article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the present work is an extension of our previous one @xcite on the investigation of unitarity effects in small-@xmath1 processes .
it was found experimentally at hera that both the total cross section of a highly virtual photon , @xmath3 , and the cross section for its diffractive dissociation have a fast increase with energy .
this is related to a fast increase of densities of quarks and gluons as the bjorken variable @xmath1 decreases .
the dynamics of such very dense partonic systems is very interesting and has been studied by many authors both in deep inelastic scattering ( see ref .
@xcite for reviews and ref .
@xcite for some recent papers ) and in high energy nuclear interactions @xcite .
unitarity effects should stop the increase of densities at extremely small @xmath1 and lead to a `` saturation '' of parton densities .
it is important to determine the region of @xmath1 and @xmath4 where the effects of saturation become important .
we study this problem using reggeon calculus @xcite with a supercritical pomeron ( @xmath5 ) and the partonic picture of @xmath0-interaction in qcd . in our previous paper @xcite we used this approach for the description of hera data in the region @xmath6 @xmath7 , where the effects of unitarity are most important .
it was shown that , with a single pomeron of intercept 1.2 and multipomeron exchanges ( unitarity effects ) , it is possible to obtain a self - consistent , simultaneous , description of both the total @xmath0-cross section and diffractive production in high - energy @xmath0-interactions . in such approach , it is convenient to consider the process of @xmath0-interaction in the laboratory frame as an interaction of the @xmath2-pair , produced by the photon , with the proton .
we separated the @xmath2-pair fluctuation into two components , `` aligned '' component with a strongly asymmetric sharing of the momentum fraction @xmath8 between @xmath9 and @xmath10 , and the rest ( `` symmetric '' ) component .
such a separation is important at large @xmath4 , where the first component has a large transverse size , while the `` symmetric '' component has a size @xmath11 and thus has a small cross section @xmath12 of interaction with the target .
both components give a contribution to the @xmath3 which behaves as @xmath13 at large @xmath4 , but the `` aligned '' component gives the main contribution to the diffraction production cross section .
triple pomeron diagrams were also included in our model . in this paper
we propose a more direct separation of the two components of the @xmath2-pair , which is valid also for small @xmath4 . the separation into a small size ( @xmath14 ) and a large size ( @xmath15 ) components of the @xmath2 pair is now made in terms of the transverse distance @xmath16 between @xmath9 and @xmath10 .
the border value , @xmath17 , is treated as a free parameter - which turn out to be @xmath18 fm .
for the @xmath14-component , with @xmath19 , we use the expression for the @xmath0 total cross - section obtained in perturbative qcd @xcite@xcite .
^*p^(tot)t(l)(s , q^2 ) = _ 0^r_0 d^2r _ 0 ^ 1 dz | ^t(l)(r , z , q ) |^2 _ s(r , s , q^2 ) , where @xmath20 and @xmath15 correspond to transverse and longitudinal polarizations of a virtual photon , @xmath21 are the corresponding wave functions of the @xmath2-pair : [ wave1 ] |^t(r , z , q ) |^2=6_e.m.4 ^ 2 _
q e_q^2\ { [ z^2+(1-z^2)]^2k_1 ^ 2(r)+m_q^2k_0 ^ 2 ( r ) } , and [ wave2 ] |^l(r , z , q ) |^2=6_e.m.4 ^ 2 _ q e_q^2\ { 4q^2z^2(1-z)^2k_0 ^ 2(r ) } , where @xmath22 . @xmath23 and @xmath24 are mcdonald functions .
the sums are over quark flavors and we have taken @xmath25 .
@xmath26 is the total cross section for the interaction of the @xmath2-pair with the proton . for the interaction of a small size dipole [ 2e ] _
s(r , s , q^2 ) = r^2 f(s , q^2 ) .
as for the @xmath15 component , we use the same parametrizations introduced in ref .
@xcite for the aligned component .
we write the @xmath0 total cross section [ 3e ] _ ^*p^(tot)(s , q^2 ) = 4 ^2 _
e.m q^2 f_2(x , q^2 ) , in the following form , using the impact parameter ( @xmath27 ) representation [ 4e ] _
^*p^(tot)(s , q^2 ) = 4 d^2b _ ^*p^(tot)(b , s , q^2 ) [ 5e ] _ ^*p^(tot)(b , s , q^2 ) = g_l^2(q^2 ) _ l^(tot)(b , s , q^2 ) + _ s^(tot)(b , s , q^2 ) .
the function @xmath28 determines the coupling of the photon to the large size @xmath2 pair and is chosen in the form @xcite [ 6e ] g_l^2(q^2 ) = g_l^2(0 ) 1 + q^2 m_l^2 where @xmath29 and @xmath30 are phenomenological parameters .
the cross section for the @xmath15-component , @xmath31 , in the impact parameter space , is chosen in the quasi - eikonal form @xcite [ 7e ] _
l^(tot)(b , s , q^2 ) = 1 - ( - c _ l(b , s , q^2 ) ) 2c , [ 8e ] _
l(s , b , q^2 ) = _ l0^p(b , ) 1 + a _ 3(s , b , q^2 ) + _ l0^f ( b , ) . the eikonal functions @xmath32 @xmath33 are written in a standard regge form [ 9e ] _
l0^k(b , ) = c_l^k _ 0k^l ( ) ( _ k - b^2 4 _ 0k^l ( ) ) , where [ 10e ] _ k = _
k(0 ) - 1 , = n s + q^2 s_0 + q^2 , _ 0k^l = r_0kl^2 + _k .
here @xmath34 is the intercept of trajectory @xmath35 and @xmath36 its slope .
the values of the radii @xmath37 , based on ref .
@xcite , are given in table 1 .
the quantity @xmath38 is chosen in such a way as to behave as @xmath39 for large @xmath4 and as @xmath40 for @xmath41 . the coefficients @xmath42 and @xmath43 determine respectively the residues of the pomeron and @xmath44-reggeon exchanges in the @xmath2-proton interaction .
the coefficient @xmath45 takes into account the dissociation of a proton @xcite .
we turn next to the denominator of eq .
( [ 8e ] ) . the constant @xmath46 si given by @xmath47 , where @xmath48 is the proton - pomeron coupling and @xmath49 is the triple pomeron coupling , both at @xmath50 . the function @xmath51 is given by eq .
( [ 29e ] ) of section 3 . with @xmath52 ,
the model described above is a standard quasi - eikonal model with born terms given by pomeron plus @xmath44 exchanges .
the denominator in eq .
( [ 8e ] ) corresponds to a resummation of triple pomeron branchings ( the so - called fan diagrams ) .
( for a full discussion on the interpretation of this denominator see ref .
thus , expressions ( [ 7e ] ) and ( [ 8e ] ) correspond to a sum of diagrams of the type shown in fig . 1
. we turn next to the @xmath14 component . in this case we put , in complete analogy with eqs .
( [ 7e])-([9e ] ) [ 11e ] _
s^(tot)(r , b , s , q^2 ) = 1 - ( - c _ s(r , b , s , q^2 ) ) 2c , [ 12e ] _ s(r , b , s , q^2 ) = _ s0(r , b , s , q^2 ) 1 + a _ 3(b , s , q^2 ) , [ 13e ] _
s0(r , b , ) = c_s^p r^2 _ 0p^s ( ) ( _ p- b^2 4 _ 0p^s ( ) ) , with @xmath53 . note that the contribution of the @xmath44-exchange to the @xmath14 component is very small and has been neglected @xcite .
the condition ( [ 2e ] ) , valid for fixed @xmath54 and @xmath4 as @xmath55 , is a property of the single pomeron exchange .
thus a factor @xmath56 has been introduced in eq .
( [ 13e ] ) .
finally @xmath26 in eq .
( [ 1e ] ) is obtained from @xmath57 , defined by eqs .
( [ 11e ] ) to ( [ 13e ] ) , as ( see eq .
( [ 4e ] ) ) [ 14e ] _
s(r , s , q^2 ) = 4 d^2b _ s(r , b , s , q^2 ) . inserting this expression in eq .
( [ 1e ] ) we obtain the transverse and longitudinal contributions of the @xmath14-component to the total @xmath0 cross - section .
following ref .
@xcite we express the total diffractive dissociation cross - section of a virtual photon as a sum of three terms [ 15e ] _
^*p^(diff ) = _ i = l , s _
i^(0 ) + _ ppp where [ 16e ] _
l^(0 ) = 4 g_l^2(q^2 ) ( _ l^(tot)(b , s , q^2 ) ) ^2 d^2b , [ 17e ] _
s^(0)t , l = 4 d^2b _ 0^r_0 d^2r _ 0 ^ 1 dz | ^t , l(z , r ) |^2 ( _ s^tot(r , b , s , q^2 ) ) ^2 [ 18e ] & & _ ppp = 2 g_l^2(q^2 ) ^l_ppp(b , s , q^2 ) e^- 2c _ l(b , s , q^2 ) d^2b + & & + 2 d^2b _ 0^r_0 ^ 2 d^2r _ 0 ^ 1 dz _ t , l | ^t , l(z , r ) |^2 _ ppp^s ( b , s , q^2 ) e^-2c_s(r , b , s , q^2 ) . here _
ppp^l(b , s , q^2 ) = a _ l^p(b , s , q^2 ) _ 3(b , s , q^2 ) [ 19e ] and _ ppp^s(r , b , s , q^2 ) = a _ s(r , b , s , q^2 ) _ 3(b , s , q^2 ) [ 20e ] where @xmath58 is given by the first term of eq .
( [ 8e ] ) and @xmath51 is defined by eq .
( [ 29e ] ) . using this expression , we see that , to first order in @xmath46 , @xmath59 consists of the sum of a triple pomeron ( @xmath60 ) term plus a @xmath61 one .
we call this sum triple pomeron , although the second one is an interference term . for the total diffractive production cross - section , that includes the diffraction dissociation of a proton , eqs .
( [ 16e])-([18e ] ) must be multiplied by the same factor @xmath62=1.5 of the total @xmath0 cross - section . at hera ,
differential diffractive cross sections are given as a function of @xmath63 , where @xmath64 is the mass of the diffractively produced system , or of @xmath65 .
they are usually integrated over @xmath66 , and the function @xmath67 is introduced [ 21e ] x_p f_2d^(3 ) = q^2 4 ^2 _ e.m . x_p
ddx_pdt dt . in our model
, this function can be written as a sum of three terms [ 22e ] f_2d^(3 ) = ( _ i = l , s f_2di^(3 ) ( x , q^2 , ) + f_2dppp^(3)(x , q^2 , ) ) . here
[ 23e ] x_p f_2dl^(3 ) = q^2 g_l^2(q^2)4 _ e.m . _
i , k = p , fd^2b _
( 1 - ) ^n_p(q^2 ) _ _ min^_max d ^_i+_k-_f(1 - ) ^n_p(q^2 ) and [ 24e ] x_p f_2ds^(3 ) = q^2 4 _ e.m .
( _ s^(0)t ^3 ( 1 - 2)^2 _ _ min^_max d ^3(1 - 2)^2 + _ s^(0)l ^3 ( 1 - ) _ _ min^_max d ^3(1 - ) ) where @xmath68 , @xmath69 and @xmath70 with @xmath71 . in eq .
( [ 23e ] ) @xmath72 corresponds to eq .
( [ 16e ] ) keeping only the linear term in @xmath31 and @xmath73 is the contribution of the @xmath74 in eq .
( [ 8e ] ) .
the @xmath75-dependence of the @xmath14-component has been taken from the qcd results of ref .
the @xmath75-dependence of the @xmath15-component was chosen according to ref . @xcite and [ 25e ] n_p(q^2 ) = - 1 2 + 3 2 ( q^2 c + q^2 ) , with @xmath76 gev@xmath77 .
the triple - pomeron ( i.e. @xmath60 plus @xmath61 ) contribution , @xmath78 , is given by [ 26e ] x_p f_2dppp^(3)(x , q^2 , ) = x_p f_2dppp^(3)b(x , q^2 , ) _ ppp _ ppp^b , where @xmath59 is given by eq .
( [ 18e ] ) , its born term , @xmath79 , by the same equation with @xmath80 , and [ 27e ] & & x_p f_2dppp^(3)b(x , q^2 , ) = q^2 4 ^2 _ e.m .
2a d^2b _ 3(b , s , q^2 , ) + & & \ { g_l^2(q^2 )
_ l^p(b , s , q^2 ) + _ t , l _ 0^r_0 d^2r _ 0 ^ 1 dz | ^t , l(r , z ) here [ 28e ] _
3(s , b , q^2 , ) = _ k = p , f _ k ( - b^2 4 _ k ( ) ) ( ) ^_k ( 1 - ) ^n_p(q^2 ) + 4 _ k ( ) where @xmath81 , @xmath82 determines the strength of the @xmath61-contribution relative to the @xmath60 one , and @xmath83 . the function @xmath84 , which enters in eqs .
( [ 8e ] ) , ( [ 12e ] ) , ( [ 19e ] ) and ( [ 20e ] ) is given by [ 29e ] _
3(s , b , q^2 ) = _ _ min^^max d _ 3(s , b , q^2 , ) .
since the triple pomeron formula is not valid for low masses , we use here @xmath85 gev .
the model was used to perform a fit of the data on structure function @xmath86 and diffractive structure function @xmath87 , in the region of small @xmath1 ( @xmath88 ) and @xmath89 gev@xmath77 only data with @xmath90 gev@xmath77 were included in the fit . ] . in table 1 ,
the full list of parameters ( fitted and fixed ) is given . in fig .
2 , the results for @xmath91 are given as a function of @xmath1 for different values of @xmath4 .
the description of the data is good . @xmath14 and
@xmath15 contributions are shown separately in order to see the different behaviors .
the @xmath14 contribution is almost negligible for very small @xmath4 , becoming comparable to the @xmath15 one at larger @xmath4 values . in figs .
3 to 6 we compare the model with the experimental data on diffraction . in order to do such a comparison ,
it is necessary to take into account that different experiments use slightly different definitions of diffractive events . in this way
we have multiplied eq .
( [ 16e])-([18e ] ) by a factor @xmath92=1.1 in order to compare with data from h1 experiment and @xmath92=1.3 for zeus .
this would improve the agreement in fig .
5.1 . ] . with these factors
we take into account the different cuts in the mass of the diffractively dissociated proton ( larger in the case of zeus ) . as in the previous case
, we plotted @xmath15 , @xmath14 and @xmath60 contributions separately . in fig .
3 , we show our results for the @xmath93dependence of @xmath94 for @xmath95=0.003 and for two values of @xmath4 . in fig .
4 , the results are given as a function of @xmath95 for different values of @xmath75 and @xmath4 .
for the highest values of @xmath4 , only comparison with @xmath75=0.4 and @xmath75=0.65 are given .
for smaller values of @xmath75 , qcd evolution becomes important . in figs .
5 , the energy dependence of the diffraction cross - section is shown for different values of @xmath64 and @xmath4 . in fig .
6 , the @xmath96dependence of the model on diffractive dissociation in photoproduction is compared with hera data for two different energies . only data with @xmath97
100 gev@xmath77 are shown for comparison . for larger values
, the effect of the non - diffractive @xmath98 contribution ( not included in the model ) is expected to be large .
we have introduced a model of the eikonal type to describe total and diffractive @xmath0 interactions .
the @xmath0 interaction is viewed as that of a @xmath2 pair , produced by the virtual photon , with the proton .
the @xmath0 total cross - section is separated into two components : large size ( @xmath15 ) for @xmath99 and small size ( @xmath14 ) for @xmath100 , where @xmath16 is the transverse distance between @xmath9 and @xmath10 .
the value of @xmath17 - treated as a free parameter - turns out to be @xmath101 fm . for the @xmath15-component ,
all the @xmath4-dependence is given by the coupling of @xmath102 to the large size @xmath2 pair - which is taken as @xmath13 at large @xmath4 ( eq . ( [ 6e ] ) ) .
for the @xmath14-component , the @xmath4-dependence is given by the wave function of the @xmath2 pair ( eqs .
( [ wave1 ] , [ wave2 ] ) ) , computed in perturbative qcd . at large @xmath4 , @xmath103 , and the unitarity corrections of the @xmath14 component are higher twist , whereas those of the @xmath15 do nt depend on @xmath4 .
a good description of the small @xmath1 data is obtained both for @xmath104 and diffractive production , in a broad region of @xmath4 ( @xmath105 gev@xmath77 ) , with a single pomeron of intercept @xmath106 . for larger values of @xmath4 , qcd evolution becomes important .
in particular it will give rise to a behavior @xmath107 with @xmath108 significantly larger than 0.2 at large @xmath4 @xcite .
for diffraction , this evolution has rather small effects at intermediate values of @xmath75 @xcite .
this allows us to use our model , in this case , without qcd evolution , up to rather large @xmath4 and moderate @xmath75 . in the region @xmath109
gev@xmath77 the unitarity effects are very important and produce a significant decrease of the effective pomeron intercept @xmath110 with decreasing @xmath4 .
this decrease is controlled by the strength of the unitarity corrections .
this , in turn , is controlled by the ratio @xmath111 and its dependence on @xmath4 .
hence the importance of describing both total cross - sections and diffractive production . in our case ,
@xmath112 and the unitarity corrections are more important in the @xmath15 component than in the @xmath14 . moreover , these corrections are higher twist at large @xmath4 in the second case .
this is more clearly seen in the diffraction , where the @xmath14 contribution to @xmath113 is much smaller than the @xmath15 one for all but the larger @xmath75 values .
an important result of our analysis is not only the fact that we can describe the data on both structure function and diffractive production in a broad region of @xmath4 with a single pomeron , but also that we can describe diffractive production at @xmath41 and at intermediate @xmath4 using the same value of the triple pomeron coupling ( which appears in our parameter @xmath46 ) .
finally , we would like to discuss the large @xmath114 limit of the total @xmath0 cross - section in our model .
the @xmath115 , given by eqs . ( [ 7e])-([9e ] ) , tend to saturate fastly with increasing @xmath54 to the value @xmath116 , due to the large @xmath117 .
the situation is different for the @xmath14 component .
let s forget for the moment about the triple pomeron contribution , i.e. consider the case @xmath46=0 .
as we have said , unitarity corrections are much smaller in the @xmath14 component , so , saturation will take place at much bigger energies , when the @xmath118 term gets large enough . for such energies , cross section in the small impact parameter will saturate to a @xmath4-independent value and @xmath119 .
this is the usual picture in perturbative qcd @xcite-@xcite .
however , by including the nonperturbative ( large distance ) @xmath60 terms ( @xmath120 0 ) we obtain a different behavior . indeed , the large @xmath121 factors in the numerator and denominator of eq .
( [ 12e ] ) cancel with each other , and we have @xmath122 ) .
thus , the @xmath13 smallness of the @xmath0 cross - section is maintained in the limit @xmath123 .
it is a pleasure to thank k. boreskov , o. kancheli , g. korchemski , u. maor and c .
merino for discussions .
this work is partially supported by nato grant outr.lg 971390 .
e. g. f. and c. a. s. thank ministerio de educacin y cultura of spain for financial support .
99 a. capella , e. g. ferreiro , a. kaidalov and c. a. salgado , orsay preprint lpt 00 - 42 , hep - ph/0005049 .
l. v. gribov , e. m. levin and m. g. ryskin , phys .
rep . * 100 * , 1 ( 1983 ) .
+ e. laenen and e. levin , ann .
part . * 44 * , 199 ( 1994 ) .
+ a. b. kaidalov , surveys high energy phys .
* 9 * , 143 ( 1996 ) .
+ a. h. mueller , hep - ph/9911289 .
a. h. mueller , nucl . phys . *
b437 * , 107 ( 1995 ) .
+ a. l. ayala , m. b. gay ducati and e. m. levin , phys .
b388 * , 188 ( 1996 ) ; nucl . phys .
* b493 * , 305 ( 1997 ) ; * 510 * , 355 ( 1998 ) .
+ e. gotsman , e. levin and u. maor , nucl .
b493 * , 354 ( 1997 ) ; phys . lett . *
b425 * , 369 ( 1998 ) ; * b452 * , 387 ( 1999 ) .
+ e. gotsman , e. levin , u. maor and e. naftali , nucl . phys . *
b539 * , 535 ( 1999 ) .
+ l l. frankfurt , w. kopf and m. strikman , phys . rev .
* d57 * , 512 ( 1998 ) ; * d54 * , 3194 ( 1996 ) .
+ m. mac dermott , l. l. frankfurt , v. guzey and m. strikman , hep - ph 9912547 .
+ k. golec - biernat and m. wsthoff , phys . rev . * d59 * , 014017 ( 1999 ) ; * d60 * , 114023 ( 1999 ) .
+ a. h. mueller , eur .
* a1 * , 19 ( 1998 ) . l. mc lerran and r. venugopalan , phys . rev . * d49 * , 2233 , 3352 ( 1994 ) ; * 50 * , 2225 ( 1994 ) ; * 53 * , 458 ( 1996 )
. + j. jalilian - marian et al . , phys . rev . *
d59 * , 014014 , 034007 ( 1999 ) .
+ a. kovner , l. mc lerran and h. weigert , phys . rev . *
d52 * , 3809 , 6231 ( 1995 ) . + yu .
v. kovchegov and a. h. mueller , nucl . phys . *
b529 * , 451 ( 1998 ) .
v. kovchegov , a. h. mueller and s. wallon , nucl . phys . *
b507 * , 367 ( 1997 ) .
+ a. h. mueller , nucl . phys . *
b558 * , 285 ( 1999 ) .
+ a. capella , a. kaidalov , j. tran thanh van , heavy ion phys . *
9 * , 169 ( 1999 ) . v. n. gribov , zhetf * 57 * , 654 ( 1967 ) .
a. di giacomo and h. panagopoulos , phys .
lett * b285 * , 133 ( 1992 ) . a. di giacomo , e. meggiolaro and h. panagopoulos , nucl
b483 * , 371 ( 1997 ) .
n. n. nikolaev and b. g. zakharov , zeit . phys . *
c49 * , 607 ( 1990 ) .
l. l. frankfurt and m. strikman , phys .
rep . * 160 * , 235 ( 1988 ) .
k. a. ter - martirosyan , nucl
b36 * , 566 ( 1972 ) .
a. haakman , a. kaidalov and j. h. koch , phys .
* b365 * , 411 ( 1996 ) .
bartels , l. l. frankfurt and m. strikman , phys .
* 160 * , 235 ( 1988 ) .
a. capella , a. kaidalov , c. merino , j. tran thanh van , phys .
lett . * b337 * , 358 ( 1994 ) ; phys
. lett . * b343 * , 403 ( 1995 ) .
a. kaidalov , c. merino , eur .
j. * c10 * , 153 ( 1999 ) , a. kaidalov , c. merino , d. perterman , hep - ph/9911331 .
a. capella , a. kaidalov , c. merino , d. perterman , j. tran than van , phys .
rev . * d53 * , 2309 ( 1996 ) . c. adloff
( h1 collaboration ) , nucl .
b497 * , 3 ( 1997 ) .
j. breitweg _
et al _ ( zeus collaboration ) , phys
. lett . * b407 * , 432 ( 1997 ) .
m. r. adams _ et al _ ( e665 collaboration ) , phys . rev .
* d54 * , 3006 ( 1996 ) .
j. breitweg _
et al _ ( zeus collaboration ) , preprint desy 00 - 071 ( hep - ex/0005018 ) .
c. adloff _ et al _
( h1 collaboration ) , z. phys . * c76 * , 613 ( 1997 ) .
j. breitweg _
et al _ ( zeus collaboration ) , eur .
j. * c6 * , 43 ( 1999 ) .
a. solano ( zeus collaboration ) , xxxv rencontres de moriond qcd and high energy hadronic interactions , les arcs , march 18 - 25 , 1999 .
c. adloff _ et al _
( h1 collaboration ) , z.phys . * c74 * , 221 ( 1997 ) .
0.5 cm * figure 1 . * a generic reggeon diagram of our model .
it contains the s - channel iteration of pomeron and @xmath44 exchanges , triple pomeron ( @xmath124 ) diagrams , as well as multiple t - channel branchings of the pomeron of the fan - diagram type .
* figure 2 . * @xmath91 as a function of @xmath1 for different values of @xmath4 compared with experimental data from h1 1995 @xcite ( open squares ) , zeus 1995 @xcite ( black circles ) , e665 @xcite ( black triangles ) ( notice that the corresponding @xmath4 values of these data are slightly different ) and zeus bpt97 @xcite ( open circles ) .
dotted curve corresponds to the @xmath15 contribution , dashed one to the @xmath14 contribution and solid one to the total @xmath91 given by the model .
* figure 3 . *
@xmath94 as a function of @xmath75 for fixed @xmath125 and for @xmath4=4.5 gev@xmath77 and @xmath4=7.5 gev@xmath77 .
experimental data are from @xcite .
dotted lines correspond to ppp contribution , dashed ones to l term and dotted - dashed to s one .
* figure 4.1 .
* @xmath94 as a function of @xmath95 for @xmath4=4.5 and 7.5 gev@xmath77 and fixed @xmath75=0.04 , 0.1 , 0.2 , 0.4 , 0.65 and 0.9 .
the curves correspond to the convention of fig .
3 . experimental data are from @xcite .
* figure 4.2 . *
@xmath94 as a function of @xmath95 for @xmath4=9 , 12 and 18 gev@xmath77 and fixed @xmath75=0.4 and 0.65 .
the curves correspond to the convention of fig .
3 . experimental data are from @xcite .
* figure 5.1 . * energy dependence of diffractive cross section for @xmath64=2 , 5 and 11 gev and @xmath4=8 and 14 gev@xmath77 .
experimental data are from @xcite .
the curves correspond to the convention of fig .
* figure 5.2 .
* energy dependence of diffractive cross section for different mass intervals and for low-@xmath4 compared with experimental data from @xcite .
the curves correspond to the convention of fig . | we investigate the role of unitarization effects in virtual photon - proton ( @xmath0 ) interactions at small @xmath1 .
the @xmath2-fluctuation of the initial photon is separated into a small distance and a large distance component and a model for the unitarization of each component is proposed .
the born approximation for the small size component is calculated using qcd perturbation theory .
reggeon diagram technique is used in order to obtain a self - consistent scheme for both total @xmath0 cross section and diffractive production .
the model gives a good description of hera data in the small-@xmath1 region , with a single pomeron of intercept 1.2 .
= 21 truecm = 15 truecm to 2 truecm * deep inelastic scattering data and the * 3 truemm * problem of saturation in small - x physics * 1 truecm * a. capella , e. g. ferreiro , c. a. salgado * laboratoire de physique thorique 8627 ] universit de paris xi , btiment 210 , f-91405 orsay cedex , france 5 truemm * a. b. kaidalov * itep , b. cheremushkinskaya ulitsa 25 117259 moscou , russia 1 truecm 1 truecm lpt orsay 00 - 50 june 2000 = 24 pt |
coalescing black hole - neutron star ( bhns ) binaries , as well as other compact binaries composed of neutron stars and/or black holes , are among the most promising sources of gravitational waves for both ground - based @xcite and space - based laser interferometers @xcite .
bhns binary mergers are also candidate central engines of short - hard gamma - ray bursts ( sgrbs ) ( see , e.g. @xcite and references cited therein ) .
the remnants of both bhns binary mergers @xcite and binary neutron star mergers @xcite are feasible progenitors for sgrbs because both may result in black holes surrounded by hot , massive accretion disks with very little , if any , baryon contamination along the polar symmetry axis . motivated by these factors ,
considerable effort has gone into the study of bhns binaries .
most approaches to date assume newtonian gravity in either some or all aspects of the calculation ( see , e.g. @xcite for quasiequilibrium calculations and @xcite for dynamical simulations ) .
more recently , several groups have also studied bhns binaries in a fully relativistic framework , both for quasiequilibrium models @xcite and dynamical simulations @xcite .
our group has pursued a systematic approach to developing increasingly realistic models of bhns binaries in quasiequilibrium circular orbits .
our first studies @xcite assumed extreme mass ratios , i.e. , black hole masses that are much greater than the neutron star mass . while this is a very natural first step from a computational point of view , binaries with comparable masses are much more interesting from the perspective of ground - based gravitational wave observations and for the launching of sgrbs .
more recently we have therefore relaxed this assumption and have extended our results to the case of comparable - mass bhns binaries @xcite .
specifically , in @xcite ( hereafter paper i ) we constructed quasiequilibrium models by solving einstein s constraint equations in the conformal thin - sandwich formalism , assuming conformal flatness and maximal slicing , together with the relativistic equations of hydrostationary equilibrium .
we accounted for the black hole by excising a coordinate sphere and imposing the equilibrium black - hole boundary conditions of cook and pfeiffer @xcite .
this original version implemented a
leading - order " approximation to nonspinning black holes , which equates an otherwise undetermined spin parameter @xmath1 that appears in the boundary condition for the shift vector with the orbital angular velocity seen by an inertial observer at infinity , @xmath2 . as for the original irrotational binary black hole models of @xcite
, this condition does not lead to simultaneous turning points of the binding energy and the total angular momentum in constant - mass sequences in paper i. such simultaneous turning points are expected for those sequences if they are truly in quasiequilibrium @xcite .
an improvement over this condition , namely to iterate over @xmath1 until the quasilocal spin angular momentum of the black hole vanishes , was suggested and implemented for binary black holes by @xcite . in this paper
we reconstruct quasiequilibrium models of bhns binaries using the same techniques as in paper i , but with the improved black hole spin angular velocity condition as suggested by @xcite .
we then compute sequences of bhns binaries in quasicircular orbits for a wider range of neutron star masses and binary mass ratios than in paper i , focusing our attention on irrotational neutron stars orbiting nonspinning black holes . here
we focus only on the irrotational state for the neutron star because it is astrophysically considered to be more realistic in a bhns binary @xcite . on the other hand , we will compute the case of spinning black holes in future work . as was the case for the irrotational black hole binaries constructed in @xcite , we find that this improved condition for the spin parameter of the black hole @xmath1 does lead to simultaneous turning points in the binding energy and the total angular momentum along constant - mass sequences .
the paper is organized as follows .
we briefly review the basic equations in section ii .
we present numerical results in section iii , and outline some qualitative considerations concerning the fate of bhns binaries in section iv . in section v we summarize our findings . throughout this paper
we adopt geometrized units with @xmath3 , where @xmath4 denotes the gravitational constant and @xmath5 the speed of light .
latin and greek indices denote purely spatial and spacetime components , respectively .
in this section we review the equations we solve to construct a quasiequilibrium bhns binary .
the equations are very similar to those in paper i , but , as we explain below , we have improved both our algorithm for the solution of the gravitational field equations and the method to compute the spin parameter @xmath1 that appears in the black hole boundary condition for the shift vector . for a more detailed discussion about the formalism we use here ,
we refer to paper i , the review articles @xcite , and sec .
ii of @xcite for the hydrostatics .
the line element in @xmath6 form is written as @xmath7 where @xmath8 is the spacetime metric , @xmath9 the lapse function , @xmath10 the shift vector , and @xmath11 the spatial metric induced on a spatial slice @xmath12 .
the spatial metric @xmath11 is further decomposed according to @xmath13 , where @xmath14 denotes the conformal factor and @xmath15 the conformal background spatial metric , defined such that @xmath16 in cartesian coordinates .
we also decompose the extrinsic curvature @xmath17 into a trace @xmath18 and a traceless part @xmath19 according to @xmath20 the hamiltonian constraint then becomes @xmath21 here @xmath22 is the covariant laplace operator , @xmath23 the covariant derivative , @xmath24 the ricci tensor , and @xmath25 the scalar curvature , all associated with the conformal background metric @xmath15 .
we employ the conformal thin - sandwich decomposition of the einstein equations @xcite . in this decomposition ,
we use the evolution equation for the spatial metric to express the traceless part of the extrinsic curvature in terms of the time derivative of the background metric , @xmath26 , and the gradients of the shift vector . under the assumption of equilibrium ,
i.e. , @xmath27 in a corotating coordinate system , the traceless part of the extrinsic curvature reduces to @xmath28 inserting eq .
( [ eq : taij ] ) into the momentum constraint we obtain @xmath29 for the construction of quasiequilibrium data it is also reasonable to assume @xmath30 in a corotating coordinate system .
the trace of the evolution equation for the extrinsic curvature then yields @xmath31 the matter terms on the right - hand side of eqs .
( [ eq : ham ] ) , ( [ eq : mom ] ) , and ( [ eq : kdot ] ) are derived from the projections of the stress - energy tensor @xmath32 into the spatial slice @xmath12 . denoting the future - oriented unit normal to @xmath12 as @xmath33 , the relevant projections of @xmath32 are @xmath34 here we write the stress - energy tensor as @xmath35 assuming an ideal fluid .
the quantity @xmath36 is the fluid 4-velocity , @xmath37 the baryon rest - mass density , @xmath38 the internal energy density , and @xmath39 the pressure . equations ( [ eq : ham ] ) , ( [ eq : mom ] ) and ( [ eq : kdot ] ) provide equations for the lapse function @xmath9 , the shift vector @xmath10 , and the conformal factor @xmath14 , while @xmath40 can be found from eq .
( [ eq : taij ] ) .
the conformally related spatial metric @xmath15 and the trace of the extrinsic curvature @xmath18 remain freely specifiable , and have to be chosen before we can solve the above equations ( note that we have already set the time derivatives of these quantities , which are also freely specifiable , to zero ) . as in paper
i we assume a flat background @xmath41 , where @xmath42 denotes a flat spatial metric , and maximal slicing , @xmath43
. in cartesian coordinates , eqs .
( [ eq : ham ] ) , ( [ eq : mom ] ) and ( [ eq : kdot ] ) can then be written as @xmath44 where @xmath45 and @xmath46 denote the flat laplace operator and the partial derivative , and @xmath47 .
( [ eq : taij ] ) becomes @xmath48 for numerical purposes we further decompose the variables and their equations into parts associated with the black hole and the neutron star . for details of this decomposition we refer to appendix a of paper i @xcite .
in paper i we solved directly for the lapse function @xmath9 , the conformal factor @xmath14 and the shift vector @xmath10 . instead of solving eq .
( [ eq : kdot ] ) for the lapse , we now solve eq .
( [ eq : trace_evolv ] ) for the combination @xmath49 .
this choice is quite common ( see e.g. @xcite ) , and has the advantage of eliminating the source term @xmath50 on the right - hand side of eq .
( [ eq : kdot ] ) above , or equivalently eq . ( 9 ) in paper i. this term falls off like @xmath51 , and hence more slowly than @xmath52 . eliminating this term therefore enhances the accuracy of the numerical solution
. we will quantify the improvement in section [ sec : iiib ] . in order to solve the gravitational field equations ( [ eq : ham_constr ] ) , ( [ eq : mom_constr ] ) , and ( [ eq : trace_evolv ] ) , we have to impose appropriate boundary conditions on two different boundaries : outer boundaries at spatial infinity and inner boundaries on the black hole horizons .
the boundary conditions at spatial infinity follow from the assumption of asymptotic flatness . with the help of a radial coordinate transformation @xmath53 in the external computational domain
, our computational grid extends to spatial infinity @xcite , and we can impose the exact boundary conditions @xmath54 here @xmath2 is the orbital angular velocity of the binary system measured at infinity , and @xmath55 is a cartesian coordinate centered on the center of mass of the binary system .
we express the shift vector @xmath10 in a corotating coordinate system that we adopt throughout our calculation . in an inertial coordinate system
, the shift vector would tend to zero at spatial infinity , while in the corotating coordinate system of the numerical code the shift vector diverges at spatial infinity . for computational purposes , it is therefore convenient to write the shift vector as a sum of the rotational shift term @xmath56 and a residual part ( which tends to zero at spatial infinity ) , and solve the equations only for the latter .
the inner boundary conditions arise from the excision of the black hole interior .
the assumption that the black hole is in equilibrium leads to a set of boundary conditions for the conformal factor and shift vector @xcite ( see also @xcite as well as the related isolated horizon formalism , e.g. @xcite ) .
the boundary condition for the conformal factor is @xmath57 where @xmath58 is the outward pointing unit vector normal to the excision surface and @xmath59 is the induced metric on the excision surface , @xmath60 .
the quantity @xmath61 is computed from the projection of the extrinsic curvature @xmath62 as @xmath63 .
the boundary condition on the normal component of the shift vector is @xmath64 the tangential components must form a conformal killing vector of the conformal metric @xmath65 on the excision surface ( see @xcite ) .
this can be achieved by choosing them to be killing vectors of a 2-sphere , @xmath66 here @xmath67 is a freely specifiable vector , related to the black - hole spin , that we take to be aligned with the @xmath68-axis , and @xmath69 is a cartesian coordinate centered on the 2-sphere .
we assume the excision surface to be a coordinate sphere .
in our previous treatment we implemented a
leading - order " approximation to nonspinning black holes and set @xmath70 , where @xmath2 is the orbital angular speed ( compare @xcite ) .
following @xcite we now iterate over @xmath1 until the black hole s quasilocal spin angular momentum @xmath71 vanishes . here @xmath72 is an approximate killing vector of @xmath59 that we find by solving the killing transport equations as described in @xcite ( see also @xcite for more detailed descriptions and @xcite for alternative methods for finding approximate killing vectors of closed 2-surfaces ) . according to @xcite ,
the boundary condition on the lapse function can be chosen freely . in this paper , we choose a neumann boundary condition @xmath73 on the excised surface .
we refer to sections ii.e and ii.f of paper i for a discussion of how the orbital angular velocity , the center of rotation , and several global quantities including total angular momentum and mass are computed . as in paper
i @xcite , we construct our numerical code based on the lorene spectral - methods library routines developed by the meudon relativity group @xcite . in our code
, the computational space is broken into multiple domains .
each domain around the black hole is covered by @xmath74 or @xmath75 collocation points , while those around the neutron star are covered by @xmath76 collocation points .
here @xmath77 , @xmath78 , and @xmath79 denote the number of collocation points for the radial , polar , and azimuthal directions , respectively .
we use a larger number of collocation points for the black hole domains than for the neutron star domains because the source terms of the black hole equations are sensitive to the resolution of the neutron star . since the black hole domains are centered on the black hole , we need a higher angular resolution to adequately resolve these source terms .
the neutron star equations , on the other hand , have large source terms only near the neutron star . since the neutron star domains are centered on the neutron star ,
a more modest angular resolution is sufficient to resolve these terms .
we refer to appendix a of paper i for a detailed discussion of the decomposition of the equations and their source terms .
throughout this paper , we model the neutron - star equation of state by the polytropic relation @xmath80 , where @xmath39 is the pressure , @xmath37 the baryon rest - mass density , @xmath81 the adiabatic index , and @xmath82 a constant .
we choose @xmath0 for the adiabatic index , and compute several different constant - mass inspiral sequences .
specifically the rest mass of the neutron star and irreducible mass of the black hole are kept constant along each sequence .
we focus on baryon rest masses for neutron stars in the range of @xmath83 ( see fig . [
fig : mass - radius ] ) . here and
in the following we normalize all dimensional quantities in terms of the polytropic length scale @xmath84 , e.g. @xmath85 . in terms of compaction ,
our models are in the range @xmath86 , where @xmath87 is the compaction of a spherical neutron star , @xmath88 the neutron star s adm mass in isolation , and @xmath89 its areal radius .
our most compact polytropic model with @xmath90 is very close to the maximum baryon rest mass for spherical @xmath0 polytropes in isolation , @xmath91 . .
the horizontal axis denotes the isotropic radius and the vertical axis the baryon rest mass ( solid curve ) and gravitational mass ( dashed curve ) , where the radius and masses are in polytropic units .
the gravitational mass is equivalent to the adm mass for an isolated , spherical neutron star .
filled circles denote the models we choose in this paper .
the compaction parameter @xmath92 for each model is also shown.,width=302 ] we consider mass ratios in the range @xmath93 , where we define the mass ratio as @xmath94 i.e. , the ratio of the irreducible mass of the black hole ( @xmath95 ) to the adm mass of a spherical , isolated neutron star ( @xmath88 ) . here
the irreducible mass @xmath95 is identical to the adm mass for an isolated nonspinning black hole .
note again that we fix the irreducible mass of the black hole and the baryon rest mass of the neutron star for the construction of constant - mass sequences .
for the definition of the mass ratio , however , we use the adm mass of a spherical isolated neutron star @xmath88 , because this turns out to be more convenient for comparisons with third - order post - newtonian ( 3pn ) results @xcite .
we tabulate our numerical results in appendix [ appendix : tables ] . in the equatorial plane for our innermost configuration with binary mass ratio @xmath96 and neutron - star baryon rest mass @xmath97 .
the cross `` @xmath98 '' indicates the position of the rotation axis.,width=302 ] we show contours of the conformal factor @xmath14 for a bhns binary with mass ratio @xmath96 and neutron - star baryon rest mass @xmath99 in fig .
[ fig : confo ] .
this figure represents the configuration at the smallest orbital separation for which our code converged .
the thick sold circle on the left - hand side denotes the position of the excised surface ( the apparent horizon ) , while that on the right - hand side denotes the position of the neutron - star surface .
the value of @xmath14 on the excised surface is not constant because a neumann boundary condition ( [ eq : psi_bc ] ) is applied . in figs .
[ fig : ej - o1m015 ] [ fig : ej - o5m015 ] we show the binding energy @xmath100 and the total angular momentum @xmath101 as a function of the orbital angular velocity @xmath102 for neutron stars with baryon rest mass @xmath97 and for mass ratios @xmath103 , 3 , and 5 . here
we define the binding energy as the difference in the total adm mass of the binary from that at infinite orbital separation , @xmath104 , and the total adm mass of the binary at infinite orbital separation as @xmath105 . in these figures
we also include , for comparison , results from paper i and 3pn approximations @xcite .
we find that our new results show much better agreement with 3pn results especially for larger separations ( smaller @xmath102 ) .
this improvement is due to the change of variables we discussed in section [ sec : iia ] . in fig .
[ fig : improve ] we show the relative difference of the total angular momentum from that of 3pn approximation as a function of the orbital angular velocity @xmath102 for neutron - star baryon rest mass @xmath99 and mass ratio @xmath106 .
the solid curve shows the new results in this paper , the dashed curve the results of the old formulation in paper i but including the new method to compute @xmath1 , and the dotted - dashed curve the old results in paper i. for closer configurations ( larger @xmath102 ) , the difference is dominated by the accuracy of the spin parameter @xmath1 , while for larger separations ( smaller @xmath102 ) , the difference comes from the change of variables .
( upper panel ) and total angular momentum @xmath101 ( lower panel ) as a function of @xmath102 for binaries of mass ratio @xmath107 and neutron - star mass @xmath99 .
the solid line with filled circles shows new results , and the dashed line the old results in paper i. the dotted - dashed line denotes the results of the 3pn approximation .
the numerical sequences end due to cusp formation and hence the onset of tidal disruption before the binary reaches the innermost stable circular orbit at the turning point of the binding energy .
3pn sequences all exhibit turning points and can not reveal cusps.,width=302 ] .,width=302 ] .
the binary encounters an isco before the neutron star is tidally disrupted.,width=302 ] for binaries of mass ratio @xmath106 and neutron - star mass @xmath99 .
the solid curve shows new results in this paper , the dashed curve the results of the old formulation in paper i but using the new method to compute @xmath1 , and the dotted - dashed curve the old results in paper i. this figure demonstrates that the new spin condition improves the results predominantly at small binary separation , while the new formulation of the equations has a larger effect at large binary separations.,width=302 ] , and different neutron star compactions.,width=302 ] before discussing these results in greater detail it is useful to anticipate some qualitative scaling . from a very crude newtonian argument , we can estimate the binary separation @xmath108 at which the neutron star will be tidally disrupted by comparing the tidal force exerted by the black hole on a test mass @xmath109 on the neutron star s surface with the gravitational force exerted by the neutron star on this test mass . equating these two forces
yields @xmath110 if @xmath108 is greater ( by a sufficient amount ) than the innermost stable circular orbit ( isco ) separation @xmath111 we may expect the neutron star to be tidally disrupted before being swallowed more or less intact , albeit distorted by tidal forces , by the black hole .
the qualitative relation ( [ eq : tidal ] ) suggests that the tidal separation decreases with increasing mass ratio @xmath112 and neutron star compaction .
our sequences terminate shortly before the onset of tidal disruption .
we therefore expect to encounter minima in the binding energy and angular momentum , which identify the isco , only for binaries with sufficiently large mass ratio @xmath112 and neutron star compaction .
comparing figs .
[ fig : ej - o1m015 ] [ fig : ej - o5m015 ] we indeed find turning points only for the largest mass ratio @xmath106 .
we also note that in this case the turning points in the binding energy and the angular momentum occur simultaneously to within our numerical accuracy .
this was not the case with for our earlier results , which adopted the leading - order " nonspinning condition @xmath70 .
when imposing the more accurate condition @xmath113 for irrotational binaries , we now do find simultaneous minima ( compare also the analogous results of @xcite for binary black holes ) . to highlight this finding we graph in fig .
[ fig : turning - point ] the binding energy versus angular momentum for sequences of mass ratio @xmath114 , but different neutron star compactions . a simultaneous turning point in
the binding energy and angular momentum leads to a cusp in these curves .
as suggested by eq .
( [ eq : tidal ] ) , we do not find turning points for small compactions ( since the sequences end at tidal disruption before encountering an isco ) , but for larger compactions these curves indeed form a cusp .
while our results agree with 3pn results very well , we do note a small deviation that increases with the neutron star compaction , as one might possibly expect .
we also clearly find that our new data agrees with 3pn results much better than the old ones in paper i for smaller binary separation ( see curves of @xmath99 in fig .
[ fig : turning - point ] ) .
we note that 3pn sequences can not identify tidal disruption and therefore always exhibit turning points .
probably the most important numerical improvement that we present here is the incorporation of a method to compute the spin angular velocity of the black hole into our numerical code .
we first obtain the killing vector on the excised surface by solving the killing transport equations , and then compute the quasilocal spin angular momentum @xcite . requiring the angular momentum to be zero ( @xmath115 ) , we iterate over the black hole spin parameter @xmath1 . in paper
i , by contrast , we simply set this parameter equal to the orbital angular velocity @xmath2 , resulting in a
leading - order " approximation . in figs .
[ fig : spinmom ] and [ fig : spinvelo ] , we compare the quasilocal spin angular momentum and black hole spin parameter @xmath1 , calculated under the two conditions @xmath116 and @xmath115 . for both computations ,
we use our set of improved variables ( as opposed to reusing our results from paper i ) . in these figures ,
solid and dotted - dashed lines represent configurations with zero quasilocal spin angular momentum for the black hole , while dashed and dotted curves represent configurations where we set the spin angular velocity of the black hole equal to the orbital angular velocity . , while dashed and dotted curves result from @xmath70 .
( compare the curves with the inset in fig . 7 in .),width=302 ]
.,width=302 ] we find from fig . [ fig : spinmom ] that the quasilocal spin angular momentum of the black hole was negative and increased in magnitude as the orbital separation decreased when we used the condition @xmath117 in our boundary conditions .
this negative spin angular momentum results in a decrease in the total angular momentum compared to the @xmath115 case .
this explains why we did not find a minimum in the total angular momentum along a sequence in paper i. here , by maintaining @xmath115 within numerical errors , we find the minimum of the total angular momentum is coincident with that of the binding energy .
similarly , the fact that the quasilocal spin angular momentum becomes negative for the condition @xmath116 implies that the spin angular velocity @xmath1 should be smaller than the orbital angular velocity @xmath2 when we require @xmath115 , as confirmed by fig .
[ fig : spinvelo ] .
among the most important questions in the context of bhns binaries is whether the coalescence leads to a tidal disruption of the neutron star , or whether the neutron star gets swallowed by the black hole more or less intact , albeit distorted by tidal forces .
clearly , this question has important consequences from the perspective of gravitational wave observations , but perhaps even more important are the ramifications for sgrbs . to launch such
a burst requires the formation of an accretion disk around the black hole , which can occur only if the neutron star is disrupted prior to reaching an isco . to explore this issue quantitatively requires a dynamical simulation ( compare @xcite ) , and part of the motivation for the work presented in this paper is the construction of suitable initial data for such calculations . in the meantime , however , we may also use our quasiequilibrium models to obtain preliminary estimates . specifically
, we will use our numerical results to construct qualitative expressions that may be used to predict whether a bhns binary of arbitrary mass ratio and neutron star compaction encounters an isco before being tidally disrupted or not . to the mass - shedding point ( @xmath118 ) .
thick lines are sequences constructed using numerical data , and thin lines are extrapolated sequences.,width=302 ] we start from investigating the binary separation ( and the orbital angular velocity ) at which tidal disruption of the neutron star by the black hole occurs . in newtonian gravity and semi - relativistic approaches , simple equations may be introduced to fit the effective radius of a roche lobe @xcite .
recently , shibata and ury introduced a fitting equation for binaries composed of a nonspinning black hole and a corotating neutron star in general relativity @xcite . in this section , we investigate our data for a nonspinning black hole and an irrotational neutron star . in order to obtain a fitting formula
, we need to determine the orbital angular velocity at the mass - shedding limit , i.e. , the point that defines the onset of tidal disruption . for this purpose
we need to extrapolate our data , because it is impossible to compute sequences up to the tidal disruption point using a spectral methods code because of gibbs phenomena . to extrapolate our results
, we introduce a sensitive mass - shedding indicator , @xmath119 , defined as the ratio between radial derivatives of the specific enthalpy @xmath120 on the neutron star surface in the direction of the companion and in the polar direction , @xmath121 ( see @xcite ) .
this indicator takes the value unity for a spherical star and reaches zero at the formation of a cusp .
we tabulate @xmath119 as a function of the orbital angular velocity and extrapolate to @xmath122 by using fitting polynomial functions to find the onset of tidal disruption . in fig .
[ fig : extrap ] , we show an example of such extrapolations for sequences of neutron - star baryon rest mass @xmath123 with mass ratios 1 , 2 , 3 , and 5 . note that the horizontal axis in fig .
[ fig : extrap ] is the orbital angular velocity in polytropic units , @xmath124 .
then we prepare the data of extrapolation for all models we compute . to obtain the fitting formula for the orbital angular velocities at the mass - shedding limit acceptable for all models we compute , we start with the qualitative newtonian expression ( [ eq : tidal ] ) , together with kepler s third law @xmath125 combining eqs .
( [ eq : tidal ] ) with ( [ eq : kepler ] ) we can eliminate @xmath108 and find @xmath126 or equivalently @xmath127 here we have identified @xmath128 with @xmath95 , @xmath129 with @xmath88 , and @xmath130 with the circumferential radius of a spherical neutron star @xmath89 .
lastly , we determined the coefficient of 0.270 from our numerical results . fitting our numerical values for the angular velocity at tidal disruption to the form ( [ eq : mass - shed ] ) resulted in a narrow range of coefficients between 0.266 and 0.273 , with a mean value of 0.270 . in fig .
[ fig : extrapall ] , we show the results of the fitting of the mass - shedding limit by eq .
( [ eq : mass - shed ] ) . ) .
the mass - shedding limit for each neutron - star mass and mass ratio is computed by the extrapolation of the numerical data.,width=302 ] it may be of interest to compare eq .
( [ eq : mass - shed ] ) with a similar expression of shibata and ury @xcite , who express the critical mass ratio @xmath112 as @xmath131 ( in @xcite , the mass ratio @xmath132 is defined as the inverse of our definition ) . in terms of the quantities used in eq .
( [ eq : mass - shed ] ) , this becomes @xmath133 which agrees with eq .
( [ eq : mass - shed ] ) for large @xmath112 .
the agreement of our result ( irrotation for the neutron star ) with that by shibata and ury ( corotation ) confirms our prediction in the limit of extreme mass ratios @xcite .
we are also interested in the binary separation ( and the corresponding orbital angular velocity ) at which the minimum of the binding energy appears , corresponding to the isco . in our numerical data we locate this point by fitting a second - order polynomial to three nearby points .
we may construct a simple empirical fit that predicts the isco angular velocity @xmath134 for an arbitrary companion orbiting a black hole as follows .
we search for expressions with three free parameters that express @xmath134 as a function of the mass ratio @xmath112 and the compaction @xmath92 of the companion .
we then fix the three parameters by matching to three known values of @xmath134 , namely ( 1 ) that of a test particle orbiting a schwarzschild black hole , @xmath135 ( for @xmath136 ) , ( 2 ) that of an equal - mass binary black hole system as computed in @xcite , @xmath137 ( for @xmath107 and @xmath138 ) , and finally ( 3 ) that of our bhns configuration with @xmath99 and @xmath139 and mass ratio @xmath106 , resulting in @xmath140 .
a further requirement arises from the fact that for a test particle ( with @xmath141 ) , the expression should be independent of the companion s compaction .
we find a good fit to our remaining numerical data with the expression @xmath142 as demonstrated in fig .
[ fig : minimum ] .
we determined the exponent of @xmath112 and @xmath92 in eq .
( [ eq : isco ] ) empirically by requiring that the fitted curves lie near the data points for all models .
clearly the agreement is not perfect , but adequate for our purposes . )
.,width=302 ] combining eqs .
( [ eq : mass - shed ] ) ( or ( [ eq : mass - shed_om ] ) ) and ( [ eq : isco ] ) , we can now identify the critical binary parameters that separate those binaries that encounter an isco before reaching mass - shedding , and vice - versa . in fig .
[ fig : endpoint_example ] we show an example for @xmath99 .
the solid curve denotes the orbital angular velocity of the mass - shedding limit , and the long - dashed for the isco . as seen from eqs .
( [ eq : mass - shed ] ) and ( [ eq : isco ] ) , both of these curves depend on the mass ratio , but in different ways , which leads to the intersection of the two curves .
an inspiraling binary evolves along horizontal lines towards increasing @xmath143 , starting at the left and moving toward the right , until reaching either the isco or the mass - shedding limit .
after the binary reaches the isco for sufficiently large mass ratio , we can not determine the exact position at which the binary disrupts via quasiequilibrium calculations because it is in the dynamical plunge region , but we nevertheless include the mass - shedding limit for unstable quasiequilibrium sequences as the dotted curve in fig . [ fig : endpoint_example ] . as shown in fig .
[ fig : endpoint_example ] , the model with mass ratio @xmath144 ( dotted - dashed line ) encounters the isco , while that of @xmath96 ( dot - dot - dashed line ) ends up at the mass - shedding limit .
the intersection between the mass - shedding and isco curve marks a critical point that separates the two distinct outcomes of the binary inspiral . .
the solid curve denotes the mass - shedding limit , and the long - dashed one the isco for each mass ratio as a function of the orbital angular velocity in polytropic units .
the dotted curve denotes the mass - shedding limit for unstable quasiequilibrium sequences.,width=302 ] since expressions ( [ eq : mass - shed ] ) and ( [ eq : isco ] ) also depend on the neutron star s compaction , the specific values of the critical point likewise depend on this compaction . in fig .
[ fig : endpoint ] , we graph the mass - shedding and isco curves for a number of different neutron star masses which have a one - to - one correspondence to the compaction .
the intersections , appearing as a knee in this figure , mark the critical point for each compaction corresponding to a binary that encounters the isco at the point of tidal disruption .
this critical point also gives the maximum value of @xmath102 in quasiequilibrium for each model .
we note again that the equation of state is fixed here to be polytropic with adiabatic index @xmath0 .
this implies that the results may change , i.e. , the coefficients in eqs .
( [ eq : mass - shed ] ) and ( [ eq : isco ] ) may change , when we change the adiabatic index or the equation of state itself . however , we find from fig . [
fig : endpoint ] that the orbital angular velocity at the isco in the form of @xmath102 has a narrow range , @xmath145 , for all models we compute .
this implies that the fitting formula ( [ eq : isco ] ) may hold approximately even for other equations of state .
we tabulate these critical values for the neutron star compactions considered in this paper in table [ table : crit ] . whether or not these critical points mark the true separation between the two outcomes will have to be verified by dynamical simulations .
we nevertheless expect that these values may provide some useful guidance
. .critical value at which tidal disruption occurs at the isco for different neutron star models . [
cols="^,^,^,<,^,^,^",options="header " , ] [ table : crit ] when we eliminate @xmath102 from eqs .
( [ eq : mass - shed_om ] ) and ( [ eq : isco ] ) , we can draw a curve of the critical mass ratio which separates bhns binaries that encounter an isco before reaching mass - shedding , and vice - versa , as a function of the compaction of the neutron star .
we show such a critical curve that separates those two regions in fig .
[ fig : critpoint ] .
the solid line denotes the critical mass ratio for each compaction .
if the mass ratio of a bhns binary is larger than the critical one , the quasiequilibrium sequence terminates by encountering the isco , while if smaller , it ends at the mass - shedding limit of the neutron star .
we have constructed new quasiequilibrium configurations of black hole - neutron star binaries in general relativity . we have solved the einstein constraint equations in the conformal thin - sandwich formalism coupled with the equations of relativistic hydrostationary equilibrium . in paper
i , we set the spin angular velocity parameter of the black hole equal to that of the orbital angular velocity in order to produce a nonspinning black hole in the leading - order " approximation @xcite , while in this paper we compute this parameter by requiring the quasilocal spin angular momentum of the black hole to be zero @xcite . we have also improved the formulation of the gravitational field equations and obtained more accurate results than in paper i. as an indication of the improvements in these calculations , a post - newtonian analysis predicts smaller binary eccentricities for these new bhns models than for those computed in paper i ( @xcite , compare @xcite ) . in @xcite , berti
_ et al . _
fit numerical results for the binding energy and angular momentum of binaries in circular orbit to post - newtonian expressions for binaries that are not necessarily in circular orbit .
deviations between the the two approaches then lead to non - zero eccentricities in the post - newtonian expressions .
these eccentricities are smaller for our new results than for those of paper i. we also remark on another finding of @xcite , namely that for a given neutron star mass @xmath146 and a given value of @xmath102 , the eccentricities in bhns models , though small , are found to be larger than in binary neutron star models @xcite .
this suggests a larger deviation from quasiequilibrium for bhns binaries than binary neutron stars .
but , for bhns binaries with a mass ratio of @xmath114 , these parameters correspond to a larger binary separation than for binary neutron stars with a mass ratio of @xmath103 ( compare eq .
( [ eq : kepler ] ) ) . for similar numerical resources
, this larger binary separation leads to a larger numerical error , which may explain the larger eccentricity found by @xcite , at least in part .
in addition to recomputing several sequences we presented in paper i , we have constructed sequences for a wider range of neutron star masses and binary mass ratios , employing a @xmath0 polytropic neutron - star equation of state throughout .
we computed several constant - mass sequences , for various mass ratios and neutron star compactions , and searched for the appearance of a cusp at the neutron star surface indicating the onset of tidal disruption and turning points on the binding energy and angular momentum curves identifying the isco .
we also included some qualitative fits that allow for a simple prediction of those binary parameters separating these two different outcomes of binary coalescence .
unlike in our earlier findings , we found simultaneous turning points along the binding energy and angular momentum quasiequilibrium curves .
this paper was supported in part by nsf grants phy-0205155 , phy-0345151 , and phy-0650377 as well as nasa grants nng04gk54 g and nnx07ag96 g to university of illinois at urbana - champaign , and nsf grant phy-0456917 to bowdoin college .
we summarize our results in tables [ table : seqm017 ] [ table : seqm013 ] . in these tables
, we tabulate the coordinate orbital separation between the center of the excised surface of the black hole and the maximum of the neutron - star baryon rest mass @xmath147 , orbital angular velocity @xmath2 , spin angular velocity of the black hole @xmath1 , binary binding energy @xmath148 , total angular momentum @xmath61 , decrease in the maximum density parameter @xmath149 from the spherical value @xmath150 , minimum of the mass - shedding indicator @xmath151 , and fractional difference @xmath152 between the adm mass @xmath153 and the komar mass @xmath154 along a sequence . here
recall that ( virial ) equilibrium requires @xmath156 .
rccllccc + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 9.76 & 0.0287 & 0.0273 & -7.86(-3 ) & 0.712 & -7.55(-3 ) & 0.981 & 1.71(-4 ) + 7.81 & 0.0388 & 0.0364 & -9.38(-3 ) & 0.668 & -1.22(-2 ) & 0.958 & 7.78(-5 ) + 6.35 & 0.0511 & 0.0473 & -1.08(-2 ) & 0.637 & -1.91(-2 ) & 0.913 & 2.22(-5 ) + 5.37 & 0.0634 & 0.0578 & -1.19(-2 ) & 0.619 & -2.72(-2 ) & 0.845 & 2.96(-5 ) + 4.88 & 0.0715 & 0.0646 & -1.24(-2 ) & 0.612 & -3.36(-2 ) & 0.782 & 4.06(-5 ) + 4.59 & 0.0773 & 0.0693 & -1.26(-2 ) & 0.609 & -3.94(-2 ) & 0.724 & 3.02(-5 ) + 4.40 & 0.0816 & 0.0728 & -1.278(-2 ) & 0.6073 & -4.46(-2 ) & 0.669 & 4.65(-6 ) + 4.30 & 0.0838 & 0.0746 & -1.285(-2 ) & 0.6066 & -4.79(-2 ) & 0.626 & 2.23(-5 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 7.81 & 0.0387 & 0.0371 & -6.67(-3 ) & 0.496 & -1.49(-2 ) & 0.983 & 4.33(-4 ) + 5.86 & 0.0564 & 0.0532 & -8.12(-3 ) & 0.466 & -2.73(-2 ) & 0.951 & 3.29(-4 ) + 4.88 & 0.0711 & 0.0663 & -8.81(-3 ) & 0.455 & -3.98(-2 ) & 0.909 & 3.46(-4 ) + 4.30 & 0.0833 & 0.0768 & -9.080(-3)&0.4504&-5.61(-2 ) & 0.856 & 3.86(-4 ) + @xmath160 4.17 & 0.0865 & 0.0795 & -9.096(-3)&0.4503&-6.01(-2 ) & 0.839 & 3.97(-4 ) + 4.04 & 0.0899 & 0.0824 & -9.084(-3)&0.4505&-6.44(-2 ) & 0.821 & 4.13(-4 ) + 3.71 & 0.0995 & 0.0904 & -8.88(-3 ) & 0.453 & -7.70(-2 ) & 0.761 & 4.79(-4 ) + 3.58 & 0.1038 & 0.0940 & -8.71(-3 ) & 0.454 & -8.30(-2 ) & 0.729 & 5.12(-4 ) + rccllccc + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 12.36 & 0.0207 & 0.0199 & -6.49(-3 ) & 0.770 & -3.50(-3 ) & 0.986 & 1.09(-4 ) + 8.24 & 0.0361 & 0.0340 & -9.08(-3 ) & 0.679 & -8.48(-3 ) & 0.943 & 3.46(-5 ) + 6.70 & 0.0476 & 0.0442 & -1.05(-2 ) & 0.645 & -1.35(-2 ) & 0.884 & 7.31(-5 ) + 5.67 & 0.0591 & 0.0542 & -1.16(-2 ) & 0.626 & -2.01(-2 ) & 0.792 & 7.37(-5 ) + 5.15 & 0.0668 & 0.0607 & -1.21(-2 ) & 0.618 & -2.65(-2 ) & 0.700 & 9.46(-5 ) + 4.95 & 0.0704 & 0.0637 & -1.23(-2 ) & 0.615 & -3.05(-2 ) & 0.641 & 1.21(-4 ) + 4.84 & 0.0722 & 0.0652 & -1.24(-2 ) & 0.614 & -3.31(-2 ) & 0.597 & 1.52(-4 ) + 4.76 & 0.0738 & 0.0665 & -1.25(-2 ) & 0.613 & -3.57(-2 ) & 0.536 & 2.19(-4 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 9.61 & 0.0292 & 0.0282 & -5.78(-3 ) & 0.526 & -7.66(-3 ) & 0.987 & 2.69(-4 ) + 6.18 & 0.0526 & 0.0498 & -7.95(-3 ) & 0.471 & -2.04(-2 ) & 0.936 & 1.91(-4 ) + 5.15 & 0.0665 & 0.0622 & -8.76(-3 ) & 0.458 & -3.09(-2 ) & 0.880 & 1.29(-4 ) + 4.47 & 0.0795 & 0.0736 & -9.13(-3 ) & 0.4530 & -4.30(-2 ) & 0.805 & 1.43(-4 ) + @xmath160 4.19 & 0.0859 & 0.0791 & -9.171(-3)&0.4526&-4.99(-2 ) & 0.756 & 1.64(-4 ) + 3.99 & 0.0914 & 0.0837 & -9.13(-3 ) & 0.4532&-5.65(-2 ) & 0.707 & 1.80(-4 ) + 3.85 & 0.0953 & 0.0870 & -9.06(-3 ) & 0.454 & -6.22(-2 ) & 0.665 & 1.81(-4 ) + 3.71 & 0.0996 & 0.0905 & -8.94(-3 ) & 0.456 & -6.91(-2 ) & 0.611 & 1.78(-4 ) + rccllccc + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 15.28 & 0.0154 & 0.0145 & -7.30(-3 ) & 1.109 & -5.93(-4 ) & 0.961 & 1.92(-5 ) + 13.10 & 0.0191 & 0.0179 & -8.34(-3 ) & 1.049 & -9.17(-4 ) & 0.934 & 1.20(-5 ) + 10.91 & 0.0246 & 0.0227 & -9.70(-3 ) & 0.986 & -1.86(-3 ) & 0.871 & 1.27(-6 ) + 9.82 & 0.0285 & 0.0261 & -1.06(-2 ) & 0.954 & -3.20(-3 ) & 0.803 & 2.90(-6 ) + 9.17 & 0.0313 & 0.0285 & -1.11(-2 ) & 0.934 & -4.83(-3 ) & 0.731 & 1.19(-5 ) + 8.73 & 0.0335 & 0.0303 & -1.15(-2 ) & 0.922 & -6.66(-3 ) & 0.648 & 2.77(-5 ) + 8.51 & 0.0347 & 0.0313 & -1.17(-2 ) & 0.916 & -8.04(-3 ) & 0.574 & 4.80(-5 ) + 8.40 & 0.0353 & 0.0318 & -1.19(-2 ) & 0.913 & -9.05(-3 ) & 0.514 & 7.56(-5 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 20.37 & 0.0102 & 0.00989&-4.99(-3 ) & 1.104 & -7.31(-4 ) & 0.993 & 3.67(-5 ) + 11.64 & 0.0225 & 0.0213 & -8.21(-3 ) & 0.894 & -2.48(-3 ) & 0.952 & 5.67(-5 ) + 8.73 & 0.0334 & 0.0311 & -1.03(-2 ) & 0.818 & -5.02(-3 ) & 0.870 & 6.10(-5 ) + 8.00 & 0.0376 & 0.0348 & -1.10(-2 ) & 0.799 & -6.56(-3 ) & 0.822 & 5.38(-5 ) + 7.28 & 0.0427 & 0.0392 & -1.17(-2 ) & 0.781 & -9.39(-3 ) & 0.742 & 5.84(-5 ) + 6.84 & 0.0464 & 0.0423 & -1.22(-2 ) & 0.770 & -1.25(-2 ) & 0.661 & 8.16(-5 ) + 6.55 & 0.0491 & 0.0446 & -1.26(-2 ) & 0.763 & -1.58(-2 ) & 0.565 & 1.24(-4 ) + 6.48 & 0.0498 & 0.0452 & -1.27(-2 ) & 0.761 & -1.69(-2 ) & 0.525 & 1.48(-4 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 19.65 & 0.0107 & 0.0105 & -4.29(-3 ) & 0.918 & -1.22(-3 ) & 0.996 & 1.36(-4 ) + 13.10 & 0.0191 & 0.0184 & -6.22(-3 ) & 0.786 & -2.70(-3 ) & 0.982 & 4.32(-5 ) + 9.82 & 0.0284 & 0.0270 & -7.94(-3 ) & 0.715 & -5.11(-3 ) & 0.953 & 4.76(-5 ) + 7.64 & 0.0399 & 0.0375 & -9.61(-3 ) & 0.666 & -9.14(-3 ) & 0.888 & 1.01(-4 ) + 6.55 & 0.0490 & 0.0455 & -1.06(-2 ) & 0.643 & -1.35(-2 ) & 0.808 & 1.09(-4 ) + 6.00 & 0.0549 & 0.0506 & -1.12(-2 ) & 0.633 & -1.76(-2 ) & 0.734 & 1.19(-4 ) + 5.46 & 0.0620 & 0.0567 & -1.18(-2 ) & 0.624 & -2.53(-2 ) & 0.595 & 1.80(-4 ) + 5.35 & 0.0636 & 0.0581 & -1.19(-2 ) & 0.622 & -2.78(-2 ) & 0.537 & 2.17(-4 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 18.19 & 0.0120 & 0.0118 & -3.32(-3 ) & 0.6605 & -2.19(-3 ) & 0.999 & 2.48(-4 ) + 11.64 & 0.0224 & 0.0218 & -4.98(-3 ) & 0.5601 & -4.58(-3 ) & 0.989 & 1.89(-4 ) + 7.28 & 0.0425 & 0.0406 & -7.22(-3 ) & 0.4882 & -1.29(-2 ) & 0.945 & 8.70(-5 ) + 5.46 & 0.0618 & 0.0580 & -8.60(-3 ) & 0.4624 & -2.53(-2 ) & 0.853 & 1.55(-5 ) + 4.59 & 0.0769 & 0.0713 & -9.12(-3 ) & 0.4554 & -3.92(-2 ) & 0.730 & 1.89(-5 ) + 4.37 & 0.0817 & 0.0755 & -9.177(-3 ) & 0.45484 & -4.51(-2 ) & 0.674 & 2.30(-5 ) + @xmath160 4.22 & 0.0851 & 0.0784 & -9.192(-3 ) & 0.45485 & -5.02(-2 ) & 0.625 & 3.25(-5 ) + 4.12 & 0.0879 & 0.0808 & -9.185(-3 ) & 0.45500 & -5.50(-2 ) & 0.570 & 4.01(-5 ) + rccllccc + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 13.95 & 0.0174 & 0.0167 & -5.91(-3 ) & 0.805 & -2.17(-3 ) & 0.978 & 6.26(-6 ) + 10.46 & 0.0260 & 0.0249 & -7.56(-3 ) & 0.729 & -4.18(-3 ) & 0.943 & 7.02(-5 ) + 8.14 & 0.0367 & 0.0346 & -9.18(-3 ) & 0.678 & -7.78(-3 ) & 0.865 & 1.09(-4 ) + 6.98 & 0.0451 & 0.0420 & -1.02(-2 ) & 0.653 & -1.23(-2 ) & 0.766 & 1.18(-4 ) + 6.63 & 0.0482 & 0.0448 & -1.05(-2 ) & 0.646 & -1.48(-2 ) & 0.714 & 1.28(-4 ) + 6.39 & 0.0505 & 0.0468 & -1.07(-2 ) & 0.642 & -1.70(-2 ) & 0.668 & 1.43(-4 ) + 6.16 & 0.0530 & 0.0490 & -1.10(-2 ) & 0.638 & -2.01(-2 ) & 0.604 & 1.70(-4 ) + 5.99 & 0.0551 & 0.0508 & -1.12(-2 ) & 0.635 & -2.32(-2 ) & 0.530 & 2.06(-4 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 10.85 & 0.0247 & 0.0240 & -5.36(-3 ) & 0.547 & -4.95(-3 ) & 0.979 & 6.41(-5 ) + 7.75 & 0.0391 & 0.0374 & -6.96(-3 ) & 0.496 & -1.08(-2 ) & 0.934 & 3.33(-6 ) + 6.20 & 0.0524 & 0.0496 & -8.08(-3 ) & 0.473 & -1.85(-2 ) & 0.859 & 8.52(-5 ) + 5.43 & 0.0622 & 0.0584 & -8.64(-3 ) & 0.463 & -2.61(-2 ) & 0.777 & 1.12(-4 ) + 5.04 & 0.0683 & 0.0639 & -8.89(-3 ) & 0.460 & -3.23(-2 ) & 0.707 & 1.25(-4 ) + 4.73 & 0.0740 & 0.0688 & -9.06(-3 ) & 0.458 & -4.00(-2 ) & 0.619 & 1.47(-4 ) + 4.58 & 0.0771 & 0.0716 & -9.12(-3 ) & 0.4572&-4.54(-2 ) & 0.546 & 1.65(-4 ) + 4.54 & 0.0780 & 0.0723 & -9.14(-3 ) & 0.4570&-4.72(-2 ) & 0.516 & 1.77(-4 ) + rccllccc + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 14.93 & 0.0158 & 0.0154 & -5.58(-3 ) & 0.825 & -1.79(-3 ) & 0.974 & 1.39(-5 ) + 11.20 & 0.0237 & 0.0227 & -7.15(-3 ) & 0.747 & -3.51(-3 ) & 0.933 & 7.52(-5 ) + 8.71 & 0.0334 & 0.0316 & -8.71(-3 ) & 0.691 & -6.94(-3 ) & 0.841 & 1.03(-4 ) + 8.09 & 0.0370 & 0.0348 & -9.19(-3 ) & 0.678 & -8.82(-3 ) & 0.792 & 1.05(-4 ) + 7.47 & 0.0412 & 0.0386 & -9.71(-3 ) & 0.665 & -1.19(-2 ) & 0.717 & 1.16(-4 ) + 6.97 & 0.0451 & 0.0421 & -1.02(-2 ) & 0.655 & -1.63(-2 ) & 0.617 & 1.45(-4 ) + 6.85 & 0.0462 & 0.0431 & -1.03(-2 ) & 0.652 & -1.79(-2 ) & 0.580 & 1.60(-4 ) + 6.72 & 0.0473 & 0.0441 & -1.04(-2 ) & 0.650 & -1.97(-2 ) & 0.531 & 1.79(-4 ) + + @xmath157&@xmath102&@xmath158&@xmath100&@xmath101 & @xmath159&@xmath151&@xmath152 + 11.62 & 0.0225 & 0.0219 & -5.06(-3 ) & 0.560 & -4.13(-3 ) & 0.975 & 6.73(-5 ) + 8.30 & 0.0356 & 0.0342 & -6.66(-3 ) & 0.505 & -9.10(-3 ) & 0.923 & 5.36(-5 ) + 7.47 & 0.0410 & 0.0392 & -7.19(-3 ) & 0.492 & -1.18(-2 ) & 0.890 & 9.24(-5 ) + 6.64 & 0.0479 & 0.0456 & -7.76(-3 ) & 0.480 & -1.61(-2 ) & 0.835 & 1.26(-4 ) + 5.81 & 0.0570 & 0.0538 & -8.36(-3 ) & 0.469 & -2.36(-2 ) & 0.735 & 1.60(-4 ) + 5.40 & 0.0627 & 0.0589 & -8.65(-3 ) & 0.465 & -3.06(-2 ) & 0.645 & 1.90(-4 ) + 5.23 & 0.0652 & 0.0611 & -8.76(-3 ) & 0.463 & -3.45(-2 ) & 0.590 & 2.09(-4 ) + 5.11 & 0.0672 & 0.0629 & -8.84(-3 ) & 0.462 & -3.81(-2 ) & 0.532 & 2.24(-4 ) + m. d. duez , y. t. liu , s. l. shapiro , m. shibata , and b. c. stephens , phys .
lett . * 96 * , 031101 ( 2006 ) ; phys .
d * 73 * , 104015 ( 2006 ) ; m. shibata , m. d. duez , y. t. liu , s. l. shapiro , and b. c. stephens , phys .
lett . * 96 * , 031102 ( 2006 ) .
an explicit proof in general relativity that simultaneous turning points exist along quasiequilibrium sequences has been given for corotating black holes in compact binaries @xcite .
no proof has yet been provided for nonspinning black holes in binaries , but the same result is expected to hold ; see also @xcite . | we construct new models of black hole - neutron star binaries in quasiequilibrium circular orbits by solving einstein s constraint equations in the conformal thin - sandwich decomposition together with the relativistic equations of hydrostationary equilibrium .
we adopt maximal slicing , assume spatial conformal flatness , and impose equilibrium boundary conditions on an excision surface ( i.e. , the apparent horizon ) to model the black hole .
in our previous treatment we adopted a
leading - order " approximation for a parameter related to the black - hole spin in these boundary conditions to construct approximately nonspinning black holes .
here we improve on the models by computing the black hole s quasilocal spin angular momentum and setting it to zero . as before ,
we adopt a polytropic equation of state with adiabatic index @xmath0 and assume the neutron star to be irrotational .
in addition to recomputing several sequences for comparison with our earlier results , we study a wider range of neutron star masses and binary mass ratios . to locate the innermost stable circular orbit we search for turning points along both the binding energy and total angular momentum curves for these sequences . unlike for our previous approximate boundary condition , these two minima now coincide .
we also identify the formation of cusps on the neutron star surface , indicating the onset of tidal disruption .
comparing these two critical binary separations for different mass ratios and neutron star compactions we distinguish those regions that will lead to a tidal disruption of the neutron star from those that will result in the plunge into the black hole of a neutron star more or less intact , albeit distorted by tidal forces . |
as we known from the xviii century , @xmath1-partitions ( as we call them nowadays ) of finite line segments are only possible if the successive adjacent parts of the @xmath1-partition are of a decreasing length .
this inevitable restriction induces a huge asymmetry in the very partition .
in fact , whatever be the length of the @xmath1-partitioned line segment and whatever be the @xmath1-partition , all its parts , except a finite number of them , will necessarily lie within an arbitrarily small final segment .
for the sake of illustration , consider an @xmath1-partition of a @xmath2 light years length segment -the assumed diameter of the universe .
whatever be the @xmath1-partition of this enormous line segment all its infinitely many parts , except a finite number of them , will inevitably lie within a final segment inconceivable less than , for instance , planck length ( @xmath3 cm ) . there is no way of performing a more equitable partition if the partition has to be @xmath1-ordered .
thus , @xmath1-partitions are @xmath1-asymmetrical .
for the same reason it is impossible to consider two proper points in the real line @xmath4 separated by an infinite euclidean distance , in spite of the assumed infiniteness of the real line .
the above simply unaesthetic consequences of @xmath1-partitions become a little more controversial if the partitioned object is a closed line as a jordan curve .
the objective of the following short discussion is just to examine one of those consequences .
the @xmath1-order makes it impossible a more equitable distribution of the available space . ]
let @xmath5 be a real valued function whose graph is a jordan curve @xmath6 in the euclidean plane @xmath7 .
if @xmath8 and @xmath9 are any two @xmath6 s points , we will write @xmath10 to denote the length of the @xmath6 s arc @xmath11 whose endpoints are @xmath8 and @xmath9 .
that is to say : + assume that @xmath6 has an infinite euclidean length .
in these conditions let @xmath13 be any proper real number greater than 0 and assume @xmath6 is partitioned clockwise from a point @xmath14 into a certain number of adjacent parts @xmath15 , @xmath16 , @xmath17 @xmath18 so that each part @xmath19 has a finite length which is equal or greater than @xmath13 : + evidently these partition @xmath21 must be infinite otherwise , and being finite the length of the parts , @xmath6 would have a finite length .
in addition , and according to cantor @xcite , @xmath21 can not be uncountably infinite .
in fact , consider the sequence of real numbers @xmath22 defined as : + the one to one correspondence @xmath24 between @xmath25 and @xmath22 defined by @xmath26 proves that both sequence have the same cardinality .
so if the first one were uncountably infinite so would be the second .
but @xmath22 can not be uncountably infinite because if that were the case we could pick a different rational number @xmath27 in each real interval @xmath28 and then we would have an uncountable set of different rational numbers , which is obviously impossible @xcite .
the partition @xmath29 must therefore be countably infinite .
on the other hand , the adjacency of @xmath15 , @xmath16 , @xmath17 @xmath18 implies successiveness and then ordinality . accordingly @xmath29 must have a certain ordinal number , and taking into account its countable nature
, this ordinal can only be transfinite , i.e. an ordinal of the second class according to cantor s classic terminology @xcite .
now consider a point @xmath30 anticlockwise from @xmath14 and such that @xmath31 . according to ( [ eqn : fjc > = r ] ) @xmath30
can only belong to the last element of @xmath29 .
consequently this sequence must have a last element and then must be @xmath32-ordered , being @xmath33 a transfinite ordinal of the second class second kind , and @xmath34 a finite ordinal ; so that @xmath35 is an ordinal of the second class first kind .
if must therefore exist an @xmath33-th second class second kind ordinal in @xmath29 so that @xmath36 can not have an immediate predecessor ( @xmath37 is the limit of a sequence of first kind ordinals ) .
now then , according again to ( [ eqn : fjc > = r ] ) , the point @xmath38 anticlockwise from @xmath37 and such that @xmath39 can only belong to a part immediately preceding @xmath40 , which is impossible if @xmath37 is of the second kind , as it has to be .
this proves that @xmath29 must be , but can not be , @xmath32-ordered , a contradiction we have derived from our initial assumption on @xmath6 infiniteness .
we must therefore conclude that jordan curves of infinite length are inconsistent objects , and so must be any metrical space compatible with them .
in addition , if @xmath6 is finite the above condition ( [ eqn : fjc > = r ] ) can no longer holds , otherwise @xmath6 would have an infinite length .
therefore , transfinite partitions of jordan curves can only be of a decreasing part - length , i.e. @xmath0-asymmetrical . | the @xmath0-asymmetry induced by transfinite partitions makes it impossible for jordan curves to have an infinite length . |
structural glasses are disordered materials which lack crystalline periodicity and are mechanically solids .
most metals crystallize on cooling , arranging their atoms into a regular pattern called lattice , but if crystallization does not occur and the atoms settle in a random arrangement , the final form is a metallic glass@xcite .
the critical cooling rate is much slower for bmg s , enabling metglass samples of larger dimensions@xcite . a similar kind of arrested kinetics of the first order magnetic transition from ferromagnetic ( fm ) to antiferromagnetic ( afm ) or vice versa has been observed in half doped manganites and in various intermetallic alloys like doped cefe@xmath3 , co doped mn@xmath3sb , ferh and ferromagnetic shape memory alloys in recent years .
these materials have been examined thoroughly by various protocols including the novel protocol chuf conceived and developed to identify the magnetic glasses and its low temperature equilibrium state @xcite .
the commonality between the structural and magnetic glass is the arrested kinetics resulting in the persistence of high temperature state as the metastable state at low temperature . in this manuscript , we highlight on the possibility of a rare - earth bulk metallic glass with composition tb@xmath0sm@xmath1al@xmath2co@xmath1 to be a magnetic glass . we have compared the magnetization behavior of as cast and annealed specimen in order to find the correlation between structural glass and magnetic glass .
the as cast bmg - tb@xmath0sm@xmath1al@xmath2co@xmath1 sample is prepared by arc melting the pure elements in an argon atmosphere and then suck - cast into a cu mold to get a cylindrical rod of 2 mm diameter .
a small piece of this sample was crystallized by annealing at 450@xmath4c .
the amorphous and crystalline nature of the sample is confirmed by x - ray diffraction using bruker ( d8 advance ) x - ray diffractometer with cu k@xmath5 radiations .
magnetization measurements were carried out using a commercial 14t - vsm ppms .
the reported glass transition temperature ( t@xmath6 ) , crystallization temperature ( t@xmath7 ) , and the melting temperature ( t@xmath8 ) of the bmg tb@xmath0sm@xmath1al@xmath2co@xmath1 is 309@xmath4c , 383@xmath4c and 659@xmath4c respectively @xcite .
bmgs are partially crystallized by isothermal annealing at a temperature in the supercooled liquid region or by fast annealing just above crystallization temperature @xcite .
this is known as partial devitrification of bmgs which gives rise to the formation of composites or multi - phases .
the magnetic behavior is complex especially in re based bmgs when multiphases are present after tailoring them by different heat treatments . in order to understand whether the inhomogeneous structure arising from multi - phases can hinder the critical dynamics ,
we have performed the zero field cooled ( zfc ) and field cooled magnetization ( fc ) measurements at various fields on as cast as well as annealed sample .
sm@xmath1al@xmath2co@xmath1 . ] figure 1(a ) shows zfc and fc behavior of the m(t ) measurements at 0.5 t , 1 t and 2 t on as cast sample .
the bifurcation between the zfc and fc curves is an indication of non - ergodicity .
this temperature where the bifurcation between two curves begin , decreases with increase in magnetic field suggesting the irreversibility process begins at lower temperature as field rises .
the difference in zfc and fc magnetization at low temperature is defined as the thermomagnetic irreversibility ( @xmath9 m = @xmath10 ) which increases upto 1 t and then decreases thereafter at 2 t for as cast sample as shown in the inset of figure1(a ) .
similar type of non - ergodic behavior and @xmath9 m has been observed in spin glasses .
figure 1(b ) shows the zfc and fc magnetization behavior of the crystalline counter part at 0.5 , 1 and 2 t which shows an antiferromagnetic transition ( t@xmath11 ) at @xmath12 30k and it decreases to 29k at 2 t .
this fm - afm transition is not seen in the as cast sample .
inset of figure 1(b ) shows the thermo - magnetic irreversibility behavior which increases with field in contrary to the as cast sample .
@xmath9 m rising with field has been observed in magnetic glasses mentioned above . to validate the kinetically arrested state in the crystalline counter part we have performed the well known protocol chuf on as cast as well as on the partially crystallized sample as shown in figure 2 .
figure 2 ( a and b ) shows the temperature dependent magnetization at 1 t and 2 t after cooling in various fields on the as cast sample .
magnetization measurements at 1 t after cooling in 0 , 0.5,1 , 2 , 3 , and 10 t demonstrates the non - ergodic behavior of as cast sample below 14k , which reduces to 13k on increasing the warming field to 2 t . on the other hand m(t )
behavior on annealed sample shows two transitions when cooling field ( h@xmath11 ) is higher than the warming field ( h@xmath13 ) while only one transition when h@xmath14 as shown in figure 2(c and d ) .
the sudden fall in magnetization at low temperatures for h@xmath15 signifies the devitrification of the magnetic glassy state and we note that for h@xmath15 we have non - ergodic behavior below @xmath16 15k where as for h@xmath17 , it is observed at a higher temperature at @xmath16 20k . [ ! h ] sm@xmath1al@xmath2co@xmath1 . ] hence , our measurements show that the as cast amorphous bmg sample does not become a magnetic glass whereas the annealed ( partially crystallized ) sample resembles a magnetic glass
a. l. greer , science * 267 * , 1947 ( 1995 ) . | * abstract*. a comparative study on as cast and annealed rare earth bulk metallic glass ( bmg ) with composition tb@xmath0sm@xmath1al@xmath2co@xmath1 has been carried out by magnetization measurements .
the as cast amorphous sample shows non - ergodic magnetization but does not show the behavior expected from magnetic glass .
after annealing , the partially crystallized bmg shows this magnetic glass behavior .
this is confirmed by the established measurement protocol of cooling and heating in unequal fields ( chuf ) . |
m - dwarfs are under intense scrutiny as potential planet hosts .
m - dwarfs have sufficiently low masses ( @xmath4 ) and small radii ( @xmath5 ) that exoplanets induce considerably larger reflex velocities and transit depths than an identical planet would around larger , more massive hosts .
accordingly , most of the lowest - mass planets detected to date orbit low - mass stars .
m - dwarfs relatively low luminosities also ensure that habitable planets will be in small , fast orbits , producing higher reflex radial velocity amplitudes , and a greater likelihood , depth , and frequency of transit events @xcite .
m - dwarfs constitute more than two thirds of the nearby stellar population @xcite ; identifying the most likely planet hosts in advance would greatly enhance the efficiency of m - dwarf exoplanet surveys .
observations of fgk - dwarfs have convincingly shown that gas giant planet frequency rises steeply with host star metallicity , a result known as the planet - metallicity correlation .
determining if this correlation extends into the m - dwarf regime requires accurate metallicity estimates for ultra - cool stars . as m - dwarf spectra are dominated by chemically complex molecular features ,
_ a priori _ spectral synthesis does not well match the observed spectra .
most attempts to calibrate m - dwarf metallicity indicators have therefore focused on observational studies of wide binaries , where the robustly estimated [ fe / h ] of the fgk - primary is assumed to describe the m - dwarf secondary as well . this assumption will hold if both binary components formed from the same well - mixed molecular cloud , and no mass transfer or dredge - up has occurred in the system . adopted this technique to develop a photometric metallicity calibration that suggested that nearby m - dwarfs , including the planet hosts gj 876 and gj 436 , were slightly metal poor .
@xcite also estimated subsolar metallicities for gj 436 , gj 581 and gj 876 from spectral synthesis of optical spectra .
though there were only a small number of m - dwarf planet hosts in their samples , the results of the b05 and @xcite studies were unexpected , since they imply that planet formation around m - dwarfs does not depend on the star s metal content .
this finding is difficult to reconcile with the planet - metallicity correlation for fgk - dwarfs , and runs counter to the predictions of the commonly accepted core accretion model @xcite .
* ja09 hereafter ) derived a new photometric metallicity calibration using 6 m - dwarfs with wide , metal - rich fgk companions not present in the b05 study .
ja09 found that the previous photometric calibration systematically underestimated the metallicities of these metal - rich stars .
ja09 attributed this discrepancy to the lack of metal - rich calibrators in the b05 sample and/or the use of visual magnitudes with poorly characterized uncertainties .
ja09 also estimated the metallicities of 7 m - dwarfs with planetary - mass companions , concluding that m - dwarf planet hosts are preferentially metal - rich , just like fgk planet hosts .
ja09 suggested that the lack of jovian planets around m - dwarfs is most likely due to the influence of stellar mass , not stellar metallicity , on the planet formation process , consistent with the work of @xcite on giant planet formation around red dwarfs .
* sg10 hereafter ) recently revised the b05 and ja09 photometric calibrations , finding that they systematically underestimate or overestimate metallicity at the extremes of their range .
their results hint that low - mass planets may be more likely to be found around metal - rich m - dwarfs , which disagrees with the almost equal probability of finding neptune - mass planets around a metal - rich or a metal - poor fgk - star .
existing methods for estimating m - dwarf metallicities rely heavily on observations at visual wavelengths , where metallic absorption lines and molecules strongly influence a star s spectrum .
mid- to late-
m - dwarfs are typically too faint for high - quality observations at visual wavelengths , which has limited most metallicity analyses to early - m - dwarfs and a few nearby stars ( e.g. * ? ? ?
lacking a satisfactory way to estimate metallicities across the entire m - dwarf sequence has limited our understanding of planet formation around the lowest mass stars .
in this letter , we present a technique for estimating the metallicity of field m - dwarfs from the strength of prominent k - band absorption features .
we calibrated our technique using spectra of m - dwarfs with existing [ fe / h ] estimates , adopted as a proxy for their overall metallicities , obtained from the analysis of a wide fgk - companion .
our technique does not depend on accurate parallaxes and v magnitudes , or high - resolution , high - signal - to - noise spectra , as required by previous techniques .
instead , our technique requires simply a k - band spectrum at moderate resolution , which can be efficiently obtained for many m - dwarfs with current near - infrared spectrographs .
our method will enable metallicity estimates to be generated for stars which are too cool or too distant for other methods , identify the most advantageous targets for planet - searches around cool stars , and thus improve our understanding of planet formation around low - mass stars . in 2 , we describe our observations and data reduction .
we present our analysis and results in 3 , and summarize and discuss our conclusions in 4 .
near - infrared spectra of the stars in this work were obtained with the triplespec spectrograph on the palomar hale telescope @xcite during 7 observing runs between 2007 and 2009 .
triplespec at palomar has no moving parts and simultaneously acquires 5 cross - dispersed orders covering 1.0 - 2.4 @xmath1 m at a resolution of @xmath6@xmath72700 .
the spectra were reduced with an idl - based data reduction pipeline developed by p. muirheadmuirhead/#downloads ] .
the data were sky - subtracted using a sky - frame made by median combining 5 exposures with the science target placed at different positions along the slit .
each sky - subtracted exposure was then divided by a normalized flat - field , wavelength calibrated and optimally extracted @xcite .
the spectra were flux - calibrated and telluric corrected using observations of an a0v star at a similar airmass with the idl - based code @xmath8_@xmath9 by @xcite .
our sample consists of : * seventeen m - dwarfs with wide ( @xmath10 5@xmath11 separation ) , common - proper - motion solar - type companions to serve as metallicity calibrators .
the fgk - primaries have spectroscopic metallicity measurements by ( * ? ? ?
* spocs catalogue ) .
the binary systems were selected from the @xcite catalogue of nearby stars , the @xcite catalogue of nearby wide binary and multiple systems , and the list of new hipparcos binaries by @xcite . * twenty - nine nearby m - dwarfs analyzed by ja09 , to compare our spectroscopic results with their photometric [ fe / h ] . *
eight m - dwarf planet hosts observable from palomar mountain : gj 876 , gj 436 , gj 581 , gj 849 @xcite , gj 1214 @xcite , gj 649 @xcite , hip 57050 @xcite , and hip 79431 @xcite .
* twenty - nine m - dwarf spectroscopic standards defined by ( * ? ? ?
* khm system ) or with khm spectral types from @xcite figure [ fig1 ] shows the @xmath12v - k@xmath13 , m@xmath14@xmath15 plane for our sample .
the ja09 calibration is only valid for stars with : 1 ) accurate johnson v magnitudes and parallaxes from the hipparcos catalogue @xcite , the yale parallax catalogue @xcite or the tass mark iv survey catalogue @xcite , and 2 ) v - k@xmath13 colors between [ 3.9,6.6 ] .
ja09 assume that the mean m@xmath14 versus v - k@xmath13 relation in the solar neighborhood is equivalent to an [ fe / h]=-0.05 metallicity contour , with stars brighter than this being metal - rich , and fainter stars being metal - poor .
most of the mid - type and all of the late - type m - dwarfs fail both of these conditions : they lack accurate parallaxes and/or johnson v magnitudes , so were not included in the ja09 calibration .
we have empirically calibrated several nir spectral features to serve as effective m - dwarf metallicity indicators .
figure [ fig2 ] shows the k - band spectra of three m4-m4.5 dwarfs in our sample , along with updated phoenix models @xcite with effective temperatures equivalent to an @xmath0m3.5 - 4 star and varying metallicities .
the overall shape of the k - band spectrum is well represented by the updated phoenix models , and the na i doublet ( 2.206 @xmath1 m & 2.209 @xmath1 m ) and the ca i triplet ( 2.261 @xmath1 m , 2.263 @xmath1 m & 2.265 @xmath1 m ) are seen to strengthen considerably with metallicity .
there is a discrepancy , however , between the strengths of these features in our spectra and in the models of comparable metallicity .
we measured the equivalent width ( ew ) of the na i and the ca i features to differentiate between metal - rich and metal - poor m - dwarfs .
each spectrum was normalized and the ews were calculated by computing the ratio of the area of a feature to a pseudocontinuum across that feature , computed from featureless regions on either side of it , using the the idl - based function ` measure_ew ` by n. konidaris and j. harker .
uncertainties in the ews were obtained from the errors in the estimated pseudocontinuum and the errors in the measured intensities of each feature , as described in @xcite .
figure [ fig3]a compares the spocs [ fe / h ] for our m - dwarf metallicity calibrators to the ews of their ca i triplets ; figure [ fig3]b compares the ews of the na i and the ca i features for our m - dwarf sample .
the correspondence between the na i and ca i features and the metallicity of m - dwarfs is evident : metal - poor m - dwarfs have lower ews than metal - rich m - dwarfs .
the strength of these features also depend on temperature and surface gravity , however , as do most atomic lines in the nir spectra of m - dwarfs @xcite . to explicitly account for the influence of stellar temperature on the strengths of na and ca lines , we include an independent , temperature sensitive spectral index in our analysis .
k - band h@xmath16o absorption bands been used extensively to diagnose m - dwarf spectral types ( e.g. , * ? ? ?
* ) ; we diagnose the effective temperatures of stars in our sample using the h@xmath16o - k index defined by @xcite : @xmath17 where @xmath18 denotes the mean flux level in the wavelength range defined by @xmath19 and @xmath20 , in microns .
the linear relation between h@xmath16o - k index and the khm spectral types of spectral standards in our sample is : @xmath21 we report the spectral types estimated from equation ( 1 ) for the metallicity calibration m - dwarfs and planet hosts in table 1 , rounded to the nearest half subtype .
a star s surface gravity also influences its spectrum , but those are typically subtle effects .
all of our targets are dwarf stars , which reduces any potential gravity ambiguity in our analysis .
@xcite reported empirical values of log [email protected] for five m - dwarfs in eclipsing binary systems , while values of log [email protected] - 4.9 have been inferred from mass and radius values for m - dwarfs in . to identify the best fit relationship between [ fe / h ] , the ews of na i and ca i features in @xmath22 , and the h@xmath16o - k index , we performed a linear regression on the parameters measured for our m - dwarfs with spocs metallicities .
we found a best - fit linear equation of : @xmath23 } & = & ~0.142(\mathrm{na~{\footnotesize i}_{ew}})~+~0.011(\mathrm{ca~{\footnotesize i}_{ew } } ) \nonumber \\ & & ~+~2.541(\mathrm{h_{2}o\!\!-\!\!k})~-~3.132 \\
\sigma\mathrm{([fe / h ] ) } & = & 0.15 \nonumber \end{aligned}\ ] ] with a residual mean square @xmath24([fe / h])=0.02 and an adjusted squared multiple correlation coefficient @xmath25@xmath26([fe / h])=0.63 , which implies , by comparison of the same criteria functions calculated in sg10 , that our model provides a better fit than the models of b05 ( @xmath24=0.05 , @xmath25@[email protected] ) , ja09 ( @xmath24=0.04 , @xmath25@[email protected] ) and sg10 ( @xmath24=0.02 , @xmath25@[email protected] ) .
we estimate systematic errors , such as the zero - point value and the inclusion of a non - physically bound binary as a metallicity calibrator , to be of the order of @xmath00.06 ( spocs data ) and @xmath00.03 , respectively .
the combined systematic error is of the same order of the structure in the [ fe / h ] residuals of the calibration sample .
we note that the predicted [ fe / h ] for some of the planet hosts lie above the [ fe / h ] boundaries of the calibrators .
the values of the ews of na i and ca i , the h@xmath16o - k index and [ fe / h ] values for the metallicity calibration m - dwarfs and planet hosts , are given in table 1 .
the best fit linear combination of the ews of the ca i and the na i features are plotted in figure [ fig4]a versus the h@xmath16o - k index of our m - dwarf sample .
there is a clear distinction between the m - dwarfs with metal - rich and metal - poor fgk - dwarf companions in figure [ fig4]a ; the spocs [ fe / h ] values of our m - dwarf metallicity calibrators are also plotted in figure [ fig4]b against the value of their h@xmath16o - k index .
the m - dwarfs with photometric metallicity estimates calculated by ja09 are also bisected between metal - rich and metal - poor using these k - band features .
moreover , we found that all of the m - dwarf planet hosts have metallicities much higher than the mean metallicity of m - dwarfs in the solar neighborhood ( [ fe / h]@xmath28@xmath0 - 0.17 ) , defined by sg10 .
the jovian planet hosts are co - located with the m - dwarf companions of metal - rich fgk stars in the upper portion of fig .
[ fig4]a , while the neptune planet hosts are located just above the m - dwarf companions of metal - poor fgk stars .
we have found that the ews of the ca i triplet , the na i doublet , and water absorption in the k - band differentiate metal - rich and metal - poor m - dwarfs , as shown in figure [ fig4 ] .
we estimate [ fe / h ] values higher than -0.05 dex for eight m - dwarf planet hosts , with the jovian planet hosts being more metal - rich than their neptune host analogs ( table 1 ) .
our results , along with the conclusions of ja09 and sg10 that m - dwarf planet hosts appear to be systematically metal - rich , are consistent with the metallicity distribution of fgk - dwarfs with planets .
we also estimated spectral types for the m - dwarf planet hosts and metallicity calibrators in our sample , using the h@xmath16o - k index defined by @xcite .
our spectral type estimates agree well with the khm spectral types @xcite , modulo the @xmath01 subtype errors ( table 1 ) .
we do find discrepancies , however , with the spectral types of some of the most metal - rich and metal - poor stars in our sample .
the metal - rich planet hosts gj 849 and gj 876 have later spectral types in the khm system than are indicated by the h@xmath16o - k index .
a possible explanation is that the khm system relies strongly on the double - metal tio and vo molecular bands in the 6300 - 9000 @xmath22 wavelength range .
these molecular bands are known to be metallicity dependent ( e.g. the subdwarf classification by * ? ? ?
* ) : they grow stronger with increasing metallicity as well as decreasing temperature . while the khm system includes the overall shape of the stellar spectrum , this does not change considerably for early m - dwarfs , such that metal - rich early - type m - dwarfs with strong molecular features could be easily mistaken for less - metal rich later - type m - dwarfs .
this reasoning can also be applied to gj 611 b , the most metal - poor star in our sample : we derive an m6 spectral type from its h@xmath16o - k index , while its khm spectral type is m4 , consistent with its weak tio and vo bands being interpreted as a temperature effect rather than a sign of its low metallicity .
measuring m - dwarf metallicities will also improve our ability to determine key physical parameters for both stars and planets .
the masses and radii of m - dwarf planet hosts are typically derived from theoretical models , and/or empirical relationships between these parameters and the star s temperature and luminosity , all of which have a strong dependence on metallicity . since minimum planetary masses inferred from rv measurements are tied to the mass of the host star , and planetary radii derived from transit observations depend directly on the adopted stellar radius , errors in stellar parameters lead directly to errors in inferred planetary properties .
report a stellar mass of 0.3 m@xmath29 for gj 876 , but with an uncertainty of @xmath030@xmath30 due primarily to the uncertainty in the host s metallicity , which affected the initial estimate of the masses of its planetary companions .
recent measured m - dwarf radii and masses have typical errors less than 10@xmath30 @xcite , but the errors may increase somewhat for stars with significantly non - solar metallicities . with poor estimates of the mass and radius of exoplanets , it is harder to constrain their true composition , atmospheres and interiors . despite the fact that k - band alkali features are temperature / gravity dependent @xcite , and prone to non - local thermodynamic equilibrium effects , our empirical analysis demonstrates that the dispersion in ew of the ca i and na i features can diagnose the metal abundance of m - dwarfs with roughly the same temperature ( h@xmath16o - k index ) and gravitym co band - head ) to distinguish between dwarfs and giants prior to interpreting na i and ca i feature strengths as metallicity indicators . ] .
our method has an advantage with respect to previous photometric and optical spectroscopic calibrations , since it does not depend on parallaxes or v magnitudes , allowing us to cover a larger sample of cooler and distant m - dwarfs .
our results enable the identification of the most advantageous m - dwarfs for planetary search surveys around cool stars , such as the triplespec exoplanet discovery instrument ( * ? ? ?
* tedi ) , mearth @xcite , the precision radial velocity spectrograph ( * ? ? ?
* pvrs ) , and the crires survey by @xcite , and will thus help unravel the formation mechanisms of planets around m - dwarfs .
we thank the staff and telescope operators of palomar observatory for their support .
we also thank kevin apps , john johnson , john bochansky , andrew west and the anonymous referee for their useful comments .
this publication makes use of the simbad database , operated at cds , strasbourg , france and the data products from the 2mass , which is a joint project of the university of massachusetts and the ipac / caltech funded by the nasa and the nsf .
this work was supported by conicyt and the aura , inc . in representation of the nsf under cooperative agreement nsf ast-9613615 .
support for this work was provided by nasa through hubble fellowship grant hst - hf-51253.01-a awarded by the stsci , which is operated by the aura , inc .
, for nasa , under contract nas 5 - 26555 .
this work was supported by nasa headquarters under the nessf program - grant nnx07ap56h .
this material is based in part on work supported by the nsf under grant ast-0905932 .
ali , b. , carr , j. s. , depoy , d. l. , frogel , j. a. , & sellgren , k. 1995 , , 110 , 2415 apps , k. , et al .
2010 , , 122 , 156 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 1998 , , 337 , 403 bean , j. l. , benedict , g. f. , & endl , m. 2006 , , 653 , l65 bean , j. l. , seifahrt , a. , hartman , h. , nilsson , h. , wiedemann , g. , reiners , a. , dreizler , s. , & henry , t. j. 2010 , , 713 , 410 bonfils , x. , delfosse , x. , udry , s. , santos , n. c. , forveille , t. , & sgransan , d. 2005 , , 442 , 635 bonfils , x. , et al .
2005 , , 443 , l15 butler , r. p. , vogt , s. s. , marcy , g. w. , fischer , d. a. , wright , j. t. , henry , g. w. , laughlin , g. , & lissauer , j. j. 2004 , , 617 , 580 butler , r. p. , johnson , j. a. , marcy , g. w. , wright , j. t. , vogt , s. s. , & fischer , d. a. 2006 , , 118 , 1685 charbonneau , d. , et al .
2009 , , 462 , 891 chabrier , g. 2003 , , 115 , 763 covey , k. r. , lada , c. j. , roman - zuniga , c. , muench , a. a. , forbrich , j. , & ascenso , j. 2010 , arxiv:1007.2192 delfosse , x. , forveille , t. , beuzit , j .- l . ,
udry , s. , mayor , m. , & perrier , c. 1999 , , 344 , 897 delfosse , x. , forveille , t. , mayor , m. , perrier , c. , naef , d. , & queloz , d. 1998 , , 338 , l67 demory , b .- o .
, et al . 2009 , , 505 , 205 droege , t. , f. , richmond , m. , w. , sallman , & m. 2007 , vizier online data catalog , 2271 , 0 edelstein , j. , et al . 2007 , , 6693 , fernandez , j. m. , et al
2009 , , 701 , 764 fischer , d. a. , & valenti , j. 2005 , , 622 , 1102 gizis , j. e. 1997 , , 113 , 806 gliese , w. , & jahreiss , h. 1991 , nasa sti / recon technical report a , 92 , 33932 gonzalez , g. 1997 , , 285 , 403 gould , a. , & chanam , j. 2004 , , 150 , 455 haghighipour , n. , vogt , s. s. , butler , r. p. , rivera , e. j. , laughlin , g. , meschiari , s. , & henry , g. w. 2010 , , 715 , 271 hauschildt , p. h. , allard , f. , & baron , e. 1999 , , 512 , 377 henry , t. j. , kirkpatrick , j. d. , & simons , d. a. 1994 , , 108 , 1437 herter , t. l. , et al .
2008 , , 7014 , horne , k. 1986 , , 98 , 609 irwin , j. et al .
2009 , iau symposium , 253 , 37 ivanov , v. d. , rieke , m. j. , engelbracht , c. w. , alonso - herrero , a. , rieke , g. h. , & luhman , k. l. 2004 , , 151 , 387 johnson , j. a. , & apps , k. 2009 , , 699 , 933 johnson , j. a. , et al .
2010 , , 122 , 149 jones , h. r. a. , et al .
2008 , , 7014 , kirkpatrick , j. d. , henry , t. j. , & mccarthy , d. w. , jr .
1991 , , 77 , 417 laughlin , g. , bodenheimer , p. , & adams , f. c. 2004 , , 612 , l73 marcy , g. w. , butler , r. p. , vogt , s. s. , fischer , d. , & lissauer , j. j. 1998 , , 505 , l147 mclean , i. s. , mcgovern , m. r. , burgasser , a. j. , kirkpatrick , j. d. , prato , l. , & kim , s. s. 2003 , , 596 , 561 nutzman , p. , & charbonneau , d. 2008 , , 120 , 317 perri , f. , & cameron , a. g. w. 1974 , icarus , 22 , 416 poveda , a. , herrera , m. a. , allen , c. , cordero , g. , & lavalley , c. 1994 , revista mexicana de astronomia y astrofisica , 28 , 43 santos , n. c. , israelian , g. , & mayor , m. 2004 , , 415 , 1153 santos , n. c. , et al .
2004 , , 426 , l19 schlaufman , k. c. , & laughlin , g. 2010 , arxiv:1006.2850 sembach , k. r. , & savage , b. d. 1992 , , 83 , 147 sousa , s. g. , et al .
2008 , , 487 , 373 vacca , w. d. , cushing , m. c. , & rayner , j. t. 2003 , , 115 , 389 valenti , j. a. , & fischer , d. a. 2005 , , 159 , 141 van altena , w. f. , lee , j. t. , & hoffleit , e. d. 2001 , vizier online data catalog , 1238 , 0 van leeuwen , f. 2008 , vizier online data catalog , 1311 , 0 woolf , v. m. , & wallerstein , g. 2005 , , 356 , 963 lcccccccccccc hip 79431 & 9.699 & 5.470 & 0.904 & m3.5 & + 0.60 & m3.0 & & + 0.16 & + 0.52 & + 0.35 & 2.1 & single jupiter + gj 849 & 8.043 & 5.635 & 0.955 & m1.5 & + 0.49 & m3.5 & & + 0.14 & + 0.58 & + 0.41 & 0.82 & single jupiter + gj 876 & 8.126 & 4.721 & 0.930 & m2.5 & + 0.43 & m4.0 & & + 0.03 & + 0.37 & + 0.23 & 2.64 & 2 jupiters +
neptune + super earth + gj 1214 & 8.520 & 4.095 & 0.895 & m4.0 & + 0.39 & m4.5 & & & + 0.03 & + 0.28 & 0.0179 & single super earth + gj 649 & 5.651 & 4.722 & 0.952 & m1.5 & + 0.14 & m1.0 & & -0.18 & + 0.04 & -0.03 & 0.328 & single neptune + hip 57050 & 6.628 & 4.410 & 0.890 & m4.5 & + 0.12 & m4.0 & & -0.02 & + 0.32 & & 0.298 & single neptune + gj 436 & 5.328 & 4.456 & 0.915 & m3.0 & -0.00 & m2.5 & & -0.03 & + 0.25 & + 0.10 & 0.072 & single neptune + gj 581 & 5.108 & 4.202 & 0.921 & m3.0 & -0.02 & m3.0 & & -0.25 & -0.10 & -0.22 & 0.0492 & neptune + 3 super earths + + gj 212 & 5.067 & 5.286 & 0.963 & m1.0 & + 0.09 & m0.5 & + 0.19 & -0.05 & & + 0.09 & & + hd 46375 b & 6.830 & 5.648 & 0.934 & m2.0 & + 0.27 & m1.0 & + 0.24 & -0.33 & + 0.22 & + 0.12 & & + gj 797 b & 4.654 & 3.604 & 0.933 & m2.5 & -0.06 & m2.5 & -0.09 & -0.09 & & + 0.11 & & + gj 872 b & 4.621 & 3.431 & 0.930 & m2.5 & -0.08 & m1.0 & -0.22 & + 0.12 & & & & + gj 250 b & 4.893 & 4.387 & 0.925 & m2.5 & -0.04 & m2.0 & + 0.14 & + 0.12 & & -0.07 & & + gj 768.1 b & 5.677 & 4.190 & 0.921 & m3.0 & + 0.06 & m3.5 & + 0.16 & + 0.16 & & + 0.22 & & + nltt 14186 & 5.937 & 5.564 & 0.921 & m3.0 & + 0.11 & m & + 0.05 & & & & & + hd 222582 b & 6.637 & 3.919 & 0.913 & m3.0 & + 0.17 & m3.5 & -0.03 & -0.04 & & & & + gj 777 b & 6.151 & 3.177 & 0.892 & m4.0 & + 0.04 & m4.5 & + 0.21 & & + 0.19 & & & + gj 231.1 b(c ) & 5.439 & 4.179 & 0.892 & m4.0 & -0.05 & m3.5 & -0.04 & + 0.16 & & + 0.12 & & + gj 324 b & 7.867 & 4.489 & 0.892 & m4.0 & + 0.30 & m4.0 & + 0.31 & & + 0.33 & + 0.26 & & + gj 783.2 b & 4.716 & 3.089 & 0.883 & m4.5 & -0.19 & m4.0 & -0.15 & -0.35 & & -0.19 & & + gj 3348 b & 5.534 & 4.154 & 0.882 & m4.5 & -0.06 & m4.0 & -0.22 & + 0.16 & & & & + gj 544 b & 5.342 & 3.605 & 0.875 & m5.0 & -0.11 & m6.0 & -0.18 & & & & & + gj 611 b & 3.105 & 2.639 & 0.856 & m6.0 & -0.49 & m4.0 & -0.69 & & & & & + nltt 15867 & 4.097 & 2.905 & 0.851 & m6.0 & -0.36 & & -0.10 & & & & & + gj 376 b & 8.418 & 3.310 & 0.821 & m7.5 & + 0.18 & m5.0 & + 0.20 & & & & & + | a metal - rich environment facilitates planet formation , making metal - rich stars the most favorable targets for surveys seeking to detect new exoplanets .
using this advantage to identify likely low - mass planet hosts , however , has been difficult : until now , methods to determine m - dwarf metallicities required observationally expensive data ( such as parallaxes and high - resolution spectra ) , and were limited to a few bright cool stars .
we have obtained moderate ( r@xmath02700 ) resolution k - band spectra of 17 m - dwarfs with metallicity estimates derived from their fgk companions .
analysis of these spectra , and inspection of theoretical synthetic spectra , reveal that an m - dwarf s metallicity can be inferred from the strength of its na i doublet ( 2.206 @xmath1 m & 2.209 @xmath1 m ) and ca i triplet ( 2.261 @xmath1 m , 2.263 @xmath1 m & 2.265 @xmath1 m ) absorption lines .
we use these features , and a temperature - sensitive water index , to construct an empirical metallicity indicator applicable for m - dwarfs with near - solar metallicities ( -0.5@xmath2[fe / h]@xmath2 + 0.5 ) .
this indicator has an accuracy of @xmath30.15 dex , comparable to that of existing techniques for estimating m - dwarf metallicities , but is more observationally accessible , requiring only a moderate resolution k - band spectrum . applying this method to 8 known m - dwarf planet hosts
, we estimate metallicities ( [ fe / h ] ) in excess of the mean metallicity of m - dwarfs in the solar neighborhood , consistent with the metallicity distribution of fgk planet hosts . |
the authors thank a. pilaftsis and c.e.m .
wagner for useful correspondence .
this work was supported by the ` fonds zur frderung der wissenschaftlichen forschung ' of austria , project no .
p13139-phy , and by the european community s human potential programme under contract hprn - ct-2000 - 00149 .
is supported by ` fonds zur frderung der wissenschaftlichen forschung ' of austria , erwin schrdinger fellowship nr .
j2095 , and partly by the ` schweizer nationalfonds ' .
m. m. nojiri , phys .
d 51 ( 1995 ) 6281 ; m. m. nojiri , k. fujii and t. tsukamoto , phys . rev .
d 54 ( 1996 ) 6756 ; a. bartl , h. eberl , s. kraml , w. majerotto , w. porod and a. sopczak , z. phys .
c 76 ( 1997 ) 549 ; a. bartl , h. eberl , s. kraml , w. majerotto and w. porod , eur .
j. direct c 6 ( 2000 ) 1 .
a. pilaftsis , phys .
b 435 ( 1998 ) 88 ; a. pilaftsis and c.e.m . wagner , nucl .
b553 ( 1999 ) 3 ; s.y .
choi , m. drees and j.s .
lee , phys .
b 481 ( 2000 ) 57 ; g.l . kane and l.t .
wang , phys .
b 488 ( 2000 ) 383 .
cohen , d.b .
kaplan and a.e .
nelson , phys .
b 388 ( 1996 ) 588 ; a.g .
akeroyd , y .- y .
keum and s. recksiegel , phys .
b 507 ( 2001 ) 252 and references therein ; ref.@xcite and references therein . m. knecht and a. nyffeler , hep - ph/0111058 ; m. knecht , a. nyffeler , m. perrottet and e. de rafael , hep - ph/0111059 ; m. hayakawa and t. kinoshita , hep - ph/0112102 . : contours of the @xmath22 decay branching ratios @xmath143 ( a , b , c ) and @xmath144 ( d , e , f ) in the @xmath145 plane for @xmath146 , @xmath147gev , @xmath119gev , @xmath148gev , @xmath149 , and ( @xmath103 , @xmath13 ) = ( 0,0 ) ( a , d ) , ( @xmath150,0 ) ( b , e ) , and ( 0,@xmath150 ) ( c , f ) in the case @xmath127 .
the blank areas are excluded by the conditions ( i ) to ( iii ) given in the text and the inequality @xmath192 .
the excluded region is different in each plot mainly due to the fact that the inequality depends on the phases ( @xmath103 , @xmath13 ) . :
contours of the @xmath22 decay branching ratios @xmath143 ( a ) , @xmath153 ( b ) , and @xmath144 ( c ) in the @xmath154 plane for @xmath146 , @xmath147gev , @xmath119gev , @xmath155gev , @xmath156gev , @xmath157gev , and @xmath149 in the case @xmath127 . : @xmath13 dependence of the @xmath22 decay branching ratios for @xmath158 ( with @xmath149 ) ( a ) and @xmath159 ( with @xmath146 ) ( b ) with @xmath160gev , @xmath119gev , @xmath161gev , @xmath156gev , @xmath162gev , and @xmath163 in the case @xmath127 . in fig.a
the solid and dashed lines are for @xmath47 = 3 and 30 , respectively . in fig.b the solid and dashed lines are for @xmath106
= 0 and @xmath150 , respectively . :
@xmath103 dependence of the @xmath9 decay branching ratios for @xmath166 ( solid lines ) and @xmath150 ( dashed lines ) in two cases : ( a ) small @xmath19-mass - splitting case with @xmath193gev , @xmath194gev , @xmath195 , @xmath196gev , @xmath156gev , @xmath175gev , @xmath149 , and @xmath126 ( for which @xmath197 ) , and ( b , c ) large @xmath19-mass - splitting case with @xmath119gev , @xmath161gev , @xmath198 , @xmath199gev , @xmath200gev , @xmath201gev , @xmath149 , and @xmath127 ( for which @xmath202 ) . in the latter case ( figs.b and c ) we have @xmath203(gev ) and @xmath204(gev ) for @xmath205 , respectively .
the branching ratios @xmath206 and @xmath207 , which are rather insensitive to @xmath103 , are not shown in figs.b and c. : @xmath13 dependence of the @xmath2 decay branching ratios for @xmath149 ( solid lines ) and @xmath150 ( dashed lines ) with @xmath146 , @xmath174gev , @xmath119gev , @xmath161gev , @xmath156gev , @xmath175gev , @xmath163 , and @xmath127 ( for which @xmath202 ) . | we study the decays of the @xmath0-sleptons ( @xmath1 ) and @xmath0-sneutrino ( @xmath2 ) in the minimal supersymmetric standard model ( mssm ) with complex parameters @xmath3 and @xmath4 ( u(1 ) gaugino mass ) .
we show that the effect of the cp phases of these parameters on the branching ratios of @xmath1 and @xmath2 decays can be quite strong in a large region of the mssm parameter space .
this could have an important impact on the search for @xmath1 and @xmath2 and the determination of the mssm parameters at future colliders .
tgu-29 + uwthph-2002 - 6 + zu - th 7/02 + hep - ph/0204071 + a. bartl@xmath5 , k. hidaka@xmath6 , t. kernreiter@xmath5 and w. porod@xmath7 [ cols= " < " , ] so far most phenomenological studies on supersymmetric ( susy ) particle searches have been performed in the minimal supersymmetric standard model ( mssm ) @xcite with real susy parameters .
studies of the 3rd generation sfermions are particularly interesting because of the effects of the large yukawa couplings .
the lighter sfermion mass eigenstates may be relatively light and they could be thoroughly studied at an @xmath8 linear collider @xcite .
they could also be copiously produced in the decays of heavier susy particles .
an analysis of the decays of the 3rd generation sleptons @xmath9 and @xmath2 in the mssm with real parameters was performed in ref.@xcite , and phenomenological studies of production and decays of the 3rd generation sfermions at future @xmath8 colliders in ref.@xcite .
the assumption that all susy parameters are real , however , may be too restrictive .
the higgsino mass parameter @xmath10 and the trilinear scalar coupling parameters @xmath11 of the sfermions @xmath12 may be complex . in minimal supergravity - type models
the phase of @xmath10 ( @xmath13 ) turns out to be restricted by the experimental data on electron and neutron electric dipole moments ( edms ) to a range @xmath14 for a universal scalar mass parameter @xmath15gev , while the phase of the universal trilinear scalar coupling parameter @xmath16 is correlated with @xmath13 , but otherwise unrestricted @xcite .
in more general models the phases of the parameters @xmath11 of the 3rd generation sfermions are not restricted at one - loop level by the edm data .
however , there may be restrictions at two - loop level @xcite . in a complete phenomenological analysis of production and decays of the susy particles one has to take into account that the @xmath10 and @xmath11 may be complex .
furthermore , explicit cp violation in the higgs sector can be induced by loop effects involving cp - violating interactions of higgs bosons to top and bottom squark ( @xmath17 and @xmath18 ) sector with complex parameters @xcite .
it is found @xcite that such effects of the complex phases on the phenomenology of the higgs boson search could be quite significant . in principle , the imaginary parts of the possible complex susy parameters involved could most directly and unambiguously be determined by measuring relevant cp - violating observables ; e.g. such analyses in @xmath0-slepton ( @xmath19 ) pair production in @xmath20 and @xmath21 colliders were performed in ref.@xcite .
+ on the other hand , the cp - conserving observables also can depend on the phases of the complex parameters because in general the mass - eigenvalues and the couplings of the susy particles ( sparticles ) involved are functions of the underlying complex parameters .
for example , the branching ratios of the higgs boson decays depend strongly on the complex phases of the @xmath17 and @xmath18 sectors @xcite . in this article
we study the effects of the complex phases of the stau and gaugino - higgsino sectors on the decay branching ratios of the staus @xmath1 and @xmath0-sneutrino @xmath2 with @xmath22(@xmath9 ) being the lighter ( heavier ) stau .
we point out that these effects can be quite strong in a large region of the mssm parameter space .
this could have an important impact on the search for @xmath1 and @xmath2 and the determination of the mssm parameters at future colliders .
first we summarize the mssm parameters in our analysis . in the mssm
the stau sector is specified by the mass matrix in the basis @xmath23
@xcite @xmath24 with @xmath25 a_\tau m_\tau & = & ( a_\tau - \mu^ * \tan\beta)\,m_\tau \,\ , = \,\ , |a_\tau m_\tau| \ , e^{i\varphi_{\sa } } \,\ , ( -\pi < \varphi_{\sa } \leq \pi ) .
\label{eq : d}\end{aligned}\ ] ] @xmath26 and @xmath27 are soft susy breaking parameters , @xmath10 is the higgsino mass parameter , and @xmath28 with @xmath29 @xmath30 being the vacuum expectation value of the higgs field @xmath31 @xmath32 . as the relative phase @xmath33 between @xmath29 and @xmath34 is irrelevant in our analysis , we adopt the @xmath35 scheme @xcite .
we take @xmath27 and @xmath10 as complex parameters : @xmath36 and @xmath37 with @xmath38 .
diagonalizing the matrix ( [ eq : a ] ) one gets the mass eigenstates @xmath22 and @xmath9 @xmath39 with the masses @xmath40 and @xmath41 ( @xmath42 ) , and the mixing angle @xmath43 @xmath44 the @xmath45 mixing is large if @xmath46 , which may be the case for large @xmath47 and @xmath48 . from eqs.([eq
: f ] ) and ( [ eq : g ] ) we see that @xmath49 and @xmath43 depend on the phases only through a term @xmath50 .
this phase dependence is strongest if @xmath51 .
the mass of @xmath2 is given by @xmath52 the properties of charginos @xmath53 ( @xmath54 ; @xmath55 ) and neutralinos @xmath56 ( @xmath57 ; @xmath58 ) are determined by the parameters @xmath59 and @xmath47 , where @xmath60 and @xmath4 are the su(2 ) and u(1 ) gaugino masses , respectively .
we assume that @xmath60 and the gluino mass @xmath61 are real and that @xmath4 is complex : @xmath62 .
inspired by the gaugino mass unification we take @xmath63 and @xmath64 . in the mssm higgs sector with explicit cp violation the mass - eigenvalues and couplings of the neutral and charged higgs bosons @xmath65 @xmath66 and @xmath67 , including yukawa and qcd radiative corrections , are fixed by @xmath68 and @xmath69 are the soft susy - breaking parameters in the @xmath17 and @xmath18 sectors , and @xmath69 are in general also complex : @xmath70 .
the neutral higgs mass eigenstates @xmath71 and @xmath72 are mixtures of cp - even and cp - odd states ( @xmath73 and @xmath74 ) due to the explicit cp violation in the higgs sector . for the radiatively corrected masses and mixings of the higgs bosons we use the formulae of ref.@xcite .
we treat @xmath75 and @xmath76 as free parameters since the ratios among them are highly model - dependent . here
we list possible important decay modes of @xmath1 and @xmath2 : _ 1 & & _ i , _ _ j [ eq : i ] + _ 2 & & _ i , _
_ j , _ 1 z^0 , _
w^- , _ 1 h_k^0 , _ h^- [ eq : j ] + _ & & _ _ i , _ j , _ 1 w^+ , _ 1 h^+ .
[ eq : k ] the decays into a gauge or higgs boson in ( [ eq : j ] ) and ( [ eq : k ] ) are possible in case the mass splitting between the sleptons is sufficiently large @xcite .
the explicit expressions of the widths of the decays ( [ eq : i])-([eq : k ] ) in case of real susy parameters are given in @xcite .
those for complex parameters can be obtained by using the corresponding masses and couplings ( mixings ) from refs.@xcite and will be presented elsewhere @xcite .
the phase dependence of the widths stems from that of the involved mass - eigenvalues , mixings and couplings among the interaction - eigenfields . here
we summarize the latter phase dependence : 1 .
@xmath77 sector : 1 .
@xmath78 are insensitive to the phases @xmath79 if the @xmath19-mixing term @xmath80 . in most cases this is naturally fulfilled , because @xmath81 is small .
2 .
the @xmath19-mixing angle @xmath43 ( given by @xmath82 ) is sensitive to @xmath79 ( via @xmath50 ) if and only if @xmath83 is small _ and _ @xmath84 . here note that @xmath85 due to the smallness of @xmath86 .
2 .
@xmath87 and @xmath88 sectors : 1 .
@xmath89 ( i=1, ... ,4 ) and the @xmath90-mixing matrix are sensitive @xmath91insensitive@xmath92 $ ] to the phases @xmath93 for small @xmath91large@xmath92 $ ] @xmath47 .
2 . @xmath94 and the @xmath95-mixing matrices are sensitive @xmath91insensitive@xmath92 $ ] to @xmath13 for small @xmath91large@xmath92 $ ] @xmath47 .
3 .
@xmath2 and @xmath67 sectors : + these sectors are independent of the phases . 4 .
@xmath96 sector : 1 .
@xmath97 ( i=1,2,3 ) are sensitive @xmath91insensitive@xmath92 $ ] to the phase sums @xmath98 for small @xmath91large@xmath92 $ ] @xmath47 @xcite .
2 . in general the @xmath96-mixing matrix ( a real orthogonal @xmath99 matrix @xmath100 ) is sensitive to @xmath98 for any @xmath47 @xcite .
5 . the couplings among the interaction - eigenfields : 1 . for the decays into fermions and gauge bosons in eqs.([eq : i])-([eq : k ] ) , they are gauge couplings and@xmath101or tau yukawa coupling ( @xmath102 ) , and are independent of the phases
. 2 . for the decays into higgs bosons in eqs.([eq : j ] ) and ( [ eq : k ] ) , the slepton-chirality " flip ( nonflip ) couplings are dependent on ( independent of ) the phases @xmath103 and @xmath13 ; c(_^_r h^+ ) & ~ & h _ ( a_^ * + ) [ eq : l ] + c(_l^_r _ 1 ) & ~ & h_a_^ * [ eq : m ] + c(_l^_r _ 2 ) & ~ & h _ [ eq : n ] + c(_l^_r a ) & ~ & h _ ( a_^ * + ) [ eq : o ] + c(_^_l h^+ ) & ~ & c(_l^_l _ 1,2 ) ~ c(_r^_r _ 1,2 ) ~ g m_w [ eq : p ] with h_= g m_/ ( m_w ) .
[ eq : q ] here @xmath104 ( i=1,2 ) and @xmath105 are the cp - even and cp - odd neutral higgs bosons , respectively @xcite . from the facts
( i)-(v ) the widths ( and hence the branching ratios ) of the decays ( [ eq : i])-([eq : k ] ) are expected to be sensitive to the phases ( @xmath103 , @xmath13 , @xmath106 , @xmath98 ) in a large region of the mssm parameter space .
now we turn to the numerical analysis of the @xmath1 and @xmath2 decay branching ratios .
we calculate the widths of all possibly important two - body decay modes of eqs.([eq : i])-([eq : k ] ) .
three - body decays are negligible in this study .
we take @xmath107 gev , @xmath108 gev , @xmath109 gev , @xmath110 gev , @xmath111 , @xmath112 , @xmath113 , and @xmath114 ( with @xmath115 for the determination of @xmath116 , @xmath117 being the number of quark flavors ) , where @xmath118 are pole masses of t and b quarks . in order not to vary too many parameters we fix @xmath119gev , @xmath120gev , @xmath121gev , and @xmath122 for simplicity . in our numerical study
we take @xmath47 , @xmath60 , @xmath41 , @xmath123 , @xmath48 , @xmath103 , @xmath13 and @xmath106 as input parameters .
note that for a given set of the input parameters we have two solutions for ( @xmath124 , @xmath125 ) corresponding to the two cases @xmath126 and @xmath127 . in the plots we impose the following conditions in order to respect experimental and theoretical constraints : 1 .
@xmath128 gev , @xmath129 gev , @xmath130 gev , @xmath131 , + @xmath132 110 gev , 2 .
@xmath133 , @xmath134 , and @xmath135 , where @xmath136 and @xmath137 , 3 .
@xmath138 @xcite using the formula of @xcite . condition ( i ) is imposed to satisfy the experimental mass bounds from lep @xcite .
( ii ) is the approximate necessary condition for the tree - level vacuum stability @xcite .
( iii ) constrains @xmath10 and @xmath47 ( in the squark sector ) .
we do not impose the @xmath139 constraint @xcite since it strongly depends on the details of the flavour structures of the squarks , including the generation - mixing . in general the experimental upper limits on the electron and neutron electric dipole moments ( edms ) strongly constrain the susy cp phases @xcite .
one interesting possibility for evading these constraints is to invoke large masses ( much above the tev scale ) for the first two generations of the sfermions @xcite , keeping the third generation sfermions relatively light ( @xmath1401 tev ) .
in such a scenario ( @xmath106 , @xmath13 ) and the cp phases in the third generation ( @xmath103 , @xmath141 , @xmath142 ) are practically unconstrained @xcite .
we take this scenario .
the deviation of the recent data on the muon g-2 from the standard model prediction is no longer significant @xcite , which allows of our scenario .
we have checked that the electron and neutron edm constraints at two - loop level @xcite are fulfilled in the numerical examples studied in this article . in fig.1
we plot the contours of the branching ratios of the @xmath22 decays @xmath143 and @xmath144 in the @xmath145 plane for @xmath146 , @xmath147gev , @xmath119gev , @xmath148gev , @xmath149 , and ( @xmath103 , @xmath13 ) = ( 0,0 ) , ( @xmath150,0 ) , ( 0,@xmath150 ) in the case @xmath127 ; in the case @xmath126 we have obtained similar results .
as expected , these branching ratios are very sensitive to the phases @xmath103 and @xmath13 in a sizable region of the @xmath145 plane .
as can be seen from item ( i ) , in this case the @xmath19-mixing angle @xmath43 is sensitive to cos@xmath151 for @xmath152 , which is the main cause for the difference between the case of ( @xmath103 , @xmath13 ) = ( 0,0 ) and those of ( @xmath150,0 ) and ( 0,@xmath150 ) .
furthermore from item ( ii ) one sees that the masses and mixing - matrices of the @xmath87 and @xmath88 are sensitive to @xmath13 , which is the main reason for the difference between the case of ( @xmath103 , @xmath13 ) = ( 0,@xmath150 ) and those of ( 0,0 ) and ( @xmath150,0 ) . here
note also item ( v)-(a ) .
in fig.2 we plot the contours of the @xmath22 decay branching ratios @xmath143 , @xmath153 , and @xmath144 in the @xmath154 plane for @xmath146 , @xmath147gev , @xmath119gev , @xmath155gev , @xmath156gev , @xmath157gev , and @xmath149 in the case @xmath127 ; in the case @xmath126 we have obtained similar results .
one sees that these branching ratios depend on the cp phases @xmath103 and @xmath13 quite strongly , as expected from items ( i ) and ( ii ) . in fig.3
we show the @xmath13 dependence of the @xmath22 decay branching ratios for @xmath158 ( with @xmath149 ) and @xmath159 ( with @xmath146 ) with @xmath160gev , @xmath161gev , @xmath156gev , @xmath162gev , and @xmath163 in the case @xmath127 ; in the case @xmath126 we have obtained similar results . from fig.3a
we see that the @xmath22 decay branching ratios are sensitive ( insensitive ) to @xmath13 for small ( large ) @xmath47 as is expected from items ( i ) , ( ii ) , and ( v)-(a ) . in general the @xmath22 decay
branching ratios are insensitive to all of the phases ( @xmath103 , @xmath13 , @xmath106 ) for large @xmath47 according to items ( i ) , ( ii ) and ( v)-(a ) . from fig.3b
we find that they are sensitive to @xmath13 and @xmath106 for small @xmath47(=3 ) where the masses and mixings of @xmath87 @xmath164 $ ] are sensitive to ( @xmath13 , @xmath106 ) @xmath165 $ ] ( though @xmath43 is insensitive to @xmath13 in this case ) as seen from items ( i ) and ( ii ) . in fig.4 we show the @xmath103 dependence of the @xmath9 decay branching ratios for @xmath166 and @xmath150 in the two cases of small ( a ) and large ( b , c ) mass difference @xmath167 .
for the former case we take ( @xmath40(gev ) , @xmath41(gev ) , @xmath47 , @xmath60(gev ) , @xmath123(gev ) , @xmath48(gev ) , @xmath106 ) = ( 240 , 260 , 6 , 500 , 600 , 150 , 0 ) with @xmath126 , whereas for the latter case we take ( @xmath40(gev ) , @xmath41(gev ) , @xmath47 , @xmath60(gev ) , @xmath123(gev ) , @xmath48(gev ) , @xmath106 ) = ( 240 , 500 , 30 , 400 , 900 , 800 , 0 ) with @xmath127 . in fig.4a ( where @xmath167 and @xmath168 are so small that the bosonic decays in eq.([eq : j ] ) are kinematically forbidden ) we see that the @xmath9 decay branching ratios are very sensitive to @xmath103 and @xmath13 as expected from items ( i ) and ( ii ) . in figs.4b and 4c ( where @xmath169 is so large that the bosonic decays in eq.([eq : j ] ) also are allowed ) we find that the branching ratios of the higgs boson modes are rather sensitive to @xmath103 and @xmath13 as expected from items ( iv)-(b ) and ( v)-(b ) ; here note that in this case @xmath170 due to the smallness of the @xmath171-@xmath172 mixing term and hence that the bosonic decays of @xmath9 are basically the decays of @xmath172 into @xmath173 . in fig.5
we show the @xmath13 dependence of the @xmath2 decay branching ratios for @xmath149 and @xmath150 in the case of @xmath146 , @xmath174gev , @xmath161gev , @xmath156gev , @xmath175gev , @xmath163 , and @xmath127 ( for which @xmath176 and hence the bosonic decays in eq.([eq : k ] ) are kinematically forbidden ) .
we see that the @xmath2 decay branching ratios are quite sensitive to @xmath13 and @xmath106 as expected from item ( ii ) . in general the @xmath2 decay
branching ratios are insensitive to all of the phases ( @xmath103 , @xmath13 , @xmath106 ) for large @xmath47 as can be seen from items ( i)@xmath177(iii ) and ( v ) .
furthermore , they tend to be insensitive to @xmath103 for any @xmath47 as can be seen from items ( i)@xmath177(iii ) and ( v ) , except for some special cases such as the case with @xmath178 , @xmath179 , small @xmath47 , and @xmath180 , where @xmath181 , @xmath182 , @xmath183 , and @xmath184 $ ] are higgsino - like [ gaugino - like ] ; in this case only the @xmath185 and @xmath186 modes dominate the @xmath2 decay , and the coupling @xmath187 can be rather sensitive to @xmath103 ( though the @xmath43 is insensitive to @xmath103 ) , which results in significant @xmath103-dependence of the @xmath2 decay branching ratios .
we have checked that this can be easily realized indeed . as for the @xmath0-lepton edm ( @xmath188 )
, we have checked that in the mssm parameter region considered here the predicted range of it is well below the current experimental limit @xmath189 @xcite and most likely also below the expected sensitivity of future experiments to measure this edm : we find that in the parameter region considered here @xmath190 , which is obtained by using the corresponding formulas in ref .
@xcite with @xmath191 replaced by @xmath81 .
the cp phases can significantly affect not only cp - violating observables such as the lepton edm but also cp - conserving quantities such as the branching ratios of the @xmath1 and @xmath2 decays .
hence the possible sizable phases could have important consequences for the determination of the fundamental mssm parameters by measurements of cp - conserving observables from which they are extracted . in conclusion
, we have shown that the effect of the cp phases of the complex parameters @xmath27 , @xmath10 and @xmath4 on the branching ratios of the @xmath1 and @xmath2 decays can be quite strong in a large region of the mssm parameter space .
this could have an important impact on the search for @xmath1 and @xmath2 and the determination of the mssm parameters at future colliders . |
the lhc has recently observed a new scalar particle with a mass around @xmath11 gev that could be consistent with the higgs boson of the standard model ( sm ) @xcite .
in addition , a study of the combined tevatron data has revealed a smaller broad excess corresponding to a mass between 115 gev and 135 gev @xcite which is consistent with this lhc discovery . with more data collected ,
the lhc is expected to be able to unveil the detailed properties of the new scalar particle and verify its nature . from the theoretical side ,
there has been a collective effort in the past decades in the search for new physics beyond the sm , that can address some of the fundamental unresolved questions in particle physics .
one simple candidate that was extensively studied in the past several years is the so called sm4 ( also referred to as naive " or
simple " sm4 ) ; the sm with a fourth sequential generation of fermions ( for reviews see @xcite ) .
this simple extension of the sm was studied for addressing some of the challenges in particle physics , such as : the hierarchy problem @xcite , the origin of matter - anti matter asymmetry in the universe @xcite and the apparent anomalies in flavor physics @xcite .
unfortunately , the recent lhc searches for the sm4 heavy 4th generation quarks have now pushed the exclusion limits to @xmath12 gev for the @xmath13 and @xmath14 gev for the @xmath15 @xcite , which is on the border of their perturbative regime . moreover , the sm4 higgs was already excluded in the mass range @xmath16 gev by the 2011 data @xcite when @xmath17 .
thus , the above reported discovery of a light higgs with a mass around 125 gev is not compatible with the sm4 that includes a heavy 4th generation neutrino with a mass @xmath18 gev , i.e. , with a mass larger than the current lower bound on @xmath19 @xcite .
in fact , it was further pointed out recently in @xcite that the interpretation of the measured higgs signals is not consistent with the sm4 also for the case @xmath20 . in particular , in the sm4 , the leading gluon - fusion light higgs production mechanism is enhanced by a factor of @xmath21 due to the contribution of diagrams with @xmath5 and @xmath22 in the loops @xcite , which in general leads to larger signals ( than what was observed at the lhc ) in the @xmath23 channels . however ,
if the 4th generation masses are of @xmath24 gev , then the decay channels @xmath25 are suppressed due to nlo corrections @xcite , and the exclusion of the sm4 is based mainly on the @xmath26 channel . in the @xmath27 channel
there is also a substantial suppression of @xmath28 due to ( accidental ) destructive interference in the loops @xcite and another @xmath28 factor due to nlo corrections @xcite .
when @xmath29 is taken to be light enough so that @xmath30 becomes @xmath31 , then the @xmath32 channel becomes further suppressed to the level that the observed excess can no longer be accounted for @xcite .
for a recent comprehensive analysis of the sm4 status in light of the latest higgs results and electroweak precision data ( ewpd ) , we refer the reader to @xcite
. however , as was noted already 20 years ago @xcite and more recently in @xcite , if heavy 4th generation fermions are viewed as the agents of electroweak symmetry breaking ( and are , therefore , linked to strong dynamics at the nearby tev - scale ) , then more higgs particles are expected at the sub - tev regime . in this case , the higgs particles may be composites of the 4th generation fermions and the low - energy composite higgs sector should resemble a two ( or more ) higgs doublet framework . indeed , the phenomenology of multi - higgs 4th generation models was studied recently in @xcite and within a susy framework in @xcite , for a review see @xcite . in @xcite
it was also shown that the current exclusion limits on the sm4 @xmath5 and @xmath22 could be significantly relaxed if the four generations scenario is embedded in a 2hdm framework .
adopting this viewpoint , i.e. , that the 4th generation setup should be more adequately described within a multi - higgs framework , we will study in this paper the expected higgs signals of a 2hdm with a 4th generation family , investigating whether the interpretation of the recently measured 125 gev higgs properties are consistent with one of the neutral scalars of the 4th generation 2hdm .
the 2hdm structure has an inherent freedom of choosing which doublet couples to which fermions .
for the three generations 2hdm , three popular setups were suggested which are usually referred to as type i , type ii and type iii 2hdm . in the case
where the 2hdm is assumed to underly some form of tev - scale strong dynamics mediated by the 4th generation fermions , we expect the higgs composites to couple differently to the 4th generation fermions .
this can be realized in a class of 2hdm models named the 4g2hdm , suggested in @xcite .
most of our analysis below is performed in this 4g2hdm framework and a comparison to a 2hdm of type ii ( which also underlies the susy higgs sector ) with and without 4th generations will also be discussed .
let us recapitulate the salient features of the 2hdm frameworks with a 4th generation of fermions ( we will focus below on the quarks sector , but a generalization to the leptonic sector is straight forward ) . assuming a common generic 2hdm potential , the phenomenology of 2hdm s is generically encoded in the texture of the yukawa interaction lagrangian . the simplest variant of a 2hdm with 4th generations of fermions , can be constructed based on the so called type ii 2hdm ( which we denote hereafter by 2hdmii ) , in which one of the higgs doublets couples only to up - type fermions and the other to down - type ones .
this setup ensures the absence of tree - level flavor changing neutral currents ( fcnc ) and is , therefore , widely favored when confronted with low energy flavor data .
the yukawa terms for the quarks of the 2hdmii , extended to include the extra 4th generation quark doublet is : @xmath33 where @xmath34 ( @xmath35 ) are left(right)-handed fermion fields , @xmath36 is the left - handed @xmath37 quark doublet , @xmath38 are general @xmath39 yukawa matrices in flavor space and @xmath40 are the higgs doublets : @xmath41 as mentioned above , motivated by the idea that the low energy scalar degrees of freedom may be the composites of the heavy 4th generation fermions , @xcite have constructed a new class of 2hdm s , named the 4g2hdm , that can effectively parameterize 4th generation condensation by giving a special status to the 4th family fermions .
the possible viable variants of this approach can be parameterized as @xcite : @xmath42 where @xmath43 are the two higgs doublets , @xmath44 is the identity matrix and @xmath45 ( @xmath46 ) are diagonal @xmath39 matrices defined by @xmath47 .
in particular , in the type i 4g2hdm of @xcite ( which we will focus on below and which will be denoted hereafter simply as the 4g2hdm ) , one sets @xmath48 , so that the heavier " higgs field ( @xmath49 ) is assumed to couple only to the 4th generation quarks , while the lighter " higgs field ( @xmath50 ) is responsible for the mass generation of all other ( lighter 1st-3rd generations ) fermions .
the yukawa interactions for these 4g2hdm models in terms of the physical states were given in @xcite .
for the lighter cp - even higgs it reads : @xmath51 \right\ } q_j h \label{sff1}~ , \\\end{aligned}\ ] ] where @xmath52 is mixing angle in the cp - even neutral higgs sector , @xmath53 is the ratio between the vevs of @xmath54 and @xmath55 and @xmath56 , @xmath57 are new mixing matrices in the down(up)-quark sectors , which are obtained after diagonalizing the quarks mass matrices .
these matrices are key parameters of the model , which depend on the rotation ( unitary ) matrices of the right - handed down and up - quarks , @xmath58 and @xmath59 , and on whether @xmath60 and/or @xmath61 are turned on " : @xmath62 thus , as opposed to standard " 2hdm s , in the 4g2hdm some elements of @xmath58 and @xmath59 are physical and can , in principle , be measured in higgs - fermion systems .
in particular , inspired by the working assumption of the 4g2hdm and by the observed flavor pattern in the up and down - quark sectors , it was shown in @xcite that , for @xmath48 , the new mixing matrices @xmath56 and @xmath57 are expected to have the following form : @xmath63 and similarly for @xmath56 by replacing @xmath64 . the new parameters @xmath65 and @xmath66 are free parameters of the model that effectively control the mixing between the 4th generation and the 3rd generation quarks .
we therefore expect @xmath67 , so that a natural choice for these parameter would be @xmath68 and @xmath69 ( see also @xcite ) . in what follows we will thus set @xmath70 and vary the @xmath3 mixing parameter in the range @xmath71 .
@xmath72 clearly , once a new higgs doublet is introduced , the phenomenology of the higgs particles production and decays becomes more complicated .
in particular , the new yukawa couplings depend on several more parameters ( i.e. , in the 4g2hdm on @xmath65 , @xmath66 , @xmath73 and @xmath52 ) and the cp - even higgs couplings to the @xmath74 and to the @xmath75 bosons have extra pre - factors of @xmath76 and @xmath77 ( the pseudoscalar @xmath10 does not couple at tree - level to the @xmath74 and the @xmath75 ) . as a result
, the one - loop @xmath78 decay and the leading @xmath79 higgs production mechanism can be significantly altered compared to their sm and sm4 values , depending on @xmath65 , @xmath73 , @xmath52 and on the 4th generation fermion masses ( i.e. , assuming @xmath67 , therefore setting @xmath80 throughout our analysis ) .
this is demonstrated in fig .
[ hgg ] , where we plot the widths @xmath81 and @xmath82 , as a function of these three parameters setting @xmath83 gev .
the dependence on @xmath84 is depicted in a narrow range around @xmath85 , for which the 4g2hdm is consistent with both ewpd @xcite and with the observed 125 gev higgs signals ( see below ) .
we see that both @xmath78 and @xmath86 have a strong dependence on the higgs mixing angle @xmath52 , while @xmath78 is also very sensitive to @xmath84 and to the new @xmath3 mixing parameter @xmath65 , due to their role in the interference between the fermion loops and the w - boson loop . in fig .
[ br_h ] we further plot the various relevant branching ratios of a 125 gev @xmath0 in the 4g2hdm , as a function of @xmath52 for @xmath87 , @xmath88 and @xmath89 gev .
let us now turn to the recently reported lhc higgs searches and the implications of the discovery of a 125 gev higgs - like particle on the 4g2hdm setup with a 4th generation of fermions .
the quantity that is usually being used for comparison between the lhc and tevatron results and the expected signals in various models is the normalized cross - section : @xmath90 for the observed ratios of cross - sections , i.e. , the signal strengths @xmath91 , and the corresponding errors @xmath92 , we use the latest results as published in @xcite : * @xmath7 : @xmath93 ( taken from @xmath94 * @xmath95 : @xmath96 * @xmath97 : @xmath98 * @xmath99 : @xmath100 * @xmath101 : @xmath102 * @xmath103 : @xmath104 @xmath105 @xmath106 the values given above are the result of a combination of the most recent data in each channel.@xmath107}$ ] the uncertainties are calculated by treating the reported experimental uncertainties as statistical and assigning 15% theoretical uncertainty to the gluon fusion production mechanism and 5% theoretical uncertainty on electroweak production mechanisms and on the branching fractions @xcite .
one can easily notice that the channels which have the highest sensitivity to the higgs signals and contributed the most to the recent 125 gev higgs discovery are @xmath108 and @xmath109 .
in all other channels the results are not conclusive and at this time they are consistent with the background only hypothesis at the level of less than @xmath110 . clearly , a sm higgs is ideally most consistent with @xmath111 in every channel , while in other models we expect some deviations in the various measured channels , depending on the parameters of the model and on the mass of the scalar candidate which should be compatible with the lhc results .
thus , the comparison to any given model can be performed using a @xmath112 fit : @xmath113 where @xmath92 are the errors on the observed cross - sections and @xmath114 is the calculated normalized cross - section in any given model . in particular , we take advantage of the fact that @xmath115 , and calculate @xmath114 using @xmath116 where @xmath117 is the higgs production mechanism , i.e. , either by gluon fusion @xmath118 , vector boson fusion @xmath119 or the associated higgs - w production @xmath120 at the tevatron .
the higgs signals in a 2hdm setup with a 4th generation of fermions have already been discussed to some extent in the literature @xcite , but with no general picture of how these signals match all the observed higgs cross - sections reported above . here
, we try to quantify how well the 2hdm scenarios ( where the lightest higgs particle , @xmath0 , has a mass of 125 gev ) fit all the available higgs data , by calculating the @xmath112 for all the relevant channels in two 2hdm realizations with four generations - the 2hdmii and the 4g2hdm ( see section [ sec2 ] ) .
we use the latest version of hdecay @xcite , with recent nlo contributions which also include the heavy 4th generation fermions , where we have inserted all the relevant couplings of the 4g2hdm and the 2hdmii frameworks described in section [ sec2 ] . for the 4th generation fermion masses involved in the loops of the decays @xmath121 ( i.e. , in the 1-loop nlo corrections for the cases @xmath122 ) , we have used the approximation of a degenerate 4th generation fermion sector , where we have tested below two representative cases : @xmath123 and @xmath124 gev ( the effect of mass splittings between 4th generation fermions on @xmath125 is negligible ) .
it is important to note that , while the first case ( @xmath89 gev ) is excluded for the sm4 @xcite , it is not necessarily excluded for the 4g2hdm , since in this model the decay patterns of @xmath5 and @xmath22 can have a completely different topology , e.g. , having @xmath126 , for which the current limits ( which are based on the standard " sm4 decays @xmath127 and @xmath128 ) do not apply , see @xcite . as mentioned earlier , we find that the simple sm4 case , with a 125 gev higgs is excluded to many @xmath129 s when confronted with the higgs search results . also ,
as was already noted in @xcite in the context of the standard " 2hdmii ( i.e. , with four generations ) , we find that the simplest case of a light 125 gev pseudoscalar @xmath10 of any 2hdm , with and without a 4th family , is not compatible with the higgs data , irrespective of the 4th generation fermion masses . in particular , the signals of the 125 gev higgs decaying into a pair of vector bosons , @xmath130 and @xmath131 , excludes this possibility due to the absence of tree - level @xmath132 and @xmath133 couplings .
we , therefore , focus below only on the case where the observed 125 gev higgs - like particle is the lighter cp - even higgs , @xmath0 .
we plot in fig .
[ chisqall1 ] the resulting @xmath112 and p - values in the 4g2hdm case ( combining all the six reported higgs decay channels above ) , with @xmath134 gev , @xmath89 and 600 gev , @xmath135 and 0.5 and for @xmath136 ( this range is allowed by ewpd and flavor physics in the 2hdm 4th generation setups , see @xcite ) .
the value of the higgs mixing angle @xmath52 is the one which minimizes the @xmath112 for each value of @xmath84 .
the sm best fit is also shown in the plot . in fig .
[ chisqall2 ] we further plot the resulting @xmath112 and p - values as a function of @xmath84 , this time minimizing for each value of @xmath84 with respect to both @xmath52 and @xmath65 ( in the 4g2hdm case ) . for comparison
, we also show in fig .
[ chisqall2 ] the @xmath112 and p - values for a 125 gev @xmath0 in the 2hdmii with a 4th generation and in the sm .
looking at the p - values in figs .
[ chisqall1 ] and [ chisqall2 ] ( which measure " the extent to which a given model can be successfully used to interpret the higgs data in all the measured decay channels ) we see that , @xmath0 of the 4g2hdm with @xmath1 and @xmath137 gev is a good candidate for the recently observed 125 gev higgs , giving a fit comparable to the sm fit .
the standard " 2hdmii setup with @xmath138 gev is also found to be consistent with the higgs data in a narrower range of @xmath139 .
we find that the fit favors a large @xmath3 mixing parameter @xmath65 , implying @xmath140 which completely changes the @xmath5 decay pattern @xcite and , therefore , significantly relaxing the current bounds on @xmath141 @xcite .
this can be seen in table [ tab1 ] where we list 6 representative sets of best fitted values ( to be used in the plots below ) for @xmath142 in the 4g2hdm , that correspond to points on the best fitted 4g2hdm curves shown in fig .
[ chisqall2 ] . in fig .
[ higgsewpd ] we further test the goodness of fit for the 125 gev @xmath0 of the 4g2hdm , where , in addition to the higgs results , we explicitly imposed the constraints on the 4g2hdm parameter space from ewpd ( from the s and t parameters and from @xmath143 ) using the results in @xcite .
evidently , our conclusions above do not change after adding the ewpd constraints to the analysis .
finally , we note that we have also tested the 3 generations type ii 2hdm and found that its lightest cp - even higgs is also a good candidate for the observed 125 gev higgs particle , giving a fit which is also comparable to the sm fit for @xmath1 .
.__six representative best fitted sets of values for @xmath142 in the 4g2hdm , corresponding to points on the best fitted 4g2hdm curves shown in fig .
[ chisqall2 ] . _ _ [ cols="^,^,^,^,^",options="header " , ]
in fig . [ indv_pulls ] we plot the individual pulls and the signal strengths for the various measured channels , @xmath144 and @xmath145 , respectively , as a function of @xmath84 , for the above best fitted 4g2hdm curve with @xmath89 gev .
we see that appreciable deviations from the sm are expected in the channels @xmath101 , @xmath7 and @xmath146 .
in particular , the most notable effects are about a @xmath147 deviation ( from the observed value ) in the vbf diphoton channel @xmath148 and a @xmath149 deviation in the @xmath101 channel .
the deviations in these channels are in fact a prediction of the 4g2hdm strictly based on the current higgs data , which could play a crucial role as data with higher statistics becomes available .
they can be understood as follows : the channels that dominate the fit ( i.e. , having a higher statistical significance due to their smaller errors ) are @xmath150 .
thus , since the @xmath79 production vertex is generically enhanced by the @xmath151 loops , the fit then searches for values of the relevant 4g2hdm parameters which decrease the @xmath152 decays in the appropriate amount .
this in turn leads to an enhanced @xmath101 ( i.e. , due to the enhancement in the @xmath79 production vertex ) and to a decrease in the @xmath7 and @xmath153 , which are independent of the enhanced @xmath154 vertex but are sensitive to the decreased @xmath155 one .
it is important to note that some of the characteristics of these predictions " can change with more data collected .
@xmath156 .
we conclude with the implications of the above results for the other two neutral scalars of the 4g2hdm . for the heavier cp - even neutral higgs , @xmath157
, we consider its decays to @xmath158 and @xmath159 , which are currently the most sensitive channels in which searches for a heavy sm higgs were performed at the lhc .
a useful approximation of the expected exclusion range on @xmath160 can be performed by comparing the calculated signal strengths : @xmath161 to the observed / measured values of this quantity , i.e. , to @xmath162 ( note that in the 4g2hdm we find @xmath163 for @xmath164 ) . in fig .
[ scalarh ] we plot @xmath165 as a function of @xmath160 for the 6 best fitted points of the relevant 4g2hdm parameter space , given in table [ tab1 ] .
we also show in fig .
[ scalarh ] an approximate exclusion line for @xmath166 , i.e. , for the observed signal strength @xmath167 , which we have extracted from the most recent cms exclusion plot in this channel ( see @xcite ) and which is currently the most stringent observed exclusion limit for a heavy higgs with a mass @xmath168 gev .
we see that @xmath169 gev is excluded by the current data in the @xmath170 channel for points p1 , p2 , p3 and p5 ( i.e. , @xmath171 for @xmath169 gev ) , while for point p4 and p6 @xmath172 gev is allowed .
the current cms and atlas higgs data in the @xmath158 and @xmath159 channels are not sensitive to the pseudoscalar @xmath10 , due to the absence of a tree level @xmath132 and @xmath133 coupling ( and due to the smallness of the corresponding @xmath132 and @xmath133 one - loop couplings @xcite ) .
therefore , the only relevant search channels which are currently sensitive to the @xmath10 decays are @xmath173 and @xmath174 , for which a search for the higgs was performed up to a higgs mass slightly below @xmath175 by both cms and atlas . defining the signal strengths for the @xmath10 signals as : @xmath176 we plot in figs .
[ scalaratau ] and [ scalaragam ] @xmath177 and @xmath178 , respectively , as a function of @xmath179 ( we assume that @xmath180 ) , for points p1-p6 of table [ tab1 ] . here
also , we plot the existing approximate exclusion lines @xmath181 and @xmath182 , based on the most recent cms analysis , which currently gives the most stringent limits in these two channels @xcite . we see that a pseudoscalar as light as 130 gev is allowed by the current data , e.g. , for for points p1 and p2 .
we have studied the recently measured higgs signals in the framework of a specific 2hdm with a fourth generation of fermions ( the 4g2hdm suggested in @xcite ) , designed and motivated by the possibility that the sub - tev higgs particles are condensates of the heavy 4th generation fermions , which are therefore , viewed as the agents of dynamical electroweak symmetry breaking .
we find that the lightest cp - even higgs state of this model , @xmath0 , is a good candidate for the recently discovered 125 gev higgs signals in all the measured channels , within a large portion of the 4g2hdm allowed parameter space , which is consistent with the current bounds from ewpd .
in particular , for typical 4th generation fermion masses in the range @xmath137 gev , @xmath1 and a large @xmath3 mixing ( the parameter @xmath65 predicted by the model ) , the lightest 4g2hdm higgs gives a good overall fit to the current 125 gev higgs data - roughly comparable to the sm fit . for these values of the 4g2hdm parameter space , in particular with @xmath183 , the flavor changing @xmath5 decay @xmath184 dominates with @xmath185 , leading to different @xmath186 signatures ( than the simple sm4 ) that can be searched for at the lhc using the methods suggested in @xcite .
finally , the heavier cp - even higgs state , @xmath157 , is found to be excluded in this model up to @xmath188 gev , while the pseudoscalar higgs state , @xmath10 , can be as light as 130 gev and can , therefore , be discovered ( or ruled out in this small mass range ) with more data collected in the @xmath189 channels .
* acknowledgments : * sbs and mg acknowledge research support from the technion .
ge thanks r. godbole and s. vempati for hospitality and discussions . the work of as was supported in part by the u.s .
doe contract # de - ac02 - 98ch10886(bnl ) .
s. chatrchyan _
et al . _ , [ cms collaboration ] , phys
b716 * , 30 ( 2012 ) ; see also , j. incandela , observation of a narrow resonance near 125 gev in cms " , talk given at the ichep 2012 , july 4 - 11th , melbourne , australia .
et al . _ , [ atlas collaboration ] , phys
b716 * , 1 ( 2012 ) ; see also , r. hawkings , atlas higgs searches and experiment overview " , talk given at the ichep 2012 , july 4 - 11th , melbourne , australia . b. holdom , w.s .
hou , t. hurth , m.l .
mangano , s. sultansoy , g. unel , talk presented at _ beyond the 3rd sm generation at the lhc era workshop _ ,
geneva , switzerland , sep 2008 , arxiv:0904.4698 [ hep - ph ] , published in pmc phys .
* a3 * , 4 ( 2009 ) . for older literature on the sm4 , see : proceedings of the first ( february 1987 ) and the second ( february 1989 ) international symposiums on the _ fourth family of quarks and leptons _ , santa monica ,
ca , published in annals of the new york academy of sciences , 517 ( 1987 ) & 578 ( 1989 ) , edited by d. cline and a. soni . for a review of the implications of stable quarks of the 4th generation on cosmology , astrophysics and accelerator physics ,
see : k. belotsky , m. khlopov and k. shibaev , arxiv:0806.1067 [ astro - ph ] and references therein .
b. holdom , phys .
lett . * 57 * , 2496 ( 1986 ) , erratum - ibid . *
58 * , 177 ( 1987 ) ; w.a .
bardeen , c.t . hill and m.
lindner , phys .
* d41 * , 1647 ( 1990 ) ; s.f . king , phys .
* b234 * , 108 ( 1990 ) ; c. hill , m. luty and e.a .
paschos , phys .
* d43 * , 3011 ( 1991 ) ; p.q . hung and g. isidori phys .
* b402 * , 122 ( 1997 ) .
hou , chin .
* 47 * , 134 ( 2009 ) ; w.s .
hou , talk given at the _ 34th international conference on high energy physics ( ichep 2008 ) _ , philadelphia , pennsylvania , july 2008 , arxiv:0810.3396 [ hep - ph ] ; s.w .
ham , s.k .
oh , d. son , phys .
rev . * d71 * , 015001 ( 2005 ) ; g. w.s .
hou , int . j. mod
d20 * , 1521 ( 2011 ) .
s.w . ham , s.k .
oh , d. son , phys . rev .
* d71 * , 015001 ( 2005 ) ; r. fok , g.d .
kribs , phys . rev . * d78 * , 075023 ( 2008 ) . w .- s .
hou , m. nagashima , a. soddu , phys . rev . *
d76 * , 016004 ( 2007 ) ; m. bobrowski , a. lenz , j. riedl and j. rohrwild , phys .
d * 79 * , 113006 ( 2009 ) ; v. bashiry , n. shirkhanghah , k. zeynali , phys . rev .
* d80 * , 015016 ( 2009 ) ; w. s. hou and c. y. ma , phys .
* d82 * , 036002 ( 2010 ) ; o. eberhardt , a. lenz and j. rohrwild , phys . rev .
d * 82 * , 095006 ( 2010 ) ; s. nandi and a. soni , phys . rev .
* d83 * , 114510 ( 2011 ) ; a.k . alok , a. dighe and d. london , phys . rev . * d83 * , 073008 ( 2011 ) 073008 ; d. choudhury , d. k. ghosh , jhep * 1102 * , 033 ( 2011 ) ; r. mohanta , a.k .
giri , phys .
rev . * d85 * , 014008 ( 2012 ) ; a. ahmed , i. ahmed , m.j .
aslam , m. junaid , m.a .
paracha , a. rehman , phys .
* d85 * , 034018 ( 2012 ) .
luk [ on behalf of the cms collaboration ] , arxiv:1110.3246v2 [ hep - ex ] ; s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b716 * , 103 ( 2012 ) ; g. aad _
_ [ atlas collaboration ] , phys .
109 * , 032001 ( 2012 ) ; s. chatrchyan _ et al . _
[ cms collaboration ] , jhep * 1205 * , 123 ( 2012 ) .
l. bellantoni , j erler , j. heckman and e. ramirez - homs , phys . rev . * d86 * , 034022 ( 2012 ) .
the cms collaboration , cms physics analysis summaries , report number : cms - pas - hig-12 - 016 , available on the cern cds information server . | we interpret the recent discovery of a 125 gev higgs - like state in the context of a two higgs doublets model with a heavy 4th sequential generation of fermions , in which one higgs doublet couples only to the 4th generation fermions , while the second doublet couples to the lighter fermions of the 1st-3rd families .
this model is designed to accommodate the apparent heaviness of the 4th generation fermions and to effectively address the low - energy phenomenology of a dynamical electroweak symmetry breaking scenario .
the physical higgs states of the model are , therefore , viewed as composites primarily of the 4th generation fermions .
we find that the lightest higgs , @xmath0 , is a good candidate for the recently discovered 125 gev spin - zero particle , when @xmath1 , for typical 4th generation fermion masses of @xmath2 gev , and with a large @xmath3 mixing in the right - handed quarks sector .
this , in turn , leads to @xmath4 , which drastically changes the @xmath5 decay pattern .
we also find that , based on the current higgs data , this two higgs doublet model generically predicts an enhanced production rate ( compared to the sm ) in the @xmath6 channel and a reduced @xmath7 and @xmath8 ones . finally , the heavier cp - even higgs is excluded by the current data up to @xmath9 gev , while the pseudoscalar state , @xmath10 , can be as light as 130 gev .
these heavier higgs states and the expected deviations from the sm in some of the higgs production channels can be further excluded or discovered with more data . |
inflation is nowadays a well established paradigm , consistent with all the observations , that solves most of the puzzles of the hot big bang model in a very simple and elegant way .
it is able to explain not only the homogeneity and isotropy of the present universe on large scales , but also the generation of almost scale invariant primordial perturbations that give rise to the structure formation @xcite .
however , the naturalness of inflation is directly related to the origin of the inflaton .
most of the inflationary models proposed so far require the introduction of new degrees of freedom to drive inflation .
the nature of the inflaton is completely unknown , and its role could be played by any candidate able to imitate a scalar condensate ( tipically in the slow - roll regime ) , such as a fundamental scalar field , a fermionic or vector condensate , or even higher order terms of the curvature invariants .
the number of particle physics motivated candidates is as big as the number of extensions of the standard model ( grand unified theories , supersymmetry , extra dimensions , etc . ) , where it is not very difficult to find a field that could play the role of the inflaton @xcite .
in addition , given a model we must find a graceful exit to inflation and a mechanism to bring the universe from a cold and empty post - inflationary state to the highly entropic and thermal friedmann universe @xcite .
unfortunately , the theory of reheating is also far from being complete , since not only the details , but even the overall picture , depend crucially on the different microphysical models .
it seems difficult to study the details of reheating in each concrete model without the experimental knowledge of the strength of the interactions among the inflaton and the matter fields . because of this , most of the work until now has focussed on models encoding the different mechanisms that could play a role in the process , with the strength of the couplings set essentially by hand .
the relative importance of each one of these mechanisms can only be clarified in light of an underlying particle physics model , able to provide us with the couplings among the inflaton and matter fields . from this point of view
it is very difficult to single out a given model of inflation , and even more difficult to understand the details of the reheating process via the experimental access to the couplings .
we may be very far away from understanding the microphysical mechanisms responsible for inflation , but maybe the natural candidate for being the inflaton was already there long ago .
if we do not want to introduce new dynamical degrees of freedom in the theory , apart from those present in the standard model ( sm ) of particle physics , and at the same time we require lorentz and gauge invariance , we are left with just one possibility : the higgs field .
early models of inflation in terms of a higgs - like scalar field @xmath2 with a quartic self - interaction potential @xmath3 need an extremely small coupling constant @xmath4 @xcite , and are also nowadays excluded at around @xmath5 by the present observational data @xcite .
however , the sm - based inflation may be rescued from these difficulties replacing the usual einstein - hilbert action by a non - minimal coupling of the higgs field to the ricci scalar @xcite , in a scalar - tensor theory fashion @xmath6 the induced gravity ( @xmath7 ) action is indeed the most natural generalization of the standard model in a curved spacetime , given the relevance of the non - minimal coupling to gravity for renormalizing the theory @xcite .
it belongs to a group of theories known as scalar - tensor theories , originally introduced by brans and dicke @xcite to explain the origin of the masses . in those theories
not only the active masses but also the gravitational constant @xmath8 , are determined by the distribution of matter and energy throughout the universe , in a clear connection with the mach principle .
the interaction that gives rise to the masses should be the gravitational one , since gravity couples to all particles , _
i.e. _ to their masses or energies .
the gravitational constant @xmath8 in these theories is replaced by a scalar function that permeates the whole space - time and interacts with all the ordinary matter content , determining how the later moves through space and time .
any measurements of an object s mass depends therefore on the local value of this new field .
the successful higgs mechanism lies precisely in the same direction of the original mach s idea of producing mass by a gravitational - like interaction .
the role of the higgs field in the standard model is basically to provide the inertial mass of all matter fields through the local spontaneous symmetry breaking mechanism .
the higgs boson couples to all the particles in the standard model in a very specific way , with a strengh proportional to their masses @xcite , and mediates a scalar gravitational interaction of yukawa type @xcite , between those particles which become massive as a consequence of the local spontaneous symmetry breaking . according to the equivalence principle
, it seems natural to identify the gravitational and particle physics approaches to the origin of the masses . from this point of view
, the induced gravity action ( [ ig ] ) would be an indication of a connection between the higgs , gravity and inertia .
indeed , the action ( [ ig ] ) is , at least at the classical level , just a different representation of the starobinsky s model of inflation @xcite , where inflation is entirely a property of the gravitational sector .
both representations of the same theory are simply related by a legendre transformation .
this fact , together with the possibility of having an inflationary expansion of the universe , makes the model extremely appealing .
unfortunately , the induced gravity model can not be accepted as a completely satisfactory inflationary scenario , since the gauge bosons acquire a constant mass in the einstein frame and totally decouple from the higgs - inflaton field @xcite , which translates into an inefficient reheating of the universe .
notice however that the action ( [ ig ] ) is not the most general one that can be written in a nontrivial background . as was shown in @xcite the simultaneous existence of a reduced _ bare _ planck mass @xmath9 and a non - minimal coupling of a symmetry breaking field to the scalar curvature @xmath10 avoid the decoupling of the gauge bosons and can give rise to an inflationary expansion of the universe together with a potentially successful reheating .
in this paper we initiate the study of the reheating process in the model presented in ref .
@xcite , in which the symmetry breaking field is the standard model higgs , strongly non - minimally coupled to gravity and playing the role of the inflaton .
the novelty and great advantage of this model is its connection with a well - known microphysical mechanism , hopefully accessible in the near future accelerator experiments .
the measurement of the higgs mass will complete the list of the couplings of the standard model and , therefore , one should be able to study all the details of the reheating mechanism .
this makes the model under consideration extremely interesting and , potentially , predictive .
reheating in the context of scalar - tensor theories has been studied , in the hartree approximation by ref .
@xcite , and perturbatively by ref .
@xcite , but without the higgs boson playing the role of the inflaton , and therefore without an explicit coupling of the fundamental scalar field to matter . studying the details of reheating in this higgs - inflaton scenario ,
we have put special attention to the relative impact of the different mechanisms that can take place .
we have found that the universe does not reheat immediately through perturbative decays , but rather initiate a complex procces in which perturbative and non - perturbative effects are mixed .
the higgs condensate starts oscillating around the minimum of its potential , producing @xmath1 and @xmath0 gauge bosons due to violation of the so called adiabaticity condition . during each semi
- oscillation , the non - perturbatively created gauge bosons decay ( perturbatively ) into fermions .
this decay prevents the development of the usual parametric resonance , since bosons do not accumulate significantly at the beginning .
the energy transferred to the decay products of the bosons is not enough to reheat the universe within a few oscillations , and therefore the resonance effects will eventually dominate over the perturbative decays . around the same time , the backreaction from the gauge bosons into the higgs condensate will also start to be significant .
soon afterwards , the universe is filled with the remnant condensate of the higgs and a non - thermal distribution of fermions and bosons ( those of the sm ) , which redshift as radiation and matter , respectively .
we end the paper computing the distribution of the energy budget among all the species present at the time of backreaction . from there on until thermalization , the evolution of the system is highly non - linear and non - perturbative , and will require a careful study via numerical simulations , to be described in a future publication .
the paper is organized as follows . in section [ higgssector ]
we present the model with the higgs field non - minimally coupled to the scalar curvature , transform it into a new frame where the action takes the usual einstein - hilbert form , and derive an approximate inflationary potential . in section [ mattersector ]
we study the effect of the conformal transformation in the matter sector , including the interaction among the higgs , vector bosons and fermions .
section [ reheating ] is devoted to the analysis of the different reheating mechanisms , both perturbative and non - perturbative , that can take place , leaving for section [ stimulateddecays ] the analysis of the combined effect of parametric resonance and perturbative decays .
we then study the backreaction of the produced particles on the higgs oscillations and the end of preheating in section [ backreaction ] .
finally the conclusions are presented in section [ conclusions ] .
the glashow - weinberg - salam @xcite action is divided into four parts : a fermion sector @xmath11 which includes the kinetic terms for the fermions and their interaction with the gauge bosons , a gauge sector @xmath12 , including the kinetic terms for the intermediate bosons as well as the gauge fixing and faddeev - popov terms , a spontaneous symmetry breaking sector @xmath13 , with a higgs potential and the kinetic term for the higgs field including its interaction with the gauge fields , and finally , a yukawa sector @xmath14 , with the interaction among the higgs and the fermions of the standard model , @xmath15 the simplest versions of this lagrangian in curved spacetime follow the principles of general covariance and locality for both matter and gravitational sectors . to preserve the fundamental features of the original theory in flat space - time
, one must also require the gauge invariance and other symmetries in flat space - time to hold for the curved space - time theory .
the number of possible terms in the action is unbounded even in this case and some additional restrictions are needed .
a natural requirement could be renormalizability and simplicity .
following this three principles ( locality , covariance and restricted dimension ) , and the previously motivated requirement of not introducing new dynamical degrees of freedom , the form of the action is fixed , except for the values of some new parameters to be determined by the physics .
this procedure leads to the non - minimal lagrangian for the standard model in the presence of gravity , given by @xmath16 where @xmath17 is the standard model part ( [ sm ] ) defined above , and @xmath18 is the new higgs - gravity sector , given by eq .
( [ shg ] ) . here
@xmath19 is the reduced planck mass , @xmath20 the ricci scalar , @xmath21 the higgs field , and @xmath22 is the announced non - minimal coupling constant . as showed in ref .
@xcite , the parameters @xmath22 and the self - coupling @xmath23 of the higgs potential are related by @xmath24 . in the unitary gauge , @xmath25 , and neglecting all gauge interactions for the time being , the lagrangian for the higgs - gravity sector in the so - called jordan ( j ) frame takes the form @xmath26 \,,\ ] ] where @xmath27 , and @xmath28 is the usual higgs potential of the standard model , with vev @xmath29 gev . in order to get rid of the non - minimal coupling to gravity , we proceed as usual , performing a conformal transformation @xcite @xmath30 such that we obtain the lagrangian in the so - called einstein ( e ) frame @xmath31 -\frac{\tilde{{\partial_\mu } } h\tilde{\partial}^\mu h}{2\omega^2}-\frac{1}{\omega^4}u(h ) \bigg\}.\ ] ] the usual einstein - hilbert term can then be obtained imposing @xmath32 , which implies the following relation between the conformal transformation and the higgs field @xmath33 this allows us to write the lagrangian ( [ higgslagrangiane ] ) completely in terms of @xmath2 @xmath34 \tilde g^{\mu\nu}\partial_\mu h\ , \partial_\nu h - { 1\over \omega^4}u(h)\bigg\},\ ] ] where we have neglected a total derivative that does not contribute to the equations of motion . as we will be working in the einstein frame
from now on , we will skip over the tilde in all the variables to simplify the notation .
notice that the conformal transformation ( [ ct ] ) leads to a non - minimal kinetic term for the higgs field , which can be reduced to a canonical one by making the transformation @xmath35 where @xmath36 is a new scalar field .
doing this , the total action in the einstein frame , without taking into account the gauge interactions , is simply @xmath37 \,,\ ] ] with @xmath38 the potential in terms of the new field @xmath36 . to find the explicit form of the potential in this new variable @xmath36
, we must find the expresion of @xmath2 in terms of @xmath36 .
this can be done by integrating eq .
( [ relchih ] ) , whose general solution is given by @xmath39 where @xmath40 .
since @xmath41 , we can take @xmath42 and , using the identity @xmath43 for @xmath44 , we can approximate eq .
( [ sol ] ) by @xmath45 or , equivalently , @xmath46 where @xmath47 and @xmath48 .
the @xmath36 field is therefore directly related in this aproximation ( just in the limit @xmath49 and far from @xmath50 ) to the conformal transformation @xmath51 in a very simple way and the inflationary potential ( [ potentiale ] ) is just given by @xmath52 ^ 2e^{-2\alpha\kappa\chi}\;.\ ] ] since @xmath53 , then @xmath54 and we can savely ignore the _
vev _ for the evolution during inflation and preheating , and simply consider the potential @xmath55 ) , the analytic formula ( [ potentiale2 ] ) for the potential ( blue dashed line ) , and their parametrization ( [ potentiale3 ] ) ( green dotted line).,width=453 ] notice that the previous potential only parametrizes partially the original potential ( [ potentialj ] ) , since it neglects the region @xmath56 , as can be seen in fig.[comparison ] , where we compare the exact solution ( red continuous line ) obtained parametrically from eq . ( [ sol ] ) , with the analitic formula ( [ potentiale2 ] ) ( blue dashed line ) . both solutions agree very well in the region of positive @xmath36 but differ substancially for @xmath56 . the conformal transformation is even ill - defined in the negative field region . from eq .
( [ relwithomega ] ) and eq .
( [ relchih2 ] ) we have @xmath57 which is inconsistent , since the left - hand side of this equation is positive definite , while the right hand is negative definite for @xmath56 . taking this into account , in order to study the different mechanisms of the post - inflationary regime , we will then use the parametrization @xmath58 which correctly describes the potential obtained from eq .
( [ sol ] ) , for the whole field range of interest . in fig .
[ comparison ] , this parametrization ( green dotted line ) is again compared to the exact solution ( red continuous line ) obtained from eq .
( [ sol ] ) . around then minimum , the potential ( [ potentiale3 ] ) can be approximated as @xmath59 where @xmath60 is the typical frequency of oscillation and @xmath61 are some corrections to the quadratic approximation , which soon become negligible after inflation ends , see section [ reheating ] .
note , nevertheless , that approximations ( [ approxsol]),([relchih2 ] ) and therefore parametrization ( [ potentiale3 ] ) , do not describe correctly the potential for very small values of the field @xmath62 . as can be read from eq .
( [ relchih ] ) , for @xmath63 , we have @xmath64 and therefore , there is a transition in the potential from ( [ potentialeapprox ] ) to @xmath65 .
however , as will be shown in section [ parametricreheating ] , the transition region , @xmath66 , is several orders of magnitude smaller than the non - adiabaticity region , @xmath67 , see eq .
( [ chia ] ) , inside which the concept of particle is not properly defined .
therefore , from now on we will neglect the change in the behaviour of the potential ( from @xmath68 to @xmath69 ) in this `` small '' field region , since @xmath70 .
see section [ reheating ] for more details .
the analysis of the inflationary potential ( [ potentiale3 ] ) can be performed either in the jordan @xcite or in the einstein frame @xcite with the same result .
the slow roll parameters can be expressed analytically , in the limit of @xmath71 , as a function of @xmath36 , @xmath72 the slow roll regime of inflation will end when @xmath73 , which correspond to the field value @xmath74 note that the slow roll parameter @xmath75 is then negative , @xmath76 , so there is a small region of negative ( mass squared ) curvature in the potential just after the end of inflation .
the effective curvature of the potential will be negative until @xmath77 , which corresponds to the inflection point , given by @xmath78 . during the slow - roll regime @xmath79 , which evaluated at @xmath80 e - folds
is approximately given by @xmath81 , where @xmath82 defines the natural inflationary energy scale of this model as well as the frequency of oscillations ( [ potentialeapprox ] ) during reheating .
at the end of inflation @xmath83 or , equivalently , @xmath84 .
the radiative corrections for a model containing a single scalar field @xmath2 non - minimally coupled to gravity were generically calculated in ref .
the specific radiative corrections for the model under consideration , estimated in ref .
@xcite , were recently reviewed in ref .
@xcite . in
what follows until the end of this section , we will summarize the results of this last work . for large @xmath22 and slow varying @xmath2 the main contribution comes from loops of the matter @xcite and the effective action can be calculated by a local @xmath85-expansion in powers of the curvature and its gradients and the gradients of the higgs field , obtaining @xcite @xmath86=\int d^{4}x\,g^{1/2}\left(-u(h)+f(h)\,r(g_{\mu\nu } ) -
\frac12\,g(h)\,(\nabla h)^2\right)\,,\label{effectiveaction}\ ] ] where the functions @xmath87 , @xmath88 and @xmath89 are given by @xmath90 here @xmath91 and @xmath92 are different combinations of the higgs , gauge and yukawa couplings and their logarithms @xcite , and @xmath93 is the normalization scale . following ref .
@xcite , for the analysis of inflation we will just consider the explict form of the combination @xmath94 , @xmath95 which is related with the local conformal anomaly .
the factor @xmath96 accounts for the polarizations of the gauge bosons and a similar factor @xmath97 for the fermions has been taken into account .
the new inflationary potential in the einstein frame for a higgs field with non - minimal coupling @xmath98 and mean value much greater than the minimum of the classical potential , is given by @xcite @xmath99 the corresponding slow roll parameters are @xmath100 with @xmath101 . using the wmap+bao+sn constraint @xcite at the @xmath102 confidence level
gives a value for the spectral index @xcite @xmath103 for @xmath104 the value for @xmath94 and for the tensor - to - scalar ratio is @xcite @xmath105 being the spectral scalar index running completely negligible @xcite @xmath106 let us compare this window with the one obtained from the standard model coupling constants at the scale @xmath82 , and obtain the higgs self - coupling at such scale .
this differs from the analysis performed in @xcite , where the coupling constants were evaluated at the electroweak scale instead of at the characteristic energy scale of inflation , @xmath82 . at that scale the coupling constants for the gauge groups @xmath107 and @xmath108 are roughly equal @xmath109 and @xmath110 .
the total anomalous scaling constant at that scale , see eq .
( [ scalinga ] ) , in terms of @xmath111 and @xmath112 , @xmath113\ ; , \label{asm}\end{aligned}\ ] ] together with the bounds obtained from the wmap+bao+sn @xmath102 c.l .
constraints @xcite , see eq .
( [ rangoa ] ) , give us the following range for the self - coupling of the higgs field at the scale @xmath82 , @xmath114 this range can be propagated back to obtain a value at the scale @xmath115 through the renomalization group equations , @xmath116 note that we have neglected the effect of the gauge and yukawa couplings , since the complete solution of the renormalization group equations is out of the scope of this preliminary study .
if we integrate this equation and take into account the present observational bounds @xcite , we obtain a higgs mass in the range @xmath117 safely within the detection range of the large hadron collider at cern . in the absence of an actual measurement of the higgs self - coupling @xmath23 , for the analysis of sections [ reheating ] and thereafter , we will just take different values compatible with the above range ( [ higgsrange ] ) .
the length scales are conventionally defined in such a way that elementary particle masses are the same for all times and in all places .
this implies for instance that if under a conformal transformation the lagrangian of a free particle transforms as @xmath118 the mass should be accordingly redefined as @xmath119 to express it in the new system of units .
the previous argument applies also for classical fields @xcite .
the rescaling of all fields ( including the metric tensor ) with an arbitrary space - time dependent factor @xmath51 , taken with a proper conformal weight for each field , will leave the physics unaffected .
the physical interpretation of this symmetry is clear : it changes all dimensional quantities ( lengths , masses , etc . ) in every point of the space time leaving their ratios unchanged . in this section
we will apply the previous prescription to the different sectors of the action ( [ sm ] ) .
consider for instance the spontaneous symmetry breaking sector , responsible of the masses of the intermediate gauge bosons @xmath0 and @xmath1 @xmath120 in the standard model , the masses of the @xmath121 bosons are due to a spontaneous symmetry breaking mechanism , realized by the constant _ vev _ of the higgs field , and therefore are constant . in our case
the higgs field evolves with time giving rise to variable effective masses for the gauge bosons @xmath122 with @xmath123 the weinberg angle defined as @xmath124 and @xmath125 and @xmath111 are the coupling constants corresponding to @xmath108 and @xmath107 at the scale @xmath82 , where the relevant physical processes during preheating will take place .
as mentioned before , numerically this correspond to a value @xmath126 , which implies @xmath127 . from now on we will use these values for numerical estimations . in agreement with the above prescription for transforming masses and fields
, the action ( [ ssbj ] ) preserves its form under the conformal transformation @xmath128 providing that we redefine the fields and masses with the corresponding conformal weights as @xmath129 the same can be applied to the interactions between fermions and gauge bosons .
let us consider , for instance , @xmath130 where @xmath131 are the charged currents carrying the information about the couplings of the @xmath132 to the standard model fermions , and @xmath133 is the neutral current with the information of the couplings of the @xmath1 boson . in the einstein
frame the action ( [ f ] ) preserves its form @xmath134 as long as we redefine the currents as @xmath135 which is equivalent to redefine the dirac fields @xmath136 finally , concerning the yukawa sector , we have , for a given family of the quark sector , @xmath137 where @xmath138 and @xmath139 stand for down- and up - type quarks respectively .
the effective masses in the jordan frame , @xmath140 , become @xmath141 in the einstein frame . + on the other hand , the total decay widths , summing over all the allowed decay channels in the standard model of the @xmath142 and @xmath1 bosons into any pair of fermions and over all the polarizations of the gauge bosons , are given respectively by @xcite @xmath143 where lips denotes the _ lorentz invariant phase - space _
factors @xmath144 the decay rates will preserve their functional form in the einstein frame , being only changed through the conformal transformation of the masses , i.e. @xmath145 where @xmath146 , are the dynamical masses in the einstein frame , see eq .
( [ redefmaswz ] ) .
in this section we will analize the different mechanisms that could give rise to efficient reheating of the universe , both perturbative and non - perturbative .
the natural mechanism , given the strength of the interactions of the higgs boson with the standard model particles , would be a perturbative reheating process right after the end of slow - roll .
however , as we will see , perturbative reheating is not efficient enough , and non - perturbative effects must be taken into account .
given the shape of the potential ( [ potentiale3 ] ) , very different ( p)reheating mechanisms could in principle take place , from a tachyonic production in the region between the end of inflation and the inflection point @xcite , an instant preheating mechanism @xcite and a parametric resonance effect around the minimum of the potential @xcite . therefore , it will crucial to disantangle the contribution of each mechanism and quantify their relative importance . in order to study the different reheating mechanisms it will be useful to have an approximate expression for the evolution of the inflaton . as mentioned before ,
we can expand potential ( [ potentiale3 ] ) for the range of interest , as @xmath147 with @xmath60 the typical frequency of oscillation .
the first terms of the corrections @xmath61 to the quadratic potential are given explicitely by @xmath148 with @xmath149 and @xmath150 .
the klein - gordon equation for the inflaton , @xmath151 can then be written , for a power - law evolution @xmath152 , as @xmath153\chi = 0\,,\ ] ] where we have neglected the higgs interactions with other fields , because they are proportional to the number density of particles of a given species and therefore they will be negligible ( see section [ backreaction ] ) during the first oscillations of the higgs field . the backreaction of other particles into the dynamics of higgs will only be relevant once their occupations numbers have grown sufficiently .
the non - linear terms of the higgs self - interaction , described by @xmath154 will be also negligible from the very beginning of reheating , @xmath155 , as we will justify _ a posteriori_. thus , neglecting such a term in the effective equation of @xmath36 , the general solution can be expressed as @xmath156\,,\ ] ] with @xmath94 and @xmath157 constants depending on the initial conditions ( end of inflation ) , and @xmath158 bessel functions of order @xmath159 , with @xmath160 . for a reasonable power index , @xmath161 for matter @xmath162 , while for radiation @xmath163 the second term in the right - hand side of eq .
( [ solutionbessel ] ) diverges in the limit @xmath164 and therefore should be discarded on physical grounds .
the physical solution is then simply given by @xmath165 which making use of the large argument expansion ( @xmath166 ) of fractional bessel functions @xcite , can be approximated by a cosinusoidal function @xmath167 the normalization constant @xmath94 can be fixed if we consider that the oscillatory behaviour starts just at the end of inflation , i.e. @xmath168 ( [ chiend ] ) . in this case , @xmath169 where we have made use of the limit of the bessel functions when @xmath170 @xmath171 the energy and pressure densities associated to the general solution ( [ physsolution ] ) are given , after averaging over several oscillations , by @xmath172 = \frac{1}{2}m^2x^2\ ; , \\
p_\chi \approx\left\langle\frac{1}{2}\dot\chi^2 - \frac{1}{2}m^2\chi^2\right\rangle & \approx & \frac{1}{2}m^2x^2[\left\langle\cos^2(mt-3\pi p/4)\right\rangle -\left\langle\sin^2(mt-3\pi p/4)\right\rangle ] = 0\ ; , \end{aligned}\ ] ] with @xmath173 .
since the averaged pressure is negligible @xmath174 , then @xmath175 with @xmath176 .
using this fact , the physical solution is finally expressed as @xmath177 rewriting the previous equation in terms of the number of times the inflaton crosses zero , @xmath178 , or equivalently in terms of the number of oscillations @xmath179 , then @xmath180 therefore , the higgs condensate oscillates with a decreasing amplitude @xmath181 .
we can obtain an upper bound , @xmath182 , on the amplitude of the higgs field after @xmath183 oscillations , which in terms of the correction of @xmath184 to 1 , see eq .
( [ eqnonlinearhiggs ] ) , implies @xmath185 or @xmath186 , after the first @xmath187 and @xmath188 oscillations , respectively .
thus , from the very beginning , the effective potential of the higgs field tends very rapidly to that of a harmonic oscillator , which justifies _ a posteriori _ the approximation @xmath189 used in the derivation of eqs .
( [ solutionbessel ] ) and ( [ chit ] ) .
note that if we neglect the presence of other fields and consider the higgs - condensate as a free field only damped by the expansion rate , then we can easily estimate the number of semi - oscillations before the amplitude of the field becomes smaller than the transition value @xmath190 , defined in section [ higgssector ] . for @xmath66
, the higgs potential will not be anymore approximated by eq .
( [ potentialeapproxii ] ) , but rather by a quartic form @xmath191 .
this will happen when @xmath192 , which implies @xmath193 therefore , if before such a moment , the higgs field has not yet transferred efficiently its energy into other fields , then after @xmath194 semi - oscillations , the transition in the behaviour of the potential will also imply a change in the expansion rate , from a matter - like , characteristic of quadratic potentials , to radiation - like , characteristic of quartic potentials .
a natural reheating mechanism , given the strength of the interactions of the higgs boson with the standard model particles , would be a perturbative decay process of the inflaton quanta into the standard model particles right after the end of slow - roll .
as we saw , soon after the end of inflation the effective potential for the higgs field can be approximated by a simple quadratic potential ( [ potentialeapprox ] ) . in this approximation ,
the masses of fermions and gauge bosons , see eqs .
( [ redefmaswz ] ) and ( [ redefmassfermion ] ) , are simply given by @xmath195 in order to have a perturbative decay two conditions must be fulfilled : + 1 ) there should be enough phase - space in the final states for the higgs field to decay , i.e. @xmath196 , which will only happen when the amplitude of the higgs field becomes smaller than a certain critial value @xmath197 . in particular , for a decay into gauge bosons and/or fermions , in the light of eq .
( [ massesapprox ] ) , one needs @xmath198 where @xmath199 for the @xmath0 and @xmath1 bosons , respectively , and @xmath200 for the fermions .
when compared to the initial amplitude ( [ chiend ] ) of the higgs field at the end of inflation , @xmath201 , we see that this critical value is much smaller than @xmath202 for the gauge bosons and the top quark ; of the same order for the bottom and charm quarks , and even greater for the rest of the quarks and the sm leptons . + 2 ) the higgs decay rate @xmath203 has to be greater than the rate of expansion @xmath204 , where we have used eq .
( [ energyinflaton ] ) .
such a condition , @xmath205 , can be translated into the following inequality @xmath206 which defines the critical number of semi - oscillations required for this second condition to be true ( again , @xmath207 and @xmath208 , for @xmath0 , @xmath1 bosons and fermions , respectively ) . + the critical amplitude ( [ chicritical ] ) below which the higgs is allowed to decay into gauge bosons is of order @xmath209 . as mentioned before
, this amplitude is much smaller than that of the higgs at the end of inflation , @xmath210 .
therefore , the higgs condensate would need to oscillate @xmath211 times before being able to decay through this channel .
the same applies to the top quark . in the case of other fermions , due to the wide range of the yukawa couplings
, several situations can take place .
for instance , the decay channel into bottom and charm quarks is opened only after a few oscillations of the higgs , while for the rest of quarks and leptons , the decay - channel has sufficient phase space from the very end of inflation . in general , the smaller the yukawa coupling of a given fermion species to the higgs , the less oscillations the higgs will go through before there is enough phase - space for it to decay into such fermion species .
notice however that the smallness of the yukawa coupling implies also a smaller decay rate .
consider for instance the decay of the higgs into electrons , whose yukawa coupling is of order @xmath212 . from the very end of inflation , see eq .
( [ chicritical ] ) , there is phase - space in this channel for the higgs to decay into .
however , it is precisely the smallness of the electron s yukawa coupling that allows the decay to be possible , which prevents the condition 2 ) to be fulfilled .
the decay width is much smaller than the hubble rate for a huge number of oscillations .
looking at eq .
( [ ncritic ] ) , we realize that the higgs condensate should oscillate @xmath213 times before the decay rate into electrons overtakes the hubble rate .
one can check that the previous conclusions also hold for the rest of fermions of the standard model .
when there is phase - space for the higgs to decay into a given species , the decay rate does not catch up with the expansion rate and , viceversa , if the decay rate of a given species overtakes the expansion rate , there is no phase - space for the decay to happen ) ( which prevents higgs decay into gauge bosons and top quarks ) assumes an average amplitude over a single higgs oscillation , while smaller values are attained around the minimum of the potential when @xmath214 .
however , when this happens the higgs field is well inside the non - adiabatic range @xmath67 ( [ chia2 ] ) , in which the very concept of particle ( gauge bosons and top quark ) is not properly defined , see section [ parametricreheating ] . ] .
therefore , during a large number of oscillations , the higgs field is not allowed to decay perturbatively in any of the standard model fields . moreover , before any of those decay channels is opened , many other interesting ( non - perturbative effects ) will take place , as we will describe in detail in the next sections . as we pointed out in section [ higgssector ] ,
the effective square mass of the higgs field @xmath36 is negative just after the end of inflation and will be so till the inflection point .
when this happens spinoidal instability takes place @xcite and long wavelengths quantum fluctuations @xmath215 , with momenta @xmath216 , grow exponentially .
the width of the tachyonic band will be limited in our case by the point of maximum particle production , the end of inflation . at this point
the effective mass @xmath217 takes a value @xmath218 which corresponds to a maximum momentum for the tachyonic band @xmath219 .
this comes from vacuum quantum fluctuations , @xmath220 , which grow exponentially . however , since the inflaton is fast rolling down the potential towards the positive curvature region , the duration of the tachyonic preheating stage is so short that the occupation numbers of those modes in the band do not grow significantly and the effect can be neglected . in particular ,
the time interval from the end of inflation till the inflection point is just @xmath221 and therefore , even for the fastest growing mode , @xmath222 ,
its growth is only @xmath223 .
this is a negligible effect and thus , one can still consider an initial spectrum of quantum vacuum fluctuations even at the inflection point . for simplicity , all our analitycal estimations have been done ignoring this period of tachyonic instability , taking as initial conditions at the end of inflation , the amplitude of the higgs condensate @xmath224 ( [ chiend ] ) and quantum vacuum fluctuations .
+ another physical effect before the higgs condensate reaches the bottom of the potential for the first time , will be the particle production in the inflection point , due to the violation of the adiabaticity condition , @xmath225 where the frequency of oscillation of the fluctuations is @xmath226 .
differenciating this and rewriting the adiabaticity condition as @xmath227 , we find that only those modes within the band @xmath228 are amplified . at the end of inflation @xmath229 which implies @xmath230 .
extrapolating the previous formula to the inflection point ( ip ) we get @xmath231 which indeed seems to be a very good approximation if we compare it with the result of a numerical solution beyond slow - roll . inserting ( [ chidot ] ) into eq .
( [ kinf ] ) we get @xmath232 which corresponds to a maximum excited wave number at the inflection point given by @xmath233 . again , here the time of production is so brief that the occupation number of modes within the band is not significantly enhanced .
we will have to wait until the next stage of consecutive oscillations of the higgs field around zero for a significant production of particles . during each oscillation of the higgs field
@xmath36 , the rest of the quantum fields that couple to it will oscillate many times .
consider for instance the interaction of the higgs field with the @xmath1 bosons around the minimum of the potential . in this region the associated action ( [ ssbe ] )
can be approximated by a trilinear interaction where the masses of the @xmath0 and @xmath1 bosons ( [ redefmaswz ] ) are given by @xmath234 which are much greater than the inflaton mass @xmath82 for the main part of the oscillation of @xmath36 . as a result ,
the typical frequency of oscillation of the gauge boson is much higher than the one of the higgs field @xmath36 .
this implies that during most of the time the effective masses of the intermediate boson are changing adiabatically and an adiabatic invariant can be defined : the number of particles .
however , for values of @xmath36 very close to zero , the adiabaticity conditions @xmath235 are violated .
in such a case , there will be an inequivalence between the vacua before and after the passage of @xmath36 through the minimum of the potential , which can be interpreted as particle production @xcite . in terms of the field
@xmath36 , the violation of ( [ adiabaticity2 ] ) corresponds to the region @xmath236 , @xmath237 where , from now on , @xmath238 for the @xmath0 or @xmath1 bosons . only outside this region , the notion of particle makes sense and an adiabatic invariants can be defined . taking into account that and approximating the velocity of the field around zero as @xmath239 , see eq .
( [ chit ] ) , the general expressions ( [ chia ] ) can be approximated as @xmath240 note that the previous regions are indeed very narrow compared to the amplitude of the oscillating higgs , @xmath241 .
therefore , the particle production that takes place in that region happens within a very short period of time as compared to the inflatons oscillation period @xmath242 , @xmath243 notice indeed that different values of @xmath23 do not change appreciably the above conclusions about the smallness of the non - adiabatic regions .
given the weak dependence of @xmath244 , many semi - oscillations ( @xmath245 ) will pass before the fraction of time spent in the non - adiabatic zone will increase from a 1% to a 10% , as compared with the period of oscillations .
this holds also independently of the species , @xmath0 or @xmath1 bosons .
moreover , despite the smallness of @xmath246 as compared to the amplitude @xmath247 , it is important to note that the field range corresponding to the region of non - adiabaticity still is much greater than those critical regions defined in section [ higgssector ] . in particular , let us recall that there is a field value , @xmath248 , below which there is a transition of the effective potential from a quadratic to quartic behaviour .
however , this is well inside the region of non - adiabaticity , @xmath70 , as we emphasized before .
moreover , there is also an interval of higgs field values , @xmath249 for which the higgs perturbative decay into @xmath0 , @xmath1 and top quarks can occur [ see eq .
( [ chicritical ] ) in section [ perturbativereheating ] ] , which nevertheless is also much smaller than the non - adiabaticity interval , @xmath250 .
we will now discuss the non - perturbative creation of particles in the non - adiabatic region .
this production is formally equivalent to the quantum mechanical problem of a particle scattering in a periodic potential . in the case under consideration
the equations of motion for the fluctuations of each gauge field with a given polarization will be given by @xmath251 , and the corresponding one for the @xmath1-fluctuations .
expanding eq .
( [ chit ] ) around the @xmath252-th zero at time @xmath253 , the evolution equation of the fluctuations can be approximated as @xmath254 notice that around the zeros of the inflaton the sinusoidal behaviour @xmath255 can be very well approximated by @xmath256 , which allows us to rewrite eqs .
( [ fluctuations ] ) and ( [ fluctuations2 ] ) as a schrdinger - like equation like @xmath257 where primes denote derivatives with respect to the rescaled time @xmath258 , @xmath259 is the rescaled momentum @xmath260 and @xmath261 are the usual resonance parameters @xcite . each time the inflaton crosses zero can be interpreted therefore as the quantum mechanical scattering problem of a particle crossing an inverted triangular potential .
let @xmath262 and @xmath263 , for either @xmath0 or @xmath1 , be the transmission and reflection probabilities for a single scattering in this periodic triangular barrier .
the number of particles just after the @xmath252-th scattering , @xmath264 , in terms of the previous number of particles @xmath265 just before that scattering , can be written as @xcite @xmath266 where @xmath267 are some accummulated phases at each scattering , that we will discuss later on in section [ stimulateddecays ] , since they will not play any role in the following discussion of this section .
the inverse of the transmission probability for the @xmath252-th scattering can be expressed as @xcite @xmath268 ^ 2\,,\end{aligned}\ ] ] with @xmath269 and @xmath270 the airy functions .
note that we have used the wronskian normalization , @xmath271 . ) for the gauge bosons created in a single zero crossing through the first term of eq .
( [ occupnum ] ) , calculated after @xmath272 and @xmath273 oscilations ( from left to right ) .
the horizontal axis represents @xmath274 , so @xmath275 is the typical width of the band of momenta of particles created at the first scattering . for later times , the distributions broaden out to greater momenta , since the argument of eq .
( [ deltan ] ) , @xmath276 behaves as @xmath277 .
the typical momenta of the distribution agree with the one calculated in section [ parametricreheating].,width=302,height=226 ] consider the situation , @xmath278 , which is certainly true in the first scattering @xmath279 , or can happen for @xmath280 if the previously produced gauge bosons have fully decayed into fermions .
in such a case , @xmath281 where we have retained only the first term of eq .
( [ occupnum ] ) .
this corresponds to the spontaneous particle creation of @xmath0 and @xmath1 bosons each time the higgs crosses zero and , therefore , tells us about the number of particles of these species that are created in each zero - crossing .
the momenta distribution is shown in fig .
[ triangulart ] . in particular , the total number of produced particles of a given species with a given polarization , just after exiting the non - adiabatic region around the @xmath252-th zero - crossing , can be obtained as @xmath282 = \frac{q}{2j}\,\mathcal{i}\,m^3\,,\ ] ] with @xmath283 ^ 2x^2dx \approx 0.0046 $ ] and @xmath284 the resonant parameters given by eq .
( [ qfactor ] ) .
thus , the only difference between the number of @xmath0 and @xmath1 bosons produced is simply encoded in the different resonance parameter , @xmath285 and @xmath286 , respectively . notice that the effect of the non - perturbative production of these particles is proportional to the coupling square .
if the couplings of the higgs field to the gauge bosons were not so large ( @xmath287 ) , then their produccion would be very supressed . strictly speaking the previous analysis
is just valid for gauge bosons .
the production of fermions through this mechanism is different and more involved than for bosons .
nevertheless , if the effect is , as expected , proportional to the yukawa coupling squared @xcite then only the top quark production would be non - negligible .
after the passage through the minimum of the effective potential the number of @xmath0 and @xmath1 particles remains ( almost ) constant while their masses grow when the field @xmath36 increases .
the @xmath0 and @xmath1 bosons tend to decay into fermions in a time @xmath288 , where @xmath289 are given by eq .
( [ widthwe ] ) , while @xmath290 represents a time average between the @xmath252- and the @xmath291-th scatterings .
given the time - dependence of the @xmath36 field ( [ chit ] ) , the typical time of decay turns out to be @xmath292 for the @xmath1 bosons and a bit bigger ,
@xmath293 , for the @xmath0 bosons , as was expected .
this implies that in a semiperiod , @xmath294 , the non - perturbatively produced gauge bosons at the @xmath252-th scattering , decay significantly before the next scattering takes place , at least for the first scatterings .
however , as the amplitude of the higgs field decreases with time due to the expansion of the universe , the probability of decay of the gauge bosons , see eq .
( [ widthwe ] ) , becomes smaller and smaller as time goes by .
this explains the @xmath295 behaviour , which essentially means that after a certain number of oscillations , the number of produced fermions through the perturbative decays per semi - oscillation , will become eventually negligible . in the jordan frame the standard model presents its usual form and
the fermions produced in the decay of the @xmath1 and @xmath0 bosons are mainly relativistic .
since both momenta and masses transform in the same way under a change of conformal frame , if a gauge boson is allowed to decay into a pair of fermions in the jordan frame , it will also be able to decay in the einstein frame .
therefore , the relation between the typical momenta and masses of those fermions ( f ) and gauge bosons ( w , z ) in the conformally transformed frame , is simply given by @xmath296 in terms of the field @xmath36 , the previous equations can be rewritten as @xmath297 for the @xmath0 and @xmath1 fields respectively .
note that the relativistic or non - relativistic nature of a given particle is something intrinsic to the particle and should not depend on the conformal frame .
as expected the transitions @xmath298 , @xmath299 are not allowed in the einstein frame . for the rest of quarks @xmath300 , which implies that all the fermions produced in the decay of the @xmath0 and @xmath1 bosons are clearly relativistic , as happened in the jordan frame .
the total number density of gauge bosons @xmath301 present just after the @xmath252-th crossing will decay exponentially fast until the next crossing , due to the perturbative decay into fermions .
therefore the total number density just previous to the @xmath291-th zero crossing , @xmath302 , is given by @xmath303 the number of fermions produced between those two scatterings , @xmath304 , is simply given by @xmath305\;,\end{aligned}\ ] ] where the factor @xmath306 takes into account that each gauge boson can have one out of three polarizations and decay into two fermions , while the extra factor @xmath307 in front @xmath308 , accounts both for the @xmath309 and @xmath310 decays .
the averaged value of the decay widths in the previous expressions can be estimated , see eqs .
( [ widthwe ] ) , as @xmath311 where @xmath242 is the typical oscillation period and we have defined @xmath312^{\frac{1}{2}}}\right ) \approx 0.3423\frac{1}{\sqrt{j}}\,.\end{aligned}\ ] ] note that the last approximated equality is simply a ( good ) fit to @xmath313 for all @xmath252 .
the constants @xmath314 are just numerical factors depending of the parameters of the model and the decaying species , @xmath315 using the notation , @xmath316 and @xmath317 for the mean energy of the fermions produced between @xmath318 and @xmath319 , from the decay of @xmath1 or @xmath0 bosons , respectively , then we find @xmath320 where we have used the fact that the produced fermions are relativistic , see eqs ( [ deltawe ] ) and ( [ deltaze ] ) , while the gauge bosons are non - relativistic , see eq .
( [ gaugebosonsmomenta ] ) in section [ stimulateddecays ] .
+ let us work now under the following hypothesis : we will consider that the perturbative decay of the gauge bosons into fermions is sufficiently effective , such that the gauge bosons do not accumulate significantly . this amounts to neglect initially a potential effect of parametric resonance .
this is of course a rough approximation , which is only valid for the first oscillations , where @xmath321 .
numerically , after @xmath322 and @xmath323 zero - crossings , the 99.5% , 98.5% , 94.2% , 87.4% , 81.9% , 77.4% respectively , of the produced @xmath1 particles have decayed into fermions ( and a similar though smaller fraction of the @xmath0 bosons ) .
this implies that there will be always a remnant of the gauge bosons produced at each scattering , that will not decay in one semi - period of the inflaton s oscillation .
let us neglect this for the time being , therefore ignoring the possibility of having parametric resonance , and estimate the energy transferred simply through the perturbative decay into fermions , during the first oscillations
. in particular , the energy density of those fermions produced after the first scattering , averaged over the first semi - oscilation between @xmath324 and @xmath325 , will be @xmath326 = \varepsilon\left(\frac{1}{2}m^2\chi_{\rm end}^2\frac{1}{\pi^2}\right)\ , f(1)\,,\end{aligned}\ ] ] with @xmath327 the energy density of the inflaton , evaluated at the maximum amplitude of the first semi - oscillation , is given ( [ energyinflaton ] ) by @xmath328 therefore , the ratio between the energy density of the fermions and of the inflaton , at that moment , @xmath329 , is @xmath330 which means that initially , for e.g. for @xmath331 , only a @xmath332 of the inflaton s energy has been transferred to the fermions .
thus , the so called instant preheating mechanism ( @xcite ) results frustrated here , because in order to make it work efficiently , the couplings of the theory must be really fine - tuned , in such a way that a significant fraction of the energy of the inflaton was transferred ( in the first semi - oscillation ) to the decay products of the bosons to which the inflaton is coupled .
moreover , in the instant preheating scenario , the produced fermions must be non - relativistic while the effective behaviour of the background inflaton should be effectively mimicing that of relativistic matter ( like @xmath333 in @xmath334 models ) . only in this case
it would be guaranteed that the remnant energy of the inflaton would decay faster that that of the fermions , thanks to the extra supression factor @xmath335 due to the expansion of the universe .
if the inflaton would effectively behave as non - relativistic matter and the produced fermions were relativistic , the energy of the inflaton could again overtake very soon that of the fermions , because now the fermion s energy would decrease faster than that of the background .
that is , precisely , the situation we have in the scenario under discussion . even if we had found that @xmath336 , the relativistic nature of the fermions and the non - relativistic effective behaviour of the higgs oscillations , would have prevented the universe to instantaneously reheat at that point .
one could hope that after a certain number of oscillations , let us say @xmath337 , that ratio would grow up to a value @xmath338 .
the succesively produced fermions , generated each semi - oscillation through the perturbative decay of the ( non - perturbatively produced ) @xmath0 and @xmath1 bosons , could perhaps accumulate a sufficiently ammount of energy that could finally equal that of the higgs condensate .
this seems not totally unresonable because the total energy stored in the higgs decreases with the expansion of the universe as @xmath339 , see eq .
( [ energyinflaton ] ) , such that the total amount of energy that we would require to transfer to the fermions would be less and less .
moreover , the number of fermions would only increase as time goes on , so one keeps adding energy each semi - oscillation in the form of new produced fermions .
thus , these two effects would contribute to the increment of the ratio of the energy between the fermions and the higgs .
on the other hand , the relativistic nature of the fermions and the decrease of their production rate with the expansion , as @xmath340 , would tend to decrease such a ratio .
therefore , one must put together all these competing effects in order to obtain the evolution in time of the energy transferred from the higgs to the fermions . to do this ,
we will assume , both for simplicity as well as for trying to make this mechanism more efficient , that since the gauge bosons do not accumulate significantly , only the first term of eq .
( [ occupnum ] ) should be considered , such that @xmath0 and @xmath1 bosons are only produced through spontaneous creation at each zero - crossing . in this case
the averaged energy density of the fermions produced between @xmath318 and @xmath319 will be given by @xmath341 = \varepsilon\left(\frac{1}{2}m^2\chi_{\rm end}^2\frac{1}{\pi^2}\right)\,\frac{f(j)}{j}\,\upsilon(j)\,,\ ] ] where @xmath342 is given by eq .
( [ factor ] ) and we have defined @xmath343 then the ratio between the energy of the fermions to the higgs condensate at the @xmath252-th zero crossing , finally reads @xmath344 where @xmath345 note that the strength of the effect , i.e. the amplitude of @xmath346 , is modulated by the gauge couplings through @xmath347 so , even in this case in which the sm gauge couplings of the higgs to the vector bosons are quite big ( @xmath348 ) , that does not help to transfer sufficient energy initially . as mentioned before , if we could apply this formalism to the production of fermions at each higgs zero crossing , by substituying the gauge couplings with the yukawa ones , we would obtain even a more ridiculous production of particles ( except perhaps for the top quarks ) . of course , the question of fermionic preheating at each zero crossing deserves more investigation and we will address it in a future publication .
the numerical values of the ratio @xmath349 after e.g. @xmath350 and @xmath323 semi - oscillations , for e.g. for @xmath331 , are respectively @xmath351 \sim 3.90 , 5.97 , 11.82 , 37.26 , 65.04 $ ] and @xmath352 . for different values of @xmath23 , these numbers do not change significantly .
thus , we see that the transferred energy from the higgs field to the fermions through the gauge bosons is generically a very slowly growing function .
after @xmath323 crossings the transferred energy is still only @xmath353 of the higgs energy at that time .
therefore , we clearly see that this _ succesive instant preheating _ mechanism is not efficient enough as to rapidly reheat the universe .
if we consider that the former formalism ( [ succesiveinstant ] ) is valid up to an arbitrary number of oscillations , then we can estimate the number @xmath337 of semi - oscllations required to achieve @xmath354 .
equating @xmath346 to 1 in eq .
( [ succesiveinstant ] ) , we obtain @xmath355 .
however , much earlier than that , parametric resonance effects should be considered , see section [ stimulateddecays ] . in other words ,
notice that we have neglected the presence of those @xmath1 and @xmath0 bosons that did not decay into fermions in each semi - oscillation .
the occupation number of the bosons produced at the bottom of the potential is not simply generated by the first term of eq .
( [ occupnum ] ) , but rather by the rest of the terms in eq .
( [ occupnum ] ) , which indeed give rise to the phenomena of resonant production of bosons . taking this into account
will have very interesting consequences .
the number density of the bosonic species will grow exponentially fast and thus will also transfer energy into the fermions exponentially rapidly .
we must therefore develop a mixed formalism that takes into account the two competing effects : that of parametric resonant production of bosons versus the effect of their perturbative decay into fermions . it is crucial to note that while the perturbative decay does not transfer enough energy ( as we have just seen ) , the fact that those bosons disappear will have very important consequences for the development of the resonant effect .
in particular , the resonance will not become effective from the beginning of the oscillations of the inflaton right after inflation , as usually assumed , but only after the inflaton has already performed a significant number of oscillations .
let us now analyze how the occupation number of the @xmath1 and @xmath0 bosons grow if we consider the effect of all the terms in eq . ( [ occupnum ] ) .
the production of gauge bosons will also occur , as before , in a very short interval of time ( [ timea ] ) when the higgs condensate crosses around zero , violating then the adiabaticity conditions ( [ chia2 ] ) , @xmath67 . in the large occupation limit @xmath356 , the first term in eq .
( [ occupnum ] ) can be neglected and therefore the spectra number density of the produced gauge bosons just after the @xmath252-th scattering is given by @xmath357 which indicates the spectral number density @xmath358 just after the @xmath252-th scattering , in terms of the spectral number density @xmath359 just before such scattering .
since the interval between successive scatterings is @xmath360 , we can define naturally a growth ( floquet ) index @xmath361 as @xcite @xmath362 comparing formulas , we obtain @xmath363 the @xmath267 are some accummulated phases at the @xmath252-th scattering , which can indeed play a very important role , since they can enhance ( @xmath364 ) or decrease ( @xmath365 ) the effect of production of particles at each scattering .
depending on the phases , we can consider the following cases : the typical behaviour of the floquet index , for @xmath366 , @xmath367 the maximum index , achieved for @xmath368 , given by @xmath369 and the average index over an oscillation , obtained as @xmath370 all these possibilities are shown in fig .
[ mubehaviour ] , as a function of @xmath371 , where @xmath284 are the resonant parameters ( [ qfactor ] ) for the @xmath1 and @xmath0 bosons , while @xmath276 is the natural argument of the transmission probability scattering functions ( [ transmision ] ) . as explained in ref .
@xcite , when @xmath372 , the effect of resonance will be chaotic , being then the phases essentially random at each scattering .
for instance , using the effective frequencies of the fluctuations ( [ fluctuations ] ) of the w field , these phases can be estimated , for the relevant range of momenta , as follows @xmath373 where @xmath313 was defined in eq .
( [ fn ] ) and we have neglected @xmath374 versus @xmath375 in the second equality , since , as will be justified later ( [ gaugebosonsmomenta ] ) , the produced bosons are non - relativistic . comparing the above formula with @xmath376 , we see that the end of the stochastic behaviour will occur after @xmath377 zero crossings , depending on @xmath23 . for the case of the @xmath1 boson the previous estimation of the end of the stochastic resonance
is modified by a factor @xmath378 , being thus the result essentially unaffected .
therefore , since for the first thousand of oscillations of the higgs , the accummulated phases of the fluctuations of the gauge bosons will be chaotic , we will average out the phases and work with @xmath379 .
on the other hand , the perturbative decay of the produced vector bosons occurs precisely just between two successive higgs zero - crossings , @xmath380 , where @xmath313 is given by eq .
( [ fn ] ) and @xmath381 , see eq .
( [ smallgamma ] ) .
taking into account eq .
( [ expgrowth ] ) and eq .
( [ floquet ] ) we can express the number of gauge bosons just after the @xmath291-th scattering in terms of the number just after the previous one @xmath382 applied recursively , this formula allows us to obtain the occupation number for each species and polarization , just after the @xmath291-th scattering in terms of the initial abundances @xmath383 , @xmath384\,\exp\big[2\pi\sum_{i=1}^{j}\mu_{k}(i+1)\big ] \;.
\end{aligned}\ ] ] the initial abundances are , of course , only generated through eq .
( [ deltan ] ) , and are given by @xmath385 where we have defined the function @xmath386 again , we used @xmath276 in light of eq .
( [ schrodinger ] ) , as the natural argument of the expression of the transmission probability ( [ transmision ] ) .
note that here , the species are only distinguished through the resonance parameters in eq .
( [ qfactor ] ) .
normalizing the scale factor at the first zero crossing as @xmath387 , then we can simply write the evolution of the scale factor as @xmath388 .
thus , the behaviour of @xmath276 with the number of zero crossings goes as @xmath277 .
then , we can define a typical momentum of the problem , @xmath389 , related in a very simple way to the resonance parameters @xmath390 , @xmath391 ( [ qfactor ] ) , as @xmath392 where @xmath393 with @xmath394 and @xmath395 for @xmath0 and @xmath1 bosons , respectively .
since @xmath389 is the natural scale for the momenta of the problem , its order of magnitude should coincide simply with the one obtained via the heisenberg uncertainty principle , see eq .
( [ timea ] ) , as it is indeed the case since @xmath396 notice that the typical momenta range will be red - shifted because of the expansion of the universe and , even the comoving typical moment @xmath397 , is not a static quantity but rather depends on @xmath252 . and @xmath1 bosons as a function of the variable @xmath398 . here
we show the maximum ( continuous red ) , the average ( short dashed green ) and the typical ( long dashed blue ) indices .
right : the initial spectral distribution @xmath383 ( lower blue curve ) and the gaussian approximation @xmath399 ( [ occupcombpre ] ) for different @xmath400 greater than 2 ( rest of the curves ) , describing the resonant behaviour .
the approximation is so good that it is hard to distinguish it from the real curve , presenting small deviations just on the tail .
the horizontal axis is @xmath401 and the curves correspond to different @xmath252 s .
it is clearly distinguishable the fact that only the range @xmath402 ( @xmath403 ) is filtered and therefore excited through parametric resonance , no matter if @xmath404.,width=264,height=188 ] let us obtain now the total number density of created particles .
just after the @xmath252-th scattering , this will be given by @xmath405 where @xmath406 ( [ typicalinitialmomenta ] ) should be evaluated with @xmath407 or @xmath395 , and @xmath408 or @xmath409 , respectively for @xmath0 or @xmath1 bosons .
this formula encodes the usual resonant behaviour discovered in the 90 s , see refs .
@xcite , in which it was implicitely assumed that the produced bosons did nt decay between succesive inflaton zero - crossings .
however , as we saw in section [ parametricreheating ] , the bosons produced each time the higgs condensate crosses zero , significantly decay before the next scattering .
therefore , we had to correct our formulas for this effect .
fortunately , this was easily done , since the resonant growth occurs in a step - like form , instantaneously ( within a time @xmath410 , see eq .
( [ timea ] ) ) when the higgs condensate crosses around zero , while the perturbative decay of the produced vector bosons occurs during the time just between two successive higgs zero - crossings .
thus , the occupation number just before the @xmath291-th scattering , in terms of the occupation number just after the @xmath252-th scattering , has been corrected by the factor @xmath411 , which accounts for the accummulated effect of the perturbative decays up to the @xmath252-th scattering .
the combined effect of the non - perturbative parametric resonant at the non - adiabatic regions at the bottom of the potential , together with the perturbative decay along the adiabatic zone during the rest of the semi - oscillation , give rise to a new phenemenology , as we will inmediately see .
therefore , to emphasize the difference from the usual parametric resonance or instant preheating - like mechanisms , we will call these effect @xmath412 @xmath413 . expanding the combination of airy functions ( [ cidef ] ) , it is possible to write @xmath414 where @xmath415 and @xmath416 substituying eqs .
( [ ciexpansion ] ) and ( [ cicoeff ] ) in eq .
( [ numberdensity ] ) , then we obtain @xmath417 where we have used @xmath388 , performed the resulting gaussian integral and defined @xmath418 for simplicity .
notice that the resonant behaviour is now encoded in the factor @xmath419 , which , for sufficiently great @xmath252 , will finally overtake the decaying factor @xmath420 , since @xmath421 .
taking also into account the factor @xmath422 due to the expansion of the universe , the first result we can read from here is that only for those values of @xmath252 for which @xmath423 , the resonant effect will dominate over both the perturbative decay and the expansion rate . between the typical momenta produced around zero and the average mass in every oscillation for the @xmath0 ( dashed blue line ) and @xmath1 bosons ( continuos red line ) as a function of the number of oscillations .
this ratio is significantly smaller than 1 for all crossings , which allows us to consider the produced gauge bosons as non - relativisitic .
right : succesive spectral distributions @xmath424 , at different @xmath252 s , including the volume factor @xmath425 .
one can see the predicted ( [ pico ] ) slow displacement of the maxima of the distribution .
the x - axis is given in terms of @xmath426,title="fig:",width=264,height=188 ] between the typical momenta produced around zero and the average mass in every oscillation for the @xmath0 ( dashed blue line ) and @xmath1 bosons ( continuos red line ) as a function of the number of oscillations .
this ratio is significantly smaller than 1 for all crossings , which allows us to consider the produced gauge bosons as non - relativisitic .
right : succesive spectral distributions @xmath424 , at different @xmath252 s , including the volume factor @xmath425 .
one can see the predicted ( [ pico ] ) slow displacement of the maxima of the distribution .
the x - axis is given in terms of @xmath426,title="fig:",width=264,height=192 ] note that inside the integral ( [ occupcombpre ] ) , the function @xmath427 has a maximum at a value @xmath428 , which implies that that the typical ( comoving ) excited momentum is @xmath429 in the right hand side of figure [ massmomentaratio ] , one can easily observe this behaviour : the value of the momentum at which the distribution peaks , slightly moves to smaller values , according to @xmath430 .
thus , it is noticeable that the typical momentum @xmath431 of the resonant fluctuations is always of order @xmath432 , independently of how many oscillations the higgs performs .
the reason of this is that the parametric resonance effect builds up initially from the spectral distribution @xmath383 , which only filters @xmath433 , see eq .
( [ formulacrecimiento ] ) and figure [ massmomentaratio ] . on the other hand , the ratio @xmath434 for both @xmath0 and @xmath1 bosons , between the typical momenta produced around zero and the average masses in every oscillation is shown in fig .
[ massmomentaratio ] .
in particular , it is easy to estimate the evolution in time of such a ratio , in terms of the resonant parameters ( [ qfactor ] ) , as @xmath435 taking into account that the previous ratio ( [ gaugebosonsmomenta ] ) is a decreasing function with @xmath252 , as well as its dependence with the gauge couplings @xmath436 , we can conclude that the vector bosons produced at the bottom of the potential are always non - relativistic .
this justifies _ a posteriori _ the calculation of the energy of the fermions as @xmath437 , see eq .
( [ fermionsenergy ] ) in section [ parametricreheating ] .
extrapolating eq .
( [ gaugebosonsmomenta ] ) to the case of fermions we realize that the produced particles at each zero crossing would be mainly relativistic , due to the smallness of the yukawa couplings , being the only exception the production of top quarks . the energy density transferred to the fermions between the @xmath252-th and the @xmath291-th scatterings , will be @xmath438 { \nonumber \\}&= & \frac{\tilde\epsilon}{2\pi^2}m^2\chi_{\rm end}^2a^{j-1}c\frac{\sqrt{\pi}}{4}\frac{k_*^3}{j^2}\left(d+b\sum_{l=2}^{j}l^{-2/3}\right)^{-\frac32}f(j)\times{\nonumber \\ } & & \times\left((1-e^{-\gamma_zf(j)})e^{-\gamma_z\gamma_\sigma(j-1 ) } + 2\cos\theta_w^3(1-e^{-\gamma_wf(j)})e^{-\gamma_w\gamma_\sigma(j-1)}\right)\,,\end{aligned}\ ] ] where we have used the energy of the fermions ( [ fermionsenergy ] ) and defined a momentum scale independent of the gauge couplings , common to both bosonic species , as @xmath439 the gauge coupling dependence is indeed incorporated on the definition of the parameter @xmath440 which modulates again the strengh of the effect as @xmath441
. the total energy density transferred into the fermions will be @xmath442 and the ratio of such an energy to that of the inflaton , @xmath443 with @xmath444 given by eq .
( [ deltarhofer ] ) . here
we can clearly see the two competing effects ; that of the perturbative decay of the bosons , given by the factors of the form @xmath445 , which tend to decrease the rate of production of bosons and fermions , while the factors @xmath446 encoded in the form of the gaussian approximation , describe the resonant effect due to the accumulation of previously produced bosons and fermions .
initially , the perturbative decay will prevent the resonance to be effective .
however , after a certain number of oscillations ( a number that we will estimate next ) , the resonant effect will overtake the perturbative decays and parametric resonance will be developed as usual , as if the produced bosons would not decay perturbatively during each semi - oscillation . in order to estimate the time in which the perturbative decays stop blocking the parametric resonance effect , we can evaluate numerically when the expression @xmath447 becomes subdominant versus @xmath448 .
in particular , we can evaluate the ratios , for either @xmath0 or @xmath1 @xmath449 for the fastest growing mode @xmath450 ( [ pico ] ) , and find the number of semi - oscillations @xmath451 for which the previous ratio becomes greater than one , @xmath452 .
we find @xmath453 for the @xmath0 bosons and @xmath454 for the @xmath1 bosons .
the fact that parametric resonance becomes important much earlier for @xmath0 s than for @xmath1 s is not a surprise , since their decay rate ( [ widthwe ] ) differ in a factor @xmath455 , which simply means that there are many more @xmath0 bosons surviving per semioscillation than @xmath1 bosons .
therefore , the _ combined preheating _ of the @xmath0 bosons is much faster driven into the parametric - like behaviour , while the evolution of the @xmath1 bosons is much more affected by the perturbative decays , delaying ( or even completely preventing ) the development of parametric resonance . obviously , after a dozen of oscillations , the transfer of energy from the inflaton to the gauge bosons will be completely dominated by the channel into the @xmath0 bosons , since by that moment they will be fully resonant while the @xmath1 bosons will still severely affected by their perturbative decay . finally , to conclude this section and achieve an overall complete picture of all the details , let us also estimate the transfer of energy from the inflaton to the gauge bosons . in particular , the total energy transferred to them just after the @xmath252-th scattering , @xmath456 , is given simply by @xmath457 where we have used the fact that the gauge bosons are non - relativistic and have 3 polarizations .
therefore , the ratio of the energy of the gauge bosons to the energy of the inflaton , can be expressed as @xmath458 where we have used eq .
( [ occupcombpre ] ) and @xmath459 , defined in eq .
( [ barepsilon ] ) , modulates again the amplitude of this growing function . using eqs .
( [ fracferm ] ) and ( [ fracbosons ] ) , we can estimate the time in which finally the energy of the inflaton would be transferred efficienty to the fermions or the bosons . defining that moment , respectively , like @xmath460 and @xmath461 , one obtains the numbers in table i. note that the bosons receive the transfer of energy from the inflaton before the fermions , since by the time that parametric resonance overtakes the perturbative decay , the fraction of bosons decaying ( per semi - oscilation ) into fermions is very small and , therefore
, the fraction of newly added fermions is less and less important , while the amount of produced bosons is more and more prominent . note also that the number of oscillations @xmath462 required for an efficient transfer of energy , depends on the parameter @xmath23 , although the overall order of magnitude does not change appreciably .
[ ratio ] .number of semi - oscillations of the higgs required , as a function of @xmath23 , for an efficient transfer of energy from the inflaton to the gauge fields and/or to the fermions . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] we clearly see that backreaction seems to become important at a time slightly earlier than that at which we were expecting the higgs to have transferred efficiently its energy to the bosons and fermions .
this means that our analytical estimates of these transfers were biased , and a careful numerical study of the process is required . beyond backreaction
, the strength of the resonance very quickly decreases due to the increased frequency of oscillations of the higgs .
eventually , the broad resonance driving the production of gauge bosons and thus their decay into sm particles becomes a narrow resonance and finally shuts off . from then on ,
the inflaton will oscillate like a matter field while the produced particle will redshift as radiation , its effect on the expansion becoming negligible after a few hundred oscillations .
we have studied the different stages of reheating after inflation in a model where the role of the inflaton is played by the higgs field of the standard model of particle physics with a non - minimal coupling to gravity .
inflation in this model takes place at the gut scale , along the lines of the starobinsky model of inflation since a conformal transformation makes these two models indistinguishable from the point of view of inflation .
the usual difficulty with large self couplings of the higgs is tamed here by the inclusion of a large non - minimal coupling to gravity , @xmath463 , which nevertheless does not leave any signature at low ( electroweak ) scales due to the fact that the higgs field acquires a vacuum expectation value and does not evolve at present , while the local spacetime curvature is negligible . the advantage of this model of inflation for the study or reheating after inflation is that all the couplings of the higgs - inflaton to matter fields are known at the electroweak scale , and can be extrapolated to the gut scale using the renormalization group equations , and therefore one can study in detail the process of reheating of the universe , without having to impose ad hoc assumptions about their values .
the surprise is that the process becomes more complicated than expected , and a series of subsequent stages take place , where essentially all different types of particle production mechanisms at preheating occur .
moreover , since the standard model couplings of the higgs to gauge and matter fields are non - negligible , nor are their couplings among themselves , the process of non - perturbative decay via parametric resonance is mixed with the usual perturbative decays of the decay products , which complicates things significantly .
inflation ends at values of the higgs field of order the planck scale and goes through a brief stage of tachyonic preheating soon after the end of inflation .
the passage is so short that particle production is not significant at that stage .
the same occurs with the production at the inflection point .
finally the higgs - inflaton field starts oscillating around the minimum of its potential with a curvature scale of order @xmath464 gev . at this stage ,
particle production occurs whenever the higgs passes through zero , creating mostly vector gauge bosons @xmath0 and @xmath1 .
these gauge bosons acquire a large mass while the higgs increases towards maximum amplitude and start to decay into all standard model leptons and quarks within half a higgs oscillation , rapidly depleting the occupation numbers of gauge bosons , like in instant preheating .
however , the fraction of energy of the higgs that goes into sm particles is still very small compared with the energy in the oscillations , and therefore the non - perturbative decay is slow .
this implies that a relatively large number of oscillations take place before a significant amount of energy is transferred to the gauge bosons and fermions .
the amplitude of higgs boson oscillations decreases as the universe expands in a matter - like dominated stage with zero pressure .
eventually , this amplitude is small enough that the gauge boson masses are not large enough for inducing a quick decay of the gauge bosons and these start to build up their occupation numbers very rapidly via parametric amplification .
the question whether this effect can give rise to the production of a significant gravitational wave background ( gwb ) potentially observable today remains to be addressed .
several papers have studied recently such an issue in the chaotic and hybrid models of inflation @xcite , but in the present model , we do nt have simply a parametric resonance phenomena but a combined preheating effect which , perhaps , could modify the properties of such a gwb .
similar arguments would affect also the production of magnetic fields at preheating @xcite or even electroweak baryogenesis @xcite .
after about a hundred oscillations the gauge bosons produced backreact on the higgs field and the resonant production of particles stops .
the higgs field acquires a large mass via its interaction with the gauge condensate and preheating ends . from there on , both higgses and gauge fields decay perturbatively until their energy is transferred to sm particles .
since the stage after backreaction is very non - linear and non - perturbative , it can not be solved analytically and we have to resort to numerical studies in the lattice . we leave the description of our numerical studies to a future publication . : upon completion of this paper , we received through the arxiv the preprint of bezrukov et al .
@xcite , where they also study preheating in the @xmath465msm .
although the formalism is common to both , our conclusions are somewhat different from those of ref .
we find that the higgs decay into gauge bosons is significantly faster , and that backreaction occurs much before thermalization .
we thus think it is not possible to determine the reheating temperature without a careful numerical analysis with lattice simulations .
a few days after this work was completed , ref .
@xcite was also posted in the arxiv , performing an analysis of the 2-loop quantum corrections to the running of all the parameters involved in the model .
this paper allows for a different range of the higgs self - coupling , which is compatible with the range ( [ higgsrange ] ) although more restricted .
the main result of ref .
@xcite is a relationship between the higgs mass and the spectral index which , in principle , could be tested in the future against data from planck and the lhc .
similar conclusions where found in ref .
@xcite .
we would like to thank geneva university , sissa - trieste and mpi - munich for hospitality during the development of parts of this research .
dgf is supported by a fpu contract with ref .
ap2005 - 1092 and jr by an i3p contract .
we also acknowledge financial support from the madrid regional government ( cam ) under the program hephacos p - esp-00346 , and the spanish research ministry ( mec ) under contract fpa2006 - 05807 .
the authors participate in the consolider - ingenio 2010 cpan ( csd2007 - 00042 ) and pau ( csd2007 - 00060 ) , as well as in the european union 6th framework marie curie network universenet " under contract mrtn - ct-2006 - 035863 . 99 a. a. starobinsky , jetp lett .
* 30 * , 682 ( 1979 ) [ pisma zh .
fiz . * 30 * , 719 ( 1979 ) ] ; phys .
b * 91 * ( 1980 ) 99 ; v. f. mukhanov and g. v. chibisov , jetp lett .
* 33 * ( 1981 ) 532 [ pisma zh .
fiz . * 33 * ( 1981 ) 549 ] ; a. h. guth , phys .
d * 23 * , 347 ( 1981 ) ; a. d. linde , phys . lett .
b * 108 * , 389 ( 1982 ) ; a. albrecht and p. j. steinhardt , phys .
lett . * 48 * , 1220 ( 1982 ) .
a. d. linde , `` particle physics and inflationary cosmology , '' chur , switzerland : harwood ( 1990 ) , arxiv : hep - th/0503203 ; d. h. lyth and a. riotto , phys .
* 314 * ( 1999 ) 1 , arxiv : hep - ph/9807278 .
a. linde , lect .
notes phys .
* 738 * ( 2008 ) 1 , arxiv:0705.0164 [ hep - th ] ; a. d. dolgov and a. d. linde , phys .
b * 116 * , 329 ( 1982 ) ; l. f. abbott , e. farhi and m. b. wise , phys .
b * 117 * , 29 ( 1982 ) ; a. albrecht , p. j. steinhardt , m. s. turner and f. wilczek , phys .
lett . * 48 * , 1437 ( 1982 ) .
e. komatsu _ et al .
_ [ wmap collaboration ] , arxiv:0803.0547 [ astro - ph ] . for a review see s. l. adler , rev
. mod .
phys . * 54 * , 729 ( 1982 ) [ erratum - ibid . * 55 * , 837 ( 1983 ) ] .
p. minkowski , phys .
b * 71 * , 419 ( 1977 ) .
l. smolin , nucl .
b * 160 * , 253 ( 1979 ) .
a. zee , phys . rev
. lett . *
42 * , 417 ( 1979 ) .
d. s. salopek , j. r. bond and j. m. bardeen , phys .
d * 40 * , 1753 ( 1989 ) .
n.d . birrell and p.c.w .
davies , quantum fields in curved space " ( cambridge u.p . , uk , 1982 ) . c. brans and r. h. dicke , phys .
rev . * 124 * , 925 ( 1961 ) .
s. l. glashow , nucl .
( 1961 ) 579 ; s. weinberg , phys .
lett 19 ( 1967 ) 1264 ; a. salam , proceedings of the nobel symposium held 1968 at lerum , sweden , stockhol ( 1968 ) , 367 .
h. dehnen , h. frommert and f. ghaboussi , int .
j. theor .
* 29 * , 537 ( 1990 ) . h. dehnen and h. frommert , int .
j. theor .
* 30 * ( 1991 ) 985 .
l. a. kofman , v. f. mukhanov and d. y. pogosian , sov .
jetp * 66 * ( 1987 ) 433 [ zh .
* 93 * ( 1987 ) 769 ] ; v. f. muhkanov and g. v. chibisov , pisma zh .
* 33 * ( 1981 ) 54 [ jetp lett . * 33 * ( 1981 ) 48 ] ; v. f. mukhanov and g. v. chibisov , sov .
jetp * 56 * ( 1982 ) 258 [ zh .
* 83 * ( 1982 ) 475 ] .
f. l. bezrukov and m. shaposhnikov , phys .
b * 659 * ( 2008 ) 703 [ arxiv:0710.3755 [ hep - th ] ] .
s. tsujikawa , k. i. maeda and t. torii , phys .
d * 60 * , 123505 ( 1999 ) [ arxiv : hep - ph/9906501 ] .
k. i. maeda , phys .
d * 39 * , 3159 ( 1989 ) .
n. makino and m. sasaki , prog .
phys . * 86 * , 103 ( 1991 ) .
r. fakir , s. habib and w. unruh , astrophys .
j. * 394 * , 396 ( 1992 ) .
a. o. barvinsky and a. y. kamenshchik , nucl .
b * 532 * , 339 ( 1998 ) [ arxiv : hep - th/9803052 ] .
a. o. barvinsky , a. y. kamenshchik and a. a. starobinsky , jcap * 0811 * , 021 ( 2008 ) [ arxiv:0809.2104 [ hep - ph ] ] .
a. o. barvinsky , a. y. kamenshchik and i. p. karmazin , phys .
d * 48 * , 3677 ( 1993 ) [ arxiv : gr - qc/9302007 ] .
g. m. shore , annals phys . * 128 * , 376 ( 1980 ) .
a. a. starobinsky , phys .
b * 117 * ( 1982 ) 175 .
g. n. felder , j. garcia - bellido , p. b. greene , l. kofman , a. d. linde and i. tkachev , phys .
lett . * 87 * , 011601 ( 2001 ) [ arxiv : hep - ph/0012142 ] .
g. n. felder , l. kofman and a. d. linde , phys .
d * 64 * ( 2001 ) 123517 [ arxiv : hep - th/0106179 ] .
j. garcia - bellido , m. garcia perez and a. gonzalez - arroyo , phys .
d * 67 * , 103501 ( 2003 ) [ arxiv : hep - ph/0208228 ] ; g. n. felder , l. kofman and a. d. linde , phys .
d * 59 * , 123523 ( 1999 ) [ arxiv : hep - ph/9812289 ] .
l. kofman , a. d. linde and a. a. starobinsky , phys . rev
. lett . *
73 * , 3195 ( 1994 ) [ arxiv : hep - th/9405187 ] .
l. kofman , a. d. linde and a. a. starobinsky , phys .
d * 56 * , 3258 ( 1997 ) [ arxiv : hep - ph/9704452 ] .
j. garcia - bellido and a. d. linde , phys .
d * 57 * , 6075 ( 1998 ) [ arxiv : hep - ph/9711360 ] .
r. easther and e. a. lim , jcap * 0604 * , 010 ( 2006 ) [ arxiv : astro - ph/0601617 ] ; r. easther , j. t. .
giblin and e. a. lim , phys .
lett . * 99 * , 221301 ( 2007 ) [ arxiv : astro - ph/0612294 ] ; j. garcia - bellido and d. g. figueroa , phys .
lett . *
98 * , 061302 ( 2007 ) [ arxiv : astro - ph/0701014 ] ; j. garcia - bellido , d. g. figueroa and a. sastre , phys .
d * 77 * , 043517 ( 2008 ) [ arxiv:0707.0839 [ hep - ph ] ] ; r. easther , j. t. .
giblin and e. a. lim , arxiv:0712.2991 [ astro - ph ] ; j. f. dufaux , g. n. felder , l. kofman and o. navros , arxiv:0812.2917 [ astro - ph ] .
a. diaz - gil , j. garcia - bellido , m. garcia perez and a. gonzalez - arroyo , phys .
lett . * 100 * , 241301 ( 2008 ) [ arxiv:0712.4263 [ hep - ph ] ] .
* 0807 * , 043 ( 2008 ) [ arxiv:0805.4159 [ hep - ph ] ] . j. garcia - bellido , d. y. grigoriev , a. kusenko and m. e. shaposhnikov , phys .
d * 60 * ( 1999 ) 123504 [ arxiv : hep - ph/9902449 ] ; l. m. krauss and m. trodden , phys .
* 83 * ( 1999 ) 1502 [ arxiv : hep - ph/9902420 ] ; a. tranberg and j. smit , jhep * 0311 * , 016 ( 2003 ) [ arxiv : hep - ph/0310342 ] , nucl . phys . proc .
* 129 * , 768 ( 2004 ) [ arxiv : hep - lat/0309082 ] ; j. garcia - bellido , m. garcia - perez and a. gonzalez - arroyo , phys .
d * 69 * ( 2004 ) 023504 [ arxiv : hep - ph/0304285 ] .
f. bezrukov , d. gorbunov and m. shaposhnikov , `` on initial conditions for the hot big bang , '' arxiv:0812.3622 [ hep - ph ] .
a. de simone , m. p. hertzberg and f. wilczek , `` running inflation in the standard model , '' arxiv:0812.4946 [ hep - ph ] .
f. l. bezrukov , a. magnin and m. shaposhnikov , `` standard model higgs boson mass from inflation , '' arxiv:0812.4950 [ hep - ph ] . | we study the details of preheating in an inflationary scenario in which the standard model higgs , strongly non - minimally coupled to gravity , plays the role of the inflaton .
we find that the universe does not reheat immediately through perturbative decays , but rather initiate a complex process in which perturbative and non - perturbative effects are mixed .
the higgs condensate starts oscillating around the minimum of its potential , producing @xmath0 and @xmath1 gauge bosons non - perturbatively , due to violation of the so called adiabaticity condition .
however , during each semi - oscillation , the created gauge bosons partially decay ( perturbatively ) into fermions .
the decay of the gauge bosons prevents the development of parametric resonance , since bosons can not accumulate significantly at the beginning .
however , the energy transferred to the decay products of the bosons is not enough to reheat the universe , so after about a hundred oscillations , the resonance effects will eventually dominate over the perturbative decays . around the same time
( or slightly earlier ) , backreaction from the gauge bosons into the higgs condensate will also start to be significant .
soon afterwards , the universe is filled with the remnant condensate of the higgs and a non - thermal distribution of fermions and bosons ( those of the sm ) , which redshift as radiation and matter , respectively .
we compute the distribution of the energy budget among all the species present at the time of backreaction . from there on until thermalization ,
the evolution of the system is highly non - linear and non - perturbative , and will require a careful study via numerical simulations .
preprint ift - uam / csic-08 - 93 |
charm mixing in the standard model is conventionally described by two small dimensionless parameters : @xmath16 where @xmath17 and @xmath18 are the masses and widths , respectively , of the neutral @xmath19 meson @xmath20 eigenstates , @xmath21 ( @xmath20-odd ) and @xmath22 ( @xmath20-even ) , defined by latexmath:[\ ] ] these @xmath59 are related to rates of single tags ( st ) , or individually reconstructed @xmath25 or @xmath26 candidates , which are obtained by summing over dt rates : @xmath61 = { \cal b}_i + { \cal b}_{\bar\imath } = a_i^2 \left ( 1 + 2 y r_i r_i \cos\delta_i + r_i^2 \right).\ ] ] here , we have used an expression for @xmath15 in terms of @xmath62 , @xmath63 , and @xmath47 , which is derived from eqs .
( [ eq : y][eq : d2 ] ) and eqs .
( [ eq : r2general][eq : coherencegeneral ] ) : @xmath64 } { \sum_i \left [ \int | a_i(\mathbf x ) - \bar a_i(\mathbf x ) |^2 d\mathbf x + \int | a_i(\mathbf x ) + \bar a_i(\mathbf x ) |^2 d\mathbf x \right ] } = -2\frac{\sum_i a_i^2 r_i r_i \cos\delta_i}{\sum_i a_i^2 ( 1 + r_i^2 ) } .\ ] ] thus , both st rates and the total rate , @xmath65 , are unaffected by quantum correlations between the @xmath25 and @xmath26 decays , and our sensitivity to mixing comes from comparing st to dt rates . table [ tab : rdelta ] gives the notation for the various @xmath62 and @xmath47 that appear in this analysis .
the final states of mixed @xmath20 ( denoted by @xmath66 and @xmath67 below ) that we consider are @xmath68 and @xmath69 .
following ref .
@xcite , the @xmath69 dalitz plot is divided into eight bins according to the strong phase of the decay amplitude .
we denote the portions of @xmath69 in phase bin @xmath43 by @xmath70 and @xmath71 , where @xmath72 for @xmath70 , and @xmath73 for @xmath71 .
the corresponding amplitude ratio magnitudes and branching fraction ratios integrated over bin @xmath43 are denoted by @xmath74 and @xmath75 , respectively . as in ref .
@xcite , we denote the real and imaginary parts of eq .
( [ eq : coherencegeneral ] ) by @xmath76 and @xmath77 , respectively , but with the opposite sign convention for @xmath77 .
semileptonic final states ( @xmath78 ) , @xmath20-even eigenstates ( @xmath79 ) , and @xmath20-odd eigenstates ( @xmath80 ) have known values of @xmath62 and @xmath47 , which give them unique leverage in determining the parameters in the other final states , as demonstrated below . note that , as shown in eqs .
( [ eq : bfcf ] ) and ( [ eq : bfdcs ] ) , the ratio @xmath81 does not equal @xmath82 in general .
.parameters describing the ratio of amplitudes @xmath83 and @xmath84 for the final states @xmath43 .
the @xmath85 indicates that we do not make explicit reference to @xmath47 for the @xmath86 modes in this article , but consider only @xmath87 and @xmath88 instead . [
cols="^,^,^,^,^,^",options="header " , ] lrrrr & @xmath15 & @xmath39 & @xmath4 & @xmath5 + + @xmath39 & @xmath89 + @xmath4 & @xmath90 & @xmath91 + @xmath5 & @xmath92 & @xmath93 & @xmath94 + @xmath38 & @xmath95 & @xmath89 & @xmath96 & @xmath97 + + @xmath39 & @xmath98 + @xmath4 & @xmath99 & @xmath100 + @xmath5 & @xmath101 & @xmath102 & @xmath103 + @xmath38 & @xmath104 & @xmath105 & @xmath106 & @xmath107 + these results include the first direct measurement of @xmath5 , as well as first branching fraction measurements for @xmath108 , @xmath109 , and @xmath110 .
we also present the first cleo - c branching fraction measurement for @xmath111 . in the standard fit ,
the statistical uncertainties on @xmath15 and @xmath112 are approximately three times smaller than in our previous analysis @xcite . because of the strong correlation between @xmath39 and @xmath4 , and because @xmath39 is determined by different inputs in the two analyses , a direct comparison of the @xmath4 uncertainties is not instructive .
accounting for correlated uncertainties , the results for @xmath4 in the standard and extended fits differ by @xmath113 .
this difference is caused primarily by the upward fluctuations in @xmath15 and @xmath39 in the standard fit , which are both negatively correlated with @xmath4 .
the above factor of three improvement in @xmath15 and @xmath112 can be attributed in equal parts to the increased size of the dataset and to the additional final states in the current analysis . in particular , the new @xmath114 modes ( including @xmath115 )
reduce the statistical uncertainties on @xmath15 and @xmath112 in the standard fit by roughly 10% , and the @xmath116 modes reduce them by 2030% .
the @xmath69 measurements from ref .
@xcite have a similar effect on @xmath15 and @xmath112 as the new @xmath114 modes , but they also provide all the information on @xmath5 in the standard fit .
the extended fit demonstrates that when our @xmath4 and @xmath5 measurements are used to combine @xmath15 and @xmath117 from other experiments , the overall uncertainty on @xmath15 is reduced by approximately 10% compared to the global average of all measurements in table [ tab : externalyyprime ] ( except @xmath118 from ref .
@xcite ) found in ref .
@xcite : @xmath119 .
note that this global average includes the results from ref .
@xcite .
figure [ fig : contoursnoextmeas ] shows the one - dimensional posterior pdfs for @xmath4 , @xmath5 , @xmath6 , and @xmath15 in the standard fit , including statistical and systematic uncertainties .
these curves are obtained by re - minimizing the @xmath35 at each point and computing @xmath120 .
also shown are the two - dimensional contours for combinations of @xmath15 , @xmath4 , and @xmath5 . because of the @xmath5 sign ambiguity , the pdfs for both @xmath5 and @xmath6 in the standard fit are symmetric around zero .
figure [ fig : contoursyypext ] shows the same distributions for the extended fit . here , the sign ambiguity is resolved by the external measurements .
all the pdfs , except those for @xmath6 , are well described by gaussians or bifurcated gaussians . in particular , the non - gaussian profile for @xmath4 in the standard fit from ref .
@xcite has been eliminated by our direct measurement of @xmath39 .
although the central value for @xmath4 in the extended fit is unphysical , we find @xmath121 to be consistent with physical boundary .
similarly , in the standard fit , @xmath122 .
the pdfs for @xmath6 in figs .
[ fig : contoursnoextmeas ] and [ fig : contoursyypext ] are obtained by probing @xmath4 and @xmath5 under the constraint @xmath123 , which reduces the height of the pdf relative to the other parameters .
the implied values for @xmath6 from these pdfs are @xmath11 for the standard fit and @xmath14 for the extended fit .
also , applying the above constraint in the standard fit improves the uncertainties on @xmath15 and @xmath38 by 15% and 8% , respectively , resulting in @xmath124 and @xmath125 ; the changes in all other parameters are negligible .
performing the extended fit with @xmath123 produces negligible shifts in all the fit parameters .
our results for @xmath76 , @xmath77 , and branching fractions do not supersede other cleo - c measurements .
for @xmath76 and @xmath77 , our fitted values are consistent with those in ref .
@xcite , after accounting for differences between the two analyses . , @xmath5 , @xmath6 , and @xmath15 .
the two - dimensional likelihoods are shown as contours in increments of @xmath126 , where @xmath127 . ] , @xmath5 , @xmath6 , and @xmath15 .
the two - dimensional likelihoods are shown as contours in increments of @xmath126 , where @xmath127 . ]
we present an updated analysis of quantum correlations in @xmath0 decays at the @xmath33 using the full cleo - c dataset , resulting in a new value of @xmath9 and a first measurement of @xmath10 , which , when combined , imply a strong phase of @xmath11 . by including external inputs on mixing parameters in the fit
, we find alternative values of @xmath12 , @xmath13 , and @xmath14 .
the effect of these measurements on the world averages of mixing parameters is to improve the uncertainty on @xmath15 by approximately 10% .
we gratefully acknowledge the effort of the cesr staff in providing us with excellent luminosity and running conditions .
this work was supported by the national science foundation , the u.s .
department of energy , the natural sciences and engineering research council of canada , and the u.k .
science and technology facilities council .
b. aubert _ et al . _ [ babar collaboration ] , phys .
d * 78 * , 011105 ( 2008 ) [ arxiv:0712.2249 [ hep - ex ] ] .
b. aubert _ et al . _
[ babar collaboration ] , phys .
d * 80 * , 071103 ( 2009 ) [ arxiv:0908.0761 [ hep - ex ] ] .
a. zupanc _ et al .
_ [ belle collaboration ] , phys .
d * 80 * , 052006 ( 2009 ) [ arxiv:0905.4185 [ hep - ex ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , jhep * 1204 * , 129 ( 2012 ) [ arxiv:1112.4698 [ hep - ex ] ] .
l. m. zhang _
[ belle collaboration ] , phys .
lett . *
99 * , 131803 ( 2007 ) [ arxiv:0704.1000 [ hep - ex ] ] .
del amo sanchez _ et al .
_ [ babar collaboration ] , phys .
* 105 * , 081803 ( 2010 ) [ arxiv:1004.5053 [ hep - ex ] ] .
l. m. zhang _
[ belle collaboration ] , phys .
lett . * 96 * , 151801 ( 2006 ) [ arxiv : hep - ex/0601029 ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
rev . lett . * 100 * , 121802 ( 2008 ) [ arxiv:0712.1567 [ hep - ex ] ] .
m. gronau , y. grossman and j. l. rosner , phys .
b * 508 * , 37 ( 2001 ) [ arxiv : hep - ph/0103110 ] .
j. l. rosner _
[ cleo collaboration ] , phys .
lett . * 100 * , 221801 ( 2008 ) [ arxiv:0802.2264 [ hep - ex ] ] ; d. m. asner _
et al . _ [ cleo collaboration ] , phys .
d * 78 * , 012001 ( 2008 ) [ arxiv:0802.2268 [ hep - ex ] ] .
m. goldhaber and j. l. rosner , phys .
d * 15 * , 1254 ( 1977 ) .
i. i. bigi and a. i. sanda , phys .
b * 171 * , 320 ( 1986 ) . i. i. bigi , slac - pub-4000 . _ invited talk given at workshop on physics simulation at high energies , madison , wi , may 5 - 16 , 1986_. d. atwood and a. a. petrov , phys .
d * 71 * , 054032 ( 2005 ) [ arxiv : hep - ph/0207165 ] .
d. m. asner and w. m. sun , phys . rev .
d * 73 * , 034024 ( 2006 ) [ erratum - ibid . * 77 * , 019901(e ) ( 2008 ) ] [ arxiv : hep - ph/0507238 ] .
r. aaij _ et al . _
[ lhcb collaboration ] , phys .
lett . * 108 * , 111602 ( 2012 ) [ arxiv:1112.0938 [ hep - ex ] ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , arxiv:1207.2158 [ hep - ex ] .
t. aaltonen _ et al .
_ [ cdf collaboration ] , phys . rev .
d * 85 * , 012009 ( 2012 ) [ arxiv:1111.5023 [ hep - ex ] ] .
n. neri , arxiv:1208.5877 [ hep - ex ] .
d. atwood and a. soni , phys .
d * 68 * , 033003 ( 2003 ) [ hep - ph/0304085 ] .
r. a. briere _ et al . _
[ cleo collaboration ] , phys .
d * 80 * , 032002 ( 2009 ) [ arxiv:0903.1681 [ hep - ex ] ] .
w. s. brower and h. p. paar , nucl .
instrum .
a * 421 * , 411 ( 1999 ) [ hep - ex/9710029 ] .
t. hokuue _ et al . _ [ belle collaboration ] , phys .
b * 648 * , 139 ( 2007 ) [ hep - ex/0604024 ] .
j. beringer _
[ particle data group ] , phys .
d * 86 * , 010001 ( 2012 ) .
d. asner _
[ heavy flavor averaging group collaboration ] , arxiv:1010.1589 [ hep - ex ] .
s. dobbs _ et al . _
[ cleo collaboration ] , phys .
d * 76 * , 112001 ( 2007 ) [ arxiv:0709.3783 [ hep - ex ] ] .
d. besson _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 80 * , 032005 ( 2009 ) [ arxiv:0906.2983 [ hep - ex ] ] . . | we analyze a sample of 3 million quantum - correlated @xmath0 pairs from 818 @xmath1 of @xmath2 collision data collected with the cleo - c detector at @xmath3 gev , to give an updated measurement of @xmath4 and a first determination of @xmath5 , where @xmath6 is the relative strong phase between doubly cabibbo - suppressed @xmath7 and cabibbo - favored @xmath8 decay amplitudes . with no inputs from other experiments , we find @xmath9 , @xmath10 , and @xmath11 . by including external measurements of mixing parameters , we find alternative values of @xmath12 , @xmath13 , and @xmath14 .
our results can be used to improve the world average uncertainty on the mixing parameter @xmath15 by approximately 10% .
= 1 |
in recent years , observations of accreting black holes with the _ rossi x - ray timing explorer ( rxte ) _ have discovered a number of sources with high frequency quasi - periodic oscillations ( hfqpos ) in their x - ray light curves .
these sources are seen predominately in the steep power law ( spl ) spectral state , suggesting a significant level of inverse - compton scattering of the emitted photons off of hot coronal electrons ( for an excellent review of the observations , see @xcite ) .
motivated by these observations , we extend a simple geodesic hot spot model @xcite to include monte carlo scattering of photons emitted isotropically in the hot spot s rest frame , which then propagate through a corona surrounding the black hole . the hot spot has a planar orbit near the inner - most stable circular orbit ( isco ) and the corona is modeled with a self - similar density profile described by the advection dominated accretion flow ( adaf ) model @xcite . in section [ classical_scattering ] , we describe the physics of classical electron scattering of unpolarized light , and show how energy is transferred from high - energy electrons to low - energy photons . section [ rel_implementation ] shows how this scattering is treated in a relativistic context in the kerr metric . assuming a thermal emitter with @xmath1 kev
, we show in section [ effect_spectra ] how the observed spectrum is modified by the hot electrons .
the shape of this modified thermal spectrum can in turn be used to infer the properties of the corona .
section [ effect_lightcurves ] shows the time - dependent effects of scattering on the x - ray light curve , particularly how the amplitude of modulation is damped for larger optical depths .
furthermore , we find this damping is more significant for photons experiencing multiple scattering events , which tend to have higher energies .
we conclude in section [ implications_qpo ] with a summary of the implications these results have for the hot spot model for qpos .
following rybicki & lightman @xcite , we use the cross section for thomson scattering of unpolarized radiation off of nonrelativistic electrons : @xmath2 where @xmath3 is the classical electron radius @xmath4 cm .
it is important to note that `` nonrelativistic '' is a reference to the _ photon _ energy , not the electron energy . in the electron rest frame , we require @xmath5 in order for the above cross sections to be valid , in which case the scattering is nearly elastic or _
coherent_. for higher energy photons , the scattering involves quantum effects and requires the `` klein - nishina '' cross section @xcite .
since we are primarily interested in the scattering of photons from a relatively cool thermal accretion disk ( @xmath6 kev ) , the classical treatment should suffice .
even if the scattering is treated as coherent in the electron frame , in the lab frame energy can be ( and often is ) transferred from the electron to the photon . to see this boosting effect , consider a photon with initial energy @xmath7 scattering off an electron with velocity @xmath8 in the @xmath9-direction in the `` lab frame '' @xmath10 . in this frame ,
the angle between the incoming photon and electron velocity is @xmath11 . in the electron rest frame @xmath12
, the photon is scattered at an angle @xmath13 with respect to the @xmath14-axis .
the doppler shift formula @xcite gives @xmath15 where @xmath16 and @xmath17 is the post - scattering energy in the lab frame . in the electron frame
, we assume elastic scattering with @xmath18 , which should be the case for the typical seed photons from a thermal emitter at @xmath0 kev .
averaging over all angles @xmath11 ( isotropic electron distribution ) and @xmath13 [ weighted by eqn .
( [ cross_unpol ] ) ] , we find that the typical scattering event boosts the photon energy by @xmath19 to determine @xmath20 , we consider a maxwell - boltzmann distribution function in electron momentum @xmath21 : @xmath22
unlike the approach taken in @xcite , where the photons were traced backwards in time from a distant observer to the emitting region , here it is conceptually easier to trace the photons _ forward _ in time from the emitter to the observer , then use monte carlo methods to determine the distribution of scattered photons . for the thomson cross section ,
the path of each photon is energy - independent , so the photon s final observed energy can be thought of as a fiducial redshift @xmath23 that can be convolved with the spectrum in the local emitter frame to produce the total spectrum seen by the observer . to determine the initial momentum of each photon
, we construct a tetrad centered on the emitter s rest frame , denoted by tilde indices @xmath24 . in the coordinate basis , the energy and angular momentum of a particle on a stable circular orbit around a kerr black hole
are given by @xmath25 and @xmath26 from these we construct the 4-velocity via the inverse metric @xmath27 , which gives @xmath28 .
then @xmath29 and @xmath30 are defined parallel to the coordinate basis vectors @xmath31 and @xmath32 , and @xmath33 is given by orthogonality . in this basis
, the initial photon direction is picked randomly from an isotropic distribution , uniform in spherical coordinates @xmath34 $ ] and @xmath35 .
all photons are given the same initial energy in the emitter frame @xmath36 , which is used as a reference energy for calculating the final redshift with respect to a stationary observer at infinity .
given @xmath37 , the photon s geodesic trajectory is integrated using the hamiltonian formulation described in @xcite .
figure [ emitter_rays ] shows an `` overhead view '' of photon trajectories in the plane of the disk , emitted isotropically by a massive test particle on a circular orbit at the isco of a black hole with @xmath38 .
the photons are colored according to their energy - at - infinity @xmath39 , either blue- or redshifted with respect to their energy in the emitter frame @xmath40 .
the relativistic beaming is done automatically by the lorentz boost from the emitter to the coordinate frame , so the blue photons are clearly bunched more tightly together , as required by the invariance of @xmath41 . at each step along the photon s path
, we determine the probability of electron scattering according to the differential optical depth @xmath42 the density @xmath43 is defined in the `` zero angular momentum observer '' ( zamo ) frame @xcite and the opacity @xmath44 is given by the classical cross section quoted above in equation ( [ cross_unpol ] ) .
the scattering is then treated classically in the electron s local frame , after which the new photon 4-momentum is transformed back to the coordinate basis and the ray - tracing continues along the new trajectory .
as we showed at the end of the section [ classical_scattering ] , one effect of the coronal scattering is generally a transfer of energy from the electron to the photon .
one way to quantify this energy transfer is through the compton @xmath45 parameter , defined as the average fractional energy change per scattering , times the number of scatterings through a finite medium . for nonrelativistic electrons , rybicki & lightman @xcite show that the average energy transfer per scattering event is @xmath46 the mean number of scatterings for an optically thin medium is simply @xmath47 , the total optical depth through the medium . for optically thick systems
, the photons must take a random walk to escape , so the number of scatterings becomes @xmath48 .
thus the compton @xmath45 parameter for a finite medium of nonrelativistic electrons is @xmath49 for a low - energy soft photon source with multiple scattering events , the final spectrum due to inverse - compton scattering can be estimated using the _
kompaneets equation_. for @xmath50 , the resulting spectrum takes the power - law form @xmath51 with @xmath52 at energies above @xmath53 , the electrons no longer efficiently transfer energy to the photons , so the spectrum shows a cutoff for @xmath54 : @xmath55 with the assumption of purely elastic scattering , we can not actually reproduce this cutoff effect ; all photons are scattered equally , and the ratio @xmath56 is independent of energy . thus equation ( [ ef_ei2 ] ) would predict infinite energy boosts until @xmath57 . in reality , higher energy photons tend to lose energy in scattering , due to the recoil of the electron .
this effect is relatively easy to calculate from conservation of energy and momentum in the electron rest frame : @xmath58 to accurately include this effect , we would have to keep track of the real `` physical '' energy of each photon , instead of the fiducial redshift method that we currently use to reconstruct the total spectrum afterwards .
ultimately , this is just a matter of computational intensity and no real conceptual difficulty . to first - order
, we can treat the thermal photon source as a monochromatic emitter at @xmath59 , which gives a reasonable approximation to the true solution . for the corona geometry
, we use a self - similar distribution described by an adaf model @xcite , with density and temperature profiles that scale as @xmath60 outside of the isco .
we have ignored the bulk velocity of the inwardly flowing gas , which will typically have @xmath61 in the adaf model . without scattering ,
the time - averaged `` numerical '' spectrum could be described by the relativistic transfer function described in @xcite , defined over an infinitesimal band in radius @xmath62 .
the inverse - compton processes in the corona serve to further broaden this transfer function , as in @xcite and @xcite .
this transfer function is then normalized to the rest energy @xmath40 and convolved with the actual emission spectrum ( e.g. a thermal blackbody at @xmath63 ) to give the simulated observed spectrum .
figure [ powerlaw_cutoff ] shows a set of these simulated spectra from a hot spot emitter around a black hole with @xmath38 .
the emission spectrum is thermal in the hot spot rest frame with @xmath64 kev .
the coronal adaf model has @xmath65 kev , and electron density @xmath66 for a variety of optical depths @xmath47 .
the spectra are plotted in units of ( # photons / s/@xmath67/kev ) , as is the convention by many observers , but the actual magnitude of the @xmath45-axis is arbitrary , and would normally depend on the distance to the source . from the slope of the power - law and the location of the cut - off , the corona temperature and optical depth can be inferred from observations @xcite .
the spectra in figure [ powerlaw_cutoff ] were created by integrating over the complete hot spot orbital period and over all observer inclination angles . however , during the monte carlo calculation , it is just as simple to bin all the photons according to their final values of @xmath11 , @xmath68 , and energy @xmath69 .
the latitude bins are evenly spaced in @xmath70 so that a comparable number of photons land in each zone .
the energy bins are spaced logarithmically to include the high energy tail and also maintain high enough spectral resolution at lower energies .
assuming the hot spot is on a circular periodic orbit , the photons detected at any azimuthal position can be mapped into the appropriate bin in @xmath68 , so that none are `` wasted . '' an excellent way to see the effects of scattering on the hot spot light curves is by plotting time - dependent spectra of a monochromatic emitter , shown in figure [ scatter_spectrogram75 ] for a range of optical depths and an inclination angle of @xmath71 .
the logarithmic color scale shows ( # photons / s/@xmath67/kev / period ) , normalized to the peak intensity in each panel . at `` 0 '' phase ,
when the emitter is on the far side of the black hole , the spectrum shows two distinct lines , one blueshifted in the forward direction of hot spot motion , and one redshifted in the backward direction . as the hot spot comes around towards the observer , the directly beamed blueshifted line dominates , and then when the phase is @xmath72 and the hot spot is on the near side of the black hole , a single line dominates .
this is due to the gravitational demagnification of the secondary images formed by photons that have to complete a full circle around the black hole to reach the observer .
while these features would most likely be unresolvable for black hole binaries , they may well be observable in x - ray flares from sgr a@xmath73 as well as other supermassive black holes ( e.g. see @xcite ) .
in the subsequent panels , the spectrum is modified by the scattering of the hot spot photons in the surrounding corona . as in section [ effect_spectra ] , the temperature and density profile of the corona is given by an adaf model with @xmath74 kev .
the four panels of figure [ scatter_spectrogram75 ] show increasing values of @xmath75 $ ] .
the effects of scattering on the spectra are really quite profound .
as we described qualitatively in @xcite , the electron corona is like a cloud of fog surrounding a lighthouse , spreading out the delta - function beam in time and energy . unlike the simple model there , where each photon was assigned some positive time delay , the full monte carlo scattering calculation shows that some photons actually arrive _ earlier _ in time by taking a `` shortcut '' to the observer instead of waiting for the hot spot to come around and move towards the observer . and
of course , the photons are also spread out in energy due to the inverse - compton effects . as the optical depth increases , the well - defined curves in figure [ scatter_spectrogram75]a is smeared out into a nearly constant blur when @xmath76 , with a broad spectral peak as in figure [ powerlaw_cutoff ] . only a slight trace of the original coherent light curve remains , composed of roughly @xmath77 of the emitted photons that do not scatter before reaching the observer or get captured by the event horizon .
when @xmath78 , multiple scattering become more common , so photon shortcuts become rarer , tending to spread the light curve preferentially to the right ( delay in observer time ) , as seen in figures [ scatter_spectrogram75]c , d . by integrating over broad energy bands such as those typically used in _ rxte _ observations , we can increase our `` signal '' strength while sacrificing spectral resolution . for millisecond periods
, there will still not be nearly enough photons to provide phase resolution , but these features may show up statistically in the power spectrum or bispectrum @xcite .
figure [ lightcurves_tes ] shows a set of integrated light curves for a variety of optical depths .
the black hole and hot spot parameters are as in figure [ scatter_spectrogram75 ] , here assuming a thermal hot spot emitter with temperature @xmath79 kev . as the optical depth to electron scattering increases ,
the rms amplitude of each light curve decreases as the photons get smoothed out in time .
similarly , due to the average time delay added to each photon by the increased path length , the relative location of each peak is shifted later in time .
energy - integrated light curves for a hot spot with orbital parameters as in figure [ scatter_spectrogram75 ] and a range of coronal optical depths .
the emitted spectrum is assumed to be thermal with a hot spot temperature @xmath79 kev , integrated over @xmath80 kev in the observer s frame.,scaledwidth=50.0% ] in all likelihood , the relative phase shifts would be nearly impossible to detect , regardless of the instrument sensitivity , since to do so would require measuring the light curve from a single coherent hot spot at two different optical depths .
it is difficult to imagine a scenario where the coronal properties could change on such short time scales ( yet it is possible that a fixed hot spot on the surface of an x - ray pulsar might actually be used for this technique ) .
however , the higher harmonic peaks of the different light curves may in fact be measurable with the next - generation x - ray timing mission , or under extremely favorable conditions , even with _
rxte_. while the absolute peak shifts for hot spot light curves at different optical depths would probably not be detectable , the relative shifts of simultaneous light curves in different energy bands may be observable , at least on a statistical level with a cross - correlation analysis .
since the average scattering event boosts photons to higher energy bands and also causes a net time delay due to the added geometric path , the light curves in higher energy bands should be delayed with respect to the lower energy light curves .
a few of the typical energy bands used for _ rxte _ observations are @xmath81 , @xmath82 , and @xmath83 kev . to fully cover the peak emission from a thermal hot spot at 1 kev , we expand the lowest energy band in our calculations to cover @xmath84 kev .
the light curves in these three bands are plotted in figure [ lightcurves_eband ] for @xmath85 .
the low energy band resembles the unscattered light curve plus a roughly flat background , while the higher energy light curves show a much smaller modulation with a significant phase shift ( @xmath86 periods ) due to the additional photon path lengths .
hot spot light curves in a few different _ rxte _ energy bands ( we have expanded the lowest energy band down to 0.5 kev to include the thermal emission of a hot spot at @xmath79 kev ) .
the hot spot inclination is @xmath71 and the coronal properties are as in figure [ scatter_spectrogram75 ] .
the optical depth to scattering is @xmath87 in ( a ) and @xmath88 in ( b ) .
the higher energy light curves are made from photons that have experienced more scattering events , boosting their energy and delaying their arrival time.,title="fig:",scaledwidth=45.0% ] hot spot light curves in a few different _ rxte _ energy bands ( we have expanded the lowest energy band down to 0.5 kev to include the thermal emission of a hot spot at @xmath79 kev ) . the hot spot inclination is @xmath71 and the coronal properties are as in figure [ scatter_spectrogram75 ] .
the optical depth to scattering is @xmath87 in ( a ) and @xmath88 in ( b ) .
the higher energy light curves are made from photons that have experienced more scattering events , boosting their energy and delaying their arrival time.,title="fig:",scaledwidth=45.0% ]
the original motivation for the application of scattering to the hot spot model was to answer a few important questions raised by _
observations : * the distinct lack of power in higher harmonics at integer multiples of the peak frequencies . * the larger significance of high frequency qpo detections in the higher energy bands ( 6 - 30 kev ) relative to the signal in the lower energy band ( 2 - 6 kev ) . * the trend for these hfqpos to exist predominantly in the steep power law ( spl ) spectral state of the black hole . beginning with the final point
, it appears to be quite reasonable that the physical mechanism producing the power law region of the spectrum is the inverse - compton scattering of cool , thermal photons off of hot coronal electrons .
steep _ power law suggests a small - to - moderate value for the compton @xmath45 parameter , inferred from equation ( [ powerlaw_alpha ] ) , in the range @xmath89 . from equation ( [ compton_y ] ) , this suggests either a small optical depth or a small electron temperature . to gain insight into which of these two options
is more likely , we need to address the other two observational clues . in @xcite , we first proposed the scattering model as an explanation for harmonic damping . with the more careful treatment in this paper , we include not only the temporal , but also the spatial effects of electron scattering .
the photons originally beamed toward the observer are now scattered in the opposite direction , while the photons emitted away from the observer can now be scattered back to him .
this smoothes out the light curve in time more effectively than the localized convolution functions used in @xcite . at the same time , the scattering is not completely isotropic [ see eqn .
( [ cross_unpol ] ) ] , so some modulation remains .
thus , to maintain a significant modulation in the observed light curve , we require a relatively small optical depth , reducing the smoothing effects of the scattering .
the fact that most hfqpos appear more significantly in higher energy bands also points towards compton scattering off hot electrons .
however , as the calculations above show ( see fig.[lightcurves_eband ] ) , with the basic thermal disk / hot spot model , the light curves actually have _ smaller _ amplitude fluctuations in the higher energy bands , as these scattered photons get smoothed out more in time . furthermore , while the higher harmonic modes are successfully damped in the scattering geometry ,
so is the fundamental peak .
thus , in order to agree with observations , the hot spot overbrightness would need to be much higher than the values quoted in @xcite . based on these arguments alone
, we find it unlikely that the hfqpos are coming from a cool , thermal hot spot getting upscattered by a hot corona . from the photon continuum spectra of the spl state
, there appears to be a hot corona with @xmath90 kev , but as figure [ lightcurves_eband]a shows , the lowest energy photons , which presumably do not scatter in the corona , have by far the greatest amplitude modulations .
it is possible that the relative modulation would appear smaller due to the added flux from the rest of the cool , thermal disk , but much of this steady - state emission should also get scattered to higher energies , further damping the modulations in the @xmath91 kev bands .
the high luminosity of the spl state ( also called the very high state ) suggests that the thermal , slim disk geometry may not be appropriate here .
perhaps it is more likely that these cases correspond to an adaf model , traditionally associated with very low or very high accretion rates @xcite . since the adaf model can not radiate energy efficiently
, the gas in the innermost regions will be much hotter than in the slim disk paradigm .
thus _ hot _ hot spots with @xmath92 kev could be forming inside a small adaf coronal region , providing seed photons that are already in the higher energy bands , and are only moderately upscattered by the surrounding corona .
future work will focus on exploring other corona geometries and applying the monte carlo scattering code to other qpo models in order to better understand the emission processes that describe the spl spectral state .
baganoff , et al . ,
nature * 413 * , 45 ( 2001 ) .
bardeen , w.h .
press , & s.a .
teukolsky , apj * 178 * , 347 ( 1972 ) .
m. gierlinsky , et al .
, mnras * 288 * , 958 ( 1997 ) . m. gilfanov , et al . , apjs * 92 * , 411 ( 1994 ) .
w. heitler , _ the quantum theory of radiation _ , ( london : oxford university press ) ( 1954 ) .
maccarone & j.d .
schnittman , mnras in press , preprint ( astro - ph/0411266 ) ( 2005 ) .
mcclintock & r.a .
remillard , r. a. , in _ compact x - ray sources _ , ed .
w. h. g. lewin & m. van del klis ( new york : cambridge univ .
press ) , preprint ( astro - ph/0306213 ) ( 2004 ) .
r. narayan & i. yi , apj * 428 * , l13 ( 1994 ) .
pozdniakov , i.m .
sobol , & r.a .
sunyaev , sov .
* 21 * , 708 ( 1977 ) .
rybicki & a.p .
lightman , _ radiative processes in astrophysics _ ,
( new york : wiley - interscience ) ( 1979 ) .
schnittman & e. bertschinger , apj * 606 * , 1098 ( 2004 ) .
schnittman , apj in press , preprint ( astro - ph/0407179 ) ( 2005 ) .
sunyaev & j. truemper , nature * 279 * , 506 ( 1979 ) .
sunyaev & l.g .
titarchuk , a&a * 86 * , 121 ( 1980 ) .
titarchuk , apj * 434 * , 570 ( 1994 ) . | using a monte carlo ray - tracing code in full general relativity , we calculate the transport of photons from a geodesic hot spot emitter through a corona of hot electrons surrounding a black hole . each photon is followed until it is either captured by the black hole or is detected by a distant observer .
the source is assumed to be a low - energy thermal emitter ( @xmath0 kev ) , isotropic in the rest frame of a massive geodesic test particle .
the coronal scattering has two major observable effects : the comptonization of the photon spectrum due to the high energy electrons , and the convolution of the time - dependent light curve as each photon is effectively scattered into a different time bin .
both of these effects are clearly present in the _ rossi x - ray timing explorer ( rxte ) _ observations of high frequency quasi - periodic oscillations ( qpos ) seen in black hole binaries .
these qpos tend to occur when the system is in the steep power law spectral state and also show no evidence for significant power at higher harmonic frequencies , consistent with the smoothing out of the light curve by multiple random time delays .
we present simulated photon spectra and light curves and compare with _ rxte _ data , allowing us to infer the properties of the corona as well as the hot spot emitter . |
we consider _ join the shortest queue _ ( jsq ) networks , where incoming jobs " ( or customers " ) are assigned to the shortest queue from among @xmath0 distinct queues , @xmath14 , with these queues being chosen uniformly from among the @xmath1 queues in the system , with @xmath15 .
when two or more of these queues each have the fewest number of jobs , each of the queues is chosen with equal probability .
after completion of service at its queue , a job leaves the network .
we assume that jobs arrive according to a rate-@xmath2 poisson process , @xmath3 , and that jobs are served independently and at rate @xmath4 at each queue .
we are interested in this article in the case where the service discipline at each queue is first - in , first - out ( fifo ) .
when the service at queues is exponentially distributed , the evolution of the system is given by a countable state markov chain where a state is given by the number of jobs at each queue .
it is not difficult to show that a unique equilibrium distribution exists ; this equilibrium is exchangeable with respect to the ordering of the queues .
let @xmath16 denote the probability that there are at least @xmath17 jobs in equilibrium for the system with @xmath1 queues .
it was shown in vvedenskaya et al .
@xcite that @xmath18 in particular , the right tail of @xmath16 decays doubly exponentially fast in the limit as @xmath5 .
this behavior is a substantial improvement over the case @xmath6 , where @xmath16 decays exponentially , and has led to substantial interest in jsq networks in the literature . for other references ,
see azar et al .
@xcite , graham @xcite , luczak - mcdiarmid @xcite,@xcite , martin - suhov @xcite , mitzenmacher @xcite , suhov - vvedenskaya @xcite , vocking @xcite and vvedenskaya - suhov @xcite .
little work has been done on the behavior of jsq networks when the service times are not exponentially distributed . in this
setting , the underlying markov process will typically have an uncountable state space , and positive harris recurrence for the process is no longer obvious .
the latter was shown in foss - chernova @xcite , and uniform bounds on the equilibria were shown in bramson @xcite .
( both articles also considered jsq networks with more general arrivals and routing of jobs . )
an important first step in analyzing the limiting behavior of the equilibria , as @xmath7 , is to show that any fixed number of queues become independent of one another and converge to a limiting distribution .
moreover , one wishes to show that such a limiting distribution is the unique equilibrium for an associated markov process at a single queue , the _ environment process _ , which corresponds in an appropriate sense to setting @xmath19 and viewing the corresponding infinite dimensional process at a single queue .
we will refer to this equilibrium as the _ equilibrium environment_. in section 2 , we will precisely define this terminology .
although it seems that this independence should hold in a very general setting , including under a wide range of service disciplines , demonstrating it appears to be a difficult problem . in bramson
@xcite , this independence and the corresponding behavior for the environment process were stated as an ansatz .
this ansatz was demonstrated in bramson et al .
@xcite in several settings including for networks where the service discipline is fifo and the service distribution has a decreasing hazard rate . in this article
, we employ the restriction of the above ansatz to fifo networks .
this version of the ansatz will be precisely stated in section 2 . here
, we summarize it for application in the current section : @xmath20 although this ansatz has only been demonstrated for service distributions having decreasing hazard rate and for general service distributions with sufficiently small @xmath21 for arrivals , our arguments here do not otherwise require either restriction .
other applications of the ansatz , but for the processor sharing and lifo service disciplines , are given in @xcite . our goal , in this article , is to investigate the limiting behavior of the right tail of the associated equilibrium environment , under the fifo service discipline and with the assigned mean-@xmath4 service distribution @xmath22 . denote by @xmath23 the probability that there are at least @xmath17 jobs in the equilibrium environment .
we will show that , when @xmath22 has a power law tail with exponent @xmath8 , for given @xmath9 , the tail of @xmath23 exhibits a wide range of behavior depending on the relationship between @xmath10 and @xmath0 .
in particular , if @xmath11 , the tail is doubly exponential and , if @xmath12 , the tail has a power law ; when @xmath13 , the tail is exponentially distributed .
when @xmath24 , the coefficient @xmath25 of @xmath17 in the doubly exponential tail converges to @xmath4 , which is the coefficient of @xmath17 in ( [ eq1.1.1 ] ) .
one obtains the same coefficient of @xmath17 when @xmath22 has an exponential tail or has bounded support .
our main results are theorems [ thm1.2.1 ] , [ thm1.3.1 ] and [ thm1.4.1 ] .
theorem [ thm1.2.1 ] covers the case @xmath11 , theorem [ thm1.3.1 ] covers the case @xmath12 and theorem [ thm1.4.1 ] covers the case @xmath13 .
we set @xmath26 .
[ thm1.2.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds and that @xmath29 with @xmath30 .
then , @xmath31 for some @xmath32 .
moreover , @xmath25 is continuous in @xmath10 and @xmath33 when ( [ eq1.2.2 ] ) holds with @xmath34 , then ( [ eq1.2.3 ] ) holds with @xmath35 . theorem [ thm1.2.1 ]
implies that , when @xmath36 as @xmath37 , for @xmath30 and @xmath38 , then @xmath39 .
[ thm1.3.1 ] consider a family of jsq networks as in theorem [ thm1.2.1 ] , with ( [ eq1.2.2 ] ) instead holding for @xmath40 .
then @xmath41 . \ ] ] theorem [ thm1.3.1 ] implies that , when @xmath36 as @xmath37 , for @xmath40 and @xmath38 , then @xmath42 , where @xmath43 is the right hand side of ( [ eq1.3.2 ] ) . note that @xmath44 as @xmath45 and @xmath46 as @xmath47
[ thm1.4.1 ] consider a family of jsq networks as in theorem [ thm1.2.1 ] , with ( [ eq1.2.2 ] ) replaced by @xmath48 for some @xmath49 .
then , for appropriate @xmath50 and @xmath51 , @xmath52 where @xmath53 theorem [ thm1.4.1 ] implies that when @xmath54 as @xmath37 , then @xmath23 decreases exponentially fast in the sense of ( [ eq1.4.3 ] ) .
because of ( [ eq1.4.4 ] ) , the exponent depends strongly on the choice of @xmath55 .
we note that the proofs of theorems [ thm1.2.1]-[thm1.4.1 ] only depend on ( [ eq1.1.2 ] ) for the existence of an equilibrium environment .
regardless of how the existence of an equilibrium environment is verified , ( [ eq1.1.2 ] ) is needed in order to relate the tail behavior of @xmath23 for the equilibrium environment to the tail behavior for the equilibria of the corresponding family of networks as @xmath7 .
this article is organized as follows . in section 2 , we provide basic background on the properties of the state space and markov process that underlie the jsq networks .
we then define equilibrium environments and formally state the ansatz . in sections 3 - 5
, we demonstrate theorems [ thm1.2.1 ] , [ thm1.3.1 ] and [ thm1.4.1 ] , respectively .
our approach will be to demonstrate lower bounds and then upper bounds that yield the theorem . in each case
, the lower bounds will be considerably easier to show . for the reader s convenience ,
we mention here some of the notation in the paper .
we will employ @xmath56 to denote positive constants whose precise value is not of importance to us . for @xmath57 , @xmath58 and @xmath59 will denote the integer part of @xmath60 , respectively , the smallest integer at least as large as @xmath60 .
in this section , we provide a more detailed description of the construction of the markov processes @xmath61 that underlie the jsq networks .
we next define the corresponding cavity process and its equilibrium environment .
we then employ these concepts to state the ansatz for jsq networks .
most of this material is included in sections 2 and 3 of bramson et al .
( related material is also given in @xcite and @xcite . )
we define the state space @xmath62 to be the set @xmath63 the first coordinate @xmath64 , @xmath65 , corresponds to the number of jobs at the @xmath66th queue ; the second coordinate @xmath67 , @xmath68 , is the amount of time the oldest job there has already been served ; and the last coordinate @xmath69 , @xmath70 , is the residual service time . when @xmath71 , set the other two coordinates equal to @xmath72 . the coordinate @xmath67 will not play a role in the evolution of @xmath61 here ; we retain it for comparison with @xcite , where it was used to demonstrate ( [ eq1.1.2 ] ) under decreasing hazard rates .
( we will employ slightly different notation here than in @xcite . ) for given @xmath73 , @xmath74 is the _ projection _ of @xmath62 obtained by restricting @xmath62 to the first @xmath75 queues ; for @xmath76 , @xmath77 is thus obtained by omitting the coordinates with @xmath78 .
one can also define projections of @xmath62 onto spaces @xmath74 corresponding to other subsets of @xmath79 analogously , although these are not needed here .
we define the metric @xmath80 on @xmath62 , with @xmath80 given in terms of @xmath81 by @xmath82 . for given @xmath83 , with the coordinates
labelled correspondingly , set latexmath:[\[\label{eq2.1.2 } d^{(n),n}(x_1,x_2)= |z_1^{n } - z_2^{n}| + |u_1^{n } - u_2^{n}| + @xmath80 is separable and locally compact ; more detail is given on page 82 of @xcite .
we equip @xmath62 with the standard borel @xmath85-algebra inherited from @xmath80 , which we denote by @xmath86 .
the markov process @xmath87 , @xmath88 , underlying a given model is defined to be the right continuous process with left limits , taking values @xmath89 in @xmath62 , whose evolution is determined by the model together with the assigned service discipline .
we denote the random values of the coordinates @xmath64 , @xmath67 and @xmath69 taken by @xmath87 , by @xmath90 , @xmath91 and @xmath92 .
jobs are allocated service according to the fifo discipline ; during the period a job is being served , @xmath91 increases at rate @xmath4 and @xmath92 decreases at rate @xmath4 . along the lines of page 85 of @xcite , a filtration @xmath93 , @xmath94 $ ] , can be assigned to @xmath61 so that @xmath61 is a piecewise - deterministic markov process , and hence is borel right .
this implies that @xmath61 is strong markov .
( we do not otherwise use borel right . )
the reader is referred to davis @xcite for more detail . in order to state the ansatz , we require some terminology .
we denote by @xmath95 the projection of the equilibrium measure @xmath96 of the @xmath1-queue system onto the first @xmath75 queues .
( since @xmath87 is exchangeable when @xmath97 is , the choice of queues will not matter . )
we wish to describe the evolution of individual queues for the limiting process , as @xmath5 . for this
, we construct a strong markov process @xmath98 , @xmath99 , on @xmath100 as follows .
let @xmath101 denote a probability measure on @xmath100 , which we refer to as the _ environment _ of the process @xmath102 ; we refer to @xmath102 as the associated _ cavity process_. we define @xmath102 so that _ potential arrivals _ arrive according to a rate@xmath103 poisson process .
when such a potential arrival to the queue occurs at time @xmath104 , @xmath105 is compared with the states of @xmath106 independent random variables , each with law @xmath101 ; we refer to these @xmath106 states at a potential arrival as the _ comparison states_. choosing from among these @xmath0 states , the job is assigned to the state with the fewest number of jobs .
( in case of a tie , each of these states is chosen with equal probability . )
if the job has chosen the state @xmath105 at the queue , it then immediately joins the queue ; otherwise , the job immediately leaves the system . in either case ,
the independent @xmath106 states employed for this purpose are immediately discarded .
jobs in the cavity process @xmath102 have the same service distribution @xmath22 as in the queueing network and are served according to the fifo service discipline .
the number of jobs in @xmath98 will be denoted by @xmath107 , the amount of time the oldest job has already been served by @xmath108 and the residual service time by @xmath109 ; we will employ @xmath89 , @xmath60 , @xmath110 and @xmath111 for the corresponding terms in the state space . when a cavity process @xmath102 , with environment @xmath101 , is stationary with the equilibrium measure @xmath101 ( i.e. , @xmath98 has the distribution @xmath101 for all @xmath104 ) , we say that @xmath101 is an _
equilibrium environment_. one can think of an equilibrium environment as being the restriction of an equilibrium measure for the jsq network , viewed at a single queue , when the total number of queues @xmath1 is infinite " .
we now state the ansatz . here , @xmath112 on @xmath74 denotes convergence in total variation with respect to the metric @xmath113 on @xmath74 .
[ ansatz ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
then , ( a ) for each @xmath75 , @xmath114 where @xmath115 is the @xmath75-fold product of @xmath116 . moreover , ( b ) @xmath116 is the unique equilibrium environment associated with this family of networks .
as was mentioned in the introduction , this ansatz was demonstrated in bramson et al .
@xcite when the service time distribution @xmath22 has a decreasing hazard rate @xmath117 ( i.e. , @xmath118 is nonincreasing in @xmath111 ) and for general service distributions when the arrival rates are small enough . in order to demonstrate theorems [ thm1.2.1]-[thm1.4.1 ] , we will analyze the cavity process @xmath102 with its unique equilibrium environment @xmath119 .
in particular , we will analyze @xmath116 over a _
cycle _ starting and ending at the state @xmath72 .
( the state where the number of jobs @xmath60 is @xmath72 . )
letting @xmath120 denote the time at which @xmath102 first returns to @xmath72 after visiting another state , the first cycle is the random time interval @xmath121 $ ] . for any @xmath122 , we will denote by @xmath123 the _ occupation time _ at states @xmath89 , with @xmath124 , over @xmath121 $ ] , that is , @xmath125 setting @xmath126 $ ] , the mean return time to @xmath72 , one has @xmath127 , \ ] ] where @xmath23 is the probability there are at least @xmath17 jobs in the equilibrium environment .
let @xmath128 denote the arrival rate of jobs for @xmath102 , when @xmath129 . in order for a potential arrival to arrive at the queue , it is necessary for all of the @xmath106 comparison states used at that time to be at least @xmath17 , in which case the probability of selecting the queue is the reciprocal of the number of states equal to @xmath17 .
this gives the bounds @xmath130 since the departure of jobs from the queue is deterministic , being a function of the residual service time @xmath111 , ( [ eq2.3.4 ] ) gives a reasonably explicit description of the transition rates for @xmath102 .
together with ( [ eq2.3.3 ] ) , ( [ eq2.3.4 ] ) will provide the basis for our demonstration of theorems [ thm1.2.1]-[thm1.4.1 ] and will be used throughout the paper .
in this section , we demonstrate theorem [ thm1.2.1 ] ; we do this by demonstrating lower and upper bounds that are needed for the theorem in propositions [ prop3.1.1 ] and [ prop3.6.1 ] .
each of these bounds is expressed in terms of a recursion relation for @xmath23 . in order to obtain theorem [ thm1.2.1 ] from these recursions , we employ proposition [ prop3p.1.1 ] , which analyzes such recursions by utilizing a standard framework involving rational generating functions . the section is organized as follows . after stating propositions [ prop3.1.1 ] and [ prop3.6.1 ] , we state and
prove proposition [ prop3p.1.1 ] .
we next employ the three propositions to demonstrate theorem [ thm1.2.1 ] .
we then provide the relatively quick proof of proposition [ prop3.1.1 ] and the longer proof of proposition [ prop3.6.1 ] , in the following subsections .
in both propositions , we set @xmath131 ( or , equivalently , @xmath132 ) and @xmath133 .
[ prop3.1.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds .
then , for appropriate @xmath134 and all @xmath17 , @xmath135 if moreover , for some @xmath136 , @xmath137 with @xmath30 , then , for appropriate @xmath134 and all @xmath17 , @xmath138 [ prop3.6.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds and that , for some @xmath136 , @xmath139 with @xmath30 .
if @xmath10 is not an integer , then , for appropriate @xmath140 and all @xmath17 , @xmath141 if @xmath10 is an integer , then , for each @xmath142 , appropriate @xmath140 and all @xmath17 , @xmath143 to employ the recursions in ( [ eq3.1.6 ] ) and ( [ eq3.6.3])([eq3.6.4 ] ) of propositions [ prop3.1.1 ] and [ prop3.6.1 ] in the proof of theorem [ thm1.2.1 ] , we will analyze the asymptotic behavior of the recursions in ( [ eq3p.1.2 ] ) .
[ prop3p.1.1 ] suppose that @xmath144 satisfies @xmath145 with @xmath146 for @xmath147 , @xmath27 and @xmath148 $ ] .
then , setting @xmath149 , @xmath150 for some @xmath151 .
moreover , @xmath152 is continuous in @xmath10 and @xmath153 exponentially fast as @xmath24 .
the recurrence ( [ eq3p.1.2 ] ) is a special case of linear recursions of the form @xmath154 with @xmath155 and general @xmath156 .
it is well known that ( see , e.g. , stanley @xcite , page 202 ) @xmath157 for each @xmath17 , where @xmath158 are distinct , @xmath159 is a polynomial in @xmath17 of degree strictly less than @xmath160 , and @xmath161 with @xmath162 .
moreover the converse holds , that is , if ( [ eq3p.2.1 ] ) and ( [ eq3p.2.2 ] ) both hold , then so does ( [ eq3p.2.0 ] ) . for @xmath144 given by ( [ eq3p.1.2 ] ) , it is not difficult to check that there is exactly one value @xmath158 , say @xmath163 , that is real and positive , that @xmath163 varies continuously in @xmath164 , and moreover that @xmath163 satisfies @xmath165 , since @xmath166 and @xmath167 .
( descartes rule of signs in fact implies that @xmath168 is a simple root . ) also , because @xmath166 , and possesses both odd and even indices , @xmath169 for @xmath170 .
since the initial data given below ( [ eq3p.1.2 ] ) are all positive , any solution of ( [ eq3p.1.2 ] ) is majorized by this particular solution , up to a multiplicative constant ; so , @xmath171 .
the limit in ( [ eq3p.1.3 ] ) , with @xmath172 , follows from these observations .
we still need to examine the limiting behavior of @xmath25 as @xmath173 . dividing both sides in ( [ eq3p.1.2 ] ) by @xmath144 , then substituting ( [ eq3p.2.1 ] ) for each of the terms , and letting @xmath174 implies that @xmath175 for @xmath176 .
this again uses @xmath177 for @xmath170 .
hence , @xmath178 note that @xmath179 and that , since @xmath25 is increasing in @xmath10 , @xmath89 is decreasing in @xmath10 . since the right hand side goes to @xmath72 exponentially fast as @xmath180 , and hence as @xmath24 , it follows that @xmath181 exponentially fast as @xmath24 , which also implies @xmath182 exponentially fast , as desired .
note that the precise exponential rate of convergence can be obtained by inserting this limit back into the right hand side of ( [ eq3p.2.3 ] ) . applying proposition [ prop3p.1.1 ] to propositions [ prop3.1.1 ] and [ prop3.6.1 ] , we now demonstrate theorem [ thm1.2.1 ] . setting @xmath183 , where @xmath144 is given in ( [ eq3p.1.2 ] )
, one has @xmath184 with @xmath185 for @xmath186 .
we proceed to compare @xmath187 with @xmath188 , where @xmath23 satisfies one of ( [ eq3.1.6 ] ) , ( [ eq3.6.3 ] ) and ( [ eq3.6.4 ] ) .
comparison of @xmath187 with @xmath188 , with @xmath189 , @xmath190 and @xmath23 satisfying ( [ eq3.1.6 ] ) , provides an upper bound on the limit in ( [ eq1.2.3 ] ) . to see this , we first set @xmath191 , for given @xmath192 .
since @xmath193 , by substituting into ( [ eq3q.1.1 ] ) , one can check that , for large enough @xmath194 and @xmath17 , @xmath195 for any fixed choice of @xmath196 and @xmath197 , in particular , for @xmath198 and @xmath199 , where @xmath200 is chosen as in proposition [ prop3.1.1 ] .
moreover , on account of ( [ eq3p.1.3 ] ) , @xmath201 where , in particular , @xmath202 , and hence @xmath203 as @xmath174 .
we observe that @xmath188 satisfies the inequality that is analogous to that for @xmath23 in ( [ eq3.1.6 ] ) , but with the inequality reversed and prefactors @xmath204 instead of @xmath205 .
comparing @xmath206 with @xmath188 therefore implies that , for large enough @xmath66 not depending on @xmath17 , @xmath207 the upper bound for ( [ eq1.2.3 ] ) therefore follows from ( [ eq3q.1.3 ] ) for the same choice of @xmath25 , which we recall is continuous in @xmath10 .
the limit in ( [ eq1.2.4 ] ) also follows from proposition [ prop3p.1.1 ] .
comparison of @xmath187 with @xmath188 also provides a lower bound on the limit in ( [ eq1.2.3 ] ) . in the case where @xmath10 is nonintegral , we choose @xmath164 and @xmath208 as before , with @xmath189 , @xmath190 ; note that @xmath23 satisfies the upper bound in ( [ eq3.6.3 ] ) .
we proceed as in the first part , but instead set @xmath209 , for given @xmath192 .
one can check that , for large enough @xmath194 and @xmath17 , @xmath210 for any choice of @xmath211 and @xmath212 . as before , ( [ eq3q.1.3 ] ) holds .
the terms @xmath188 satisfy the inequality that is the analog of ( [ eq3.6.3 ] ) . also , @xmath213 as @xmath174 .
comparing @xmath206 with @xmath188 therefore implies that , for large enough @xmath66 not depending on @xmath17 , @xmath214 the lower bound for ( [ eq1.2.3 ] ) therefore follows from ( [ eq3q.1.3 ] ) when @xmath10 is nonintegral . the reasoning in the case where @xmath10 is integral is similar , but with the difference that we now choose @xmath215 , @xmath216 , where @xmath217 is arbitrary .
now , @xmath23 satisfies the upper bound in ( [ eq3.6.4 ] ) .
we proceed as in the nonintegral case , once again obtaining ( [ eq3q.1.4 ] ) . comparing @xmath188 with @xmath206
again produces ( [ eq3q.1.3 ] ) , except that the limit is now @xmath218 because of our choice of @xmath164 . by proposition [ prop3p.1.1 ] , @xmath219 is continuous in its argument .
therefore , letting @xmath220 produces the same limit as in the nonintegral case , and hence implies the lower bound for ( [ eq1.2.3 ] ) in the case where @xmath10 is integral . we still need to demonstrate that when ( [ eq1.2.2 ] ) holds with @xmath34 , then ( [ eq1.2.3 ] ) holds with @xmath221 .
the lower bound in ( [ eq1.2.3 ] ) holds on account of ( [ eq1.2.4 ] ) .
the upper bound is not difficult to show and does not require proposition [ prop3p.1.1 ] ; we proceed to show the bound .
we will show by induction that , for all @xmath17 , @xmath222 where @xmath200 is as chosen as in ( [ eq3.1.2 ] ) , which we assume wlog is at most @xmath4 . to see ( [ eq3q.1.6 ] ) , note that if it holds for all @xmath223 then this , together with ( [ eq3.1.2 ] ) , implies that @xmath224^{d-1 } \\
& \ge ( { c_{1}}/8k)^{(k-1)(d^k -1 ) + k } \ge ( { c_{1}}/8k)^{k d^k}. \end{split}\ ] ] the upper bound in ( [ eq1.2.3 ] ) , with @xmath35 , follows immediately from ( [ eq3q.1.6 ] ) .
the proof of proposition [ prop3.1.1 ] is quick . to obtain the lower bounds in both ( [ eq3.1.2 ] ) and ( [ eq3.1.6 ] )
, it suffices to construct a path along which @xmath107 increases from @xmath72 to @xmath17 within the first cycle .
this is done , in both cases , by allocating the same amount of time to each of the first @xmath17 arrivals , which are also required to occur before the first departure .
consider the cavity process @xmath102 with @xmath225 . in order to show ( [ eq3.1.2 ] ) and ( [ eq3.1.6 ] )
, we obtain lower bounds on the expected amount of time @xmath226 $ ] over which @xmath227 before @xmath102 returns to @xmath72 .
we first show ( [ eq3.1.2 ] ) .
we consider the event @xmath228 where the first service time @xmath229 is at least @xmath230 and the first @xmath17 arrivals occur by time @xmath231 .
the latter event contains the event where each of the first @xmath17 arrivals occurs not more than @xmath232 units of time after the previous arrival , starting at time @xmath72 .
conditioned on there being @xmath233 jobs in the queue , jobs arrive at rate @xmath234 , and so the probability of such an arrival occurring over an interval of length @xmath232 is at least @xmath235 .
so , given that @xmath236 , the probability that all @xmath17 of these arrivals occur by time @xmath231 is at least @xmath237 the event @xmath238 occurs with some positive probability @xmath55 depending on @xmath22 and , under the event @xmath228 , the departure time for the first job occurs at least @xmath231 after the last of the first @xmath17 arrivals .
so , the expected amount of time in @xmath239 $ ] , during which @xmath227 and before @xmath102 has returned to @xmath72 , is at least @xmath240 which is therefore a lower bound for @xmath226 $ ] .
it therefore follows from ( [ eq2.3.3 ] ) that @xmath241 which implies ( [ eq3.1.2 ] ) for appropriate @xmath200 .
we next show ( [ eq3.1.6 ] ) under the assumption ( [ eq3.1.5 ] ) . for this
, we set @xmath242 one can reason analogously as through ( [ eq3.7.2 ] ) , but by replacing the time interval @xmath243 $ ] by @xmath244 $ ] and employing @xmath245 for the allotted time for each of the @xmath17 arrivals .
one obtains that the expected amount of time in @xmath246 $ ] , during which @xmath227 and before @xmath102 has returned to @xmath72 , is at least @xmath247 choose @xmath17 large enough so that @xmath248 , where @xmath249 is as in ( [ eq3.1.5 ] ) and @xmath250 is as in ( [ eq3.7.4 ] ) .
since @xmath251 for @xmath252 , this is at least @xmath253 where the inequality follows from ( [ eq3.7.4 ] ) and @xmath254 .
consequently , @xmath255 \ge 2^{-k}(\alpha/4k)^{\beta}\left(\prod_{i = k_1 + 1}^{k-1}p_{i}^{d-1}\right ) p_{k_1}^{\hat{\beta}(d-1)}.\ ] ] again applying ( [ eq2.3.3 ] ) , it follows that , for large enough @xmath17 ( depending on @xmath21 and @xmath10 ) , @xmath256 which implies ( [ eq3.1.6 ] ) . in order to demonstrate proposition [ prop3.6.1 ]
, we will employ lemma [ lem3.9.1 ] below ; the lemma will also be employed in the demonstration of propositions [ prop4.4.1 ] and [ prop5.3.1 ] .
lemma [ lem3.9.1 ] provides upper bounds involving @xmath257 , @xmath258 and @xmath259 , for @xmath122 , @xmath260 and @xmath261 , which we define as follows . for @xmath262
, @xmath257 is the expected return time of the cavity process @xmath102 to the empty state @xmath72 , from @xmath263 with @xmath264 and @xmath265 .
we set @xmath266 , which is also the expected return time to @xmath72 just after departure of a job , but without knowledge of the residual service time of the job that is beginning service .
the quantity @xmath258 is the number of jobs , for this process , at the time when the @xmath267st job has just departed , e.g. , @xmath268 is the number of jobs just after departure of the job originally in service .
the stopping time @xmath259 is the first time @xmath66 at which @xmath269 .
we also denote by @xmath270 the service time of the @xmath267st job ( with @xmath271 being the service time of the job originally in service ) , and set @xmath272 .
note that @xmath273 are i.i.d . with distribution function @xmath22 , which , as always , is assumed to have mean @xmath4 . [ lem3.9.1 ]
let @xmath274 and @xmath275 be defined as above .
then , for large enough @xmath276 , @xmath277 and @xmath278 \le 2(k+s/2 + n_0 ) , \ ] ] for all @xmath17 and @xmath111 . it is not difficult to see that ( [ eq3.9.2 ] ) follows from ( [ eq3.9.3 ] ) . by applying wald s equation to @xmath279 and @xmath275 ( with respect to the underlying @xmath85-algebra generated by @xmath102 ) ,
one obtains @xmath280 = e\left[\sum_{n=1}^{\rho(k , s)}y_n\right ] + s = e[\rho(k , s)]e[y_1 ] + s \le 2(k+s+n_0 ) , \ ] ] with the inequality following from ( [ eq3.9.3 ] ) and @xmath281 = 1 $ ] . in order to show ( [ eq3.9.3 ] )
, we consider the process @xmath282 for appropriate choices of @xmath283 and @xmath284 , we claim @xmath285 is a supermartingale , with respect to the filtration @xmath286 , after restricting to times @xmath66 , with @xmath287 , and then stopping the process .
these three constants are chosen as follows .
we choose @xmath288 large enough so that @xmath289 . for @xmath290 , one can check that the supermartingale inequality @xmath291 \le m(n ) \ ] ] is satisfied the arrival rate of jobs is at most @xmath230 over the time interval @xmath292 $ ] during which the @xmath267st job is served , which has mean length @xmath4 , and so @xmath293 \le h(n ) - 1/2.\ ] ] in order to analyze @xmath294 when @xmath295 , we set @xmath296 we choose @xmath297 large enough so that , for some @xmath298 and all @xmath299 , @xmath300 \le m_1(n ) - \epsilon . \ ] ] this requires a standard computation using the convexity of the exponential function and the upper bound @xmath301 on the arrival rate of jobs .
( since @xmath302 may have positive drift , @xmath297 may need to be chosen large . )
we also choose @xmath303 so that @xmath304 .
together with ( [ eq3.10.3 ] ) , this implies ( [ eq3.10.2 ] ) also holds for @xmath295 .
consequently , @xmath285 is a supermartingale , as claimed . in order to demonstrate ( [ eq3.9.3 ] )
, we will apply the optional sampling theorem to @xmath305 stopped at times @xmath306 . first note that @xmath307 \le e[h(0 ) ] \le k + s/2 \ ] ] for @xmath308 , since the arrival rate of jobs is bounded above by @xmath230 .
also , for given @xmath111 , @xmath309 $ ] is increasing as a function of @xmath17 , the number of jobs in the cavity process at time @xmath72 .
together with ( [ eq3.10.4 ] ) , this implies that , for all @xmath17 , @xmath310 \le ( k\vee k_0 ) + s/2 \le k + s/2 + k_0 . \ ] ] since the supermartingale @xmath305 is bounded from below , application of the optional sampling theorem to @xmath311 implies that @xmath312 \le e[m(0 ) ] \le k + s/2 + k_0,\ ] ] and hence @xmath313 \le k + s/2 + k_0 + n_1 - e[\rho_n(k , s)]/2.\ ] ] solving for @xmath314 $ ] implies @xmath315 \le 2(k + s/2 + k_0 + n_1 ) = 2(k + s/2 + n_0)\ ] ] for @xmath316 .
letting @xmath317 implies ( [ eq3.9.3 ] ) .
lemma [ lem3.9.1 ] provides an upper bound on the expected time over a cycle during which there are at least @xmath17 jobs , provided such a state has already been attained .
below , we will obtain an upper bound on the probability of attaining such a state and combine this with ( [ eq3.9.2 ] ) . in order for @xmath102 , starting at @xmath72 , to attain a state with @xmath17 jobs
, it must first attain states with @xmath318 jobs , where @xmath319 has been specified in the previous subsection .
( it turns out that including states with fewer jobs in this sequence will not improve our bounds . )
we let @xmath320 denote the number of jobs that have already departed when such a state is first attained ( e.g. , @xmath321 means that the first job is still being served at the time @xmath104 when @xmath322 first occurs ) .
one trivially has @xmath323 partition @xmath324 so that @xmath325 are in the same subset if @xmath326 , i.e. , the times @xmath327 and @xmath328 at which @xmath329 and @xmath330 first occur are in the same service time interval .
one can write such a partition as @xmath331 with @xmath332 and @xmath333 , when the partition consists of @xmath334 sets ( where @xmath334 is random ) .
we denote by @xmath335 the set of all such partitions and by @xmath336 an element in the set , with the notation @xmath337 being used when convenient .
we will say that a partition @xmath338 occurs during a cycle when the corresponding sequence of events occurs , and denote by @xmath339 the event associated with the partition .
for each of the sets in ( [ eq3.11.1 ] ) except the last , there is a corresponding service interval , @xmath340 , with @xmath341 , at the beginning of which there are strictly less than @xmath342 jobs and at the end exactly @xmath343 jobs .
( since such an interval ends with a departure , the number of jobs at the beginning of the next service interval must be one less , which requires the cavity process to retrace some of its steps " before the number of jobs reaches @xmath343 again . ) for @xmath344 , there may be strictly more than @xmath17 jobs at @xmath345 ; instead , we consider the restricted interval @xmath346 $ ] , where @xmath347 is the first time at which there are at least @xmath17 jobs .
unlike at the end of the other intervals @xmath340 , the residual service time @xmath111 will not be @xmath72 .
when @xmath111 is large , this will increase the occupation time where @xmath348 , which will require us to exercise some care with our computations .
since @xmath349 , the number of distinct partitions in ( [ eq3.11.1 ] ) is at most @xmath350 . in proposition [ prop3.12.1 ] below , we compute an upper bound on @xmath23 using an upper bound on the expected occupation time corresponding to each partition , and then by multiplying by @xmath350 .
the upper bound in ( [ eq3.12.2 ] ) includes a factor @xmath351 obtained by employing lemma [ lem3.9.1 ] repeatedly .
the form of the bounds in ( [ eq3.12.2 ] ) and ( [ eq3.12.5 ] ) varies in different ranges of @xmath111 ; we will therefore find it useful to employ the notation @xmath352 . \ ] ] ( @xmath353 implicitly depends on the partition @xmath338 through @xmath342 and @xmath343 . )
we will employ @xmath354 when @xmath233 goes from @xmath319 to @xmath355 , which corresponds to the trivial partition in ( [ eq3.11.1 ] ) consisting of a single set . in the proof of proposition [ prop3.12.1 ] , we will use the following elementary chebyshev integral inequality , which states that , if @xmath356 and @xmath357 are both integrable functions that are increasing in @xmath111 , then , for any distribution function @xmath22 , @xmath358 [ prop3.12.1 ] consider a family of jsq networks , with the same assumptions holding as in proposition [ prop3.6.1 ] , except that ( [ eq3.6.2 ] ) is not assumed . then , for large enough @xmath17 , @xmath359 we first claim that the probability of the cavity process @xmath102 , with @xmath360 and @xmath361 , attaining @xmath343 jobs before time @xmath111 is at most @xmath362 \\ & = l_{\ell}(s ) .
\end{split}\ ] ] under this event , arrivals must occur sequentially over @xmath363 $ ] at times @xmath327 when @xmath364 , for @xmath365 , and the rate of such arrivals is at most @xmath366 . since there is at most time @xmath111 for each arrival ,
multiplying the corresponding upper bounds on the probability of an arrival at each step gives the first bound in ( [ eq3.12.5 ] ) .
the following inequality is then obtained by applying the inequality @xmath367 .
recall that @xmath123 denotes the occupation time over a cycle when @xmath348 . in order for @xmath368
, the event @xmath339 must occur for some @xmath336 ; hence @xmath226 = \sum_{\pi \in \pi_k } e[v_k;\ , a_{\pi}]$ ] .
we claim that , for any partition @xmath336 and large enough @xmath17 , @xmath369 \le ( 3k ) ^{m_{\pi } } \prod_{\ell = 1 } ^{m_{\pi}-1 } \left(\int_{0}^{\infty}l_{\ell}(s)\,f(ds)\right ) \times \\ & \qquad \qquad \qquad\qquad\qquad\quad\times 3\int_{0}^{\infty}(k+s)l _ { m_{\pi } } ( s)\,f(ds ) .
\end{split}\ ] ] to obtain ( [ eq3.12.6 ] ) , we argue by induction , applying ( [ eq3.12.5 ] ) at each step . it suffices to show that , for each step with @xmath370 , one obtains an additional factor @xmath371 and , for @xmath372 , one obtains the factor @xmath373 . for @xmath374 ,
the factor @xmath375 is obtained by applying ( [ eq3.9.3 ] ) , with @xmath376 , which gives an upper bound on the expected number of service intervals occurring over the remainder of the cycle , after the service interval corresponding to the @xmath377st step ends ; also , @xmath378 , which equals the expected number of service intervals at the beginning of the cycle . the other factor is obtained from ( [ eq3.12.5 ] ) by integrating against @xmath22 and , for @xmath372 , by employing ( [ eq3.9.2 ] ) to provide an upper bound on the expected occupation time @xmath123 , again employing ( [ eq3.12.5 ] ) and then integrating against @xmath22 . on the other hand , by repeatedly applying the chebyshev integral inequality ( [ eq3.11.5 ] ) to ( [ eq3.12.6 ] )
, it follows that , for an arbitrary partition in ( [ eq3.11.1 ] ) , ( [ eq3.12.6 ] ) is maximized for the trivial partition .
that is , for any partition @xmath336 , the quantity in ( [ eq3.12.6 ] ) is bounded above by @xmath379 since @xmath380 , it follows from ( [ eq3.12.6 ] ) and ( [ eq3.12.7 ] ) that @xmath381 = m_0^{-1}\sum_{\pi \in \pi_k}e[v_k;\ , a_{\pi } ] \\ & \le 3m_0^{-1 } ( 6k)^{\beta } \int_{0}^{\infty}(k+s)l(s)\,f(ds ) , \end{split}\ ] ] which implies ( [ eq3.12.2 ] ) we now complete the proof of proposition [ prop3.6.1 ] .
we employ the upper bound for @xmath23 given by ( [ eq3.12.2 ] ) for large enough @xmath17 . the integral in ( [ eq3.12.2 ] )
is bounded above by @xmath382 by integrating by parts and absorbing the first term into the second ; note that @xmath354 is increasing in @xmath111 on account of ( [ eq3.11.3 ] ) .
we decompose this last integral using intervals of the form @xmath383 , @xmath384 , for @xmath385 , and @xmath386 ; we need to consider the cases where @xmath10 is and is not an integer separately .
suppose that @xmath10 is not an integer . applying ( [ eq3.11.3 ] ) to the above integral over @xmath383 , one has the upper bound @xmath387 for @xmath385
, one has , over @xmath384 , the upper bounds @xmath388 for the last interval @xmath386 , one has the upper bound @xmath389 where we recall that @xmath390 . note that the lower limits of integration supply the dominant term in ( [ eq3.14.1 ] ) and ( [ eq3.14.2 ] ) , whereas the upper limit supplies the dominant term in ( [ eq3.14.3 ] ) , because of the choice of @xmath319 .
since @xmath391 is decreasing in @xmath233 , if one ignores the coefficients not involving powers of @xmath391 on the right hand sides of ( [ eq3.14.1])([eq3.14.3 ] ) , the largest bounds in ( [ eq3.14.1])([eq3.14.3 ] ) are given in ( [ eq3.14.2 ] ) , with @xmath392 , and in ( [ eq3.14.3 ] ) , in each case by the powers of @xmath391 , @xmath393 the coefficients of these powers are bounded above by terms not involving @xmath17 . employing ( [ eq3.12.2 ] ) of proposition [ prop3.12.1 ] , together with ( [ eq3.13.1 ] ) , one obtains the bound ( [ eq3.6.3 ] ) for @xmath23 , for appropriate @xmath140 and all @xmath17 .
when @xmath10 is an integer , the computations are similar .
the inequalities in ( [ eq3.14.1 ] ) and ( [ eq3.14.3 ] ) are the same as before , as are all of the cases in ( [ eq3.14.2 ] ) except for @xmath392 . rather than ( [ eq3.14.2 ] ) ,
one obtains the following inequality when @xmath392 : @xmath394 by comparing terms involving @xmath391 and ignoring the other coefficients , one can check that the largest bound is given in ( [ eq3.14.5 ] ) . since the logarithm term there is dominated by @xmath395 , for given @xmath142 and small enough @xmath396
, it follows that ( [ eq3.6.4 ] ) holds for @xmath23 , for appropriate @xmath140 and all @xmath17 .
in this section , we demonstrate theorem [ thm1.3.1 ] . we do this by demonstrating the lower and upper bounds needed for the theorem in propositions [ prop4.1.1 ] and [ prop4.4.1 ] . here
, we set @xmath398 $ ] .
[ prop4.1.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds and that @xmath399 with @xmath400 and some @xmath136 .
then , for appropriate @xmath401 and all @xmath17 , @xmath402 [ prop4.4.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds and that @xmath403 with @xmath400 and some @xmath136 .
then , for each @xmath142 , appropriate @xmath404 , and all @xmath17 , @xmath405 theorem [ thm1.3.1 ] follows immediately from propositions [ prop4.1.1 ] and [ prop4.4.1 ] upon letting @xmath220 in ( [ eq4.4.3 ] ) . as in section 3
, the demonstration of the lower bound is much quicker than that of the upper bound .
we first demonstrate the lower bound , proposition [ prop4.1.1 ] , and then , in the remainder of the section , derive the upper bound , proposition [ prop4.4.1 ] . as in section 3 , when we considered the case where @xmath11 , for the lower bound , it suffices to construct a path along which @xmath107 increases from @xmath72 to @xmath17 within the first cycle .
as before , we allocate the same amount of time for each of the first @xmath17 arrivals , which are also required to occur before the first departure .
consider the cavity process @xmath102 with @xmath225 .
we obtain a lower bound on the expected amount of time over which @xmath227 before @xmath102 returns to @xmath72 , assuming that @xmath406 .
we consider the event where the first service time is at least @xmath407 and the first @xmath17 arrivals occur by time @xmath408 .
we note that the probability of the latter event occurring is greater than the probability of at least @xmath17 events occurring by time @xmath408 for a rate-@xmath409 poisson process , which , by a simple large deviations estimate , is at least @xmath410 for large enough @xmath17 and an appropriate constant @xmath411 . together with ( [ eq4.1.2 ] ) , this implies that the expected amount of time in @xmath246 $ ] , during which @xmath227 and before @xmath102 has returned to @xmath72 , is at least @xmath412 inequality ( [ eq4.1.4 ] ) implies that @xmath413 where @xmath414 is the mean return time to @xmath72 . solving for @xmath23
, it follows from this that , for large @xmath17 , @xmath415 which implies ( [ eq4.1.3 ] ) for all @xmath17 .
the demonstration of the upper bound ( [ eq4.4.3 ] ) for theorem [ thm1.3.1 ] is considerably more involved than is the lower bound .
the basic idea is to consider two cases , depending on whether or not there is a service time @xmath111 with @xmath416 , for preassigned @xmath417 , before a state @xmath89 with @xmath129 is reached in the first cycle , and to obtain upper bounds for each case .
the two bounds are given in propositions [ prop4.5.1 ] and [ prop4.6.1 ] , which are then combined in corollary [ cor4.7.1 ] .
employing corollary [ cor4.7.1 ] , the proof of proposition [ prop4.4.1 ] provides an iteration scheme where a sequence of values @xmath418 for @xmath250 are given that provide successively better upper bounds for @xmath23 , and that yield ( [ eq4.4.3 ] ) in the limit .
the demonstration of proposition [ prop4.6.1 ] involves the construction of a supermartingale , whose details are postponed until the end of the section .
let @xmath347 , for given @xmath419 , denote the first time @xmath104 in the first cycle at which @xmath420 . for propositions [ prop4.5.1 ] and [ prop4.6.1 ] ,
we denote by @xmath421 the set of realizations on which some service time that is strictly greater than @xmath250 , with @xmath417 , occurs up to and including the service time interval that contains @xmath347 .
proposition [ prop4.5.1 ] considers the case where @xmath421 occurs ; the demonstration of the proposition is quick , using lemma [ lem3.9.1 ] . as in sections 2 and 3 ,
we denote by @xmath123 the occupation time at states @xmath89 , with @xmath124 .
[ prop4.5.1 ] consider a family of jsq networks with the same assumptions holding as in proposition [ prop4.4.1 ] .
then , for appropriate @xmath422 and all @xmath17 , @xmath423 \le { c_{7}}s_1^{-\beta}(k + s_1 ) .
\ ] ] we apply lemma [ lem3.9.1 ] at the beginning of the first service time that is greater than @xmath250 .
since there are less than @xmath17 jobs under @xmath421 then , it follows that , for appropriate @xmath424 and large enough @xmath17 , @xmath425 & \le 3(p(b_{s_1,k})/\bar{f}(s_1))\int_{s_1}^{\infty}(k+s)\,f(ds ) \\ & \le { c_{8}}\int_{s_1}^{\infty}(k+s)\,f(ds ) .
\end{split}\ ] ] for the latter inequality , note that there are only a finite expected number of service times in the first cycle , and that , by wald s equation , the expected number of such times that are at most @xmath111 , for given @xmath426 , is proportional to @xmath427 . since @xmath428 is increasing in @xmath111 , integration by parts together with ( [ eq4.4.2 ] )
implies that the last quantity in ( [ eq4.5.3 ] ) is at most @xmath429 , for appropriate @xmath422 . in order to consider the behavior of @xmath102 on @xmath430 , we find it convenient to employ the service time distribution @xmath431 that is given by @xmath432 we define @xmath433 analogously to @xmath102 , but where the service time distribution of the process is @xmath431 up to and including the service time interval containing @xmath347 , and is given by @xmath22 afterwards ;
@xmath434 and @xmath435 are defined analogously .
one has @xmath436
\le e[v_k^{s_1 } ] , \ ] ] where @xmath437 is the occupation time at states @xmath89 with @xmath124 for @xmath433 .
note that the mean of @xmath431 is at most @xmath4 .
in contrast to proposition [ prop4.5.1 ] , proposition [ prop4.6.1 ] requires us to restrict our choice of @xmath250 in terms of @xmath17 . for this
, we set @xmath438 and introduce the abbreviation @xmath439 the required restriction on @xmath250 is that @xmath440 where @xmath441 . in the proof of proposition [ prop4.4.1 ]
, we will introduce an iterative scheme that involves explicit choices of @xmath250 based on our knowledge of @xmath442 at each step .
proposition [ prop4.6.1 ] gives us the following upper bound for @xmath443 $ ] .
[ prop4.6.1 ] consider a family of jsq networks with the same assumptions holding as in proposition [ prop4.4.1 ] .
suppose that @xmath444 and @xmath441 are given , and that @xmath250 satisfies ( [ eq4.6.7 ] ) then , for appropriate @xmath445 and all @xmath17 , @xmath446 \le { c_{9}}(k + s_1 ) \exp\{-\delta k^{\eta}\}. \ ] ] the demonstration of proposition [ prop4.6.1 ] depends on an appropriate supermartingale . in order to construct the supermartingale , we employ the following notation .
we fix @xmath447 , which will not depend on @xmath17 as @xmath17 increases , and set @xmath448 , where @xmath319 is as defined earlier .
we set @xmath449 where @xmath450 , @xmath451 and @xmath452 , and where @xmath444 and @xmath441 are as in proposition [ prop4.6.1 ] ; the function @xmath453 is sketched in figure [ figurenew2:f ] .
the terms @xmath23 will continue to refer to the probabilities defined at the beginning of the paper with respect to the cavity process with the original service distribution @xmath22 ( not @xmath431 ) .
we let @xmath258 , with @xmath454 , denote the number of jobs for the process @xmath433 , with @xmath455 , at the time when the @xmath66th job has just departed ; we set @xmath456 , and we let @xmath457 denote the first time @xmath66 at which either @xmath458 or @xmath459 . using this notation
, we define the analog of @xmath305 in ( [ eq3.10.1 ] ) , @xmath460 note that , unlike for @xmath305 in ( [ eq3.10.1 ] ) , @xmath305 here depends strongly on the choice of @xmath17 . also , unlike @xmath305 in ( [ eq3.10.1 ] )
, it was not necessary to wait until the first departure in defining @xmath268 , since @xmath455 , and hence there is no initial residual service time ; in both cases , @xmath461 is the change in the number of jobs during the service time of the first job that begins service when @xmath462 .
[ prop4.6.6 ] consider a family of jsq networks with the same assumptions holding as in proposition [ prop4.4.1 ] .
suppose that @xmath444 and @xmath441 are given , and that @xmath305 is defined as above .
also , assume that @xmath250 satisfies ( [ eq4.6.7 ] ) .
then , for large enough @xmath17 , @xmath305 is a supermartingale , with respect to the filtration @xmath463 , for small enough @xmath444 , and appropriate @xmath464 , with @xmath465 , @xmath297 , and @xmath303 not depending on @xmath17 .
the demonstration of proposition [ prop4.6.6 ] will be given at the end of the section .
employing proposition [ prop4.6.6 ] , we now demonstrate proposition [ prop4.6.1 ] . we suppose that the terms @xmath465 , @xmath297 and @xmath303 are chosen so that , for large enough @xmath17 , @xmath305 is a supermartingale .
set @xmath466 , for given @xmath467 , which will depend on @xmath17 . since @xmath305 is bounded below by @xmath468 and @xmath469 , by the optional sampling theorem , @xmath470 on the other hand , denoting by @xmath471 the service interval during which @xmath472 first occurs and by @xmath473 the end of that interval , @xmath474 . substituting this into ( [ eq4.6.3])-([eq4.6.5 ] ) and recalling that @xmath475 , one obtains @xmath476 for large @xmath17
let @xmath477 denote the first time @xmath104 , during the first cycle , at which @xmath472 .
plugging @xmath478 into ( [ eq4.6.8 ] ) , substituting in for @xmath479 and recalling that @xmath480 , it follows that , for large @xmath17 , @xmath481 lemma [ lem3.9.1 ] applied to @xmath22 , which is the service distribution of new service times after @xmath477 , provides the upper bound @xmath482
2(k+s+n_0),\ ] ] given that @xmath483 .
since the residual service time for @xmath484 is at most @xmath250 for @xmath485 , it therefore follows from ( [ eq4.6.10 ] ) that , for large @xmath17 , @xmath486 \le 3(k+s_1)\exp\{-\delta k^{\eta}/3\}. \ ] ] the inequality in ( [ eq4.6.2 ] ) follows upon applying ( [ eq4.5.5 ] ) to ( [ eq4.6.12 ] ) and substituting in a smaller choice of @xmath164 .
we combine the upper bounds given in propositions [ prop4.5.1 ] and [ prop4.6.1 ] for @xmath487 $ ] and @xmath488 $ ] to obtain the following upper bound on @xmath226 $ ] .
since we will always assume @xmath489 in our application of the corollary , this allows us to omit the exponential term inherited from ( [ eq4.6.2 ] ) .
[ cor4.7.1 ] consider a family of jsq networks with the same assumptions holding as in proposition [ prop4.4.1 ] .
fix @xmath490 and assume that @xmath491 \wedge k^{n } , \ ] ] for some @xmath492 . then , for appropriate @xmath493 and all @xmath17 , @xmath494 \le { c_{10}}s_1^{-\beta}(k+s_1 ) . \ ] ] it follows from propositions [ prop4.5.1 ] and [ prop4.6.1 ] that @xmath255 \le { c_{7}}s_1^{-\beta}(k+s_1 ) + { c_{9}}(k+s_1)\exp\{-\delta k^{\eta}\}\ ] ] for appropriate @xmath422 and @xmath445 .
the assumption @xmath495 allows us to absorb the second term into the first .
the following elementary lemma will be employed in the proof of proposition [ prop4.4.1 ] .
[ lem4.8.1 ] suppose that r(n ) satisfies @xmath496 with @xmath497 , for @xmath498 and @xmath499 .
then , @xmath500 if @xmath501 , then the sequence @xmath502 is increasing , and if @xmath503 , then the sequence is decreasing . setting @xmath504
, it follows from ( [ eq4.8.2 ] ) that @xmath505 with @xmath506 .
all of the claims follow by iterating ( [ eq4.8.4 ] ) .
we will employ the lemma in the following multiplicative format .
[ cor4.8.7 ] suppose that @xmath507 satisfies @xmath508 with @xmath509 , for @xmath510 and @xmath441 .
then , @xmath507 satisfies @xmath511 , where the sequence @xmath502 is increasing in @xmath66 and @xmath512 with @xmath513 $ ] .
the limit in ( [ eq4.8.9 ] ) follows from ( [ eq4.8.3 ] ) upon setting @xmath514 , @xmath515 and @xmath516 .
the sequence @xmath502 is increasing since @xmath517 .
we now employ corollaries [ cor4.7.1 ] and [ cor4.8.7 ] to demonstrate proposition [ prop4.4.1 ] . for given @xmath17 and @xmath441 , we define @xmath507 as in corollary [ cor4.8.7 ] and set @xmath518 where @xmath438 . using @xmath519
, we will inductively show that , for large @xmath17 ( depending on @xmath164 ) , @xmath520 letting @xmath317 , it therefore follows from the corollary that @xmath521 this implies ( [ eq4.4.3 ] ) in proposition [ prop4.4.1 ] , with @xmath522
. to show ( [ eq4.9.2 ] ) holds for @xmath523 , we note that @xmath524 satisfies ( [ eq4.7.2 ] ) .
therefore , by ( [ eq2.3.3 ] ) and corollary [ cor4.7.1 ] , for large @xmath17 , @xmath525 where the constants in the second expression are absorbed in the third expression by using the @xmath526 term . note that , in this application of ( [ eq4.7.3 ] ) , @xmath527 . in the application of ( [ eq4.7.3 ] ) given next , @xmath528 for all @xmath454 .
suppose that ( [ eq4.9.2 ] ) holds with @xmath529 in place of @xmath66 .
choosing @xmath519 as in ( [ eq4.9.1 ] ) and employing the lower bound for @xmath507 given in ( [ eq4.8.9 ] ) , one can check that @xmath519 satisfies ( [ eq4.7.2 ] ) , with @xmath530 . also note that , by corollary [ cor4.8.7 ] , @xmath531 for large @xmath17 and all @xmath66 . applying ( [ eq2.3.3 ] ) and corollary [ cor4.7.1 ] again
, we therefore obtain that , for large @xmath17 , @xmath532 this demonstrates ( [ eq4.9.2 ] ) . in order to complete the demonstration of proposition [ prop4.4.1 ]
, we need to prove proposition [ prop4.6.6 ] , which asserts that @xmath305 , given by ( [ eq4.6.5 ] ) , is a supermartingale .
we need to show the supermartingale inequality ( [ eq3.10.2 ] ) for @xmath533 .
we do this separately over the intervals @xmath534 $ ] and @xmath535 . the basic idea
for the first interval will be to show that , on @xmath534 $ ] , ( [ eq3.10.2 ] ) will be satisfied for the same reasons as was @xmath305 , for @xmath305 given by ( [ eq3.10.1 ] ) , the point being that , since @xmath536 is large , the role played by the additional terms @xmath537 in ( [ eq4.6.3 ] ) is negligible . on the second interval @xmath535 ,
the strong negative drift of @xmath434 will be enough to compensate for both the @xmath538 and @xmath537 terms .
we do the latter interval first . we claim that for large @xmath17 and @xmath539 , @xmath540 \le e[\exp\{\phi h(n)\}. \ ] ] we first note that , because of ( [ eq4.6.4 ] ) , for @xmath539 , the number of arrivals over the @xmath267st service interval is dominated by a mixture of poisson rate-@xmath541 random variables , with @xmath111 being distributed according to @xmath431 . therefore , @xmath542 \le e^{-\phi}\int_{0}^{s_1}\exp\{ps(e^{\phi } -1)\}f^{s_1}(ds ) . \ ] ] since the integrand is convex and the mean of @xmath431 is at most @xmath4 , the right hand side is at most @xmath543 . \ ] ] on account of the definitions of @xmath544 and @xmath545 given between ( [ eq4.6.4 ] ) and ( [ eq4.6.5 ] ) , both @xmath544 and @xmath546 are at most @xmath465 .
using @xmath547 for @xmath60 close to @xmath72 , one can therefore check that , for given @xmath298 and small enough @xmath142 , ( [ eq4.10.2 ] ) is at most @xmath548 . \ ] ] for @xmath549 , the above quantity is at most @xmath4 , which holds here since @xmath550 as @xmath174 .
this implies ( [ eq4.10.1 ] ) . for @xmath551 , it is easy to see that ( [ eq3.10.2 ] ) follows from ( [ eq4.10.1 ] ) , since @xmath552 where @xmath553 . for @xmath554
$ ] , ( [ eq3.10.2 ] ) follows from ( [ eq4.10.1 ] ) with a bit more work . in place of ( [ eq4.10.4 ] )
, one uses @xmath555 for all @xmath60 , where @xmath556 . to check ( [ eq4.10.5 ] ) , note that equality holds for @xmath557 ; we claim that the maximum of @xmath558 is taken there .
one has @xmath559 because of our definition of @xmath560 ; @xmath561 for @xmath562 and @xmath563 for @xmath564 because of the concavity of @xmath558 there ; and since @xmath565 , for @xmath566 , it is easy to see that @xmath563 there .
this shows ( [ eq4.10.5 ] ) and hence ( [ eq3.10.2 ] ) for @xmath554 $ ] as well .
we still need to show ( [ eq3.10.2 ] ) for @xmath567 $ ] . for this , we compare @xmath305 with @xmath568 , where @xmath569 and @xmath570 set @xmath571 . for @xmath567 $ ] ,
one has @xmath572 since @xmath568 is the supermartingale in ( [ eq3.10.1 ] ) , except with a different initial state , @xmath568 satisfies ( [ eq3.10.2 ] ) if @xmath297 and @xmath303 are chosen as in ( [ eq3.10.1 ] ) . in a moment
, we will show that @xmath573 \le \gamma /2 \ ] ] for @xmath574 and large @xmath17 . using ( [ eq4.10.8 ] ) and
( [ eq4.10.9 ] ) , ( [ eq3.10.2 ] ) therefore also follows for @xmath305 for @xmath574 .
it suffices to show ( [ eq4.10.9 ] ) for @xmath575 . to do this
, we need to control the right tail of @xmath576 .
the number of arrivals over the @xmath267st service interval for the cavity process is dominated by a mixture of poisson mean-@xmath577 random variables , with the mixture distributed according to @xmath578 .
this mixture is in turn dominated by a poisson mean-@xmath250 random variable .
therefore , the left hand side of ( [ eq4.10.9 ] ) is at most @xmath579e^{\phi k^{\prime}}. \ ] ] setting @xmath580 , one has @xmath581 where the last inequality follows from stirling s formula .
substituting @xmath208 into ( [ eq4.10.10 ] ) , applying this bound , and employing @xmath582 , it follows that ( [ eq4.10.10 ] ) is at most @xmath583 for appropriate @xmath584 , where the inequality employs ( [ eq4.6.7 ] ) and @xmath585 , for small @xmath544 . as @xmath174 , the right hand side of ( [ eq4.10.11 ] ) goes to @xmath72 .
it follows that the left hand side of ( [ eq4.10.9 ] ) , with @xmath575 , goes to @xmath72 as @xmath174 .
this implies ( [ eq4.10.9 ] ) holds for @xmath586 and large @xmath17 , which completes the proof of the proposition .
in this section , we demonstrate theorem [ thm1.4.1 ] . we do this by demonstrating the lower and upper bounds needed for the theorem , in propositions [ prop5.1.1 ] and [ prop5.3.1 ] .
[ prop5.1.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 .
assume that ( [ eq1.1.2 ] ) holds and that @xmath587 for some @xmath588 and @xmath136
. then , for appropriate @xmath589 and @xmath51 , @xmath590 where @xmath591 [ prop5.3.1 ] consider a family of jsq networks , with given @xmath27 and @xmath28 , where the @xmath1th network has poisson rate-@xmath2 input , with @xmath3 , and where service at each queue is fifo , with distribution @xmath22 having mean @xmath4 . assume that ( [ eq1.1.2 ] ) holds and that @xmath592 for some @xmath593 and @xmath136 .
then , for appropriate @xmath594 and @xmath50 , @xmath595 where @xmath596 as in the previous two sections , the demonstration of the lower bound is substantially quicker than that of the upper bound .
we first demonstrate the lower bound , proposition [ prop5.1.1 ] and then , in the remainder of the section , derive the upper bound proposition [ prop5.3.1 ] . as in sections 3 and 4 , where we considered the cases @xmath11 and @xmath40 , for the lower bound
, it suffices to construct a path along which @xmath107 increases from @xmath72 to @xmath17 within the first cycle .
in contrast to the previous two settings , we allocate geometrically increasing amounts of time to the sequence of arrivals , up through the @xmath17th arrival ; as before , these arrivals are required to occur before the time of the first departure . the argument is similar to that for proposition [ prop4.1.1 ] in that we examine the cavity process @xmath102 with @xmath597 , and obtain a lower bound on the expected amount of time @xmath123 over which @xmath348 before @xmath102 returns to @xmath72 . here , we argue by induction , and assume that @xmath598 for given @xmath17 , where @xmath599^{-1}$ ] , and @xmath600 will be specified later .
we consider the following event @xmath228 that leads to a lower bound on @xmath23 that is compatible with ( [ eq5.2.1 ] ) .
we stipulate that the first service time is at least @xmath601 where @xmath602 .
note that @xmath603 .
we also assume that the interarrival time for the @xmath604st arrival at the queue , @xmath223 , is at most @xmath605 a little estimation shows that the sum of the terms in ( [ eq5.2.3 ] ) , over @xmath223 , is bounded above by @xmath606 which is one - half of ( [ eq5.2.2 ] ) . on account of the induction hypothesis in ( [ eq5.2.1 ] ) , the probability that the @xmath604st arrival occurs within the interarrival time in ( [ eq5.2.3 ] ) is at least @xmath607 so , the probability that the corresponding events for @xmath608 all occur within the allotted time is at least @xmath609 where @xmath610 for @xmath611 and does not depend on @xmath17 or @xmath0 , with @xmath612 as @xmath613 ; the inequality requires a little computation .
it follows from the previous two paragraphs that the event @xmath228 , given by the service time and interarrival times restricted as in ( [ eq5.2.2 ] ) and ( [ eq5.2.3 ] ) , has probability at least @xmath614 on @xmath228 , @xmath227 over the interval @xmath246 $ ] , which has length @xmath615 .
so , @xmath255 \ge \frac{1}{2}{c_{14}}\psi ( { a_{1 } } ) \exp\{{a_{1}}(d-1)k\ } \bar{f}({c_{14}}\exp\{{a_{1}}(d-1)k\ } ) . \ ] ] by substituting the bound in ( [ eq5.1.2 ] ) for @xmath616 and employing @xmath617 $ ] , one obtains @xmath618 for given @xmath619 and large enough @xmath620 , the last quantity in the above display is at least @xmath621 .
this implies the induction hypothesis in ( [ eq5.2.1 ] ) for @xmath17 and this choice of @xmath620 . since ( [ eq5.2.1 ] ) obviously holds for @xmath622 , ( [ eq5.1.3 ] ) follows , with @xmath623 .
similarly , for given @xmath620 , one obtains the lower bound @xmath621 , if @xmath619 is chosen large enough , which implies ( [ eq5.1.4 ] ) .
this completes the proof .
the demonstration of the upper bound ( [ eq5.3.3 ] ) is substantially more involved than is the lower bound .
the basic idea is similar to that employed for the upper bound in section 3 , where we classified different paths for attaining @xmath624 , for given @xmath17 and some @xmath104 , in terms of partitions @xmath338 given by ( [ eq3.11.1 ] ) .
there , the probability of the event associated with the trivial partition dominated the probabilities for the other partitions .
computing an upper bound for the probability for the trivial partition and multiplying by the upper bound @xmath350 for the total number of partitions gave us our desired upper bounds on @xmath23 .
the details of our setup here will be different .
the partitions we consider will be defined somewhat differently , and we will need to be more careful in summing up probabilities we will compute the probability of the event associated with the trivial partition separately , and will then sum up the probabilities for the other partitions , which will be negligible in comparison .
we will also require an upper bound on @xmath23 from proposition [ prop4.4.1 ] , at the beginning of our argument .
on the other hand , the computations of these upper bounds will be substantially easier here than the corresponding bounds were in section 3 .
the key difference is that here the probabilities @xmath23 will decrease sufficiently slowly in @xmath17 so that , for our estimates , not too much will be lost if we consider @xmath391 to be approximately the same for @xmath625 , which will simplify our computations . in order to show ( [ eq5.3.3 ] ) and ( [ eq5.3.4 ] ) of proposition [ prop5.3.1 ]
, we will argue by induction , assuming that , for preassigned @xmath626 and @xmath627 , @xmath628 for given @xmath17 with @xmath629 .
for appropriate choices of these preassigned values , we will show that the inequality in ( [ eq5.4.1 ] ) holds with @xmath630 .
we set @xmath631 and @xmath632 where @xmath633 is as in ( [ eq5.3.2 ] ) . these particular choices of @xmath634 and @xmath635 are not needed for most of the argument , and will only be inserted at the very end . in order to specify @xmath594 and @xmath288 , we note that , since ( [ eq4.4.2 ] ) is satisfied for every @xmath12 because of ( [ eq5.3.2 ] ) and since @xmath636 as @xmath637 , it follows from proposition [ prop4.4.1 ] that , for any @xmath1 , @xmath638 . here , we set @xmath639 .
we choose @xmath288 large enough so that @xmath640 , @xmath641 , @xmath642 and @xmath643 all hold , where @xmath644 , @xmath249 is as in ( [ eq5.3.2 ] ) and @xmath276 is as in lemma [ lem3.9.1 ] . setting @xmath645 implies ( [ eq5.4.1 ] ) holds for @xmath646 , which we will need in order to begin our induction argument .
it follows from the definition of @xmath594 and the first two conditions on @xmath288 that @xmath647 setting @xmath648 , it follows from this that @xmath649 which we will use throughout the induction argument for ( [ eq5.4.1 ] ) . in order to follow the basic induction argument
, the reader should keep in mind ( [ eq5.4.1 ] ) and ( [ eq5.4.1e ] ) , without worrying much about the other inequalities . in order to demonstrate the inequality in ( [ eq5.4.1 ] ) with @xmath630
, we proceed as outlined in the beginning of the subsection , employing the partitions @xmath338 given in ( [ eq3.11.1 ] ) and the events @xmath339 , on which a sequence of arrivals and departures occurs in the first cycle that induces the partition @xmath338 .
we define @xmath335 , as before , as the set of all partitions with final element @xmath650 ; here , the first element will be @xmath651 , with @xmath652 , where @xmath653 . in the present setting , we will pay more attention than in section 3 to the length of each of the sets in a partition @xmath338 , setting @xmath654 , for @xmath655 , for the number of elements in the @xmath208th set @xmath656 of the partition ; one has @xmath657 .
an important step in computing an upper bound for @xmath23 is proposition [ prop5.5.1 ] , which is the analog of proposition [ prop3.12.1 ] . rather than employing @xmath658 as in the proof of proposition [ prop3.12.1 ] for the upper bound for a set in the partition , we employ @xmath659 the quantity @xmath660 is the probability of at least @xmath661 events occurring for a mean-@xmath662 poisson random variable , and dominates the probability that , over the time interval @xmath663 $ ]
, at least @xmath661 arrivals occur for a cavity process @xmath102 with @xmath664 and @xmath665 .
this bound follows from the upper bound in ( [ eq2.3.4 ] ) , together with the induction hypothesis ( [ eq5.4.1 ] ) and our definition of @xmath194 . [ prop5.5.1 ] consider a family of jsq networks with the same assumptions holding as in proposition [ prop5.3.1 ] , except that ( [ eq5.3.2 ] ) is not assumed .
suppose that the induction assumption ( [ eq5.4.1 ] ) holds for given @xmath634 and for @xmath666 , where @xmath276 is as in lemma [ lem3.9.1 ] .
then , @xmath667 one can reason similarly to the argument for ( [ eq3.12.6 ] ) , in the proof of proposition [ prop3.12.1 ] , by computing an upper bound on @xmath668 $ ] .
summation over @xmath669 and application of ( [ eq2.3.3 ] ) will then imply ( [ eq5.5.2 ] ) .
the assumption @xmath666 is needed only to absorb the term @xmath276 when applying lemma [ lem3.9.1 ] .
one argues inductively , repeating the argument for ( [ eq3.12.6 ] ) , except for the substitution of @xmath670 for @xmath658 and a minor change involving the factors of @xmath671 .
for each step with @xmath672 , one obtains an additional factor @xmath673 and , for @xmath372 , one obtains the factor @xmath674 , where @xmath675 for @xmath374 and @xmath676 , with @xmath414 being the mean return time to @xmath72 for @xmath102 . for @xmath672 , the integral part of the factor is obtained by employing the comparison given directly before the statement of the proposition , comparing @xmath670 with the probability of at least @xmath661 arrivals over a service time of at least @xmath111 , and then by integrating against @xmath111 ; for @xmath372 , one also employs ( [ eq3.9.2 ] ) to provide an upper bound on the expected occupation time @xmath123 . for @xmath677 , the factor @xmath678
is obtained by applying ( [ eq3.9.3 ] ) , with @xmath376 , which gives an upper bound on the expected number of service intervals occurring over the remainder of the cycle , after the service interval corresponding to the @xmath377st step ends . for @xmath679 , instead of the factor @xmath680 , one can employ @xmath414 , since this is the expected number of service intervals over an entire cycle , and no conditioning is needed for this first step . since each of the remaining factors is at most @xmath671 , the product of all of the factors is at most @xmath681 , and since @xmath682 $ ] , the @xmath414 factors cancel , and one obtains the @xmath683 factor in ( [ eq5.5.2 ] ) .
( the improved bound just obtained by removing a factor of @xmath671 will only be needed when bounding the right hand side of ( [ eq5.5.2 ] ) for the trivial partition , in proposition [ prop5.6.1 ] . ) in propositions [ prop5.6.1 ] and [ prop5.7.1 ] , we provide upper bounds for the summands on the right hand side of ( [ eq5.5.2 ] ) , which we denote by @xmath684 . in proposition
[ prop5.6.1 ] , we do this for the trivial partition consisting of a single set , for which we write @xmath685 . in proposition
[ prop5.7.1 ] , we do this for each of the other partitions .
the sum over @xmath686 of the bounds for @xmath684 that are obtained in proposition [ prop5.7.1 ] will be negligible in comparison with the bound obtained for @xmath687 in proposition [ prop5.6.1 ] .
this last bound will therefore dominate the upper bound for @xmath23 that will be obtained by inserting these bounds into ( [ eq5.5.2 ] ) of the preceding proposition . both propositions [ prop5.6.1 ] and [ prop5.7.1 ] employ the elementary upper bounds for @xmath660 , @xmath688 which one obtains by dominating the series in ( [ eq5.4.3 ] ) by the geometric series @xmath689 , for @xmath690 .
[ prop5.6.1 ] suppose that @xmath691 where @xmath22 satisfies ( [ eq5.3.2 ] ) and @xmath692 is chosen as above , with @xmath693 , and suppose that @xmath308 , with ( [ eq5.4.1c ] ) and [ eq5.4.1e ] ) both holding .
then , @xmath694 one has @xmath406 for @xmath249 in ( [ eq5.3.2 ] ) .
applying ( [ eq5.3.2 ] ) and @xmath702 , and substituting @xmath703 , one sees that the second integral is bounded above by @xmath704 since @xmath705 , one can check that @xmath706 and @xmath707 for @xmath708 . therefore , ( [ eq5.6.5 ] ) is bounded above by @xmath709 applying ( [ eq5.3.2 ] ) , the third integral is at most @xmath710 on account of ( [ eq5.4.1e ] ) and @xmath711 , the bound for the third integral is clearly the dominant term .
combining the bounds for the three integrals therefore implies that @xmath712 which is the bound in ( [ eq5.6.2 ] ) .
[ prop5.7.1 ] suppose that @xmath713 where @xmath22 satisfies ( [ eq5.3.2 ] ) and @xmath670 is chosen as above , with @xmath714 , and suppose that @xmath308 , with ( [ eq5.4.1c ] ) and [ eq5.4.1e ] ) both holding .
then , @xmath715 for each @xmath716 . in order to demonstrate proposition [ prop5.7.1 ]
, we will categorize each partition in @xmath717 as one of three types , based on the sizes and indices of its constituent sets @xmath656 , @xmath718 .
we will say @xmath656 is _ large _ if @xmath719 and _ small _ if @xmath720 ; we will also distinguish between sets @xmath656 with @xmath672 and @xmath372 .
we will say that a partition @xmath338 is of type ( i ) if at least one of its sets @xmath656 , with @xmath672 , is large ; that it is of type ( ii ) if @xmath721 is large , but all of the other sets are small ; and that it is of type ( iii ) if none of its sets is large , but at least two sets @xmath722 and @xmath723 , with @xmath724 are small .
it is easy to check that , for any @xmath714 , the three types of sets partition @xmath717 .
we will show separately that ( [ eq5.7.2 ] ) holds when @xmath338 is a member of any of the above three types of partitions .
we will first bound the above integrals for the large and small sets @xmath656 , for both @xmath372 and @xmath672 , and will then apply these bounds to the three types of partitions .
when convenient , we abbreviate by setting @xmath725 . applying almost the same reasoning as in the proof of proposition [ prop5.6.1 ] , one obtains , for large @xmath721 , @xmath726 one decomposes the integral into the parts @xmath727 , @xmath728 and @xmath729 .
a bound for the first integral is again given by the right hand side of ( [ eq5.6.4 ] ) and a bound for the third integral is given by @xmath730 .
for the second integral , one obtains the bound @xmath731 , after substituting @xmath703 as before . instead of ( [ eq5.6.5 ] ) ,
one employs @xmath732 as an intermediate bound for the second integral , to which one applies @xmath733 ; the acquired factor @xmath734 will not cause difficulties in the present context . for @xmath308 ,
the bound in ( [ eq5.8.1 ] ) follows from the bounds on the three integrals , on account of ( [ eq5.4.1e ] ) and @xmath711 .
similar reasoning can be applied for large @xmath656 , with @xmath672 , to obtain the upper bound @xmath735 one decomposes the integral into the parts @xmath736 , @xmath737 and @xmath729 .
the first integral is at most @xmath738 and the third integral is at most @xmath739 . for the second integral , one obtains the upper bound @xmath740 , after substituting @xmath703 .
instead of ( [ eq5.6.5 ] ) or ( [ eq5.8.2 ] ) , one employs @xmath741 as an intermediate bound for the second integral , to which one applies @xmath742 . since @xmath743 , the bound in ( [ eq5.8.1 ] ) follows from the bounds on the three integrals . for small @xmath656 with @xmath672 ,
one obtains the upper bound @xmath744 as in the previous case , one decomposes the integral into the parts @xmath736 , @xmath737 and @xmath729 .
the same estimates show that the first integral is at most @xmath745 and the third integral is at most @xmath739 . for the second
integral , one obtains the upper bounds @xmath746 with the inequality using @xmath747 .
the bound in ( [ eq5.8.5 ] ) follows from the bounds on the three integrals .
we now combine the upper bounds in ( [ eq5.8.1 ] ) , ( [ eq5.8.3 ] ) , ( [ eq5.8.5 ] ) ( [ eq5.9.2 ] ) and ( [ eq5.9.3 ] ) to obtain upper bounds for the right hand side of ( [ eq5.7.3 ] ) , for large @xmath17 .
when @xmath338 is a type ( i ) partition , it follows from ( [ eq5.8.3 ] ) , ( [ eq5.9.2 ] ) and ( [ eq5.9.3 ] ) that @xmath751 when @xmath338 is a type ( ii ) partition , it follows from ( [ eq5.8.1 ] ) , ( [ eq5.8.5 ] ) and ( [ eq5.9.3 ] ) that @xmath752 and when @xmath338 is a type ( iii ) partition , it follows from ( [ eq5.8.5 ] ) , ( [ eq5.9.2 ] ) and ( [ eq5.9.3 ] ) that @xmath753 the right hand side of ( [ eq5.7.2 ] ) is greater than each of the quantities in ( [ eq5.9.4])([eq5.9.6 ] ) .
consequently , ( [ eq5.7.2 ] ) holds for all @xmath754 , as desired .
we will demonstrate that the inequality in ( [ eq5.4.1 ] ) holds for @xmath630 , provided it holds for @xmath755 , for @xmath634 and @xmath635 satisfying ( [ eq5.4.1a ] ) and ( [ eq5.4.1b ] ) , and for @xmath288 satisfying the inequalities in ( [ eq5.4.1c ] ) and on each side . by induction
, it will follow that @xmath756 by proposition [ prop5.5.1 ] , @xmath757 on account of ( [ eq5.4.1c ] ) and ( [ eq5.4.1e ] ) , it follows from the bounds in ( [ eq5.6.2 ] ) and ( [ eq5.7.2 ] ) , for @xmath687 and for @xmath684 , @xmath758 , that the first term on the right hand side of ( [ eq5.10.1 ] ) dominates the second term , and therefore @xmath759 substituting for @xmath760 and then for @xmath194 , this is at most @xmath761 upon substitution of the value for @xmath634 in ( [ eq5.4.1a ] ) and @xmath762 , the quantity inside the parentheses in ( [ eq5.10.3 ] ) is less than @xmath4 .
also , by replacing the term @xmath763 by @xmath4 , it is easy to see that the quantity inside the parentheses is again less than @xmath4 , for @xmath764 .
so , in either case , the inequality in ( [ eq5.4.1 ] ) holds for @xmath630 .
this implies ( [ eq5.9.8 ] ) . with a large enough choice of @xmath594 , ( [ eq5.9.8 ] )
extends to all @xmath765 .
this implies ( [ eq5.3.3 ] ) of proposition [ prop5.3.1 ] with @xmath766 , for this choice of @xmath594 .
moreover , as @xmath767 , one has @xmath768 , and so ( [ eq5.3.4 ] ) also holds .
this completes the proof of proposition [ prop5.3.1 ] .
99 azar , y. , broder , a. , karlin , a. and upfal , e. ( 1994 ) .
balanced allocations .
26@xmath769 acm symp .
theory comp .
_ , 593 - 602 .
bramson , m. ( 2008 ) . _ stability of queueing networks . _
cole dt de probabilits de saint - flour xxxvi - 2006 , lecture notes in mathematics 1950 .
springer - verlag , berlin .
bramson , m. ( 2010 ) . _ stability of join the shortest queue networks_. submitted to _ ann .
_ bramson , m. , lu , y. and prabhakar , b. ( 2010 ) .
randomized load balancing with general service time distributions . to appear in _ proc .
acm sigmetrics 2010_. bramson , m. , lu , y. and prabhakar , b. ( 2011 ) .
asymptotic independence of queues under randomized load balancing .
submitted to _ queueing systems_. davis , m.h.a .
_ markov models and optimization_. chapman & hall , london .
foss , s. and chernova , n. ( 1998 ) . on the stability of a partially accessible multi - station queue with state - dependent routing .
_ queueing systems _ * 29 * , 55 - 73 .
graham , c. ( 2000 ) .
chaoticity on path space for a queueing network with selection of the shortest queue among several .
_ j. appl .
prob . _ * 37 * , 198 - 211 .
luczak , m. and mcdiarmid , c. ( 2005 ) . on the power of two choices : balls and bins in continuous time .
probab . _ * 15 * , 1733 - 1764 .
luczak , m. and mcdiarmid , c. ( 2006 ) . on the maximum queue length in the supermarket model .
probab . _ * 34 * , 493 - 527 .
martin , j.b . and
suhov , y.m .
fast jackson networks .
probab . _ * 9 * , 840 - 854 .
mitzenmacher , m. ( 2001 ) .
the power of two choices in randomized load balancing .
_ ieee transactions on parallel and distributed systems _ * 12:10 * , 1094 - 1104 .
stanley , r.p .
_ enumerative combinatorics , volume 1_. cambridge university press , cambridge .
suhov , y.m . and vvedenskaya , n.d .
fast jackson networks with dynamic routing .
_ problems inform .
_ * 39 * , 136 - 153 .
vocking , b. ( 1999 ) .
how asymmetry helps load balancing .
_ ieee symp . found .
, 131 - 140 .
vvedenskaya , n.d . ,
dobrushin , r.l . and karpelevich ( 1996 ) .
queueing system with selection of the shortest of two queues : an asymptotic approach .
_ problems of information transmission _ * 32 ( 1 ) * , 15 - 29 .
vvedenskaya , n.d . and suhov , y.m .
dobrushin s mean - field approximation for a queue with dynamic routing .
_ markov proc .
fields _ * 3 * , 493 - 527 . | in join the shortest queue networks , incoming jobs are assigned to the shortest queue from among a randomly chosen subset of @xmath0 queues , in a system of @xmath1 queues ; after completion of service at its queue , a job leaves the network .
we also assume that jobs arrive into the system according to a rate-@xmath2 poisson process , @xmath3 , with rate-@xmath4 service at each queue .
when the service at queues is exponentially distributed , it was shown in vvedenskaya et al .
@xcite that the tail of the equilibrium queue size decays doubly exponentially in the limit as @xmath5 . this is a substantial improvement over the case @xmath6 , where the queue size decays exponentially .
the reasoning in @xcite does not easily generalize to jobs with nonexponential service time distributions . a modularized program for treating general service time distributions
was introduced in bramson et al . @xcite .
the program relies on an ansatz that asserts , in equilibrium , any fixed number of queues become independent of one another as @xmath5 .
this ansatz was demonstrated in several settings in bramson et al .
@xcite , including for networks where the service discipline is fifo and the service time distribution has a decreasing hazard rate . in this article , we investigate the limiting behavior , as @xmath7 , of the equilibrium at a queue when the service discipline is fifo and the service time distribution has a power law with a given exponent @xmath8 , for @xmath9 .
we show under the above ansatz that , as @xmath7 , the tail of the equilibrium queue size exhibits a wide range of behavior depending on the relationship between @xmath10 and @xmath0 . in particular , if @xmath11 , the tail is doubly exponential and , if @xmath12 , the tail has a power law . when @xmath13 , the tail is exponentially distributed . |
noncollinear magnetism , i.e. magnetic ordering where the local magnetic moments are not parallel or antiparallel to a global direction , exists in a variety of systems , for example , in topologically frustrated antiferromagnets , in spiral spin - density waves and spin - spiral states .
these types of orderings can be called interatomic noncollinear magnetism since the different atomic moments are noncollinear .
generally , noncollinear configurations occur more easily in a low symmetry or a disordered magnetic system.@xcite thus small clusters , which have less symmetry constraints than the bulk , are likely candidates for noncollinear structures . indeed , as a consequence of spin - spiral ground state of fcc fe ( refs . 3 and 4 ) and spin density wave ground state of bcc cr ( ref .
5 ) , the small fe@xmath0 and cr@xmath0 clusters were found to have noncollinear magnetic structures.@xcite from the structural and magnetic point of view , manganese is one of the most complex of all metals and has attracted considerable attention .
the electronic configuration in a mn atom is 4@xmath6 3@xmath7 and consequently mn atoms have high magnetic moments of 5 @xmath8 .
high 4@xmath63@xmath9 @xmath10 4@xmath113@xmath12 promotion energy means that mn atoms do not bind strongly when they are brought together to form a cluster or a bulk solid .
the most stable polymorph , @xmath13mn , has an exotic crystal structure containing 58 atoms in a cubic unit cell .
this @xmath13mn exhibits a complex antiferromagnetic order below the nel temperature of 95 k and is nonmagnetic at room temperature.@xcite the magnetic transition of @xmath14-mn is coupled to a tetragonal distortion .
recent density functional calculations indicate a noncollinear magnetic ground state for @xmath13mn.@xcite the other polymorph of mn , known as @xmath15mn , also has a stable noncollinear configuration even though the ground state is weakly ferrimagnetic.@xcite these results are in good agreement with the neutron scattering,@xcite magnetic torque@xcite and nuclear magnetic resonance@xcite experiments , where the experimental results are interpreted in terms of noncollinear antiferromagnetic structure . as mentioned earlier small mn clusters , which have less symmetry constraints , are likely candidates for the occurrence of noncollinear magnetic structure .
however , majority of the theoretical calculations have been made under the collinear spin assumption.@xcite among them , most extensive study has been done by kabir _
et al . _ , where the structural and magnetic properties of mn@xmath0 clusters in the size range @xmath12@xmath220 have been investigated using density functional theory.@xcite it was found that mn@xmath16 , mn@xmath17 and mn@xmath18 exhibit ferromagnetic ordering , whereas a magnetic transition to the ferrimagnetic ordering takes place at @xmath1 5 and the ferrimagnetic state continues to be the ground state for clusters with @xmath19 5 .
the predicted magnetic moments are in well agreement with the stern - gerlach ( sg ) clusters beam experiments.@xcite however very recently,@xcite we pointed out that the noncollinear treatment of spins is indeed necessary at the size range @xmath46 .
_ studied mn@xmath20 and mn@xmath21 clusters and found mn@xmath21 to be the smallest cluster to exhibit noncollinear ground state.@xcite however , using local spin density approximation longo _
_ found that all the clusters in the size range @xmath1 3@xmath27 have noncollinear ground states.@xcite recently , noncollinear magnetic ordering has also been found in small mn - clusters supported on a cu(111 ) surface.@xcite in this study however , the cluster geometries have been assumed planer and they were placed on a regular face centered - cubic cu lattice with experimental lattice parameters .
no structural relaxation for either the surface or the cluster was considered .
the present paper is the sequel of our earlier reports.@xcite . in this paper , we allow noncollinear ordering of atomic moments and extensively study how collinear and noncollinear magnetic ordering evolves in small mn@xmath0 clusters and how the magnetic ordering gets perturbed if we dope a single nonmagnetic atom ( for example an as - atom ) into it .
this is important as ( iii , mn)v semiconductors show ferromagnetism and the effect of mn - clustering is being debated for a long time.@xcite moreover , the noncollinear ferromagnetism is common in ( iii , mn)v semiconductors.@xcite experimental study by potashnik _
et al._@xcite indicated that the curie temperature , the ground state saturation magnetization @xmath22 and the shape of the @xmath23 curve all depend upon the temperature and annealing .
these can be explained only if the noncollinearity in the localized mn magnetic moments is considered in the ground state.@xcite from this point of view , it will be very interesting to investigate whether small as@mn@xmath0 clusters do show noncollinearity or not .
in the theoretical approaches to noncollinear magnetism,@xcite the wave functions are described by two - component spinors , @xmath24 , where @xmath14 and @xmath25 are the spin indices .
the density matrix is defined as , @xmath26 where @xmath27 is the occupation number of the single - particle state .
+ the charge @xmath28 and magnetization @xmath29 parts of the density matrix can be extracted by expanding in terms of the pauli spin matrices @xmath30 @xmath31 , @xmath32,\ ] ] where @xmath33 is the cartesian components of @xmath29 .
the exchange - correlation potential , @xmath34 , contains nonmagnetic and magnetic parts .
the nonmagnetic part @xmath35 and @xmath36 are given as a function of @xmath37 and @xmath38 in the same way that is done in the case of collinear magnetism , but here in addition @xmath39 is always parallel to @xmath29 . in this scheme , the individual eigenstates can have different spin quantization directions and furthermore , the spin quantization axis of the each state can vary with position .
calculations have been performed using density functional theory ( dft ) within the pseudopotential plane wave method .
the projector augmented wave method@xcite has been used and for the spin - polarized gradient approximation to the exchange - correlation energy we used the perdew - burke - ernzerhof functional,@xcite as implemented in the vienna _ ab - initio _
simulation package.@xcite the 3@xmath40 , 4@xmath41 for mn and 4@xmath41 , 4@xmath42 orbitals for as are treated as valence states .
the spinor wave functions are expanded in a plane wave basis set with the kinetic energy cutoff 337.3 ev .
we have ensured that the cutoff is sufficient by varying it up to 600 ev and found that the energy difference between ferromagnetic and antiferromagnetic states converges within @xmath43 1 mev for mn@xmath16 and as@mn@xmath16 .
calculations have been carried out at the @xmath44-point .
we adopted periodic boundary conditions and described the cluster within a large simple cubic supercell such that the periodic images are separated by at least 12 of vacuum .
this essentially ensures negligible interactions between the periodic images .
symmetry unrestricted optimizations were performed using quasi - newtonian and conjugate gradient methods until all the force components become less than 0.005 ev/. for a particular sized mn@xmath0 and as@mn@xmath0 cluster , several different initial structures were studied to ensure that the globally optimized geometry does not correspond to a local minima , and to be extensive , both noncollinear and collinear magnetic structures have been considered separately . moreover , we explicitly considered all possible spin multiplicities for the collinear case . for a collinear spin cluster
all the spins are parallel ( @xmath45 ) or antiparallel ( @xmath46 ) to each other . on the other hand , the angle between any two moments could be anything in between @xmath45 and @xmath46 for a noncollinear case and the deviation from @xmath45 or @xmath46 is termed degree of noncollinearity " .
the average degree of noncollinearity for a particular cluster can be defined as , @xmath47 where @xmath48 runs for all independent spin pairs and @xmath49 is the total number of such independent spin(atom ) pairs with @xmath50 being the number of atoms in the cluster .
@xmath51 is either @xmath45 or @xmath46 and @xmath52 is the angle between @xmath53-th and @xmath54-th moment .
in an earlier work , we have studied the magnetic ordering of pure mn@xmath0 ( @xmath55 ) clusters within the collinear atomic moment assumption.@xcite above a certain cluster size ( @xmath37=5 ) , the magnetic ordering is found to be ferrimagnetic and the calculated magnetic moments were in good agreement@xcite with stern - gerlach experiments.@xcite however , the validity of the collinear spin assumption should be checked rigorously because , in principle , the complex ferrimagnetic ordering and magnetic ` frustration ' @xcite could definitely lead to the noncollinear ordering in magnetic clusters .
the type of magnetic ordering along with the total magnetic moment and relative energy difference with the ground state are given in the table [ tab : table1 ] and the optimal noncollinear structures are shown in fig.[fig : composite ] ( @xmath13@xmath210 ) and fig.[fig : mn13 ] ( @xmath1 13 , 15 and 19 ) for pure mn@xmath0 clusters .
.[tab : table1 ] type of magnetic ordering , average degree of noncollinearity ( @xmath56 ) , total magnetic moment ( @xmath57 ) and the relative energy difference ( @xmath58e = e@xmath2e@xmath59 ) for pure mn@xmath0 clusters for @xmath12@xmath210 , 13 , 15 and 19 . [ cols= " < , < ,
< , < , < " , ] the asmn dimer has much higher binding energy of 1.12 ev / atom and much shorter bond length ( 2.21 ) than those of mn@xmath16 dimer : 0.52 ev / atom and 2.58 , respectively.@xcite we have repeated our calculations of asmn dimer including the mn 3@xmath42 as valence electrons and obtained an optimized bond length of 2.22 and binding energy of 1.08 ev / atom , with the same total magnetic moment , which confirms that the inner mn-3@xmath42 electrons contributes insignificantly to the bonding .
the mn - mn coupling is found to be ferromagnetic ( fig.[fig : mnas_compo]a ) for as@mn@xmath16 cluster , which has a total magnetic moment of 9 @xmath8 .
the mn - mn distance in this collinear as@mn@xmath16 is same with the pure mn@xmath16 dimer .
however , this collinear ground state is nearly degenerate with a noncollinear magnetic structure ( fig.[fig : mnas_compo]b ) .
another collinear structure with antiferromagnetic mn - mn coupling ( fig.[fig : mnas_compo]c ) is found to be the next isomer which lies 111 mev higher than that of the ground state .
the pure mn@xmath17 cluster was found to be ferromagnetic with large 15 @xmath8 magnetic moment and is collinear in nature .
the single as - atom doping reduces its magnetic moment considerably to 4 @xmath8 for as@mn@xmath17 cluster in its ground state .
this structure is also collinear in nature , which is shown in fig.[fig : mnas_compo]d .
however , as in the pure case , there exist several isomers with different magnetic structures ( table [ tab : table2 ] and fig.[fig : mnas_compo]e , f ) .
the next three isomers are all noncollinear and they lie close in energy , 5 , 8 and 43 mev higher , respectively . the optimal collinear structure with 20 @xmath8 magnetic moment was found to have 31 mev less energy than the optimal noncollinear structure for the pure tetramer . on the other hand for as@mn@xmath18 cluster , the optimal noncollinear structure ( fig.[fig : mnas_compo]h )
is found to be nearly degenerate with the collinear ferromagnetic ground state ( fig.[fig : mnas_compo]g ) .
the next isomer ( table [ tab : table2 ] ) is also noncollinear in nature and all these three structures have comparable magnetic moments .
the ground state of the as@mn@xmath20 cluster is found to be collinear ( fig.[fig : mnas_compo]i ) , which is nearly degenerate with the optimal noncollinear structure ( fig.[fig : mnas_compo]j ) . both of these structures have equal magnetic moments , which are particularly small .
the next two isomers are noncollinear with high noncollinearity and have comparatively large magnetic moments ( table [ tab : table2 ] ) .
similar to the pure mn@xmath21 cluster , the as@mn@xmath21 is found to be the smallest doped cluster , which show noncollinear magnetism ( fig.[fig : mnas_compo]l ) in its ground state .
this noncollinear structure lies 159 mev lower than the next isomer , which is collinear ( fig.[fig : mnas_compo]m ) .
the noncollinear ground state has small magnetic moment ( 1.28 @xmath8 ) compared to the optimal collinear ( 9 @xmath8 ) state .
the noncollinear ground state has a as - capped mn - octahedral geometry , whereas the next ( collinear ) isomer is a pentagonal bi - pyramid where the as - atom sits in the pentagonal ring .
another noncollinear structure lies 196 mev higher which has large noncollinearity ( 24.04@xmath60 ) and comparable magnetic moment with the collinear structure ( see table [ tab : table2 ] ) .
the ground state of the as@mn@xmath61 cluster is collinear ( fig.[fig : mnas_compo]n ) with a total magnetic moment of 6 @xmath8 .
the next isomer is noncollinear in nature with 4.94 @xmath8 magnetic moment ( fig.[fig : mnas_compo]o ) , which lies sightly higher ( 16 mev ) in energy .
however , for pure mn@xmath61 cluster , the energy difference between the collinear ground state and the optimal noncollinear state is large , 232 mev ( table [ tab : table1 ] ) ; i.e. the single as doping reduces the energy difference between the collinear ground state and the optimal noncollinear state .
moreover , there exist several collinear and noncollinear structures ( table [ tab : table2 ] ) , which are close in energy . in an early work,@xcite we found a collinear ground state with a magnetic moment of 3 @xmath8 for as@mn@xmath62 cluster . however , after a more rigorous scan@xcite of the potential energy surface we found few different magnetic states ( table [ tab : table2 ] ) , which are lower in energy than the previously reported one . among them
a noncollinear state with very low noncollinearity ( fig.[fig : mnas_compo]q ) is found to be the ground state .
the next isomer is also noncollinear ( fig.[fig : mnas_compo]r ) , which is followed by a collinear state ( fig.[fig : mnas_compo]s ) with 7 @xmath8 magnetic moment .
the ground state of as@mn@xmath63 cluster is noncollinear with high noncollinearity , 43.24@xmath60 .
however , the magnetic moment of this structure is nearly zero , 0.1 @xmath8 ( fig.[fig : mnas_compo]t ) , which is much smaller than that of the pure mn@xmath63 cluster in its ground state .
the next isomer is collinear in nature ( fig.[fig : mnas_compo]u ) and has comparatively large magnetic moment , 10 @xmath8 .
however , this collinear structure lies much higher , 249 mev , in energy .
we have found several isomers for the as@mn@xmath64 cluster ( table [ tab : table2 ] ) . among all a noncollinear magnetic structure ( fig.[fig : mnas_compo]v ) with high noncollinearity ( 45.22@xmath60 ) is found to be the ground state .
this structure has a total magnetic moment of 3.09 @xmath8 .
the next isomer is also noncollinear which lies only 36 mev higher in energy .
the optimal collinear structure have comparatively high magnetic moment and lies 90 mev higher ( fig.[fig : mnas_compo]w ) .
binding energies of the optimal collinear and noncollinear magnetic structures have been shown in fig.[fig : be ] for both pure mn@xmath0 and as@mn@xmath0 clusters .
it has been understood@xcite that due to the lack of hybridization between the filled 4@xmath41 states and the half - filled 3@xmath40 states and due to high ( 2.14 ev ) 4@xmath6 3@xmath7 @xmath10 4@xmath11 3@xmath12 promotion energy , mn atoms do not bind strongly when they form clusters or crystals .
this is manifested through their low binding energy ( see fig.[fig : be ] ) , which is the lowest among all other 3@xmath40- transition metal clusters .
this is also experimentally evidenced through recent photodissociation experiments for cationic clusters.@xcite however , the bonding improves considerably due to doping of single nonmagnetic as - atom.@xcite when an as - atom is attached to the pure mn@xmath0 clusters , due to enhanced hybridization through the as-@xmath42 electrons , the binding energy of the resultant as@mn@xmath0 cluster increases substantially .
it is found that the energy gain in adding an as - atom to an existing pure mn@xmath0 cluster is much larger than that of adding a mn - atom to an existing as@mn@xmath65 cluster.@xcite this clearly indicates that the as - atom may act as a nucleation center for mn - atoms .
the total energy difference , @xmath66e , between the optimal collinear and noncollinear structure for a particular sized cluster is plotted in the inset of fig.[fig : be ] as a function of @xmath37 for both pure and doped clusters . by definition
if @xmath66e is positive(negative ) the corresponding ground state is collinear(noncollinear ) . for both pure mn@xmath0 and as@mn@xmath0 clusters ,
the collinear states are found to be lower in energy than the corresponding optimal noncollinear states up to five mn - atoms in the cluster and in the size range @xmath67 6 the noncollinear states start to be the ground state with the exception for @xmath17 for mn@xmath0 and @xmath17 and 8 for as@mn@xmath0 clusters . however , it should be noted here that for the entire size range there exists several collinear / noncollinear isomers which are close in energy with the corresponding ground state ( see table [ tab : table1 ] and [ tab : table2 ] ) .
this establishes the importance to treat the moments noncollinearly .
the total magnetic moment of pure mn@xmath0 and as@mn@xmath0 clusters of the corresponding ground states and closely lying isomers are given in table [ tab : table1 ] and table [ tab : table2 ] , respectively , and plotted in fig.[fig : mag ] for the ground states .
pure mn@xmath0 clusters in the size range @xmath3 4 have collinear ground state and the mn - mn coupling is ferromagnetic .
these clusters have a 5 @xmath8/atom magnetic moment which is the hund s rule value for isolated mn - atom .
the magnetic moment per atom decreases drastically at @xmath1 5 and the magnetic ground states start to be noncollinear for @xmath4 6 .
however , we have found many collinear and noncollinear isomers with varying magnetic moment which are close in energy ( not shown in fig.[fig : mag ] , see for instance table [ tab : table1 ] ) .
the ground state magnetic moments of pure mn@xmath0 clusters are compared with the sg experiment in fig.[fig : mag ] .
calculated magnetic moments are in good agreement with the sg experiment .
however , the quantitative value differ for few clusters sizes . for an example
, we find a much larger magnetic moment ( 2.14 @xmath8/atom ) for mn@xmath21 cluster compared to the experimental value of 0.55 @xmath8/atom.@xcite however , our predicted value agrees with previous dft calculation.@xcite this quantitative discrepancy between the calculated and measured magnetic moment may arise from few reasons : ( i ) if the true ground state is missed out during the search through the potential energy surface , which is very much unlikely in the present case as we have extensively considered many different geometric and magnetic structures as initial guess .
( ii ) in every magnetic deflection measurement the magnetic moments are calculated assuming a model ( e.g. superparamagnetic , locked moment , adiabatic magnetization etc . ) which might affect the outcome . ( iii ) due to the isomer distribution in the experimental cluster beam .
the large moment of pure mn@xmath0 and as@mn@xmath0 clusters arise from localized 3@xmath40 electrons at mn - atoms .
however , the strong @xmath5 hybridization induces a small negative polarization to the as - atom .
for example , this negative polarization is 0.26 @xmath8 for asmn dimer .
however , the magnitude of this negative polarization decreases monotonically to 0.01 @xmath8 for as@mn@xmath64 cluster . generally , the total magnetic moment of as@mn@xmath0 cluster is lower than the corresponding pure mn@xmath0 cluster due to @xmath5 hybridization .
similar to their pure counterparts , the mn - mn coupling in as@mn@xmath16 and as@mn@xmath18 is collinear ferromagnetic and the emergence of noncollinear ground state is seen at @xmath37=6 .
moreover , it has been seen that mn - mn exchange coupling in as@mn@xmath0 clusters show anomalous behavior compared to ruderman - kittel - kasuya - yosida ( rkky ) type predictions.@xcite for collinear ground state of as@mn@xmath16 cluster exchange coupling oscillates between positive and negative with mn - mn separation ( @xmath68 ) favoring the ferromagnetic and antiferromagnetic solutions , respectively , and dies down as 1/@xmath69 , which is a typical rkky - type behavior .
however on the other hand , the average exchange coupling decreases with @xmath37 and has strong environment dependency , which is in contradiction with the rkky - type prediction .
this study presents a systematic investigation of electronic structure and emergence of noncollinear magnetism in pure mn@xmath0 and as@mn@xmath0 clusters within gradient - corrected dft approach .
no considerable structural change has been found due to noncollinear treatment of atomic moments for both pure and doped manganese clusters
. moreover , single as - doping to mn@xmath0 clusters does not give rise any considerable structural change , but only a moderate perturbation to their pure counterpart .
the pure mn@xmath0 clusters have low binding energies , which improves substantially due to single as - doping through strong @xmath5 hybridization .
the ground state of both pure mn@xmath0 and as@mn@xmath0 clusters for @xmath3 5 is collinear and emergence of noncollinear ground states is seen for @xmath4 6 .
however , for both kind of clusters , there exist many collinear and noncollinear isomers . due to strong @xmath5 hybridization
the magnetic moments of as@mn@xmath0 are smaller compared to their pure counterpart .
this also induces a negative polarization to the as - atom in as@mn@xmath0 cluster .
although the results presented here are specific to the mn@xmath0 and as@mn@xmath0 clusters , they also contain more general picture : noncollinear magnetic ordering is possible in small magnetic clusters .
electronic address : [email protected] r. lorenz , j. hafner , s. s. jaswal , and d. j. sellmyer , phys .
* 74 * , 3688 ( 1995 ) .
m. liebs , k. hummler and m. fhnle , phys .
b * 51 * , 8664 ( 1995 ) .
y tsunoda , j. phys . : condens .
matter * 1 * , 10427 ( 1989 ) .
o. n. mryasov , a. i. liechtenstein , l. m. sandratskii and v a gubanov , j. phys . : condens .
matter * 3 * , 7683 ( 1991 ) .
h. zabel , j. phys . : condens .
matter * 11 * 9303 ( 1999 ) .
t. oda , a. pasquarello and r. car , phys .
lett . * 80 * , 3622 ( 1998 ) . o. ivanov and v. p. antropov , j. appl .
phys . * 85 * , 4821 ( 1999 ) .
d. hobbs , g. kresse and j. hafner , phys .
b * 62 * , 11556 ( 2000 ) . c. kohl and g. f. bertsch , phys . rev .
b * 60 * , 4205 ( 1999 ) .
m. a. ojeda , j. dorantes - dvila and g. m. pastor , phys .
b * 60 * , 6121 ( 1999 ) .
r. s. tebble and d. j. craik , _ magnetic materials _
( wiley , new york , 1969 ) p. 48 .
d. hobbs , j. hafner , and d. spik , phys .
b * 68 * , 14407 ( 2003 ) .
j. hafner and d. hobbs , phys .
b * 68 * , 14408 ( 2003 )
. c.g . shull and m.k .
wilkinson , rev .
phys . * 25 * , 100 ( 1953 ) ; j.s .
kasper and b.w .
roberts , phys . rev .
* 101 * 537 ( 1956 ) ; j.a .
oberteuffer , j.a .
marcus , l.h .
schwartz , and g.p .
felcher , phys . lett . *
28a * , 267 ( 1969 ) ; n. kunitomi , y. yamada , y. nakai , and y. fujii , j. appl . phys . * 40 * , 1265 ( 1969 ) ; t. yamada , n. kunitomi , y. nakai , d.e .
cox , and g. shirane , j. phys .
* 28 * , 615 ( 1970 ) ; a.c .
lawson , a.c .
larson , m.c .
aronson , s. johnson , z. fisk , p.c .
canfield , j.d .
thompson , and r.b .
von dreele , j. appl .
phys . * 76 * , 7049 ( 1994 ) .
t. yamada and s. tazawa , j. phys .
* 28 * , 609 ( 1970 ) .
h. yamagata and k. asayama , j. phys .
* 33 * , 400 ( 1972 ) .
bobadova - parvanova , k. a. jackson , s. srinivas and m. horoi , phys .
a * 67 * , 61202(r ) ( 2003 ) .
p. bobadova - parvanova , k. a. jackson , s. srinivas and m. horoi , j. chem .
phys * 122 * , 14310 ( 2005 ) .
n. o. jones , s. n. khanna , t. baruah and m. r. pederson , phys .
b * 70 * , 45416 ( 2004 ) t. m. briere , m. h. f. sluiter , v. kumar and y. kawazoe , phys . rev .
b * 66 * , 64412 ( 2002 ) .
m. kabir , a. mookerjee and d. g. kanhere , phys .
b * 73 * , 224439 ( 2006 ) .
m. b. knickelbein , phys .
lett . * 86 * , 5255 ( 2001 ) .
m. b. knickelbein , phys .
b * 70 * , 14424 ( 2004 ) .
m. kabir , d. g. kanhere and a. mookerjee , phys .
b * 73 * , 75210 ( 2006 ) .
t. morisato , s. n. khanna and y. kawazoe , phys .
b * 72 * , 014435 ( 2005 ) .
r. c. longo , e. g. noya , and l. j. gallego , phys .
b * 72 * , 174409 ( 2005 ) .
a. bergman , l. nordstrm , a. b. klautau , s. frota - pessa and o. eriksson , phys .
b * 73 * , 174434 ( 2006 ) . j. m. sullivan , g. i. boishin , l. j. whitman , a. t. hanbicki , b. t. jonker , and s. c. erwin , phys . rev .
b * 68 * , 235324 ( 2003 ) .
m. zajac , j. gosk , e. grzanka , m. kamiska , a. twardowski , b. strojek , t. szy szko , and s. podsiadlo , j. appl .
phys . * 93 * , 4715 ( 2003 ) .
d. c. kundaliya , s. b. ogale , s. e. lofland , s. dhar , c. j. metting , s. r. shind e , z. ma , b. varughese , k.v .
ramanujachary , l. salamanca - riba and t. venkatesan , nature materials * 3 * , 709 ( 2004 ) .
p. mahadevan and a. zunger , phys .
b * 68 * , 75202 ( 2003 ) ; m. van schilfgaarde and o. n. mryasov , phys . rev .
b * 63 * , 233205 ( 2001 ) . j. l. xu , m. van schilfgaarde and g. d. samolyuk , phys .
lett . * 94 * , 97201 ( 2005 ) .
s. j. potashnik , k. c. ku , s. h. chun , j. j. berry , n. samarth , and p. schiffer , appl . phys .
lett . * 79 * , 1495 ( 2001 ) .
s. j. potashnik , k. c. ku , r. mahendiran , s. h. chun , r. f. wang , n. samarth , and p. schiffer , phys .
b * 66 * , 012408 ( 2002 ) .
j. schliemann and a. h. macdonald , phys .
lett . * 88 * , 137201 ( 2002 ) .
j. schliemann , phys .
b * 67 * , 045202 ( 2003 ) .
g. a. fiete , g. zarnd , b. jank , p. redliski , and c. p. moca , phys .
b * 71 * , 115202 ( 2005 ) .
u. von barth and l. hedin , j. phys .
c 5 , 1629 ( 1972 ) .
sandratskii and p.g .
guletskii , j. phys .
16 , l43 ( 1986 ) .
j. kbler , k. h. hck , and j. sticht , j. appl .
63 , 3482 ( 1988 ) .
j. kbler , k.h .
hck , j. sticht , and a.r .
williams , j. phys .
phys . 18 , 469 ( 1988 ) . g. kresse and d. joubert , phys .
b * 59 * , 1758 ( 1999 ) .
j. p. perdew , k. burke , and m. ernzerhof , phys .
* 77 * , 3865 ( 1996 ) .
g. kresse and j. furthmuller , phys .
b. * 54 * , 11169 ( 1996 ) .
s. t. bramwell and m. j. p. gingras , science * 294 * , 1495 ( 2001 ) .
g. w. ludwig , h. h. woodbury , and r. o. carlson , j. phys .
solids * 8 * , 490 ( 1959 ) . due to the negative polarization of
the as - atom , the summation in the equation 3 includes the as - atom too for as@mn@xmath0 clusters .
after our initial report,@xcite we have considered several more initial structures to confirm the true ground state , for each cluster .
however , we did not find any new structures with energy lower than those previously reported ground states except only for as@mn@xmath62 cluster .
a. terasaki , s. minemoto and t. kondow , j. chem . phys .
* 117 * , 7520 ( 2002 ) .
k. tono , a. terasaki , t. ohta and t. kondow , j. chem . phys . * 123 * , 174314 ( 2005 ) . | using first - principles density functional calculations , we have studied the magnetic ordering in pure mn@xmath0 ( @xmath12@xmath210 , 13 , 15 , 19 ) and as@mn@xmath0 ( @xmath11@xmath210 ) clusters .
although , for both pure and doped manganese clusters , there exists many collinear and noncollinear isomers close in energy , the smaller clusters with @xmath35 have collinear magnetic ground state and the emergence of noncollinear ground states is seen for @xmath46 clusters . due to strong @xmath5 hybridization in as@mn@xmath0 clusters ,
the binding energy is substantially enhanced and the magnetic moment is reduced compared to the corresponding pure mn@xmath0 clusters . |
let @xmath2 be the empirical distribution function of independent random variables @xmath3 with unknown distribution function @xmath4 on the real line .
let us recall some well - known facts about @xmath2 ( cf .
shorack and wellner 1986 ) : the stochastic process @xmath5 has the same distribution as @xmath6 , where @xmath7 is the empirical distribution of independent random variables @xmath8 with uniform distribution on @xmath9 $ ] .
this enables us to construct confidence bands for the distribution function @xmath4 .
a well - known classical method are kolmogorov - smirnov confidence bands : let @xmath10 and let @xmath11 be the @xmath12-quantile of @xmath13 } |\mathbb{u}_n(t)| .\ ] ] then with probability at least @xmath14 , @xmath15 \quad\text{for all } \ x \in { \mathbb{r}}.\ ] ] equality holds if @xmath4 is continuous . since @xmath16 converges in distribution in @xmath17)$ ] to standard brownian bridge @xmath18 , @xmath11 converges to the @xmath12-quantile @xmath19 of @xmath20 .
in particular , the simultaneous confidence intervals in have width @xmath21 uniformly in @xmath22 .
another method , based on a goodness - of - fit test by berk and jones ( 1979 ) , was introduced by owen ( 1995 ) : let @xmath23 be the @xmath12-quantile of @xmath24 where @xmath25 for @xmath26 $ ] and @xmath27 .
this leads to an alternative confidence band for @xmath4 : with probability at least @xmath14 , @xmath28 as shown by jager and wellner ( 2007 ) , the asymptotic distribution of @xmath29 remains the same if one replaces @xmath30 by a more general function ; in particular , one may interchange its two arguments .
moreover , @xmath31 from this one can deduce that leads to confidence intervals with length at most @xmath32 uniformly in @xmath22 ; see ( k.5 ) in section [ subsec : k ] . hence they are substantially shorter than the kolmogorov - smirnov intervals for @xmath33 close to @xmath34 or @xmath35 .
but in the central region , i.e. when @xmath33 is bounded away from @xmath34 and @xmath35 , they are of width @xmath36 rather than @xmath21 .
an obvious goal is to refine these methods and combine the benefits of the kolmogorov - smirnov and berk - jones confidence bands .
methods of this type have been proposed by various authors , see mason and schuenemeyer ( 1983 ) and the references cited therein .
a key for understanding the asymptotics of @xmath29 but also the new methods presented later are suitable variants of the law of the iterated logarithm ( lil ) . for brownian
bridge @xmath18 the lil states that @xmath37 almost surely .
various refinements of this result have been obtained .
one particular consequence of kolmogorov s upper class test ( cf .
erds 1942 , or ito and mckean 1974 , chapter 1.8 ) is the following result : for @xmath27 define @xmath38 then for any fixed @xmath39 , @xmath40 almost surely .
note that @xmath41 , @xmath42 , and , as @xmath43 , @xmath44 this explains why follows from kolmogorov s test and shows the connection between and .
note also that @xmath45 in the present paper we prove statements similar to for general stochastic processes on @xmath46 . in section [ sec : lil ] we state a general condition on a stochastic process @xmath47 such that for any fixed @xmath48 , @xmath49 almost surely . in particular , the stochastic process @xmath50 satisfies this condition . then in section [ sec : ep ] these general results are applied to @xmath51 it turns out that for any fixed @xmath48 , @xmath52 converges in distribution to @xmath53 asymptotic statements like this refer to @xmath54 , unless stated otherwise .
moreover , if @xmath55 are the order statistics of @xmath8 , then for fixed @xmath48 , @xmath56 where @xmath57 to test the null hypothesis that @xmath4 is equal to a given continuous distribution function @xmath58 , consider the test statistic @xmath59 under the null hypothesis , @xmath60 has the same distribution as @xmath61 .
hence if @xmath62 denotes the @xmath12-quantile of @xmath61 , one may reject the null hypothesis at level @xmath63 if @xmath60 exceeds @xmath62 . in section [ sec : gof ]
we investigate the power of this new test in more detail .
in particular we show that it attains the detection boundary for gaussian mixture models as specified by donoho and jin ( 2004 ) .
the statistic @xmath64 leads to a new confidence band for @xmath4 : let @xmath65 be the order statistics of @xmath3 , and let @xmath66 and @xmath67 be the @xmath12-quantile of @xmath64 and @xmath68 , respectively .
then @xmath69 , and with probability at least @xmath14 , the following is true : for @xmath70 and @xmath71 , @xmath72 , \ ] ] where @xmath73 , @xmath74 and @xmath75 : n k(t_{nj},u ) \le c(t_{nj } ) + \nu d(t_{nj } ) + \tilde{\kappa}_{n,\nu,\alpha } \bigr\ } \ \ \text{if } \ j > 0 , \\
b_{nj } & : = \max \bigl\ { u \in [ 0,1 ] : n k(t_{n , j+1},u ) \le c(t_{n , j+1 } ) + \nu d(t_{n , j+1 } ) + \tilde{\kappa}_{n,\nu,\alpha } \bigr\ } \ \ \text{if } \ j <
n .\end{aligned}\ ] ] since @xmath76 is no larger than @xmath77 for @xmath78 , our confidence bands have similar accuracy as those of owen ( 1995 ) in the tail regions while achieving the usual root-@xmath79 consistency everywhere .
a more precise comparison is provided in section [ sec : cb ] .
thereafter we relate our methods to a negative result of bahadur and savage ( 1956 ) about the nonexistence of confidence bands with vanishing width in the tails .
finally we discuss briefly an interesting alternative approach to goodness - of - fit tests and confidence bands by aldor - noiman et al .
( 2013 ) and eiger et al .
( 2013 ) .
all proofs and technical arguments are deferred to section [ sec : proofs ] .
section [ sec : supplements ] contains supplementary material including a quantitative version of bahadur and savage ( 1956 , theorem 2 ) and decision theoretic considerations about the gaussian mixture model of donoho and jin ( 2004 ) .
our conditions and results involve the function @xmath80 with @xmath81 its inverse is the logistic function @xmath82 with @xmath83 and @xmath84 we consider stochastic processes @xmath85 on subsets @xmath86 of @xmath46 which have locally uniformly sub - exponential tails in the following sense : [ c ] there exist a real constant @xmath87 and a non - increasing funtion @xmath88 $ ] such that @xmath89 as @xmath90 , and @xmath91 \cap { \mathcal{t } } } x(t ) > \eta \bigr ) \ \le \ m \exp(- l(c ) \eta)\ ] ] for arbitrary @xmath92 , @xmath93 and @xmath94 . [ thm : lil ]
suppose that @xmath95 satisfies condition [ c ]
. for arbitrary @xmath48 and @xmath96 there exists a real constant @xmath97 depending only on @xmath98 , @xmath99 , @xmath100 and @xmath101 such that @xmath102 [ rem : limit.at.01 ] suppose that @xmath95 satisfies condition [ c ] , where @xmath103 and @xmath104 . for any @xmath48 ,
the supremum @xmath105 of @xmath106 over @xmath86 is finite almost surely .
but this implies that @xmath107 almost surely . for if @xmath108 , then @xmath109 so the claim follows from @xmath110 almost surely and @xmath111 as @xmath112 .
our definition of the function @xmath113 may look somewhat arbitrary .
indeed , we tried various choices , e.g. @xmath114 .
theorem [ thm : lil ] is valid for any nonnegative function @xmath115 on @xmath46 such that @xmath116 and @xmath117 as @xmath43 .
the special choice @xmath113 yielded a rather uniform distribution of @xmath118 when @xmath119 and @xmath100 close to one .
our first example for a process @xmath95 satisfying condition [ c ] is squared and standardized brownian bridge : [ lem : bb ] let @xmath120 and @xmath121 with standard brownian bridge @xmath18
. then condition [ c ] is satisfied with @xmath122 and @xmath123 .
in particular , lemma [ lem : bb ] and theorem [ thm : lil ] yield for any @xmath48 .
as indicated in the introduction , theorem [ thm : lil ] may be applied to the uniform empirical process @xmath7 in two ways . a first version concerns @xmath120 and @xmath51 [ lem : uniformep.1 ] the stochastic process @xmath124 satisfies condition [ c ] with @xmath122 and @xmath123 . combining this lemma , theorem [ thm : lil ] and donsker s theorem for the uniform empirical process yields the following result : [ thm : uniformep.1 ] for any fixed @xmath48
, @xmath125 converges in distribution to the random variable @xmath53 for the computation of confidence bands it is more convenient to work with the following stochastic process on @xmath126 : @xmath127 [ lem : uniformep.2 ] the stochastic process @xmath128 satisfies condition [ c ] with @xmath122 and @xmath123 .
again we may combine this with theorem [ thm : lil ] and donsker s theorem for partial sum processes to obtain a new limit theorem : [ thm : uniformep.2 ] for any fixed @xmath48 , @xmath129 converges in distribution to the random variable @xmath68 defined in theorem [ thm : uniformep.1 ] .
as explained in the introduction , we may reject the null hypothesis that @xmath4 is a given continuous distribution function @xmath58 at level @xmath130 if @xmath131 exceeds @xmath62 .
note also that the latter supremum may be expressed as the maximum of @xmath132 terms , replacing the argument @xmath133 with @xmath134 and @xmath135 for @xmath136 or with @xmath137 . as shown in the next lemma , for any fixed citical value @xmath138 , the probability that @xmath139 is small if the quantity @xmath140 is large , where @xmath141 for @xmath142 $ ] with @xmath143 .
note that @xmath144 as @xmath145 .
[ lem : power ] for any critical value @xmath138 there exists a constant @xmath146 such that @xmath147 here and subsequently , the subscript @xmath4 in @xmath148 or @xmath149 specifies the true distribution function of the random variables @xmath3 .
now consider an arbitrary sequence @xmath150 of distribution functions .
then for any fixed level @xmath63 , lemma [ lem : power ] and the fact that @xmath151 imply that @xmath152 provided that @xmath153 in particular , is satisfied if @xmath154 for all sample sizes @xmath79 .
thus our test has asymptotic power one for any fixed distribution function different from @xmath58 .
[ [ detecting - gaussian - mixtures . ] ] detecting gaussian mixtures .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider a testing problem studied in detail by donoho and jin ( 2004 ) .
the null hypothesis is given by @xmath155 , the standard gaussian distribution function , whereas @xmath156 for certain numbers @xmath157 and @xmath158 . by means of lemma [
lem : power ] one can derive the following result : [ lem : donohojin ] * ( a ) * suppose that @xmath159 for some fixed @xmath160 .
further let @xmath161 for some @xmath162 .
then @xmath163 if @xmath164 , \\ \bigl ( 1 - \sqrt{1 - \beta } \bigr)^2 & \text{if }
\ \beta \in [ 3/4,1 ) .
\end{cases}\ ] ] * ( b ) * suppose that @xmath165 such that @xmath166 .
then @xmath163 if @xmath167 for some @xmath168
. as explained by donoho and jin ( 2004 ) , any goodness - of - fit test at fixed level @xmath63 has trivial asymptotic power @xmath130 whenever @xmath169 for some @xmath160 and @xmath161 with @xmath170 .
thus our new test provides another example of an asymptotically optimal procedure in this particular setting .
other procedures with asymptotic power one whenever @xmath171 are tukey s higher criticism test ( donoho and jin 2004 ) or the generalized berk jones tests ( jager and wellner 2007 ) . in the setting of part
( b ) , the latter two classes of tests can fail to have asymptotic power one if @xmath167 for fixed @xmath168 but @xmath172 sufficiently slow . on the other hand
, one can show that any level-@xmath130 test of @xmath58 versus @xmath173 has trivial asymptotic power whenever @xmath174 for an arbitrary fixed @xmath175 . a rigorous proof is provided with the supplementary material .
parts ( a ) and ( b ) of lemma [ lem : donohojin ] are well connected .
for let @xmath159 for some @xmath176 $ ] , and @xmath161 for some @xmath177 .
then @xmath178 and with @xmath179 we may rewrite @xmath180 as @xmath181
the confidence bands of owen ( 1995 ) may be described as follows : for @xmath70 let @xmath182 . with confidence
@xmath14 we may claim that for @xmath70 and @xmath71 , @xmath183 , \ ] ] where @xmath184 and @xmath185 our new method is analogous : with confidence @xmath14 , for @xmath70 and @xmath71 , the value @xmath33 is contained in @xmath186 $ ] , where @xmath187 and @xmath188 for @xmath189 .
asymptotically the new confidence band is everywhere at least as good as owen s ( 1995 ) band , and in the central region it is infinitely more accurate : [ thm : asefficiency ] for any fixed @xmath63 , @xmath190 while @xmath191 to be honest , the asymptotic statement in the first part of theorem [ thm : asefficiency ] requires huge sample sizes to materialize . in our numerical experiments it turned out that for sample sizes @xmath79 up to @xmath192 and very small indices @xmath193 , the ratio @xmath194 is between @xmath195 and @xmath196 but drops off quickly as @xmath193 gets larger .
[ [ numerical - example . ] ] numerical example .
+ + + + + + + + + + + + + + + + + + the left panel in figure [ fig : asefficiency.1 ] depicts for @xmath197 , @xmath198 and @xmath199 the confidence limits @xmath200 and @xmath201 as functions of @xmath202 .
the dotted ( yellow ) line in the middle represents the values @xmath203 .
the corresponding quantile @xmath66 was estimated in @xmath204 monte - carlo simulations as @xmath205 , and this leads to the maximal value @xmath206 . in the right panel
one sees the centered boundaries @xmath207 and @xmath208 .
in addition the centered boundaries @xmath209 and @xmath209 are shown as dashed ( and cyan ) lines , based on the estimated quantile @xmath210 and @xmath211 .
the additional horizontal lines are the values @xmath212 for the kolmogorov - smirnov bands .
figure [ fig : asefficiency.2 ] shows the same as the right panel in figure [ fig : asefficiency.1 ] , but with sample sizes @xmath213 and @xmath214 in the left and right panel , respectively .
( left panel ) and the centered confidence limits @xmath215 ( right panel ) for @xmath197 , @xmath198 and @xmath199.,title="fig:",scaledwidth=49.0% ] ( left panel ) and the centered confidence limits @xmath215 ( right panel ) for @xmath197 , @xmath198 and @xmath199.,title="fig:",scaledwidth=49.0% ] and @xmath198 , @xmath199.,title="fig:",scaledwidth=49.0% ] and @xmath198 , @xmath199.,title="fig:",scaledwidth=49.0% ] [ [ accuracy - in - the - tails . ] ] accuracy in the tails .
+ + + + + + + + + + + + + + + + + + + + + + the confidence bands described here yield an upper bound for @xmath4 with limit @xmath216 or @xmath217 at @xmath218 and a lower bound for @xmath4 with limit @xmath219 or @xmath220 at @xmath221 .
the proof of theorem [ thm : asefficiency ] reveals that @xmath222 on the other hand , the proof of theorem 2 of bahadur and savage ( 1956 ) shows that we can not expect substantially more accuracy in the tails .
their arguments can be adpated to show that for any @xmath12-confidence band and any @xmath223 , the limit of the upper band at @xmath224 is smaller than @xmath225 with probability at most @xmath226 .
the same bound holds true for the probability that the limit of the lower bound at @xmath227 is greater than @xmath228 . for a proof
we refer to the supplementary material .
[ [ an - alternative - approach - via - the - union - intersection - principle . ] ] an alternative approach via the union - intersection principle .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + aldor - noiman et al .
( 2013 ) and eiger et al .
( 2013 ) propose to use a union - intersection type goodness - of - fit test and related confidence bands . under the null hypothesis that @xmath229 , the test statistic @xmath230 and @xmath231 follow a beta distribution with parameters @xmath232 and @xmath233 .
denoting its distribution function with @xmath234 , two resulting p - values would be @xmath235 and @xmath236 .
thus one can reject the null hypothesis at level @xmath130 if the test statistic @xmath237 is lower or equal to the @xmath130-quantile @xmath238 of @xmath239 a corresponding @xmath12-confidence band for @xmath4 may be constructed as follows : with confidence @xmath14 one may claim that for @xmath70 and @xmath71 , @xmath240 , \ ] ] where @xmath241 , @xmath242 , and @xmath243 the results of eiger et al . (
2013 ) indicate that this goodness - of - fit test has similar properties as the one of berk and jones ( 1979 ) .
indeed , if one considers the closely related test statistic @xmath244 one may consider @xmath245 as a simple surrogate for the minimum of the two p - values @xmath235 and @xmath236 .
a possible weakness of the union - intersection approach is that it ignores correlations between the random variables @xmath231 .
elementary calculations reveal that for @xmath246 , @xmath247 thus the correlation of two neighbors @xmath231 and @xmath248 is rather large if @xmath249 is close to @xmath250 but much smaller if @xmath249 is close to @xmath34 or @xmath35 . as a result
, the minimum in tends to be attained for indices @xmath232 such that @xmath249 is close to @xmath34 or @xmath35 . with
our additive correction term @xmath251 we try to account for such effects .
for symmetry reasons it suffices to prove upper bounds for @xmath252 note first that for @xmath253 , @xmath254 consequently , @xmath255 and since @xmath256 has derivative @xmath257 , @xmath258 now let @xmath259 be sequence of real numbers with @xmath260 such that @xmath261 then it follows from @xmath262 for @xmath263 $ ] that @xmath264 } & \bigl ( x - c - \nu d ) \\ & \le \ \sup_{{\mathcal{t}}\cap [ \ell(a_k),\ell(a_{k+1 } ) ] } x \ , - \ , c(\ell(a_k ) ) - \nu d(\ell(a_k ) ) \ , + \ , ( 1 + \nu ) \delta_k \\ & \le \ \sup_{{\mathcal{t}}\cap [ \ell(a_k),\ell(a_{k+1 } ) ] } x \ , - \ , c(\ell(a_k ) ) - \nu d(\ell(a_k ) ) \ , + \ , ( 1 + \nu ) \delta_*\end{aligned}\ ] ] with @xmath265 .
thus condition [ c ] implies that @xmath266 } ( x - c - \nu d ) > \eta \bigr ) \\ & \le \
\sum_{k \ge 0 } \pr \bigl ( \sup_{{\mathcal{t}}\cap [ \ell(a_k ) , \ell(a_{k+1 } ) ] } x \ , > \eta - ( 1 + \nu ) \delta _
* + c(\ell(a_k ) ) + \nu d(\ell(a_k ) ) \bigr ) \\ & \le \
m \exp ( ( 1 + \nu)\delta _ * ) \exp(- \eta l(\delta _ * ) ) \cdot g , \end{aligned}\ ] ] where @xmath267 for any number @xmath268 , @xmath269 .\ ] ] now we define @xmath270 for some @xmath271 such that @xmath272 .
note that @xmath273 is a continuously differentiable function on @xmath274 with @xmath275 , limit @xmath276 and derivative @xmath277 this implies that is indeed satisfied with @xmath278 moreover , as @xmath279 , @xmath280 since @xmath281 , this implies that @xmath282 .
hence the asserted inequality is true with @xmath283 . to verify condition [ c ] here ,
recall that if @xmath284 is standard brownian motion , then @xmath285 has the same distribution as @xmath286 with @xmath287 .
hence for @xmath92 and @xmath93 , @xmath288 } x(t ) \ & = _ { { \mathcal{l } } } \ \sup_{t \in [ \ell(a),\ell(a+c ) ] } \frac{(1 - t)^2 \mathbb{w}(s(t))^2}{2t(1 - t ) } \\ & = \
\in [ \ell(a),\ell(a+c ) ] } \frac{\mathbb{w}(s(t))^2}{2 s(t ) } \\ & = \
\in [ e^a , e^{a+c } ] } \frac{\mathbb{w}(s)^2}{2 s } \\ & = _ { { \mathcal{l } } } \ \sup_{u \in [ 1 , e^c ] } \frac{\mathbb{w}(u)^2}{2u } .\end{aligned}\ ] ] but it is well - known that @xmath289 is a reverse martingale . thus @xmath290 is a nonnegative reverse submartingale for arbitrary real numbers @xmath291 .
hence it follows from doob s inequality for nonnegative submartingales that for any @xmath292 , @xmath293 } \frac{\mathbb{w}(u)^2}{2u } \ge \eta \bigr ) \ & \le \ \pr \bigl ( \sup_{u \in [ 1,e^c ] } \frac{\mathbb{w}(u)^2}{u^2 } \ge \frac{2\eta}{e^c } \bigr ) \\
& \le \ 2 \pr \bigl ( \sup_{u \in [ 1,e^c ] } \mathbb{w}(u)/u \ge \sqrt{2e^{-c}\eta } \bigr ) \\ & = \ 2 \inf_{\lambda > 0 } \ , \pr \bigl ( \sup_{u \in [ 1,e^c ] } \exp \bigl ( \lambda \mathbb{w}(u)/u \bigr ) \ge \exp \bigl ( \lambda \sqrt{2e^{-c}\eta } \bigr ) \bigr ) \\ & \le \ 2 \inf_{\lambda > 0 } \ ,
\ex \exp \bigl ( \lambda \mathbb{w}(1)/1 \bigr ) \exp \bigl ( - \lambda \sqrt{2e^{-c}\eta } \bigr ) \\ & = \ 2 \inf_{\lambda > 0 } \ , \exp \bigl ( \lambda^2/2 - \lambda \sqrt{2e^{-c}\eta } \bigr ) \\ & = \ 2 \exp ( - e^{-c } \eta ) .\end{aligned}\ ] ] + before starting with a function @xmath294 itself , let us introduce two auxiliary functions : @xmath295 elementary algebra shows that for @xmath296 , @xmath297 this representation will be useful for @xmath298 close to @xmath34 or @xmath35 .
[ lem : h ] both functions @xmath299 and @xmath300 are bijective , strictly increasing and strictly convex .
moreover , @xmath301
\quad\text{for } \ x \in [ 0,\infty ) , \\
\tilde{h}(z ) \ & \in \ \bigl [ - \log(1 - z^2)/2 , - \log(1 - z ) \bigr ] \quad\text{for } \ z \in [ 0,1 ) .\end{aligned}\ ] ] the inverse functions @xmath302 and @xmath303 are strictly increasing and strictly concave with @xmath304 , \\
\tilde{h}^{-1}(y ) \ & \in \
\bigl [ 1 - e^{-y } , \sqrt{1 - e^{-2y } } \bigr ] .\end{aligned}\ ] ] the proof of this lemma is elementary and thus omitted .
now we are ready to state essential properties of @xmath294 : [ [ k.0 ] ] ( k.0 ) + + + + + with the convention that @xmath305 one can easily verify that the function @xmath306 \times ( 0,1 ) \to { \mathbb{r}}$ ] is continuous .
in particular , @xmath307 and @xmath308 . moreover , @xmath309 for arbitrary @xmath26 $ ] and @xmath27 . [ [ k.1 ] ] ( k.1 ) + + + + + for @xmath296 , @xmath310 ( the latter formula is true even for @xmath26 $ ] . ) in particular , @xmath311 with equality if , and only if , @xmath312 . [
[ k.2 ] ] ( k.2 ) + + + + + for @xmath296 , @xmath313 in particular , the hessian matrix of @xmath314 at @xmath315 has positive diagonal elements and non - negative determinant @xmath316 .
this implies that @xmath314 is convex on @xmath9 \times ( 0,1)$ ] .
[ [ k.3 ] ] ( k.3 ) + + + + + for fixed @xmath317 and arbitrary @xmath318 , @xmath319 since @xmath320 , it follows from ( k.1 ) that @xmath321 these formulae remain true if we replace @xmath322 with @xmath34 . on the other hand , since @xmath323 for @xmath324 , a suitable version of taylor s formula and ( k.2 ) imply that @xmath325 but for any @xmath326 , @xmath327 thus for @xmath318 , @xmath328 and this entails the asserted inequalities for the three ratios @xmath329 , @xmath330 and @xmath331 .
[ [ k.4 ] ] ( k.4 ) + + + + + to verify theorems [ thm : uniformep.1 ] , [ thm : uniformep.2 ] and [ thm : asefficiency ] we have to approximate @xmath314 by a simpler function @xmath332 given by @xmath333 indeed , for arbitrary @xmath296 and @xmath334 , @xmath335 .\ ] ] it follows from ( k.1 - 2 ) and taylor s formula that @xmath336 for some @xmath337 between @xmath338 and @xmath339 .
hence @xmath340 are both contained in @xmath341 $ ] , according to .
[ [ k.5 ] ] ( k.5 ) + + + + + for arbitrary @xmath342 and @xmath26 $ ] , @xmath27 , the inequality @xmath343 implies that @xmath344 in particular , @xmath345 the first inequality has been proved by dmbgen ( 1998 ) , but for the reader s convenience and the proof of the new part , a complete derivation is given here : for symmetry reasons , it suffices to consider the case @xmath346 and derive the upper bounds for @xmath347 .
let us first treat the case @xmath348 : here @xmath349 .
thus @xmath350 implies that @xmath351 .
moreover , @xmath352 , and elementary considerations show that @xmath353 .
now let @xmath354 and @xmath355 .
it follows from @xmath356 and ( k.1 ) that @xmath357 in case of @xmath358 , the latter integral is not smaller than @xmath359 , and @xmath343 implies the upper bound @xmath360 . in case of @xmath361 , we obtain the bound @xmath362 with @xmath363 , @xmath364 and the auxiliary function @xmath365 from lemma [ lem : h ] .
consequently , the inequality @xmath343 entails that @xmath366 , so @xmath367 on the other hand , @xmath368 in case of @xmath369 , the latter integral is at least @xmath370 , and we may conclude from @xmath343 that @xmath371 is bounded by @xmath372 . in case of @xmath373 , we define @xmath374 , @xmath375 and may write @xmath376 the second inequality in the previous display follows from the fact that @xmath377 is strictly decreasing on @xmath378 $ ] .
thus @xmath379 equals @xmath380 and is strictly positive .
hence the preceding considerations yield the upper bound @xmath381 for @xmath371 .
[ [ k.6 ] ] ( k.6 ) + + + + + for @xmath382 and @xmath342 let @xmath383 solve the equation @xmath384 then @xmath385 @xmath386 @xmath387 .\ ] ] with @xmath388 we may write @xmath389 since @xmath390 , this implies that @xmath391 , which is equivalent to @xmath392 . on the other hand
, it follows from the expansion @xmath393 that @xmath394 as @xmath395 , it follows from @xmath396 that @xmath397 whence @xmath398 consequently , @xmath399 the latter inequality following from concavity of @xmath400 .
this proves . as to ,
let @xmath401 , and define the points @xmath402 for @xmath403 $ ] .
then @xmath404 consequently , by ( k.4 ) , @xmath405 uniformly in @xmath403 $ ] .
this shows that @xmath406 as @xmath407 .
finally , let @xmath408 .
then it follows from @xmath409 that @xmath410 consequently , by concavity of @xmath411 , @xmath412 which yields . before proving lemma [ lem : uniformep.1 ]
let us recall that for @xmath413 and @xmath27 , @xmath414 , \\ \infty & \text{else } .
\end{cases}\ ] ] indeed , hoeffding ( 1963 ) showed that for a random variable @xmath415 and @xmath413 , @xmath416 we imitate and modify a martingale argument of berk and jones ( 1979 , lemma 4.3 ) which goes back to kiefer ( 1973 ) .
note first that @xmath417 is a reverse martingale in @xmath27 , that means , @xmath418 consequently , for @xmath318 and @xmath419 , @xmath420 } \hat{g}_n(s)/s \le u \bigr ) \ & = \
\inf_{\lambda \le 0 } \ , \pr \bigl ( \sup_{s \in [ t , t ' ] } \exp(\lambda \hat{g}_n(s)/s - \lambda u ) \ge 1 \bigr ) \\ & \le \
\inf_{\lambda \le 0 } \ , \ex \exp(\lambda \hat{g}_n(t)/t - \lambda u)\end{aligned}\ ] ] by doob s inequality for non - negative submartingales .
but @xmath421 , so @xmath422 thus @xmath423 } \hat{g}_n(s)/s \le u \bigr ) \ \le \ \exp(- n k(tu , t ) ) \quad\text{for all } \ u \in [ 0,1 ] .\ ] ] one may rewrite this inequality as @xmath424 } n k \bigl ( t \min\{\hat{g}_n(s)/s , 1\ } , t \bigr ) \ge \eta \bigr ) \ \le \
\exp(- \eta ) \quad\text{for all } \ \eta \ge 0 .\ ] ] for if @xmath425 , the probability on the left hand side equals @xmath34 .
otherwise there exists a unique @xmath426 $ ] such that @xmath427 .
but then @xmath428 finally , it follows from property ( k.3 ) of @xmath294 that for @xmath429 , @xmath430 with @xmath431 .
hence @xmath424 } n k \bigl ( \min\{\hat{g}_n(s ) , s\ } , s \bigr ) \ge
\eta \bigr ) \ \le \
\exp(- e^{-c } \eta ) \quad\text{for all } \
\eta \ge 0 .\ ] ] since @xmath432 has the same distribution as @xmath433 , and because of the symmetry relations @xmath434 and @xmath435 , the previous inequality implies further that @xmath436 } n k \bigl ( \max\{\hat{g}_n(s ) , s\ } , s \bigr ) \ge \eta \bigr ) \\ & = \ \pr \bigl ( \sup_{s \in [ t , t ' ] } n k \bigl ( \min\{1 - \hat{g}_n(s ) , 1 - s\ } , 1 - s \bigr ) \ge \eta \bigr ) \\ & = \ \pr \bigl ( \sup_{s \in [ 1-t',1-t ] } n k \bigl ( \min\{\hat{g}_n(s ) , s\ } , s \bigr ) \ge \eta \bigr ) \\ & \le \
\exp(- e^{-c } \eta ) \quad\text{for all } \
\eta \ge 0 .\end{aligned}\ ] ] consequently , since @xmath437 , @xmath424 } n k(\hat{g}_n(s),s )
\ge \eta \bigr ) \ \le \ 2 \exp(- e^{-c } \eta ) \quad\text{for all } \
\eta \ge 0 .\ ] ] + for any fixed @xmath438 , it follows from donsker s invariance principle for the uniform empirical process and the continuous mapping theorem that @xmath439 } \bigl ( \frac{\mathbb{u}_n(t)^2}{2t(1-t ) } - c(t ) - \nu d(t ) \bigr ) \ \to_{{\mathcal{l } } } \
\sup_{[-\delta,\delta ] } \bigl ( x - c - \nu d \bigr ) , \ ] ] where @xmath119 .
with @xmath440 it follows from property ( k.4 ) of @xmath294 that @xmath441 with @xmath442 } |r_n(t)| \ \le \ \bigl ( 1 - n^{-1/2 } \delta^{-1 }
\|\mathbb{u}_n\|_\infty \bigr)^{-2 } - 1
o_p(n^{-1/2 } ) .\ ] ] thus @xmath443 } \bigl ( x_n - c - \nu d \bigr ) \ \to_{{\mathcal{l } } } \
\sup_{[-\delta,\delta ] } \bigl ( x - c - \nu d \bigr ) .\ ] ] but theorem [ thm : lil ] implies that for any @xmath108 , the random variables @xmath444 and @xmath445 satisfy the inequalities @xmath446 and @xmath447 for arbitrary @xmath94 and some constants @xmath96 , @xmath97 .
consequently for any @xmath448 , @xmath449 } ( x_{n } - c - \nu d ) < \sup_{(0,1 ) } ( x_{n } - c - \nu d ) \bigr ) \\ & \le \ \pr \bigl ( t_{n,\nu ' } - ( \nu - \nu ' ) d(\delta ) > - \rho \bigr ) + \pr \bigl ( x_{n}(1/2 ) \le - \rho \bigr ) \\ & \le \ m_o \exp \bigl ( l_o\rho - l_o ( \nu - \nu ' ) d(\delta ) \bigr ) + \pr(x(1/2 ) \le - \rho ) + o(1 ) \intertext{because $ x_n(1/2 ) \to_{{\mathcal{l } } } x(1/2)$ , and } \pr \bigl ( & \sup_{[\delta,1-\delta ] } ( x - c - \nu d ) < \sup_{(0,1 ) } ( x - c - \nu d ) \bigr ) \\ & \le
\ m_o \exp \bigl ( l_o\rho - l_o ( \nu - \nu ' ) d(\delta ) \bigr ) + \pr(x(1/2 ) \le - \rho ) .\end{aligned}\ ] ] setting @xmath450 , the limits of the right hand sides become arbitrarily small for sufficiently small @xmath371 .
this shows that @xmath451 converges in distribution to @xmath68 .
our proof of lemma [ lem : uniformep.2 ] involves an exponential inequality for beta distributions from dmbgen ( 1998 ) . for the reader s
convenience , its proof is included in the supplementary material .
[ lem : beta ] let @xmath296 , and let @xmath452 for some @xmath453 .
then @xmath454 we utilize a well - known representation of uniform order statistics : let @xmath455 be independent random variables with standard exponential distribution , i.e. @xmath456 , and let @xmath457 .
then @xmath458 in particular , @xmath459 and @xmath460 .
furthermore , for @xmath461 , the random vectors @xmath462 and @xmath463 are stochastically independent .
this implies that @xmath464 is a reverse martingale , because for @xmath465 , @xmath466 consequently , for @xmath467 and @xmath468 , it follows from doob s inequality and lemma [ lem : beta ] that @xmath469 again one may reformulate the previous inequalities as follows : for any @xmath292 , @xmath470 but property ( k.3 ) of @xmath294 implies that for @xmath471 , @xmath472 with @xmath473 .
consequently , @xmath474 since @xmath475 has the same distribution as @xmath476 , a symmetry argument as in the proof of lemma [ lem : uniformep.1 ] reveals that @xmath477 + one can use essentially the same arguments as in the proof of theorem [ thm : uniformep.1 ] .
this time one has to utilize the well - known fact that @xmath478 where the uniform quantile process @xmath479 with @xmath480 converges in distribution in @xmath481)$ ] to a brownian bridge @xmath482 ; see e.g. shorack and wellner ( 1986 ) , pages 86 , 93 , and 637 - 644 .
suppose that @xmath139 .
then the inequalities in ( k.5 ) imply that @xmath483 where @xmath484 .
multiplying this inequality with @xmath79 and utilizing the triangle inequality @xmath485 leads to @xmath486 now our goal is to get rid of the term @xmath487 on the right hand side . defining the auxiliary stochastic process @xmath488 with the convention @xmath489
, we may rewrite as @xmath490 where we utilized the inequalities @xmath491 for @xmath492 $ ] and @xmath493 for @xmath494 .
note that inequality is of the form @xmath495 with the nonnegative processes @xmath496 and @xmath497 .
but @xmath495 is equivalent to @xmath498 , and this may be rewritten as @xmath499 , so @xmath500 consequently , @xmath501 because @xmath502 .
finally , since @xmath503 , we obtain the inequality @xmath504 on the left hand side stands a function @xmath505 , and its supremum over @xmath506 equals @xmath507 . thus it suffices to show that for a suitable constant @xmath146 , @xmath508 for any @xmath22 .
indeed , @xmath509 is equivalent to @xmath510 since @xmath511 , it follows from markov s inequality that the latter inequality occurs with probability at most @xmath512 with a certain constant @xmath513 .
this bound is trivial if @xmath514 , which is equivalent to @xmath515 .
otherwise , @xmath516 + in what follows we use frequently the elementary inequalities @xmath517 where @xmath518 .
in particular , as @xmath519 , @xmath520 now consider two sequences @xmath521 and @xmath522 tending to @xmath227 and @xmath155 , @xmath523 . then the inequalities imply that @xmath524 moreover , @xmath525 because @xmath526 while @xmath527 consequently , @xmath163 if @xmath528 in part ( a ) with @xmath159 and @xmath160 we imitate the arguments of donoho and jin ( 2004 ) and consider @xmath529 with @xmath530 .
then by , @xmath531 so the left hand side of equals @xmath532 the exponent in the enumerator is maximal in @xmath533 $ ] if @xmath534 , i.e. @xmath535 , and this leads to @xmath536 thus when @xmath537 we should choose @xmath538 and @xmath539 . when @xmath540 we should choose @xmath541 and @xmath542 . as to part
( b ) , we consider the more general setting that @xmath159 for some @xmath543 , where @xmath544 .
the latter constraint is trivial when @xmath545 but relevant when @xmath546 .
now we consider @xmath547 with arbitrary constants @xmath548 . now
@xmath549 so the left hand side of equals @xmath550 the exponent of @xmath551 becomes minimal in @xmath552 if @xmath553 , i.e. @xmath554 .
then we obtain @xmath555 and this converges to @xmath227 if the exponents of @xmath551 and @xmath556 are negative and non - positive , respectively .
this is the case if @xmath557 .
( note that @xmath558 because @xmath559 . ) by symmetry it suffices to analyze the differences @xmath208 and @xmath560 for @xmath561 .
recall the notation @xmath562 for the unique number @xmath563 such that @xmath564 , introduced in ( k.6 ) .
there we considered only @xmath382 , but it follows from @xmath565 that @xmath566 as @xmath567 . for @xmath561 , we may write @xmath568 recall that @xmath569 moreover , since @xmath570 is convex on @xmath571 , the numbers @xmath572 are concave in @xmath573 . in particular , with @xmath574 denoting the maximum of @xmath575 and @xmath576 , @xmath577 uniformly in @xmath561 . hence it suffices to show that @xmath578 first we consider indices @xmath579 .
note that for @xmath580 , @xmath581 and we may deduce from and @xmath582 that uniformly in @xmath583 , @xmath584 on the other hand , since @xmath585 we may conclude that uniformly in @xmath586 , @xmath587 hence holds true if we restrict @xmath193 to the interval @xmath588 .
next we consider indices @xmath193 between @xmath589 and @xmath590 , i.e. @xmath591 and @xmath592 uniformly in @xmath593 .
then it follows from , together with @xmath594 and monotonicity of @xmath595 , that uniformly in @xmath596 , @xmath597 hence is satisfied with @xmath598 in place of @xmath599 .
now consider @xmath600 .
uniformly in @xmath601 , the product @xmath602 is larger than @xmath603 , so @xmath604
. moreover , @xmath605 , and it follows from that @xmath606 uniformly in @xmath601 . finally , we may conclude from , concavity of @xmath411 and the inequality @xmath607 that that uniformly for @xmath608 , @xmath609 these considerations prove .
it remains to analyze the maximum of @xmath560 and @xmath208 , respectively , over @xmath610 .
note first that by ( k.5 ) , @xmath611 on the other hand , for @xmath612 , implies that @xmath613 this proves the assertion about @xmath614 . as to the new confidence bounds ,
note first that by ( k.5 ) , @xmath615 where @xmath616 is a continuous function on @xmath46 with limit @xmath34 as @xmath112 .
consequently , @xmath617 is finite and @xmath618 + [ [ acknowledgements . ] ] acknowledgements .
+ + + + + + + + + + + + + + + + + we are grateful to guenther walther for stimulating talks about likelihood ratio tests in nonparametric settings .
in particular , rivera and walther ( 2013 ) inspired us to reformulate the law of the iterated logarithm in terms of processes with subexponential tails .
many thanks go to rudy beran for drawing our attention to bahadur and savage ( 1956 ) .
10 , _ the power to see : a new graphical test of normality _ , amer .
, 67 ( 2013 ) , pp .
249260 . , _ the nonexistence of certain statistical procedures in nonparametric problems _ , ann
, 27 ( 1956 ) , pp .
11151122 . ,
_ goodness - of - fit test statistics that dominate the kolmogorov statistics _ , z. wahrsch .
gebiete , 47 ( 1979 ) , pp .
4759 . , _ higher criticism for detecting sparse heterogeneous mixtures _ , ann .
, 32 ( 2004 ) , pp .
962994 . , _ new goodness - of - fit tests and their application to nonparametric confidence sets _ , ann
, 26 ( 1998 ) , pp .
288314 . , _ the calibrated kolmogorov - smirnov test _ , tech . rep .
, department of computer science , weizmann institute of science ; department of statistics , texas a&m university , 2013 .
( http://arxiv.org/abs/1311.3190 ) , _ on the law of the iterated logarithm _ ,
ann . of math .
( 2 ) , 43 ( 1942 ) , pp .
419436 . , _ probability inequalities for sums of bounded random variables _ , j. amer .
assoc . , 58 ( 1963 ) , pp .
1330 . , _ diffusion processes and their sample paths _ , springer - verlag , berlin , 1974 .
second printing , corrected , die grundlehren der mathematischen wissenschaften , band 125 . , _ goodness - of - fit tests via phi - divergences _ , ann .
, 35 ( 2007 ) , pp .
20182053 .
, _ iterated logarithm analogues for sample quantiles when @xmath619 _ , in proceedings of the 6th berkeley symposium on mathematical statistics and probability , vol . 1 , university of california , 1973 , pp .
227244 . ,
_ asymptotics in statistics some basic concepts _ , springer series in statistics , springer - verlag , new york , second ed . , 2000 . , _ a modified kolmogorov - smirnov test sensitive to tail alternatives _ , ann .
, 11 ( 1983 ) , pp .
933946 . , _ nonparametric likelihood confidence bands for a distribution function _ , j. amer .
assoc . , 90 ( 1995 ) , pp .
516521 . ,
_ optimal detection of a jump in the intensity of a poisson process or in a density with likelihood ratio statistics _ , scand . j. statist .
, 40 ( 2013 ) , pp .
752769 . , _ empirical processes with applications to statistics _ , wiley series in probability and mathematical statistics : probability and mathematical statistics , john wiley & sons inc . , new york , 1986 .
somewhat hidden in our proofs of lemmas [ lem : uniformep.1 ] and [ lem : uniformep.2 ] is a basic fact about moment generating functions which is stated in a slightly different form by rivera and walther ( 2013 ) and possibly of independent interest : suppose that @xmath95 is a real - valued random variable with mean @xmath620 and moment - generating function @xmath621 , @xmath622 we assume that @xmath623 in a neighborhood of zero .
in particular , all moments of @xmath95 are finite . a standard application of markov s inequality yields @xmath624 where @xmath625 the latter facts follow from the fact that @xmath626 is a convex function with derivative @xmath620 at @xmath34 .
note also that @xmath627 $ ] is a convex , lower semi - continuous function with @xmath628 and @xmath629 . from this one
can derive the following inequalities : by symmetry , it suffices to show that @xmath632 is not greater than @xmath633 . since @xmath634 $ ] is convex and lower semi - continuous with @xmath628 and @xmath635 , the point @xmath636 is well - defined .
when @xmath637 , convexity of @xmath314 and @xmath628 imply that @xmath638 for all @xmath639 . hence @xmath640 when @xmath641 , we may conclude from monotonicity and lower semicontinuity of @xmath314 that @xmath642 for all @xmath643 .
but this implies that @xmath644 + in case of @xmath645 , markov s inequality yields that @xmath646 the latter step is trivial but convenient for the next consideration : we may write @xmath647 with independent random variables @xmath648 and @xmath649 . moreover , it is well - known that @xmath650 and @xmath651 are stochastically independent with @xmath652 . consequently , by jensen s inequality and fubini s theorem , @xmath653 for @xmath654 .
( for @xmath655 the expectation of @xmath656 would be infinite . )
elementary calculations show that @xmath657 is maximal for @xmath658 , and this yields the bound @xmath659 in case of @xmath660 , the previous result may be applied to @xmath661 : @xmath662 + as in section [ sec : gof ] we consider the standard gaussian distribution function @xmath663 and the alternative distribution functions @xmath664 where @xmath665 and @xmath666 .
optimal tests of @xmath667 versus @xmath668 reject for large values of the log - likelihood ratio statistic @xmath669 with @xmath670 if @xmath522 is chosen such that @xmath671 then for any sequence of tests @xmath672 $ ] , @xmath673 see lecam and yang ( 2000 ) .
let @xmath684 be a @xmath685-confidence band for @xmath686 with a given class @xmath687 of distribution functions .
that means @xmath688 and @xmath689 are non - decreasing functions on the real line depending on the data vector @xmath690 such that @xmath691 we assume that @xmath687 is convex and satisfies @xmath692 for any @xmath686 and @xmath693 .
this is true if , for instance , @xmath687 corresponds to all mixtures of gaussian distributions with variance one .
then theorem 2 of bahadur and savage ( 1956 ) may be modified as follows : by symmetry , it suffices to prove the claim about @xmath700 . by monotonicity of @xmath700 , @xmath701
hence it suffices to show that @xmath702 for any single point @xmath22 and @xmath703 . to this end
consider @xmath704 for our given @xmath705 and some @xmath693 .
note that @xmath706 describes the distribution of @xmath707 with @xmath708 independent random variables @xmath709 and @xmath710 .
in particular , for any event @xmath711 , @xmath712 consequently , since @xmath713 , too , we may conclude from @xmath714 that @xmath715 but for sufficiently small ( negative ) @xmath620 , the value @xmath716 is greater than or equal to @xmath371
. then we may conclude that @xmath717 . | we present a general law of the iterated logarithm for stochastic processes on the open unit interval having subexponential tails in a locally uniform fashion .
it applies to standard brownian bridge but also to suitably standardized empirical distribution functions .
this leads to new goodness - of - fit tests and confidence bands which refine the procedures of berk and jones ( 1979 ) and owen ( 1995 ) . roughly speaking , the high power and accuracy of the latter procedures in the tail regions of distributions
are essentially preserved while gaining considerably in the central region .
@xmath0work supported by swiss national science foundation .
@xmath1work supported in part by nsf dms 11 - 04832 and ni - aid grant 2r01 ai291968 - 04 .
[ [ ams - subject - classifications . ] ] ams subject classifications .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + 62e20 , 62g15 , 62g30 .
[ [ key - words . ] ] key words .
+ + + + + + + + + + confidence band , limit distribution , sub - exponential tails , submartingale , tail regions . |
the discovery of the two strange - charmed mesons @xmath0 and @xmath4 with spin - parity @xmath5 and @xmath6 respectively has triggered hot debate on their nature and under - structures @xcite .
there have been a lot of explanations for their nature , for example , conventional @xmath7 states @xcite , two - meson molecular states @xcite , @xmath8 mixing states @xcite and four - quark states @xcite , etc . the mass of the @xmath0 is significantly lower than the values of the @xmath5 state mass from the quark models and lattice simulations @xcite .
those two states @xmath0 and @xmath4 lie just below the @xmath9 and @xmath10 threshold respectively which are analogous to the situation that the scalar mesons @xmath11 and @xmath12 lie just below the @xmath13 threshold and couple strongly to the nearby channels .
if we take the scalar mesons @xmath11 and @xmath12 as four - quark states with the constituents of scalar diquark - antidiquark sub - structures , the masses of the scalar nonet mesons below @xmath14 can be naturally explained . the mechanism responsible for the low - mass charmed scalar meson may be the same as the light scalar nonet mesons , the @xmath15 , @xmath16 , @xmath17 and @xmath18 @xcite . the one - gluon exchange force and the instanton induced force lead to significant attractions between the quarks in the @xmath5 diquark channels
, we can take the scalar diquark and antidiquark as the basic constituents in constructing the interpolating current @xcite , furthermore , in the color superconductivity theory , the attractive interactions in this channel lead to the formulation of nonzero condensates and the breaking of both the color and flavor @xmath19 symmetries for the light flavors @xcite . in this article
, we take the point of view that the charmed scalar meson @xmath0 be a tetraquark state consist of scalar diquark and antidiquark , and devote to calculate its mass in the heavy quark limit with the qcd sum rules @xcite .
the masses of the ground state and lowest excited state heavy mesons have been studied with qcd sum rules in heavy quark effective theory via the @xmath20 expansion @xcite .
the article is arranged as follows : we derive the qcd sum rules for the bound energy @xmath21 of the @xmath0 in the heavy quark limit in section ii ; in section iii , numerical results and discussions ; section vi is reserved for conclusion .
in the following , we write down the two - point correlation function @xmath22 in the framework of the qcd sum rules approach @xcite , @xmath23 here the @xmath24 are color indices and the @xmath25 is the charge conjugation matrix . in the heavy quark limit , the @xmath26 quark field can be approximated by the static heavy quark field @xmath27 with the propagator , @xmath28|0 \rangle=\frac{1+\!\not\ !
v}{2}\int_0^{\infty}\delta(x - vt)dt,\end{aligned}\ ] ] here the @xmath29 is a four - vector with @xmath30 .
the calculation of the operator product expansion can be performed in the coordinate space , and does not need the mixed picture both in coordinate and momentum spaces @xcite , in the heavy quark limit , the calculation can be greatly facilitated . according to the basic assumption of current - hadron duality in the qcd sum rules approach @xcite , we insert a complete series of intermediate states satisfying the unitarity principle with the same quantum numbers as the current operator @xmath31 into the correlation function in eq.(1 ) to obtain the hadronic representation . after isolating the pole term of the lowest @xmath0 state ,
we obtain the following result in the heavy quark limit , @xmath32 here @xmath33 and @xmath21 is the bound energy .
we perform operator product expansion up to the vacuum condensates of dimension-9 to obtain the correlation function @xmath22 at the level of quark - gluon degrees of freedom .
once the analytical results are obtained , then we can take the current - hadron duality below the threshold @xmath34 and perform the borel transformation with respect to the variable @xmath35 , finally we obtain the following sum rule , @xmath36 here the @xmath37 is the borel parameter . differentiate the above sum rule with respect to the variable @xmath38 , then eliminate the quantity @xmath39 , we obtain @xmath40 it is easy to integrate over the variable @xmath35 , we prefer this formulation for simplicity .
the parameters for the condensates are chosen to be the standard values at the energy scale @xmath41 , although there are some suggestions for updating those values , for reviews , one can consult ref.@xcite .
@xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 and @xmath48 .
small variations of those condensates will not lead to large changes for the numerical values and impair the predictive ability , we can neglect the uncertainties for the vacuum condensates for simplicity .
the mass of the @xmath49 quark from the particle data group is about @xmath50 @xcite , the values ( listed in the first article of ref.@xcite ) from the qcd sum rules , lattice qcd and @xmath51 decays vary in a large range , @xmath52 , here we take the average values from lattice qcd , @xmath53 .
the variations of the @xmath54 about @xmath55 can only lead to tiny changes for the final result , the uncertainties can be safely neglected , we take the value @xmath56 in numerical calculation for simplicity .
the values of the mass of the @xmath26 quark are @xmath57 at the energy scale @xmath58 from the particle data group @xcite .
the average values ( listed in the first article of ref.@xcite ) at the energy scale @xmath59 are @xmath60 from the qcd sum rules and lattice qcd .
we take the values @xmath60 and evolve them to the energy scale @xmath41 with the renormalization group equation , @xmath61 . in the following ,
we discuss the criterion for selecting the threshold parameter @xmath62 ( or @xmath34 ) and borel parameter @xmath63 ( or @xmath37 ) in the qcd sum rules dealing with the multiquark states .
the qcd sum rules have been extensively applied to the hadronic physics and given a lot of successful descriptions @xcite .
for the conventional ( two - quark ) mesons and ( three - quark ) baryons , the hadronic spectral densities are experimentally well known , the separations between the ground state and excited states are large enough , the `` single - pole @xmath64 continuum states '' model works well in representing the phenomenological spectral densities . in the phenomenological analysis ,
the continuum states can be approximated by the contributions from the asymptotic quarks and gluons , and the single - pole dominance condition can be well satisfied , @xmath65 for the conventional qcd sum rules , and @xmath66 for the qcd sum rules in the heavy quark limit , here the @xmath67 and @xmath68 stand for the contributions from the perturbative and non - perturbative part of the spectral density respectively . from the conditions in eqs.(7 - 8 ) , we can obtain the maximal value for the borel parameter @xmath69 ( or @xmath70 ) , exceed this value , the single - pole dominance will be spoiled . on the other hand , the borel parameter must be chosen large enough to warrant the convergence of the operator product expansion and contributions from the high dimension vacuum condensates which are poorly known are of minor importance , the minimal value for the borel parameter @xmath71 ( or @xmath72 ) can be determined . for the conventional ( two - quark ) mesons and ( three - quark ) baryons ,
the borel window @xmath73 ( or @xmath74 ) is rather large and the reliable qcd sum rules can be obtained . however , for the multiquark states i.e. tetraquark states , pentaquark states , etc , the spectral densities @xmath75 with @xmath76 is larger than the corresponding ones for the conventional hadrons , the integral @xmath77 ( or @xmath78 ) converges more slowly @xcite . if one do not want to release the conditions in eqs.(7 - 8 ) , we have to either postpone the threshold parameter @xmath62 ( or @xmath34 ) to very large values or choose very small values for the borel parameter @xmath69 ( or @xmath70 ) . with large values for the threshold parameter @xmath62 ( or @xmath34 ) , for example , @xmath79 , here the @xmath80 stands for the ground state , the contributions from the excited states are already included in if there are really some ones , the single - pole approximation for the spectral densities is spoiled ; on the other hand , with very small values for the borel parameter @xmath69 ( or @xmath70 ) , the operator product expansion is broken down , and the borel window shrinks to zero and beyond .
this may lead to the pessimistic opinion that the qcd sum rules can not be successfully applied to the multiquark states , the sum rules concerning the tetraquark states and pentaquark states should be rejected , however , we are optimistical participators for the qcd sum rules and take the point of view that the qcd sum rules can be successfully applied to the multiquark states , and one should resort to the `` multi - pole @xmath64 continuum states '' to approximate the phenomenological spectral densities .
the onset of the continuum states is not abrupt , the ground state , the first excited state , the second excited state , etc , the continuum states appear sequentially ; the excited states may be loose bound states and have large widths .
the threshold parameter @xmath62 ( or @xmath34 ) is postponed to large value , at that energy scale , the spectral densities can be well approximated by the contributions from the asymptotic quarks and gluons , and of minor importance for the sum rules .
the present experimental knowledge about the phenomenological hadronic spectral densities for the tetraquark states is rather vague , even the existence of the tetraquark states is not confirmed with confidence , and no knowledge about either there are high excited states or not . in this article , the following criteria are taken .
we choose the suitable values for the borel parameter @xmath63 ( or @xmath37 ) , on the one hand the minimal values @xmath71 ( or @xmath72 ) are large enough to warrant the convergence of the operator product expansion , on the other hand the maximal values @xmath69
( @xmath70 ) are small enough to suppress the contributions from the high excited states and continuum states i.e. we choose the naive analysis @xmath81 ( or @xmath82 ) , furthermore , there exist a borel platform which is insensitive to the variations of the borel parameter . for the hadronic spectral density , the more phenomenological analysis is preferred , we approximate the spectral densities with the contribution from the single - pole term ,
the threshold parameter @xmath62 ( or @xmath34 ) is taken slightly above the ground state mass , @xmath83 ( or @xmath84 ) , to subtract the contributions from the excited states and continuum states .
one may reject taking the values from the more phenomenological analysis as quantitatively reliable , the results are qualitative at least . in the heavy quark limit @xmath85 ,
the threshold parameter @xmath34 is chosen to vary between @xmath86 .
the values are reasonable for the scalar meson @xmath0 with narrow width , @xmath87 \geq(2.6 - 2.8)gev > m_{d_{s0}}+\gamma_{d_{s0}}$ ] , the contributions from the @xmath0 can be correctly taken into account .
for the values @xmath88 , the contributions from the perturbative term and the linear quark condensate terms proportional to the @xmath89 , @xmath90 are dominating i.e. @xmath91 with the minimal borel parameter @xmath92 , in this region , we can warrant the convergence of the operator product expansion . for the intervals ,
@xmath93 and @xmath94 , the main contributions to the bound energy @xmath21 come from the linear quark condensates terms proportional to the @xmath95 and @xmath96 , about @xmath97 , the perturbative contributions are suppressed by the large numerical denominator and of minor importance , about @xmath98 , which is significantly in contrary to the ordinary qcd sum rules for the conventional mesons and baryons where the main contributions come from the perturbative terms .
furthermore , the contributions from the gluon condensates are of minor importance due to the large numerical denominator @xcite . from the fig.1
, we can see that the predicted bound energy @xmath21 is almost independent on the borel parameter @xmath37 in the region @xmath99 . if we restrict the values of the borel parameter @xmath37 to the naive analysis @xmath100 , the predicted values are @xmath101 with @xmath94 . for @xmath102 ,
the contributions from the perturbative term and the linear quark condensate terms proportional to the @xmath89 , @xmath90 are larger than @xmath103 with the threshold parameter @xmath104 , the high dimensional condensates ( non - perturbative terms ) are greatly suppressed , this may be an indication the onset of the continuum states . for @xmath94 and @xmath105 , @xmath106gev , \nonumber \\ & = & [ ( 1.3 - 1.5)+(0.8 - 1.0)]gev , \nonumber \\ & = & ( 2.1 - 2.5)gev , \nonumber \\ & = & ( 2.3\pm0.2)gev .\end{aligned}\ ] ] comparing with the experimental data , the tetraquark configuration gives reasonable values for the mass of the scalar meson @xmath0 ; there must be some tetraquark component in the charmed scalar meson @xmath0 . with the borel parameter @xmath37 .
in this article , we take the point of view that the charmed scalar meson @xmath0 be a tetraquark state and devote to calculate its mass within the framework of the qcd sum rule approach in the heavy quark limit .
the numerical values for the mass of the @xmath0 are consistent with the experimental data .
there must be some tetraquark component in the scalar meson @xmath0 .
this work is supported by national natural science foundation , grant number 10405009 , and key program foundation of ncepu .
the authors are indebted to dr .
j.he ( ihep ) , dr .
x.b.huang ( pku ) and dr .
l.li ( gscas ) for numerous help , without them , the work would not be finished .
99 babar collaboration , phys .
* 90 * ( 2003 ) 242001 ; belle collaboration , phys . rev .
* 91 * ( 2003 ) 262002 ; cleo collaboration , phys . rev .
* d68 * ( 2003 ) 032002 ; babar collaboration , phys . rev . *
d69 * ( 2004 ) 031101 .
w. bardeen , e. eichten and c. hill , phys . rev .
* d68 * ( 2003 ) 054024 ; a. deandrea , g. nardulli and a. polosa , phys . rev . * d68 * ( 2003 ) 097501 ; s. godfrey , phys
* b568 * ( 2003 ) 254 ; p. colangelo and f. de fazio , phys . lett .
* b570 * ( 2003 ) 180 ; fayyazuddin and riazuddin , phys . rev .
* d69 * ( 2004 ) 114008 ; a. hayashigaki and k. terasaki , hep - ph/0411285 ; s. narison , phys
. lett . * b605 * ( 2005 ) 319 ; w. wei , p. z. huang and s. l. zhu , hep - ph/0510039 .
h. y. cheng and w. s. hou , phys . lett . * b566 * ( 2003 ) 193 ; k. terasaki , phys . rev . * d68 * ( 2003 ) 011501 ; r. l. jaffe and f. wilczek , phys . rev . lett . * 91 * ( 2003 ) 232003 ; a. hayashigaki and k. terasaki , hep - ph/0410393 ; t. browder , s. pakvasa and a.a .
petrov , phys . lett .
* b578 * ( 2004 ) 365 ; u. dmitrasinovic , phys . rev .
* d70 * ( 2004 ) 096011 ; l. maiani , f. piccinini , a.d .
polosa , v. riquer , phys . rev .
* d71 * ( 2005 ) 014028 ; u. dmitrasinovic , phys .
lett . * 94 * ( 2005 ) 162002 .
s. godfrey and n. isgur , phys . rev .
* d32 * ( 1985 ) 189 ; s. godfrey and r. kokoshi , phys . rev .
* d43 * ( 1991 ) 1679 ; g.s .
bali , phys . rev .
* d68 * ( 2003 ) 071501 ; a. dougall , r. d. kenway , c. m. maynard and c. mc - neile , phys
* b569 * ( 2003 ) 41 . t. v. brito , f. s. navarra , m. nielsen and m. e. bracco , phys .
b608 * ( 2005 ) 69 ; z. g. wang and w. m. yang , eur .
j. * c42 * ( 2005 ) 89 ; z. g. wang , w. m. yang and s. l. wan , j. phys . * g31 * ( 2005 ) 971 . a. de rujula , h. georgi and s. l. glashow , phys .
d12 * ( 1975 ) 147 ; t. degrand , r. l. jaffe , k. johnson and j. e. kiskis , phys . rev . * d12 * ( 1975 ) 2060 ; g. t hooft , phys . rev .
* d14 * ( 1976 ) 3432 [ erratum - ibid .
phys . rev .
* d18 * ( 1978 ) 2199 ] ; e. v. shuryak , nucl . phys .
* b203 * ( 1982 ) 93 ; t. schafer and e. v. shuryak , rev .
* 70 * ( 1998 ) 323 ; e. shuryak and i. zahed , phys . lett .
* b 589 * ( 2004 ) 21 .
m. a. shifman , a. i. and vainshtein and v. i. zakharov , nucl .
* b147 * ( 1979 ) 385 , 448 ; l. j. reinders , h. rubinstein and s. yazaky , phys
* 127 * ( 1985 ) 1 ; s. narison , qcd spectral sum rules , world scientific lecture notes in physics * 26*. d. j. broadhurst and a. g. grozin , phys .
lett . * b274 * ( 1992 ) 421 ; e. bagan , v. m. brown and h. g. dosch , phys . lett . * b278 * ( 1992 ) 457 ; m. neubert , phys . rev .
* d45 * ( 1992 ) 2451 ; phys .
* 245 * ( 1994 ) 259 ; p. ball and v. m. braun , phys . rev . * d49 * ( 1994 ) 2472 ; t. huang and c. w. luo , phys . rev .
* d50 * ( 1994 ) 5775 .
y. b. dai , c. s. huang , m. q. huang and c. liu , phys . lett .
* b390 * ( 1997 ) 350 ; y. b. dai , c. s. huang and m. q. huang , phys . rev .
* d55 * ( 1997 ) 5719 ; y. b. dai and s. l. zhu , phys . rev .
* d58 * ( 1998 ) 074009 ; p. colangelo , f. de fazio and n. paver , phys . rev . *
d58 * ( 1998 ) 116005 . | in this article , we take the point of view that the charmed scalar meson @xmath0 be a tetraquark state and devote to calculate its mass within the framework of the qcd sum rules approach in the heavy quark limit .
the numerical values for the mass of the @xmath0 are consistent with the experimental data , there must be some tetraquark component in the scalar meson @xmath0 .
detailed discussions about the threshold parameter and borel parameter for the multiquark states are also presented . + zhi - gang wang@xmath1 and shao - long wan@xmath2 + @xmath1 department of physics , north china electric power university , baoding 071003 , p. r. china + @xmath2 department of modern physics , university of science and technology of china , hefei 230026 , p. r. china + pacs number : 12.38.aw , 12.38.qk key words : @xmath3 , qcd sum rules |
when observed with high - sensitivity spectropolarimeters , the linearly polarized spectrum of the solar radiation coming from quiet regions close to the limb ( the so - called `` second solar spectrum '' ) reveals a complex structure and an impressive richness of spectral details @xcite . we know that the second solar spectrum is the observational signature of atomic polarization ( i.e. , population imbalances and quantum interference between different magnetic sublevels ) , produced by the absorption and scattering of anisotropic radiation .
however , the physics of scattering polarization in an optically thick plasma , such as the extended solar atmosphere , is very complicated and a general theory for multilevel systems is still lacking .
it is thus natural that many of the @xmath0 signals of the second solar spectrum were considered as enigmatic by their discoverers , as well as a true challenge for the theorists ( see the review by * ? ? ?
@xmath2 , with the reference direction for positive @xmath3 parallel to the nearest solar limb . ]
at present , the most challenging @xmath0 signals are those observed in the d@xmath1 lines of na i and ba ii at 5896 and 4934 , respectively ( see figures 2 and 5 of * ? ? ? * ) .
the aim of this letter is to point out a physical mechanism that might be of key importance for explaining such enigmatic linear polarization profiles .
the polarization signals observed by @xcite and @xcite in the d@xmath1 lines of na i and ba ii were considered enigmatic because these lines were expected to be intrinsically unpolarizable .
this is because they result from transitions between an upper and a lower level with @xmath4 ( @xmath5 being the total angular momentum ) .
such levels can not carry atomic alignment , so that no linear polarization can in principle be produced in these lines in weakly magnetized regions of the solar atmosphere .
( @xmath6 being the magnetic quantum number ) are unevenly populated .
when the upper and/or lower level of a line transition is aligned , then the emergent radiation is in general linearly polarized , even in the absence of a magnetic field ( e.g. , * ? ? ?
] however , 100% of sodium and about 18% of barium have hyperfine structure ( hfs ) .
both sodium ( @xmath7na ) and the barium isotopes with hfs ( @xmath8ba and @xmath9ba ) have nuclear spin @xmath10 . in such isotopes ,
the upper and lower @xmath5-levels of the d@xmath1 line transition thus split into two hfs @xmath11-levels ( @xmath12 and @xmath13 , with @xmath11 the total , electronic plus nuclear , angular momentum ) , which can carry atomic alignment .
in fact , taking the hfs of sodium into account , @xcite concluded that if there is a substantial amount of atomic alignment in the ground level ( i.e. , in the lower @xmath12 and @xmath13 levels of the sodium d@xmath1 and d@xmath14 lines ) , then a significant @xmath0 signal is produced in the d@xmath1 line . in his investigation of the sodium doublet ,
@xcite considered the unmagnetized case and assumed frequency - coherent scattering and a constant pumping radiation field within each d - line . in landi
deglinnocenti s ( 1998 ) modeling of the sodium d@xmath1 line , the ensuing @xmath0 profile vanishes if the alignment of the ground @xmath11-levels is destroyed .
such lower @xmath11-levels are long - lived and therefore their atomic alignment is very sensitive to depolarizing mechanisms , such as elastic collisions with neutral hydrogen atoms and/or the hanle effect of very weak ( mg ) non - vertical magnetic fields .
accordingly , @xcite concluded that the @xmath0 signal detected by @xcite in the sodium d@xmath1 line implies that depolarization does not occur in the lower solar chromosphere ( where the line - core of the sodium d - lines originates ) , which would seem to rule out the existence of tangled magnetic fields and of inclined , canopy - like fields stronger than 0.01 g. he also pointed out that this is difficult to understand because there is substantial evidence from other types of observations for both types of magnetic field .
this paradox was later reinforced by detailed theoretical investigations on the magnetic sensitivity of the d - lines of na i @xcite and of ba ii @xcite , as well as on the depolarizing role of elastic collisions with neutral hydrogen atoms @xcite .
it is also noteworthy that the scattering polarization profile obtained by @xcite is antisymmetric , in contrast with the symmetric peak observed by @xcite , but in agreement with the observations of other researchers ( see * ? ? ?
* for a review ) .
the physical mechanism discussed in this letter does not require the presence of atomic polarization in the lower level .
indeed , we will show that by properly taking into account hfs and partial frequency redistribution ( prd ) effects , even without any atomic polarization in the ( long - lived ) lower @xmath11-levels , the scattering of anisotropic d@xmath1-line radiation in the solar atmosphere can produce significant linear polarization in the d@xmath1 lines of ba ii and na i.
we model the intensity and linear polarization profiles of the d@xmath1 lines of ba ii and na i by solving the full non - lte radiative transfer problem for polarized radiation in one - dimensional semi - empirical models of the solar atmosphere , taking prd effects into account . to this aim
, we apply the redistribution matrix approach outlined in @xcite , which can be applied also to the case of a two - level atom with hfs , under the assumptions that the lower @xmath11-levels are unpolarized and infinitely sharp .
as previously discussed , these approximations appear to be suitable since the lower level of the d@xmath1 lines of ba ii and na i is the ( long - lived ) ground level .
we recall that our redistribution matrix is given by the linear combination of two terms , which describe purely coherent scattering ( @xmath15 ) and completely redistributed scattering ( @xmath16 ) in the atom rest frame , respectively .
we investigate the d@xmath1 line of barium accounting for the contribution of all its seven stable isotopes .
the two odd isotopes ( @xmath8ba and @xmath9ba , both with nuclear spin @xmath10 ) are described through a two - level model atom with hfs , the five even isotopes ( about 82.18% in abundance ) are described through a two - level model atom without hfs .
the d@xmath1 line of sodium is modeled through a two - level model atom with hfs ( sodium has a single stable isotope , @xmath7na , with @xmath10 ) . as clear from the grotrian diagram shown in the left panel of figure 1 , in the isotopes with hfs
, each upper @xmath11-level can be excited from the two lower @xmath11-levels through two transitions that are very close in frequency ( about 70 m apart in the barium isotopes with hfs , and about 20 m apart in sodium ) . due to such small frequency separation , the assumption that the pumping radiation field is the same in the two hfs transitions appears to be an excellent approximation for modeling these lines .
however , under this hypothesis , and assuming that the lower level is unpolarized , no atomic polarization can be produced in the upper @xmath11-levels , and therefore no polarization can be obtained in the emergent radiation .
this can be seen , for example , from the statistical equilibrium equations for a two - level atom with hfs derived in landi deglinnocenti & landolfi ( 2004 , hereafter ll04 ) . under the hypothesis of unpolarized lower level ,
such equations can be analytically solved , and the multipole components of the density matrix of the upper levels result to be given by @xmath17 where @xmath18 and @xmath19 are the einstein coefficients for spontaneous emission and for absorption , respectively , @xmath20 is the total number density of atoms , @xmath21 is the number density of atoms in the lower @xmath5-level , and @xmath22 is the frequency separation between the hfs levels @xmath23 and @xmath24 . since the incident field ( described in equation ( 1 ) through the radiation field tensor @xmath25 , see equation ( 5.157 ) of ll04 )
is assumed to be constant across the various hfs transitions ( this is one of the hypotheses the theory described in ll04 is based on ) , it is possible to perform the sum over @xmath26 , thus obtaining @xmath27 observing that the first 6-@xmath28 symbol in the second line of equation ( 2 ) vanishes for @xmath29 and @xmath30 , we immediately see that under the above - mentioned hypotheses , no atomic polarization ( either alignment or @xmath11-state interference , both being described by the multipole components with @xmath30 ) can be induced in the upper @xmath11-levels of the d@xmath1 line transition . indeed , in the previous investigations on this line , in which the pumping radiation field was assumed to be identical in the various hfs transitions , either because the investigation was carried out by applying the complete frequency redistribution ( crd ) theory presented in ll04 ( see * ? ? ?
* ) , or because this was assumed by definition , though considering coherent scattering ( see * ? ? ?
* ) , the authors concluded that the only way to obtain a scattering polarization signal in the core of the d@xmath1 line is by assuming that a given amount of atomic polarization is present in the lower level . in this investigation , on the contrary , we neglect lower level polarization but we take into account , through the @xmath15 term of the redistribution matrix , that the pumping radiation field can , in principle , be different in the various hfs transitions ( see the right panel of figure 1 for the case of barium ) .
although such difference is in many cases very small , we find that this allows atomic polarization to be induced in the upper @xmath11-levels , giving rise to an appreciable polarization signal in the core of the d@xmath1 line ( see the next section ) .
according to this physical mechanism , the amplitude of the linear polarization signal produced in the d@xmath1 line is expected to be higher when the hfs splitting of the lower level is larger . the two - level atom models ( either with or without hfs ) considered in this investigation
do not allow us to take quantum interference between the upper @xmath5-levels of the d@xmath1 and d@xmath14 lines into account .
this physical ingredient can actually be safely neglected for the investigation of the ba ii d@xmath1 line , since it is about 380 away from d@xmath14 , but it is known to play an important role in the modeling of the na i d - lines , which are only 6 apart .
on the other hand , an accurate modeling of the scattering polarization in the ba ii d@xmath1 and d@xmath14 lines would require to account for the radiative and collisional coupling of the upper levels of such lines with the metastable @xmath31 and @xmath32 levels .
this coupling , which is not present in neutral sodium since it does not have such metastable levels , can not be accounted for within our two - level atom modeling .
as we shall see below , in spite of such limitations , our approach is suitable to show that the physical mechanism identified in this investigation might be of key relevance for understanding the enigmatic d@xmath1 line polarization .
figure 2 shows the emergent stokes @xmath2 and @xmath0 profiles of the ba ii d@xmath1 line calculated in the semi - empirical model c of fontenla et al .
( 1993 ; hereafter , fal - c model ) .
the profiles in the left panel of figure 2 have been calculated in the limit of crd , following the theoretical approach of ll04 . as discussed in section 2 , within the framework of such theory
, the absorption of radiation from an unpolarized lower level can not induce any atomic polarization in the upper @xmath11-levels and therefore the d@xmath1 line must behave as intrinsically unpolarizable .
indeed , as shown in the lower left panel of figure 2 , the d@xmath1 line only produces a depolarization of the continuum polarization level ( our calculations account for the contribution of a coherent polarized continuum ) .
the right panels of figure 2 show the results of prd calculations . in this case , through the @xmath15 term of the redistribution matrix , we account for the exact frequency dependence of the pumping radiation field within the d@xmath1 line .
as expected , no signal is obtained if only @xmath33ba ( which is devoid of hfs ) is considered ( see the dotted line ) . however ,
if the contribution of the small fraction ( @xmath34% ) of the barium isotopes with hfs is taken into account , a conspicuous polarization signal is obtained for a line - of - sight ( los ) with @xmath35 ( @xmath36 being the cosine of the heliocentric angle ) .
the @xmath0 signal shows a large positive peak ( about 0.45% ) slightly blueshifted with respect to the line center , and a smaller negative peak ( about @xmath37% ) slightly redshifted . such @xmath0 peaks decrease , though remaining significant , for larger @xmath36 values .
as can be noted by comparing the top panels of figure 2 , the stokes @xmath2 profile is not very sensitive to prd effects , except at the very line center where the prd calculation gives slightly smaller intensity values .
particularly interesting , on the other hand , is the significant broadening of the stokes @xmath2 profile produced by the small fraction of barium isotopes with hfs ( compare the solid and dotted lines ) .
the results of figure 3 are also noteworthy .
the bottom panel shows the @xmath0 profiles calculated in three different semi - empirical models of the solar atmosphere ( the temperature structure of such models is plotted in the upper panel ) .
as can be observed , the @xmath0 signals are not very different .
this small sensitivity to the thermal model is good news , because the @xmath0 signal of the ba ii d@xmath1 line produced by the physical mechanism discussed in section 2 is sensitive to the upper - level hanle effect and therefore to magnetic fields in the gauss range . in figure 4
, we show the results that we have obtained for the d@xmath1 line of na i. the sensitivity of the @xmath0 signal of this stronger line to the thermal model is more noticeable , with the ( relatively cool ) fal - x model producing the largest peaks .
such @xmath0 peaks are much weaker than those observed by @xcite and than those obtained in the ba ii d@xmath1 line .
the calculated @xmath0 signals are much weaker in sodium than in barium because the hfs splitting of the ground level is about four times smaller in na i.
two physical mechanisms able to produce a scattering polarization signal in the na i and ba ii d@xmath1 lines have been identified up to now .
hfs is a fundamental ingredient in both of them .
such physical mechanisms are not mutually exclusive , and are probably at work simultaneously in the solar atmosphere , especially in the case of the na i d@xmath1 line .
one mechanism is that pointed out by @xcite , which requires the presence of a substantial amount of ground - level polarization . as clarified by @xcite , this mechanism , which operates even when the pumping radiation is constant within the d lines
( see also * ? ? ?
* ) , requires taking into account the absorption of anisotropic radiation in the d@xmath14 line ( whose upper level has @xmath38 ) and to assume that depolarizing mechanisms do not completely destroy the atomic alignment that _ repopulation pumping _ produces in the long - lived lower @xmath11-levels . under such circumstances , the atomic alignment of the lower and upper @xmath11-levels of the d@xmath1 line ( and therefore its scattering polarization signal ) are sensitive to the same ( mg ) magnetic field strengths . in this letter
, we have shown that the absorption and scattering of anisotropic d@xmath1-line radiation within the solar atmosphere can directly produce linear polarization in the d@xmath1 line , without any need for ground - level polarization .
this is possible if one takes into account that the solar d@xmath1-line radiation has a non - negligible spectral structure over the short frequency interval spanned by the hfs transitions .
the fact that small variations of the pumping radiation field between very close hfs transitions is able to create a significant amount of atomic polarization in the upper @xmath11-levels of the d@xmath1 lines is certainly an unexpected and important result .
the resulting @xmath0 signals are not so easily destroyed by elastic collisions and/or magnetic fields .
the amplitude and shape of the calculated @xmath0 profile of the ba ii d@xmath1 line are similar to the observed ones ( see figure 5 of * ? ? ?
the main discrepancy concerns the slightly redshifted peak , which reaches a negative value in the theoretical profile , while it remains positive in the observation . in our opinion
, such discrepancy does not invalidate our conclusion that the physical mechanism we have discussed in this letter may be a key ingredient for explaining the enigmatic linear polarization of the ba ii d@xmath1 line .
this is because our ba ii model atom does not include the metastable @xmath31 and @xmath32 levels , which may have a significant impact on the d@xmath1 line polarization ( see * ? ? ?
* concerning the ca ii ir triplet ) .
the amplitudes of the @xmath0 profiles observed in the d@xmath1 lines of na i and ba ii are of the same order of magnitude , but the calculated @xmath0 signal of the na i d@xmath1 line is one order of magnitude smaller .
this is due to fact that the hfs splitting of the ground level of sodium is about four times smaller than that of barium .
we believe that this discrepancy with the observation does not rule out the possibility that the physical mechanism described in this letter may play a key role also for explaining the enigmatic linear polarization of the na i d@xmath1 line .
we plan to include in our modeling the quantum interference between the @xmath11-levels pertaining to the two upper @xmath5-levels of the sodium doublet . this additional physical ingredient will probably influence the shape of the @xmath0 profile , even in the neighborhood of the sodium d@xmath1 line core .
in addition to this , we believe that if there is a non - negligible amount of ground - level polarization , then the physical mechanism discussed in this letter should induce a larger @xmath0 signal .
financial support by the spanish ministry of economy and competitiveness and the european feder fund through project aya2010 - 18029 ( solar magnetism and astrophysical spectropolarimetry ) is gratefully acknowledged .
avrett , e.h .
1995 , in infrared tools for solar astrophysics : what s next ?
, eds . j.r .
kuhn & m.j .
penn ( singapore : world scientific ) , 303 belluzzi , l. , & trujillo bueno , j. 2012 , , 750 , l11 belluzzi , l. , trujillo bueno , j. , & landi deglinnocenti , e. 2007 , , 666 , 588 casini , r. , landi deglinnocenti , e. , landolfi , m. , & trujillo bueno , j. 2002 , , 573 , 864 fontenla , j.m . ,
avrett , e.h . , & loeser , r. 1993 , , 406 , 319 kerkeni , b. , & bommier , v. 2002 , , 394 , 707 landi deglinnocenti , e. 1998 , , 392 , 256 landi deglinnocenti , e. , & landolfi , m. 2004 , polarization in spectral lines ( dordrecht : kluwer ) manso sainz , r. , & trujillo bueno , j. 2003 , phrvl , 91 , 111102 stenflo , j.o . , & keller , c.u .
1997 , , 321 , 927 stenflo , j.o , keller , c.u . , &
gandorfer , a. 2000 , , 355 , 789 trujillo bueno , j. 2009 , in asp conf .
405 , solar polarization 5 , eds .
berdyugina , k.n .
nagendra , and r. ramelli , ( san francisco , ca : asp ) , 65 trujillo bueno , j. , casini , r. , landolfi , m. , & landi deglinnocenti , e. 2002 , , 566 , 53 line transition are drawn on the diagram .
right panel : anisotropy ( expressed through the ratio between the @xmath39 and @xmath40 components of the radiation field tensor ) of the solar radiation across the ba ii d@xmath1 line at 800 km in the fal - c model atmosphere ( the height where the line center optical depth is unity for a los with @xmath41 ) .
the vertical lines indicate the wavelength positions of the four hfs components of @xmath9ba , and of the fs component of @xmath33ba.,scaledwidth=100.0% ] ( lower panel ) profiles of the ba ii d@xmath1 line , as calculated by applying the crd theory presented in ll04 , considering the fal - c atmospheric model , and a los with @xmath41 .
the solid line represents the solution obtained by considering the contribution of all the seven isotopes of barium .
the dotted line represents the solution obtained by assuming that 100% of barium is @xmath33ba ( which is devoid of hfs ) .
right : same as left panels , but for prd calculations.,scaledwidth=100.0% ] profiles of the na i d@xmath1 line calculated in the semi - empirical models c and f of @xcite , and in the semi - empirical model m@xmath42 ( also known as fal - x ) of @xcite , for a los with @xmath41.,scaledwidth=70.0% ] | the linearly polarized spectrum of the solar limb radiation produced by scattering processes is of great diagnostic potential for exploring the magnetism of the solar atmosphere .
this spectrum shows an impressive richness of spectral details and enigmatic @xmath0 signals , whose physical origin must be clearly understood before they can be exploited for diagnostic purposes .
the most enduring enigma is represented by the polarization signals observed in the d@xmath1 resonance lines of na i ( 5896 ) and ba ii ( 4934 ) , which were expected to be intrinsically unpolarizable .
the totality of sodium and 18% of barium have hyperfine structure ( hfs ) , and it has been argued that the only way to produce a scattering polarization signal in such lines is through the presence of a substantial amount of atomic polarization in their lower hfs levels . the strong sensitivity of these long - lived levels to depolarizing mechanisms led to the paradoxical conclusion that the observed d@xmath1-line polarization is incompatible with the presence in the lower solar chromosphere of inclined magnetic fields sensibly stronger than 0.01 g. here we show that by properly taking into account the fact that the solar d@xmath1-line radiation has a non - negligible spectral structure over the short frequency interval spanned by the hfs transitions , it is possible to produce scattering polarization signals in the d@xmath1 lines of na i and ba ii without the need of ground - level polarization .
the resulting linear polarization is not so easily destroyed by elastic collisions and/or magnetic fields . |
the top - mass is a key parameter in the standard model .
for example , it plays an important role for analyzing electroweak precision constraints , for predicting rare decays like @xmath3 and @xmath4 , as well as for unraveling the higgs sector in supersymmetric models .
the latest measurement from the tevatron @xcite , @xmath5 , is of very high precision , and brings to mind several theoretical questions . in what mass - scheme is the value quoted ?
are measurements of the mass of a colored particle with a precision better than @xmath6 possible using jets , or is this an irreducible uncertainty ?
how do perturbative and nonperturbative effects modify the relation between the experimental observable and the underlying lagrangian mass parameter ? for the lhc a @xmath7 systematic uncertainty was obtained from preliminary atlas studies @xcite ( where hurdles to go beyond this level include understanding the jet - energy scale to better than 1% ) . at this level ,
understanding the answers to the above questions is important . to achieve a high precision top - mass , measurements exploit kinematic information by considering leptons + jets , @xmath8 , or the associated dilepton or all - hadronic channels , and use monte carlo ( mc ) simulations to reconstruct observables sensitive to the top - quark four vector , and
hence @xmath9 .
one of the most sensitive observables in this reconstruction are the invariant masses @xmath10 and @xmath11 of the jets and other decay products produced by the top and antitop respectively . here
the sets @xmath12 and @xmath13 depend on the jet - algorithm and cuts . in such analyses
it might appear natural to think of the reconstructed top - mass as being associated to the pole - mass @xmath14 , since it is designed for sensitivity to the on - shell region @xmath15 .
however , as we will argue in section 5 below , it is not @xmath14 that is being measured by the tevatron analyses . the relation of the pole - mass and any other lagrangian mass - scheme @xmath16 can be expressed as a perturbative series , @xmath17 where @xmath18^n \ln^k\!\big(\frac{\mu}{r}\big),\end{aligned}\ ] ] with @xmath19 a dimension-1 scale intrinsic to the scheme , and @xmath20 finite numerical coefficients .
theoretically the pole - mass is a poor scheme choice since in qcd its definition is ambiguous by an amount of @xmath21 ( in perturbation theory this is referred to as the infrared renormalon problem ) .
nice mass schemes avoid this problem by a suitable choice for the @xmath20 s , and are known as short - distance mass schemes . using observables expressed in terms of short - distance schemes the accuracy that the mass of a colored particle can be measured
is not limited by @xmath22 .
this fact is important to obtain the @xmath23 uncertainties for current measurements of the @xmath13-quark mass @xcite . for current top - measurements based on reconstruction
there is another important restriction on viable short - distance schemes . in these kinematic - based analyses
the top - quark decay is treated with a breit - wigner in the mc s , and only `` top - resonance mass schemes '' with a small @xmath24 are compatible with the breit - wigner form @xcite .
this restriction rules out directly measuring the @xmath25 mass , which has a much larger @xmath26 . to explore these issues in detail
it is useful to consider a case with full analytic control , namely jets produced in @xmath0 with c.m .
energy @xmath27 . here the jet - invariant masses @xmath28 sum over particles in the top and antitop hemisphere defined with respect to the thrust axis , and are examples of event shape variables for massive quarks .
the resonance region @xmath29 is most sensitive to the top - mass .
the appropriate factorization theorem was derived in ref .
@xcite , @xmath30 where @xmath31 and @xmath32 .
( [ fthm ] ) is valid to all orders in @xmath33 and has power corrections of @xmath34 @xmath35 . here
@xmath36 and @xmath37 are perturbative coefficients describing the hard scales @xmath38 and @xmath39 , as well as the summation of large logarithms from @xmath40 , where @xmath41 .
the @xmath42 are perturbatively calculable heavy - quark jet functions , which describe the evolution and subsequent decay of the top / antitop quark to jets , and include a resummation of large logs between @xmath43 and the soft scale @xmath44 @xmath45 .
a complete resummation of large logs at next - to - leading order ( nll ) was carried out in ref .
@xcite , where the reader can find further details . in eq .
( [ fthm ] ) @xmath39 and @xmath46 are specified in a top - resonance mass - scheme , of which an example known as the jet - mass is discussed below in section 2 . finally , @xmath47 is the hemisphere soft - function describing the soft - radiation between jets .
@xmath47 is perturbatively calculable for @xmath48 and non - perturbative for @xmath49 .
the parameter @xmath50 is a gap in the soft function @xcite , while @xmath51 indicates that the scheme for this gap is defined so that both perturbative and nonperturbative contributions are included without inducing a leading renormalon ambiguity , as discussed below in section 3 .
the result in eq .
( [ fthm ] ) relates an experimentally measurable hadronic observable , @xmath52 , to the short distance mass @xmath39 . as an example , consider the peak position of the invariant mass distribution @xmath53 .
schematically this relation has the form @xmath54 where the perturbative shifts @xmath55 can be calculated from the jet - functions , @xmath42 , and the nonperturbative shift @xmath56 is completely determined by the soft - function @xmath47 . having a systematic separation of the perturbative and non - perturbative shifts into @xmath42 and @xmath47 is important for a precision determination of the lagrangian mass @xmath39 .
in particular the value of the terms in the perturbative series in eq .
( [ mpeak ] ) are related to the choice of mass - scheme for @xmath39 , and are modified by different scheme choices .
thus , control over the mass - scheme requires control over these corrections .
note that the non - perturbative shift is always positive for hemisphere masses , and quite sizeable , @xmath57@xmath58 for typical values of @xmath38 .
in sections 2 and 3 below we discuss the functions responsible for these perturbative and nonperturbative shifts in more detail .
then in section 4 we quantitatively analyze the relation between @xmath59 and @xmath39 using eq .
( [ fthm ] ) .
finally , in section 5 we return to implications for the tevatron mass , and discuss a very useful mass - scheme for measuring both top and bottom heavy - quark masses , the msr - scheme .
since @xmath60 the jet function can be formulated in the heavy - quark limit with hqet , @xmath61 here @xmath62 encodes the mass - scheme choice , and @xmath63 is the total width of the top - quark which suffices since our observable is inclusive in the decay products .
the heavy - quark jet function is defined as the imaginary part of a @xmath64 propagator connected by a light - like wilson line @xcite , @xmath65 $ ] where @xmath66 at tree level in @xmath33 we have a breit - wigner , @xmath67 $ ] . from eq .
( [ lh ] ) the jet function obeys the shift identity @xmath68 .
it also has an rge @xmath69 where the anomalous dimension has a plus - function term with the universal cusp anomalous dimension and a @xmath70-function term , @xmath71/\mu [ \mu\theta(\hat s)/\hat s]_+ + \gamma_b[\alpha_s ] \delta(\hat s)$ ] .
summing logarithms from @xmath43 to @xmath44 gives @xmath72 which is the function @xmath73 appearing in eq .
( [ fthm ] ) .
( [ brun ] ) sums all logs that can affect the shape of @xmath74 @xcite .
there is some freedom in combining fixed order results for @xmath75 at @xmath43 with the rge in @xmath76 .
we define a counting that elevates the importance of fixed order results @xmath77 one - loop results for @xmath75 and @xmath78 were computed in ref .
@xcite , and two - loop results in ref .
the three - loop result for @xmath79 is known from ref .
@xcite , so @xmath75 is known at nnll .
results for @xmath75 in moment space have also been computed @xcite .
a different heavy quark jet function also occurs in the massive quark form factor , whose anomalous dimension shares several coefficients in common with @xmath78 @xcite . in order to implement a short - distance mass - scheme and subtract the infrared contributions related to the pole - mass renormalon we must systematically expand in @xmath46 to the same level in @xmath33 that we determine @xmath75 itself .
writing @xmath80 ^ 2 \delta m_2 \big\ } \,,\end{aligned}\ ] ] with @xmath24 , the remaining freedom in specifying the mass - scheme at two - loop order corresponds to defining @xmath81 .
the jet - function itself can be used to define a jet - mass scheme , @xmath82 @xcite .
it is convenient to use the position space jet - function @xmath83 so that @xmath84 has a consistent @xmath85-anomalous dimension @xcite , and we define @xmath86 this definition for the jet mass is possible for any value of @xmath19 , and the anomalous dimension in @xmath19 can also be consistently derived @xcite . in fig .
[ fig : b ] ( top panel ) we show results for eq .
( [ brun ] ) using @xmath87 , the jet - mass with reference value @xmath88 , @xmath89 , and three curves at each of ll , nll , and nnll order corresponding to @xmath90 .
convergence by nnll is evident . in the bottom panel of fig .
[ fig : b ] we compare the pole and jet - mass schemes and show that very good convergence for the peak position is achieved in the jet - mass scheme .
the hemisphere soft function is defined by a matrix element of wilson lines @xmath91 where @xmath92 and @xmath93 is similar but in the @xmath94 representation . here
@xmath95 is the sum of all plus - momenta of particles in @xmath96 in hemisphere @xmath12 , and @xmath97 is the sum of all minus - momenta for particles in hemisphere @xmath13 .
thus the @xmath70-functions in eq .
( [ sdef ] ) are a reflection of the prescription for assigning soft - radiation to the invariant masses @xmath98 and @xmath99 .
this soft function is universal @xcite , the same function appears in the factorization theorem for the hemisphere invariant mass distribution as well as the thrust and heavy jet mass distributions of massless jets @xcite .
@xmath100 has both perturbative and non - perturbative components , and a convenient way to account for this is @xcite @xmath101 here @xmath102 refers to a partonic computation of eq .
( [ sdef ] ) and encodes the proper @xmath44 dependence and large @xmath103 behavior into @xmath47 .
@xmath104 is a model for the non - perturbative @xmath105 region of @xmath47 .
a gap @xmath106 is contained in @xmath104 where @xmath50 is defined in a scheme specified by @xmath107 .
the partonic soft - function obeys @xmath108 . to avoid a @xmath109 renormalon ( which has nothing to do with the @xmath109 renormalon of quark masses )
, we must systematically expand in @xmath110 to the same level in @xmath33 as we expand @xmath111 @xcite .
we write @xmath112 ^ 2
\delta \delta_2 \big\ } \,.\end{aligned}\ ] ] removing the renormalon with @xmath51 avoids having to make large changes to @xmath104 each time an additional order in @xmath33 is included in @xmath102 .
much like eq .
( [ dmj ] ) we can define @xmath110 with the position space soft function @xcite , which yields a scheme for @xmath113 with a consistent rge in @xmath85 and @xmath19 @xmath114 results for @xmath115 are known at one - loop order @xcite and two - loop order @xcite , and for the thrust soft - function in refs @xcite .
scheme and scheme from eq .
( [ ddelta ] ) .
plot from @xcite . ] in fig .
[ fig : s ] we compare results for eq .
( [ sconv ] ) at lo ( black ) , nlo ( blue ) , and nnlo ( red ) , in the @xmath116 scheme ( dotted & dot - dashed ) and in the @xmath50-scheme ( dashed & solid ) . for @xmath104 we use the 2d - exponential model of ref .
@xcite with the addition of the gap @xmath50 .
the negative dips at small @xmath117 that are present in the @xmath116 scheme , are removed in the renormalon free @xmath50-scheme , and a stable result is obtained for the peak - position of the soft function .
we now proceed to put together all the ingredients in eq .
( [ fthm ] ) at nll order @xcite ( since the full nnll results have not yet been published ) .
the cross - section depends on @xmath118 , @xmath119 , @xmath43 , and @xmath44 , and the dependence on these four renormalization scales is reduced from ll to nll order .
the largest variation occurs for @xmath43 and @xmath44 , but is highly correlated .
it becomes significantly smaller if we vary fixing @xmath120 , which is consistent with the scaling given below eq .
( [ fthm ] ) .
fig.[fig : sig ] ( top panel ) shows the invariant mass distribution for @xmath121 where the three curves at each order have @xmath122 and the same reference top - mass of @xmath123 was used as in fig .
[ fig : b ] .
( bottom ) .
vertical black lines indicate theory uncertainties and the colored bands linear extrapolations to @xmath124 .
plots from @xcite . ]
( bottom ) .
vertical black lines indicate theory uncertainties and the colored bands linear extrapolations to @xmath124 .
plots from @xcite . ]
the @xmath125 renormalons associated to the pole - mass and the partonic soft - function , both would cause bigger shifts to the location of the invariant mass peak than the difference ( @xmath126 ) between ll and nll order shown in fig .
[ fig : sig ] . thus using mass and gap schemes that avoid
these renormalons was important to obtain perturbative stability .
the largest correction to @xmath127 is due to the soft radiation , which causes the majority of the difference between the dashed and solid vertical lines of fig.[fig : sig ] ( top panel ) .
the advertised shift , @xmath128 , is demonstrated in fig .
[ fig : sig ] ( bottom ) where we show that linear dependence on @xmath38 occurs for six different models for @xmath129 . for all models the extrapolations to @xmath124 converge to the underlying lagrangian mass value @xmath39 . in an @xmath130 environment
the correction @xmath127 can be determined a ) from measurements of @xmath47 for massless jets that together with perturbative computations determine the shift as in fig .
[ fig : sig ] ( top ) , or b ) from measurements at different @xmath38 s with a linear extrapolation that removes the soft - radiation effect as in fig . [ fig : sig ] ( bottom ) .
the situation at high energy hadron colliders is somewhat different from @xmath130 collisions due to the more complicated initial state and because the jet algorithms needed for the reconstruction of the invariant mass distribution are implemented in terms of mc programs rather than a factorization theorem . nevertheless , the principles upon which final state interactions in mc s for tops are based , have close analogues with elements in the factorization theorem ( [ fthm ] ) .
these analogies make it obvious that the top quark masses contained in mc programs , which are measured at the tevatron and lhc , are very similar to the jet mass definition we discussed above .
moreover , the tevatron top mass measurements are statistically dominated by the reconstructed invariant mass distribution in the peak region , which allows more reliable statements to be made about the scheme of the tevatron top mass .
the final state shower in mc s describes the perturbative aspects of collinear and soft radiation of partons .
starting at transverse scales of order the momentum transfer of the primary partons the parton shower evolves the system down to the shower cutoff scale @xmath131 which is typically in the 1 gev range . for the top quark the width @xmath132 gev provides an additional natural shower cutoff for radiation off the top quark . in the factorization theorem ( [ fthm ] ) the same physics
is described by the renormalization group running of the factors @xmath36 and @xmath37 , the partonic contributions in the soft function and , in particular , the jet functions @xmath42 .
as far as the question of the mass scheme implemented in mc s is concerned , the analogue of the shower cutoff @xmath131 is the modification of the jet functions @xmath42 due to the residual mass term @xmath46 , which subtracts low - energy fluctuations of the jet functions , absorbing them into the mass definition .
it is therefore the implementation of the parton shower for the top quark and the size of the shower cutoff @xmath131 which determines the top mass definition used in the mc s .
another important ingredient of mc s is the description of nonperturbative effects in the hadronization process through models that depend on many parameters .
these are fixed from reference processes . for the factorization theorem ( [ fthm ] )
the analogue of the hadronization models is the soft function which can be determined from event - shapes involving light quark jets .
we note that at hadron colliders the treatment of nonperturbative effects is more involved due to the partonic initial state , and does not necessarily lead to a positive shift of the peak position . for hadron collisions at the tevatron and the lhc
there are additional complications due to for example underlying events , more involved combinatorial background and initial state showers .
however , if such effects are modeled correctly in the mc s they do not affect the correspondences mentioned above .
we therefore conclude that the top quark mass that is implemented in mc s has the property @xmath133 + \ldots\end{aligned}\ ] ] with @xmath134 gev .
the exact coefficients in the series on the rhs depend on how the parton shower is implemented and are currently unknown . however , it is reasonable to assume that they are of order one .
numerically , the mc masses are therefore quite close to the jet mass scheme defined in ( [ dmj ] ) for which @xmath135 or to any other short - distance mass scheme that falls in the category of ( [ mrdef ] ) with @xmath135 .
a good way to illustrate the numerical size of the scheme uncertainty due to the ignorance of the coefficients in ( [ mcmassdef ] ) is the associated uncertainty in the @xmath136 top mass @xmath137 .
this uncertainty is essential for fits of electroweak precision observables and the resulting indirect higgs mass bounds . to avoid large logarithms in the perturbative relation of the mc or the jet mass ( with @xmath138 gev ) and the @xmath136 mass ( with @xmath139 ) , one can use the evolution equation in @xmath19 from ref .
conversions involving the @xmath136 mass @xmath137 are particularly simple when the r - evolution is carried out with the msr - mass scheme @xmath140 , which is a short - distance mass scheme that also falls in the category of ( [ mrdef ] ) and has in addition the property @xmath141 .
the msr mass definition is obtained from the @xmath25-pole series by a simple replacement rule @xcite .
we can identify the msr mass at a low scale @xmath19 in the 1 gev range with the mass that is contained in mc programs , @xmath142 the variation of @xmath19 between 1 and 9 gev parameterizes our remaining ignorance .
we now identify @xmath143 with the tevatron measurement @xmath144 @xcite . the result for @xmath137
is illustrated in fig .
[ fig : m ] .
the vertical error bars on the tevatron mass are the experimental uncertainties and the horizontal error bars reflect the current scheme uncertainty due to eq .
( [ mteverror ] ) .
the three red lines show the 3-loop r - evolution of the scheme uncertainty with the msr mass .
the intersection of the red lines with the black line that shows the equation @xmath145 , gives the @xmath136 top mass @xmath146 gev @xcite .
here the first error is the combined experimental one , and the second error is from the scheme uncertainty in the tevatron mass .
the above correspondence between the shower mc and the factorization theorem is compelling for large @xmath147 events where top quarks are fast .
a potential concern is that tops at the tevatron are produced with predominantly small @xmath147 and are therefore slow .
however given that the mc framework applies to this situation , the concept of the top - mass intrinsic to the mc should be independent of whether the mc is applied to energetic tops or to soft tops , and hence the correspondence we have discussed should be applicable for the tevatron as well .
further study of this is warranted .
this work was supported in part by the department of energy office of nuclear science under the grant de - fg02 - 94er40818 , and the eu network contract mrtn - ct-2006 - 035482 ( flavianet ) .
iws was also supported by the doe oji program and sloan foundation .
t. t. e. group _ et al .
arxiv:0803.1683 [ hep - ex ] . i. borjanovic _ et al .
_ , eur .
j. c * 39s2 * , 63 ( 2005 ) [ arxiv : hep - ex/0403021 ] . w. m. yao _ et al .
_ [ particle data group ] , j. phys .
g * 33 * , 1 ( 2006 ) .
s. fleming , a. h. hoang , s. mantry and i. w. stewart , phys .
d * 77 * , 074010 ( 2008 ) [ arxiv : hep - ph/0703207 ] .
s. fleming , a. h. hoang , s. mantry and i. w. stewart , phys .
d * 77 * , 114003 ( 2008 ) [ arxiv:0711.2079 [ hep - ph ] ] .
a. h. hoang and i. w. stewart , phys .
b * 660 * , 483 ( 2008 ) [ arxiv:0709.3519 [ hep - ph ] ] .
a. jain , i. scimemi and i. w. stewart , phys .
d * 77 * , 094008 ( 2008 ) [ arxiv:0801.0743 [ hep - ph ] ] .
s. moch , j. a. m. vermaseren and a. vogt , nucl .
b * 688 * , 101 ( 2004 ) [ arxiv : hep - ph/0403192 ] .
a. mitov and s. moch , jhep * 0705 * , 001 ( 2007 ) [ arxiv : hep - ph/0612149 ] .
u. aglietti , l. di giustino , g. ferrera and l. trentadue , phys .
b * 651 * , 275 ( 2007 ) [ arxiv : hep - ph/0612073 ] ; arxiv:0804.3922 [ hep - ph ] .
a. h. hoang , a. jain , i. scimemi and i. w. stewart , arxiv:0803.4214 [ hep - ph ] .
g. p. korchemsky and g. sterman , nucl .
b * 555 * , 335 ( 1999 ) [ arxiv : hep - ph/9902341 ] ; c. w. bauer , c. lee , a. v. manohar and m. b. wise , phys .
d * 70 * , 034014 ( 2004 ) [ arxiv : hep - ph/0309278 ] .
a. h. hoang and s. kluth , arxiv:0806.3852 [ hep - ph ] . m. d. schwartz , phys .
d * 77 * , 014026 ( 2008 ) [ arxiv:0709.2709 [ hep - ph ] ] ; t. becher and m. d. schwartz , jhep * 0807 * , 034 ( 2008 ) [ arxiv:0803.0342 [ hep - ph ] ] .
g. p. korchemsky and s. tafat , jhep * 0010 * , 010 ( 2000 ) [ arxiv : hep - ph/0007005 ] . | theoretical issues are discussed for the measurement of the top - mass using jets , including perturbative and non - perturbative effects that relate experimental observables to the lagrangian mass , and appropriate choices for mass schemes . full account for these issues
is given for @xmath0 using a factorization theorem for event shapes for massive quarks .
implications for the tevatron top - mass measurement are discussed .
a mass - scheme , the `` msr - mass '' , is introduced which allows for a precise description of observables sensitive to scales @xmath1 , but at the same time does not introduce perturbative matching uncertainties in conversion to the @xmath2 mass . |
the notion of a _ weyl group _ is crucial in linear algebraic group theory @xcite .
the seminal example occurs when one considers the _ algebraic group _ @xmath0 . in that case
, the associated weyl group is isomorphic to the group of monomial matrices , that is to the permutation group @xmath1 .
weyl groups are special examples of _ finite coxeter groups_. hence , they possess a group presentation of a particular type , and an associated length function .
it turns out that this presentation and this length function are deeply related with the geometry of the associated algebraic group .
linear algebraic monoid theory , mainly developed by putcha , renner and solomon , has deep connections with algebraic group theory . in particular , the _ renner monoid _ @xcite plays the role that the weyl group does in linear algebraic group theory . as far as i know , in the case of renner monoids , there is no known theory that plays the role of coxeter group theory .
therefore it is natural to look for such a theory , and , therefore , to address the question of monoid presentations for renner monoids . in @xcite ,
we considered the particular case of the _ rook monoid _ defined by solomon @xcite .
we obtained a presentation of this group and introduced a length function , that is nicely related to the hecke algebra of the rook monoid .
our objective here is to consider the general case .
we obtain a presentation of every renner monoid and introduce a length function . in the case of the rook monoid , we recover the results obtained in @xcite .
our length function is not the classical length function on renner monoids @xcite .
we remark that the former shares with the latter several nice geometrical and combinatorial properties .
we believe that this new length function is nicely related to the hecke algebra , as in the special case of the rook monoid .
let us postpone to the next sections some definitions and notations , and state here our main results .
consider the renner monoid @xmath2 of a linear algebraic monoid @xmath3 .
denote by @xmath4 the unit group of @xmath2 and consider its associated coxeter system @xmath5 .
denote by @xmath6 a _ cross section lattice _ of the monoid @xmath7 of idempotent elements of @xmath8 , and by @xmath9 the set of elements of @xmath6 that are distinct from the unity .
denote finally by @xmath10 the associated _ type map _ of @xmath2 ; roughly speaking , this is a map that describes the action of @xmath4 on @xmath7 .
the renner monoid @xmath2 admits the monoid presentation whose generating set is @xmath11 and whose defining relations are : + [ cols= " < , < , < " , ] [ thepreserennmonid ] note that when @xmath12 and @xmath13 , then relation ( tym3 ) becomes @xmath14 .
more generally , one has @xmath15 . by remark [ remgenrset ] the submonoids @xmath16 and @xmath17 generate the monoid @xmath2 .
as @xmath18 is a generating set for @xmath17 , it follows from the definition of @xmath6 that the set @xmath11 generates @xmath2 as a monoid .
clearly , relations ( cox1 ) and ( cox2 ) hold in @xmath17 , relations ( tym1 ) and ( tym2 ) hold in @xmath2 . relations ( tym3 ) hold in @xmath2 by proposition [ propconslong ] .
it remains to prove that we obtain a presentation of the monoid @xmath2 .
let @xmath19 belong to @xmath2 with @xmath20 as normal form .
consider any word @xmath21 on the alphabet @xmath11 that represents @xmath19 .
we claim that starting from @xmath21 , one can obtain the word @xmath22 using the relations of the above presentation only .
this is almost obvious by induction on the number @xmath23 of letters of the word @xmath21 that belong to @xmath9 .
the property holds for @xmath24 ( in this case @xmath25 and @xmath26 ) because ( cox1 ) and ( cox2 ) are the defining relations of the presentation of @xmath17 .
the case @xmath27 is also clear , applying relations ( cox1 ) , ( cox2 ) , ( tym1 ) and ( tym2 ) .
now , for @xmath28 , the case @xmath23 can be reduced to the case @xmath29 using relations ( tym3 ) ( and the other relations ) .
the presentation in proposition [ thepreserennmonid ] is not minimal ; some relations can be removed in order to obtain the presentation stated in theorem [ thepreserennmonidintro ] .
let us introduce a notation used in this theorem : if @xmath30 lies in @xmath6 , we denote by @xmath31 the set @xmath32 .
similarly , we denote by @xmath33 the set @xmath34 .
\(i ) @xmath35 by proposition [ propopo ] .
+ ( ii ) @xmath36 the reader may note that for @xmath12 one has @xmath37 .
we need to prove that every relations @xmath38 of type ( tym3 ) in proposition [ thepreserennmonid ] can be deduced from relations ( rbi ) of theorem [ thepreserennmonidintro ] , using the other common defining relations of type ( cox1 ) , ( cox2 ) , ( tym1 ) and ( tym2 ) .
we prove this by induction on the length of @xmath19 . if @xmath39 then @xmath19 is equal to @xmath40 and therefore belongs to @xmath41 .
assume @xmath42 and @xmath19 does not belong to @xmath41 .
assume furthermore @xmath19 does not lie in @xmath43 ( the other case is similar ) .
choose @xmath44 in @xmath9 such that @xmath45 and @xmath19 does not lie in @xmath46 .
then , applying relations ( rib ) , we can transform the word @xmath47 into the word @xmath48 . using relations ( cox2 ) ,
we can transform the word @xmath49 into a word @xmath50 where @xmath51 belongs to @xmath46 and @xmath52 belongs to @xmath53 .
then , applying relations ( cox2 ) and ( tym1 ) , we can transform the word @xmath48 into the word @xmath54 . by hypothesis on @xmath19 , we have @xmath55 and , therefore , @xmath56 .
assume @xmath52 belongs to @xmath57
. we can apply relation ( rib ) to transform @xmath54 into @xmath58 . using relations ( cox2 )
, we can transform @xmath59 into a word @xmath60 with @xmath61 in @xmath62 , @xmath63 in @xmath64 and @xmath65 in @xmath66 .
then @xmath58 can be transformed into @xmath67 . since @xmath68
, we can apply an induction argument to transform the word @xmath58 into the word @xmath69 .
now , by the unicity of the normal decomposition , @xmath70 as to belong to @xmath71 .
therefore we can transform @xmath69 into @xmath72 using relations ( tym2 ) .
note that the letters @xmath72 and @xmath73 has to be equal as they represent the same element in @xmath6 .
assume finally that @xmath52 does not belong to @xmath57 .
by similar arguments we can , applying relations ( cox2 ) and ( tym1 ) , transform the word @xmath54 into a word @xmath74 where @xmath75 in @xmath9 and @xmath76 with @xmath77 in @xmath78 and @xmath79 in @xmath80 . at this stage
we are in position to apply relation ( rbi ) .
thus , we can transform the word @xmath54 into @xmath81 . since we have @xmath82 we can proceed as in the first case to conclude .
here we focus on some special renner monoids considered in @xcite . in each case
, we provide an explicit monoid presentation using the general presentation obtained in section 1 .
consider @xmath83 and choose @xmath84 for borel subgroup ( see example [ exemplecrossselat ] ) . in this case , the weyl group is the symmetric group @xmath1 .
its generating set @xmath18 is @xmath85 where @xmath86 is the transposition matrix corresponding to @xmath87
. the cross section section lattice @xmath88 is linear ( we have @xmath89 for every @xmath23 ) . for every @xmath23 one
has @xmath90 and @xmath91 .
in particular , @xmath92 , and for @xmath93 we have @xmath94 .
( 250,75 ) ( 18,0 ) ( 148,15)@xmath95 ( 125,15)@xmath96(40,15)@xmath97(17,15)@xmath98 ( 225,0)@xmath99 ( 225,20)@xmath44 ( 225,70)@xmath100 therefore , we recover the monoid presentation of the rook monoid @xmath2 stated in @xcite : the generating set is @xmath101 and the defining relations are @xmath102 let @xmath103 be[simplectmonoid ] a positive even integer and @xmath104 be the _
symplectic algebraic group _
@xcite : assume @xmath105 lies in @xmath106 , and consider the matrix @xmath107 in @xmath108 .
let @xmath109 in @xmath110 , where @xmath111 .
then , @xmath104 is equal to @xmath112 , where @xmath113 is the transpose matrix of @xmath114 .
we set @xmath115 .
this monoid is a regular monoid with @xmath116 whose associated reductive algebraic unit group is @xmath117 .
it is called the _ symplectic algebraic monoid _
let @xmath84 be the borel subgroup of @xmath118 as defined in example [ exemplecrossselat ] , and set @xmath119 .
this is a borel subgroup of the unit group of @xmath3 .
it is shown in @xcite that the cross section lattice @xmath6 of @xmath3 is @xmath120 where the elements @xmath121 correspond to the matrices of @xmath110 defined in example [ exemplecrossselat ] .
in particular the cross section lattice @xmath6 is linear . in this case , the weyl group is a coxeter group of type @xmath122 . in other words ,
the group @xmath4 is isomorphic to the subgroup of @xmath123 generated by the permutation matrices @xmath124 corresponding to @xmath125 , @xmath126 , @xmath127 , @xmath128 , and @xmath129 , respectively .
one has @xmath130 and @xmath131 . therefore ,
@xmath132 , and for @xmath133 in @xmath134 one has @xmath92 .
a direct calculation proves that @xmath135 for every @xmath133 , and @xmath136 .
( 250,75 ) ( 18,0 ) ( 148,15)@xmath95 ( 125,15)@xmath96(40,15)@xmath137(30,28)@xmath138(17,15)@xmath139 ( 225,0)@xmath99 ( 225,20)@xmath44 ( 225,47)@xmath140(225,70)@xmath100 hence , a monoid presentation of @xmath2 is given by the generating set @xmath141 and the defining relations@xmath142 let @xmath103 be a positive integer and @xmath143 be defined as in section [ simplectmonoid ] .
the _ special orthogonal group _
@xmath144 is defined as @xmath145 .
the group @xmath146 is a connected reductive group .
following @xcite , we define the _ special orthogonal algebraic monoid _ to be the zariski closure @xmath147 of @xmath148 .
this is an algebraic monoid @xcite , and @xmath149 is a borel subgroup of its unit group . in this case
, the cross section lattice depends on the parity of @xmath103 .
furthermore , the weyl group is a coxeter group whose type depends on the parity of @xmath103 too .
( 250,100 ) ( 18,0 ) ( 148,15)@xmath95 ( 125,15)@xmath96(40,15)@xmath150(17,15)@xmath137(40,52)@xmath151 ( 229,0)@xmath99 ( 229,20)@xmath44 ( 223,53)@xmath152 ( 210,73)@xmath153 ( 270,73)@xmath140 ( 227,100)@xmath100 assume @xmath111 is even . in this case , @xmath4 is a coxeter group of type @xmath154 .
the standard generating set of @xmath4 is @xmath155 where for @xmath156 , the element @xmath86 is the permutation matrix associated with @xmath157 , and @xmath151 is the permutation matrix associated with @xmath158 .
it is shown in @xcite that the cross section @xmath6 is equal to @xmath159 .. the elements @xmath121 correspond to the matrices of @xmath110 defined in example [ exemplecrossselat ] ; the element @xmath153 is the diagonal matrix @xmath160 .
the hasse diagram of @xmath6 is as represented in figure [ fig : hassecrldn ] . for @xmath23 in @xmath161 one
has @xmath90 and @xmath91 .
furthermore , one can verified that @xmath162 @xmath163 @xmath164 @xmath165 therefore , for @xmath133 in @xmath166 one has @xmath92 . furthermore , @xmath167
@xmath168 @xmath169 the monoid @xmath2 has a presentation with @xmath170 for generating set and @xmath171 for defining relations .
assume @xmath172 is odd .
in that case , @xmath4 is a coxeter group of type @xmath122 .
it is shown in @xcite that the cross section lattice is linear as in the case of the symplectic algebraic monoid .
it turns out that the renner monoid of @xmath173 is isomorphic to the renner monoid of symplectic algebraic monoid @xmath174 , and that we obtain the same presentation than in the latter case .
let @xmath175 be a simple algebraic group , and denote by @xmath176 its lie algebra .
let @xmath3 be the algebraic monoid @xmath177 in @xmath178 .
the cross section lattice of @xmath3 and the type map of @xmath3 has been calculated for each dynkin diagram ( see ( * ? ? ?
therefore one can deduce a monoid presentation for each of the associated renner monoid .
in @xcite , renner and putcha consider among regular irreducible algebraic monoids those who are _
@xmath179-irreducible _ , that is those whose cross section lattices have a unique minimal non - zero element .
it is easy to see that the @xmath179-irreducibility property is related to the existence of irreducible rational representations ( * ? ? ?
renner and putcha determined the cross section lattice of those @xmath179-irreducible that arise from special kind of dominant weigths ( * ? ? ?
2,3 ) . using ( * ? ? ?
* theorem 4.13 ) , one can deduce the associated type maps and therefore a monoid presentation of each corresponding renner monoids .
in this section we extend the length function defined in @xcite to any renner monoid . in all this section , we assume @xmath3 is a regular irreducible algebraic monoid with a zero element .
we denote by @xmath175 the unit group of @xmath3 .
we fix a maximal torus @xmath180 of @xmath175 and we denote by @xmath17 the weyl group @xmath181 of @xmath175 .
we denote by @xmath18 the standard generating set associated with the canonical coxeter structure of the weyl group @xmath4 .
we fix a borel subgroup @xmath182 of @xmath175 that contains @xmath180 , and we denote by @xmath6 the associated cross section lattice contained in @xmath2 . as before we set @xmath183 .
the definition of the length function on @xmath4 and of a reduced word has been recalled in section [ sousectionrappelcgt ] . if @xmath184 is a set , we denote by @xmath185 the set of finite words on @xmath184 .
[ deflenfun ] ( i ) we set @xmath186 for @xmath187 in @xmath18 and @xmath188 for @xmath30 in @xmath6 .
let @xmath189 be in @xmath190 and consider the word @xmath191 .
then , the _ length _ of the word @xmath192 is the integer @xmath193 defined by @xmath194 .
+ ( ii ) the _ length _ of an element @xmath19 which belongs to @xmath2 is the integer @xmath195 defined by @xmath196 [ propproplenght]let @xmath19 belong to @xmath2
. + ( i ) the length function @xmath105 on @xmath2 extends the length function @xmath105 defined on @xmath17 .
+ ( ii ) if @xmath39 then @xmath19 lies in @xmath6 .
+ ( iii ) if @xmath187 lies in @xmath18 then @xmath197 . + ( iv ) if @xmath198 belongs to @xmath2 , then @xmath199 .
point @xmath200 and @xmath201 are clear ( the letters of every representative word of an element in @xmath17 are in @xmath18 ) .
if @xmath19 lie in @xmath2 and @xmath187 lie to @xmath18 , then @xmath202 . since @xmath203
, the inequality @xmath204 holds too .
point @xmath205 follows , and point @xmath206 is a direct consequence of @xmath205 .
[ propproplenght2 ] let @xmath19 belong to @xmath2 .
+ ( i ) if @xmath20 is the normal decomposition of @xmath19 , then @xmath207 .
+ ( ii ) if @xmath208 are two representative words of @xmath19 on @xmath11 such that the equalities @xmath209 hold , then using the defining relations of the presentation of @xmath2 in theorem [ thepreserennmonid ] , we can transform @xmath210 into @xmath211 without increasing the length .
\(i ) let @xmath21 be a representative word @xmath19 on @xmath11 such that @xmath212 .
it is clear that we can repeat the argument of the proof of theorem [ thepreserennmonid ] without using the relation @xmath213
. therefore @xmath214 .
+ ( ii ) is a direct consequence of the proof of @xmath200 .
let @xmath19 lie in @xmath2 and @xmath30 belongs to @xmath9 .
denote by @xmath215 the normal decomposition of @xmath19 .
+ ( i ) one has @xmath216 and @xmath217 .
+ ( ii ) @xmath218 if and only if the normal decomposition of @xmath219 is @xmath220 . furthermore , in this case
, @xmath52 lies in @xmath221 .
\(i ) is a direct consequence of the definition of the length and of proposition [ propproplenght2](i ) : @xmath222 .
the same arguments prove that @xmath217 .
+ ( ii ) decompose @xmath52 as a product @xmath223 where @xmath224 lies in @xmath225 , @xmath63 lies in @xmath226 , @xmath227 lies in @xmath228 and @xmath229 .
then , @xmath230 .
in particular , @xmath231 .
assume @xmath218 .
we must have @xmath232 .
the element @xmath63 , that is @xmath52 , must belong to @xmath233 , and the element @xmath51 must belong to @xmath234 . in particular , @xmath52 lies in @xmath221 .
furthermore , @xmath52 lies in @xmath235 since @xmath236 by proposition [ propopo](i ) .
conversely , if the the normal decomposition of @xmath219 is @xmath220 , then @xmath237 .
in proposition [ propgeominterpr ] below we provide a geometrical formula for the length function @xmath105 defined in the previous section .
this formula extends naturally the geometrical definition of the length function on a coxeter group .
another length function on renner monoids has already been defined and investigated @xcite .
this length function has nice properties , which are similar to the ones in propositions [ propproplenght ] , [ propproplenght2](i ) and [ propgeominterpr ] .
this alternative length function has been firstly introduced by solomon @xcite in the special case of rook monoids in order to verify a combinatorial formula that generalizes rodrigues formula @xcite .
that is why we call this length function the _ solomon length function _ in the sequel .
we proved in @xcite that our length function for the rook monoid verifies the same combinatorial formula .
we also proved in @xcite that in the case of the rook monoid , our presentation of @xmath2 and our length function are related to the hecke algebra .
we believe this is still true in the general case .
[ lemtechimpo ] let @xmath19 belong to @xmath2 and denote by @xmath20 its normal decomposition .
let @xmath187 be in @xmath18 .
+ ( i ) we have one of the two following cases : + ( a ) there exists @xmath238 in @xmath239 such that @xmath240 . in this case
, one has @xmath241 ; + ( b ) the element @xmath242 lies in @xmath243 and @xmath244 is the normal decomposition of @xmath245 .
+ ( ii ) denote by @xmath246 the solomon length function on @xmath2 .
then , @xmath247 @xmath200 if @xmath248 lies in @xmath243 , then by theorem [ fnrenner ] , the triple @xmath244 is the normal decomposition of @xmath245 . assume now that @xmath248 does not belong to @xmath243 . in that case , @xmath30 can not be equal to @xmath40 .
since @xmath51 belongs to @xmath243 , by the exchange lemma , there exists @xmath238 in @xmath249 such that @xmath240 .
therefore , @xmath250 .
+ @xmath201 the solomon length @xmath251 of an element @xmath19 in @xmath2 can be defined by the formula @xmath252 where @xmath20 is the normal decomposition of @xmath19 and @xmath253 is a constant that depends on @xmath30 only ( * ? ? ?
* definition 4.1 ) .
therefore the result is a direct consequence of @xmath200 . | we extend the result obtained in @xcite to every renner monoid : we provide renner monoids with a monoid presentation and we introduce a length function which extends the coxeter length function and which behaves nicely . |
let @xmath10 be a compact smooth boundary - less riemannian manifold of dimension @xmath11 . denote by @xmath12 the laplace operator associated to the metric @xmath13 . in the case
@xmath14 , we discovered in @xcite a bilinear generalization of the well - known sogge estimates @xcite for @xmath15 ( @xmath16 ) norms of @xmath17 normalized eigenfunctions of @xmath12 .
these bilinear estimates play a central role in the analysis of @xcite concerning the nonlinear schrdinger equation ( nls ) posed on @xmath0 .
the goal of this paper is to generalize our bilinear estimate of @xcite to all higher dimensions and to deduce new results regarding the global existence of solutions for nls when @xmath18 .
we consider thus the cauchy problem for nls @xmath19 in ( [ intr-1 ] ) , @xmath20 is a complex valued function on @xmath0 .
the nonlinear interaction @xmath21 satisfies @xmath22 and is supposed of the form @xmath23 with @xmath24 satisfying @xmath25 and , for some @xmath26 , @xmath27 the number @xmath28 involved in the second condition on @xmath29 corresponds to the `` degree '' of the nonlinearity @xmath30 in ( [ intr-1 ] ) . under these assumptions on @xmath21 ,
nls can be seen as a hamiltonian equation in an infinite dimensional phase space .
it follows from that hamiltonian structure that smooth solutions of ( [ intr-1 ] ) enjoy the conservation laws @xmath31 where the energy functional @xmath32 reads as follows , @xmath33 in view of ( [ intr-2 ] ) and ( [ intr-3 ] ) , the local well - posedness of ( [ intr-1 ] ) in @xmath34 ( with time existence depending upon the @xmath1 norm ) is of particular importance . if for example @xmath35 and @xmath36 , ( [ intr-2 ] ) provides an @xmath1 a priori bound and thus the local well - posedness of ( [ intr-1 ] ) in @xmath1 implies the global well - posedness in @xmath1 .
let us notice , on the other hand , that the local well - posedness of ( [ intr-1 ] ) in @xmath37 , @xmath38 can be obtained by the classical energy method ( see @xcite ) .
if @xmath0 is two dimensional , the well - posedness of ( [ intr-1 ] ) in @xmath34 is established in @xcite . in this case , the issue is to get an improvement of @xmath39 derivatives with respect to the energy method . in @xcite , this @xmath39 gain
is achieved by a strichartz type inequality ( with derivative loss ) .
therefore , for @xmath14 , the @xmath1 well - posedness theory for ( [ intr-1 ] ) is completed .
moreover , in the recent paper @xcite , we establish a sharp @xmath37 theory in the case @xmath40 , as far as cubic nonlinearities are concerned . in three dimensions ,
the @xmath1 theory for ( [ intr-1 ] ) becomes much harder . in the case
@xmath18 , the strichartz type inequalities established in @xcite yield the local well - posedness of ( [ intr-1 ] ) in @xmath37 , @xmath41 , as far as @xmath42 .
notice that this is already a significant improvement with respect to the energy approach .
unfortunately , it barely misses the crucial @xmath1 regularity .
however , in @xcite , we succeeded in using the conservation laws ( [ intr-2 ] ) in order to get _ global _
@xmath37 , @xmath41 strong solutions . by `` strong @xmath37 solutions '' ,
we mean the existence , the uniqueness , the propagation of regularity and the uniform continuous dependence in bounded subsets of initial data in @xmath37 . moreover , the methods of @xcite also yield uniqueness of @xmath1 weak solutions . on the other hand ,
if @xmath0 is the torus @xmath43 and @xmath44 , the global existence of @xmath1 strong solutions of ( [ intr-1 ] ) was established by bourgain @xcite .
the approach in @xcite is based on an ingenious use of multiple fourier series and thus relies deeply on the particular structure of the torus . in this paper
, we will prove the counterpart of this result of bourgain to the cases of the sphere @xmath45 and the product manifold @xmath46 , where @xmath47 , @xmath48 is the embedded sphere of radius @xmath49 in @xmath50 .
[ thm1 ] let @xmath2 or @xmath51 endowed with the standard metrics .
suppose that @xmath44 and @xmath52 , @xmath53 .
then there exists a space @xmath54 continuously embedded in @xmath55 such that for every @xmath56 there exists a unique global solution @xmath57 of the cauchy problem ( [ intr-1 ] ) . moreover 1 . for every @xmath58 , the map @xmath59\,;\
, h^{1}(m))$ ] is lipschitz continuous on bounded sets of @xmath34 .
2 . if @xmath60 , @xmath61 , then for every @xmath62 , @xmath63 . let us make some comments about this result .
the condition @xmath64 is classically ( see e.g. cazenave @xcite ) imposed to ensure that the energy controls the @xmath34 norm ( defocusing case ) .
the space @xmath54 will be defined in section [ sec.3 ] as a local version of bourgain space @xmath65 .
it is used to ensure the uniqueness of solutions .
however , observe that if @xmath66 , then the uniqueness holds in the class @xmath67 . in particular , our theorem implies that for any smooth data @xmath68 , there exists a unique global smooth solution . in the appendix of this paper
, we show that theorem [ thm1 ] can not hold for @xmath69 . the proof is based on an adaptation of an argument of a recent paper of christ - colliander - tao @xcite to the setting of compact riemannian manifolds .
the critical case @xmath70 is still open .
let us recall that the result of theorem [ thm1 ] was known if we replace @xmath0 with the euclidean space @xmath4 ( see ginibre - velo @xcite and kato @xcite ) . to get the @xmath71 well - posedness of ( [ intr-1 ] ) , for @xmath44
, it is sufficient to apply the picard iteration scheme to the duhamel formulation of f ( [ intr-1 ] ) in the space @xmath72 where @xmath73 depends only on @xmath74 .
the approach on @xmath4 breaks down in the case of a compact manifold since the corresponding strichartz type estimates have to encounter some unavoidable derivative losses ( see @xcite ) . in order to deal with such losses ,
bilinear improvements of the strichartz inequalities are very useful ( see e.g. @xcite ) .
this is the approach that we will adopt in the proof of theorem [ thm1 ] when @xmath2 .
the proof in the case @xmath75 is more intricate .
the bilinear strichartz estimates that we are able to prove in the case @xmath75 are considerably weaker compared to the corresponding estimates for @xmath2 .
however , they are sufficient to treat the case @xmath76 .
the crucial new point involved in the analysis on @xmath77 is that we can prove a trilinear improvement of the strichartz estimate which enables one to treat the case @xmath78 for data in @xmath79 , @xmath41 .
a suitable interpolation ( in the framework of a littlewood - paley analysis ) between the bilinear and the trilinear approach finally completes the argument in the case @xmath75 .
the results of ginibre - velo @xcite on @xmath80 , of bourgain @xcite on @xmath81 ( and more recently on the irrational three dimensional torus @xcite ) , and theorem [ thm1 ] were obtained for seemingly different reasons in each case .
for the torus the eigenfunctions enjoy very good algebraic properties and @xmath15 bounds whereas the spectrum is `` badly '' localized . on the other hand for the sphere @xmath45 , the eigenfunctions present `` bad '' concentration properties but the spectrum is very well localized , and the manifold @xmath82 has an intermediate behavior .
the balance between these properties ( concentration of eigenfunctions and repartition of the spectrum ) leads to the suggestion that a similar result might hold for _ any _ three dimensional manifold
. the proof of this conjecture would necessitate a general analysis of the schrdinger group , unifying these different approaches , which seems to be out of reach at the present moment .
the @xmath1 theory for ( [ intr-1 ] ) in dimensions @xmath83 remains an open problem .
the only known result in this direction is that of bourgain @xcite who gets _ global _
@xmath84 solutions , if @xmath85 , @xmath41 .
it seems that the obstructions to extending our approach to high dimensions are not only of technical nature since in @xcite we have shown that for no @xmath861,2]$ ] ( even very close to @xmath87 ) , the cauchy problem ( [ intr-1 ] ) , posed on @xmath88 can have strong @xmath1 solutions in the sense explained above .
interestingly , the result of @xcite is in strong contrast with the situation on @xmath89 ( see @xcite ) .
we now turn to the crucial step in the proof of theorem [ thm1 ] . to that purpose
, we introduce the following notation : given @xmath90 , we set @xmath91 with this notation , we have the following multilinear eigenfunction estimates .
[ thm2 ] there exists @xmath92 such that , if @xmath93 and @xmath94 are two spherical harmonics of respective degrees @xmath95 and @xmath96 , @xmath97 moreover for any @xmath98 , the following trilinear estimates hold @xmath99^{\frac{1}{4 } } \|h_p\|_{l^{2}(s^2)}\|h_q\|_{l^{2}(s^2)}\|h_r\|_{l^{2}(s^2)}\ , .\ ] ] estimates ( [ thm2 - 1 ] ) and ( [ thm2 - 2 ] ) are sharp , apart from the logarithmic loss in ( [ thm2 - 1 ] ) for @xmath18 . as an easy consequence of ( [ thm2 - 1 ] ) , one can prove the corresponding estimate to ( [ thm2 - 2 ] ) , for @xmath100 , @xmath101 indeed it suffices to use that the @xmath102 norm of @xmath103 is bounded by @xmath104 ( weyl bound ) and ( [ thm2 - 1 ] ) for the product @xmath105 .
in view of further possible developments , we will also prove in section [ sec.2 ] that for every @xmath1060,1]$ ] there exists @xmath107 such that @xmath108 in fact , we deduce theorem [ thm2 ] as a consequence of a more general statement concerning the approximated spectral projectors @xmath109 , @xmath110 , @xmath111 , where @xmath12 is the laplace operator on an arbitrary compact riemannian manifold @xmath10 ( see theorem [ thm3 ] below ) .
notice that when @xmath112 , apart from the @xmath113 loss in @xmath114 , we recover some particular case of the @xmath115 _ linear _ estimates of sogge @xcite . in the proof of theorem [ thm1
] , we typically apply theorem [ thm2 ] for @xmath116 and thus estimates ( [ thm2 - 1 ] ) , ( [ thm2 - 2 ] ) are used in their full strength . in the case
@xmath14 , estimate ( [ thm2 - 1 ] ) has already appeared in our previous paper @xcite . in @xcite , the proof is inspired by hrmander s work @xcite on carleson - sjlin type operators .
the proof we present here is different even for @xmath14 and relies on a `` bilinearization '' of the arguments in @xcite .
after several preliminaries , we reduce the matters to two micro - local _ linear _ estimates of quite a different nature .
the first one is applied to the higher frequency eigenfunction and is in the spirit of the @xmath17 boundedness of spectral projectors .
the second one is applied to the smaller frequency eigenfunctions and relies on a dispersive ( curvature ) effect .
as far as the optimality of ( [ thm2 - 1 ] ) , ( [ thm2 - 2 ] ) is concerned , we notice that it is achieved either by testing the estimates against eigenfunctions concentrating on an equator or by testing against zonal eigenfunctions concentrating on a point .
let us mention that estimates ( [ thm2 - 1 ] ) , ( [ thm2 - 2 ] ) and a sketch of the proof of ( [ thm2 - 1 ] ) appeared in @xcite . the rest of this paper is organized as follows . in section [ sec.2 ]
we prove theorem [ thm2 ] . in section [ sec.3 ]
we set up the framework of bourgain s spaces and reduce the proof of theorem [ thm1 ] to the obtaining of nonlinear estimates in this framework .
section [ sec.4 ] consists in two parts .
first we prove bilinear strichartz estimates for the linear schrdinger group on @xmath45 .
then we show that theorem [ thm1 ] holds for any three dimensional manifold on which these estimates are true .
section [ sec.5 ] also consists in two parts .
first we prove trilinear strichartz estimates for the linear schrdinger group on the product manifold @xmath117 and then we show that theorem [ thm1 ] holds for any three dimensional manifold on which these estimates are true .
an appendix is devoted to the proof of the optimality of the quintic threshold .
* acknowledgements .
* we are grateful to j. bourgain for sending us his manuscript @xcite and h. koch and d. tataru for interesting discussions about spectral projectors .
in this section we prove theorem [ thm2 ] , and more generally the corresponding result for spectral projectors on arbitrary compact manifolds .
we first consider the optimality of ( [ thm2 - 1 ] ) in the case @xmath118 .
let us see @xmath119 as a hyper - surface in @xmath50 , i.e. @xmath120 let us define the highest weight spherical harmonics @xmath121 which concentrate , for @xmath122 , on the closed geodesic ( a big circle ) @xmath123 .
an easy computation shows that @xmath124 clearly @xmath125 and therefore there exist constants @xmath126 , @xmath127 such that for every @xmath128 , @xmath129 therefore , for @xmath118 , estimate ( [ thm2 - 1 ] ) turns out to be optimal , modulo the logarithmic loss in @xmath114 . in the same way , since @xmath130 , estimate ( [ thm2 - 2 ] )
is optimal by testing it on @xmath131 , @xmath132 and @xmath133 .
let us now consider the case @xmath83 . in this case the optimality of ( [ thm2 - 1 ] ) is given by the zonal spherical harmonics .
let us a fix a pole on @xmath119 .
if we consider functions on @xmath119 depending only on the geodesic distance to the fixed pole , we obtain the zonal functions on @xmath119 .
the zonal functions can be expressed in terms of zonal spherical harmonics which in their turn can be expressed in terms of the classical jacobi polynomials ( see e.g. @xcite ) . using asymptotics for the jacobi polynomials ( see @xcite , ( * ? ? ? * lemma 2.1 ) ) we can obtain the following representation for the zonal spherical harmonics @xmath134 of degree @xmath95 , in the coordinate @xmath135 , @xmath136+\frac{{\cal o}(1)}{p\ , \sin\theta } \big\},\quad \frac{c}{p}\leq \theta\leq\pi-\frac{c}{p}\ , , \ ] ] where @xmath28 and @xmath137 are some fixed constants depending only on @xmath138 .
moreover , we have a point - wise concentration @xmath139 .
let @xmath140 . then @xmath141 using ( [ point ] ) , we get @xmath142 in view of ( [ szego ] ) , @xmath143,\ ] ] where @xmath144d \theta\geq \frac{c}{p } \quad { \rm and } \quad i_{2}=\frac{1}{q^2}\int_{c / q}^{c / p}\frac{1}{(\sin\theta)^2}d\theta\leq \frac{c}{q}\ll \frac{c}{p}\ , .\ ] ] therefore @xmath145 if @xmath146 .
let finally @xmath147 . using ( [ point ] ) , we get @xmath148 therefore , collecting ( [ lll1 ] ) and ( [ lll2 ] )
, we obtain @xmath149 which proves the optimality of ( [ thm2 - 1 ] ) , for @xmath100 , modulo the logarithmic loss in @xmath114 .
let us finally notice that similarly we can prove that for @xmath150 @xmath151 which proves the optimality of ( [ thm2 - 2-novo ] ) , for @xmath100 , apart from the logarithmic loss in @xmath114 , and the optimality of ( [ thm2 - 2-novo - bis ] ) apart from the @xmath152 shift .
let @xmath10 be a compact smooth riemannian manifold without boundary of dimension @xmath138 and @xmath12 be the laplace operator on functions on @xmath0 .
it turns out that estimates ( [ thm2 - 1 ] ) , ( [ thm2 - 2 ] ) and ( [ thm2 - 2-novo - bis ] ) can be deduced from the following more general result .
[ thm3 ] let @xmath111 . for @xmath153 ,
denote by @xmath154 the approximated spectral projector around @xmath155 .
there exists @xmath126 such that for any @xmath156 , @xmath157 , @xmath158 moreover , in the case @xmath14 , for any @xmath159 , @xmath160 , the following trilinear estimate holds @xmath161 finally , in the case @xmath18 , for any @xmath162 , @xmath160 , @xmath1060,1]$ ] , the following trilinear estimate holds @xmath163 if one is only interested in estimates for single eigenfunctions , the bounds provided by theorem [ thm3 ] seem to be relevant for `` sphere like manifolds '' but they are far from the optimal ones in the case of the torus .
for example , the classical result of zygmund @xcite says that there exists a constant @xmath126 such that for every couple @xmath164 of eigenfunctions of the laplace operator on the torus @xmath165 , one has @xmath166 we refer to bourgain @xcite for further extensions of zygmund s result .
a first reduction in the proof of theorem [ thm3 ] is that it suffices to prove it for one fixed non trivial function @xmath167 .
[ chi ] suppose that the assertion of theorem [ thm3 ] holds for a bump function @xmath168 which is not identically zero .
then it holds for any other choice of the bump function .
suppose that ( [ eqbilin ] ) holds for a nontrivial @xmath168 .
then , there exists @xmath169 such that @xmath170 and moreover there exists @xmath171 such that @xmath172 for @xmath173 satisfying @xmath174 . using a partition of unity argument
, we can find @xmath175 supported in @xmath176 such that @xmath177 thanks to the support properties of @xmath178 and @xmath167 , we can write @xmath179 notice that the second factor in the right hand - side of ( [ div ] ) is uniformly bounded . therefore , using that ( [ eqbilin ] ) holds for @xmath167 , we obtain the estimate @xmath180 let us now take an arbitrary function @xmath181 . using ( [ pu ] ) ,
we can write @xmath182 let @xmath183 be equal to one on the support of @xmath178 . then clearly @xmath184 using the expansion ( [ eee ] ) together with ( [ loc - psi ] ) and ( [ ddd ] ) yields @xmath185 hence ( [ eqbilin ] ) holds for @xmath186 .
the proof of the independence of ( [ eqtrilin ] ) and ( [ eqtrilin-3 ] ) with respect to the bump function @xmath167 is very similar and thus we will omit it . following ( * ? ? ? * chap .
4 ) , thanks to lemma [ chi ] , it is sufficient to prove theorem [ thm3 ] with @xmath167 such that @xmath187 is supported in the set @xmath188 where @xmath189 is a _ small _ number to be determined later .
we can write @xmath190 for @xmath191 and @xmath192 , using a partition of the unity on @xmath0 , we can represent @xmath193 as a fourier integral operator ( see e.g. @xcite ) .
therefore @xmath194 can also be represented as such . after a stationary phase argument ( see ( * ? ? ?
5 ) ) we can represent @xmath195 as follows .
[ sogge ] there exists @xmath196 such that for every @xmath1970,\varepsilon_{0}[$ ] , every @xmath198 , we have the splitting @xmath199 with @xmath200 moreover there exist @xmath171 and , for every @xmath201 , a system of coordinates @xmath202 , containing @xmath203 such for @xmath204 , @xmath205 , @xmath206 where @xmath207 is a polynomial in @xmath208 with smooth coefficients supported in the set @xmath209 and @xmath210 is the geodesic distance between @xmath211 and @xmath212 .
let us notice that one can use @xmath213 as approximated spectral projector instead of @xmath109 . in that case one should use semi - classical calculus for the approximation of @xmath214 , @xmath110 , as we did in @xcite . in view of lemma [ sogge ] , to prove ( [ eqbilin ] )
, it is enough to show @xmath215 uniformly for @xmath216 . indeed , using ( [ reste ] ) , one has to evaluate in @xmath17 the products @xmath217 the products involving @xmath218 are straightforward to estimate while for @xmath219 , using the @xmath17 boundedness of @xmath220 , we write @xmath221 furthermore , we notice that once is proved ( at least for @xmath14 ) , to prove ( [ eqtrilin ] ) it is enough to show that for @xmath14 , @xmath222 uniformly for @xmath223 . in this case there are more remainder terms to estimate .
the most difficult one is @xmath224 .
this term can be evaluated , by using ( [ eqbilin - pak ] ) for @xmath14 , as follows @xmath225 similarly , to prove ( [ eqtrilin-3 ] ) , it is enough to show that for @xmath18 , @xmath226 uniformly for @xmath223 . in this case
, we estimate @xmath224 , by another use of ( [ eqbilin - pak ] ) , as follows @xmath227 where @xmath228 .
.5 cm next , we represent @xmath212 in geodesic ( polar ) coordinates as @xmath229 , @xmath230 , @xmath231 . for @xmath205 and @xmath232
, we define the frozen phase @xmath233 , @xmath234 we now state the main property of the phase @xmath235 .
[ pak ] there exists @xmath189 such that for every @xmath236 $ ] , every @xmath237 we have the identity , @xmath238 the proof for @xmath14 is given in @xcite .
the extension to an arbitrary @xmath138 is straightforward as we explain below .
for @xmath191 , let @xmath239 , @xmath240 and @xmath241 be the unique unit vector in the tangent space to @xmath0 at @xmath212 such that @xmath242 differentiating with respect to @xmath211 this identity , we get for @xmath243 , and any @xmath244 , @xmath245 where @xmath73 denotes the tangential map . on the other hand , we have @xmath246 consequently ,
using gauss lemma ( see @xcite ) , we get @xmath247 let us now take the scalar product of ( [ geom ] ) with @xmath248 .
collecting ( [ geom ] ) , ( [ geom3 ] ) and yield @xmath249 which completes the proof of lemma [ pak ] .
let us notice that there exists a smooth positive function @xmath250 such that @xmath251 . for @xmath252 $ ] and @xmath253 , we define the operator @xmath254 , acting on functions on @xmath255 via the identity @xmath256 where @xmath257 . then clearly @xmath258 where @xmath259 .
similarly , with @xmath260 , @xmath261 and the minkowski inequality shows that ( [ eqbilin ] ) will be a consequence of @xmath262 uniformly for @xmath216 and @xmath263 $ ] . similarly , to prove ( [ eqtrilin ] ) , it is enough to show @xmath264 uniformly for @xmath223 and @xmath265 $ ] . finally , to prove ( [ eqtrilin-3 ] )
, it is enough to show @xmath266 uniformly for @xmath223 and @xmath265 $ ] .
fix a point @xmath267 .
the set @xmath268 is a smooth hyper - surface in @xmath269 .
indeed assuming for instance @xmath270 , then @xmath271 is a system of coordinates on @xmath255 and according to lemma [ pak ] , @xmath272 has rank @xmath273 .
following stein @xcite and sogge @xcite , we now state the crucial curvature property .
[ lem.courb ] the hyper - surface @xmath274 has non - vanishing principal curvatures : for @xmath275 a local coordinate system near @xmath276 , if we denote by @xmath277 the normal unit vectors to the surface @xmath274 at the point @xmath278 , then for @xmath211 close to @xmath279 , @xmath280 the relation ( [ eq1.2 ] ) is equivalent to the fact that @xmath281 is a local diffeomorphism .
indeed , dropping the @xmath211 variable for conciseness and denoting by @xmath282 we have @xmath283 as a consequence , the determinant in is non vanishing if and only if the system of vectors @xmath284 is of maximal rank in @xmath285 . we deduce that is independent of the choice of coordinates @xmath286 and it suffices to prove it for a particular choice of a coordinate system near @xmath287 .
we can suppose that @xmath288 and we choose as coordinates @xmath289 we can also assume that at the point @xmath290 , the metric is diagonal , @xmath291 .
using lemma [ pak ] , we get @xmath292 and consequently follows by continuity . denote by @xmath293 the formal adjoint of @xmath294 .
the kernel of the operator @xmath295 , @xmath296 , is given by the relation @xmath297 the curvature property of the phase @xmath235 in lemma [ lem.courb ] implies a dispersion inequality for the kernel @xmath298 .
[ lem.dispersion ] there exist @xmath92 such that for any @xmath299 , @xmath300 where @xmath301 with @xmath302 , we can write @xmath303 where @xmath304 now we want to prove , with @xmath305 , @xmath306 where @xmath307 from the definition of the normal @xmath308 , we have @xmath309 for @xmath310 , @xmath311 , @xmath312 . according to the curvature property , we have @xmath313 for @xmath310 , @xmath311 , @xmath312 .
from the implicit function theorem , there exist @xmath314 , such that if @xmath315 has a unique non - degenerate critical point @xmath316 and , by stationary phase , under the assumption ( [ under ] ) , the kernel ( [ eq.kernel ] ) is bounded by @xmath317 let us next assume that latexmath:[\[\label{rakia } then for @xmath286 close to @xmath279 and @xmath319 small enough , we obtain by continuity @xmath320 is one dimensional and spanned by @xmath308 .
coming back to the definition of @xmath321 , we deduce that implies ( for @xmath322 small enough ) @xmath323 consequently , integrating by parts in , we obtain that under the assumption ( [ rakia ] ) the kernel ( [ eq.kernel ] ) is bounded by @xmath324 which is even better than needed .
this completes the proof of lemma [ lem.dispersion ] .
the second property of the phase we need is the following : [ curvature ] let @xmath325 where @xmath326 and @xmath327
. then for every @xmath328 with @xmath329 there exist a neighborhood @xmath330 of @xmath287 , @xmath189 and @xmath171 such that , for @xmath331 and @xmath332 , the phase @xmath333 , where @xmath275 is a local coordinate in @xmath334 , is uniformly non degenerate with respect to @xmath335 .
more precisely @xmath336 since ( [ eq1.0 ] ) is independent of the choice of coordinates @xmath286 , it suffices to prove it for a particular choice of a coordinate system near @xmath287 . for @xmath337 in a small neighborhood of @xmath287 , we choose @xmath286 as @xmath289 which is a coordinate system thanks to the assumption @xmath329 .
we can also assume that at the point @xmath338 , the metric is diagonal @xmath291 .
using lemma [ pak ] , we get @xmath339 we now obtain ( [ eq1.0 ] ) from ( [ eq1.0-bis ] ) by continuity .
we next state a corollary of lemma [ curvature ] .
[ curvature - bis ] let @xmath340 be @xmath341 points on @xmath255 .
then there exists a splitting of the variable @xmath325 and neighborhoods @xmath342 , @xmath343 of @xmath344 such that @xmath333 satisfies ( [ eq1.0 ] ) , where @xmath286 is a coordinate in @xmath345 . obviously , there exists a unit vector @xmath346 such that @xmath347 by performing a rotation , we can assume that @xmath348 and consequently it suffices to apply lemma [ curvature ] .
the dispersion inequality of lemma [ lem.dispersion ] leads to the following estimate .
[ linfty ] let @xmath349 be any local system of coordinate near @xmath350 .
then the operator @xmath351 is continuous with norm bounded by @xmath352 .
recall that @xmath353 let consider the formal adjoint of @xmath354 defined as @xmath355 according to the classical duality argument which reduces the study of @xmath354 to the study of @xmath356 , it is sufficient to show that the norm of the operator @xmath357 is bounded by @xmath358^{2}$ ] . but
according to lemma [ lem.dispersion ] , the kernel of this operator satisfies and as a consequence , there exists @xmath92 such that for every @xmath90 , @xmath359 ) and the young inequality , we get @xmath360 but clearly @xmath361 it remains to observe that the right hand - side of the above inequality is equal to @xmath358^{2}$ ] which completes the proof of lemma [ linfty ] . in two space dimensions , we shall need the following extension of lemma [ linfty ] .
[ linfty - bis ] let @xmath14 and @xmath349 be any local system of coordinate near @xmath350 .
the operator @xmath362 is continuous with norm bounded by @xmath363 .
similarly to the proof of lemma [ linfty ] , it is sufficient to show that the norm of the operator @xmath364 is bounded by @xmath365 .
the kernel @xmath366 of @xmath367 satisfies the bound ( [ eq.dispfble ] ) with @xmath14 . from the hardy - littlewood inequality , @xmath368 therefore @xmath369 which completes the proof of lemma [ linfty - bis ] . in the proof of ( [ tri ] ) , we need the following extension of lemma [ linfty ] for @xmath100 .
[ linfty - tris ] let @xmath100 , @xmath370 and @xmath349 be any local system of coordinate near @xmath350 .
the operator @xmath371 is continuous with norm bounded by @xmath372 .
let @xmath373 be such that @xmath374 .
it is sufficient to show that the norm of the operator @xmath375 is bounded by @xmath376 . since for @xmath370 , @xmath377 using the young inequality , we get the bound @xmath378 which completes the proof of lemma [ linfty - tris ] thanks to the bound ( [ eq.dispfble ] ) on the kernel of @xmath367 . notice that for @xmath379 , the proof above still works in dimensions @xmath83 . in the case
@xmath379 , @xmath18 , we have the same difficulty as in the case of the end point strichartz estimates on @xmath380 ( see remark [ kochtat ] below ) .
a consequence of lemma [ curvature ] is the following statement .
[ l2 ] under the assumptions of lemma [ curvature ] , the operator @xmath381 is continuous with norm bounded by @xmath382 . in view of ( [ eq1.0 ] ) , the statement of lemma [ l2 ] , which can be understood as a refinement of the @xmath17 boundedness of the spectral projector , is an immediate consequence of the following generalization of plancherel s identity , which we borrow from @xcite .
let us consider @xmath383 and @xmath384 such that @xmath385\neq 0.\end{aligned}\ ] ] there exists @xmath92 such that for every @xmath253 , the operator @xmath386 @xmath387 satisfies , @xmath388 let us first prove ( [ edno ] ) .
we can write @xmath389 we need to evaluate the above expression in @xmath390 . after a partition of unity
, we can suppose that on the support of @xmath391 @xmath392 is close to a fixed point @xmath393 .
we can therefore use the splitting @xmath394 of lemma [ curvature - bis ] with @xmath395 . using hlder s inequality , lemma [ l2 ] and lemma [ linfty ] ,
we infer @xmath396 this completes the proof of ( [ edno ] ) .
we next prove ( [ dve ] ) .
let us write @xmath397 after a partition of unity we can assume that @xmath398 is close to a fixed point @xmath399 .
with the splitting @xmath394 of lemma [ curvature - bis ] with @xmath400 , using lemma [ l2 ] , lemma [ linfty - bis ] and the hlder inequality , we get @xmath401 this completes the proof of ( [ dve ] ) .
we finally prove ( [ tri ] ) .
we can again use the splitting @xmath394 of lemma [ curvature - bis ] with @xmath400 . for @xmath402 and @xmath403
such that @xmath404 , a use of lemma [ linfty - tris ] gives the bound @xmath405 where @xmath406 .
this ends the proof ( [ tri ] ) and completes the proof of theorem [ thm2 ] .
[ kochtat ] as pointed to us by koch and tataru @xcite , another approach to these multilinear estimates would be , after a suitable micro - localization , to particularize one variable ( @xmath407 in the exposition above ) and see the equation satisfied by the approximated spectral projector @xmath408 as a _ semi - classical _ evolution equation of the type @xmath409 then lemmas [ linfty ] , [ linfty - bis ] are simply the ( semi - classical ) strichartz estimates which can be proved by using the approach in @xcite .
in this section @xmath10 is an arbitrary riemannian manifold of dimension @xmath138 .
our first purpose is to introduce the basic localization operators @xmath410 and @xmath411 which are naturally related to the sobolev spaces and the bourgain spaces on @xmath0 respectively .
we establish some basic bounds related to @xmath410 and @xmath411 .
the main purpose of this section is to show that the well - posedness of the cauchy problem ( [ intr-1 ] ) in the sobolev space @xmath412 is a consequence of nonlinear estimates in the bourgain spaces associated to the laplace operator @xmath12 .
this reduction is now classical ( see e.g. @xcite ) .
since @xmath0 is compact , @xmath12 has a compact resolvent and thus the spectrum of @xmath12 is discrete .
let @xmath413 , @xmath414 be an orthonormal basis of eigenfunctions of @xmath12 associated to eigenvalues @xmath415 .
denote by @xmath416 the orthogonal projector on @xmath417 .
the sobolev space @xmath418 is therefore equipped with the norm ( with @xmath419 ) , @xmath420 the bourgain space ( or conormal sobolev space ) @xmath421 is equipped with the norm @xmath422 where @xmath423 denotes the fourier transform of @xmath424 with respect to the time variable .
let us first recall that for @xmath425 the space @xmath426 is continuously embedded in @xmath427 . for @xmath428 , we write @xmath429 for @xmath430 , we get by the cauchy - schwarz inequality , applied in @xmath431 , @xmath432 squaring ( [ eq2.400 ] ) , integrating over @xmath0 and summing over @xmath433 yields , @xmath434 for @xmath435 and @xmath198 , we define the projector @xmath410 as @xmath436 we now state a basic bound for @xmath410 .
[ sob - g ] there exists a constant @xmath126 such that for every @xmath437 $ ] , every @xmath438 , @xmath439 the assertion clearly holds for @xmath440 .
we next prove it for @xmath441 .
let us write @xmath442 where @xmath443 due to the weyl estimate ( see lemma [ sogge ] ) @xmath444 and due to the triangle and cauchy - schwarz inequalities @xmath445 by hlder s inequality ,
this completes the proof of lemma [ sob - g ] .
for @xmath198 a dyadic integer , i.e. @xmath446 , @xmath447 , we define the operator @xmath448 as @xmath449 where the sum is taken over all dyadic integers @xmath450 smaller or equal to @xmath341 .
we also define @xmath451 by @xmath452 .
next , for @xmath428 and @xmath453 positive integers , we define the localization operators @xmath411 as @xmath454 it is easy to check that @xmath411 is a projector .
it follows from this definition that for every @xmath455 , @xmath456 there exists @xmath92 such that @xmath457 and @xmath458 where the sums is taken over all dyadic values of @xmath341 and @xmath459 , i.e. @xmath446 , @xmath460 , @xmath461 .
we now state a basic bound for @xmath411 .
[ sob - co - g ] there exists a constant @xmath126 such that for every @xmath462 $ ] , every @xmath463 , @xmath464 since @xmath465 , a use of lemma [ sob - g ] yields @xmath466 therefore , we only need to consider the case @xmath440 . using that @xmath467 ,
we can write @xmath468 since the integration over @xmath431 is on a region of size @xmath459 , using the cauchy - schwarz inequality in @xmath431 and the plancherel identity yields @xmath469 applying the last inequality to @xmath470 instead of @xmath20 and using that @xmath411 is a projector gives @xmath471 the assertion of the lemma trivially holds for @xmath472 and therefore the proof of lemma [ sob - co - g ] is completed by hlder s inequality . the starting point is to consider the integral equation ( duhamel form ) @xmath473 at least for classical ( smooth ) solutions @xmath20 the integral equation ( [ duhamel ] ) is equivalent to ( [ intr-1 ] ) .
for that reason we solve ( [ duhamel ] ) by the picard iteration in a suitable functional setting and thus we get solutions of ( [ duhamel ] ) . notice that this is achieved classically if @xmath474 by taking @xmath475 ; h^s(m))$ ] .
therefore we shall restrict the study to the case @xmath476 $ ] . in this case of low regularity solutions of ( [ duhamel ] ) the information we have for @xmath20 and @xmath30
should be strong enough to conclude that we get solutions of ( [ intr-1 ] ) too . for @xmath58
, we define the restriction space @xmath477\times m)$ ] , equipped with the norm @xmath478}=u\}.\ ] ] for @xmath425 , the space @xmath479 is continuously embedded in @xmath480\,;\,h^{s}(m))$ ] ( see ( [ sobolev ] ) ) and @xmath479 will be the space for the solutions of ( [ duhamel ] ) on @xmath481 $ ] .
the next proposition contains the basic reduction to nonlinear estimates in @xmath482 .
[ basic - reduction ] assume that there exists @xmath483 satisfying @xmath484 such that for every @xmath485 there exists a constant @xmath126 and @xmath95 such that for every @xmath486 , @xmath487 and for every @xmath488 , @xmath489 then 1 . [ item-1 ] for every bounded subset @xmath490 of @xmath34 there exists @xmath58 such that if @xmath491 then there exists a unique solution @xmath20 of ( [ duhamel ] ) on @xmath481 $ ] in the class @xmath492 .
moreover the map @xmath59\,;\ , h^{1}(m))$ ] is lipschitz continuous on @xmath490 .
[ item-2 ] if in addition @xmath493 then @xmath494\,;\ , h^{s}(m))$ ] .
the function @xmath20 is a solution of ( [ intr-1 ] ) in the distributional sense .
4 . if in addition we suppose that @xmath52 , @xmath495 then the results above hold for any arbitrarily large @xmath73 .
5 . for @xmath493 ,
@xmath474 the solution is unique in @xmath480\,;\ , h^{s}(m))$ ] .
let @xmath175 be equal to @xmath87 on @xmath496 $ ] .
the identity @xmath497 follows from the definition of @xmath421 and therefore for @xmath498 @xmath499 the bound @xmath500 holds for @xmath498 and @xmath501 satisfying ( [ restrictions ] )
. indeed ( see ( * ? ? ?
* proposition 2.11 ) ) estimate ( [ eq2.17 ] ) follows from the _ one dimensional _
inequality @xmath502 a proof of ( [ basic ] ) can be found in @xcite . using ( [ eq2.17 ] ) and the assumptions of the proposition we obtain the estimates @xmath503 and @xmath504 provided @xmath498 and @xmath501 satisfying ( [ restrictions ] ) .
let @xmath490 be a bounded subset of @xmath34 .
it results from ( [ free ] ) , ( [ tame ] ) and ( [ diff ] ) with @xmath505 that there exists @xmath506 such that for every @xmath507 the right hand - side of ( [ duhamel ] ) is a contraction in a suitable ball of @xmath492 with a unique fixed point which is the solution of ( [ duhamel ] ) .
the uniqueness in the class @xmath492 and the lipschitz continuity of the flow map follow from ( [ diff ] ) .
suppose now that @xmath493 .
then as before it follows from ( [ free ] ) , ( [ tame ] ) and ( [ diff ] ) that we can find @xmath508 such that we can identify @xmath509}$ ] as the unique solution of ( [ duhamel ] ) on @xmath510 $ ] in the class @xmath511 .
in particular @xmath512 for @xmath513 $ ] .
then by a bootstrap and the tame estimate ( [ tame ] ) we conclude that @xmath512 for @xmath514 $ ] .
thanks to ( [ parvo - red ] ) , we obtain that @xmath515 and since @xmath516 , we infer that @xmath517= e^{-it{\mathbf \delta}}f(u(t))\ ] ] in the distributional sense which implies that @xmath20 is a solution of the original pde ( [ intr-1 ] ) in the distributional sense .
if @xmath518 then , thanks to the propagation of the @xmath37 regularity assertion , one can take the scalar product of ( [ intr-1 ] ) with @xmath20 and @xmath519 and it results that the conservations laws ( [ intr-2 ] ) hold . if @xmath520 , we can approximate in @xmath34 the function @xmath68 with a sequence @xmath521 such that @xmath522 .
if we denote by @xmath523 the corresponding solutions of ( [ intr-1 ] ) , thanks to the propagation of the regularity we obtain then @xmath523 enjoy the conservation laws ( [ intr-2 ] ) for @xmath407 on the time of existence of @xmath524 .
finally we can pass to the limit @xmath525 and thanks to the @xmath1 continuity of the conservation laws functionals , we deduce that @xmath524 satisfies the conservation laws ( [ intr-2 ] ) .
if we suppose that @xmath526 satisfies @xmath52 , @xmath495 , using the gagliardo - nirenberg inequalities , we obtain that there exists @xmath5270,2[$ ] such that @xmath528 therefore , the conservation laws ( [ intr-2 ] ) yield a bound independent with respect to @xmath407 for @xmath529 which allows to reiterate the local existence argument and thus to achieve the existence of @xmath524 on an arbitrary time interval .
finally , thanks to the sobolev embedding @xmath530 , @xmath474 and the propagation of regularity , we easily obtain that if @xmath493 , @xmath474 then the uniqueness holds in the class @xmath480\,;\ , h^{s}(m))$ ] .
this completes the proof of proposition [ basic - reduction ] . with proposition [ basic - reduction ] in hand
the assertion of theorem [ thm1 ] follows from the following statement .
[ thm4 ] let @xmath2 or @xmath51 endowed with the standard metrics . for every @xmath531 and @xmath532 there exists @xmath483 satisfying ( [ restrictions ] ) such that ( [ parvo - red ] ) and ( [ vtoro - red ] ) hold .
the next two sections are devoted to the proof of theorem [ thm4 ] .
in this section we prove theorem [ thm4 ] for @xmath2 with the standard metric . in the case
@xmath2 the eigenvalues of @xmath533 are @xmath534 , @xmath535 and this fact plays a key role in the analysis .
the starting point is the following bilinear improvement of the @xmath536 strichartz inequality on @xmath45 established in @xcite .
[ str - bl - s3 ] for every interval @xmath537 , every @xmath189 there exists a constant @xmath126 such that for every @xmath538 , every @xmath539 , @xmath540 by a time translation we can suppose that @xmath541 $ ] .
moreover , for @xmath542 the function @xmath543 is periodic with respect to @xmath407 and therefore it suffices to give the proof with @xmath544 .
let us write @xmath545 using the parseval identity with respect to @xmath407 we get @xmath546\times m)}^{2 } = \sum_{\tau\in\z } \big\| \sum_{\tau=\lambda_{k_1}+\lambda_{k_2 } } p_{k_1}(f_1)p_{k_2}(f_2 ) \big\|_{l^{2}(m)}^{2},\ ] ] where the summation over @xmath547 is restricted to @xmath548 , @xmath549 . applying the triangle inequality for the @xmath550 norm , the cauchy - schwarz inequality in the summation over @xmath547 , and the bilinear estimate of theorem [ thm2 ] for @xmath18 yields that for every @xmath189 , @xmath551\times m)}^{2 } \\\leq c_{\varepsilon}\,(\min(n_1,n_2))^{1+\varepsilon } \sup_{\tau\in\z}\alpha_{n_1,n_2}(\tau)\prod_{j=1}^{2}\|\delta _ { n_j}f_j\|_{l^{2}(m)}^{2}\ , , \end{gathered}\ ] ] where @xmath552 we claim that @xmath553 .
indeed this follows from the next lemma .
[ le3.2 ] for every @xmath189 there exists @xmath92 such that for every positive integers @xmath431 and @xmath341 , @xmath554 this lemma already appeared in @xcite ( see ( * ? ? ?
* lemma 3.2 ) ) .
we recall the proof .
for @xmath555 it follows from the divisor bound in the ring of gaussian integers which is a euclidean division domain .
for @xmath556 there is at most one value of @xmath547 satisfying the imposed restriction since in this case @xmath557 should range in an interval of size smaller than one . hence for @xmath556 the left hand - side of ( [ chudo ] )
is bounded by @xmath87 .
this completes the proof of lemma [ le3.2 ] .
proposition [ str - bl - s3 ] now readily follows from lemma [ le3.2 ] . from now on
we simply assume that @xmath0 is a three dimensional compact manifold satisfying proposition [ str - bl - s3 ] . proceeding as in ( * ? ? ?
* section 3.2 ) one can show , for instance , that three dimensional zoll manifolds have this property . as a consequence it can be remarked that in fact theorem [ thm1 ] holds for any such manifold . .5
cm first we deduce from proposition [ str - bl - s3 ] the following bilinear estimate in the @xmath482 context .
[ bolen ] for every @xmath189 there exist @xmath558 and @xmath92 such that for every @xmath559 , every @xmath560 , @xmath561 let us suppose that @xmath562 .
using lemma [ sob - co - g ] and the hlder inequality we can write @xmath563 estimate ( [ parva - pista ] ) is better than the needed one with respect to the @xmath564 localization but is far from the needed one with respect to the @xmath565 localization .
we now estimate @xmath566 by means of proposition [ str - bl - s3 ] .
it is indeed possible thanks to the following lemma .
[ claim ] for every @xmath567\frac{1}{2},1]$ ] , every @xmath171 , there exists @xmath568 such that for every @xmath569 , every @xmath570 , @xmath571 let us set @xmath572 , @xmath573 then we can write @xmath574 therefore @xmath575 using the triangle inequality and proposition [ str - bl - s3 ] gives that for every unit interval @xmath537 , every @xmath171 there exists @xmath576 such that @xmath577 hence using the cauchy - schwarz inequality in @xmath578 gives for @xmath425 , @xmath579 using a partition of unity , we can find a @xmath175 , supported in @xmath580 $ ] such that @xmath581 notice that if for @xmath582 , @xmath583 $ ] , we set @xmath584 then @xmath585 indeed ( [ split3 ] ) is straightforward for @xmath586 and @xmath587 and it follows by complex interpolation for @xmath5670,1[$ ] . using the almost disjointness of the supports of @xmath588 , @xmath589 , the triangle inequality , estimates ( [ split1 ] ) , ( [ split ] ) and ( [ split3 ] ) complete the proof of lemma [ claim ]
. next we apply lemma [ claim ] with @xmath590 , @xmath549 in the place of @xmath591 and it follows from the definition of @xmath482 that , for any @xmath425 and any @xmath592 , @xmath593 it is now clear that the proof of proposition [ bolen ] can be completed from and by a suitable hlder inequality let us now turn to the proof of ( [ parvo - red ] ) . set @xmath594 and @xmath595 .
thanks to ( [ gauge ] ) and using that @xmath22 , we obtain that the function @xmath596 is vanishing at least of order @xmath597 at the origin .
therefore , in order to prove ( [ parvo - red ] ) , it suffices to prove @xmath598 assuming that @xmath30 is vanishing to order @xmath597 in zero . we can write @xmath599,\ ] ] where the sum is taken over all dyadic values of @xmath450 ( recall that @xmath600 ) .
we have for @xmath601 , @xmath602 therefore @xmath603 with @xmath604 we have thus the splitting @xmath605 where @xmath606 thanks to the growth assumption on @xmath526 , we have the bound @xmath607 we will provide a bound only for @xmath608 . the analysis for @xmath609 is exactly the same .
we have for dyadic integers @xmath450 , @xmath610 @xmath611 and hence we can write that for dyadic integers @xmath450 , @xmath610 , @xmath612 is equal to @xmath613 using the vanishing property of @xmath21 at the origin allows us to write @xmath614 with @xmath615 where @xmath616 are the derivatives of @xmath617 with respect to the first and the second arguments respectively
. moreover @xmath618 notice that @xmath619 we can write @xmath620 we will provide a bound only for @xmath621 .
the analysis for @xmath622 is exactly the same .
similarly to the analysis for @xmath623 , using once again the vanishing property of @xmath21 at the origin , allows us to expand @xmath624 as follows @xmath625 where , due to the growth assumptions on @xmath29 , @xmath626 satisfies latexmath:[\[\label{alpha-2 } of course we can write explicit formulas for @xmath626 as we did for @xmath628 but it will not be needed in the sequel .
the only information for @xmath626 that we will use is the bound ( [ alpha-2 ] ) .
now , we can write @xmath629 we will provide a bound only for @xmath630 .
the analysis for @xmath631 is exactly the same .
notice that @xmath632 where the sum is taken over all dyadic values of @xmath459 .
for @xmath633 , we set @xmath634 where the sum is taken over dyadic values of @xmath635 , @xmath636 . by duality , to prove ( [ parvo - red - bis ] ) it suffices to establish the bound @xmath637 set @xmath638 we split @xmath639 as @xmath640 , where we define @xmath641 and @xmath642 to be the sums of the terms @xmath643 associated to indexes such that @xmath644 and @xmath645 respectively , and @xmath646 is a large constant to be determined later .
we first evaluate @xmath641 . using proposition [ bolen ] , and the hlder inequality , we get , that for every @xmath189 there exists @xmath558 such that , @xmath647 thanks to ( [ alpha-2 ] ) we can write @xmath648 using lemma [ sob - co - g ] , ( [ harcterizatzia ] ) and the cauchy - schwarz inequality yield , for @xmath425 , @xmath649 we next estimate @xmath650 .
@xmath651 provided @xmath425 . using the last two estimates , we obtain the bound @xmath652 with ( [ l - infty ] ) in hand , we estimate @xmath641 .
let us recall a discrete schur lemma .
[ ds ] for every @xmath653 , every @xmath654 there exists @xmath92 such that if @xmath655 and @xmath656 are two sequences of nonnegative numbers indexed by the dyadic integers , then , @xmath657 let us set @xmath658 summing geometric series imply that there exists @xmath92 such that @xmath659 therefore the schur lemma implies the boundedness on @xmath660 of the bilinear form with kernel @xmath661 .
this completes the proof of lemma [ ds ] . in estimation @xmath641 ,
we first sum with respect to @xmath662 .
writing @xmath663 using lemma [ ds ] and ( [ l - infty ] ) , after summing geometric series in @xmath664 , @xmath665 , we can write for @xmath425 and @xmath666 , @xmath667 since @xmath44 and @xmath668,we have , choosing @xmath189 small enough , @xmath669 therefore , by summing geometric series in @xmath670 , we get the bound @xmath671 it remains to estimate @xmath642 .
this is performed by using the following proposition and summing geometric series .
[ prop.rough ] let @xmath672 .
then there exists @xmath653 , @xmath673 satisfying , @xmath674 and @xmath675 such that for every @xmath633 , @xmath676 , if @xmath677 , @xmath450 , @xmath610 , @xmath678 satisfy @xmath679 then @xmath680 to prove proposition [ prop.rough ] , we consider three regimes : we use proposition [ bolen ] , the hlder inequality and ( [ l - infty ] ) . for every @xmath189 , there exists @xmath558 such that , @xmath682 therefore @xmath683 since @xmath44 , we observe that there exist @xmath189 , @xmath171 and @xmath674 such that @xmath684 the parameter @xmath189 being fixed , we choose @xmath137 as imposed by proposition [ bolen ] .
finally we chose @xmath483 satisfying ( [ restrictions ] ) such that @xmath685 .
we start with a rough bound for @xmath687 . by a repetitive use of leibniz rule and the sobolev embeddings
, we obtain the following statement .
[ rough ] there exists @xmath688 such that for every coordinate patch @xmath689 every @xmath690 , there exists @xmath691 such that for every @xmath692 , @xmath693 we next state a bound for products of eigenfunctions .
[ rough - bis ] let @xmath694 .
there exists @xmath653 such that if @xmath695 then for every @xmath674 there exists @xmath126 and @xmath95 such that for every @xmath696 , @xmath697 provided @xmath698 $ ] , @xmath636 .
a similar argument already appeared in lemma 2.6 of our previous paper @xcite .
the new point here is the presence of @xmath687 . working in local coordinates , due to lemma [ sogge ]
, we can substitute @xmath699 with the oscillatory integral @xmath700 indeed the remainder term can be estimated thanks to the sobolev embeddings and lemma [ rough ] .
we consider three cases .
* case 1.[case111 ] suppose first that @xmath701 .
using lemma [ pak ] , we integrate by parts in the variable @xmath211 by means of the oscillating factor @xmath702 and after @xmath96 integrations , we gain a factor @xmath703 .
on the other hand , due to lemma [ rough ] , the assumption @xmath701 and the sobolev inequality , we obtain that the derivation of the amplitude is causing at most a factor @xmath704 . by taking @xmath705 ,
this completes the proof in the case @xmath701 .
* case 2.[case222 ] suppose next that @xmath706 but @xmath707 .
in this case we can substitute @xmath699 with ( [ oscil-1 ] ) and @xmath708 with @xmath709 indeed in the considered case the remainders in the approximation for @xmath699 and @xmath708 given by lemma [ sogge ] are both @xmath710 as operators from @xmath17 to the sobolev spaces .
thanks to lemma [ pak ] , if we take @xmath711 , we can again integrate by parts in @xmath211 with the slightly modified oscillatory factor @xmath712 where @xmath713 * case 3.[case.333 ] suppose finally that @xmath706 and @xmath714 .
then we can substitute @xmath699 , @xmath708 and @xmath715 with the corresponding oscillatory integrals and we can then argue as in case 2 .
this completes the proof of lemma [ rough - bis ] .
lemma [ rough - bis ] ( with @xmath716 ) is now used to prove proposition [ prop.rough ] for space time functions in this regime .
define @xmath717 as follows @xmath718 further we set @xmath719 since @xmath720 we get the bound@xmath721 since @xmath722 , under the assumption ( [ ass - n ] ) a use of lemma [ rough - bis ] yields , @xmath723 for @xmath425 , a use of ( [ sobolev ] ) and the hlder inequality implies that @xmath724 since @xmath725 , for @xmath726 such that @xmath727 , using ( [ eq - bis ] ) , we get @xmath728 @xmath729 and using lemma [ sob - co - g ] , for @xmath730 , @xmath731 using a crude form of weyl asymptotics , we get a bound @xmath732 ) , ( [ ii - s ] ) , ( [ iii - s ] ) and ( [ weyl ] )
. we shall denote by @xmath735 any quantity bounded by @xmath736 for some @xmath737 .
let @xmath738 there exists @xmath739 such that @xmath740 for any @xmath433 . indeed , working in local coordinates , according to ( * ? ? ?
* proposition 2.1 ) , there exists @xmath741 such that for any @xmath433 @xmath742 therefore , modulo negligible terms , the spectrum of @xmath743 lies in a ring of size @xmath744 which proves the lemma . next we take advantage of this spectral localization to perform integrations by parts
: we have ( for some function @xmath745 ) @xmath746 and modulo negligible terms @xmath747\ , dt\ , dx\right|\ ] ] applying proposition [ bolen ] we obtain @xmath748 but , by sobolev embedding , we have @xmath749 and thus , since @xmath750 and @xmath751 , @xmath752 and we can choose @xmath189 small enough such that @xmath753 .
this completes the proof of proposition [ prop.rough ] ( and thus of ( [ parvo - red ] ) ) in the case of a three dimensional compact manifold satisfying proposition [ str - bl - s3 ] .
[ zabelejka ] let us notice that the estimate ( [ parvo - red ] ) holds for any sub - quintic nonlinearity , not necessarily satisfying the gauge condition ( [ gauge ] ) .
we used ( [ gauge ] ) in the reduction to @xmath21 vanishing of order three at zero performed above because it simplifies a bit the analysis .
more precisely for an arbitrary @xmath21 in the expansions of @xmath623 and @xmath754 above one should add a constant .
this would force one to analyze quadratic nonlinearities separately which can be done with our methods .
thanks to the multilinear nature of our arguments , the proof of ( [ vtoro - red ] ) is essentially the same as for ( [ parvo - red ] ) . indeed for suitable @xmath755 ,
@xmath756 one writes @xmath757 then we expand @xmath758 and for @xmath549 , @xmath759.\ ] ] one then further expand the difference and after a duality argument the proof of ( [ vtoro - red ] ) is reduced to a bound for a @xmath760-linear expression multiplied with a factor similar to @xmath761 appeared in the proof of ( [ parvo - red ] ) .
we omit the details .
in this section we prove theorem [ thm4 ] for @xmath51 with the standard metric
. we do not know whether proposition [ str - bl - s3 ] holds in this case .
instead , we shall prove a trilinear strichartz - type estimate .
let us first introduce some notation . as usual
we identify @xmath762 with @xmath763 .
the eigenfunctions of @xmath12 in the considered case are @xmath764 let us denote by @xmath765 the spectral projector on spherical harmonics of degree @xmath766 on @xmath767 . for @xmath768 , we set @xmath769 the crucial estimate is the following . by a time translation
we can suppose that @xmath541 $ ] .
since @xmath773 is not necessarily integer , we can not employ the argument of proposition [ str - bl - s3 ] which reduces the analysis to the case @xmath774 $ ] .
we shall instead use the following lemma , already used in a similar context in @xcite .
let @xmath779 be such that @xmath780 on the interval @xmath781 $ ] .
set @xmath782 then @xmath783 and the problem is to show that @xmath784 next , we write @xmath785 where @xmath786 it is clear that @xmath787 implies @xmath788 and therefore , using that @xmath779 , we deduce that for every @xmath789 there exists @xmath790 such that @xmath791 a use of the schur lemma completes the proof of lemma [ cska ] .
we expand @xmath792 where the sum is taken over @xmath793 , @xmath794 such that @xmath795 . using the parseval identity with respect to @xmath135 and lemma [ cska ] , we obtain @xmath796\times m ) } \\
\leq c_{t}\,\ , \sum_{(l,\xi)\in\z^{2 } } \big\|\displaystyle\sum_{\stackrel{|l-\lambda_{m_1,n_1}-\lambda_{m_2,n_2}-\lambda_{m_3,n_3}|\leq 1/2}{\xi = m_1+m_2+m_3 } } \,\,\ , \prod_{j=1}^{3}\bigl|\pi_{n_{j}}\theta_{m_{j}}f_{j}\bigr| \big\|_{l^{2}(s^2_{\rho})}^{2}\,,\end{gathered}\ ] ] where the summation over @xmath797 is restricted to @xmath798 such that @xmath799 , @xmath794 .
applying the triangle inequality for the @xmath800 norm , the cauchy - schwarz inequality in the summation over @xmath801 and the trilinear estimate ( [ thm2 - 2 ] ) of theorem [ thm2 ] yields @xmath796\times m ) } \\
\leq \sum_{(l,\xi)\in\z^{2 } } \displaystyle\sum_{\stackrel{|l-\lambda_{m_1,n_1}-\lambda_{m_2,n_2}-\lambda_{m_3,n_3}|\leq 1/2}{\xi = m_1+m_2+m_3 } } ( n_2
n_3)^{\frac{1}{2}}\prod_{j=1}^{3}\|\pi_{n_{j}}\theta_{m_{j}}f_{j}\|_{l^{2}(s^2_{\rho})}^{2 } \\
\leq ( n_2 n_3)^{\frac{1}{2 } } \sup_{(l,\xi)\in\z^{2}}|\lambda(l,\xi)|\prod_{j=1}^{3}\|f_{j}\|_{l^{2}(m)}^{2}\ , , \end{gathered}\ ] ] where @xmath802 it remains to bound the size of @xmath803 .
the number of possible @xmath804 is bounded by @xmath805 .
the number of possible @xmath806 is bounded by @xmath807 .
thus the number of possible @xmath808 is bounded by @xmath809 .
let us now * fix * a possible triple @xmath808 .
our goal is evaluate the number of possible @xmath810 such that @xmath811 . in view of the imposed restrictions
, we can eliminate @xmath812 by concluding that @xmath813 should satisfy @xmath814\right|\leq \frac{1}{2}\ ] ] or equivalently @xmath815 where @xmath816\,.\ ] ] using lemma [ le3.2 ] , uniformly with respect to @xmath817 , the number of integer solutions @xmath818\times [ 0 , cn_2]$ ] of the inequality ( [ broene ] ) is bounded by @xmath819 which implies the estimate @xmath820 the proof of proposition [ str - tl - s21 ] is now completed . from now on
we simply assume that @xmath0 is a three dimensional compact manifold satisfying proposition [ str - tl - s21 ] .
proceeding as in ( * ? ? ?
* section 3.2 ) one can show , for instance , that the product of any zoll surface with @xmath762 has this property . as a consequence it can be remarked that in fact theorem [ thm1 ] holds for any such manifold .
one can show that proposition [ str - bl - s21 ] implies ( [ lille-1 ] ) exactly as we did in the proof of proposition [ bolen ] .
the proof of ( [ lille-2 ] ) follows similar lines .
first , using lemma [ sob - co - g ] and the hlder inequality we get @xmath827 next , exactly as in the proof of proposition [ bolen ] , we obtain that for every unit interval @xmath828 , every @xmath425 , every @xmath171 there exists @xmath568 such that for every @xmath829 , every @xmath830 , @xmath831 using the partition of unity ( [ split ] ) , we get the bound @xmath832 finally , a suitable interpolation between ( [ parva - pista - bis ] ) and ( [ vtora - pista - bis ] ) completes the proof of proposition [ bolen - bis ] .
let us now turn to the proof of theorem [ thm4 ] in the case @xmath833 ( or more generally any manifold satisfying proposition [ str - tl - s21 ] ) .
we can again suppose that @xmath30 is vanishing at least at order three at zero .
we expand @xmath30 as we did in section [ sec.4 ] and are led to estimating terms of the form @xmath834 as in section [ sec.4 ] , we consider two cases and denote by @xmath641 the contribution corresponding to @xmath644 , where @xmath711 is a large constant and by @xmath642 the contribution corresponding to @xmath835 . to study @xmath641 ,
we even make one more expansion of the terms @xmath836 and it results that estimate ( [ parvo - red ] ) is a consequence of the bounds @xmath837 and @xmath838 @xmath839 where @xmath840 and @xmath841 @xmath842 with sums taken over the dyadic values of @xmath843 and @xmath844 , @xmath845 .
moreover @xmath846 enjoys the bound @xmath847 in addition for @xmath848 , we can further expand @xmath849 and we can get the bound @xmath850 the proof of ( [ j ] ) is a consequence of the bilinear estimate ( [ lille-1 ] ) .
more precisely , using ( [ lille-1 ] ) and the hlder inequality , we obtain that for every @xmath189 there exists @xmath851 such that @xmath852 since for @xmath853 we have @xmath854 , we can complete the proof of ( [ j ] ) as we did in section [ sec.4 ] . a similar argument
( using both ( [ lille-1 ] ) and ( [ lille-2 ] ) ) is valid for , if @xmath855 . to prove if @xmath856 , we use proposition [ bolen - bis ] in its full strength . set @xmath857 in order to estimate @xmath858 , we use the following form of hlder s inequality .
@xmath8590,1 ] , \qquad \left|\int_{\r\times m}\ , fg\right| \leq\big(\int_{\r\times m}|f|\big)^{1-\gamma } \big(\int_{\r\times m}|f| @xmath860 , @xmath8610,1[$ ] . using ,
we can write , @xmath862^{1-\gamma } [ k_{l_0l_1l_2l_3l_4}^{n_0n_1n_2n_3n_4}]^{\gamma}\ , , \ ] ] where @xmath863 thanks to ( [ posledno ] ) , the second factor @xmath864 in ( [ inter ] ) enjoys the bound @xmath865 .
using hlder inequality and proposition [ bolen - bis ] ( both ( [ lille-1 ] ) and ( [ lille-2 ] ) ) , we obtain that for every @xmath189 there exists @xmath851 such that @xmath866 next we estimate @xmath864 . by writing @xmath867 , using hlder s inequality and proposition [ bolen - bis ] ( twice ( [ lille-2 ] ) )
, we obtain that for every @xmath189 there exists @xmath851 such that @xmath864 is bounded by @xmath868 writing @xmath869 , using the cauchy - schwarz inequality , we get for @xmath870 , @xmath871^{2 } \big)^{\frac{1}{2}}\|u\|_{x^{1,b } } \leq cn_{4}^{\frac{1}{4}+\varepsilon}\|u\|_{x^{1,b}}\ , .\end{gathered}\ ] ] therefore , we have the estimate , @xmath872 coming back to ( [ inter ] ) , we obtain the following estimate @xmath873 let us take @xmath189 such that @xmath874 or equivalently , @xmath875 note that a proper choice of @xmath39 is possible thanks to the sub critical assumption @xmath44
. therefore there exists @xmath876 such that for @xmath877 , @xmath878 thanks to ( [ glad ] ) and ( [ crucial ] ) , we obtain @xmath879 for @xmath880 by summing geometric series in @xmath664 , @xmath665 , @xmath881 , @xmath882 , @xmath883 , @xmath610 , @xmath678 , @xmath884 while the sum over @xmath885 is performed by invoking lemma [ ds ] . in the first regime , we apply the same strategy as when @xmath644 .
indeed , in this regime , @xmath890 and we obtain ( after expanding once more the non linear term ) with @xmath228 @xmath891 which gives the summability in @xmath892 . the second regime can be dealt with in the same way as in the previous section ( in this regime , we gain arbitrary powers of @xmath893 ) .
finally we concentrate on the last regime .
let @xmath894 as in section 4 , we have using that for some function @xmath895 we have @xmath896 we can integrate by parts in the integral defining @xmath897 : @xmath898 \end{gathered}\ ] ] now to estimate the first term in the right hand side of
we simply apply the strategy already used in the case @xmath899 , the only difference being the additional factor @xmath900 ( the factor @xmath901 coming from the action of the laplace operator on @xmath902 ) , which allows , since @xmath903 , to exchange the roles of @xmath744 and @xmath450 and gain summability .
remark here that the additional operator @xmath904 applied to @xmath905 plays no role since it disappears when one takes @xmath17 norms .
next we estimate the two other terms .
we expand the derivatives and observe that @xmath906 @xmath907 next we use that for @xmath908 , @xmath909 is an @xmath910-semi - classical operator and consequently the gradient of a spectrally localized function is essentially spectrally localized .
this allows to apply the trilinear estimate to @xmath743 on one hand and to @xmath911 on the other hand ( remark that the moduli in do not spoil the estimate since we take @xmath17 norms ) .
we obtain that the contribution of these two terms is bounded ( for any @xmath189 ) by @xmath912 using that @xmath592 and @xmath751 , for @xmath913 small enough , this term can be bounded by @xmath914 giving the required summability in @xmath915 .
a careful examination of the proof above shows that theorem [ thm4 ] still holds for a three dimensional manifold @xmath0 satisfying the more general trilinear strichartz estimate , @xmath916\times m ) } \leq c n_{3}^{1+a}n_{2}^{1-a+\varepsilon}\prod_{j=1}^{3}\|\delta_{n_j}f_j\|_{l^{2}(m)}\ , .\end{gathered}\ ] ] this appendix is devoted to the optimality of the assumption @xmath44 in theorem [ thm1 ] .
let us again consider a 3d - manifold @xmath0 endowed with a riemannian metric @xmath13 and @xmath12 the laplace - beltrami operator acting on functions of @xmath0 .
we consider the following non - linear schrdinger equation on @xmath0 @xmath917 where @xmath918 and @xmath69 .
let us fix @xmath474 .
equation ( [ eq.surcrit ] ) is well - posed for data in @xmath412 by the energy method . in particular , for every bounded set @xmath919 there exists @xmath920 such that for every @xmath921 the cauchy problem ( [ eq.surcrit ] ) has a unique solution on the interval @xmath922 $ ] in the class @xmath923\,;\,h^{s}(m))$ ] .
moreover the flow map @xmath924 is continuous ( and even lipschitz continuous ) from @xmath490 , endowed with the @xmath418 metric , to @xmath923\,;\,h^{s}(m))$ ] .
the next statement shows that @xmath321 can not be extended as a continuous map on bounded sets of @xmath34 .
the result of theorem [ thm6 ] , in the particular case @xmath931 , endowed with the standard metric , can be found in @xcite .
we also refer to @xcite for more ill - posedness results for nls on @xmath269 , @xmath48 , with power - like nonlinearities and data in @xmath37 , @xmath654 . the proof of theorem [ thm6 ] is strongly inspired by @xcite .
the only observation we make here is that the dilation arguments involved in the proof in @xcite are not essential .
it is clear from the proof we present that the discussed phenomenon is completely local , i.e. the whole analysis is close to a point of @xmath0 for very small times .
we work in a local coordinate patch around @xmath279 and consider as initial data the sequence @xmath932 , where @xmath933 is a fixed non negative smooth compactly supported @xmath934 with @xmath171 to be fixed later .
remark that @xmath935 let us set @xmath936 . then @xmath937 is the solution of the equation @xmath938 let us give a basic bound for @xmath939 .
the change of variable @xmath943 gives @xmath944\big\|_{l^2 } \\ &
\geq \smash{\kappa_n \bigl(2 t \kappa_n n^{1/2}\,\|\varphi(y ) \partial_z
f(\kappa_n n^{1/2}\varphi(y))\cdot \nabla_y \varphi(y ) \|_{l^2 } - \|\nabla_y \varphi\|_{l^2}\bigr ) } \\ & \geq c\,t\,\kappa ^{\alpha}_n n^{(\alpha-1)/2 } - c\kappa_n \end{aligned}\ ] ] which implies the lemma . since the initial data are in @xmath953 , @xmath474 , we know that @xmath523 exist on a ( non empty ) maximal time interval @xmath954 .
consequently , to prove lemma [ le.est ] , we simply prove the _ a priori _ estimates which ensure , by a classical bootstrap argument , both the existence and the control on @xmath955 for @xmath951\cap [ 0 , \widetilde { t}_n[$ ] .
let us set @xmath956 .
for the sake of conciseness , in the rest of the proof of lemma [ le.est ] , we drop the subscript @xmath957 of @xmath946 , @xmath958 and @xmath959 .
the a priori estimates involved in the proof are simply energy inequalities in the equations @xmath960 where @xmath961 from the explicit formula for @xmath743 , we have for @xmath962 , @xmath963 , @xmath964 according to the gagliardo - nirenberg inequalities , @xmath965 we deduce @xmath966 to estimate @xmath775 , we proceed similarly .
more precisely , thanks to ( [ gn ] ) , we estimate systematically the terms involving @xmath743 or @xmath286 below the @xmath967 sign in @xmath968 . the only term which can not be estimated by invoking ( [ gn ] ) is @xmath969 in order to evaluate ( [ new ] ) , we use the bound @xmath970 and we obtain @xmath971 we are therefore conducted to the following estimate for @xmath775 @xmath972 next , thanks to the formula for @xmath743 , for @xmath962 , we estimate the source terms , @xmath973 according to , and , we obtain @xmath974 suppose first that @xmath975 which is clearly the case at least for @xmath976 since @xmath977 . notice that @xmath978 therefore @xmath979 \leq c\ , n^{4-\frac{\alpha-1}{2}}\log^{3/4}(n)\ , e^{-c\,t\ , n^{(\alpha-1)/2 } \log^{1/4}(n ) } \ , .\ ] ] integrating the last inequality between @xmath279 and @xmath407
gives the estimate @xmath980 for every @xmath674 there exists @xmath981 such that for @xmath928 $ ] , @xmath982 since @xmath983 , by taking @xmath674 small enough , we obtain that there exists @xmath189 such that for @xmath928 $ ] , we have @xmath984 finally the usual bootstrap argument allows to drop the assumption @xmath975 .
this completes the proof of lemma [ le.est ] .
it is clear that by interpolation , the quantity @xmath985 controls uniformly with respect to @xmath957 the @xmath1 norm of @xmath20 .
consequently , it follows from lemma [ le.expl ] and lemma [ le.est ] that for @xmath986 and @xmath987 , @xmath988 this completes the proof of theorem [ thm6 ] .
10 j. bourgain .
fourier transform restriction phenomena for certain lattice subsets and application to nonlinear evolution equations i. schrdinger equations .
, 3:107156 , 1993 . j. bourgain .
exponential sums and nonlinear schrdinger equations .
, 3:157178 , 1993 . j. bourgain .
eigenfunction bounds for the laplacian on the @xmath957-torus . , 3:6166 , 1993 . j. bourgain .
remarks on strichartz inequalities on irrational tori . , 2004 .
n. burq , p. grard and n. tzvetkov .
strichartz inequalities and the nonlinear schrdinger equation on compact manifolds . , 126 - 3 : 569605 , 2004
. n. burq , p. grard , and n. tzvetkov .
an instability property of the nonlinear schrdinger equation on @xmath989 .
, 9(2 - 3):323335 , 2002
. n. burq , p. grard , and n. tzvetkov .
the cauchy problem for the nonlinear schrdinger equation on compact manifold .
, 10 : 12 - 27 , 2003 .
n. burq , p. grard and n. tzvetkov .
bilinear eigenfunction estimates and the nonlinear schrdinger equation on surfaces . to appear in _ inventiones mathematicae _
n. burq , p. grard and n. tzvetkov .
multilinear estimates for laplace spectral projectors on compact manifolds . c. r. acad .
paris , ser .
i 338 : 359 - 364 , 2004 .
t. cazenave .
semi - linear schrdinger equations .
courant lecture notes in mathematics , 10 .
new york university .
american mathematical society , providence , ri , 2003 .
m. christ , j. colliander and t. tao .
ill - posedness for nonlinear schrdinger and wave equations . , 2003 .
s. gallot , d. hulin and j. lafontaine , riemannian geometry , 1990 j. ginibre and g. velo . on a class of nonlinear schrdinger equations . ,
32 : 1 - 71 , 1979 .
j. ginibre .
le problme de cauchy pour des edp semi - linaires priodiques en variables despace ( daprs bourgain ) .
, 237 : 163187 , 1996 .
l. hrmander . . , 121:193218 , 1968 . l. hrmander .
oscillatory integrals and multipliers on @xmath990 .
, 11:111 , 1973 . t. kato .
on nonlinear schrdinger equations . , 46 : 113 - 129 , 1987
. s. klainerman and m. machedon .
remark on strichartz - type inequalities .
, 5:201220 , 1996 . with appendices by j. bourgain and d. tataru
. s. klainerman and m. machedon .
finite energy solutions of the yang - mills equations in @xmath991 , 142 ( 1 ) : 39119 , 1995 .
h. koch and d. tataru .
l . lions .
quelques mthodes de rsolution des quations aux drives partielles non linaires 1969 .
oscillatory integrals and spherical harmonics . , 53:4365 , 1986 .
c. sogge . concerning the @xmath15 norm of spectral clusters for second order elliptic operators on compact manifolds .
, 77:123138 , 1988 . c. sogge .
fourier integrals in classical analysis . , 1993 .
e.m . stein . .
, princeton university press , princeton , nj , 1993 . g. szeg .
orthogonal polynomials . , 1974 .
a. zygmund . 50:189201 , 1974 . t. tao .
multilinear weighted convolutions of @xmath17 functions , and applications to non - linear dispersive equations . 123 : 839908 , 2001 . | we study nonlinear schrdinger equations , posed on a three dimensional riemannian manifold @xmath0 .
we prove global existence of strong @xmath1 solutions on @xmath2 and @xmath3 as far as the nonlinearity is defocusing and sub - quintic and thus we extend results of ginibre - velo and bourgain who treated the cases of the euclidean space @xmath4 and the torus @xmath5 respectively .
the main ingredient in our argument is a new set of multilinear estimates for spherical harmonics .
on tudie lquation de schrdinger sur une varit de dimension trois @xmath0 .
on dmontre lexistence globale en temps de solutions fortes @xmath1 si @xmath6 ou @xmath7 , pour les non linarits sous quintiques et dfocalisantes . on tend ainsi
les rsultats de ginibre et velo et bourgain qui ont trait les cas de lespace euclidien @xmath8 et du tore @xmath9 respectivement .
lingrdient essentiel de notre dmonstration est lobtention de nouvelles estimes multilinaires pour les harmoniques sphriques . |
magnetic frustration have provided us with many intriguing topics e.g. the phenomena of order - by - disorder , the residual entropy at absolute zero temperature , disordered spin liquids , etc@xcite .
there are a variety of models which are known to exhibit the so - called frustration effects . among them , the @xmath4 @xmath0-@xmath1 model on a square lattice has been extensively investigated over the last two decades as one of the simplest models to study how frustration destroys magnetic orders and stabilizes paramagnetic phases .
the model is defined by adding antiferromagnetic interactions on diagonal bonds to the ordinary heisenberg antiferromagnet on a square lattice ( see fig .
[ fig : plafull ] ) : @xmath5 where the summations ( n.n . ) and ( n.n.n . ) are taken for the nearest - neighbor- and the second - neighbor ( diagonal ) pairs , respectively . in the classical ( @xmath6 )
limit , the ground state is readily obtained by computing fourier transform @xmath7 of the exchange interactions and minimize it in the @xmath8-space : * @xmath9 , @xmath10 : the ground state has nel antiferromagnetic order ( naf ) . *
@xmath11 : the ground state consists of two interpenetrating nel - ordered square lattices .
first quantum correction fixes the relative angle between the two ordering directions and selects the so - called _ collinear antiferromagnetic order _
* otherwise : the ferromagnetic ( fm ) ground state is stabilized . for @xmath12 ,
the next - nearest - neighbor ( diagonal ) interaction gives rise to no frustration and only the case with @xmath13 is non - trivial .
the case @xmath14 has been extensively studied in the context of spin - gap phases stabilized by the frustrating interactions .
chandra and doucot@xcite investigated the model in the large-@xmath15 limit and concluded that a non - magnetic ( neither naf nor caf ) phase appeared around the classical phase boundary @xmath16 .
the most quantum case @xmath4 has been studied later both by numerical@xcite and by analytical methods @xcite ( for other literatures , see , for instance , refs . and references cited therein ) . by now
it is fairly well established that we have spin gapped phase(s ) in the window @xmath17 although the nature of the spin - gap phase(s ) is still in controversy . the case with @xmath18 , @xmath13 has been less investigated and recent analyses@xcite suggested that there is another non - magnetic ( probably spin - nematic ) phase around the classical boundary @xmath19 between caf and fm . from an experimental viewpoint , most compounds@xcite found so far correspond to the ordered phase ( caf ) of the @xmath20-@xmath21 model . recently , kageyama _ et al . _
reported@xcite a new two - dimensional cu - based compound ( cucl)la@xmath22 . in this compound ,
two - dimensional sheets consisting of @xmath23 and @xmath24 are separated from each other by non - magnetic [ la@xmath22 ] layers and within each sheet the @xmath23 ions form a square lattice .
the @xmath24 ions are located at the center of plaquettes and from a naive goodenough - kanamori argument the @xmath4 @xmath20-@xmath21 model with @xmath18 and @xmath13 is suggested as the model hamiltonian for ( cucl)la@xmath22 .
what is remarkable with this compound is that inelastic neutron scattering experiments @xcite observed a finite spin gap 2.3mev(=26.7k ) above the spin - singlet ground state .
subsequently , high - field magnetization measurements@xcite were carried out to show that magnetization monotonically increased between two critical fields @xmath26 t and @xmath27 t .
the data for ( i ) the weiss temperature and ( ii ) the saturation field @xmath28 in principle determine the coupling constants @xmath20 and @xmath21 .
unfortunately , none of the solutions @xmath29 obtained in this way reproduced the spin - gap behavior@xcite .
therefore , the usual @xmath20-@xmath21 model does not seem to work .
the second intriguing point concerns the magnetization process . from the standard scenario @xcite ,
the onset of magnetization at @xmath30 in spin - gapped systems is understood as bose - einstein condensation ( bec , or superfluid onset , more precisely ) of the lowest - lying triplet excitation ( magnon ) and the lower critical field @xmath31 at @xmath32 is given by the spin gap @xmath33 as @xmath34 .
this bec scenario has been confirmed in various spin gap compounds@xcite .
recent specific - heat- and magnetization measurements @xcite for ( cucl)la@xmath22 exhibited behavior typical of spin - bec transitions and suggested that the magnetization - onset transition at @xmath31 may be described by bec of a certain kind of magnetic excitations .
however , we immediately find a serious difficulty when we try to understand this within the standard bec scenario ; the lower critical field @xmath35 t expected from the observed spin gap @xmath36mev at the zero field ( where the experimental value @xmath37 is used ) in the standard scenario is much larger than the observed value@xcite @xmath26 t .
one possible explanation for this discrepancy may be that a lower - lying triplet excitation which is responsible for the bec was not observed in the neutron - scattering experiments because of selection rules .
however , this seems unlikely since powder samples were used and usually one can hardly expect a perfect extinction of a certain triplet excitation in such powder samples .
neither susceptibility measurements @xcite nor nmr data @xcite indicate such a hidden triplet excitation .
an alternative and a more appealing scenario would be that the bec occurs not in a single - particle channel but in a multi - particle channel .
that is , what condenses to support a spin - superfluid is a bound state of magnon excitations .
the possibility of multi - magnon condensation has been proposed theoretically @xcite in the context of a _
kinetic _ quintet bound state in the shastry - sutherland model ( see ref . and references cited therein ) .
in fact , gapped quintet excitations which come down as the external field is increased were observed in the esr experiments @xcite carried out for sr@xmath38 , whereas small dzyaloshinskii - moriya interactions hindered a quintet bec from being observed in that compound ( see also ref . ) .
one of the simplest @xmath20-@xmath21-like models which realize the above scenario and have a finite spin gap would be the @xmath4 @xmath20-@xmath21 model with a plaquette structure ( see fig .
[ fig : plafull ] ) .
a similar model ( @xmath39 ) has been investigated to develop a plaquette series expansion@xcite . in this paper
, we mainly focus on the region @xmath18 , @xmath13 where the quintet excitation is expected to play an important role in low - energy physics .
the organization of the present paper is as follows . in section
ii , we briefly recapitulate the problem of a single plaquette mainly to establish the notations .
the coupling among plaquettes will be taken into account in section [ sec3 ] by two different methods : ( i ) a plaquette extension of the bond - operator mean - field theory@xcite and ( ii ) a perturbation expansion with respect to the inter - plaquette couplings .
we find gapped triplets and quintet for small enough inter - plaquette couplings in both methods .
for larger values of inter - plaquette couplings , one of the gapped excitations softens and the form of the effective interactions among the soft excitations determines the resulting magnetic phases . by using the gaps obtained in the perturbation expansion , we determine the semi - quantitative phase diagram in section [ sec : phase ] ( see fig .
[ fig : souzug ] and fig .
[ fig : nf ] ) . the effect of high magnetic field will be considered in section [ sec : magnet ] . for high enough field compared with the spin gaps , we can approximate the low - energy sector by using only the singlet and the lowest excited state . for @xmath18
, we may expect that the quintet touches the singlet ground state first and a multi - particle bec occurs . on general grounds ,
a single - particle ( magnon ) bec phase is expected to have finite transverse magnetization .
actually , in the bec phase of tlcucl@xmath40 , the transverse magnetization has been observed in the experiment@xcite . in the case of a multi - particle bec , however , the transverse magnetization does not appear . to investigate the magnetization process , we shall keep only the singlet and the quintet to derive a hardcore boson model as the effective hamiltonian valid in high enough magnetic field .
a mean - field approximation@xcite will be applied to the resulting effective hamiltonian to draw a full magnetization curve .
interesting phases ( a 1/2-plateau and supersolids ) will be discussed . according to the value of the parameters
, we shall roughly classify the magnetization curve in fig .
[ fig : magsouzu ] . a summary of the main results and the discussion on the connection to the spin - gap compound @xmath3 will be given in sections [ sec : cucl ] and [ sec : conclusion ] , respectively .
the equations omitted in the text will be summarized in the appendices .
we consider a spin-1/2 @xmath0-@xmath1 model with a plaquette structure where the interactions among spin-1/2s are explicitly tetramerized ( see fig .
[ fig : plafull ] ) .
the model is made up of four - spin units ( _ plaquettes _ ) and the four sites constituting a single plaquette are connected by the nearest - neighbor- ( @xmath20 ) and the second - neighbor ( @xmath21 ) interactions as is shown in fig .
[ fig : plas ] ) .
the inter - plaquette interactions ( both the nearest - neighbor- and the diagonal ) which connect those units are multiplied by a distortion constant @xmath41 @xmath42 .
this parameter may be thought of as modeling the distortion of the underlying lattice in a simple way . in the case of @xmath43 ,
this model reduces to the homogeneous @xmath0-@xmath1 model , while when @xmath44 , the plaquettes are decoupled from each other . on these bonds.[fig : plafull ] ] and @xmath1 between @xmath4 spins.[fig : plas ] ] let us begin by analyzing a single isolated plaquette , which corresponds to the case @xmath45 .
the eigenstates of a single plaquette can be easily obtained as follows .
first we note that a plaquette hamiltonian can be rewritten as @xmath46 where @xmath47 , @xmath48 and @xmath49 . therefore , all the @xmath50 eigenstates are classified by the three quantum numbers as @xmath51 .
the eigenvalues @xmath52 are given by @xmath53 [ sec2:ee ] here a constant @xmath54 has been dropped just for simplicity . the energy of these states is shown in fig . [
fig : elebel ] . for @xmath55 ,
the spin - singlet state @xmath56 is the ground state , the triplets @xmath57 , @xmath58 are the first excited states , and the quintet @xmath59 is the second excited state . for @xmath60 ,
the singlet @xmath56 is the ground state , quintet @xmath59 is the first excited state , and triplets @xmath57 , @xmath58 are the second excited state .
+ , @xmath58 and the quintet @xmath59 .
we take the units of energy as @xmath1 , and the energy is plotted as a function of @xmath61[fig : elebel ] ] the singlet @xmath56 is written as @xmath62 in what follows , the single - spin states in ket will be shown in the order of 1,3,2,4 , i.e. @xmath63 in fig .
[ fig : plas ] . for later convenience ,
we name the two triplets @xmath57 and @xmath58 as @xmath64 and @xmath65 ( @xmath66 ) respectively .
the explicit expressions of the two triplets are given as : @xmath67 @xmath68 to label the quintet @xmath59 states , we use the eigenvalues of @xmath69 , i.e. @xmath70 whose expressions are given explicitly as : @xmath71 [ qq ]
for @xmath45 and @xmath72 , all plaquettes are in the singlet state @xmath56 .
finite inter - plaquette interactions @xmath41 induce various tunneling processes among plaquettes to change both the ground state and the excitations over it . for finite @xmath41
, we calculate the excitation energy by two different approaches .
one is the bond - operator mean - field theory ( mft)@xcite , which gives the excitation energy of the triplets @xmath64 , @xmath65 .
another is the second - order perturbation theory in @xmath41 , and it gives the energy of the quintet @xmath59 as well as that of @xmath64 and @xmath65 .
for sufficiently small @xmath41 , both approximations yield finite energy gaps for these excitations and when one of these gaps closes , the corresponding ( bosonic ) excitation condenses to form a magnetically ordered state .
the energy of triplet excitations can be observed by inelastic neutron scattering experiments .
both approximations may not be reliable for large @xmath41 and small @xmath73 .
let us begin with the bond - operator mft@xcite .
for @xmath74 and @xmath45 , @xmath56 is the ground state and the degenerate triplets @xmath57 , @xmath58 are the first- or the second excited state ( see fig . [
fig : elebel ] ) .
therefore , we may truncate the hilbert space and consider a subspace spanned by the singlet @xmath56 and the triplets @xmath64 , @xmath65 .
this approximation is reliable to estimate the excitation energy of the triplets , unless @xmath59 condenses . in this subspace ,
nonzero matrix elements of @xmath75 is @xmath76 [ psgs ] where @xmath77 . using boson operators
@xmath78 satisfying the standard commutation relations , @xmath79=1 $ ] , @xmath80=\delta_{\alpha\beta}$ ] , @xmath81=\delta_{\alpha\beta}$ ] , @xmath82=0 $ ] , etc , the local spin operator @xmath75 can be written as @xmath83 where the summation over repeated indices is implied . since the restriction that each plaquette has exactly one particle leads to the local constraint @xmath84 , we introduce the lagrange multiplier @xmath85 and add a term @xmath86 to each plaquette hamiltonian .
we may assume that @xmath85 for each plaquette takes the same value @xmath87 for all plaquettes because of the translation invariance .
next , we replace @xmath88 by its expectation value @xmath89 , since the @xmath88 boson condenses in the ground state .
moreover , since the triplet is dilute when the energy gap is positive , we may ignore the terms consisting of three or four triplet operators . in this way
, we obtain the mean - field hamiltonian @xmath90 consists only of bilinear terms in @xmath91 and @xmath92 .
the mean - field parameters @xmath93 are determined by requiring the expectation values of the derivatives of @xmath90 with respect to the mean - field ( mf ) ground state vanish : @xmath94 or equivalently by finding the extrema of the mean - field ground - state energy @xmath95 : @xmath96 in particular , @xmath95 must be minimum for @xmath97 . in this approximation ,
the inter - plaquette interactions associated with the site @xmath98 reads @xmath99 where the site labels @xmath100 and @xmath101 are defined in fig .
[ fig : heizu ] .
( shown by a thick line).[fig : heizu ] ] summing up all four interactions and doing fourier transformation , the total hamiltonian @xmath90 reads @xmath102 where we have defined @xmath103 if we introduce a vector @xmath104 , the mf hamiltonian @xmath90 can be written compactly as @xmath105 where @xmath106 denotes the total number of plaquettes and the kernel @xmath107 is given as @xmath108 @xmath109 using a @xmath110 real matrix @xmath111 ( see appendix [ sec : appendix-1 ] for the detail ) , we can diagonalize @xmath107 by the bogoliubov transformation : @xmath112 as is shown in appendix [ sec : appendix-1 ] , @xmath90 then reduces to @xmath113 where the mean - field ground state energy is given as : @xmath114^{\frac{1}{2 } } \ .\label{pmpm}\end{aligned}\ ] ] in eq .
( [ pmpm ] ) , the order of signs @xmath115 coincides on both sides . since @xmath116 , condensation of the triplets @xmath91 and @xmath92 occurs when the equality holds at some @xmath117 .
otherwise , there is no condensation , and @xmath118 .
therefore , there exist rotational symmetry and no magnetic order . in this case , @xmath119 are the excitation energy of triplets .
we looked for the solutions @xmath93 to the set of equations ( [ eqn : variational ] ) numerically .
for example , we found @xmath120 for the set of parameters @xmath121 .
the dispersion relation of the excitation energy @xmath122 is shown in fig .
[ omega1 ] . in ( [ gaph ] ) , which has the lower energy of the two triplets , for the parameters @xmath123
if the excitation becomes soft @xmath124 at some @xmath117 , the system is in a magnetically ordered phase . from the known results@xcite
, we expect that ordered phase appears for @xmath41 sufficiently close to 1 . to determine the phase boundary between the paramagnetic phase and magnetically ordered ones , we searched the @xmath125 plane for the points where the mean - field gap vanishes .
unfortunately we found that the gap did not close in the relevant parameter region @xmath126 , and that the disordered singlet phase persisted ; the gap vanished only for larger @xmath127 .
this unacceptable result may be attributed to the fact that the bond - operator mean - field theory probably overestimates the stability of the plaquette phase . in this section ,
we compute the energy gap of triplets @xmath57 , @xmath58 and the quintet @xmath59 by the second order perturbation theory in the distortion parameter @xmath41 .
the naive expansion in @xmath41 is ill - behaved in the vicinity of the point @xmath128 and we have to use another perturbation scheme for that region .
let us consider the states where there exists only one triplet and all the other plaquettes are in the singlet @xmath56 state . if the coupling constant of inter - plaquette interaction @xmath41=0 , these states are @xmath106-fold degenerate , where @xmath106 is the number of plaquettes . for finite @xmath41 ,
the second order perturbation induces hopping of the triplet to nearest or next nearest neighbors and lifts the degeneracy .
rotational symmetry forbids the hopping which changes the spin label @xmath129 or the magnetic quantum number . on the other hands ,
the transitions between two different triplets @xmath130 and @xmath131 of the same label @xmath132 occur .
for example , the hopping amplitude of @xmath133 to the nearest - neighbor plaquette is given by @xmath134 the degeneracy is partially resolved by the hopping of @xmath133 .
the transition between @xmath130 and @xmath131 will be considered later . in the second - order perturbation ,
the processes that the triplet returns to the original site is also allowed . including this effect ,
the energy change of @xmath135-particle is given by @xmath136 in ( [ ap2:p ] ) .
similarly , that of @xmath137 is given by @xmath138 in ( [ ap2:q ] ) .
next , we consider the transition between @xmath135 and @xmath139 .
the transition amplitude is given by @xmath140 therefore , for each @xmath141 , eigenstates @xmath142 satisfies @xmath143 the expressions of @xmath144 are given in appendix [ sec : ap2 ] .
after this procedure , the degeneracy with respect both to the position and to the species @xmath130 and @xmath131 is resolved .
there also exists an energy shift in the ground state .
taking all these into account , we obtain the energy of the triplets : @xmath145 where @xmath146 denotes the energy shift of the bare ground state where all plaquettes are occupied by the singlet @xmath56 and is given by eq.([ap2:es ] ) .
the dispersion relation of the lower branch @xmath147 is shown in fig .
[ bunsan1 ] of triplet at @xmath148 .
[ bunsan1 ] ] the lower branch @xmath149 takes its minimum at the @xmath150-point @xmath151 , and @xmath152 gives spin gap @xmath153 .
the second order expression of @xmath153 is given in ( [ ap2:tmin ] ) .
the expression tells us that @xmath154 has a pole at @xmath155 and that the standard perturbation breaks down near the pole . to remedy this
, we introduce another perturbation parameter @xmath156 and carry out a double expansion in both @xmath41 and @xmath157 .
then , we obtain the energy gap in a modified method @xmath158 given in eq.([ap2:tmind ] ) .
this improved energy gap is expressed to give a better approximation around @xmath159 next , we consider states containing only one quintet in a background of the singlet plaquettes . as
before , the degeneracy with respect to the position of the quintet plaquette is resolved by hopping .
up to the second order in @xmath41 , the hopping to nearest neighbor is given by @xmath160 and the hopping to next nearest neighbor does not occur .
taking into account the processes that the quintet returns to the original site and the energy shift of the ground state , the excitation energy of quintet is given by @xmath161 in ( [ ap2:quin ] ) .
the dispersion relation is shown in fig .
[ bunsanq ] of quintet at @xmath148.[bunsanq ] ] since the quintet dispersion @xmath161 takes its minimum at @xmath162 , the quintet gap is given by @xmath163 we note that there is the pole at @xmath164 and the approximation becomes poor for @xmath165 .
if the inter - plaquette coupling @xmath41 is increased , one of the energy gaps of the triplets ( ( [ ap2:tmin ] ) and ( [ ap2:tmind ] ) ) and the quintet ( [ ap2:quin ] ) becomes @xmath166 at a certain critical value of @xmath41 . when it happens , the corresponding particle condenses and a phase transition occurs from the gapped spin - singlet phase to superfluid phases with magnetic long - range order .
therefore , we can classify the phases according to what kind of particles condense and what kind of magnetic orders is stabilized by a given set of interactions among them . in figs .
[ fig : qt ] and [ fig : qtmod ] , we plot the value of @xmath41 at which the smallest energy gap becomes 0 .
at which the energy gaps @xmath153 and @xmath167 close .
the energy gap of the triplets @xmath153 is given in ( [ ap2:tmin ] ) , which is not reliable near @xmath168 because of the pole there , and that of the quintet @xmath167 is in ( [ quingap ] ) , which is not reliable near @xmath169.[fig : qt ] ] when the energy gaps are equal to 0 .
we use @xmath170 ( eq.([ap2:tmind ] ) ) for the triplets , which is reliable even in the vicinity of @xmath168 .
the curve for the quintet is the same as in fig .
[ fig : qt ] .
we only plot the region @xmath171.[fig : qtmod ] ] if we assume that no further condensation occurs in the other kinds of particles once the triplets or the quintet condenses , the phase diagram fig .
[ fig : souzug ] is obtained . when we mapped out the phase diagram fig .
[ fig : souzug ] , we have used two different expressions ( [ ap2:tmind ] ) and ( [ ap2:tmin ] ) for the energy gap of the lowest triplet in the vicinity of @xmath168 and away from it ( @xmath172 ) , respectively . we have also neglected the quintet around @xmath169 since the collapse of the quintet gap there ( see fig .
[ fig : qt ] ) can be attributed to the existence of a pole and is just an artifact of the perturbative approximation .
note that the phase boundary between the two regions covered by eq.([ap2:tmin ] ) and eq.([ap2:tmind ] ) is only schematic .
( eq.([hqumean])).[fig : souzug ] ] now let us discuss the nature of the ordered phases appearing after the condensation . in the region shown as `` caf '' ( highlighted in red ) in fig .
[ fig : souzug ] , condensation occurs to the singlet and the triplets .
then , we may expect : @xmath173 ) , implies @xmath174 provided that @xmath175 and @xmath176 .
note that all the plaquettes are in the same state , since the energy of the triplet takes its minimum at the @xmath150-point @xmath151 ( see fig .
[ bunsan1 ] ) .
when the combination @xmath177 of the two bosons condenses , the relation @xmath178 holds and the ground state has the transversely aligned ( i.e. @xmath179 ) collinear antiferromagnetic order . in the case where @xmath180 condenses , on the other hand , we have @xmath181 instead and the system is in the collinear antiferromagnetic ground state in the longitudinal ( @xmath182 ) direction .
this is consistent with the known results@xcite .
now we move on to a more interesting case . in the green region in fig .
[ fig : souzug ] , frustration is strong ( @xmath183 ) and nontrivial order may be expected .
in fact , shannon _
et al._@xcite analyzed the uniform @xmath184 model by numerical exact diagonalizations up to clusters of 36 spins and found a spin - nematic phase with @xmath185-wave ( or @xmath186 ) symmetry for @xmath187 . in the state with the nematic order
, the expectation value of the rank-1 tensor vanishes @xmath188 , while we have a finite expectation value of the following traceless rank-2 tensor : @xmath189 where @xmath190 and @xmath191 label the lattice sites . as is shown in fig .
[ fig : souzug ] , the singlet and the quintet condense in the region of interest .
this is analogous to the spinor bose - einstein condensation of spin-2 particles ( here particles are defined not on the lattice sites but on the plaquettes ) .
we consider a single plaquette ( see fig .
[ fig : plas ] ) and , as before , denote the singlet and the quintet respectively by @xmath192 and @xmath193 . to investigate
what kind of magnetic order is stabilized in the condensate , let us introduce the following mean - field ansatz for the ground state : @xmath194 where the product is over all plaquettes and the complex numbers @xmath195 satisfy @xmath196 .
then , since the rank-1 tensor can not give rise to transitions between the spin-0 states and the spin-2 ones by the wigner - eckart theorem@xcite , we have @xmath197 , @xmath198 and consequently @xmath199 which translates the state as @xmath200 , we obtain @xmath201 from ( [ ss1 ] ) and ( [ qq ] ) .
therefore , the spin - nematic tensor @xmath202 defined on the bond @xmath203 satisfies @xmath204 .
this implies that the spinor condensate @xmath205 of our quintet boson has the same ( @xmath185-wave ) symmetry as the spin - nematic state discussed in ref ..
however , this is not the end of the story . since the local spin operator with @xmath206 assumes several different states ( e.g. _ polarized _ , _ nematic _ , etc . ) and it is not obvious if latexmath:[$\langle\{\theta_{{\mathbf{r}}}\},\{\psi_{{\mathbf{r}}}\}|q_{ij}^{\alpha\beta } for our @xmath20-@xmath21 model . to determine the actual value of @xmath208 .
since we are considering the situation where the gap between the singlet ground state and the quintet excitation is vanishingly small , it would be legitimate to keep only the singlet @xmath209 and the quintet for each plaquette to write down the effective hamiltonian .
the form of the effective hamiltonian is determined by using the second - order perturbation theory and it contains the kinetic part describing the hopping of the quintet particles and the magnetic part which concerns the interactions among them . since within a mean - field treatment the spinor part @xmath195 is determined by the magnetic interactions , it suffices to consider only the magnetic part of the effective hamiltonian : @xmath210 , \end{split } \label{hqumean}\ ] ] where @xmath211 denotes the @xmath212 spin operator , and the symbols @xmath213 and @xmath214 mean the nearest - neighbor- and the next - nearest - neighbor pairs , respectively . for different types of three - plaquette clusters
@xmath215 , we assign different three - body ( i.e. three - plaquette ) interactions @xmath216 ( @xmath217 ) in ( [ hqumean ] ) . the correspondence between six types of clusters and the strength of the three - plaquette interaction @xmath216 is shown in fig .
[ fig:3point ] .
the full expressions of @xmath218 , @xmath219 and @xmath216 are given in appendix .
[ sec : ap2 ] . note that our effective hamiltonian in its full form contains the kinetic term and charge interactions as well as magnetic ones @xmath220 . in this sense ,
our effective model is a generalization of the bose hubbard hamiltonian for @xmath221 cold atoms in optical lattices@xcite and the determination of the full phase diagram and the identification of various phases found in systems of cold atoms in our magnetic system would be interesting in its own right . ) .
the plaquette corresponding to @xmath222 is always located on the center of the clusters .
we identify all clusters obtained from a given one by rotation by @xmath223 and reflection .
[ fig:3point ] ] we investigate this hamiltonian by means of a mean - field theory by assuming an @xmath224-independent uniform @xmath225 , for simplicity .
since the parametrization of the spin-2 states is cumbersome , we adopt the method used by bacry@xcite and barnett _
first we note that arbitrary ( normalized ) spin-@xmath15 states are parametrized by a set of @xmath226 unit vectors except for obvious gauge redundancy . using rotational symmetry
, we can further reduce the number of free parameters needed to express arbitrary spin-2 states to @xmath227 ( see appendix . [ classify ] ) .
we numerically minimized the mean - field energy with respect to these five parameters . the result
is shown in fig .
[ fig : nf ] . at @xmath228 ,
the system is in the ferromagnetic state for @xmath229 and is in the spin - nematic state for @xmath230 .
this result slightly differs from the numerical results@xcite @xmath231 .
however , this is not surprising since our results are based on a mean - field treatment of the magnetic hamiltonian @xmath232 obtained by perturbation expansion in @xmath41 . our result may be improved by taking the number of sublattice larger , since @xmath233 and there are various 3-site interactions @xmath234 and @xmath235 . .
we zoom up the region around @xmath128 in fig . [
fig : souzug ] . in the two regions on the left ( green and yellow ) , the quintet and the singlet condense and we determined the resulting magnetic orders by a mean - field approximation to the magnetic hamiltonian @xmath232 . in the green region , the quintet and the singlet condense , and the spin - nematic phase appears . in the red , on the other hand ,
conventional ferromagnetic order is stabilized .
the red and the blue region are the same as fig .
[ fig : souzug ] .
[ fig : nf ] ]
having mapped out the phase diagram in the absence of external magnetic field , we consider next the magnetization process of the plaquette model by mapping the original model onto a hardcore boson model or an equivalent @xmath4 pseudo - spin model .
tachiki and yamada@xcite applied this method to obtain the magnetization curve of the spin - dimer model , which consists of pairs of @xmath4 spins .
the coupling to the external magnetic field is incorporated into the hamiltonian by adding the zeeman term @xmath236 . for convenience
, we set @xmath237 and assume that @xmath238 is pointing the @xmath239-direction : @xmath240 .
although the original treatment in ref . is for a coupled dimer systems , we can readily generalize the method to our plaquette system as follows .
we denote the plaquette states by @xmath241 , where @xmath242 are defined in ( @xmath243 ) . from ( [ sec2:ee ] ) , the energies of a single plaquette satisfy @xmath244 for @xmath245 . as is shown in fig .
[ fig : jikas ] , with increasing the magnetic field , the quintet level @xmath246 comes down to @xmath56 faster than the lowest triplet levels @xmath247 and @xmath248 . .
] therefore , in order to describe the low - energy physics in the presence of strong magnetic field ( @xmath249 ) , we may keep only the two lowest - lying states @xmath56 and @xmath246 for each plaquette and restrict ourselves to the subspace spanned by them . in
what follows , we regard the singlet @xmath56 and the quintet @xmath246 respectively as the up- and the down state of a pseudo spin-1/2 .
that is , @xmath250 then , the resulting effective hamiltonian is written in terms of the pauli matrices ( @xmath4 spins ) defined on each strongly - coupled plaquette . note that the approximation to treat only the subspace spanned by @xmath56 and @xmath246 probably breaks down for @xmath251 where all the components ( @xmath252 ) of the quintet come into play .
also the validity of the approximation may be questionable for sufficiently large @xmath41 where the singlet - triplet gap may be much smaller than the singlet - quintet gap , since the triplet states @xmath253 and @xmath254 are important there ( see fig .
[ fig : souzug ] ) . if we simply project the original @xmath4 hamiltonian to the restricted subspace as in ( [ psgs ] ) , no spin - flipping term ( or , hopping term , in terms of hardcore bosons ) appears .
this is because the projection is equivalent to the ordinary first - order perturbation theory and no transition between the singlet and the quintet occurs in the first - order processes .
therefore , we need take into account the second - order processes to obtain the meaningful effective hamiltonian .
the amplitude that a quintet state ( spin ` down ' ) @xmath246 hops to the adjacent plaquette is given by @xmath255 the hopping to the next nearest - neighbor does not occur at this order of approximation .
the energy gap between the state where there exists only one static ` down ' spin ( @xmath246 ) in a background of the ` up ' spins ( singlet @xmath56 plaquettes ) and the one where all plaquettes are ` up ' is given by @xmath256 in ( [ ap2:mu ] ) .
the interaction between the two adjacent ` up ' spins ( @xmath246 ) is given by @xmath257 in ( [ ap2:je1 ] ) and that between the next - nearest - neighbor pair is given by @xmath258 in ( [ ap2:je2 ] ) .
we note that this approximation becomes poor near the pole of @xmath259 and @xmath260 at @xmath169 . on top of them
, we have several three-`site ' processes and putting them all together , we obtain the effective hamiltonian : @xmath261 where @xmath262s denote the pauli matrices and @xmath263 .
the symbols @xmath213 and @xmath214 mean that the summation is taken over the nearest - neighbor- and the next - nearest - neighbor plaquettes , respectively . as in section [ sec : phase ] , there are six types of @xmath264 for different bond configurations @xmath215 ( see fig .
[ fig:3point ] ) .
we label the different three - plaquette interactions by @xmath265 @xmath266 and the corresponding bond - configurations are shown in fig .
[ fig:3point ] .
the concrete expressions of @xmath267 and @xmath268 are given in appendix .
[ sec : ap2 ] .
we note that the transverse components @xmath269 and @xmath270 can be translated to the creation- @xmath271 and the annihilation @xmath272 operator of a hardcore boson , respectively .
we analyze the hamiltonian ( [ heff ] ) within a mean - field approximation . since @xmath259 , which have the first order contributions in @xmath41 , are dominant for small @xmath41 , we may assume two different two - sublattice structures : ( i ) `` checkerboard '' and ( ii ) `` stripe '' shown in fig .
[ fig : sublattice ] in the calculation .
.[fig : sublattice ] ] by using the relations @xmath273 we can rewrite ( [ heff ] ) in terms of @xmath274 . since we are interested in the ground state energy at @xmath32 , we can simply replace operators in ( [ heff ] ) by their expectation values on each site , e.g. @xmath275 for the `` checkerboard '' case . for convenience ,
we introduce the following two - component vector : @xmath276 since there is rotational symmetry in the @xmath277-@xmath278 plane , the mean - field energy is parametrized by @xmath279 , @xmath280 , @xmath281 , @xmath282 and the angle @xmath283 between @xmath284 and @xmath285 .
the hamiltonian ( [ heff ] ) reduces to @xmath286 \end{split } \label{eeff}\ ] ] where @xmath287 denotes the total number of plaquettes and @xmath288 are given in appendix [ sec : ap2 ] both for the case of `` checkerboard '' and for the `` striped '' case .
correspondingly , the total magnetization is given simply as @xmath289 in both cases , @xmath290 and @xmath291 is minimized for @xmath292 . since any spin-1/2 states satisfy the following relation among the expectation values ( see eq .
( [ sigmai ] ) ) @xmath293 the transverse magnetization @xmath294 can be expressed in terms of the longitudinal one @xmath279 .
hence , there remain two variational parameters @xmath279 and @xmath280 in @xmath295 . from the definition ( [ eqn : def - pseudospin ] ) , the expectation values @xmath296 and @xmath297 respectively correspond to the singlet state and the fully polarized ( or , saturated ) state .
the critical field @xmath298 which marks the onset of magnetization is given by @xmath299 after substituting @xmath300 , i.e. @xmath301 the right - hand side is exactly the same as ( [ quingap ] )
. once spin - gap closes at @xmath298 , the quintet particle @xmath246 condenses , i.e. @xmath302 , @xmath303 . if @xmath304 , @xmath305 and there exists a finite expectation value of @xmath306 . in the hardcore boson language discussed below ( [ heff ] ) , @xmath307 can be viewed as the boson annihilation operator @xmath271 and its finite expectation value @xmath308 implies that bose - einstein condensation of the quintet particle occurs . in particular , if @xmath309 and @xmath310 in bec phase , the state is in the so - called `` supersolid '' phase@xcite . for convenience
, we shall call the bec phase satisfying @xmath311 a _ normal bec_. it should be noted that even when @xmath305 , the transverse magnetization @xmath312 vanishes unlike the bec in the spin - dimer model@xcite . in fact , since the creation operator @xmath272 of the quintet particle can be written in terms of the original spin operators as @xmath313 the existence of the condensate @xmath314 ( or @xmath315 ) implies that we have a finite expectation value of the following spin - nematic operator : @xmath316 the form ( [ eqn : quintet - creation ] ) of the quintet creation operator suggests that we should think of the plaquette quintet @xmath317 as a tightly - bound magnon pair ( or _ magnon molecule _ ) . the critical field @xmath28 where the saturation occurs is given by @xmath318 after substituting @xmath300 , i.e. @xmath319 we minimized @xmath320 numerically and we found that the energy in the `` stripe '' case was always equal to or smaller than that in the `` checkerboard '' case . we show various types of magnetization curves obtained in this way in fig . [
fig : magcurve ] . in fig .
[ fig : magsouzu ] , we also classified the parameter regions ( in the @xmath321-plane ) according to the qualitative behavior of the magnetization curve
. there appears ( i ) the normal bec phase , ( ii ) the `` striped '' supersolid phase and ( iii ) the `` striped '' 1/2-plateau . at the 1/2-plateau ,
the pseudo - spins @xmath322 are ordered in a collinear manner @xmath323 and @xmath324 ( see fig .
[ fig : sublattice ] ) . .
the phase boundary is only schematic.[fig : magsouzu ] ] .
the frustration parameters is fixed to @xmath325 .
the colors of the curves correspond to those used in fig .
[ fig : magsouzu ] ( except for yellow ) .
the blue curve has supersolid phase around the 1/2-plateau and the phase transition between the normal bec and the supersolid phase is of second - order .
all curves in bec and supersolid phase is convex down because of the 3-point interaction @xmath326 in ( [ eeff ] ) .
[ fig : magcurve ] ] the magnetization curve in the bec and the supersolid phase is convex down because of 3-point interaction @xmath326 in ( [ eeff ] ) which breaks the particle - hole symmetry .
the `` striped '' supersolid phase always appears around the 1/2-plateau and the width of the supersolid phase appearing on the left of the 1/2 plateau is broader than that on the right due to the convex down character .
the equivalent hamiltonian ( [ heff ] ) without the 3-point interactions has been investigated by using the mean field theory@xcite and monte - carlo simulations@xcite .
they found that the `` striped '' supersolid phase around the 1/2-plateau is stable@xcite .
therefore , our result that the supersolid phase exists may be correct beyond the mean - field approximation , since the 3-point interaction in ( [ heff ] ) is weak .
there are other models accompanied by the supersolid phase , e.g. spin dimer xxz model@xcite , spin-1/2 xxz model on the triangular lattice@xcite , etc .
in this section , we compare our results with the experimental data obtained for @xmath3 . since we have three parameters @xmath0,@xmath1 and @xmath41 , three experimental inputs in principle determine the set of coupling constants .
then , we use those values of coupling constants to compare the magnetization curve of our model with the experimental one@xcite .
we use the triplet gap @xmath328 observed in inelastic neutron scattering@xcite , the lower critical field @xmath329 ( or @xmath330 if @xmath37 is used ) , which marks the onset of magnetization , and the saturation field@xcite @xmath331 @xmath332 as the experimental input .
the triplet gap has been calculated in sec .
[ sec3 ] and are given by eq.([ap2:tmin ] ) or ( [ ap2:tmind ] ) . in sec .
[ sec : magnet ] , we have obtained the critical field @xmath31 ( eq.([tatiagari ] ) ) and @xmath28 ( eq.([hc2 ] ) ) .
we compare these results with the experimental ones to determine two exchange couplings @xmath20 , @xmath21 and the distortion parameter @xmath41 .
the result is : @xmath333 where we have used ( [ ap2:tmind ] ) for the excitation energy of the triplet .
the magnetization curve for the ratio @xmath334 and the distortion @xmath335 obtained above is shown in fig .
[ fig : cucl ] ( see fig . [
fig : magsouzu ] ) .
this curve is similar to that obtained in the high - field magnetization measurement@xcite except for the little convex down character . ) by using the parameter set ( [ cuclpara ] ) .
[ fig : cucl ] ] however , a remark is in order here .
recent nmr experiments@xcite suggest the displacement patterns of @xmath24 which yield different magnetic interactions from what have been assumed here . in particular , the system does not have any explicitly tetramerized structure ( see fig.[fig : plafull ] ) , although @xmath3 has period 2 both in the @xmath271- and the @xmath336 direction .
therefore , our results should not be taken literally . instead
, our plaquette model should be thought of as one of the simplest hamiltonians realizing the bec of magnon bound states which is applicable to much wider class of systems including our simple @xmath20-@xmath21 model .
motivated by the recent discovery of a new two - dimensional spin - gap compound @xmath3 , we have studied spin-1/2 @xmath0-@xmath1 model with a plaquette structure . for the small inter - plaquette interactions , i.e. , for small @xmath41
, there exists a finite spin gap over the spin - singlet ground state .
we have computed the excitation energy of the triplets and the quintet in section [ sec3 ] in two different methods . if the gap of the lowest excitation closes , the corresponding particle condenses and a phase transition occurs from a paramagnetic phase to magnetically ordered phases . for the case of ferromagnetic @xmath20 considered here , we have two possibilities . for relatively small @xmath337 , the triplet particles ( @xmath338 and @xmath339 ) condense and generically we may expect caf appears after the condensation ( see fig . [
fig : souzug ] ) . for larger values of @xmath337 , however , the quintet excitation matters and we may have various phases .
in the situation of relevance , we have either a usual ferromagnetic phase or a less conventional spin - nematic phase . one of these phases is selected by magnetic interaction among the quintet particles .
we have derived an effective hamiltonian governing the magnetic part by using the second - order perturbation and mapped out the magnetic phase diagram ( see fig .
[ fig : nf ] ) .
a mean - field calculation predicted a finite window of the spin - nematic phase ( green region in fig .
[ fig : nf ] ) in agreement with recent numerical results@xcite obtained for @xmath228 . from the properties of the condensing particle
, we found the nematic order for @xmath340 in the homogeneous ( @xmath228 ) @xmath0-@xmath1 model in section [ sec : phase ] .
we remark that our effective hamiltonian is closely related to that for @xmath221 cold atoms in optical lattices @xcite .
we have studied the magnetization process in section [ sec : magnet ] . in the region of interest
, magnetization is carried by spin-2 particles , which should be identified with a tightly - bound magnon pair ( _ magnon molecule_),and we have constructed an effective hardcore boson ( or , pseudo spin-1/2 ) model for these spin-2 particles . by using a mean - field ansatz , we have determined the ground state of the above effective hamiltonian as a function of the external field @xmath341 .
we have found three different phases : ( i ) the normal bec phase , ( ii ) the ` striped ' supersolid phase and ( iii ) the ` striped ' 1/2-plateau . in the normal bec phase , the transverse magnetization @xmath312 vanishes unlike the conventional bec in the spin - dimer model@xcite .
we have compared the results obtained for our @xmath20-@xmath21 model with the experimental data of @xmath3 in section [ sec : cucl ] . although we have found that our model could qualitatively explain the magnon gap in the inelastic neutron scattering experiments@xcite and the magnetization curve@xcite , the structure suggested by nmr measurements@xcite is inconsistent with our tetramerized @xmath20-@xmath21 model and this agreement
should not be taken literally .
nevertheless , we hope that our scenario ` molecular spin - bec ' based on a simple @xmath20-@xmath21 model will capture the basic physics which underlies the magnetism of the compound @xmath3 .
we would like to thank h. kageyama for sharing his unpublished results , many helpful discussions , and comments on the manuscript .
we are also grateful to a. kitada and t. miki for many useful discussions , and mike zhitomirsky for careful reading of the manuscript .
this work is supported by the grant - in - aid for the 21st century coe of education , culture , sports , science and technology ( mext ) of japan .
for convenience , we briefly summarize the method of bogoliubov transformation .
we want to diagonalize @xmath342 where @xmath343 now @xmath344 are boson operators , and @xmath345 .
we introduce bogoliubov transformation @xmath346 where @xmath347 is @xmath110 real matrix , and @xmath348 @xmath349 & = [ l_{1\mu}v_\mu , l_{1\nu } v_\nu^{\dagger } ] = l_{11}^2 + l_{12}^2 -l_{13}^2 -l_{14}^2 = 1\ , \\ [ p^{\prime},q^{\prime\dagger } ] & = l_{11}l_{21 } + l_{12}l_{22 } -l_{13}l_{23 } -l_{14}l_{24 } = 0\ , \\ [ p^{\prime\dagger},p^{\prime } ] & = l_{31}^2 + l_{32}^2 -l_{33}^2 -l_{34}^2 = -1\ , \\ & \vdots \end{split } \label{ap1:cd2}\ ] ] where the summation over repeated indices is implied . by using @xmath350 and @xmath351 the condition ( [ ap1:cd2 ] )
can be rewritten as @xmath352 where @xmath353 is a 4-dimensional column vector and the summation over @xmath132 is not taken in ( [ beq ] ) .
@xmath354 can be considered as a metric .
the condition ( [ ap1:cd1 ] ) can be rewritten as @xmath355 where @xmath356 and @xmath357 is a 2-dimensional column vector .
we denote @xmath358 in ( [ ap1:a ] ) as @xmath359 and @xmath360 as @xmath361 .
then , with regard to eigenvectors satisfying @xmath362 , we obtain @xmath363 since @xmath359 is symmetric . therefore , if @xmath364 , @xmath365 , i.e. eigenvectors of a different eigenvalue are orthogonal each other .
we define @xmath366 where * l*@xmath367 are eigenvectors of @xmath368 , and @xmath369 then , @xmath370 .
therefore , @xmath371 where @xmath372 is identity matrix .
now we can write @xmath373 where @xmath374 are @xmath375 matrix , and @xmath345 .
therefore , we can take * l*@xmath367 satisfying ( [ lkkai ] ) . defining @xmath376 as the eigenvalue of * l*@xmath367 , this leads to @xmath377 from ( [ ap1:koutou ] ) , ( [ ap1:1 ] ) reduces to @xmath378 where @xmath379 , and @xmath380 is omitted . comparing to ( [ ap1:l ] )
, we obtain @xmath381 this @xmath347 satisfies ( [ beq ] ) and ( [ lkkai ] ) .
the eigenvalues of @xmath382 ( see eq .
( [ ap1:a ] ) ) are given by @xmath383^{\frac{1}{2}}\ .
\end{split}\ ] ] we note that + @xmath384 are the eigenvalues of the eigenvectors whose norm is positive , + @xmath385 are the eigenvalues of the eigenvectors whose norm is negative , + and ( [ omegar ] ) is satisfied . moreover , using the boson s commutation relation , @xmath386 reduces to @xmath387 where @xmath388 .
in this section , we briefly summarize the method of parametrizing arbitrary spin-2 states used in a mean - field calculation of section [ sec : phase ] .
the method is based on a geometrical representation of the spin-@xmath15 states used by bacry@xcite and barnett _
et al._@xcite .
since our model has rotational symmetry , the mean - field energy has a trivial degeneracy with respect to the global rotation of the spin states .
to mod out this degeneracy and find only essentially different solutions , this geometric method is quite efficient .
first we introduce the maximally polarized spin-@xmath15 state ( _ spin coherent state _ ) @xmath389 which is pointing the direction of @xmath390 i.e. @xmath391 . if we introduce the schwinger boson operators @xmath392 ( @xmath393 ) which destroys a spin parallel ( anti - parallel ) to the @xmath239-direction , the operator which creates a spin parallel to the @xmath394-direction is given by @xmath395 where @xmath396 and @xmath397 is an arbitrary gauge function . by using @xmath398 , the coherent state @xmath389 can be written simply as @xmath399 where the combinatorial symbol @xmath400 is defined by @xmath401 .
next , we introduce a complex number @xmath402 and the corresponding unnormalized ket @xmath403 : @xmath404 we note that the vector @xmath394 rotates on the unit sphere @xmath405 , when the su(2 ) rotation operator @xmath406 acts on @xmath403 .
we denote an arbitrary spin-@xmath15 state by @xmath407 .
then , it is convenient to introduce the following ` wave function ' which is in a one - to - one ( except for an unphysical overall phase factor ) correspondence with @xmath408 under the condition @xmath409 : @xmath410 where @xmath411s are the @xmath226 roots of @xmath412 and are parametrized as @xmath413 .
if the degree @xmath414 of the above polynomial is smaller than @xmath226 , @xmath415 roots of @xmath416 are at the infinity ( @xmath417 or the north pole ) .
since the stereographic projection uniquely maps a set of @xmath226 complex roots @xmath418 onto a set of @xmath226 points on a two - dimensional sphere @xmath405 , we can parametrize arbitrary spin-@xmath15 states by specifying @xmath226 points on a sphere . if @xmath419 , the limit @xmath420 for any @xmath421 must be taken ( @xmath422 ) . in the case of spin-2 , @xmath423 is given in terms of four complex numbers @xmath418 by @xmath424 and hence the coefficients @xmath425 read @xmath426 where @xmath283 is the phase of @xmath427 .
therefore , as has been described above , arbitrary spin-2 states are parametrized by a set of four unit vectors and an overall phase factor .
the rotational symmetry enables us to further reduce the number of free parameters by fixing @xmath428 and @xmath429 as : @xmath430 equations ( [ eqn : abyalpha-1])-([eqn : abyalpha-3 ] ) express arbitrary ( except for global rotation ) spin-2 states in terms of five free parameters .
there exists a simple relation among the expectation values of spin-@xmath15 operators . by spin-@xmath15 operators , here we mean all independent ( traceless ) polynomials made up of the usual spin-@xmath15 operators @xmath431 .
the spin 1 case has been considered by chen and levy@xcite in the context of spin - nematic order .
an arbitrary spin-@xmath15 ket is written as @xmath432 where @xmath433 .
it is convenient to consider the lie group @xmath434 which naturally acts on the above @xmath435-dimensional space .
let us denote the generators @xmath436 @xmath437 of @xmath438 and normalize them as @xmath439 then , they satisfy @xmath440 using this relation , we obtain @xmath441 where the summation over repeated indices is implied . in spin-1/2 ( @xmath442 )
case , @xmath436 can be written as @xmath443 , where @xmath322 is pauli matrix .
therefore , this relation can be written as @xmath444
: + the elements of the second - order hopping matrix ( eq .
( [ acb ] ) ) are given by : @xmath445 @xmath446 @xmath447 the second - order energy shift for the singlet ground state is calculated as(see eq . ( [ etpm ] ) ) : @xmath448 the excitation gap of triplets from the second - order perturbation is given by : @xmath449 the modified excitation gap of triplets which is free from the pole @xmath450 is given by : @xmath451 the excitation energy of quintet from the second - order perturbation is given by : @xmath452 : + the parameters of the effective hamiltonian ( [ hqumean ] ) where the quintet condenses are given by : @xmath453 @xmath454 @xmath455 @xmath456 @xmath457 : + the parameters of the effective hamiltonian @xmath458 ( [ heff ] ) in the magnetization process are given in a series in @xmath41 by : 101 h.t .
diep , _ frustrated spin systems _ , edited by h.t . diep ( world scientific , singapore , 2004 ) .
p. chandra and b. doucot , phys.rev .
b * 38 * , 9335 ( 1988 ) .
schulz , t.a.l .
ziman and d. poilblanc , j.phys .
i * 6 * 675 ( 1996 ) .
l. capriotti , f. becca , a. parola , and s. sorella , phys.rev.lett .
* 87 * , 097201 ( 2001 ) .
singh , z. weihong , c.j .
hamer , and j. oitmaa , phys .
b * 60 * , 7278 ( 1999 ) . v. kotov , j. oitmaa , o. sushkov and z. weihong , phil . mag .
b * 80 * , 1483 ( 2000 ) .
g. misguich and c. lhuillier , in ref . .
j. oitmaa , c. hamer and w. zheng , _ series expansion methods for strongly interacting lattice models _
( cambridge university press , uk , 2006 ) .
n. shannon , b. schmidt , k. penc and p. thalmeier , eur .
j. b * 38 * , 599 ( 2004 ) .
n. shannon , t. momoi and p. sindzingre , phys .
lett . * 96 * , 027213 ( 2006 ) .
r. melzi , p. carretta , a. lascialfari , m. mambrini , m. troyer , p. millet , and f. mila , phys .
rev . lett . * 85 * , 1318 ( 2000 ) .
kaul , h. rosner , n. shannon , r.v .
shapanchenko , and c. geibel , j. mag . mag . mat . * 272 - 276 * , 922 ( 2004 ) .
h. kageyama , t. kitano , n. oba , m. nishi , s. nagai , k. hirota , l. viciu , j.b .
wiley , j. yasuda , y. baba , y. ajiro and k. yoshimura , j. phys .
. jpn . * 74 * , 1702 ( 2005 ) .
h. kageyama , j. yasuda , t. kitano , k. totsuka , y. narumi , m. hagiwara , k. kindo , y. baba , n. oba , y. ajiro and k. yoshimura , j. phys .
. jpn . * 74 * , 3155 ( 2005 ) .
t. nikuni , m. oshikawa , a. oosawa and h. tanaka , phys .
rev . lett .
* 84 * , 5868 ( 2000 ) .
h. tanaka , a. oosawa , t. kato , h. uekusa , y. ohashi , k. kakurai and a. hoser , j. phys .
jpn . * 70 * , 939 ( 2001 ) .
m. jaime _ et al .
* 93 * , 087203 ( 2004 ) ; s.e .
sebastian _ et al .
b * 72 * , 100404 ( 2005 ) . m.b
. stone , c. broholm , d.h .
reich , o. tchernyshyov , p. vorderwisch and n. harrison , phys .
lett . * 96 * , 257203 ( 2006 ) .
a. kitada , z. hiroi , y. tsujimoto , t. kitano , h. kageyama , y. ajiro and k. yoshimura , j. phys .
jpn . * 76 * , 093706 ( 2007 )
. m. yoshida _ et al .
_ , [ arxiv : cond - mat/07063559 ] ; submitted to j. phys . soc
t. momoi and k. totsuka , phys .
b * 62 * , 15067 ( 2000 ) .
k. totsuka , s. miyahara and k. ueda , phys.rev.lett .
* 86 * , 520 ( 2001 ) . s. miyahara and k. ueda , j.phys . : condens .
matter * 15 * , r327 ( 2003 ) .
h. nojiri , h. kageyama , y. ueda and m. motokawa , j.phys.soc.jpn .
* 72 * , 3243 ( 2003 ) .
r. bendjama , b. kumar and f. mila , phys .
lett . * 95 * , 110406 ( 2005 ) .
s. sachdev and r. n. bhatt , phys .
b * 41 * , 9323 ( 1990 ) .
sakurai , _ modern quantum mechanics , ch.3 _ ( benjamin / cummings publishing company , inc . 1985 )
. h. bacry , j.math.phys . *
15 * , 1686 ( 1974 ) .
r. barnett , a. turner and e. demler , phys.rev.lett .
* 97 * , 180412 ( 2006 ) .
fei zhou and g.w .
semenoff , phys.rev.lett . * 97 * , 180411 ( 2006 ) .
m. lewenstein , a. sanpera , v. ahufinger , b. damski , a. sen and u. sen , adv.phys .
* 56 * , 243 ( 2007 ) . m. tachiki and t. yamada , j. phys .
jpn , @xmath463 , 1413 ( 1970 ) .
zhitomirsky and k. ueda , phys .
b @xmath464 , 9007 ( 1996 ) .
a.f . andreev and i.m .
lifshitz , jetp , * 29 * 1107 , ( 1969 ) ; a.j . leggett , phys
lett . * 30 * , 1543 ( 1970 ) ; k - s .
liu and m.e .
fisher , j. low . temp .
* 10 * , 655 ( 1973 ) .
scalettar , g.g .
batrouni , a.p .
kampf and g.t .
zimanyi , phys .
b , * 51 * , 8467 ( 1995 ) .
batrouni and r.t .
scalettar , phys .
lett , * 84 * , 1599 ( 2000 ) ; f. h@xmath465bert , g.g .
batrouni , r.t .
scalettar , g. schmid , m. troyer and a. dorneich , phys .
b , * 65 * , 014513 ( 2001 ) .
ng and t.k .
lee , phys .
lett , * 97 * , 127204 ( 2006 ) ; n. laflorencie and f. mila , phys .
lett , * 99 * , 027202 ( 2007 ) .
g. murthy _ et al .
_ , phys . rev
. b , * 55 * , 3104 ( 1997 ) ; m. boninsegni , j. low temp . phys . * 132 * , 39 ( 2003 ) ; s. wessel _ et al .
lett . * 95 * , 127205 ( 2005 ) ; d. heidarian _ et al .
_ , _ ibid _ * 95 * , 127206 ( 2005 ) ; r. g. melko _ et al .
_ , _ ibid _ * 95 * , 127207 ( 2005 ) ; m. boninsegni _ et al .
_ , _ ibid _ * 95 * , 237204 ( 2005 ) . h.h .
chen and p.m. levy , phys .
b , * 7 * , 4267 ( 1973 ) . | we study the ground state and the magnetization process of a spin-1/2 @xmath0-@xmath1 model with a plaquette structure by using various methods . for small inter - plaquette interaction ,
this model is expected to have a spin - gap and we computed the first- and the second excitation energies . if the gap of the lowest excitation closes , the corresponding particle condenses to form magnetic orders . by analyzing the quintet gap and magnetic interactions among the quintet excitations ,
we find a spin - nematic phase around @xmath2 due to the strong frustration and the quantum effect .
when high magnetic moment is applied , not the spin-1 excitations but the spin-2 ones soften and dictate the magnetization process .
we apply a mean - field approximation to the effective hamiltonian to find three different types of phases ( a conventional bec phase , `` striped '' supersolid phases and a 1/2-plateau ) . unlike the bec in spin - dimer systems
, this bec phase is not accompanied by transverse magnetization .
possible connection to the recently discovered spin - gap compound @xmath3 is discussed . |
gamma - ray bursts ( grbs ) are the brightest explosions in the universe .
they release energy up to @xmath2 , which is on the order of @xmath3 , in the form of gamma - rays within a few seconds . from the spectral features and light curves of their prompt and afterglow emission , they are considered to be produced in ultrarelativistic jets .
although the picture of the central engine has not yet been established , several constraints are required from the observations .
it should be able to launch ultrarelativistic jets , meaning that relativistic objects should be involved .
it should have two different characteristic timescales : the variability timescale of their prompt emission ( @xmath4 , e.g. , @xcite ) and the duration timescale of the prompt emission ( @xmath5 ) .
the latter timescale is much longer than the dynamical timescale of relativistic objects , but there is a way to explain all the requirements . the most promising model of the grb central engine is a hyperaccretion flow @xcite . in this model
an ultrarelativistic jet would be launched from a massive accretion disk around a stellar - mass black hole whose mass accretion rate is at most @xmath6 .
such a system is expected to form after a cataclysmic event such as the gravitational collapse of a massive star or the merger of a neutron star - neutron star binary or a black hole - neutron star binary @xcite .
since such an accretion flow is extremely optically thick with respect to photons , it can not cool via electromagnetic radiation . instead ,
since the density and temperature of this accretion flow are very high ( @xmath7 , @xmath8 ) , it would cool via neutrinos , and such an accretion flow is often called as `` neutrino - dominated accretion flows '' ( ndaf ; @xcite ; see also section 10.6 of @xcite ) . in this model , the variability timescale can be interpreted as the dynamical ( i.e. , rotation ) timescale of an accretion disk , while the burst duration timescale corresponds to the accretion timescale . as for how a hyperaccretion flow can produce an ultrarelativistic jet , two physical processes
are mainly discussed in the literature : neutrino pair annihilation ( @xmath9 ; @xcite ) and magnetohydrodynamic ( mhd ) mechanisms such as the blandford - znajek ( bz ) process ( @xcite ) . in either scenario , it is implied that the energy deposition rate ( corresponding to the jet luminosity ) as a function of mass accretion rate has a step - function - like behavior at a certain critical mass accretion rate .
according to the neutrino pair annihilation scenario , in fact , we expect a big jump in the neutrino emission efficiency across the critical accretion rate , @xmath10 for fiducial parameters ( see @xcite ) .
that is , when the mass accretion rate is above @xmath11 the accretion flow becomes an efficient neutrino emitter ( `` ndaf '' regime ) and the energy deposition via neutrino annihilation is efficient , while when @xmath1 is below @xmath11 the accretion flow is in the `` advection - dominated '' ( adaf ) regime and neutrino pair annihilation emission is not at all efficient . similar behavior is expected in the mhd scenario , as well .
@xcite investigated the mhd jet luminosity predicted from a hyperaccretion flow and found that it shows step - function - like behavior at @xmath12 . in either case , therefore , one can expect a drastic change of jet luminosity from a hyperaccretion flow when its mass accretion rate attains the critical value @xmath13 .
if there is any physical mechanism which can modulate the mass accretion rate around @xmath11 , the jet luminosity would change violently , which may explain to the highly variable grb emission . theoretically , several types of disk instabilities are known @xcite , and the large - amplitude luminosity variations of accreting systems , such as those observed in dwarf novae and x - ray binaries are often ascribed to the accretion disk instabilities ( e.g. , @xcite , see a concise review by @xcite chapter 5 and 10 ) .
however , little was known about the disk instabilities of a hyperaccretion flow , especially that with mass accretion rate around @xmath11 .
recently , @xcite found that the thermal equilibrium curve at the innermost region of a hyperaccretion flow on the @xmath0-@xmath1 plane has the structure like a character `` n '' , which has a negative slope at @xmath12 .
this means the existence of a secular ( or viscous ) instability of the accretion flow . in our study , we for the first time solve the time evolution of the unstable hyperaccretion flow and show that the luminosity evolution occurs intermittently as a consequence of a local transition which takes place in a narrow region .
the plan of this paper is as follows : in section 2 , we describe the equations for disk evolution and the modeled n - shaped thermal equilibrium curve . in section 3
we show the results of our calculations with an emphasis on the consequence of the instability .
section 4 is devoted to the discussion on the dependence on the n - curve , the observable behavior , the comparison with the case of s - shaped curve , and future issues .
in the present study , we solve the height - averaged basic equations for viscous diffusion of an axisymmetric accretion disk ( see , e.g. , @xcite chapter 3 ) : @xmath14 and @xmath15 = -2\pi \frac{\partial}{\partial r } \left(r^{3 } \nu \sigma \frac{d\omega}{dr } \right ) .
\label{basic2}\ ] ] here , @xmath16 ( with @xmath17 being mass density ) is the surface density , @xmath18 is the distance from the center of a black hole , @xmath19 ( with @xmath20 being the radial velocity ) , @xmath21 is the kinematic viscosity , and @xmath22 is the angular velocity . equation ( [ basic1 ] ) describes mass conservation , while equation ( [ basic2 ] ) represents angular momentum conservation . in our calculations ,
we adopt the pseudo - newtonian force for a kerr black hole ( artemova et al .
1996 ) for calculating @xmath22 in equation ( [ basic2 ] ) ; that is , from @xmath23 we find @xmath24 where @xmath25 is the black hole mass , @xmath26 is the radius of the event horizon , @xmath27 ( with @xmath28 being the radius of the marginally stable orbit ) is a constant to be specified later , and we assumed the balance between the gravitational force and the centrifugal force . inserting equation ( [ omega ] ) to equation ( [ basic2 ] ) , we obtain @xmath29 r^{\frac{\beta-1}{2 } } } \frac{\partial}{\partial r } \left[\frac{r^{\frac{\beta+1}{2}}(r-\frac{3-\beta}{3}r_{\rm h})}{(r - r_{\rm h})^{\frac{\beta+2}{2 } } } \nu \sigma \right ] .
\label{basic2-pk}\ ] ] we next transform equations ( [ basic1 ] ) and ( [ basic2-pk ] ) to non - dimensional forms . for this purpose
we introduce the following dimensionless variables : @xmath30 here , @xmath31 is the size of the innermost region of the disk , which we are concerned with , @xmath32 and @xmath33 represent typical surface density and mass accretion rate of ndaf , and @xmath34 corresponds to the viscous timescale at the outer edge of the disk
. these normalization constants will be specified later ( see section 2.2 ) .
we can then rewrite the basic equations with dimensionless variables as follows : @xmath35 @xmath36x^{\beta } } \frac{\partial}{\partial x } \left[\frac{x^{\beta+1}(x^{2}-\frac{3-\beta}{3}x_{\rm h}^{2})}{(x^{2}-x_{\rm h}^{2})^{\frac{\beta+2}{2 } } } \mu \right ] .
\label{basic2'}\ ] ] to solve the basic equations ( [ basic1 ] ) and ( [ basic2 ] ) , we need to prescribe the functional form of @xmath37
. this can be done by using the thermal equilibrium curve , on which heating and cooling are balanced .
the thermal equilibrium curve of the hyperaccretion flow has a characteristic n - shape , as is shown in the left panel of figure 1 taken from @xcite .
they adopted @xmath38 viscosity ;
@xmath39 with @xmath40 , where @xmath41 is the half thickness of the disk and @xmath42 is the speed of sound .
the vertical axis is @xmath1 ( mass accretion rate ) and the horizontal axis is @xmath0 ( surface density ) in the left panel of figure 1 .
we model this equilibrium curve in a simplified form as is shown in the right panel of figure 1 : for @xmath43 @xmath44 for @xmath45 @xmath46 for @xmath47 @xmath48 where @xmath49 and @xmath50 are the values of @xmath51 at the two critical points , ` a ' and ` b ' , respectively , where the slope of the thermal equilibrium curve changes ; @xmath52 likewise , @xmath53 and @xmath54 are the values of @xmath55 at the critical points , ` a ' and ` b ' , respectively , and their explicit forms are calculated by using equations ( [ n - curve - a ] ) and ( [ n - curve - b ] ) as functions of @xmath56 .
[ model - kawanaka ] ( 80 mm , 50mm)f1.eps [ n - model ] ( 80 mm , 50mm)mu - model.eps equations ( [ n - curve - a])([n - curve - c ] ) represent the lower , middle , and upper branches of the thermal equilibrium curve , respectively .
the principle for constructing this formula is as follows : 1 .
the value of the kinematic viscosity is constant ( i.e. , @xmath57 ) on the lower and upper branches , while @xmath58 on the middle branch in the innermost region . the bigger the radius is , the smaller is the slope of the middle branch .
the entire equilibrium curve up - shifts as @xmath56 increases . in other words ,
the bigger @xmath56 is , the bigger is the value of @xmath55 for a fixed @xmath51 .
the n - shape is more enhanced at smaller radii , whereas the unstable branch ( with a negative slope ) disappears at @xmath59 . in the present study , we set @xmath60 . here
, @xmath61 is the gravitational radius . by comparing figure 1(a ) with figure 1(b ) , we assign the following values to @xmath33 and @xmath62 : @xmath63 , \label{mdot0}\ ] ] @xmath64 , \label{sigma0}\ ] ] for the black hole mass of @xmath65 . inserting these normalization constants into equation ( 9 ) and simply assuming that @xmath66
, we find @xmath67 . \label{t0}\ ] ] as long as the system is in the thermal equilibrium , it evolves along the equilibrium curve .
it starts to deviate from the curve , however , once the instability takes place . in order to describe the disk evolution in the state out of equilibrium
, we introduce the following simple form of energy equation : @xmath68 where @xmath69 is the value of @xmath55 on the thermal equilibrium curve for a given @xmath51 , and @xmath70 represents the thermal timescale ; @xmath71 here , @xmath72 represents the diffusion timescale at radius @xmath18 and is shorter at smaller @xmath18 , and @xmath73 is a constant set to be @xmath74 in the present study .
there are two boundary conditions .
one is the choice of @xmath75 , mass injection rate at the outer edge of the disk ( the outer boundary condition ) and the other is the zero torque condition at the inner edge of the disk ( the inner boundary condition ) ;
@xmath76 in this study , we examine the following three cases for the way of mass injection from the outer boundary : @xmath77 @xmath78 @xmath79 note that the value of @xmath80 of models 1 and 2 ( or model 3 ) has been chosen in order that the whole disk can be on the lower ( upper ) branch ; i.e. , @xmath81 ( @xmath82 ) . also note that the rise timescale of model 1 is @xmath83 [ s ] [ see equation ( [ t0 ] ) ] .
the initial condition is a steady disk corresponding to @xmath80 . by integrating equation ( [ basic2 ] ) with respect to @xmath56 ,
we obtain @xmath84 x^{\beta+1 } } \left[\frac{x^{\beta+1}}{(x^{2}-x_{\rm h}^{2})^{\frac{\beta}{2 } } } - \frac{x_{\rm in}^{\beta+1}}{(x_{\rm in}^{2}-x_{\rm h}^{2})^{\frac{\beta}{2 } } } \right ] , \label{steady - mu}\ ] ] and the initial value of @xmath55 is calculated by inserting @xmath80 into the equation above .
we can then solve time - dependent equations ( [ n - curve - a])([n - curve - c ] ) to obtain surface density evolution at each @xmath56 .
actual numerical procedure is explained in appendix 1 .
we set the inner boundary at @xmath85 or @xmath86 .
the inner radius corresponds to the innermost stable circular orbit of a black hole with the dimensionless spin parameter @xmath87 of 0.6 , for which @xmath88 .
we finally get @xmath89 .
how does a transition from the lower branch to the upper branch proceed in the disk when the mass injection rate gradually increases ? to answer to this question , we first show in figures 2(a ) and 3(a ) a series of the surface - density distribution expected to be the steady disk model for several mass injection rates ; @xmath90 with @xmath91 , and @xmath92 in figure 2(a ) while @xmath93 in figure 3(a ) , respectively [ recall equation ( [ mdot - inj-1 ] ) ] .
we , here , assume that the disk is on the lower ( @xmath51 ) branch of the n - shaped curve as long as @xmath43 but that it is on the upper branch otherwise .
note that each curve in figure 3(a ) except for the lowest one is vertically offset by the value depending on the elapsed time . in these plots
the mass accretion rates are taken to be the same everywhere by the assumption that every portion of the disk can immediately adjust to any changes in @xmath75 given at the outer boundary .
as @xmath75 increases , @xmath94 at each radius gradually increases , so does the value of @xmath51 .
eventually a transition from the lower adaf branch to the upper ndaf branch takes place when @xmath95 .
we can see in figures 2(a ) and 3(a ) that the surface density rapidly increases at the radius where an upward transition occurs .
after the transition , neutrino cooling is efficient .
hence , neutrino is emitted from the high-@xmath51 regions .
the steady model is simple and easy to calculate , since it requires no numerical technique , however , there exist certainly limitations . for example , it should take time for the influence of varying mass injection rates to prevail over the disk plane , but such a time delay is totally neglected in the steady model .
also an upward transition can not occur in an infinitesimal time but a transition is assumed to be instantaneous in the steady model .
further , mass inflow into and/or outflow from the transition region , which should be necessary to make a local change in @xmath51 , are not properly taken into account .
we thus need to perform numerical simulations .
the first case is model 1 , in which the mass injection rate slowly increases .
figures 2(b ) and 3(b ) display the global evolution of the surface - density distribution in slightly different ways .
the expansion of the high-@xmath51 region , where neutrino emission is significant , is clear in these figures .
the upward transition to the upper ( ndaf ) branch is indicated by a rapid increase of @xmath51 .
[ steady - sigma ] ( 80 mm , 50mm)steady - sigma-1a.eps [ sigma - hikaku ] ( 80 mm , 50mm)sigma - hikaku-1b.eps [ steady1 ] ( 80 mm , 50mm)steady1-az.eps [ sigma1 ] ( 80 mm , 50mm)sigma1-bz.eps let us carefully examine to what extent the steady model depicted in figure 2(a ) reproduces the time - dependent calculations shown in figure 2(b ) . in the steady model ,
the mass accretion rate ( @xmath94 ) is everywhere equal to the mass injection rate ( @xmath75 ) ; in other words , the whole disk is assumed to immediately respond to a change of @xmath75 at the outer boundary .
another big assumption in the steady model is that a transition from one branch to another at a certain radius does not affect its environment . from these two panels ,
we notice that the time of an upward transition at a certain radius in the time - dependent model is delayed , compared with that at the same radius in the steady model .
this feature can also be confirmed by comparing figures 3(a ) and ( b ) and is obviously the effect of the finite accretion timescale .
we also find that the @xmath51 profile shows fluctuating patterns .
these fluctuating patterns are more clearly displayed in figure 3(b ) .
this is caused by rapid mass exchanges between the neighboring regions associated with branch transitions .
hereafter we call such fluctuations as ` non - steady ' effects
. obviously they are induced by ` non - steady ' mass accretion flow ( i.e. , locally enhanced mass flow ) .
we will discuss their physical meaning in details in the next subsection .
+ ( 80 mm , 50mm)nusigma78.eps in order to understand the non - steady effects , more careful analysis is necessary regarding the behavior of each part of the disk .
this can be done by using the @xmath51-@xmath55 plane at fixed radii .
a typical example is displayed in figure [ nusigma78 ] .
we see that the evolutionary path shown by the solid line grossly deviates from the thermal equilibrium curve indicated by the dashed line .
we can understand this fact in the following way .
suppose that the critical point a is reached at some radius .
although there are no longer low branch solutions above that point , @xmath51 should steadily increase , since mass is continuously supplied from the neighboring outer zone .
thus , the disk should evolve in the right direction on the @xmath51-@xmath55 plane .
this is exactly what we see in figure [ nusigma78 ] . after reaching the upper branch , however , unexpected behavior appears .
that is , the evolutionary path fluctuates around the equilibrium point , on the upper branch .
this is a result of complex mass exchanges between the neighboring sites . to understand how this occurs
, we make the following thought experiments . at first
we wish to note that the exact behavior of the disk on the @xmath51-@xmath55 plane depends critically on the ratio of the thermal timescale ( @xmath70 , on which @xmath55 changes ) to the viscous timescale ( @xmath72 , on which @xmath51 changes ) .
let us , hence , consider two hypothetical cases : case a transition in which @xmath96 as is expected in a standard - type disk ( see , e.g. , kato et al .
2008 , chapter 3 ) and case b transition in which @xmath97 .
( the actual simulation will be shown later . )
[ mangaa ] ( 150 mm , 50mm)manga1-mu.eps + [ mangab ] ( 150 mm , 50mm)manga2-mu.eps + [ mangac ] ( 150 mm , 50mm)manga3-mu.eps in case a transition ( see the upper four panels of figure [ manga-1 ] ) , we assume @xmath98 .
a short thermal timescale means that a system can quickly adjust to any changes added externally so that it should always stay around the thermal equilibrium state .
[ we assume that the disk is thermally stable ( see , e.g. , @xcite ) , as in the case that we are discussing .
] let us denote by @xmath99 the ignition radius , where the instability sets out .
since the disk should evolve along the thermal equilibrium curve , @xmath51 should increase accordingly , while @xmath55 should decrease during an upward transition at @xmath99 ( see panel a2 of figure [ manga-1 ] ) . a non - steady mass flow ( @xmath100 )
is then induced from its neighboring sites ( with high @xmath55 values ) to the ignition region ( with low @xmath55 values ) as is illustrated in panel a4 , since crudely we have @xmath101 from equation ( [ basic2 ] ) . here
, the non - steady mass flow is defined by @xmath102 where the averaged value , @xmath103 , is the mass flow rate averaged over a region around the radius , @xmath18 .
this non - steady mass flow assures an increase of @xmath51 at the ignition point , whereas @xmath51 in the neighboring sites should decrease ( see panels a1 and a3 of figure [ manga-1 ] ) .
to conclude , a transition at a certain point does not trigger further transitions in the surrounding region . in order for a transition to propagate into the neighboring regions
, these regions should acquire mass from the ignition site ( at @xmath99 ) . for this to occur
, the value of @xmath55 has to increase at @xmath99 , as is illustrated in panel b2 of figure [ manga-1 ] ( referred to as case b transition ) ; i.e. , the disk evolutionary path on the @xmath51-@xmath55 plane should largely deviate from the equilibrium curve . for this sort of behavior to take place
we should require @xmath97 , but this never happens in a standard type disk with no instability , since roughly we estimate @xmath104 ( with @xmath41 being the scale - height of the disk ) , and since the aspect ratio is expected to be small , @xmath105 .
we wish to emphasize , however , that this inequality no longer holds , once an instability grows and produces abrupt , spatial variations of physical quantities .
under such circumstances , equation ( [ standard - relation ] ) should be written as @xmath106 with @xmath107 being the spatial scale , on which the value of @xmath55 varies .
if this were the case , the evolution of @xmath51 profile would be like that illustrated in panel b4 of figure [ manga-1 ] . which is actually the case in the time - dependent calculation ?
the resultant time evolution of an unstable disk is illustrated as case c in the lower four panels of figure [ manga-1 ] .
when a transition occurs , @xmath55 only slightly decreases at @xmath99 [ see figure [ nusigma78 ] ] .
this apparently looks like case a except for one important difference ; that is , the disk trajectory starts to oscillate around its equilibrium value through interactions with neighboring regions after the transition to the upper branch .
( the meaning of ` interactions ' will be discussed in more details in sec .
owing to these fluctuations , the values of @xmath55 in the surrounding regions also begin to oscillate around their equilibrium values , and such fluctuations in @xmath55 eventually induce mass accretion flow from or into the neighboring region , thereby occasionally triggering transitions [ see figure 5(c ) ] .
these are purely non - steady effects .
importantly , the onset of the instability at some radius does not always trigger the instability in adjacent regions .
the instability zone is confined in a relatively narrow region . in other words
, the instability does not promote coherent behavior of the entire disk .
this point will be discussed in section 4.4 in comparison with the case of disks with the ` s'-shaped equilibrium curve .
+ there is another interesting feature unique to unstable accretion disks ; the influence of rapid variations in @xmath94 propagates from @xmath99 in the radial direction , as is depicted in figure 6(a ) .
we can see a wavy pattern in the @xmath94 distribution propagating over the disk plane from @xmath99 both in the inner and outer directions .
fluctuations in @xmath94 drive small amplitude fluctuations in @xmath51 [ see equation ( [ basic1 ] ) ] .
[ mdot - wave-1 ] ( 80 mm , 50mm)mdot - wave-1.eps [ mdot - variation-1 ] ( 80 mm , 50mm)mdot - variation-1z.eps figure 6(b ) illustrates the time developments of the accretion rate ( @xmath94 ) at several fixed radii . here , by @xmath94 at @xmath18 (= @xmath108 , @xmath109 ) we mean the mass flow from the radius @xmath110 to the radius @xmath18 .
the upper three curves ( or the lower three curves ) represent non - steady mass inflow ( outflow ) towards the ignition point from the outer ( inner ) adjacent zones .
it is important to note time delays in the @xmath94 response at the distant points ( in comparison with that at the closer points ) to the ignition radius , indicating that the fluctuations are not locally confined but spatially propagate and cause @xmath51 variations in the neighboring sites .
when the values of @xmath51 exceed its critical value , @xmath49 , an upward transition can be triggered .
the transition can occur , even when @xmath51 is expected to be below @xmath49 in the steady model for a given @xmath75 . here , we discuss what is meant by `` interactions '' in sec
the unique features in case c transitions reside in back - reactions of the non - steady mass flow ( @xmath111 ) [ see the upper and lower panels of figure 5 ] . according to our simulations , @xmath51 and @xmath55 values
change in the neighboring region after their receiving non - steady mass inflow .
the changes in @xmath55 will then modify the mass flow pattern in a next moment , since @xmath112 .
the modified mass flow pattern in the next moment produces further changes in @xmath51 and @xmath55 values , which sometimes cause a negative feedback .
this is the back - reaction of the non - steady mass flow , which is not considered in case a transition .
the transition wave eventually damps but on longer timescale than the thermal time .
this explains why the solution oscillates around the equilibrium curve even after the thermal timescale .
+ the neutrino luminosity can , in principle , be calculated by @xmath113 with @xmath114 cautions should be taken here , since @xmath94 in equation ( [ lumi - neu ] ) should be mass flow rate on the upper ( ndaf ) branch .
we , hence , need special arrangements when calculating neutrino luminosity from the part on the lower branch , which has very little contribution to the neutrino luminosity , even if @xmath94 is relatively large . in actual calculations , therefore , we use equation ( [ n - curve - c ] ) even when @xmath115 to calculate @xmath55 from @xmath51 , and then calculate @xmath94 from equation ( [ steady - mu ] ) by assuming the steady - disk relation . the resultant neutrino light curve is shown in figure 7(a ) . in the same figure
, we plot two more light curves representing the cases with no n - shaped equilibrium curves nor instability .
the upper ( or lower ) straight line labeled with ` steady 1'(`steady 2 ' ) is the light curve that we would expect if the lower ( upper ) branch would extend up ( down ) to high ( low ) @xmath51 regimes . comparing these light curves , we find two key features .
first , the overall shape of the light curve is steeper than the cases with no instability .
the main reason for this resides in the ` n'-shape ; that is , the value of @xmath51 and , hence , the @xmath55-value suddenly increases on a shorter timescale than the viscous timescale ( on which @xmath75 increases ) after the onset of the instability .
the second key feature of the light curve is that the luminosity abruptly increases every time a transition takes place at a certain radius , resulting in step - function - like variations .
this is because that the instability is confined in a narrow region and does not produce coherent variations over a wide range of the disk plane , as mentioned above .
we compare with the mass accretion rate at radius @xmath116 with the mass injection rate in figure 7(b ) .
we can see significant variability in the mass accretion rate .
this may produce rapid variations in the electromagnetic radiation via the blandford - znajek process ( to be discussed in section 4.3 ) .
[ case1-lumi - neu ] ( 80 mm , 50mm)lneu - mono-1.eps [ mdot - acc-1 ] ( 80 mm , 50mm)acc - inj-1.eps the second case is model 2 , in which the mass injection rate increases more rapidly .
we show the global evolution of the surface - density distribution in figure 8(a ) .
we can see that the region on the upper ( ndaf ) branch expands from @xmath99 both in the inner and the outer directions more rapidly than in model 1 .
this is simply because of more rapid increase in @xmath75 in model 2 than in model 1 .
if we have a closer look at this figure , moreover , we see that the out - going transition wave propagates faster than the in - going wave .
we show in figure 8(b ) the neutrino luminosity in model 2 .
we can see that the luminosity rises more smoothly in model 2 than in model 1 because the speed of the instability propagation is faster than in model 1 .
[ sigma - rapid ] ( 80 mm , 50mm)sigma-2a.eps [ lumi - neu - rapid ] ( 80 mm , 50mm)lneu - mono-2b.eps in the last model ( model 3 ) , we let the mass injection rate decrease on the viscous timescale . we compare the steady model ( left ) and the time - dependent model ( right ) in the same way as we did in model 1 [ see figure 9 ] .
the adopted mass injection rates are @xmath117 with @xmath118 , and @xmath92 [ see equation ( [ mdot - inj-3 ] ) ] .
we clearly see in both these panels that the high-@xmath51 region producing intense neutrino emission gradually shrinks .
in addition , we notice a tendency that the high-@xmath51 regions are systematically narrower in the steady model than in the time - dependent model .
this is because it takes about the viscous timescale until the information of reduced @xmath75 is transported to each portion of the disk so that it can lose the surface density to reach point b in the right panel of figure 1 .
+ [ steady - sigma-2 ] ( 80 mm , 50mm)steady - sigma-3kaia.eps [ sigma - hikaku-2 ] ( 80 mm , 50mm)sigma - hikaku-3kaib.eps figure [ nusigma54 ] shows the simulation results of model 3 on the @xmath51-@xmath55 plane , while figure [ mangad ] is a schematic picture explaining the time evolution of each portion of the disk on same plane .
when point b is reached at some radius ( referred to as @xmath99 ) , a downward transition is triggered there , thereby the value of @xmath55 sightly increasing .
this then causes non - steady mass flow ( @xmath100 ) from @xmath99 to the neighboring regions so that @xmath51 should decrease at @xmath99 .
even after the downward transition is completed , the @xmath55 and @xmath51 are oscillating around their equilibrium values more remarkably than in model 1 .
we explain why these violent oscillations occur in sec .
3.4.3 . such fluctuations in @xmath55
induce mass accretion flow from or into the neighboring regions and occasionally trigger transitions to the lower branch in the surrounding region .
it is also important to note that such propagation of the instability is restricted in a narrow zone . in other words
, it can not produce coherent oscillation .
+ ( 80 mm , 50mm)nusigma54.eps ( 120 mm , 50mm)manga4-mu.eps next , we show in figure [ mdot - wave-3 ] the radial distribution of the mass accretion rate after the downward transition .
it is clear in this figure that the @xmath94 distribution is never flat , meaning that each part of the disk at any times suffers fluctuations .
especially , @xmath94 rapidly goes up and down with a large amplitude when a transition takes place .
the influence propagates in a form of @xmath94 variation wave from @xmath99 to both in the inner and outer directions .
the situations are similar to those in model 1 but are more pronounced in model 3 .
fluctuations in @xmath94 drive variations of @xmath51 [ see equation ( [ basic1 ] ) ] . here , we think the violent fluctuations mentioned in sec .
3.4.2 over again [ see figure [ mangad ] ] .
the fluctuations are the results of the back - reactions of the non - steady mass flow as mentioned in model 1 .
the values of @xmath51 and @xmath55 change in the neighboring zones after their receiving non - steady mass outflow .
the changes in @xmath55 will then modify the mass flow pattern in a next moment . the modified mass flow pattern in the next moment further changes @xmath51 and @xmath55 values , which sometimes cause a negative feedback .
for this reason , the @xmath51 and @xmath55 values violently oscillates around the equilibrium curve even after the thermal timescale .
+ ( 80 mm , 50mm)mdot - wave-3.eps we show in figure 13(a ) the neutrino luminosity in model 3 . comparing the changes of the time - dependent luminosity with the changes of the luminosities in the steady models , we can see that the luminosity rapidly decreases when downward transitions from the upper regime ( ndaf ) to the lower ( adaf ) branch take place .
we plot the accretion rate at the inner edge of the disk for model 3 in figure 13(b ) .
we again see the significant variability in the mass accretion rate , as we saw in model 1 .
[ case3-lumi - neu ] ( 80 mm , 50mm)lneu - mono-3.eps [ mdot - acc-3 ] ( 80 mm , 50mm)acc - inj-3.eps
the thermal structure of the hyperaccretion flows is characterized by an n - shaped thermal equilibrium curve : the middle branch , which is secularly unstable , is sandwiched by the lower adaf branch and the upper ndaf branch . in this study , we examine the time evolution of a disk , which has n - shaped thermal equilibrium curves , by means of simple numerical simulations .
we solve time - dependent , height - averaged basic equations for viscous disks receiving with variable mass injection to see how the transition between the lower and upper branches occurs , how it propagates over a disk plane , and what differences arise from the cases without n - shaped equilibrium curves .
we studied three models , which have different time dependence of the mass injection rate : model 1 ( slow rise ) , model 2 ( rise ) , and model 3 ( decay ) .
what we find can be summarized as follows : 1 . when the mass injection rate increases ( or decreases ) with time , the region initially on the lower ( upper ) branch successively undergoes upward ( downward ) transitions .
the transition associated with rapid changes in the surface density and propagates over the disk , giving rise to an expansion ( shrinkage ) of the neutrino - emission region .
non - steady mass flow ( @xmath100 ) , the mass flow deviated from the spatially averaged flow , arises in a narrow region around the ignition radius @xmath99 during a transition .
this is because the middle branch is secularly unstable ; that is , the larger ( or the smaller ) the surface density of the unstable region is , the more ( less ) mass enters the ignition region than its neighboring regions .
resultant @xmath94 variation wave propagates over the whole disk .
3 . such non - steady effects can occasionally cause transitions even in the region at which the transition would not occur , if the disk were in a steady state .
the more slowly the mass injection rate changes , the more remarkable the non - steady effects become .
the effects are , however , not strong enough to produce coherent transition over the entire disk plane . 4 .
the transition between the lower and upper branches yields abrupt changes in the neutrino luminosity .
in fact , the neutrino luminosity calculated based on the time - dependent results changes much more rapidly than that in the case without n - shaped equilibrium curves .
non - steady effects can be seen as fluctuations of the neutrino luminosity and in mass accretion rates .
we have argued that the instability is a source of producing the non - steady effects .
if so , the properties of the disk evolution may critically depend on the precise shape of the n - shaped thermal equilibrium curve . in this subsection , we modify the thermal equilibrium curve in such a way to enhance the instability to see if it is possible for the instability to propagate more widely on the disk plane .
@xcite adopted the @xmath38 viscosity prescription in calculating the thermal equilibrium curve of a hyperaccretion shown in the left panel of figure 1 .
although they set @xmath38 to be constant (= 0.1 ) , there are no strong reasons to believe so . as a thought experiment we set @xmath119 on the upper branch , keeping the location of the lower branch and the slope of the middle branch unchanged .
the value of @xmath50 is now 45.0 , instead of 15.0 ( see figure
[ n - curve - hikaku ] ) . by this change ,
the middle unstable branch in the equilibrium curve is more elongated than in the original model .
this model is referred to as the modified n - shape model .
( 80 mm , 50mm)n - hikaku - mu.eps ( 80 mm , 50mm)sigma1-d - az.eps [ lneu - case1-d ] ( 80 mm , 50mm)lneu - mono-1-d.eps [ acc-1d ] ( 80 mm , 50mm)acc - inj-1d.eps as to the mass injection rate , we only calculate the case of slow rise [ i.e. , model 1 in section 3 ; see also equation ( [ mdot - inj-1 ] ) ] .
first , we show the global evolution of the surface - density distribution in figure 15 .
we can see that the region expands in a similar way to that in figure 3(b ) .
note also that the high-@xmath51 region with efficient neutrino emission values in the upper branch are higher .
next , we calculate the luminosity and show it in figure 16(a ) . comparing it with figure 7(a
) , we understand that the modification of the n - shape does yield more rapid increase in neutrino luminosity , but the speed of the upward transition wave is not so much affected .
we also calculate the mass accretion rate at radius @xmath116 in figure 16(b ) . comparing it with figure 7(b ) , we find that the accretion rate varies more or less continuously .
these variations look similar to the observed ones in gamma - ray bursts ( see the next subsection ) . here
we briefly discuss the observable features of the unstable hyperaccretion flows . the short - term variability of a hyperaccretion flow would be the origin of the highly variable prompt emissions of grbs because the variable energy deposition from an accretion flow would produce a spatially inhomogeneous outflow , and it would result in producing internal shocks , which give rise to the highly variable emission from shock - accelerated electrons . according to our calculations
, the neutrino luminosity shows abrupt changes with time during the transition of a hyperaccretion flow . in model 1 with slowly increasing mass injection , especially ,
the step - function - like changes in the neutrino luminosity occur repeatedly .
if we adopt the neutrino pair annihilation scenario as the energy deposition process from a hyperaccretion flow , we can expect the intermittent energy ejection from an accretion flow , which can result in the highly variable emission of grbs .
as for model 3 , where the injection rate decreases with time , the mass accretion rate at the inner edge shows strong variability .
this may be related to the short - term variation of the blandford - znajek jet luminosity since it is proportional to the magnetic energy density on the event horizon and is , hence , likely to be proportional to the mass accretion rate ( kawanaka et al .
2013a ) .
we wish to note that the behavior of the secular instability is not always controlled by the variation timescale of the mass injection rate , but that rapid variations can occur on a much shorter timescale , less than @xmath120 s [ see figure 7(b ) and equation ( [ t0 ] ) ] .
we here need to distinguish two timescales : the timescale of variable mass injection rate ( which is roughly the viscous timescale at large radii ) and that of the fluctuations in the @xmath51 and @xmath55 values around the ignition point ( which is much shorter than the former ) .
the variability timescale observed in grbs is the order of @xmath121 ( e.g. , @xcite ) .
the observed rapid variability of the grbs should be related to the latter process , and so we conclude that the variability timescale in our model can match the observations of grbs .
it is well known that the outbursts of dwarf - novae , a subclass of cataclysmic variables , and probably x - ray nova eruptions are triggered by a disk instability caused by the partial ionization of hydrogen in the disk ( see , e.g. , kato et al .
2008 , chapter 5 and references therein ) .
this is the so - called dwarf - nova type instability ( or hydrogen ionization instability ) and is characterized by the s - shaped equilibrium curve .
it can be easily shown that the middle branch is thermally and secularly unstable , while the other two branches are stable . under the circumstances relaxation oscillations between the bi - modal stable branches spontaneously occur , even if the mass injection rate does not vary in time .
this is the most successful time - dependent theory of the viscous accretion disk .
there are similarities and differences between the dwarf - nova type instability and the present case of hyperaccretion disk instability with the n - shaped equilibrium curve .
the most prominent common feature is bi - modal nature of the equilibrium curve .
there are three states for a given @xmath75 and the middle state being secularly unstable in the present case , while dwarf - nova type disks undergo bi - modal transitions , although only one state is found for a given @xmath75 .
another common feature is global propagation of bi - modal transitions from the ignition point to its neighboring sites .
a big distinction resides in the fact that the solution is only secularly unstable in the present case .
this is a crucial point when discussing the global nature of the instability in the present case , since the thermal timescale is much shorter than the viscous timescale , on which a secular instability grows .
we in fact find non - steady features not so much pronounced , judging from the fact that the transition is restricted in a narrow region and light curves exhibit small - amplitude fluctuations .
this contrasts the case in the dwarf - nova type disks which undergoes coherent oscillation and produce large - amplitude luminosity variations .
to summarize , the big distinction arises from the fact that the @xmath55 value can change with a large amplitude , thereby producing significant non - steady effects in a thermally unstable disk , while it can not in a thermally stable disk
. it might be important in this respect that we have employed one - dimensional formulation for the basic equations in the present study , assuming that the vertical structure of the disk can immediately respond to changes in @xmath51 .
this is not precisely correct , since it takes about a sound crossing time over the disk height ( @xmath122 ) for the entire vertical structure to adjust to any changes in @xmath51 . in future work , we thus need two - dimensional simulations to see vertical propagation of the instability .
further , discontinuous transitions in the neutrino light curve , which seems to depend on a finite mesh size , can be altered by considering the two - dimensional effects .
this is because changes in the vertical structure take longer time than the propagation time of the instability in the radial direction over a distance of one mesh size .
we naively expect that the region with the width of @xmath123 may simultaneously undergo the branch transition , if we properly consider the two - dimensional effects . in order to solve the time evolution of a hyperaccretion disk strictly
, we have to solve equation ( [ thermal.eq ] ) without modeling it .
in other words , we have to evaluate advection cooling rate ( @xmath124 ) as the function of the radial velocity ( @xmath125 ) and the @xmath18 differentiation of the entropy ( @xmath126 ) and neutrino cooling rate ( @xmath127 ) as the function of the temperature ( @xmath128 ) .
we are grateful to the anonymous referee for insightful comments .
this work is supported in part by grants - in - aid of the ministry of education , culture , sports , science and technology ( mext ) ( 26400229 sm ) .
in our calculations , we use the implicit method for time integration of equations ( [ basic1 ] ) and ( [ basic2 ] ) . at each time step
, we solve these equations , as well as the energy equation in non - dimensional forms [ see equation ( [ thermal.eq ] ) ] . the radial coordinate ( @xmath56= 0.0 1.0 )
is divided into 200 mesh points ( with a constant mesh spacing of @xmath129 ) ; @xmath130 with @xmath131 .
we distinguish cell variables ( such as @xmath51 and @xmath55 ) and interface variables ( such as @xmath132 ) .
the former variables are defined at the same place of the radial mesh points ; e.g. , @xmath133 is the surface density at @xmath134 ( @xmath135 ) , while the latter variables are defined between the mesh points ; e.g. , @xmath132 is the mass flow rate from the cell at @xmath136 to the cell at @xmath137 .
second , we set @xmath138 to be constant ( @xmath139 ) being independent of @xmath56 . in the present study
we assign @xmath140 so as to satisfy the two conditions , @xmath141 and @xmath142 at every radius .
when either of these conditions is unsatisfied at some mesh point(s ) , we re - calculate by using smaller @xmath138 so that conditions should be met . one exception is model 3 , in which we omit calculations in the innermost region ( at @xmath143 ) to avoid numerical difficulties .
as a test of the numerical code , we check the mass conservation , we compute @xmath144 , the disk mass in the following two methods and the difference between them .
we use trapezoidal rule for this integral calculus . in method
a we calculate we have confirmed that the total mass is conserved within the accuracy of less than 0.2@xmath147 in model 1 , less than 0.8@xmath147 in model 2 , and less than 3.6@xmath147 in model 3 for time integration of @xmath148 ( @xmath149 ) . | hyperaccretion flows with mass accretion rates far above the eddington rate have an n - shaped equilibrium curve on the @xmath0-@xmath1 plane ( with @xmath0 and @xmath1 being surface density and mass accretion rate , respectively ) .
the accretion flow on the lower @xmath0 branch of the n - shape is optically thick , advection - dominated accretion flow ( adaf ) while that on the upper one is neutrino - dominated accretion flow ( ndaf ) .
the middle branch has a negative slope on the @xmath0-@xmath1 plane , meaning that the flow on this branch is secularly unstable . to investigate how the instability affects the flow structure and what observable signatures are produced
, we study the time evolution of the unstable hyperaccretion flow in response to variable mass injection rates by solving the height - averaged equations for viscous accretion flows . when a transition occurs from the lower branch to the upper branch ( or from the upper branch to the lower branch ) , the surface density rapidly increases ( decreases ) around that transition region , which induces locally enhanced mass flow ( referred to as non - steady mass flow ) into ( out of ) that region .
we confirm that the non - steady flow can create a kind of disturbance and that it propagates over the whole disk .
however , the non - steady mass flow is not strong enough to induce coherent transition over the whole disk , unless the mass injection rate varies with time .
when the injection rate continuously changes , the neutrino luminosity varies intermittently , thus producing step - function - like light curves , as the radiation efficiency discontinuously changes every time the local transition occurs .
the effects of changing the n - shape and possible observational consequences on the gamma - ray burst emission are briefly discussed in relation to gamma - ray bursts . |
through lattice qcd simulations ( lqcd ) it is possible to study qcd at current quark masses which are not restricted to their physical values .
data points obtained for such a scenario represent a powerful pool of information to fix free low - energy constants ( lecs ) of the ( baryon ) chiral perturbation theory ( b@xmath6pt ) ; lecs that also enter calculations for experimental observables .
thus , adjusting the b@xmath6pt to lqcd data in the unphysical region allows for predictions at the physical point .
in our study we performed such a matching for the quark - mass ( @xmath7 ) dependence of the nucleon mass @xmath8 .
we fit lqcd data for the @xmath9 flavors , two light degenerated quarks , as well as for the @xmath10 ensembles , two light degenerated and one heavy quarks . here , we concentrate on our @xmath3 results and refer to ref .
@xcite for more details on the @xmath9 ones . a measure of the contribution from the explicit chiral symmetry breaking to the nucleon mass is given by the so - called @xmath11-term .
it is the nucleon scalar form factor at zero momentum transfer squared which can be isolated from @xmath12-scattering experiments .
alternatively , the @xmath11 can also be obtained by the hellmann - feynman theorem from @xmath8 : @xmath13 as a result , the function @xmath8 , which has mainly unphysical values , connects lqcd , b@xmath6pt and experiment .
the b@xmath6pt up to @xmath0 describes the nucleon mass by : @xmath14 with @xmath15 and @xmath16 as the pion mass and pion decay constant .
we fit the three lecs @xmath17 , @xmath18 and @xmath19 to lqcd data and obtain through eq .
( [ eq : hf ] ) a @xmath11 value .
explicit @xmath20 contributions can appear in the loop - contributions @xmath21 and we refer to ref .
@xcite for all technical details .
generally , lqcd data is given in the dimensionless form @xmath22 where each collaboration sets its scale through a specific value for the lattice spacing @xmath23 . in the @xmath3 case
, we fit the data after converting to physical units .
since all three quantities are subject to statistical uncertainties , the @xmath23 uncertainty translates into a correlated uncertainty for the normalization of data points belonging to the same set .
we take this correlation into account by defining a correlation matrix @xmath24 containing all these uncertainties , and minimize the function : @xmath25 with @xmath26 being the data and @xmath27 and @xmath28 the b@xmath6pt infinite volume nucleon mass and its finite volume corrections .
the constants @xmath29 parametrize finite spacings effects for each lqcd - action separately in our @xmath1 b@xmath6pt analysis we only include @xmath3 lqcd data for which the strange quark mass is kept constant approximately at its phenomenological value .
for this case the strange quark mass corrections to the nucleon mass can be seen as being integrated out into the lecs we are fitting .
furthermore , to ensure controllable finite volume corrections and a reasonable chiral convergence of our results , we only account for data points with @xmath30 and @xmath31 mev , with @xmath32 being the lattice box size . following this , we include data from the bmw @xcite , hsc @xcite , lhpc @xcite , milc @xcite , nplqcd @xcite , pacs @xcite and rbcuk - qcd @xcite collaborations .
fits . right : the blue - solid ( green - dashed ) line shows our result for the @xmath33 b@xmath6pt fit with(out ) @xmath20 contributions by excluding the two data points from the hsc and nplqcd collaborations . the red - dotted line is a b@xmath6pt fit with @xmath20
including also this data .
left : same results depicted for a smaller @xmath34 range.,title="fig:",width=245 ] fits .
right : the blue - solid ( green - dashed ) line shows our result for the @xmath33 b@xmath6pt fit with(out ) @xmath20 contributions by excluding the two data points from the hsc and nplqcd collaborations . the red - dotted line is a b@xmath6pt fit with @xmath20 including also this data .
left : same results depicted for a smaller @xmath34 range.,title="fig:",width=245 ] the fig .
[ fig-1 ] shows several b@xmath6pt fit strategies to the above listed @xmath3 lqcd data .
we obtain consistent descriptions of the pion - mass dependence of the nucleon mass where the slope - variations are below the resolution of the current data . however , by looking to the pion - mass dependence of the @xmath11 term , left graph of fig .
[ fig-2 ] , we observe that these small variations translate into noticeable changes . for successively including explicit @xmath20 contributions and finite spacing corrections in our fit formula , the @xmath11 term at the physical point changes by up to @xmath35 mev from @xmath36 mev to @xmath37mev down to @xmath38 mev
in view of this , we take a conservative standpoint and give the following weighted average as our final @xmath3 @xmath11 value at the physical point : @xmath39 where the second ( systematical ) uncertainty spans all the above central values . the right graph of fig .
[ fig-2 ] compares our b@xmath6pt results from separate fits to lqcd data from @xmath9 and @xmath3 ensembles . also here , differences are within the spread and uncertainties of the input data but result in a slightly smaller @xmath11 value for the @xmath40 case : @xmath41 concerning the uncertainties , these two values are compatible , however , show a slight tension which we mostly trace back to the different data point distributions . on the one hand
, the current @xmath9 data does not constrain the low-@xmath34 region as much as the @xmath3 data .
this region is dominated by the lec @xmath17 which is the @xmath11 at leading order .
consequently , we obtain by @xmath42 % different @xmath17-values with correspondingly different contributions to the total @xmath11 . on the other hand , in the @xmath9 case one direct data point @xcite of the @xmath11 at @xmath43 mev
could be included in the fit .
a positive outcome was that the above spread of our results with and without inclusion of @xmath20 contributions is lifted .
since no such data point at low-@xmath34 is available for the @xmath3 case , a spread of @xmath44 mev remains .
note that in the hb@xmath6pt formalism @xcite this spread is more than @xmath45 mev . to compare our above numbers also with phenomenology
, we cite the latest value extracted from pure @xmath46-n scattering data : @xmath47 mev @xcite . in conclusion
, we performed fits of the @xmath0 @xmath1 covariant b@xmath6pt with and without explicit @xmath20 contributions to lqcd nucleon mass data . even though the present data set is extensive , we spotted systematic effects coming mainly from the distribution of the data points .
these systematic effects would further be reduced by new data for : a ) more @xmath9 data points in the low-@xmath34 region , b ) already one direct calculation of the @xmath11 at @xmath48 mev for the @xmath49 case .
term from @xmath3 fits .
the color code is the same as in fig .
[ fig-1 ] .
the red - diamond is our predicted @xmath11 value at the physical point .
right : comparison of b@xmath6pt @xmath0 fits with @xmath20 to @xmath9 ( red - triangles ) and @xmath3 ( blue - squares ) lqcd data .
the blue - circle is the physical nucleon mass.,title="fig:",width=245 ] term from @xmath3 fits . the color code is the same as in fig .
[ fig-1 ] .
the red - diamond is our predicted @xmath11 value at the physical point .
right : comparison of b@xmath6pt @xmath0 fits with @xmath20 to @xmath9 ( red - triangles ) and @xmath3 ( blue - squares ) lqcd data .
the blue - circle is the physical nucleon mass.,title="fig:",width=245 ]
the work has been supported by the spanish ministerio de economa y competitividad and european feder funds under contracts fis2011 - 28853-c02 - 01 and fis2011 - 28853-c02 - 02 , generalitat valenciana under contract prometeo/2009/0090 and the eu hadron - physics3 project , grant no .
jmc also acknowledges support from the science technology and facilities council ( stfc ) under grant st / j000477/1 , the grants fpa2010 - 17806 and fundacin sneca 11871/pi/09 .
0 l. alvarez - ruso , t. ledwig , j. martin - camalich , m.j .
vicente - vacas , _ phys .
d _ * 88 * , 054507 ( 2013 ) .
s. durr et al .
( bmw collaboration ) , _ phys .
d _ * 85 * , 014509 ( 2012 ) .
lin et al .
( hs collaboration ) , _ phys .
d _ * 79 * , 034502 ( 2009 ) .
bratt et al .
( lhpc collaboration ) , _ phys .
d _ * 82 * , 094502 ( 2010 ) .
d. toussaint et al .
( milc collaboration ) , private communication .
beane et al .
( nplqcd collaboration ) , _ phys .
d _ * 84 * , 014507 ( 2011 ) .
s. aoki et al .
( pacs collaboration ) , _ phys .
d _ * 79 * , 034503 ( 2009 ) . c. jung et al .
( rbc and ukqcd collaborations ) , _ pos lattice2012 _ , 164 ( 2012 ) .
bali et al .
( qcdsf collaboration ) , nucl .
b * 866 * , 1 ( 2013 ) .
a. walker - loud , _
pos lattice2008 _ , 005 ( 2008 ) .
alarcon , j. martin camalich , j.a .
oller , _ phys .
d _ * 85 * , 051503 ( 2012 ) . | fits of the @xmath0 covariant @xmath1 baryon chiral perturbation theory to lattice qcd nucleon mass data from several collaborations for 2 and 2 + 1 flavors are presented .
we consider contributions from explicit @xmath2 degrees of freedom , finite volume and finite spacing corrections .
we emphasize here on our @xmath3 study .
we obtain low - energy constants of natural size that are compatible with the rather linear pion - mass dependence of the nucleon mass observed in lattice qcd .
we report a value of @xmath4 mev in the 2 flavor case and @xmath5 mev for 2 + 1 flavors . |
the truebeam linac is the most recently marketed linear accelerator by varian medical systems ( palo alto , california , usa ) .
the truebeam linac can produce flattening - filter - free ( fff ) photon beams of 6 and 10 mv .
removing the flattening filter is an effective way to boost the dose rate in high - dose treatments with multiple narrow fields such as those employed in stereotactic ablative radiotherapy .
since the seminal works of petti and co - workers @xcite and of mary udale @xcite on monte carlo simulation of a linac until more recent times ( see @xcite and references therein ) researchers have tried to improve the accuracy with which monte carlo models reproduce experimental dose distributions in water and heterogeneous phantoms .
it has become clear that this goal requires precise knowledge of the dimensions and materials of the linac head assembly components that influence the beam .
usually , these data are provided by the manufacturers to monte carlo users who agree to keep this information undisclosed . however , in the case of the truebeam linac , varian has not made available the characteristics of the head components situated above the collimation jaws .
instead , the company distributes , through its website ( http://www.myvarian.com/montecarlo ) , phase - space files of the fff beams tallied at a plane located just upstream the jaws which can be downloaded by truebeam users .
these phase - space files can be used as sources to transport particles through the geometry of the jaws , and other beam modifiers , for subsequently estimating the absorbed dose in a patient computerized tomography or phantom .
those phase - space files were produced by varian @xcite in simulations employing the monte carlo code geant4 @xcite and are coded in the international atomic energy agency ( iaea ) phase - space format @xcite .
some authors have validated the varian distributed phase - space files and they have generally found a reasonable agreement of the estimated dose with measurements @xcite . however , monte carlo simulations based on third party tallied phase spaces are subject to several limitations : ( i ) the initial beam parameters can not be adapted to match dose profiles measured in the user s linac ; ( ii ) the latent variance @xcite of the phase - space file imposes a limit in the statistical uncertainty of the estimated observables ; ( iii ) simulations are subject to type b uncertainties associated to limitations of the monte carlo code used for generating the phase - space files @xcite ; and ( iv ) the energy deposited in the ionization chamber can not be determined since this component is located in the upper part of the linac , thus , absolute dosimetry in units of gy / mu can not be accurately computed by the method of popescu _ et .
al _ @xcite . in this work we present a monte carlo model of the truebeam linac , inferred by comparing dose measurements and simulations , which is suitable for the simulation of the 6 and 10 mv fff beams .
the geometry of the linac in this model includes thin filters for the 6 and 10 mv beams that are situated approximately in the position occupied by the standard flattening filters of a varian clinac 2100 linac .
we have called this geometry fakebeam to emphasize that it is not the actual geometry of the truebeam machine , but an _ ad hoc _ one created with the purpose of providing monte carlo users a way to circumvent some of the aforementioned limitations of using the varian distributed phase - space files .
simply removing the flattening filters of a clinac 2100 linac is not enough to reproduce the dose distributions produced by the truebeam fff beams in a water phantom .
the sole elimination of the flattening filter would allow high - energy electrons escaping from the target to produce a considerably higher than expected dose at shallow depths .
the insertion of a thin sheet of a material such as bronze or copper at some place in the beam path can absorb those contaminant electrons .
however , this simple configuration still fails to reproduce the off - axis absorbed dose . according to the manufacturer ( information provided to truebeam users under non - disclosure agreement )
, there is a large similarity between the head assembly geometries of the clinac 2100 and the truebeam linacs downstream of a plane situated just above the jaws .
thus , in the fakebeam geometry modifications on the head assembly of the clinac 2100 linac were introduced in the part upstream of the jaws . a trial and error process was used to find the shape of thin filters which , situated approximately at the location of the standard flattening filters of the clinac 2100 machine , reproduce the dose distribution produced by the fff beams of the truebeam linac in a water phantom .
we shall name these filters 6-fff and 10-fff filters .
the detailed geometry of these filters is shown in figure [ fig : filters ] .
the material composition of the 6-fff and 10-fff filters is bronze ( 70% copper , 30% zinc , @xmath38.412 g/@xmath4 ) and tantalum ( @xmath316.654 g/@xmath4 ) , respectively .
the selection of these materials was inspired by the composition found in other flattening filters from varian .
a minor change was made to the original geometry of the clinac 2100 , so that the beryllium window was repositioned at the upstream plane of the primary collimator .
consequently , only the target is at vacuum in the fakebeam geometry .
the fakebeam geometry includes the modifications to the primary collimator and the secondary collimator ( lead shield ) proposed by chibani and ma for the clinac 2100 @xcite .
all the simulations included in this work were performed with the primo code @xcite , a user - friendly and freely distributed software ( http://www.primoproject.net ) specifically designed for simulating varian and elekta linacs and estimating absorbed dose distributions in water phantoms and computerized tomographies .
primo is designed with a layered structure having the monte carlo general - purpose radiation transport code penelope @xcite and the geometry packages penvox and pengeom as the lowermost layers .
the main program used in primo for steering the penelope radiation transport subroutines is peneasy @xcite .
the modified geometry of the clinac 2100 including the 6- and 10-fff filters were coded as a hierarchic structure of homogeneous bodies limited by quadric surfaces according to the rules of the pengeom package and were incorporated into primo as a new linac named fakebeam .
the penelope absorption energies ( eabs ) were set to 100 kev for charged particles and 20 kev for photons everywhere .
for bremsstrahlung emission by electrons and positrons @xcite , the energy of the emitted photons is sampled from the scaled cross - section tables of seltzer and berger @xcite and its angular distribution is described by an approximation to the partial - wave shape functions of kissel , quarles and pratt @xcite .
penelope uses a mixed simulation scheme @xcite for electrons and positrons and detailed simulation for photons .
the five penelope transport parameters for mixed simulation are c1 , c2 , wcc , wcr and dsmax .
c1 and c2 are the electron average angular deflection and the maximum fractional energy loss allowed in one step , respectively .
wcc and wcr are the energy cutoffs for inelastic and bremsstrahlung interactions , respectively .
dsmax is the maximum allowed step length . in all simulations
the condensed history transport parameters were set to c1@[email protected] , wcc@xmath5100 kev and wcr@xmath520 kev for all materials , except for the materials composing the target where they were set to c1@[email protected] , wcc@xmath51 kev and wcr@xmath520 kev .
we have observed that those parameters are adequate for producing an accurate transport in the target @xcite . the maximum step length ( dsmax )
was set to @xmath6 of the thickness of each material body .
the initial electron source for the nominal 6 mv beam had a gaussian energy distribution with mean of 5.8 mev and full width at half maximum ( fwhm ) of 1% .
its spatial distribution was also modeled by a gaussian function centered at the linac central axis with a fwhm of 0.15 cm . for the nominal 10 mv beam
the initial electron source was monoenergetic with an energy of 10.8 mev and spatially distributed by a gaussian function with a fwhm of 0.1 cm .
these parameters were found optimal to match the measurements .
several variance - reduction techniques were employed . the splitting - roulette @xcite , which is a technique that combines photon splitting and russian roulette applied to photons and electrons ,
was used in the target .
the moveable skins technique @xcite was applied in the jaws and the primary collimator .
moveable skins is a particular implementation of range rejection in which external layers of high atomic number collimation structures are geometrically separated from the internal regions .
these layers are called ` skins ' . in skins ,
all particles undergo analogue simulation ( _ i.e. _ , no variance - reduction technique is applied ) , whereas in internal regions electrons are terminated by setting arbitrarily high energy cutoffs .
additionally , simple splitting was applied to all particles crossing the surface of the water phantom .
the average standard statistical uncertainty of all bins scoring more than 50% of the maximum dose was 0.3% for all simulations of the 10 mv fff beam and ranged between 0.4% and 0.6% for simulations of the 6 mv fff beam .
the @xmath1 @xmath2 field was used to iteratively tune the model and the rest were used to validate it . additionally , simulations employing the second generation of the varian distributed truebeam phase - space files as sources of particles were also run .
the phase - space files that belong to the second generation are those tallied on a plane , while those that belong to the first generation were tallied on a section of a cylindrical surface and employed a substantially lower number of primary particles @xcite .
the phase - spaces were imported into primo .
the region downstream of the jaws was simulated as described above for the fakebeam geometry and following the varian recommendations .
splitting in the water phantom was used with a variable splitting factor depending on the field .
the factors were selected to be sufficiently large to approximate the dose uncertainty to the latent variance of the phase space @xcite . the obtained average standard statistical uncertainty of all bins scoring more than 50% of the maximum dose was about 0.8% .
we refer to these simulations as truebeam .
validation of the geometry was based on comparisons between monte carlo estimated dose distributions and two sets of measurements : ( i ) the varian golden beam data set ( gbds ) which employed an iba ( schwarzenbruck , germany ) cc13 ionization chamber ; ( ii ) in - house measurements performed with an iba pfd@xmath7 detector @xcite .
comparisons were made for field sizes of 3@xmath83 , 6@xmath86 , 10@xmath810 , 20@xmath820 and 40@xmath840 @xmath2 .
crossline profiles were compared at depth of maximum dose @xmath9 , at 5.0 cm and at 10.0 cm depths .
measurements with the diode were used for comparing the depth - dose curves of the 3@xmath83 @xmath2 field and lateral dose profiles of the 3@xmath83 , 6@xmath86 and 10@xmath810 @xmath2 .
the gbds was employed for the rest of depth - dose and crossline dose profiles comparisons .
the ssd was 100 cm and the phantom bin size was selected according to the field size : for the 3@xmath83 , 6@xmath86 and 10@xmath810 @xmath2 fields it was [email protected]@xmath80.1 @xmath4 ; for larger fields it was [email protected]@xmath80.2 @xmath4 ( crossline@xmath8inline@xmath8depth ) .
the larger bin size in the inline direction was chosen in order to reach a low uncertainty in the simulations with the truebeam phase - space files .
the percentage of points passing a gamma analysis @xcite with criteria of 1% , 1 mm ( @xmath10 ) , the dose difference relative to the maximum dose ( diff% ) and the root - mean - squared error ( rms ) of the dose difference were employed as figures of merit to determine agreement between the measured and the monte carlo estimated dose distributions . the rms was computed only for depth - dose curves at depths greater than 3 mm . in spite of the fact that the reference data set in the gamma analysis is a 1d array of dose points with arbitrary separation
, primo performs a re - gridding process of the evaluated 3d dose distribution that allows to apply an arbitrarily small distance - to - agreement criterion .
re - gridding is performed using trilinear interpolation . in the same manner ,
the dose profiles used for evaluating the percentage difference were calculated in the monte carlo estimated 3d dose distribution by trilinear interpolation at the measurement points . in order to reduce the effect of uncertainties in the analysis , estimated depth - dose curves were normalized so that their areas between @xmath9 and @xmath11 coincided with the experimental values @xcite .
crossline profiles were normalized to 100% at the central axis .
we performed a series of simulations with the purpose of assessing the sensitivity of the dose profiles produced by the fakebeam model to the initial beam parameters of the beam .
the depth - dose curves and lateral profiles at 5.0 cm of depth obtained with the initial parameters described in section [ sec : mc ] for a 40@xmath840 @xmath2 field were used as reference in the gamma analysis . for the 6 mv beam we varied the initial energy of electrons from 5.6 mev to 6.0 mev in steps of 0.1 mev .
the fwhm of the focal spot size was varied from 0 to 0.15 cm in steps of 0.05 cm .
for the 10 mv beam the initial energy ranged from 10.6 to 11.0 mev with a step of 0.2 mev and the fwhm of the focal spot size was varied from 0 to 0.1 cm in steps of 0.05 cm .
the fwhm of the initial energy was set to 0 in all cases .
additionally , we performed simulations with the initial beam parameters that were used to generate the varian s phases - space files .
these parameters are declared in the phase - space file headers and are , for the 6 mv beam , 5.9 mev , 0.06 mev and 0.078 cm for the initial energy , fwhm of the energy and fwhm of the focal spot size , respectively . likewise ,
for the 10 mv beam they were 10.2 mev , 0.102 mev and 0.096 cm . the obtained average standard statistical uncertainty of the dose distributions varied between 0.5% and 1% .
figure [ fig : pdds ] shows the comparison between measured and monte carlo estimated depth - dose curves . for fakebeam ,
differences are less than 1% in most of the points for both beams and all fields evaluated , except in the high dose gradient build - up region , near the phantom surface , where an over - response of the cylindrical ionization chamber is expected .
this over - response arises from the fact that ionization chambers , even those with a small cavity volume such as the cc13 employed for measuring the varian s gbds , overestimate the dose in the build - up region where lack of charged particle equilibrium exists .
depth - dose curves for truebeam also show a good agreement for both beams with lower differences in the build - up region than fakebeam but slightly larger beyond @xmath9 , which results in relatively lower rms values for fakebeam as shown in figure [ fig : rms ] .
values of @xmath10 obtained were in all cases , for both the fakebeam and the truebeam simulations , greater than 98% .
for the three depths considered in the analysis .
notice that results for the @xmath1 @xmath2 field are shown in a different scale . ]
comparison of lateral dose profiles shown in figures [ fig : profs ] and [ fig : pdiff ] reveals a good agreement of fakebeam with the experimental data with differences less than 1% for large fields ( @xmath12 @xmath2 and @xmath1 @xmath2 ) and less than 1.5% for smaller fields , except in the regions of high dose gradient .
these differences are similar to those found for truebeam for the smaller fields but lower for large fields where truebeam agreement is poorer , particularly for the 6 mv fff beam .
values of the combined @xmath10 obtained for the lateral dose profiles at the three depths considered are shown in figure [ fig : gamma ] .
the agreement is very good for fakebeam with at least 96% of the points passing the test for all fields and energies
. gamma analysis also reveals the poor agreement obtained with the phase - space files distributed by varian for truebeam for the @xmath1 @xmath2 field with 42% and 84% of the points passing the test , for the 6 mv and 10 mv beams , respectively .
these values improved to 88% and 98% when the gamma analysis criteria were relaxed to 2% , 2 mm .
results of the sensitivity analysis show that the dose is quite insensitive to the initial parameters of the beam in the range analyzed for both nominal energies .
dose profiles and depth dose curves matched the reference data sets with @xmath13 results of 98% or better .
for the lateral dose profiles obtained with the parameters used to generate the varian s phase - space files , the 6 mv beam produced a @xmath10 and @xmath14 of 97% and 100% , respectively . in contradistinction
, for the 10 mv beam the @xmath10 and @xmath14 were 53% and 95% , respectively .
this larger variation is caused by the marked difference between the initial energy of the reference and the analyzed beams , which is of about 0.5 - 0.7 mev .
a geometrical model , named fakebeam , for the monte carlo simulation of fff beams produced by the truebeam linac has been developed and validated by comparison of computed pdds and profiles with experimental measurements .
the description of the filters developed in this work allows those researchers having access to the varian clinac 2100 geometry to implement fakebeam in their own monte carlo systems .
fakebeam has been incorporated into primo as a new linac model , so users can simulate it without any coding effort .
primo includes the geometries of the millennium 120 and 120 hd multileaf collimators .
users of the primo code can either simulate the fakebeam geometry or import the varian s distributed phase - space files for truebeam .
our approach overcomes some of the limitations of using phase - space files as a source of particles .
now it is possible to : ( i ) adapt the initial beam parameters to match measured dose profiles ;
( ii ) reduce the statistical uncertainty to arbitrary low values ; and ( iii ) assess systematic uncertainties ( type b ) by employing different monte carlo codes .
our model does not take into account the actual geometry of the ionization chamber of the truebeam linac .
instead , it uses the ionization chamber of the varian clinac 2100 .
since knowledge of the ionization chamber geometry is essential for some methods of monte carlo absolute dosimetry , our model is limited in this regard .
notwithstanding , the fakebeam geometry could be used in conjunction with the method employed by zavgorodni _ et .
al _ @xcite to determine backscattered radiation in the monitor chamber . with this combined approach
it is possible to provide dose distributions expressed in gy / mu with an accuracy close to 1% .
mr , js and lb are thankful to the spanish ministerio de economa y competitividad ( project fis2012 - 38480 ) and to the deutsche forschungsgemeinschaft ( project br 4043/3 - 1 ) .
js is grateful to the spanish networking research center ciber - bbn for financial support . | monte carlo simulation of linear accelerators ( linacs ) depends on the accurate geometrical description of the linac head .
the geometry of the varian truebeam linac is not available to researchers .
instead , the company distributes phase - space files of the flattening - filter - free ( fff ) beams tallied at a plane located just upstream the jaws . yet ,
monte carlo simulations based on third party tallied phase spaces are subject to limitations . in this work ,
an experimentally - based geometry developed for the simulation of the fff beams of the varian truebeam linac is presented .
the monte carlo geometrical model of the truebeam linac uses information provided by varian which reveals large similarities between the truebeam machine and the clinac 2100 downstream of the jaws .
thus , the upper part of the truebeam linac was modeled by introducing modifications to the varian clinac 2100 linac geometry .
the most important of these modifications is the replacement of the standard flattening filters by _
ad hoc _ thin filters .
these filters were modeled by comparing dose measurements and simulations .
the experimental dose profiles for the 6 mv and 10 mv fff beams were obtained from the varian golden data set and from in - house measurements performed with a diode detector for radiation fields ranging from @xmath0 to @xmath1 @xmath2 at depths of maximum dose , 5 and 10 cm .
indicators of agreement between the experimental data and the simulation results obtained with the proposed geometrical model were the dose differences , the root - mean - square error and the gamma index .
the same comparisons were done for dose profiles obtained from monte carlo simulations using the phase - space files distributed by varian for the truebeam linac as the sources of particles .
results of comparisons show a good agreement of the dose for the ansatz geometry similar to that obtained for the simulations with the truebeam phase - space files for all fields and depths considered , except for the @xmath1 @xmath2 field where the ansatz geometry was able to reproduce the measured dose more accurately .
our approach overcomes some of the limitations of using the varian phase - space files .
it makes possible to : ( i ) adapt the initial beam parameters to match measured dose profiles ; ( ii ) reduce the statistical uncertainty to arbitrarily low values ; and ( iii ) assess systematic uncertainties ( type b ) by employing different monte carlo codes .
one limitation of using phase - space files that is retained in our model is the impossibility of performing accurate absolute dosimetry simulations since the geometrical description of the truebeam ionization chamber remains unknown . |
a tunable filter ( tf ) is an imaging device that can isolate an arbitrary spectral band @xmath7 at an arbitrary wavelength @xmath5 over a broad continuous spectral range .
those filters are ideally suited for surveys of emission - line galaxies ( elgs ) in different environments and are a powerful tool to detect distant line emitters @xcite .
the use of tfs reduces significantly the sky contamination , an important limitation of broad band surveys , because they cover a small wavelength range , thus increasing the contrast between the emission lines and the continuum , and allowing a moderate 2d coverage in a single pointing depending on the instrument .
also , tfs are narrower than most narrow band filters generally used , thus increasing emission line object detection ratio . among the first tf systems for non solar astronomy
, we have the goddard fabry - perot imager ( gfpi ; * ? ? ?
* ) , which is an optical scanning interferometer and ccd imaging system .
also , @xcite developed a narrowband imaging survey using a fabry - perot imaging interferometer .
another known tf system is the taurus tunable filter ( ttf ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
now decommissioned , it was in operation from 1996 to 2003 on the 3.9 m anglo - australian telescope and from 1996 to 1999 on the 4.2 m william hershel telescope ( wht ; * ? ? ?
the ttf was used , among other things , for several extragalactic surveys ( e.g. * ? ? ?
* ) , and has shown that there is a rich field of science awaiting exploration with large ground based telescopes equipped with these narrow band imagers ( for a review see * ? ? ?
* ) . among other instruments with tfs
, we have the maryland - magellan tunable filter ( mmtf ; * ? ? ?
* ) , installed on the magellan
baade 6.5 m telescope , located at las campanas observatory , chile , and the robert stobie spectrograph ( rss ; * ? ? ?
* ) for the 11 m south african large telescope ( salt ) , which provides spectroscopic imaging at any desired wavelength from 430 to 860 nm .
otelo ( osiris tunable emission line object survey ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) is an emission line object survey using the red tf of osiris ( optical system for imaging and low resolution integrated spectroscopy ) @xcite . the possibility to measure
h@xmath4 and [ nii]@xmath56583 lines and discriminating agns , makes otelo a unique emission line survey . observing in selected atmospheric windows relatively free of sky emission lines
, it is expected to reach a 5 sigma depth of 10@xmath8 erg/@xmath2/s , detecting objects with ew @xmath3 0.3 . with a total survey area of 0.1 square degrees distributed in different fields , such as the extended groth strip , goods - n , sxds , and cosmos
, otelo will be the deepest emission line survey to date .
the expected number of emitters distributes as follows : 1000 h@xmath4 star forming emitters up to a redshift 0.4 , from which about 100 would correspond to low luminosity star forming galaxies , 6000 star forming emitters in other optical emission lines up to a redshift 1.5 , 400 ly@xmath4 emitters at redshifts up to 6.7 , 400 qso at different redshifts , and about 1000 agns .
the otelo survey observations are being presently carried out in the groth field .
the project has produced previous @xmath9 broad - band photometry @xcite , as well as optical properties of x - ray emitters @xcite on this field .
one of the aims of the otelo survey is to estimate metallicities of elgs . among the different indirect methods to estimate metallicities in elgs we can distinguish between theoretical models , such as [ nii ] @xmath56583/[oii ] @xmath53727 @xcite , empirical calibrations , for example the @xmath10 ratio @xcite , or a combination of both , e.g. the n2 method @xcite .
a detailed description of the different metallicity methods and calibrations are given in @xcite and @xcite .
the n2 @xmath11 [ nii ] @xmath56583/ h@xmath4 method has been used and calibrated by several authors @xcite , and has demonstrated to work accurately estimating metallicities from 1/50th to twice the solar value @xcite .
one of the most important advantages of this method is that an extinction correction is not required because it only uses the h@xmath4 and [ nii]@xmath56583 lines , both close in wavelength .
it requires only a narrow spectral range , making it suitable for surveys of limited spectral coverage like otelo .
finally , the n2 method has demonstrated to work accurately at high redshift ( @xmath12 2.2 ) , making it suitable for detecting evolution @xcite .
the work presented here has been developed within the otelo project .
as part of the otelo preparatory activities , the aims of this study is to determine the optimal sampling and fwhm combination for the osiris tfs that allows to determine the n2 ratio , by analyzing and recovering flux errors of the h@xmath4 and [ nii]@xmath56583 lines . with the selected instrumental configuration it will be possible to deblend both lines ,
classify galaxies as star forming and agns using the n2 ratio @xcite , to estimate the star formation rates ( sfrs ) with the h@xmath4 flux ( e.g. * ? ? ?
* ) , and calculating the chemical abundances using the n2 method @xcite in star forming galaxies .
this paper is structured as follows : in sect .
2 we give a review of the osiris instrument , in sect .
3 we detail the scanning tunable imaging technique , in sect .
4 we analyze the error estimates , in sect .
5 we try our method using sdss data , and finally , in sect .
5 we give the conclusions .
osiris is the spanish day one instrument for the gtc 10.4 m telescope . with a field of view of 8.5x8.5 arcminutes and sensitive in the wavelength range from 3650 through 10000 @xmath13 ,
osiris is a multiple purpose instrument for imaging and low - resolution long slit and multiple object spectroscopy ( mos ) .
the main characteristic of osiris is the use of two tfs , one for the blue ( 3700 - 6700 @xmath14 ) , and another for the red ( 6400 - 9600 @xmath14 ) that overlap in wavelength and allow to cover most of the full osiris wavelength range @xcite .
tunable narrow - band filters , also known as fabry - perot filters ( fpfs ) , consist essentially of two glass or quartz parallel plates with flat surfaces enclosing a plane
parallel plate of air .
the inner surfaces are coated with films of high reflectivity and low absorption .
the general equation for the intensity transmission coefficient of an ideal fpf ( an airy function ) , as a function of wavelength is : @xmath15^{-1}\ ] ] where @xmath16 is the transmission coefficient of each coating , @xmath17 is the reflection coefficient , @xmath18 is the plate separation , @xmath19 is the refractive index of the medium in the cavity , usually air with @xmath20 , and @xmath21 is the angle of incidence .
the instrumental response of an ideal fpf , given by equation 1 , is periodic in wavelength and formed by airy profiles , as show in fig .
1 . see @xcite and @xcite for a detailed theory explanation .
according to the osiris characteristics , the available tf fwhm as function of wavelength span in a range from @xmath228 to @xmath2220 .osiris@xmath23tf@xmath23filter@xmath23widths ]
we can define scanning tunable imaging as taking a set of images of the same fov with the tf tuned at different contiguous wavelengths , which is similar to low resolution mos spectroscopy .
each wavelength is shifted by a certain fraction of the tf fwhm with respect to the others ( e.g. * ? ? ?
* ; * ? ? ?
* ) . as part of the preparatory activities for otelo
, we simulate the scanning using a tunable filter with different fwhm of the spectra of several hii galaxies , aimed at selecting the best combination of tunable filter fwhm and sampling .
this combination will allow deblending h@xmath4 from [ nii]@xmath56583 lines with a flux relative error lower than 20@xmath6 ( 5@xmath0 error ) , which is the maximum error for reliable sources and flux emission - line detection according to the project requirements ( see also * ? ? ?
. given the low fwhm of the tf , it will be possible to estimate the object s chemical abundances using the n2 method even for very low metallicity systems .
as a first step , we generated the response of the tf ( an airy function ) , with equation 1 , observing that a difference of at least 3 in fwhm is required for obtaining significant differences in the recovered lines - flux error .
then we perform several tests with fwhms of 6 , 9 , 12 , and 15 . however , according to the characteristics of osiris as explained in sect .
2 , a fwhm of 6 is not available , and a fwhm of 15 gives errors larger than 25@xmath6 for [ nii]/h@xmath4 , which is out of our upper limit error .
therefore , we selected fwhms of 9 and 12 @xmath13 , as shown in fig 2 . , in solid and dot - dash lines , respectively . ]
we generated synthetic spectra based on data from real hii galaxies , with h@xmath4 and [ nii]@xmath56583 lines in emission centered on 6563 and 6583 respectively , fwhm(h@xmath4)=4.7 , and [ nii]@xmath56583/h@xmath4=0.43 , which correspond to a maximum rotation velocity ( @xmath24 ) of 215 km / s .
median values of @xmath24 decrease from 300 to 220 to 175 km / s for the sa , sb , and sc types , respectively @xcite , then , our synthetic spectra are representative of spiral galaxies .
we redshifted the spectra to z=0.24 and z=0.4 , the two redshifts of the chosen atmospheric windows of 150 and 180 width , respectively .
the wavelengths at z=0.24 and z=0.4 are of 8138 & 9188 for h@xmath4 , respectively , and of 8163 & 9216 for [ nii]@xmath56583 , respectively . at redshift zero , h@xmath4 and
[ nii]@xmath56583 lines are separated by @xmath2220 . as redshift increases ,
the separation between both lines increases as 1+@xmath25 ( 25 at z=0.24 and 28 at z=0.40 ) , making it easier to deblend h@xmath4 from [ nii ] .
+ the intermediate observed redshift populations at 0.24 and 0.4 are representative of the transition from the relative quiet local universe to the starbursting universe at z@xmath221 .
for instance , galaxies at redshift 0.4 have shown lower metallicity and higher sfr than those of the local universe @xcite . to add a continuum , we used h@xmath4 ews of 5 , 10 , 20 , 30 , 40 and 50 , with the ew defined by ew@xmath26 , where @xmath27 is the continuum flux density at the h@xmath4 line , and @xmath28 is the h@xmath4 flux of the elg @xcite .
the adopted ews assure the inclusion of several morphologies and types of galaxies @xcite , as shown in table 1 . finally , to add noise to the spectra , we considered the equation of the signal to noise ( @xmath29 ) of a charge - coupled device ( ccd ) , or the @xmath30ccd equation@xmath31 @xcite : @xmath32 where @xmath33 is the total number of photons ( signal ) , @xmath34 is the number of pixels under consideration for the @xmath29 calculation , @xmath35 is the total number of photons per pixel from the background or sky , @xmath36 is the total number of dark current electrons per pixel , and @xmath37 is the total number of electrons per pixel resulting from the read out noise .
we can see from equation 2 that if the total noise for a given measurement is dominated by the first noise term , @xmath33 , the equation 2 becomes @xmath38 , which is a measurement of a single poisson behaved value .
therefore , we add a poisson noise to the simulated spectra .
we adopted a @xmath29 of 5 , which ensures the detection of the object within an error of @xmath39 .
the magnitude error of the observed object is @xmath40mag @xmath41 , which means a 0.2 magnitude error for a @xmath29 of 5 .
this procedure is valid when sky noise is not dominant .
.equivalent widths of galaxies with different morphological types .
e , sab , and scd / im from @xcite , @xcite , @xcite ; hii / bcd from @xcite ; and sy2/sy1 from @xcite . [ cols="^,^",options="header " , ]
in this work we generated spectra of typical star forming galaxies with different ews ( 5 , 10 , 20 , 30 , 40 , 50 ) , at redshifts 0.24 and 0.4 , which are the two windows of the otelo survey for the h@xmath4 line .
we convolved those spectra with the tunable filter response of the osiris instrument of fwhm of 12 and 9 , subtracting the continuum , and estimating the relative errors of the recovered h@xmath4 and [ nii]@xmath56583 fluxes .
we have concluded the following : * using an airy function with fwhm larger than 15 , the errors of the recovered fluxes are larger than @xmath22 25@xmath6 .
therefore , the convolutions were performed using a fwhm of 9 and 12 . * as a result of the convolutions , it was not possible to recover the fwhm of the h@xmath4 or [ nii]@xmath56583 lines , because the fwhm of the airy function is larger than that of those lines . however , if the fwhm of any observed line is larger or of similar size than that of the convolved function ( e.g. , quasars ) , it will be possible to recover the fwhm of the observed line through a deconvolution . in those cases , to estimate the observed flux of the line , all the data points of the pseudo spectrum will be used . * the resulting pseudo spectra show a decrement of the integrated flux at the edges , which is an effect of the limits of the wavelength window that in a real case would correspond to the wavelength limits of the order sorted used . *
the initial wavelength and the sampling interval are of the highest importance , because both will determine how near the airy function will be with respect to the observed emission line .
the estimated flux error of the detected sources will be smaller when the peak of the airy function is close to the peak of the emission line of the source . * the highest difference in wavelength of the pseudo
spectrum peak ( detected line ) with respect to the observed one , will be half of the sampling interval .
this means that for a sampling of 6 , the redshift error of the pseudo spectra will be of @xmath42x@xmath43 .
however , a fit to the pseudo spectrum would reduce the error . * as a result of the convolutions , an airy function with fwhm of 9 allows minimizing contamination by closer lines , but generates large errors when recovering the h@xmath4 and [ nii]@xmath56583 fluxes . however , the error of the line ratio is smaller than that using a fwhm of 12 .
* an airy function with fwhm of 12 produces smaller errors when recovering the fluxes of the lines .
however , it favours the cross contamination of the fluxes of both lines . also , the error of the line ratio h@xmath4/[nii]@xmath56583 is larger than that using a fwhm of 9 .
* as a result of our simulations , we concluded that the combination of an airy function of fwhm of 12 , sampling every 5 , will allow separating the h@xmath4 and [ nii]@xmath56583 emission lines with an error lower than 20@xmath6 .
however , in the appendix , the flux error estimates of the emission lines fluxes up to a sampling of 10 are also presented . *
although we selected the combination given above , according to fig .
8 , sampling every 5 , 6 , and 7 , would give acceptable errors in the flux line measurements as well .
* in order to test our method , we selected spectra from four sdss - dr7 galaxies at redshifts 0.24 and 0.4 , and convolved them with an airy function of fwhm of 12 and sampling every 5 , obtaining in all cases errors lower than 20@xmath6 , even in those cases where the [ nii]@xmath56583 line was weak ( e.g. [ nii]/h@xmath4=0.16 ) . as a result of our simulations we concluded that with the osiris s tf is possible to estimate metallicities using the n2 method in galaxies spanning a wide range of ews and morphological types , to discriminate star forming from agn galaxies , and to estimate the sfr using the h@xmath4 flux . the selected combination of tf fwhm and sampling that will allow deblending h@xmath4 and [ nii]@xmath56583 lines , and estimating their fluxes with an error lower than 20@xmath6 , is a tf fwhm of 12 and a sampling of 5 .
this work was supported by the spanish _
plan nacional de astronoma y astrofsica _ under grant aya2008 - 06311-c02 - 01 .
the sloan digital sky survey ( sdss ) is a joint project of the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , the max
institute for astronomy , princeton university , the united states naval observatory , and the university of washington .
apache point observatory , site of the sdss , is operated by the astrophysical research consortium .
funding for the project has been provided by the alfred p. sloan foundation , the sdss member institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , and monbusho .
the official sdss web site is www.sdss.org .
we thank the anonymous referee for all his / her constructive comments .
maritza a. lara - lpez is supported by a conacyt and sep mexican fellowships .
abazajian , k. n. , et al .
2009 , , 182 , 543 bland , j. , & tully , r. b. 1989 , , 98 , 723 bland - hawthorn , j. , & jones , d. h. 1998 , publications of the astronomical society of australia , 15 , 44 bland - hawthorn , j. , & jones , d. h. 1998 , , 3355 , 855 bland - hawthorn , j. , & kedziora - chudczer , l. 2003 , pasa , 20 , 242 born , m. , wolf , e. , 1999 , principles of optics , 7th expanded edition , cambridge university press cepa , j. , alfaro , e. j. , bland - hawthorn , j. , et al . 2003 , rev .
astrofis . , 16 , 64 cepa , j. , alfaro , e. j. , bland - hawthorn , j. ; et al .
2005 , rev .
astrofis . , 24 , 82 cepa , j. , aguiar , m. , castaeda , h. o. , et al . 2005 , rev .
, 24 , 1 cepa , j. , alfaro , e. j. , castaeda , h. o. , et al . 2007 , rev .
astrofis . , 29 , 168 cepa , j. , prez - garca , a. m. , bongiovanni , a. et al . 2008 , , 490 , 1 cepa , j. , et al .
in preparation denicol , g. , terlevich , r. , & terlevich , e. 2002 , , 330 , 69 erb , d. k. , shapley , a. e. , pettini , m. , steidel , c. c. , reddy , n. a. , & adelberger , k. l. 2006 , , 644 , 813 gallego , j. , zamorano , j. , rego , m. , & vitores , a. g. 1997 , , 475 , 502 gavazzi , g. , boselli , a. , cortese , l. , arosio , i. , gallazzi , a. , pedotti , p. , & carrasco , l. 2006 , , 446 , 839 gelderman , r. , woodgate , b. e. , & brown , l. w. 1995 , iau colloq . 149 : tridimensional optical spectroscopic methods in astrophysics , 71 , 89 gil de paz , a. , madore , b. f. , & pevunova , o. 2003 , , 147 , 29 gullixson , c. a. 1992 , astronomical ccd observing and reduction techniques , 23 , 130 gunn , j. e. , et al . 2006 , , 131 , 2332 jones , d. h. , & bland - hawthorn , j. 2001 , , 550 , 593 kennicutt , r. c. , jr . 1998 , , 36 , 189 kennicutt , r. c. , jr . , & kent , s. m. 1983 , , 88 , 1094 kewley , l. j. , & dopita , m. a. 2002 , , 142 , 35 kewley , l. j. , & ellison , s. l. 2008 , , 681 , 1183 lara - lpez , m. a. , cepa , j. , bongiovanni , a. , et al .
2009a , , 493 , l5 lara - lpez , m. a. , cepa , j. , bongiovanni , a. , et al .
2009b , , 505 , 529 lara - lpez , m. a. , bongiovanni , a. , cepa , j. , et al . 2010a , , 519 , a31 lara - lpez , m. a. , et al .
2010b , highlights of spanish astrophysics v , 507 liang , y. c. , hammer , f. , yin , s. y. , flores , h. , rodrigues , m. , & yang , y. b. 2007 , , 473 , 411 lowenthal , j. d. , hogan , c. j. , green , r. f. , et al .
1991 , , 377 , l73 macchetto , f. , lipari , s. , giavalisco , m. , turnshek , d. a. , & sparks , w. b. 1993 , , 404 , 511 mortara , l. , & fowler , a. 1981 , , 290 , 28 newberry , m. v. 1991 , , 103 , 122 pettini , m. , & pagel , b. e. j. 2004 , , 348 , l59 pilyugin , l. s. 2001 , , 369 , 594 pilyugin , l. s. , & thuan , t. x. 2005 , , 631 , 231 povi , m. , snchez - portal , m. , prez - garca , a. m. , et al .
2009 , , 706 , 810 rangwala , n. , williams , t. b. , pietraszewski , c. , & joseph , c. l. 2008 , , 135 , 1825 roberts , m. s. 1978 , , 83 , 1026 rubin , v. c. , burstein , d. , ford , w. k. , jr . , & thonnard , n. 1985 , , 289 , 81 sandage , a. 2000 , , 112 , 504 sofue , y. , & rubin , v. 2001 , , 39 , 137 stasiska , g. , cid fernandes , r. , mateus , a. , sodr , l. , & asari , n. v. 2006 , , 371 , 972 steidel , c. c. , adelberger , k. l. , shapley , a. e. , pettini , m. , dickinson , m. , & giavalisco , m. 2000 , , 532 , 170 thompson , d. , djorgovski , s. , & trauger , j. 1995 , , 110 , 963 veilleux , s. 2005 , rev .
astrofis . , 24 , 70 veilleux , s. , weiner , b. j. , rupke , d. s. n. , et al .
2010 , , 139 , 145 waller , w. h. 1990 , , 102 , 1217 york , d. g. , et al .
2000 , , 120 , 1579 | emission - line galaxies are important targets for understanding the chemical evolution of galaxies in the universe .
deep , narrow - band imaging surveys allow to detect and study the flux and the equivalent widths ( ew ) of the emission line studied . the present work has been developed within the context of the otelo project , an emission line survey using the tunable filters ( tf ) of osiris , the first generation instrument on the gtc 10.4 m telescope located in la palma , spain , that will observe through selected atmospheric windows relatively free of sky emission lines . with a total survey area of 0.1 square degrees distributed in different fields , reaching a 5 @xmath0 depth of @xmath1 erg/@xmath2/s and detecting objects of ew @xmath3 0.3 , otelo will be the deepest emission line survey to date .
as part of the otelo preparatory activities , the objective of this study is to determine the best combination of sampling and full width at half maximum ( fwhm ) for the osiris tunable filters for deblending h@xmath4 from [ nii ] lines by analyzing the flux errors obtained .
we simulated the otelo data by convolving a complete set of synthetic hii galaxies in ew with different widths of the osiris tfs .
we estimated relative flux errors of the recovered h@xmath4 and [ nii]@xmath56583 lines .
we found that , for the red tf , a fwhm of 12 and a sampling of 5 is an optimal combination that allow deblending h@xmath4 from the [ nii]@xmath56583 line with a flux error lower than 20@xmath6 .
this combination will allow estimating sfrs and metallicities using the h@xmath4 flux and the n2 method , respectively . |
[ [ par_intro_colloid ] ] prelude : overdamped colloidal suspensions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + dense , essentially incompressible simple liquids with conserved momentum are known to exhibit long - range correlations of the particle displacement field @xcite . as first shown in the seminal molecular dynamics ( md ) simulations by alder and wainwright @xcite ,
the coupling of displacement and momentum fields manifests itself by an algebraic decay of the velocity correlation function ( vcf ) , @xmath19 with @xmath20 being the particle velocity at time @xmath21 , @xmath22 the spatial dimension and @xmath23 the typical particle displacement with exponent @xmath15 @xcite .
interestingly , even if the momentum conservation is dropped , as justified for overdamped dense colloidal suspensions @xcite , scale - free albeit much weaker correlations of the displacement field are to be expected due to the incompressibility constraint @xcite . as illustrated in fig .
[ fig_dipole ] , the motion of a tagged colloid is coupled to the collective density dipole field @xcite , @xmath24 created by the colloid s own displacement @xmath25 at @xmath26 . after averaging over the typical displacements of the test particle and assuming a cahn - hilliard response proportional to the gradient @xmath27 of the chemical potential @xmath28 of the density field , this leads to a _ negative _ algebraic long - time decay of the vcf @xmath29 e.g. , @xmath30 in @xmath0 dimensions .
this phenomenological scaling picture agrees with mode - coupling calculations @xcite and has been confirmed computationally by means of lattice - boltzmann simulations @xcite , md simulations @xcite and even monte carlo ( mc ) simulations with local moves ( as discussed below ) @xcite . [ [ par_intro_static ] ] deviations from flory s ideality hypothesis .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this study we explore the dynamics of another complex fluid for which momentum conservation is generally believed to be irrelevant @xcite : melts of long and flexible homopolymers of length @xmath4 in @xmath0 dimensions .
following flory s ideality hypothesis " @xcite , one expects these chains to obey a gaussian statistics with a typical chain size @xmath31 where @xmath32 stands for the effective bond length of asymptotically long chains @xcite .
recently , this cornerstone of polymer physics has been challenged both theoretically and numerically for three - dimensional ( 3d ) melts @xcite , for effectively two - dimensional ( 2d ) ultrathin films @xcite and one - dimensional ( 1d ) thin capillaries @xcite .
the physical idea behind the predicted long - range correlations is related to the segmental correlation hole " , @xmath33 , of a subchain of arc - length @xmath34 of typical size @xmath35 @xcite . due to the overall incompressibility of the melt this sets an entropic penalty @xmath36 ( with @xmath8 being the total monomer density ) against bringing two subchains together @xcite . in @xmath0 dimensions , the segmental correlation hole effect is weak , @xmath37 , and a perturbation calculation
can be performed @xcite .
the detailed calculation yields , e.g. , for the intrachain angular correlations function @xmath38 @xcite an algebraic decay @xcite , @xmath39 at variance to flory s hypothesis with @xmath40 being the subchain length spanning the screening length of the density fluctuations @xcite . since @xmath41 , the angular correlation function provides a direct measure of the curvature of the mean - squared subchain size @xmath42 and eq .
( [ eq_psd3 ] ) implies that @xmath43 @xcite .
interestingly , @xmath40 is related to the isothermal compressibility @xmath44 of the solution @xcite and can thus be determined directly from the total monomer structure factor @xmath45 @xmath46 being the position of monomer @xmath47 , @xmath48 the conjugated wavevector , @xmath49 the volume of the system and @xmath50 the temperature .
( boltzmann s constant @xmath51 is set to unity throughout this paper . ) due to its definition , @xmath40 is often called dimensionless compressibility " @xcite .
remarkably , eq . ( [ eq_psd3 ] ) does no depend _ explicitly _ on @xmath40 .
this reflects the fact that the deviations arise due to the incompressibility of blobs " @xcite on scales corresponding to @xmath52 @xcite .
[ [ par_intro_aim ] ] aim of this study .
+ + + + + + + + + + + + + + + + + + naturally , eq . ( [ eq_psd3 ] ) and related findings beg the question of whether a similar interplay between the connectivity of the chains and the incompressibility of the melt may cause measurable scale - free and @xmath40-independent _ dynamical _ correlations between chains and between subchains . to avoid additional physics and to simplify the problem we focus on polymer melts where hydrodynamic @xcite and topological constraints may be considered to be negligible , as in the pioneering work by paul _ et al . _
@xcite , or are deliberately switched off @xcite .
deviations from the expected rouse - type dynamics @xcite have indeed been reported for such systems in various numerical @xcite and experimental studies @xcite .
characterizing , e.g. , the motion of the chain center - of - mass ( com ) @xmath53 by its mean - square - displacement ( msd ) @xmath2 , it was found that @xmath54 with @xmath7 being the longest ( rouse ) relaxation time and @xmath55 an empirical exponent .
this is of course at variance to the key assumption of the rouse model that the random forces acting on the monomers ( and thus on the chain ) are uncorrelated , which implies an exponent @xmath56 for all times @xcite . in this paper
we attempt to clarify this problem by means of mc simulations of a variant of the bond - fluctuation model ( bfm ) @xcite using local hopping moves which do _ not _ conserve topology @xcite and assuming a finite monomer overlap penalty @xmath57 @xcite . in analogy to our work on the angular correlation function @xmath58 @xcite ,
our analysis focuses on the com displacement correlation function @xmath59 , measuring the curvature of @xmath2 @xcite .
this allows us to probe directly the colored forces from the molecular bath @xcite acting on a tagged reference chain displaced at @xmath26 and , in turn , to elaborate the scaling theory sketched below .
[ [ par_intro_key ] ] key results .
+ + + + + + + + + + + + we demonstrate numerically that the vcf @xmath60 does indeed _ not _ vanish as it would if all the random forces acting on the chains _ were _ uncorrelated @xcite .
instead , @xmath60 is found to scale as @xmath61 with @xmath10 being a universal scaling function .
note that the postulated eq .
( [ eq_key ] ) does not depend explicitly on the compressibility of the solution .
the squared characteristic velocity " @xmath62 arises for dimensional reasons .
the prefactor @xmath63 is motivated by the correlation hole penalty , i.e. the incompressibility constraint which ultimately couples the displacements of ( sub)chains @xcite .
as one expects from eq .
( [ eq_vcfcolloid ] ) , the scaling function decays as @xmath64 with an exponent @xmath65 for @xmath16 .
we show that this long - time behavior is preceded by a much weaker algebraic decay with an exponent @xcite @xmath66 due to the much slower relaxation , @xmath13 , of the collective dipole field of subchains which was generated by the initial displacement of a tagged subchain at @xmath26 .
the gradient of the chemical potential @xmath27 pulling the reference subchain back to its original position is of course not only due to the density fluctuation of the subchain density field but also to the tensional forces along the chains caused by the displacement .
( subchain density fluctuations and tensions are coupled and associated both to a free energy fluctuation of order @xmath67 . ) since @xmath59 , it follows from eq .
( [ eq_keyomega ] ) that for sufficiently short times , such that the white forces acting on the chains are negligible , we expect to find @xmath68 which is rather similar to the fit , eq .
( [ eq_keyomega ] ) , suggested in the literature .
see refs .
@xcite and @xcite for two closely related theoretical studies .
interestingly , schweizer s mode - coupling theory approach @xcite is consistent with eq .
( [ eq_omega2beta ] ) .
we stress that the presented numerical study is necessarily incomplete since important dynamical correlations are expected to arise in more realistic models due to topological constraints and , even more importantly , due to not fully screened hydrodynamic interactions which have recently been shown to matter @xcite . [ [ par_intro_outline ] ] outline .
+ + + + + + + + the paper is organized as follows . in sec .
[ sec_algo ] the numerical algorithm is introduced and some technical details are discussed .
our computational results are presented in sec .
[ sec_comp ] where we focus on essentially incompressible melts ( sec .
[ sub_comp_jean ] ) but also discuss effects of finite excluded volume ( sec .
[ sub_comp_robust ] ) .
our results are summarized in sec .
[ sub_conc_summary ] .
the paper concludes in sec .
[ sub_conc_outlook ] with a comment on what we would expect if topology conservation is switched on again .
the classical bfm is an efficient lattice mc algorithm for coarse - grained polymer chains where each monomer occupies _ exclusively _ a unit cell of @xmath69 lattice sites on a @xmath22-dimensional simple cubic lattice @xcite .
( the fraction @xmath70 of occupied lattice sites is thus @xmath71 . )
the bfm was proposed in 1988 by carmesin and kremer @xcite as an alternative to single - site self - avoiding walk models , which retains the computational efficiency of the lattice without being plagued by ergodicity problems .
the key idea is to increase the size of the monomers and the number of bond vectors to allow a better representation of the continuous - space behavior of real polymer melts .
a widely used choice of bond vectors for the 3d variant of the bfm is given , e.g. , by all the permutations and sign combinations of the six vectors @xcite @xmath72 if only _
mc moves of the monomers to the six nearest neighbor sites are performed called l06 " moves @xcite this vector set ensures automatically that polymer chains can not cross .
( these l06-moves are represented in fig .
[ fig_algo](b ) by the filled circles . )
consequently , several authors report a reptation - type dynamics for chain lengths above @xmath73 at a melt " volume fraction @xmath74 @xcite .
.various properties for bfm polymers melts of arbitrarily large chain length @xmath4 at volume fraction @xmath75 : the dimensionless compressibility @xmath40 , the root - mean - square bond length @xmath76 , the effective bond length @xmath77 , the acceptance rate @xmath78 , the local monomer mobility @xmath79 and the self - diffusion coefficient @xmath80 .
the dynamical data refer to l26-moves to the nearest and next - nearest lattice sites .
[ tab_overlap ] [ cols="^,^,^,^,^,^,^",options="header " , ] for consistency with these studies we keep eq .
( [ eq_bondset ] ) , although the non - crossing constraint is irrelevant for us , and use the same volume fraction @xmath81 . with respect to the classical variant of the bfm in @xmath0
our algorithm differs in two important points : _ ( i ) _ as described in ref .
@xcite , we use a finite excluded volume penalty @xmath57 which has to be paid if two monomers fully overlap .
the overlap of two cube corners is sketched in fig .
[ fig_algo](a ) .
temperature is arbitrarily set to unity .
technically , the finite monomer interaction penalty is implemented using a potts spin representation of the discrete local density field @xmath82 @xcite .
the monomer overlap penalty parameter @xmath57 is in fact a laplace multiplier controlling the fluctuations of spins and imposing thus the incompressibility of the melt . by reducing the fluctuations of the spins the laplace multiplier causes thus the effective entropic forces which imply the dynamical correlations discussed in sec .
[ sec_comp ] .
this modification allows us to check at one constant volume fraction ( @xmath83 ) that the dynamical correlations do not depend _ explicitly _ on the monomer excluded volume as stated in eq .
( [ eq_key ] ) . _
( ii ) _ we use local hopping moves to the 26 next and next - nearest lattice sites ( so - called
l26 " moves ) . due to the larger jumps ,
chains cross and the dynamics is of rouse - type : the chain self - diffusion coefficient @xmath80 , e.g. , scales as @xmath84 for all chain lengths @xmath4 even if monomer overlap is disallowed ( @xmath85 ) @xcite .
although the use of l26-moves and finite monomer interactions does speed up the relaxation dynamics , it remains obviously impossible to equilibrate dense polymer solutions with chain lengths up to @xmath86 just using _ local _ hopping moves .
taken advantage of our previous studies on static properties @xcite , we have used configurations equilibrated using a mix of _ global _ slithering snake and double bridging moves together with local l26-moves .
we use periodic simulation boxes of linear dimension @xmath87 . thus at @xmath74 these systems contain @xmath88 monomers and even for chains with @xmath86 we still have @xmath89 chains . using these well characterized configurations we have computed time series over @xmath90 or @xmath91 monte carlo steps ( mcs ) using local l26-moves .
we stress that the aim is not to cover necessarily the full relaxation time @xmath7 for our larger chains , but to precisely describe the dynamical correlations which are the most pronounced at short times . as main working point
we take @xmath92 for which the chain length has been scanned up to @xmath86 . for other penalties
@xmath57 we present results obtained using @xmath93 .
some relevant properties are summarized in table [ tab_overlap ] and are represented in fig .
[ fig_static ] and fig .
[ fig_aw ] .
panel ( a ) of fig .
[ fig_static ] shows the mean - square bond length @xmath94 and the effective bond length @xmath95 for asymptotically long chains as determined in ref .
obviously , @xmath96 in the small-@xmath57 limit .
@xmath95 then increases in the intermediate @xmath57-window before it levels off at our main working point @xmath92 .
as one may expect , systems with @xmath97 can not be distinguished from systems computed using the classical bfm without monomer overlap ( @xmath85 ) .
panel ( b ) presents the dimensionless compressibility @xmath98 obtained from total static structure factor , eq .
( [ eq_gdef ] ) .
the dashed line corresponds to the asymptotics for weak interactions , @xmath99 @xcite .
the compressibility levels off for large penalties where @xmath100 ( bold line ) .
the angular correlation function @xmath38 @xcite is presented for several overlap penalties @xmath57 and one chain length @xmath93 in panel ( c ) . the theoretical prediction eq .
( [ eq_psd3 ] ) is nicely confirmed for @xmath101 by the power - law asymptote indicate by the bold line .
since in mc simulations there is no monomer mass " , no ( conserved or non - conserved ) monomer momentum " and not even an instantaneous velocity , it might at first sight appear surprising that a well - posed velocity correlation function " ( vcf ) can be defined and measured . to illustrate that this is indeed the case is the first purpose of this subsection .
the second is to verify that the negative analytic decay of the vcf expected for overdamped colloids , eq .
( [ eq_vcfcolloid ] ) , is also of relevance for dense bfm beads ( @xmath102 ) diffusing through configuration space by means of _ local _ hopping moves on the lattice .
the systems presented in fig . [ fig_bfmbeads ]
correspond to three different effective dimensions @xmath22 .
the effectively 1d systems ( @xmath103 ) have been obtained by confining the beads to a thin capillary of square cross - section , the 2d systems ( @xmath104 ) by confining the beads to a thin slit as shown in fig .
[ fig_algo](c ) .
the distance @xmath105 between parallel walls allows the free crossing of the beads .
( for the 1d case this is crucial , of course . )
the data has been obtained for beads without monomer overlap ( @xmath85 ) by means of l26-moves on a 3d cubic lattice for a volume fraction @xmath81 .
we average over the @xmath106 beads contained in each configuration .
one standard measure characterizing the monomer displacements is the msd @xmath107 displayed in the inset of fig .
[ fig_bfmbeads ] ( with @xmath108 being the particle position at time @xmath21 ) .
as one expects , the displacements become uncorrelated for @xmath109 , i.e. @xmath110 with @xmath111 being the monomer self - diffusion constant : @xmath112 for @xmath103 , @xmath113 for @xmath104 , @xmath114 for @xmath0 .
the typical displacement @xmath115 thus scales as @xmath23 with @xmath116 . however , small deviations are clearly visible for short times if @xmath117 is plotted in log - linear coordinates .
the deviations are particulary strong for @xmath103 .
the effective random forces acting on the beads are thus not completely white . obviously , one might try to characterize these deviations by fitting various polynomials to the measured msd . since one needs to substract , however , the huge free diffusion contribution from the measured signal to obtain tiny deviations this is a numerical difficult if not impossible route . in analogy to the static angular correlation function @xmath58 allowing to make manifest deviations from the gaussian chain assumption , it is numerically much better to _ directly _ compute the second derivative of @xmath118 with respect to time .
how this can be done is illustrated in the main panel of fig .
[ fig_bfmbeads ] .
we sample equidistant series of configurations at time intervals @xmath119 as indicated by the open symbols .
each time series contains @xmath120 configurations .
averaging over all possible pairs of configurations @xmath121 we compute the displacement correlation function @xmath122 , i.e. a four - point correlation function of the monomer trajectories with @xmath123 being the monomer displacement vector at time @xmath21 in a time interval @xmath124 . by construction
@xmath125 if both displacement vectors are uncorrelated .
note that if one computed times @xmath21 shorter than @xmath124 , both displacement vectors would become trivially correlated , since they describe in part the same particle trajectory .
hence , @xmath126 and @xmath127 ( not shown ) .
more importantly , for times @xmath128 one has @xmath129 as one readily sees by applying finite - difference operators with respect to time to the monomer msd .
as can be seen for @xmath103 , the @xmath124-dependence drops indeed out for @xmath130 .
we thus often avoid the second index @xmath124 and write @xmath131 for the displacement correlation function .
obviously , the statistics deteriorates for large @xmath132 where fewer configuration pairs contribute to the average ( taking apart that the signal itself decays ) .
it is for this reason that we need a hierarchy of time series of different @xmath124 .
taking for each time series only the first decade of data ( @xmath133 ) , these reduced data sets are pasted together and averaged logarithmically .
these logarithmic cummulants are given for each dimension and compared to the exponents @xmath134 ( thin line ) , @xmath135 ( dashed line ) and @xmath30 ( bold line ) predicted by eq .
( [ eq_vcfcolloid ] ) for @xmath136 , @xmath137 and @xmath0 , respectively .
the data agrees over more than two orders of magnitude in time with the prediction , especially for @xmath103 . if one is satisfied with less orders of magnitude it is sufficient to check the exponents using just a time window @xmath138 as can be seen for the open spheres .
the superposition of data from time series with different @xmath124 is just a numerical trick which reduces the number of configurations to be stored and the number of configuration pairs to be computed for a given time @xmath21 .
having shown that scale - free dynamical correlations exist for dense bfm beads as expected for overdamped colloids ( fig . [ fig_dipole ] ) , we turn now our attention to 3d melts of long and flexible homopolymers .
we focus here on systems with finite monomer overlap penalty @xmath139 and volume fraction @xmath81 .
due to the use of the finite overlap penalty and the topology non - conserving local l26-moves , the dynamics is indeed of rouse - type as can be seen from the various msd presented in fig .
[ fig_msd_jean ] for chains of length @xmath140 ( diamonds ) and @xmath86 ( spheres ) .
the monomer msd @xmath118 is indicated by the open symbols .
note that for the large chain lengths sampled here it is inessential whether this average is performed over all monomers ( as in sec .
[ sub_comp_beads ] ) or only over a few monomers in the ( curvilinear ) center of chains at @xmath141 as in refs .
as expected from the rouse model we obtain for short times @xcite @xmath142 as indicated by the dashed line , i.e. the typical displacement increases as @xmath143 with @xmath144 . as can be seen for @xmath140 , the monomers diffuse again freely with a power - law slope @xmath145 for times larger than the rouse time @xmath146 .
we remind that it was not the aim of the present work to sample for our larger chains ( @xmath147 ) over the huge times needed to make this free diffusion regime accessible . following paul _
@xcite , the short - time power law , eq .
( [ eq_msdmonshort ] ) , can be used to determine the effective monomer mobility .
we obtain @xmath148 for our main working point .
mobilities for other penalties are listed in tab .
[ tab_overlap ] . as may be seen from fig .
[ fig_aw ] , @xmath149 decays with increasing excluded volume just as the acceptance rate @xmath150 ( spheres ) of the hopping attempts , but the decay is even more pronounced for @xmath149 , i.e. increasingly more accepted moves do not contribute to the effective motion ( cage effect " ) @xcite .
the full symbols displayed in fig .
[ fig_msd_jean ] refer to the msd @xmath2 of the chain com defined in eq .
( [ eq_betadef ] ) . as one expects for rouse chains , the amplitude of @xmath2 decreases inversely with @xmath4 and the diffusion appears to be uncorrelated ( @xmath151 ) at least for times @xmath152 as indicated by the vertical arrow .
obviously , the center - of - mass msd @xmath2 and the monomer msd @xmath118 merge for times beyond the rouse time ( @xmath153 ) .
fortunately , since @xmath2 becomes linear for @xmath154 , it is always possible , even for our largest chain length @xmath4 , to measure the chain self - diffusion coefficient @xmath80 by plotting @xmath155 in log - linear coordinates . for our main working point
we thus obtain @xmath156 for all @xmath4 .
values for other @xmath57 are again given in tab .
[ tab_overlap ] and are represented in fig .
[ fig_aw ] ( diamonds ) . that these values are consistent with rouse dynamics can be checked by comparing the measured diffusion coefficients with the values obtained using the local mobilities @xmath79 according to @xcite @xmath157 as shown by the dashed line in fig .
[ fig_aw ] , eq .
( [ eq_w2dn ] ) agrees well with the directly measured diffusion coefficients .
similarly , it is possible ( at least for our shorter chains ) to measure the longest rouse relaxation time @xmath146 by an analysis of the rouse modes and to compare it with @xmath158 @xcite .
we obtain again a nice agreement between directly and indirectly computed relaxation times ( not shown ) . up to now
we have insisted on the fact that our systems are to leading order of rouse type and we have characterized them accordingly .
however , deviations from the rouse picture are clearly revealed for short times , especially for @xmath2 in agreement with the literature @xcite .
( that the monomer msd @xmath118 also deviates for very short times is due to the lattice model , i.e. to a trivial lower cut - off effect associated to the discretization . ) in agreement with eq .
( [ eq_betadef ] ) , the short - time com motion can be characterized by an effective exponent @xmath55 ( bold lines ) . since in our bfm version topological
constraints are irrelevant , this confirms the finding by shaffer @xcite that the deviations found for short chains using the classical bfm algorithm with topological constraints @xcite can not alone be attributed to precursor effects to reptational dynamics ( as discussed in sec .
[ sub_conc_outlook ] ) .
before we turn to the more precise numerical characterization of these deviations by means of the associated displacement correlation function , let us ask whether the observed colored forces acting for short times on the com of the entire chain are also relevant on the scale of subchains of arbitrary arc - length @xmath34 ( @xmath159 ) . to answer this question
we compute the msd displacement @xmath160 associated to the subchain center - of - mass @xmath161 as shown in fig .
[ fig_msd_jean ] for subchains of length @xmath162 in the middle of total chains of length @xmath86 ( crosses ) . since for short times
the subchain does not know " that it is connected to the rest of the chain , one expects it to behave as a total chain of the same length ( @xmath163 ) .
this is indeed borne out by our data which are well described by @xmath164 for all chain length @xmath4 and subchain length @xmath34 studied .
the subchains reveal thus for sufficiently short times the _ same _ colored forces ( @xmath165 ) as the total chain as can be clearly seen from the example given in fig .
[ fig_msd_jean ] . for larger times , @xmath166 ,
the subchain becomes aware " that it is connected to the rest of the chain and gets enslaved by the monomer msd .
we thus observe @xmath167 and , obviously , @xmath168 for even larger times @xmath153 .
two comments are in order here .
first , it should be noticed that eq .
( [ eq_msdlocality ] ) expresses the fact that the effective forces acting on the @xmath170 subchains of length @xmath34 in a chain of total length @xmath4 add up _ independently _ to the forces acting on the total chain . in this sense eq .
( [ eq_msdlocality ] ) states that the deviations from the rouse picture must be _ local _ @xcite .
we will explicitly verify this below ( fig .
[ fig_vcf_jeansub ] ) .
second , if one chooses following eq .
( [ eq_ctht ] ) an arbitrary time window @xmath124 to characterize the displacement correlations , this corresponds to dynamical blobs containing @xmath171 adjacent monomers , which must move together due to chain connectivity .
( [ eq_msdcmsintermediate ] ) implies now that the dipole field @xcite associated with the com of these @xmath34-subchains and created at @xmath26 by a tagged @xmath34-subchain must decay according to a typical displacement @xmath172 with @xmath144 .
it is this exponent @xmath169 which is mentioned in the introduction , eq .
( [ eq_keyomega ] ) . since this exponent is smaller than for freely diffusing colloids ( @xmath116 )
, the dipole field of the com of the subchains must decay more slowly and one expects , accordingly , a much weaker decay of the associated vcf .
following the numerical strategy used in sec .
[ sub_comp_beads ] for bfm beads , we characterize now in more detail the dynamical correlations already visible from the chain msd @xmath2 and the subchain msd @xmath173 by computing directly their second derivative with respect to time , i.e. the displacement correlation functions @xmath1 ( figs .
[ fig_vcf_jeanconstruct ] , [ fig_vcf_jean ] and [ fig_vcf_jeantau ] ) and @xmath174 ( fig . [ fig_vcf_jeansub ] ) .
the displacement correlation function @xmath175 for the displacement vector @xmath176 of the chain com @xmath53 is shown in fig .
[ fig_vcf_jeanconstruct ] for two chain lengths , @xmath177 ( top ) and @xmath86 ( bottom ) .
averages are again performed over all configuration pairs @xmath178 possible in the set of @xmath120 configurations sampled for each @xmath124 indicated . as in sec .
[ sub_comp_beads ] we find that @xmath179 for @xmath128 .
for clarity , only the data subset is indicated which is used to construct the cummulated final vcf @xmath60 ( as shown below in fig .
[ fig_vcf_jean ] ) .
the bold lines represent the predicted short - time exponent @xmath180 .
we emphasize that this exponent can be observed for @xmath86 over nearly five orders of magnitude .
for @xmath177 we also indicate the exponent @xmath30 expected for times @xmath153 where the chains should behave as colloids according to eq .
( [ eq_vcfcolloid ] ) .
the magnitude of the signal decreases strongly with @xmath4 , which together with the fact that fewer chains per box are available makes the determination of the vcf more difficult with increasing chain length .
the @xmath4-dependence of the vcf for @xmath18 is further analysed in fig .
[ fig_vcf_jean ] where we plot the reduced vcf @xmath181 _ vs. _ the reduced time @xmath182 using the effective monomer mobility @xmath79 and the diffusion coefficient @xmath80 determined above .
this scaling makes the axes dimensionless and rescales the vertical axis by a factor @xmath4 . as shown by the successful data collapse for chain lengths ranging from @xmath177 up to @xmath86 on the power - law slope indicated by the bold line , the vcf scales exactly as @xmath183 for @xmath18 .
this confirms the already stated locality " of the correlations @xcite , eq .
( [ eq_msdlocality ] ) , i.e. the effective forces acting on subchains add up independently to the forces acting on the entire chain .
we have still to motivate the precise form used for the rescaling of the axes . according to eq .
( [ eq_key ] ) we claim that the vcf @xmath60 scales as a function of the reduced time @xmath184 . substituting the typical chain size @xmath31 and the chain relaxation time @xmath185 and using @xmath180 this yields @xmath186 with @xmath187 being an empirical dimensionless constant .
the bold slope indicated in the plot corresponds to a value @xmath188 .
interestingly , since @xmath1 , it follows from eq .
( [ eq_vcf_wt ] ) that @xmath189 as may be seen from fig .
[ fig_msd_jean ] ( stars ) for @xmath86 , eq .
( [ eq_ct2msd ] ) with @xmath190 provides an excellent fit of the measured @xmath2 .
we also note that the second term in eq .
( [ eq_ct2msd ] ) dominates the dynamics for @xmath191 .
this is indicated by the vertical arrow in fig .
[ fig_msd_jean ] .
hence , for @xmath192 the stars correspond to an exponent @xmath193 , eq . ( [ eq_omega2beta ] ) , being close to the phenomenological exponent @xmath55 which motivated our study .
the central advantage of computing the vcf @xmath60 lies in the fact that it allows us thus to make manifest that ( negative algebraic ) deviations from the rouse behavior exist for _ all _ times and not just for @xmath192 . returning to our discussion of fig .
[ fig_vcf_jean ] we emphasize that the vcf of shorter chains decays more rapidly for large times following roughly the dashed line corresponding to the exponent @xmath30 expected for effective colloids .
we verify in fig .
[ fig_vcf_jeantau ] that this bending down of the vcf is described by the announced scaling in terms of a reduced time @xmath194 and a vertical axis @xmath195 using the amplitude @xmath196 stated in eq .
( [ eq_key ] ) . as shown by the data collapse ,
there is only one relevant time scale in this problem , namely the chain relaxation time @xmath146 , for which the deviations for short times , where polymer physics matters ( @xmath180 ) , and for large times , where polymer chains behave as effective colloids ( @xmath30 ) , nicely match .
it is worthwhile to emphasize that eq .
( [ eq_key ] ) together with the locality of the deviations , @xmath183 , immediately imply the exponent @xmath197 .
this can be seen by counting the powers of the chain length @xmath4 , @xmath198 which implies @xmath199 in agreement with eq .
( [ eq_keyomega ] ) and the numerically observed time dependence . assuming eq .
( [ eq_key ] ) , the exponents for @xmath4 and @xmath21 thus contain the same information .
the subchain vcf @xmath200 presented in fig .
[ fig_vcf_jeansub ] has been obtained as the total chain vcf @xmath60 , the only difference being that the com @xmath161 of the subchain defines now the displacement vector @xmath201 .
the data presented in the main panel is rescaled as in fig .
[ fig_vcf_jean ] with @xmath202 setting now the diffusion constant . for small times
@xmath203 , all data sets collapse on the same slope with exponent @xmath180 ( bold line ) as in eq . ( [ eq_vcf_wt ] ) .
this confirms that @xmath204 in agreement with eq .
( [ eq_msdlocality ] ) .
this shows that the same deviations occur for arbitrary subchains and that the colored forces acting on the subchains add up independently to the effective forces acting on the total chain .
since for intermediate times @xmath205 the subchains are enslaved by the monomer motion according to eq .
( [ eq_msdcmsintermediate ] ) , it follows using eq .
( [ eq_ctht ] ) that @xmath206 ( dash - dotted line ) .
this scaling can be better seen in the inset of fig .
[ fig_vcf_jeansub ] where we plot @xmath207 as a function of @xmath208 setting @xmath209 and @xmath210 .
this allows to bring to a nice collapse the subchain vcf for @xmath16 .
the dash - dotted line corresponds to the expected decay @xmath211 for intermediate times .
we have focused above on one specific choice of operational parameters . in order to check whether our key predictions eq .
( [ eq_key ] ) and ( [ eq_keyomega ] ) remain valid more generally under different conditions , additional simulations have been performed which we summarize here : _ ( i ) _ switching to the l06-moves of the original bfm algorithm , but keeping the same static conditions ( @xmath81 , @xmath139 ) , the same scaling is obtained for the msds and the vcfs as before ( not shown ) . using the directly measured local monomer mobility @xmath212 and the self - diffusion coefficient
@xmath213 we get for the vcf exactly the same scaling as in fig .
[ fig_vcf_jean ] .
this is consistent with the fact that eq .
( [ eq_key ] ) does not depend explicitly on the specific local dynamics . _
( ii ) _ using again l26-moves we have varied the overlap penalty @xmath57 at constant chain length @xmath93 and volume fraction @xmath81 . since eq . ( [ eq_key ] ) does only depend _
implicitly _ on @xmath40 , this suggest that eq .
( [ eq_vcf_wt ] ) for times @xmath18 should also remain valid if one uses for rescaling of the axes the measured effective bond length @xmath95 , the mobility @xmath149 and the diffusion constant @xmath214 indicated in tab .
[ tab_overlap ] .
this is in fact borne out for the asymptotic behavior of the rescaled vcf displayed in fig .
[ fig_vcf_overlap ] . in analogy to the angular correlation function @xmath38 for soft melts presented in panel ( c ) of fig .
[ fig_static ] , the deviations from eq .
( [ eq_vcf_wt ] ) for short times are expected since the incompressibility constraint is only felt if distances corresponding to the static screening length are probed . _
( iii ) _ a similar plot has been obtained for @xmath60 if we change the dimensionless compressibility by varying the volume fraction @xmath70 at constant chain length @xmath93 and overlap penalty @xmath139 ( not shown ) .
as long as the chains remain sufficiently entangled , all data merge for large times on the asymptotic behavior of incompressible melts , eq .
( [ eq_vcf_wt ] ) , for @xmath215 .
the incompressibility constraint of polymer melts is known to restrict the fluctuations of ( sub)chains and generates thus scale - free _ static _ deviations from the gaussian chain statistics @xcite . in this paper
we addressed the question of whether the incompressibility constraint also causes _ dynamical _ correlations of the ( sub)chain displacements . to avoid additional correlations
we have supposed that the dynamics is locally perfectly overdamped ( no momentum conservation " ) and that the chains may freely intercept ( no reptation " ) , i.e. the relaxational dynamics is assumed to be to leading order of rouse type @xcite .
we have shown that the above conditions may be realized computationally by a variant of the bfm using finite monomer excluded volume penalties and local topology non - conserving mc moves ( fig .
[ fig_algo ] ) . sampling chain lengths up to @xmath86 allowed us to carefully check the @xmath4-scaling of the deviations ( figs .
[ fig_vcf_jean ] , [ fig_vcf_jeantau ] ) .
such deviations are visible from the short - time scaling of the com msd @xmath2 presented in fig .
[ fig_msd_jean ] confirming published computational and experimental work @xcite .
we have shown that a better characterization of these deviations can be achieved by means of the com displacement auto - correlation function @xmath216 .
computing directly the curvature of @xmath2 this removes the large free diffusion contribution to the chain motion and allows to focus on the correlated random forces which cause the second term contributing to @xmath2 according to eq .
( [ eq_ct2msd ] ) .
how such a velocity correlation function " can be computed within a mc scheme has been first illustrated for dense bfm beads ( fig .
[ fig_bfmbeads ] ) confirming the negative algebraic decay expected for overdamped colloids , eq .
( [ eq_vcfcolloid ] ) . the observed exponent @xmath217 is expected @xcite due to the coupling of a tagged colloid to the gradient of the collective density dipole field ( fig . [ fig_dipole ] ) decaying in time by free diffusion ( @xmath116 ) .
as shown in fig .
[ fig_vcf_jeantau ] , the same exponents ( dashed lines ) characterize the motion of polymer chains for large times ( @xmath153 ) where the chains behave as effective colloids . more importantly , we have demonstrated that the observed short - time deviations for @xmath2 , eq .
( [ eq_betadef ] ) , can be traced back to the negative analytic decay of the correlation function , @xmath218 for @xmath18 with @xmath219 in @xmath0 ( fig .
[ fig_vcf_jean ] ) in agreement with eq .
( [ eq_keyomega ] ) @xcite .
that the correlation functions decays inversely with mass shows that the process is _ local _
@xcite , i.e. the displacement correlations of subchains add up independently ( fig . [ fig_vcf_jeansub ] ) .
assuming according to the key scaling relation eq .
( [ eq_key ] ) the relaxation time @xmath146 to be the only characteristic time , both asymptotic regimes can be brought to a data collapse ( fig .
[ fig_vcf_jeantau ] ) .
the short - time exponent @xmath220 is implied by eq .
( [ eq_key ] ) and the locality of the correlations , eq .
( [ eq_ncounting ] ) .
a deeper insight is obtained by generalizing the well - known displacement correlations of overdamped colloids ( fig .
[ fig_dipole ] ) to the displacement field of subchains of length @xmath221 with @xmath124 being the time window used to define the displacements .
since subchains repel each other due to the incompressibility constraint , a tagged subchain is pulled back to its original position by the subchain dipole field @xcite . since for times
@xmath222 the relevant dipole field decays much slower than for colloids ( @xmath223 ) , the correlations are much more pronounced . in this study
we have deliberately tuned our model to avoid topological constraints .
obviously , these constraints are expected to matter for the dynamics of real polymer melts @xcite .
it is thus of interest to see how the presented picture changes if topology is again switched on by using the topology conserving l06-moves of the classical bfm algorithm .
preliminary data comparing systems of chain length @xmath86 is presented in fig .
[ fig_rept ] .
the dynamics is still of rouse - type at small overlap penalties as can be seen for @xmath139 .
with increasing @xmath57 the crossing of the chains gets more improbable and the topological constraints become more relevant .
this can be seen for @xmath224 ( diamonds ) and even more for @xmath97 ( triangles ) .
the latter data set clearly approaches the power - law slope @xmath225 ( dashed line ) expected from reptation theory for times larger than the entanglement time @xmath226 . the vertical arrow indicates a value for this time obtained from an analysis of @xmath118 .
apart from the much larger chain length used , our data for @xmath97 is consistent with the results obtained using the classical bfm with @xmath85 @xcite . as in fig .
[ fig_msd_jean ] the bold line represents the effective exponent @xmath227 , eq .
( [ eq_betadef ] ) .
superficially , it does a better job for systems with conserved topology due to the broad crossover to the entangled regime . however , this apparent exponent is by no means deep as revealed by the displacement correlation function @xmath60 plotted in the inset . for short times
_ all _ systems are well - described by the _
exponent @xmath180 ( bold line ) in agreement with eq .
( [ eq_keyomega ] ) .
note that the final points given for @xmath97 decay slighly more rapidly .
unfortunately , neither the length of the analyzed time series nor the precision of our data does currently allow us to show that @xmath228 as expected for reptating chains with @xmath229 being the tube diameter @xcite .
much longer time series are currently under production to clarify this issue .
the numerical demonstration is obviously challenging , since the difference between the exponents @xmath230 and @xmath231 is rather small . in any case
it is thus due to the dynamical correlations first seen in the bfm simulations of paul _ et al .
_ @xcite that the crossover between rouse and reptation regimes becomes broader and more difficult to describe than suggested by the standard rouse - reptation theory @xcite which does not take into account the ( static and dynamical ) correlations of the composition fluctuations imposed by the incompressibility constraint @xcite .
we thank s.p .
obukhov ( gainesville ) and a.n .
semenov ( strasbourg ) for helpful discussions .
acknowledges the miur ( italian ministry of research ) for support within the program
incentivazione alla mobilit di studiosi stranieri e italiani residenti allestero " , and p.p . a grant by the irtg soft matter science " . | by monte carlo simulations of a variant of the bond - fluctuation model without topological constraints we examine the center - of - mass ( com ) dynamics of polymer melts in @xmath0 dimensions .
our analysis focuses on the com displacement correlation function @xmath1 , measuring the curvature of the com mean - square displacement @xmath2 .
we demonstrate that @xmath3 with @xmath4 being the chain length ( @xmath5 ) , @xmath6 the typical chain size , @xmath7 the longest chain relaxation time , @xmath8 the monomer density , @xmath9 the self - density and @xmath10 a universal function decaying asymptotically as @xmath11 with @xmath12 where @xmath13 for @xmath14 and @xmath15 for @xmath16 .
we argue that the algebraic decay @xmath17 for @xmath18 results from an interplay of chain connectivity and melt incompressibility giving rise to the correlated motion of chains and subchains . |
the formation of envelope solitons , through the nonlinear interaction of high - frequency ( hf ) electric fields and low - frequency ( lf ) ion - acoustic waves ( iaws ) is one of the most interesting features , and is extensively studied problem in the context of turbulence in modern plasma physics .
formation of these structures not only plays a role in turbulence , but also for plasma heating , particle transport etc . , and so their properties are of fundamental interest . in plasmas ,
such envelope solitons are well - known langmuir solitons that are hf langmuir waves ( lws ) trapped by the density troughs associated with lf iaws ( see e.g. , refs .
@xcite ) . during the past few years
several attempts have been made to investigate the dynamics of solitons including dressed solitons @xcite , soliton collapse @xcite , nucleation of cavitons @xcite , radiation of langmuir solitons @xcite , landau damping in partially incoherent lws @xcite etc .
an experimental observation for the excitation of iaws has also been reported in a langmuir turbulence regime by prado _
et al _ @xcite .
moreover , an increasing interest has also been found to study the langmuir turbulence as a result of chaos @xcite .
when the electric field intensity is strong enough to reach the modified decay instability threshold , the interaction between lws and iaws results to ` weak turbulence ' , and then the lws are scattered off iaws under @xmath0 where @xmath1 is the electron ( ion ) temperature . on the other hand ,
when the field intensity is so strong that the modulational instability ( mi ) threshold is exceeded , the lws are then essentially trapped by the density cavities associated with iaws .
such phenomenon can occur frequently in plasmas .
the interaction is then said to be in a ` strong turbulence ' regime , in which transfer or redistribution of energy to higher harmonic modes with small wavelengths may be possible .
in general , it is believed that some sort of chaotic process may be responsible for the energy transfer from large to small spatial scales @xcite .
this energy transfer may become faster when the chaotic process is well developed in a subsystem of the full set of equations @xcite .
one of the most important models in this context is the so - called zakharov equations ( zes ) @xcite , which have been derived by assuming a quasineutrality as well as neglecting the ion - wave nonlinearity ( iwn ) due to density fluctuation .
however , the effects of dispersion due to charge separation ( deviation from the quasineutrality ) may be important in the radiation of iaws , i.e. , when the lws resonantly interact with iaws at the same speed @xcite .
on the other hand , when wave amplitude becomes relatively large , the iwn can be important in the enhancement of ion - acoustic wave emission as well as in the excitation of many unstable harmonic modes at the initial stage of interaction , which later turns out to spatiotemporal chaos ( stc ) .
the basic purpose of this work is to investigate numerically the dynamics of coupled lws and iaws in presence of these important effects , and also to modify the results of langmuir solitons not observed in the one - dimensional ( 1d ) zes .
we show that transfer of energy to fewer modes , indeed , occurs and it is faster when the chaotic process in a low - dimensional model of the full system is well developed by the iwn .
such low - dimensional model is constructed on the basis of three unstable harmonic modes @xcite , and which depends on a particular regime of the wave number of excitation .
numerical simulation of this low - dimensional model reveals that transition from order to chaos is possible , when the largest length scale of the iaws is greater than that of the soliton characteristic size .
solitons will then much be distorted by the iaw emission until a mostly chaotic state emerges . for details of the idea of constructing such model ,
readers are referred to the work in ref .
@xcite .
since the low - dimensional model is restricted to a few number of modes , and also it is valid for a certain vale of the wave number of excitation , one needs to investigate , e.g. , numerically the the full space - time problem , especially when this wave number is small enough to excite many unstable modes .
we show that these solitary patterns , thus formed , ultimate lose their strengths after a long time through the random collision and fusion under the strong iaw emission .
the stc state of the system is then said to emerge . however , the mechanism leading to such stc is still not clear .
a few works in this direction can be found in the literature @xcite .
the paper is organized as follows . in sec .
ii , we present the governing equations , and analyze the evolution of langmuir solitons in some particular cases of interest . a low - dimensional three - wave model is formed and its linear stability analysis is carried out in sec .
numerical results for the temporal evolution of the low - dimensional model and the spatiotemporal evolution of the full system are presented in sec .
iv and sec .
v respectively .
finally , sec .
vi contains some discussion and conclusion .
the nonlinear interaction of hf lws and lf iaws is governed by the following coupled set of equations @xcite @xmath2@xmath3 where @xmath4 is the slowly varying langmuir wave electric field normalized by @xmath5 with @xmath6 denoting the equilibrium number density , @xmath7 is the boltzmann constant , @xmath8 is the square root of the electron - ion mass ratio and @xmath9 is the plasma relative density fluctuation .
the time @xmath10 is normalized by the ion plasma period @xmath11 and the space @xmath12 by the electron debye radius @xmath13 in eq .
( [ e2 ] ) the terms with fourth - order mixed derivatives describe the effects due to charge separation ( i.e. , deviation from the quasineutrality ) .
the third and fourth terms in the right - hand side of eq .
( [ e2 ] ) arise due to the expansion of @xmath14 up to the second order of @xmath15 in the equation for lf perturbation of electrons@xmath16 this enforces us to assume that the relative density fluctuation associated with iaws is not so small@xmath16 we note that the analytic treatment of the eqs .
( [ e1 ] ) and ( [ e2 ] ) without iwn has been studied in detail for the modification of langmuir solitons by karpman _
et al _ @xcite .
however , when the iwn is included in the system , the soliton properties may be changed . to look for this effect
, we simply disregard the mixed derivatives , and solve numerically the system of eqs .
( [ e1 ] ) and ( [ e2 ] ) by runge - kutta scheme . in the scheme ,
we approximate the spatial derivatives by the centered second - order difference formulas , and use @xmath17 at the boundaries@xmath16 the results are presented in fig.1 . when the ion - acoustic radiation is weak , we see that around @xmath18 the ponderomotive force dominates over the iwn , giving rise an increased amplitude of the envelope ( highly correlated with the density depletion )
. however , in the other regimes of @xmath12 , iwn dominates and makes the solitons wider .
later , we will see that as time progresses , the iaw emission increases by the iwn , causing solitons to lose their strengths . .
the solid and dashed ( red ) lines respectively correspond to the case with iwn and without iwn after @xmath19.,width=336,height=336 ] some particular cases can be of interest in the context of langmuir solitons .
for example , under the small amplitude condition , i.e. , disregarding the third and fourth terms in the right - hand side of eq .
( [ e2 ] ) , and assuming the quasineutrality ( disregarding the mixed fourth order derivatives ) , the dynamics of small amplitude iaws are then governed by @xmath20 this equation when coupled with eq .
( [ e1 ] ) describes under suitable renormalizations the classical zes @xcite , which has been extensively studied in the context of langmuir wave turbulence ( see e.g. , @xcite ) .
again , assuming a slow space - time response to the perturbations as @xmath21 i.e. , ion perturbations move with the ion - acoustic speed , @xmath22 , and asuming the dependent variables to vary@xmath23 the iaws can be described by the following modified korteweg - de vries ( mkdv ) equation @xcite @xmath24 the equation ( [ e1 ] ) together with eq .
( [ e4 ] ) describe the dynamics of slowly varying langmuir wave electric field coupled to the slow response of the density perturbations .
these coupled equations predict a class of solutions with a bell - shaped langmuir wave electric field profile arrested into the density cavity @xcite .
we now turn out to the global behaviors of the system . to this end , we first renormalize the original variables ( similar to the zes ) as @xmath25 and obtain the condition for mi as given below .
thus , eqs .
( [ e1 ] ) and ( [ e2 ] ) reduce to @xmath26@xmath27 clearly , the term proportional to @xmath28 becomes smaller and can be neglected .
however , the terms proportional to @xmath29 can no longer be neglected when the soliton attenuation becomes relatively larger .
next , the linear stability analysis of the perturbations @xmath30 @xmath31 [ where @xmath32 @xmath33 is the wave number ( frequency ) of modulation ] for eqs .
( [ e5 ] ) and ( [ eq6 ] ) gives the following growth rate of mi .
@xmath34 @xmath35 and the mi sets in for @xmath32 satisfying @xmath36
we note that the system of eqs .
( [ e5 ] ) and ( [ eq6 ] ) is , in general , multi - dimensional .
however , there may be the situation in which a few number of modes is more active than the remaining modes .
such cases are quite typical not only for langmuir wave turbulence , but are rather general for parametric instabilities of hf waves interacting with lf ones close to the instability threshold . in this way ,
galerkin expansions and truncations to a few normal modes are commonly used to describe the basic features of the full dynamics by a low - dimensional model .
however , we will see that the specific details of such model are highly dependent on the range for the basic wave number of modulation @xmath32 .
so , considering the dynamics of few coupled waves , we expand the envelope @xmath4 and the density @xmath37 as @xcite @xmath38 where @xmath39 $ ] represents the number of modes , @xmath40 .
since in our case , @xmath41 @xmath42 etc . , we can express the fields as ( see for details @xcite ) @xmath43 where @xmath44 @xmath45 is the conserved plasmon number .
also , @xmath46
@xmath47 @xmath48 and @xmath49 are each a function of time @xmath50 . inserting eqs .
( [ e8 ] ) and ( [ e9 ] ) into the eqs .
( [ e5 ] ) and ( [ eq6 ] ) , and following , e.g. , refs .
@xcite we obtain the following temporal system @xmath51@xmath52@xmath53 where the ` dot ' represents differentiation with respect to @xmath54 @xmath55 and @xmath56 @xmath57 and @xmath58 are given by @xmath59 here the constants @xmath60 are @xmath61 the system of eqs .
( [ e10])-([e12 ] ) can be recast as a set of four ordinary differential equations ( odes ) : @xmath62 where @xmath63 are now functions of @xmath64 @xmath65 and we redefine the variables as @xmath66 the basic properties of the system can now be in order .
it can be shown that the adiabatic limit of eqs .
( [ eq15])-([e18 ] ) gives an integrable system , and hence nonchaotic @xcite .
next , the local stability of the system about the fixed points @xmath67 given by @xmath68 @xmath69@xmath70 where @xmath71 can be examined by linearizing the eqs .
( [ eq15])-([e18 ] ) , and solving the corresponding eigenvalue problem .
the linearized set of equations in a neighborhood of @xmath72 is @xmath73 where @xmath74 @xmath75 $ ] is a column vector and @xmath76 is the jacobian matrix at @xmath77 , i.e. , @xmath78{cccc}% 0 & -p_{0 } & 0 & 0\\ q_{0 } & 0 & -r_{0 } & 0\\ 0 & 0 & 0 & 1\\ a_{1 } & a_{2 } & a_{3 } & 0 \end{array } \right ) , \label{e22}%\ ] ] the eigenvalue equation @xmath79 then reads @xmath80 @xmath81 where @xmath82 @xmath83 @xmath84
we numerically examine a set of fixed points satisfying eqs .
( [ e19 ] ) and ( [ e20 ] ) , and the linear stability of the system by the routh - hurwitz ( rh ) criteria . to this end , we use carden s method to look for a real solution of eq .
( [ e20 ] ) when expressed it as a cubic in @xmath85 .
the range of @xmath32 and @xmath44 for such real @xmath86 and hence for @xmath87 are @xmath88 and @xmath89 thus , for a given set of values of @xmath32 and @xmath44 , one can obtain a set of fixed points from eqs .
( [ e19 ] ) and ( [ e20 ] ) . having obtained a set of fixed points , one can then numerically examine the coefficients of the depressed quartic [ eq .
( [ e23 ] ) ] for @xmath90 and @xmath91 .
we find that in these regimes , coefficients of @xmath92 and @xmath93 are always positive , and the constant term changes sign as in fig .
the shaded ( white ) region indicates the positive ( negative ) values of the constant term .
the constant term is also positive for @xmath94 and @xmath89 thus , by the rh criteria eq .
( [ e23 ] ) has at least one root with positive real part when the constant term of the quartic is negative . the system of eqs .
( [ eq15])-([e18 ] ) is then said to be linearly unstable .
as an illustration , the eigenvalues for @xmath95 , @xmath96 are a pair of complex conjugates , one real positive @xmath97 and one negative real roots . and @xmath32 , showing its positive ( shaded colored region ) and negative ( white region ) values .
the coefficient is also positive ( not shown ) for @xmath94 and @xmath89,width=288,height=240 ]
from eq . ( [ e7 ] ) , the mi sets in only if @xmath98 and @xmath99 defines the curve along which pitchfork bifurcation takes place @xcite . also
, the dynamics is really subsonic in @xmath100 where the mi growth rate is small @xcite . as
, @xmath32 decreases from @xmath101 many more unstable modes ( since @xmath39)$ ] with higher harmonics will be excited . in this case , even though the dynamics is not subsonic , a description with three modes should be relatively accurate for @xmath102 our basic assumption at this point is that the langmuir modes for @xmath103 are stable , and the ion - acoustic modes with @xmath104 remain as they are , already excited by the presence of @xmath105 langmuir wave envelopes .
this results a system of odes [ eqs .
( [ eq15])-([e18 ] ) ] describing the temporal dynamics of the field variables .
notice , however , that although the fourier modes are truncated , no assumption on the slowness of the dynamics is made .
we would like to see how the temporal system behaves as @xmath32 approaches @xmath106 from @xmath101 and the condition for the subsonic region is progressively relaxed by reducing @xmath32 from @xmath107 to some extent@xmath16 here @xmath32 is not so small that the three - wave model becomes inappropriate , because smaller the @xmath32 , the larger is the number of unstable modes . in the latter case , full simulation of eqs .
( [ e5 ] ) and ( [ eq6 ] ) will be necessary to give more appropriate information about the interaction .
thus , for the temporal evolution , we numerically integrate the system of eqs .
( [ eq15])-([e18 ] ) by the runge - kutta scheme with step size @xmath108 and initial values @xmath109 by suitably varying the system parameters @xmath32 and @xmath44 together with the initial choices , we have obtained different sets of parameters which can exhibit periodic as well as chaotic attractors .
the latter are established by the lyapunov analysis with at least one positive exponent indicating chaos . on the other hand ,
the regular or limit cycles may be described on the basis of their structures , amplitude oscillations with time or lyapunov exponents with negative or zero values . for @xmath110 , @xmath111 and @xmath112 in @xmath113 fig .
3 shows a periodic limit cycle .
this suggests that within the plane wave region , strong chaotic activity is completely absent , i.e. , most of the kolmogorov - arnold - moser ( kam ) surfaces , especially the central fixed point remains unaffected by the nonlinear interaction .
in fact , the absence of chaos is not only for these parameter values , but also for some other values smaller or larger .
it also explains that within this plane wave region no massive redistribution of energy takes place
. this information well agrees with the full simulation of the space - time problem @xcite . as @xmath32 is lowered from @xmath107 , e.g. , @xmath114 , fig .
4 * * shows that the periodicity or integrability of the temporal system tends to become poorer exhibiting chaotic orbits .
this can be verified by the set of corresponding lyapunov exponents as shown in fig . 5 , where one positive exponent indicates chaos .
we also find that as @xmath32 becomes smaller , the largest lyapunov tends to increase from a positive small value to a large one @xmath97 .
so , it may be possible that as the basic wave number @xmath32 is decreased further , strong chaotic features develop into the system . and
@xmath111 , showing that the three - wave dynamics is stable.,width=384,height=240 ] and @xmath111 , showing that the three - wave dynamics is unstable.,width=384,height=240 ] corresponding to the same parameter values as in fig .
one of the exponents is clearly positive , indicating that the three - wave dynamics is chaotic.,width=384,height=240 ] and @xmath111 , showing the bifurcation of an attracting periodic orbit.,width=384,height=240 ] thus , when @xmath115 the system s periodicity breaks down .
this is expected , since for smaller @xmath32 , the solitons get more affected by the ion - acoustic radiation until they are completely destroyed . in the strong chaotic process
one can expect that solitons will deliver their energy to the remaining modes so as to gradually disappear after some time .
however , one can no longer rely on the low - dimensional model for some smaller values of @xmath32 , as it gives rise an excitation of large number of unstable harmonic modes .
this is the situation where full simulation of eqs .
( [ e5 ] ) and ( [ eq6 ] ) will give better understanding for the interaction .
in addition to the above results , one interesting can also be to study the existence of hopf - bifurcation in the eqs .
( [ eq15])-([e18 ] ) by choosing @xmath32 as the bifurcation parameter .
generally , hopf - bifurcation is defined as the change in qualitative behavior of the system when a pair of complex conjugate eigenvalues passes through the imaginary axis .
a pair of such roots can be obtained by numerically examining the eigenvalue equation ( [ e23 ] ) .
thus , one can perform a series of analysis with different values of @xmath32 .
figure 6 shows the supercritical diagram corresponding to @xmath116 and @xmath117 in which an attracting periodic orbit bifurcates .
the detailed analysis of such bifurcation is , however , beyond the scope of the present work .
in the previous section we have observed that as the value of @xmath32 is lowered from @xmath101 the system exhibits strong chaotic features .
so , in order to investigate the global behaviors of the solitons , namely the soliton interaction with strong iaw emission , formation of incoherent patterns due to collision and fusion , it is reasonable to consider the system of eqs .
( [ e5 ] ) and ( [ eq6 ] ) without the mixed derivatives ( valid for long - wavelengths ) . in our numerical scheme ,
we assume the spatial periodicity with the simulation box length @xmath118 such that @xmath119 the resonant wave length@xmath16 the integer @xmath39,$ ] representing the number of modes , is considered as a number of grid points , e.g. , @xmath120 and @xmath121 as the case may be .
we choose the initial conditions as @xcite@xmath122 , \nonumber\\ \text { } n(x,0 ) & = -2be_{0}\cos(kx)/l , \label{e27}%\end{aligned}\ ] ] where @xmath123 is the amplitude of the pump langmuir wave , @xmath124 is a suitable constant , and @xmath125 is of the order of @xmath126 to emphasize that the perturbation is relatively small .
the spatiotemporal system was advanced in time using the runge - kutta scheme with a time step @xmath127 .
we consider the grid size @xmath128 for lower values of @xmath32 , and @xmath129 for relatively large @xmath130 so that @xmath18 corresponds to the grid position @xmath131 and @xmath132 respectively . in the simulation , we consider @xmath133 @xmath134and @xmath135 , unless otherwise mentioned .
figures 7 ( in presence of iwn ) and 8 ( in absence of iwn ) show the profiles of the electric filed ( upper panel ) and density fluctuation ( lower panel ) associated with the iaws at the end of simulation for @xmath136 i.e. , for the of @xmath137=4 $ ] harmonic modes . at the central part
, we observe an excited electric field , @xmath138 highly correlated with the density depletion , @xmath139 in presence of iwn . whereas in absence of iwn , these are @xmath140 and @xmath141 clearly , the effect of iwn is to enhance the magnitude of the wave electric field or to decrease that of the density fluctuation . from fig
. 9 ( contour plot corresponding to fig .
7 ) , we observe that the solitary master patterns are first formed by the master mode , and then begin to appear from the unstable harmonic modes . during the pattern formation , ions are usually driven out ( by the resultant of ponderomotive force and the iwn ) from the regions of the master mode as well as from some harmonic patterns , and then they form density distributions with @xmath142 ( see the lower panels of figs . 7 and 8) along the spatial axis .
these ion density humps move stochastically on either side of the patterns , and do not arrest the langmuir wave fields .
we observe from figs . 9 and 10 * * ( contour plot corresponding to fig .
8) that the pattern selection leads to three @xmath143 harmonic modes initially peaked at @xmath144 and @xmath145 these three modes are basically generated by the master mode and the unstable harmonic modes .
the system is then in the coexistence of temporal chaos ( tc ) and spatio - partial coherence ( spc ) state similar to the case of plane wave region @xmath146 explained in the previous section .
however , as the iwn is relaxed , only two solitary patterns initially peaked at @xmath147 and @xmath148 are seen to collide with the master mode at @xmath149 and @xmath150 respectively ( see fig .
10)@xmath16 also , the collision is so weak that it does not influence the soliton dynamics .
the reason is that in absence of iwn and for relatively small @xmath123 , the ponderomotive force was the only force to cause the weak ion - acoustic wave emission . as a result , though the collisions take place but are not sufficiently strong to fuse into a new one . on the other hand ,
when @xmath123 is relatively large , e.g. , @xmath151 i.e. , when the amplitude of both the initial profiles increase , one may observe ( see fig .
11 ) more collision and fusion due to strong iaw emission by the iwn . in this case , @xmath152 increases , but @xmath153 decreases as compared to the case of no iwn ( fig . 12 ) .
figure 11 also shows that two pairs of solitary patterns initially peaked at the either side of @xmath154 collide and fuse into two new patterns , one of which again collides with the other two master patterns peaked at @xmath155 and @xmath156 at @xmath157 and @xmath158 after these collisions , there remain only three distorted patterns .
the system is in tc , but still in the coexistence of tc and stc , since a few collisions will not suffice to cause stc @xcite .
lower panel ) with respect to the space @xmath159 after time @xmath160 in the numerical simulation of eqs .
( 5 ) and ( 6 ) for @xmath161 and @xmath111 with iwn.,width=336,height=336 ] const . with respect to the space @xmath162 and time @xmath163 corresponding to the evolution as in fig .
this shows that the pattern selection leads to three harmonic patterns .
the system is in the coexistence of tc and spc state.,width=240,height=240 ] const . with respect to the space @xmath162 and time @xmath163 corresponding to the evolution as in fig .
this shows that the pattern selection leads to three harmonic patterns .
two collisions occur at @xmath149 and @xmath164 the system is still in the coexistence of tc and spc state.,width=240,height=240 ] now , if @xmath32 is further lowered , e.g. , @xmath165 many solitary patterns are formed from the master mode and unstable harmonic modes ( see fig . 13 with iwn and fig . 14 without iwn ) by means of pattern selection .
we observe that when iwn influences the dynamics , many unstable harmonic modes are excited compared to the other case . as a result ,
not only two solitary patterns collide ( as mostly in the case of no iwn ) to fuse into a new pattern , but more than two also collide repeatedly to form new ones which again collide with other new pattern(s ) fused by the other collisions .
notice from figure 13 that the first collision occurs at @xmath166 between two solitary patterns .
these two patterns then fuse to form a new one , which again collides with other three master patterns initially peaked at @xmath167 @xmath168 and @xmath169 at @xmath170 and @xmath171 respectively .
there are also other collision and fusion that can be explained similarly . in general ,
as time progresses , the patterns seem to be much distorted in losing their strengths .
the original @xmath172 ( fig .
13 ) or @xmath173 ( fig .
14 ) solitary patterns are then finally fused into a few incoherent ones . as solitons gradually vanish , irregular radiation increases due to stochastic motion .
that is , as soon as a chaotic subset with less degrees of freedom is formed , the subset will then act as a pump delivering a net amount of energy to its neighboring modes .
this energy transfer takes place because of the very random nature of the couplings .
the system is then said to be in the stc state . in this case , for @xmath174 the spatial correlation function will approach to a zero value @xcite . a certain amount of energy , which was initially distributed among the @xmath173 or @xmath172 solitary waves ,
will now be transferred to a few incoherent patterns as well as to some stable higher harmonic modes with short - wavelengths .
so , if initially there exist many modulational unstable modes with different modulation lengths to form solitary envelopes , collision and fusion among most of them can lead to the state of stc .
there must exist a critical value of this wavelength or wave number @xmath32 at which the transition from tc to stc takes place .
this value of @xmath32 is roughly in @xmath175 comparing the pattern formation , collision and fusion in figs .
11 , 12 and 13 , 14 one finds that iaw emission is more stronger in presence of iwn .
thus , in order to observe the interactions between lws and iaws at longer time scale and for long - wavelength limits , the effect of iwn is more important , which can no longer be neglected especially for relatively large amplitude waves . in this way , the term containing the iwn in eq .
( [ e2 ] ) would act as a correction to the dynamics of arbitrary amplitude coupled langmuir and iaws . , but for relatively large @xmath123 , i.e. , @xmath176 , indicating that more collision and fusion take place due to iwn.,width=240,height=240 ] const .
with respect to the space @xmath162 and time @xmath163 for @xmath177 and @xmath111 in presence of iwn .
the pattern selection shows that seventeen solitary patterns , which were formed initially , collide and get fused into other incoherent patterns after some time .
the collision is random and not confined between two patterns .
the iaw emission also occurs , and the stc state emerges.,width=240,height=240 ]
from the analysis of the previous sections , we find that the three - wave model can be a good approximation @xcite for the interaction of coupled lws and iaws in the plane wave region @xmath113 where the system exhibits stable oscillations .
it can relatively be accurate in the region @xmath178 where the system exhibits temporal chaos for a given value of the pump electric field @xmath123 . for values of @xmath179 which indicates the excitation of many ( @xmath180 ) unstable harmonic modes
, the low - dimensional model fails to describe the dynamics of the coupled waves .
we have found that the three - wave model exhibits periodic orbits in @xmath181and chaos in @xmath182 for a given initial pump @xmath111 and a given constant @xmath183 the space - time evolution , on the other hand , reveals that the system is in the state of tc for relatively a lower value of @xmath130 i.e. , in the regime @xmath184 and the stc state emerges for @xmath175 to conclude , we have investigated the nonlinear interaction of coupled lws and iaws by considering the effects of dispersion due to charge separation , and the nonlinearity associated with the ion density fluctuations .
the latter can cause the excitation of more unstable harmonic modes , and can lead to strong iaw emission in the space - time evolution .
the low - dimensional model shows that a transition from periodicity to chaos is possible for a suitable choice of the wave number of modulation .
it basically predicts some basic features of the system as well as suggests an approximate region of @xmath32 for the existence of regular and chaotic structures .
the space - time evolution indicates that solitary patterns are first formed by the unstable harmonic modes .
these modes are then excited by a master mode due to initial mi .
as solitons experience oscillatory motions , they emit radiation due to irregular interaction with the ion - acoustic fields .
such radiations are broad - band in nature , causing a growing number of modes to be involved in the stochastic dynamics . depending on the modulation length scale , the system may either be in the tc and spc states or in tc and stc states . if few harmonic patterns coexist with the master pattern , and iaw emission remains weak , the system is still in the state of tc and spc . however , if many solitary patterns are formed at the early stage due to mi , the system may experience the state of spc . over longer time scales , when solitons are affected by the free ion - acoustic radiation , collision and fusion among the patterns occur frequently as well as repeatedly to form several new incoherent patterns . the stc state is then said to emerge in which energy can flow faster the smaller is the wave number of modulation @xmath32 . since our model is in one spatial dimension , energy transfer is not a result of soliton collapse , rather it is purely generated by the nonintegrable nature as well as chaotic aspects of the system @xcite .
the results may have physical significance to related problems , such as turbulence in pulsar radiation where strong magnetic field plays an important role @xcite .
s. b. is thankful to m.pizzi of micro and nanotechnology division , techfab s.r.l .
, chivasso , italy , for some useful discussion .
m. gratefully acknowledges support from the kempe foundations , sweden . | a simulation study is performed to investigate the dynamics of coupled langmuir waves ( lws ) and ion - acoustic waves ( iaws ) in an unmagnetized plasma .
the effects of dispersion due to charge separation and the density nonlinearity associated with the iaws , are considered to modify the properties of langmuir solitons , as well as to model the dynamics of relatively large amplitude wave envelopes .
it is found that the langmuir wave electric field , indeed , increases by the effect of ion - wave nonlinearity ( iwn ) .
use of a low - dimensional model , based on three fourier modes shows that a transition to temporal chaos is possible , when the length scale of the linearly excited modes is larger than that of the most unstable ones .
the chaotic behaviors of the unstable modes are identified by the analysis of lyapunov exponent spectra .
the space - time evolution of the coupled lws and iaws shows that the iwn can cause the excitation of many unstable harmonic modes , and can lead to strong iaw emission .
this occurs when the initial wave field is relatively large or the length scale of iaws is larger than the soliton characteristic size .
numerical simulation also reveals that many solitary patterns can be excited and generated through the modulational instability ( mi ) of unstable harmonic modes .
as time goes on , these solitons are seen to appear in the spatially partial coherence ( spc ) state due to the free ion - acoustic radiation as well as in the state of spatiotemporal chaos ( stc ) due to collision and fusion in the stochastic motion .
the latter results the redistribution of initial wave energy into a few modes with small length scales , which may lead to the onset of langmuir turbulence in laboratory as well as space plasmas . |
finite unified theories are @xmath4 supersymmetric grand unified theories ( guts ) which can be made finite even to all - loop orders , including the soft supersymmetry breaking sector .
the method to construct guts with reduced independent parameters @xcite consists of searching for renormalization group invariant ( rgi ) relations holding below the planck scale , which in turn are preserved down to the gut scale . of particular interest is the possibility to find rgi relations among couplings that guarantee finitenes to all - orders in perturbation theory @xcite . in order to achieve the latter
it is enough to study the uniqueness of the solutions to the one - loop finiteness conditions @xcite .
the constructed _ finite unified _ @xmath4 supersymmetric su(5 ) guts , using the above tools , predicted correctly from the dimensionless sector ( gauge - yukawa unification ) , among others , the top quark mass @xcite .
the search for rgi relations and finiteness has been extended to the soft supersymmetry breaking sector ( ssb ) of these theories @xcite , which involves parameters of dimension one and two .
eventually , the full theories can be made all - loop finite and their predictive power is extended to the higgs sector and the supersymmetric spectrum ( s - spectrum ) .
the purpose of the present article is to start an exhaustive search of these latter predictions , as well as to provide a rather dense review of the subject .
here let us review the main points and ideas concerning the _ reduction of couplings _ and _ finiteness _ in @xmath4 supersymmetric theories .
a rgi relation among couplings @xmath5 , @xmath6 has to satisfy the partial differential equation @xmath7 where @xmath8 is the @xmath9-function of @xmath5 . there
exist ( @xmath10 ) independent @xmath11 s , and finding the complete set of these solutions is equivalent to solve the so - called reduction equations ( res ) @xcite , @xmath12 where @xmath13 and @xmath14 are the primary coupling and its @xmath9-function . using all the @xmath15 s to impose rgi relations
, one can in principle express all the couplings in terms of a single coupling @xmath13 .
the complete reduction , which formally preserves perturbative renormalizability , can be achieved by demanding a power series solution , whose uniqueness can be investigated at the one - loop level .
finiteness can be understood by considering a chiral , anomaly free , @xmath4 globally supersymmetric gauge theory based on a group g with gauge coupling constant @xmath13 .
the superpotential of the theory is given by @xmath16 where @xmath17 ( the mass terms ) and @xmath18 ( the yukawa couplings ) are gauge invariant tensors and the matter field @xmath19 transforms according to the irreducible representation @xmath20 of the gauge group @xmath21 .
the one - loop @xmath9-function of the gauge coupling @xmath13 is given by ^(1)_g&= & = [ _ il(r_i)-3c_2(g ) ] , [ betag ] where @xmath22 is the dynkin index of @xmath20 and @xmath23 is the quadratic casimir of the adjoint representation of the gauge group @xmath21 .
the @xmath9-functions of @xmath18 , by virtue of the non - renormalization theorem , are related to the anomalous dimension matrix @xmath24 of the matter fields @xmath19 as : _
c^ijk = c^ijk = c^ijp _ n=1_p^k(n ) + ( k i ) + ( kj ) .
[ betay ] at one - loop level @xmath24 is given by _
i^j(1)=c_ipqc^jpq-2g^2c_2(r_i)_i^j , [ gamay ] where @xmath25 is the quadratic casimir of the representation @xmath20 , and @xmath26 .
all the one - loop @xmath9-functions of the theory vanish if the @xmath9-function of the gauge coupling @xmath27 , and the anomalous dimensions @xmath28 , vanish , i.e. @xmath29 where @xmath30 is the dynkin index of @xmath31 , and @xmath32 is the quadratic casimir invariant of the adjoint representation of @xmath21 .
a very interesting result is that the conditions ( [ zoup - fini ] ) are necessary and sufficient for finiteness at the two - loop level @xcite .
the one- and two - loop finiteness conditions ( [ zoup - fini ] ) restrict considerably the possible choices of the irreducible representations @xmath31 for a given group @xmath21 as well as the yukawa couplings in the superpotential ( [ supot ] ) . note in particular that the finiteness conditions can not be applied to the supersymmetric standard model ( ssm ) , since the presence of a @xmath33 gauge group is incompatible with the condition ( [ zoup - fini ] ) , due to @xmath34=0 $ ] .
this leads to the expectation that finiteness should be attained at the grand unified level only , the ssm being just the corresponding low - energy , effective theory .
the finiteness conditions impose relations between gauge and yukawa couplings .
therefore , we have to guarantee that such relations leading to a reduction of the couplings hold at any renormalization point .
the necessary , but also sufficient , condition for this to happen is to require that such relations are solutions to the reduction equations ( res ) to all orders .
the all - loop order finiteness theorem of ref.@xcite is based on : ( a ) the structure of the supercurrent in @xmath4 sym and on ( b ) the non - renormalization properties of @xmath4 chiral anomalies @xcite .
alternatively , similar results can be obtained @xcite using an analysis of the all - loop nsvz gauge beta - function @xcite .
the above described method of reducing the dimensionless couplings has been extended @xcite to the soft supersymmetry breaking ( ssb ) dimensionful parameters of @xmath4 supersymmetric theories . recently very interesting
progress has been made @xcite-@xcite concerning the renormalization properties of the ssb parameters , based conceptually and technically on the work of ref .
@xcite . in this work
the powerful supergraph method @xcite for studying supersymmetric theories has been applied to the softly broken ones by using the `` spurion '' external space - time independent superfields @xcite . in the latter method
a softly broken supersymmetric gauge theory is considered as a supersymmetric one in which the various parameters such as couplings and masses have been promoted to external superfields that acquire `` vacuum expectation values '' . based on this method the relations among the soft term renormalization and
that of an unbroken supersymmetric theory have been derived .
in particular the @xmath9-functions of the parameters of the softly broken theory are expressed in terms of partial differential operators involving the dimensionless parameters of the unbroken theory . the key point in the strategy of refs .
@xcite-@xcite in solving the set of coupled differential equations so as to be able to express all parameters in a rgi way , was to transform the partial differential operators involved to total derivative operators @xcite .
it is indeed possible to do this on the rgi surface which is defined by the solution of the reduction equations .
in addition it was found that rgi ssb scalar masses in gauge - yukawa unified models satisfy a universal sum rule at one - loop @xcite .
this result was generalized to two - loops for finite theories @xcite , and then to all - loops for general gauge - yukawa and finite unified theories @xcite . in order to obtain a feeling of some of the above results , consider the superpotential given by ( 1 ) along with the lagrangian for ssb terms @xmath35 where the @xmath36 are the scalar parts of the chiral superfields @xmath37 , @xmath38 are the gauginos and @xmath39 their unified mass . since only finite theories are considered here , it is assumed that the gauge group is a simple group and the one - loop @xmath9-function of the gauge coupling @xmath13 vanishes .
it is also assumed that the reduction equations admit power series solutions of the form c^ijk = g_n=0^ijk_(n ) g^2n . according to the finiteness theorem @xcite , the theory is then finite to all - orders in perturbation theory , if , among others , the one - loop anomalous dimensions @xmath40 vanish
. the one- and two - loop finiteness for @xmath41 can be achieved by @xcite h^ijk = -m c^ijk+ =-m
^ijk_(0)g+o(g^5 ) .
an additional constraint in the ssb sector up to two - loops @xcite , concerns the soft scalar masses as follows @xmath42 for i , j , k with @xmath43 , where @xmath44 is the two - loop correction @xmath45~t(r_l),\ ] ] which vanishes for the universal choice @xcite , i.e. when all the soft scalar masses are the same at the unification point .
if we know higher - loop @xmath9-functions explicitly , we can follow the same procedure and find higher - loop rgi relations among ssb terms .
however , the @xmath9-functions of the soft scalar masses are explicitly known only up to two loops . in order to obtain higher - loop results ,
we need something else instead of knowledge of explicit @xmath9-functions , e.g. some relations among @xmath9-functions .
the recent progress made using the spurion technique @xcite leads to the following all - loop relations among ssb @xmath9-functions , @xcite-@xcite _ m & = & 2o(_gg ) , [ betam ] + _
h^ijk&=&^i_lh^ljk+^j_lh^ilk + ^k_lh^ijl + & & -2_1^i_lc^ljk -2_1^j_lc^ilk-2_1^k_lc^ijl , + ( _ m^2)^i_j & = & ^i_j , [ betam2 ] + o & = & ( mg^2 -h^lmn ) , [ diffo ] + & = & 2oo^ * + 2|m|^2 g^2 + _ lmn + ^lmn , where @xmath46 , @xmath47 , and ^ijk&= & ( m^2)^i_lc^ljk+(m^2)^j_lc^ilk+(m^2)^k_lc^ijl . [ tildec ]
it was also found @xcite that the relation h^ijk & = & -m ( c^ijk) -m , [ h2 ] among couplings is all - loop rgi .
furthermore , using the all - loop gauge @xmath9-function of novikov _ et al . _ @xcite given by _ g^nsvz & = & , [ bnsvz ]
it was found the all - loop rgi sum rule @xcite , m^2_i+m^2_j+m^2_k & = & |m|^2 \ { + } + & & + _
[ sum2 ] in addition the exact-@xmath9-function for @xmath48 in the nsvz scheme has been obtained @xcite for the first time and is given by _
m^2_i^nsvz & = & _
in this section we examine two concrete @xmath49 finite models , where the reduction of couplings in the dimensionless and dimensionful sector has been achieved . a predictive gauge - yukawa unified @xmath49 model which is finite to all orders , in addition to the requirements mentioned already , should also have the following properties : 1 .
one - loop anomalous dimensions are diagonal , i.e. , @xmath50 .
2 . three fermion generations , in the irreducible representations @xmath51 , which obviously should not couple to the adjoint @xmath52 .
3 . the two higgs doublets of the mssm should mostly be made out of a pair of higgs quintet and anti - quintet , which couple to the third generation . in the following we discuss two versions of the all - order finite model .
the model of ref .
@xcite , which will be labeled @xmath53 , and a slight variation of this model ( labeled @xmath54 ) , which can also be obtained from the class of the models suggested by kazakov _
@xcite with a modification to suppress non - diagonal anomalous dimensions .
the superpotential which describes the two models takes the form @xcite w & = & _ i=1 ^ 3 [ g_i^u * 10*_i*10*_i h_i+ g_i^d*10*_i _ i _ i ] + g_23^u*10*_2 * 10*_3 h_4 + & & + g_23^d*10*_2 _ 3 _ 4 + g_32^d*10*_3 _ 2 _ 4 + _
a=1 ^ 4g_a^fh_a * 24*_a+ ( * 24*)^3 , [ zoup - super ] where @xmath55 and @xmath56 stand for the higgs quintets and anti - quintets .
the non - degenerate and isolated solutions to @xmath57 for the models @xmath58 are : ( g_1^u)^2 & = & \ { , } g^2 , ( g_1^d)^2 = \{,}g^2 , ( g_2^u)^2=(g_3^u)^2=\{,}g^2 , [ zoup - sol5 ] + ( g_2^d)^2 & = & ( g_3^d)^2=\{,}g^2 , ( g_23^u)^2 = \{0,}g^2 , ( g_23^d)^2=(g_32^d)^2=\{0,}g^2 , + ( g^)^2 & = & g^2 , ( g_2^f)^2 = ( g_3^f)^2=\{0,}g^2 , ( g_1^f)^2=0 , ( g_4^f)^2=\{1,0}g^2 . according to the theorem of ref . @xcite these models are finite to all orders . after the reduction of couplings the symmetry of @xmath59 is enhanced @xcite . the main difference of the models @xmath53 and @xmath54 is that three pairs of higgs quintets and anti - quintets couple to the @xmath60 for @xmath54 so that it is not necessary to mix them with @xmath61 and @xmath62 in order to achieve the triplet - doublet splitting after the symmetry breaking of @xmath49 . in the dimensionful sector
, the sum rule gives us the following boundary conditions at the gut scale @xcite : m^2_h_u+ 2 m^2_*10 * & = & m^2_h_d+ m^2_+ m^2_*10*=m^2 * a * ; + m^2_h_u+ 2 m^2_*10 * & = & m^2 , m^2_h_d-2m^2_*10*=- , + m^2_+ 3m^2_*10*&= & * b * , where we use as free parameters @xmath63 and @xmath64 for the model @xmath53 , and @xmath64 for @xmath54 , in addition to @xmath39 .
since the gauge symmetry is spontaneously broken below @xmath65 , the finiteness conditions do not restrict the renormalization properties at low energies , and all it remains are boundary conditions on the gauge and yukawa couplings ( [ zoup - sol5 ] ) , the @xmath66 relation , and the soft scalar - mass sum rule ( [ zoup - sumr ] ) at @xmath65 , as applied in the two models .
thus we examine the evolution of these parameters according to their rges up to two - loops for dimensionless parameters and at one - loop for dimensionful ones with the relevant boundary conditions . below @xmath65 their evolution
is assumed to be governed by the mssm .
we further assume a unique supersymmetry breaking scale @xmath67 ( which we define as the average of the stop masses ) and therefore below that scale the effective theory is just the sm .
the predictions for the top quark mass @xmath68 are @xmath69 and @xmath70 gev in models @xmath53 and @xmath54 respectively . comparing these predictions with the most recent experimental value @xmath71 gev @xcite , and recalling that the theoretical values for @xmath68 may suffer from a correction of @xmath72 @xcite
, we see that they are consistent with the experimental data .
in addition the value of @xmath73 is found to be @xmath74 and @xmath75 for models @xmath53 and @xmath54 respectively . in the ssb sector , besides the constraints imposed by finiteness there are further restrictions imposed by phenomenology . in the case where all the soft scalar masses are universal at the unfication scale ,
there is no region of @xmath39 below @xmath76 in which @xmath77 is satisfied ( where @xmath78 is the lightest @xmath79 mass , and @xmath80 the lightest neutralino mass , which is the lightest supersymmetric particle ) . but once the universality condition is relaxed this problem can be solved naturally ( thanks to the sum rule ) . more specifically , using the sum rule ( [ zoup - sumr ] ) and imposing the conditions a ) successful radiative electroweak symmetry breaking , b ) @xmath81 and c ) @xmath82 , a comfortable parameter space for both models ( although model @xmath54 requires large @xmath83 tev ) is found .
as function of @xmath84 for different values of @xmath39 for model @xmath85 , for @xmath86.,width=302 ] as an additional constraint , we take into account the @xmath87 @xcite .
we do not take into account , though , constraints coming from the muon anomalous magnetic moment ( g-2 ) in this work , which excludes a small region of the parameter space . in the graphs we show the * futa * results concerning @xmath88 , @xmath80 , and @xmath89 , for different values of @xmath39 , for the case where @xmath90 and the lsp is a neutralino @xmath91 .
the results for @xmath92 are slightly different : the spectrum starts around 500 gev .
the main difference , though , is in the value of the running bottom mass @xmath93 , where we have included the corrections coming from bottom squark - gluino loops and top squark - chargino loops @xcite . in the @xmath90 case ,
@xmath94 gev is just below the experimental value @xmath95 gev @xcite , whereas in the @xmath96 case , @xmath97 gev , i. e. above the experimental value .
as function of @xmath84 for different values of @xmath39 for model @xmath85 , for @xmath86.,width=302 ] .a representative example of the bottom ( running ) and top ( pole ) masses , plus the supersymmetric spectrum for model futa , with @xmath98 697 gev , @xmath99 806 gev , @xmath100 1681 gev , @xmath101 .
all masses in the table are in gev . [ cols="^,^,^,^",options="header " , ] the higgs mass prediction of the two models is , although the details of each of the models differ , in the following range m_h = ~112 - 132 gev , where the uncertainty comes from variations of the gaugino mass @xmath39 and the soft scalar masses , and from finite ( i.e. not logarithmically divergent ) corrections in changing renormalization scheme .
the one - loop radiative corrections have been included @xcite for @xmath88 , but not for the rest of the spectrum . in making the analysis ,
the value of @xmath39 was varied from @xmath102 gev for * futa*. we have also included a small variation , due to threshold corrections at the gut scale , of @xmath103 of the fut boundary conditions . this small variation does not give a noticeable effect in the results at low energies . from fig .
1 we already see that the requirement @xmath104 gev @xcite ( neglecting the theoretical uncertainties ) excludes the possibility of @xmath105 gev for futa , which is taken into account in the presentation of the next graphs . in tables 1 and 2
we present representative examples of the values obtained for the s - spectra for each of the models .
a more detailed numerical analysis , where the results of our program and of the known programs feynhiggs @xcite and suspect @xcite are combined , is currently in progress @xcite .
we acknowledge very useful discussions with n.benekos , g. daskalakis , r.kinnunen and e. richter - was , as well as with j. erler .
supported by the projects papiit - in116206 and conacyt 42026-f , and partially by the rtn contract hprn - ct-2000 - 00148 , the greek - german bilateral programme ikyda-2001 and by the ntua programme for fundamental research `` thales '' .
natexlab#1#1 [ 1]``#1 '' url#1`#1 ` urlprefix kubo , j. , mondragn , m. and zoupanos , g. ( 1994 ) _ nucl .
_ * b424 * 291 . zimmermann , w. , ( 1985 ) _ com . math . phys . _ * 97 * 211 ; oehme , r. and zimmermann , w. ( 1985 ) _ com .
phys . _ * 97 * 569 ; ma , e .
phys.rev _ * d 17 * 623;ibid ( 1985 ) * d31 * 1143 .
lucchesi , c. , piguet , o. and sibold , k. ( 1988 ) _ helv .
acta _ * 61 * 321 ; piguet , o. and sibold , k. ( 1986 ) _ intr .
phys . _ * a1 * 913 ; ( 1986 ) _ phys
. lett . _ * b177 * 373 ; see also lucchesi , c. and zoupanos , g. ( 1997 ) _ fortsch . phys .
_ * 45 * 129 .
ermushev , a.z . , kazakov , d.i . and tarasov , o.v .
( 1987 ) _ nucl .
phys . _ * 281 * 72 ; kazakov , d.i . ( 1987 ) _ mod .
. lett . _ * a9 * 663 .
kapetanakis d. , mondragn , m. and zoupanos , g. , ( 1993 ) _ zeit .
f. phys . _
* c60 * 181 ; mondragn , m. and zoupanos , g. ( 1995 ) _ nucl . phys .
suppl . ) * 37c * ) 98 .
jack , i. and jones , d.r.t .
( 1995 ) _ phys .
lett . _ * b349 * 294 .
kubo , j. , mondragn , m. and zoupanos , g. ( 1996 ) _ phys .
lett . _ * b389 * 523 .
jones , l. mezincescu and y .- p .
yao , _ phys .
lett . _ * b148 * 317 ( 1984 ) .
jack , i. and jones , d.r.t .
( 1994 ) _ phys .
_ * b333 * 372 .
leigh , r.g . , and strassler , m.j .
( 1995 ) _ nucl .
phys . _ * b447 * 95 .
novikov , v. , shifman , m. , vainstein , a. , and zakharov , v. ( 1983 ) _ nucl .
phys._*b229 * 381 ; ( 1986 ) _ phys .
lett . _ * b166 * 329 ; shifman , m. ( 1996 ) _
int.j . mod . phys._*a11 * 5761 and references therein .
avdeev , l.v . ,
kazakov , d.i . and kondrashuk , i.n .
( 1998 ) _ nucl .
_ * b510 * 289 ; kazakov , d.i .
_ phys.lett . _ * b449 * 201 .
t. kobayashi , j. kubo and g. zoupanos , ( 1998 ) phys .
* b427 * 291 ; t. kobayashi et al , in proc . of `` supersymmetry , supergravity and superstrings '' , pp .
242 - 268 , seoul , 1999 ; t.kobayashi et .
al ( 2001 ) _ surveys high energy phys . _ * 16 * 87 - 129 .
yamada , y. ( 1994 ) _
phys.rev . _ * d50 * 3537 .
j. hisano and m. shifman , _ phys . rev . _
* d56 * ( 1997 ) 5475 .
kawamura , t. , kobayashi , t. and kubo , j. ( 1997 ) _ phys .
lett . _ * b405 * 64 .
delbourgo , r. ( 1975 ) _
nuovo cim _ * 25a * 646 ; salam , a. and strathdee , j. ( 1975 ) _ nucl .
phys . _ * b86 * 142 ; fujikawa , k. and lang , w. ( 1975 ) _ nucl . phys . _ * b88 * 61 ; grisaru , m.t . ,
rocek , m. and siegel , w. ( 1979 ) _ nucl .
phys . _ * b59 * 429 .
girardello , l. and grisaru , m.t .
( 1982 ) _ nucl .
phys . _ * b194 * 65 ; helayel - neto , j.a .
( 1984 ) _ phys .
_ * b135 * 78 ; feruglio , f. , helayel - neto , j.a . and legovini , f. ( 1985 ) _ nucl . phys . _
* b249 * 533 ; scholl , m. ( 1985 ) _ zeit . f. phys . _ *
c28 * 545 . i. jack , d.r.t .
jones and a. pickering , _ phys .
_ * b426 * ( 1998 ) 73 .
kobayashi , t. , kubo , j. , mondragn , m. and zoupanos , g. ( 1998 ) _ nucl .
phys . _ * b511 * 45 ; in proc . of ichep 1998 ,
2 , p. 1597
( vancouver 1998 ) ; acta phys . polon . *
b30 * ( 1999 ) 2013 ; in proc . of hep99 ,
( tampere 1999 ) ; m. mondragon and g. zoupanos , acta phys
. polon .
b * 34 * ( 2003 ) 5459 .
k. s. babu , t. enkhbat and i. gogoladze ( 2003 ) phys .
b * 555 * 238 [ arxiv : hep - ph/0204246 ] ; j. erler , to appear in the next edition of the pdg , private communication .
for an extended discussion and a complete list of references see : kubo , j. , mondragn , m. and zoupanos , g. ( 1997 ) _ acta phys .
polon . _
* b27 * 3911 .
p. cho , m. misiak and d. wyler , _ phys .
* d 54 * , 3329 ( 1996 ) ; a. kagan and m. neubert , _ eur .
j. _ * c 7 * ( 1999 ) 5 , k. chetyrkin , m. misiak and m. munz , _ phys .
* b 400 * , ( 1997 ) 206 , [ erratum - ibid .
* b 425 * ( 1998 ) 414 ] p.
gambino and m. misiak , _ nucl .
phys . _ * b 611 * ( 2001 ) 338 , r. barate et al .
[ aleph collaboration ] , _ phys .
* b 429 * ( 1998 ) 169 ; s. chen et al .
[ cleo collaboration ] , _ phys .
* 87 * ( 2001 ) 251807 , k. abe et al .
[ belle collaboration ] , _ phys .
_ * b 511 * ( 2001 ) 151 , b. aubert et al .
[ babar collaboration ] ; l. j. hall , r. rattazzi and u. sarid , phys .
d * 50 * , ( 1994 ) 7048 [ arxiv : hep - ph/9306309 ] ; r. hempfling , z. phys .
c * 63 * , ( 1994 ) 309 [ arxiv : hep - ph/9404226 ] ; f. m. borzumati , m. olechowski and s. pokorski , phys .
b * 349 * , ( 1995 ) 311 [ arxiv : hep - ph/9412379 ] ; t. blazek , s. raby and s. pokorski , phys .
d * 52 * , ( 1995 ) 4151 [ arxiv : hep - ph/9504364 ] .
m. carena , m. olechowski , s. pokorski and c. e. m. wagner , nucl .
b * 426 * ( 1994 ) 269 [ arxiv : hep - ph/9402253 ] .
k. hagiwara et al .
d66 _ , 010001 ( 2002 ) .
gladyshev , a.v . ,
kazakov , d.i .
, de boer , w. , burkart , g. and ehret , r. ( 1997 ) _ nucl .
phys . _ * b498 * 3 ; carena , m. et .
al . , ( 1995 ) _ phys .
lett . _ * b355 * 209 .
lep higgs working group , _ phys .
* b 565 * ( 2003 ) 61 , hep - ex/0306033 .
s. heinemeyer , w. hollik and g. weiglein , comput .
. commun .
* 124 * ( 2000 ) 76 [ arxiv : hep - ph/9812320 ] ; ibid eur .
j. c * 9 * ( 1999 ) 343 [ arxiv : hep - ph/9812472 ] .
a. djouadi , j. l. kneur and g. moultaka , arxiv : hep - ph/0211331 .
djouadi , a. , heinemeyer , s. , mondragn , m. and zoupanos , g. , work in progress . | finite unified theories ( futs ) are n=1 supersymmetric grand unified theories , which can be made all - loop finite , both in the dimensionless ( gauge and yukawa couplings ) and dimensionful ( soft supersymmetry breaking terms ) sectors .
this remarkable property provides a drastic reduction in the number of free parameters , which in turn leads to an accurate prediction of the top quark mass in the dimensionless sector , and predictions for the higgs boson mass and the supersymmetric spectrum in the dimensionful sector .
here we examine the predictions of two futs taking into account a number of theoretical and experimental constraints . for the first one we present the results of a detailed scanning concerning the higgs mass prediction , while for the second we present a representative prediction of its spectrum . * finite unified theories and the higgs mass prediction * + abdelhak djouadi@xmath0 , sven heinemeyer@xmath1 , myriam mondragn@xmath2 + and george zoupanos@xmath3 + @xmath0laboratoire de physique mathmatique et thorique , + universit de montpellier ii , france + ` [email protected] ` + @xmath1 dept . of physics , cern ,
th division , 1211 geneva 23 , switzerland + ` [email protected] ` + @xmath2 instituto de fsica , unam , apdo .
postal 20 - 364 , mxico 01000 , d.f .
, mxico + ` [email protected] ` + @xmath3 physics dept .
, nat .
technical university , gr-157 80 zografou , athens , greece + ` [email protected] ` |
a strongly regular graph srg@xmath1 , is a graph with @xmath2 vertices such that each vertex lies on @xmath3 edges ; any two adjacent vertices have exactly @xmath4 common neighbours ; and any two non - adjacent vertices have exactly @xmath5 common neighbours .
we consider the strongly regular graphs constructed from a non - singular quadric @xmath6 in @xmath7 . the _ point - graph _
@xmath8 of @xmath6 has vertices corresponding to the points of @xmath6 .
two vertices in @xmath8 are adjacent if the corresponding points of @xmath6 lie on a line contained in @xmath6 .
it is well known ( see for example @xcite ) that @xmath8 is a strongly regular graph . in this article
we let @xmath9 , and construct from @xmath8 approximately @xmath10 new strongly regular graphs with the same parameters as @xmath8 ( see table [ table3 ] for a precise count ) .
this article proceeds as follows .
section [ background ] contains several preliminary results we need .
section [ sec : const ] describes our construction of a series of infinite families of strongly regular graphs , the proof of the construction is given in section [ sec : pf ] . in section [ sec - cliques ] , we classify and count the maximal cliques in the new graphs . in section
[ sec - new ] we prove that our construction yields new families of strongly regular graphs . finally , in section [
sec : autgp ] , we determine the automorphism group of the new graphs . in previous work , kantor @xcite constructed a strongly regular graph from @xmath8 with the same parameters in the case when @xmath6 contains a spread .
kantor conjects that his graph is not isomorphic to @xmath8 .
we show in section [ sec : kantor ] that the graph constructed by kantor is not isomorphic to any of our new graphs .
abiad and haemers @xcite construct several strongly regular graphs from the symplectic graph over @xmath11 .
the dual of these graphs have the same parameters as the point - graph of a non - singular parabolic quadric , so @xmath12 is even .
it is not known if these graphs are isomorphic to our examples with @xmath12 even .
background in @xcite , godsil and mckay take a graph @xmath13 , and use a vertex partition to construct a new graph @xmath14 that has the same spectrum as @xmath13 .
it is well - known ( see for example @xcite ) that if a graph @xmath14 has the same spectrum as a strongly regular graph @xmath13 , then @xmath14 is also strongly regular with the same parameters as @xmath13 . specialising the godsil - mckay construction to a partition of size two in a strongly regular graph
gives the following result .
gmk 1 .
a _ godsil - mckay partition _ of a graph is a partition of the vertices into two sets @xmath15 satisfying : * the set @xmath16 induces a regular subgraph . *
each vertex in @xmath17 is adjacent to @xmath18 , @xmath19 or @xmath20 vertices in @xmath16 .
2 . _ godsil - mckay construction .
_ let @xmath13 be a strongly regular graph with godsil - mckay partition @xmath15 .
construct the graph @xmath14 with the same points and edges as @xmath13 , except : for each vertex @xmath21 in @xmath17 with @xmath19 neighbours in @xmath16 , delete these @xmath19 edges and join @xmath21 to the other @xmath19 vertices in @xmath16 .
then the graph @xmath14 is strongly regular with the same parameters as @xmath13 .
let @xmath6 be a non - singular quadric in @xmath7 .
the _ projective index _
@xmath22 of @xmath6 is the dimension of the largest subspace contained in @xmath6 .
a @xmath22-space contained in @xmath6 is called a _ generator _ of @xmath6 .
if @xmath23 is even , then a non - singular quadric is called a parabolic quadric , denoted @xmath24 , which has projective index @xmath25 . if @xmath26 is odd , then there are two types of non - singular quadrics : the elliptic quadric denoted @xmath27 has projective index @xmath25 ; and the hyperbolic quadric denoted @xmath28 has projective index @xmath29 .
the points and generators of @xmath6 also form a polar space of rank @xmath30 .
we repeatedly use the following two properties of quadrics and polar spaces , see ( * ? ? ?
* chapter 22 ) for more information on quadrics , and ( * ? ? ?
* section 26.1 ) for more information on polar spaces .
quadric - deg let @xmath6 be a non - singular quadric in @xmath7 and let @xmath31 be a @xmath3-space . if the quadric @xmath32 contains a @xmath33-space , then @xmath32 is either @xmath31 , or one or two @xmath33-spaces .
property - star let @xmath6 be a non - singular quadric in @xmath0 , with projective index @xmath22 .
let @xmath34 be a generator of @xmath6 , and @xmath35 a point of @xmath6 not in @xmath34 .
then there is a unique generator @xmath31 of @xmath6 that contains @xmath35 and meets @xmath34 in a @xmath36-space .
further , the points in @xmath34 which lie on a line of @xmath6 through @xmath35 are exactly the points in @xmath37 .
sec : const | in this article we construct a series of new infinite families of strongly regular graphs with the same parameters as the point - graphs of non - singular quadrics in @xmath0 . corresponding author : dr susan barwick , university of adelaide , adelaide 5005 , australia .
phone : + 61 8 8313 3983 , fax : + 61 8 8313 3696 , email : [email protected] keywords : strongly regular graphs , projective geometry , quadrics ams codes : 51e20 , 05b25 , 05c62 |
the objective of the nemo 3 experiment is to search for the double beta decay process with two ( @xmath0 decay ) or zero ( @xmath1 decay ) neutrinos in the final state in the seven different @xmath5 isotopes listed in table [ tab : t12 ] .
the experimental search for @xmath1 decay is of major importance in particle physics .
if this process is observed then it will reveal the majorana nature of the neutrino ( @xmath6 ) and may allow an access to the absolute neutrino mass scale .
the process @xmath1 decay also violates the principle of lepton number conservation and is , therefore , a direct probe for physics beyond the standard model . in the case of the neutrino - mass mechanism
the @xmath1 decay rate can be written as @xmath7^{-1 } = \langle m_{\nu}\rangle^2 \cdot |m^{0\nu}(a , z)|^2 \cdot g^{0\nu}(q_{\beta\beta},z ) , \label{eq : nme}\ ] ] where @xmath8 is the effective neutrino mass , @xmath9 is the nuclear matrix element ( nme ) , and @xmath10 is the kinematical factor proportional to the transition energy to the fifth power , @xmath11 . the @xmath0 decay process is a rare second order weak interaction process .
the accurate measurement of its rate of decay is important since it constitutes the ultimate background in the search for @xmath1 decay signal and is a valuable input for the theoretical calculations of the nme .
the detector @xcite is located in the lsm in the frejus tunnel at the depth of 4800 m w.e .
it is cylindrical with thin source foils ( @xmath12 ) situated in the middle of the tracking volume surrounded by the calorimeter .
the source foils are composed of almost 10 kg of enriched @xmath5 isotopes , listed in table [ tab : t12 ] .
observation of the @xmath5 decay is accomplished by fully reconstructing the tracks of the two electrons and measuring their energy .
a tracking chamber containing 6180 open drift cells , operating in the geiger mode , provides a vertex resolution of about 1 cm and a 25 g magnetic field is used to curve the tracks for charge identification .
a calorimeter consisting of 1940 plastic scintillator blocks with photomultiplier readout gives an energy resolution of 14 17%/@xmath13 fwhm .
a time resolution of 250ps allows excellent suppression of the external background due to electrons crossing the detector .
the detector is capable of identifying e@xmath14 , e@xmath15 , @xmath16 and @xmath17 particles and allows good discrimination between signal and background events .
the detector is covered by two layers of shielding against external @xmath16 rays and neutrons .
the @xmath18 events are selected by requiring two reconstructed electron tracks with a curvature corresponding to the negative charge , originating from a common vertex in the source foil .
the energy of each electron measured in the calorimeter should be higher than 200kev .
each track must be incident upon a separate scintillator block and the time difference of the two pmt signals is compared to the estimated time as if the two electrons had originated from a common vertex .
the background can be classified in three groups : external one from incoming @xmath19 , radon inside the tracking volume and internal radioactive contamination of the source .
all three were estimated from the nemo-3 data with events of various topologies .
in particular , radon was measured with e@xmath20 and e@xmath21 events and internal @xmath22bi with e@xmath21 .
the e@xmath20@xmath20 , and e@xmath20@xmath20@xmath20 events are used to measure the @xmath23tl activity requiring the detection of the 2.615 mev @xmath20-ray typical of the @xmath23tl @xmath24 decay .
single electron events are used to measure the foil contamination by @xmath24-emitters .
the external background is measured with the events with the detected incoming @xmath20-ray producing an electron in the source foil .
the external background is checked with two - electron events originating from pure copper and natural tellurium foils .
measurements performed with an hpge detector and radon detectors are used to verify the results .
measurements of the @xmath25 decay half - lives have been performed for the seven isotopes available in nemo-3 and for completeness previous results as well as new results are given in table [ tab : t12 ] .
new preliminary results based on higher statistics than used previously are presented here for the three isotopes , @xmath26 , @xmath27 and @xmath28 .
.nemo-3 results of @xmath25 decay half - life measurements for seven isotopes .
[ tab : t12 ] [ cols="^,^,^,^,^",options="header " , ] the measurement of the @xmath27 half - life was performed using the data collected for 925 days .
a total of 678 events was selected , with an expectation of 328 background events .
the largest background contribution is due to the internal @xmath29 contamination of the sample .
the distributions of the sum of the energies of the two electrons and their angular separation are shown in fig .
[ fig : bb_zr96 ] , demonstrating good agreement between the data and the monte carlo simulation .
the @xmath25 efficiency is estimated to be 7.6% .
the measured half - life is @xmath30\cdot10^{19}y$ ] .
zr.,title="fig:",scaledwidth=41.0%,height=162 ] zr.,title="fig:",scaledwidth=41.0%,height=162 ] the measurement of the @xmath26 half - life was performed using the data collected for 943 days .
a total of 133 events was selected , with 17 background events expected .
the two - electron energy sum distribution and single - electron energy spectrum are presented in fig .
[ fig : bb_ca48 ] .
the @xmath31ca sample is known to be contaminated with @xmath32sr ( t@xmath33=28.79y , @xmath34=0.546mev ) .
its daughter @xmath32y ( t@xmath35=3.19h , @xmath36=2.282mev ) is the major background source in this case .
an activity of approximately 1700mbq / kg was measured for @xmath32y using single - electron events .
both @xmath37sr and @xmath32y are essentially pure @xmath38 emitters and imitate @xmath18 events through mller scattering . to suppress this background contribution , events with the energy sum greater than 1.5 mev and @xmath39
are selected .
the @xmath25 efficiency is 3.3% , and the measured half - life is @xmath40\cdot10^{19}y$ ] .
ca.,title="fig:",scaledwidth=41.0%,height=162 ] ca.,title="fig:",scaledwidth=41.0%,height=162 ] the measurement of the @xmath28 half - life was performed using the data collected within 939 days .
a total of 2853 events was selected , with 765 background events expected .
the distributions of the sum of the energies of the two electrons and their angular separation are shown in fig .
[ fig : bb_nd150 ] , demonstrating good agreement between the data and the monte carlo simulation .
the @xmath25 efficiency is estimated to be 7.6% and the half - life is measured to be @xmath41 \cdot10^{19}y$ ] .
nd.,title="fig:",scaledwidth=41.0%,height=162 ] nd.,title="fig:",scaledwidth=41.0%,height=162 ] in the case of the mass mechanism , the @xmath1 decay signal is expected to be a peak in the energy sum distribution at the position of the transition energy @xmath42 ( with a long non - gaussian tail due to energy losses in the source foil ) .
since no excess is observed at the tail of the distribution for @xmath43zr , see fig .
[ fig : bb_zr96 ] ( left ) , nor for @xmath44ca , fig .
[ fig : bb_ca48 ] ( left ) , limits are set on the neutrinoless double beta decay t@xmath45 using the cls method @xcite . a lower half - life limit is translated into an upper limit on the effective majorana neutrino mass @xmath8 ( see eq .
[ eq : nme ] ) .
the following results are obtained using the nme values from @xcite for @xmath43zr and from @xcite for @xmath44ca : t@xmath46 ( 90% c.l . ) , @xmath47 7.4 20.1ev and t@xmath48 ( 90% c.l . ) , @xmath47 29.7ev .
similarly a limit has been set for @xmath8 for @xmath28 .
it is complicated by whether the deformation of the nucleus is taken into account .
the limit has been calculated for both scenarios : t@xmath49 ( 90% c.l . )
with @xmath47 1.7 2.4ev using qrpa model ( no deformation ) @xcite and @xmath47 4.8 7.6ev using so(3 ) model ( with deformation ) @xcite .
all of these limits should be compared with the one set previously by nemo-3 of @xmath47 0.8
1.3ev for @xmath50 @xcite .
no evidence was found for @xmath1 decay and limits on the neutrino mass , @xmath8 , have been set .
new measurements for the half - lives due to @xmath25 decay for the three isotopes , @xmath26 , @xmath27 and @xmath28 has been presented .
the nemo-3 experiment continues taking data but will eventually be superceded with a new experiment called supernemo .
the baseline design of supernemo envisages using 100 kg of @xmath51 or @xmath28 to reach a sensitivity to the majorana neutrino mass of 0.05 0.1ev . | the nemo-3 experiment is located in the modane underground laboratory ( lsm ) and has been taking data since 2003 with seven isotopes .
it is searching for the double beta decay process with two ( @xmath0 ) or zero ( @xmath1 ) neutrinos emitted in the final state .
precision measurements of the half - life of the isotopes due to @xmath0 decay have been performed and new results for @xmath2\cdot10^{19}y$ ] , @xmath3\cdot10^{19}y$ ] and @xmath4 \cdot10^{19}y$ ] are presented here .
measurements of this process are important for reducing the uncertainties on the nuclear matrix elements .
no evidence for @xmath1 decay has been found and a 90% confidence level lower limit on the half - life of this process is derived . from this an upper limit can be set on the effective majorana neutrino mass using the most recent nuclear matrix elements ( nme ) calculations . |
during last two decades fe - rich intermetallic systems stays under investigation from both experimental and theoretical points of view because its anomalous magnetic properties . at present
the intermetallic compounds with rare - earth elements draw a significant interest because of their well pronounced invar properties and also as possible new cheap high - energy storing materials for permanent magnets .
@xcite rare - earth intermetallic compounds r@xmath0fe@xmath1 ( r a rare - earth ion ) are well - known due to large magnetic moment of fe ions and yet their comparatively low curie temperature @xmath8 .
further increase of the @xmath8 of the compounds can be achieved by implantation of interstitial atoms @xcite or introducing a low content of non - magnetic impurities substituting fe atoms.@xcite in such extended systems , compared to the parent r@xmath0fe@xmath1 compounds , upon the doping @xmath8 grows more than twice in some cases .
this increment of @xmath8 was explained primarily as a result of the lattice expansion with substitution of fe ions for larger radii ions . according to this concept , fe - fe interactions
are ferromagnetic or antiferromagnetic for interatomic distance larger or smaller then the critical value 0.25 nm.@xcite however , a number of experiments @xcite showed that in the substitutional r@xmath0fe@xmath9si@xmath10 alloys the crystal lattice contracts and the magnetic moment per unit volume decreases , while @xmath8 grows .
later experimental facts are in apparent contradiction with the model .
@xcite clearly , a simple approach based on distance - dependent exchange interaction is not sufficient to explain the changes of magnetic properties of these compounds with substitution of fe for non - magnetic ions .
past life several results of band structure calculations of the systems from the family are available .
for instance the spin - polarized calculations of electron spectra of y@xmath0fe@xmath1n@xmath11,@xcite sm@xmath0fe@xmath9m@xmath10 ( m = al , ga , si),@xcite and nd@xmath0fe@xmath9m@xmath10 ( m = si , ga ) @xcite compounds showed that non - magnetic impurities modify the shape and width of the spin - up and spin - down densities of states @xmath12 .
jaswal @xmath13 @xmath14 .
@xcite suggested to explain the increase of @xmath8 in y@xmath0fe@xmath1n@xmath11 via changes of the density of states at the fermi level @xmath15 owing to the lattice expansion by means of the mohn - wohlfarth static spin - fluctuation model .
@xcite an analysis of the experimental optical , @xcite low - temperature heat capacity@xcite and photoemission@xcite data of some pseudobinary alloys r@xmath0fe@xmath9m@xmath10 with m = al , si revealed a qualitative correlation between the curie temperature and parameters of electronic structure in frame of such an approach .
recent experimentally measured optical conductivity of ce@xmath0fe@xmath1 was interpreted in terms of band structure obtained within local density approximation ( lda).@xcite in accordance with the foregoing systematic study of the electronic structure of r@xmath0fe@xmath1 compounds and their modifications with substituted fe ions further investigations are of fundamental importance . in this work theoretical calculations of the electronic properties together with experimental magnetic and optical measurements
were performed .
the paper is organized as follows : in the section [ exp ] experimental details and results for pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 are presented .
for instance , subsection [ mag_mes ] is devoted to sample preparation , x - ray diffraction structural analysis and magnetic measurements conditions .
subsection [ opt_mes ] provides description of optical experiments .
section [ lsdau ] contains results of lsda+u @xcite computations of electronic structure and magnetic properties of pr - fe and gd - fe systems .
structure of experimentally observed optical conductivity curves is anatomized in the section [ opt_theor ] . at the end
we briefly summarize our paper with the section [ summary ] .
the compounds pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 were prepared by induction melting in alumina crucible under argon atmosphere .
the ingots were homogenized in the high - purity argon atmosphere at @xmath161300 k. the purity of the alloys was checked using standard x - ray diffractometry in cu k@xmath17 radiation .
the samples were found to be polycrystalline , single - phase and have rhombohedral structure of the @xmath18-type@xcite ( space group r@xmath19 m ) for pr@xmath0fe@xmath1 and hexagonal crystal structure of the @xmath20-type@xcite ( space group p6@xmath11/mmc ) for gd@xmath0fe@xmath1 .
the measured lattice parameters @xmath21 and @xmath22 are close to those published earlier@xcite and are shown in the table [ tab1 ] .
spherical specimens of 2 - 3 mm in diameter were used for magnetic measurements . for following optical studies the specular surface of the samples
was prepared by mechanical polishing with diamond pastes .
the curie points of the compounds were determined from temperature dependencies of @xmath23 susceptibility .
the @xmath23 susceptibility was measured by a differential method in an @xmath23 magnetic field of 8 oe with a frequency of 80 hz .
the saturation magnetizations at @xmath24=4.2 k were determined from the isothermal magnetization measurements carried out by means of vibrating sample magnetometer in magnetic fields up to 20 koe .
the magnetic parameters of the compounds are listed in table [ tab1 ] .
they are a little different from previously reported data .
@xcite the differences among authors may originate from some deviations of their samples from stoichiometry . [ cols="^,^,>,^,^,^,^,^,^",options="header " , ] [ tab1 ] investigation of optical properties of pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 were carried out at room temperature by ellipsometric beattie technique .
@xcite spectroscopic ellipsometry is based on the fact that the state of polarization of incident light is changed on reflection .
the optical constants refractive index @xmath25 and absorption coefficient @xmath26 were measured in the spectral range of @xmath27=0.0775.6 ev ( @xmath28 is a cyclic frequency of light ) with accuracy of 24% . from @xmath25 and @xmath26 ,
the real @xmath29 and imaginary @xmath30 parts of the complex dielectric function @xmath31 , the optical conductivity @xmath32 , and the reflectance @xmath33/[(n+1)^2+k^2]$ ] were derived .
measurements of reflection spectra followed by the kramers - kronig analysis were applied to determine the optical parameters in the short - wave range ( @xmath27=5.68.5 ev ) .
the results of optical study ( @xmath34 as functions of @xmath28 ) for the pr and gd compounds are shown in fig .
it is seen that the optical properties of both compounds are rather similar .
all directly measured dispersions and further derived quantities are characterized by the broad feature with maximum at the photon energies near 1 ev . as it follows from the lineshape of the @xmath35 ,
there is a strong absorption region at energies @xmath361 ev .
with increase of the wavelength to infrared range ( @xmath372 @xmath2 m ) ( inset of fig .
[ fig1]a ) the nonmonotonic behaviour of @xmath25 and @xmath26 is changed by a smooth growth related to the domination of free - electron absorption .
such a character of the frequency dependencies for the optical constants together with negative quantities @xmath38 , as a rule , is typical for the metal - like solids .
as one can see from dispersion curves @xmath39 , in the low - energy range the reflectance exhibits the high values .
the analysis of the energy dependence of @xmath38 and @xmath40 in this spectral interval , corresponding to intraband electronic excitations , makes it possible to determine the plasma frequency @xmath3 and the effective relaxation frequency @xmath4 of free charge carriers . within the assumption that light absorption for these energies has drude character , the parameters @xmath41 and @xmath42 were computed . in the long - wave region @xmath438
@xmath2 m , @xmath4 and @xmath44 become frequency independent .
the values of @xmath4 and @xmath44 in this energy interval were used then to estimate the effective concentration of conduction electrons @xmath45 ( @xmath46 and @xmath47 are the mass and the charge of a free electron respectively ) .
all these parameters obtained from experimental data treatment are presented in table [ tab1 ] .
optical conductivities @xmath48 for both compounds are displayed in fig .
a monotonic increment of the experimental @xmath48 dispersion observed in the low - energy range is related to the drude mechanism of electron excitations .
for both systems intraband ( drude - like ) contribution to the optical conductivity were computed according to the relation @xmath49 and drown in fig .
[ fig2 ] black dotted lines .
corresponding values of @xmath44 and @xmath4 are given in table [ tab1 ] .
the magnitude of these contributions falls down sharply with energy and becomes insignificant above 0.5 ev .
the values of static conductivity @xmath50 are estimated to be 0.89 * 10@xmath51 s@xmath52 for pr@xmath0fe@xmath1 and 1.03 * 10@xmath51 s@xmath52 for gd@xmath0fe@xmath1 correspondingly .
with increase of photon energy the @xmath48 curves show a very intense asymmetric stuctures in the near infrared region of spectra at @xmath161.2 ev .
these structures have a pronounced shoulder on high - energy side and abrupt low - energy edge .
such a behaviour is a typical manifestation of the predominance of the interband absorption in this energy interval .
the similar shape of the @xmath48 curves was early observed in y@xmath0fe@xmath1 , ce@xmath0fe@xmath1 , and lu@xmath0fe@xmath1 compounds .
to calculate electronic structure of the intermetallic compounds under investigation lsda+u method @xcite within the tb - lmto - asa package ( tight binding , linear muffin - tin orbitals , atomic sphere approximation ) @xcite was applied .
experimentally obtained values of lattice constants for both pr and gd systems given in table [ tab1 ] were used in our calculations .
atomic spheres radii were chosen as @xmath53(pr ) = 3.91 a.u . and @xmath53(fe )
= 2.62 a.u . for pr@xmath0fe@xmath1 and @xmath53(gd ) = 3.72 a.u . and @xmath53(fe )
= 2.66 a.u . for gd@xmath0fe@xmath1 .
orbital basis consists of 6@xmath54 , 6@xmath55 , 5@xmath56 , and 4@xmath57 muffin - tin orbitals for pr or gd and 4@xmath54 , 4@xmath55 , and 3@xmath56 for fe sites .
the calculations were performed with 32 irreducible * k*-points ( 6@xmath586@xmath586 spacing ) in the first brillouin zone .
parameters of direct @xmath59 and exchange @xmath60 coulomb interactions for gd and pr ions were calculated by constrained lda method .
@xcite for gd@xmath0fe@xmath1 we obtained @xmath61=6.7 ev and @xmath62=0.7 ev ( similar values were determined previously for elemental gd @xcite ) , and for pr@xmath0fe@xmath1 @xmath63=4.9 ev and @xmath64=0.6 ev .
corresponding values of @xmath59 and @xmath60 were applied to gd and pr compounds in frame of the lsda+u method . to note , in the present work we do not take into account local coulomb interaction on fe ions since constrained lda gives surprisingly large @xmath59 values.@xcite however , for elemental fe it was shown that account of coulomb correlations is important to describe semiquantitatively temperature dependence of magnetization.@xcite but on the other hand such approach reproduces main structures of lsda dos with slight modifications of the fe 3@xmath56 dos and thus in our case will not affect resulting dispersions of optical conductivity strongly .
total magnetic moments per formula units obtained within lsda+u method ( 33.79 @xmath65 for pr@xmath0fe@xmath1 and 21.89 @xmath65 for gd@xmath0fe@xmath1 ) are in good agreement with measured experimental data ( see table [ tab1 ] ) .
values of local magnetic moments for different sites for pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 compounds are listed in table [ tab2 ] .
pr and gd 4@xmath57-shells are computed to be fully polarised with local moments close to its ionic values while fe ions have local moments values almost equal to its elemental fe magnitude .
it is remarkable that local moments on rare - earth sites during selfconsistent loops become oppositely directed to those on fe sites .
one should also mention that initial value of local moments on pr and gd ions was taken to be zero .
thus r and fe sublattices in our calculations are obtained to be antiferromagnetically ordered .
calculated in the present work within the lsda+u method partial fe @xmath66 and rare - earth @xmath67 dos for spin - up ( @xmath68 ) and spin - down ( @xmath69 ) projections of local spin moments for pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 are shown in figs .
[ fig3 ] and [ fig4 ] .
it is seen that main spectral weight is located in the @xmath70 5 ev energy range around the fermi level .
structures of these dos are rather similar for both compounds and are qualitatively close to the dos of ferromagnetic elemental iron in @xmath71 structure .
@xcite both these spin - polarized doss have a two - peak structure related to the @xmath6 states of fe ions . the fermi level @xmath72 set to zero lies near the minimum between these two peaks in the spin - down channel and on the upper edge in the spin - up channel .
`` spin - up '' states of fe ions are almost completely occupied while `` spin - down '' are nearly half - filled .
the narrow intensive peaks at 3 ev ( @xmath69 ) and 2.7 ev ( @xmath68 ) ( pr@xmath0fe@xmath1 ) also at @xmath73 ev ( @xmath69 ) and 3.5 ev ( @xmath68 ) ( gd@xmath0fe@xmath1 ) belong to @xmath74 states of rare - earth ions .
the intensities of fe @xmath75 and rare - earth @xmath76 , and @xmath77 contributions to the dos are considerably smaller .
ldld & + site & & site & + pr(@xmath78 ) & -2.08 & gd(@xmath79 ) & -7.13 + fe(@xmath78 ) & 2.21 & gd(@xmath80 ) & -7.20 + fe(@xmath81 ) & 2.23 & fe(@xmath74 ) & 2.31 + fe(@xmath82 ) & 2.13 & fe(@xmath83 ) & 2.10 + fe(@xmath84 ) & 2.34 & fe(@xmath85 ) & 2.40 + & & fe(@xmath86 ) & 1.87 + [ tab2 ]
calculated lsda+u band structures presented in the preceding section were used to interpret experimental optical conductivity @xmath48 for both intermetallic systems under consideration . in order to calculate theoretical optical conductivity @xmath87 and anatomize its ingredients we applied rather simplified technique similar in spirit to ref .
namely , we computed following convolutions representing all possible optical transitions @xmath88 where @xmath89 and @xmath90 are partial doss of the same ion @xmath91 with the same spin @xmath54 and orbital quantum numbers @xmath92 and @xmath93 shown in figs .
[ fig3 ] and [ fig4 ] .
the orbital quantum numbers are related by dipol selection rule @xmath94 .
total theoretical optical conductivity is linear combination of different contributions ( [ occ ] ) : @xmath95 where @xmath96 is the degrees of site degeneracy ( wyckoff positions )
. figs .
[ fig2 ] and [ fig5 ] display corresponding @xmath87 as solid lines .
the overall shape of optical conductivity dispersion curves for both compounds exhibits a broad structure with the peak at @xmath162 ev . on the whole
the theoretical calculations reproduce main features of the experimental @xmath48 curve ( fig .
[ fig2 ] ) : ( i ) the width of the intense absorption band with distinct maximum , ( ii ) the sharp threshold in the 0.51 ev range and ( iii ) the gradual diminution of the high - energy slope . at the same time
there are certain discrepancies .
the theoretical peaks are slightly shifted towards the higher energies in contrast to the experimental ones
. also high - energy contributions at 4 ev are slightly overestimated .
it may have following reasons .
as was shown by lichtenstein @xmath97@xcite inclusion of local dynamical coulomb interactions into consideration in elemental fe will lead to about 30% of correlation bandwidth narrowing . in our case dos in the vicinity of the fermi level
consists mostly from fe 3@xmath56 states which are almost identical to those of elemental @xmath71 iron .
roughly one can say that for our intermetallic compounds it can slightly improve comparison with experiment .
first of all peak at 2 ev could be moved on 0.50.6 ev towards lower energies because of correlation narrowing .
second spectral weight at 4 ev can be transfered to lower energies and it can give high - energy tail of theoretical optical conductivity closer to experimental one .
furthermore for both systems @xmath87 shows a significant interband absorption in the low - energy range ( @xmath980.6 ev ) which was not confirmed by the measurements .
it is possible that the low - energy absorption , predicted in theory , may be partially disguised in the experimental @xmath48 curves by the strong drude rising .
the drude contribution was not taken into consideration in our theoretical model .
densities of states near the fermi energy for both compounds mostly consist of the 3@xmath56-states of various crystallographically inequivalent iron ions ( see figs .
[ fig3 ] and [ fig4 ] ) .
a line shape of our lsda calculated fe ions dos is very similar to the known dos of elemental fe in the @xmath71 ( @xmath99 ) phase .
@xcite for the @xmath71 iron it was shown @xcite that the optical conductivity determined using the berglung - spicer approach @xcite agrees rather well with theoretical calculations accounting matrix elements@xcite and also with experimental results.@xcite this fact enables us to suppose that matrix elements formalism does not play an important role in case of intermetallic compounds r@xmath0fe@xmath1 . in this work
we can report our theoretical curves to be in a reasonable qualitative agreement with experimental data . to analyze the line shape of experimental @xmath48 we provide detailed description of different contributions to @xmath87 . as it is seen in fig .
[ fig5 ] for both systems biggest contribution to @xmath87 comes from @xmath6@xmath7 interband optical transitions for fe ions through the spin - down channel .
that gives a high absorption peak at @xmath162 ev ( see dotted curves in fig . [ fig5 ] ) .
these particular transitions mostly govern low - energy range behaviour below 1 ev as well .
second largest contribution is @xmath6@xmath7 transitions for fe ions but in the spin - up channel . in fig .
[ fig5 ] one can see it as a rather broad structure with maxima at @xmath164 ev ( dot - dashed curves ) .
the contributions from fe @xmath100-@xmath7 ( fig . [ fig5 ] , dashed - dot - dot lines ) and pr(gd ) @xmath74-@xmath77 ( fig .
[ fig5 ] , thin solid line ) transitions are substantially smaller ; and magnitude of convolutions of rare earths @xmath77-@xmath101 and @xmath102-@xmath101 states is negligible .
comprehensive experimental investigation of structural , magnetic and optical properties of the intermetallic compounds pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 was performed during this work .
refined structural parameters are found to be in good agreement with previous data .
measured magnetic properties : ( i ) curie temperatures @xmath8=294k ( 466k ) and ( ii ) saturation magnetizations ( at @xmath24=4.2k ) 36.1 ( 21.2 ) @xmath65/f.u . for pr - fe ( gd - fe )
systems respectively also agree well with available in the literature experimental data . we also report measured for the first time optical constants @xmath25 and @xmath26 observed at room temperature by ellipsometric beattie technique in the spectral range of 0.2215 @xmath2 m .
this experimental data allows us to determine charge carriers parameters ( plasma @xmath3 and relaxation @xmath4 frequencies ) and optical conductivity @xmath103 . to model magnetic and optical properties of pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 we did self - consistent spin - resolved calculations within the lsda+u method .
calculated by lsda+u method magnetic moments per formula unit describe well observed experimental values . furthermore experimental optical conductivity @xmath5 was interpreted in terms of convolutions between partial densities of states for the same ion applying dipole selection rule for orbital quantum number .
overall the line shape of experimental and theoretical optical conductivity curves was found to be qualitatively very similar . by anatomizing different contributions to @xmath87
it was understood that transitions between @xmath6 and @xmath7 states of fe ions give the biggest contribution .
the intense peaks in optical conductivity below 1 ev and around 2 ev are predominantly formed by the transitions in the `` spin - down '' channel while the second large contribution from `` spin - up '' channel transitions provides broad feature with the maximum at about 4 ev .
the other contributions from rare - earth @xmath74-@xmath77 and iron @xmath100-@xmath7 optical transitions are almost negligible .
this work was supported by grants of russian foundation for basic research 05 - 02 - 17244 , 04 - 02 - 16086 and 05 - 02 - 16301 and in part by programs of the presidium of the russian academy of sciences ( ras ) `` quantum macrophysics '' and of the division of physical sciences of the ras `` strongly correlated electrons in semiconductors , metals , superconductors and magnetic materials '' .
two of us ( in , al ) acknowledge dynasty foundation and international center for fundamental physics in moscow and russian science support foundation ( in ) .
99 j. m. d. coey and h. sun , j. magn .
magn . mater . *
87 * , l251 ( 1990 ) ; h. sun , j. m. d. coey , y. otani , and d. p. f. hurley , j. phys . : condens .
matter * 2 * , 6465 ( 1990 ) .
d. b. de mooij , k. h. j. buschow , j. less - common . met .
* 142 * , 349 ( 1988 ) .
p. c. m. gubbens , a. m. van der kraan , t. h. jacobs , and k. h. j. buschow , j. less - common met .
* 159 * , 173 ( 1990 ) .
t. h. jacobs , k. h. j. buschow , g. f. zhou , x. li , and f. r. de boer , j. magn .
magn . mater . *
116 * , 220 ( 1992 ) .
z. wang and r. a. dunlap , j. phys .
: condens .
matter * 5 * , 2407 ( 1993 ) .
b. g. shen , f. w. wang , l. s. kong , and l. cao , j. phys . : condens .
matter * 5 * , l685 ( 1993 ) .
a. g. kuchin _ et al .
( a ) * 155 * , 479 ( 1996 ) .
d. givord , r. lemaire , ieee trans . magn . *
mag-10 * , 109 ( 1974 ) .
r. van mens , j. magn .
magn . mater . *
61 * , 24 , ( 1986 ) .
z. w. li , x. z. zhow , and a. h. morrish , phys .
b * 51 * , 2891 ( 1995 ) .
a. g. kuchin , a. s. ermolenko , v. i. khrabrov , phys . met .
metallogr . * 86 * , 276 ( 1998 ) . s. s. jaswal , w. b. yelon , g. c. hadjipanayis , y. z. wang , and d. j. sellmyer , phys . rev . lett .
* 67 * , 644 ( 1991 ) .
r. f. sabirianov and s. s. jaswal , j. appl .
79 * , 5942 ( 1996 ) .
ming - zhu huang and w. y. ching , j. appl . phys . * 79 * , 5545 ( 1996 ) .
p. mohn and e. p. wohlfarth , j. phys .
* 17 * , 2421 ( 1987 ) .
v. knyazev , a. g. kuchin , and yu . i. kuzmin , phys . met .
metallogr . * 89 * , 558 ( 2000 ) .
v. knyazev , a. g. kuchin , and yu .
i. kuzmin , j. alloys comp .
* 327 * , 34 ( 2001 ) .
j. p. woods _
et al_. , phys .
b * 51 * 1064 ( 1995 ) .
i. a. nekrasov , yu . v. knyazev , yu . i. kuzmin , a. g. kuchin , and v. i. anisimov , phys . met .
metallogr . * 97*(2 ) , 129 ( 2004 ) .
v. i. anisimov , f. aryasetiawan , and a. i. lichtenstein , j. phys . : condens .
matter * 9 * , 767 ( 1997 ) .
q. johnson , d. h. wood , and g. s. smith , acta crystallographica b * 24 * , 274 ( 1968 ) .
f. givord and r. lemaire , j. less - common met .
* 21 * , 463 ( 1970 ) .
x. c. kou , f. r. de boer , r. grossinger , g. wiesinger , h. suzuki , h. kitazava , t. takamasu , g. kido , j. magn .
magn . mater . * 177 - 181 * , 1002 ( 1998 ) .
r. c. mohanty , c. zhang , s. a. shaheen , a. murugaiah , a. raman , c. g. grenier , and r. e. ferrel jr .
, j. phys . : condens .
matter * 12 * , 9657 ( 2000 ) .
j. r. beattie and g. m. conn , philos . mag . * 46 * , 235 ( 1955 ) .
o. k. andersen , phys .
b * 12 * , 3060 ( 1975 ) .
o. gunnarsson , o. k. andersen , o. jepsen , and j. zaanen , phys .
b * 39 * , 1708 ( 1989 ) . v. i. anisimov and o. gunnarsson , phys . rev .
b * 43 * , 7570 ( 1991 )
. a. i. lichtenstein , m. i. katsnelson , and g. kotliar , phys .
lett . * 87 * , 067205 ( 2001 ) .
o. gunnarsson , physica bc * 91 * , 329 ( 1977 ) . c. n. berglung and w. e. spicer , phys
. rev . * 136*. a1030 ( 1964 ) ; c. n. berglung and w. e. spicer , phys . rev .
* 136*. a1044 ( 1964 ) .
s. wakon , j. yamashita , j. phys .
japan * 21 * , 1712 ( 1966 ) . g. a. bolotin , m. m. kirillova , v. m. maevskii , phys . met .
metallogr . * 27*(2 ) , 224 ( 1969 ) .
m. singh , c. s. wang , and j. callaway , physical review b * 11 * , 287 ( 1975 ) ; k. j. kim , t. c. leung , b. n. harmon , and d. w. linch , j. phys .
: condens .
matter * 6 * , 5069 ( 1994 ) .
v. p. shirokovskii , m. m. kirillova , n. a. shilkova , jetp * 82*(3 ) , 784 ( 1982 ) . | in this paper we report comprehensive experimental and theoretical investigation of magnetic and electronic properties of the intermetallic compounds pr@xmath0fe@xmath1 and gd@xmath0fe@xmath1 . for the first time electronic structure of these two systems
was probed by optical measurements in the spectral range of 0.2215 @xmath2 m . on top of that charge carriers parameters (
plasma frequency @xmath3 and relaxation frequency @xmath4 ) and optical conductivity @xmath5 were determined .
self - consistent spin - resolved bandstructure calculations within the conventional lsda+u method were performed .
theoretical interpetation of the experimental @xmath5 dispersions indicates transitions between @xmath6 and @xmath7 states of fe ions to be the biggest ones .
qualitatively the line shape of the theoretical optical conductivity coincides well with our experimental data .
calculated by lsda+u method magnetic moments per formula unit are found to be in good agreement with observed experimental values of saturation magnetization . |
at present , the best evidence in favor of @xmath2 oscillations comes from the super - kamiokande ( sk ) atmospheric @xmath2 experiment @xcite , corroborated the macro and soudan2 measurements @xcite .
the sk data , in general , indicate that ( at least ) two mass eigenstates ( say , @xmath3 and @xmath4 ) are nondegenerate @xmath5 , and that @xmath3 is mixed with active flavor states ( @xmath6 ) and , possibly , with a fourth sterile state ( @xmath7 ) , @xmath8 where one has @xmath9 from the data and @xmath10 from unitarity .
i briefly review some mixing scenarios relevant for atmospheric @xmath2 phenomenology , pointing out also their interplay with solar @xmath2 s @xcite and with lsnd @xcite , as well as some outstanding issues .
such scenarios are parametrized in table 1 , where @xmath11 , @xmath12 , and @xmath13 are appropriate neutrino mixing angles @xmath14 , and @xmath15 is the relevant atmospheric @xmath2 squared mass difference ( @xmath16 ) in the various cases ( see @xcite for details on the @xmath17 , @xmath18 , and @xmath19 notation ) . assuming @xmath17 mixing , the present atmospheric data @xcite provide us with two solid bounds , @xmath20 \sin^2\psi & \simeq & 0.50\pm 0.17\ , \\
\end{array}\right.\ ] ] that are not significantly modified @xcite when passing to @xmath18 and @xmath19 scenarios .
( or @xmath21 ) rather than @xmath22 , especially in contexts where physics can be asymmetric in the two octants of a mixing angle @xmath23 @xcite .
] ccccccc + + case & @xmath3&@xmath15&@xmath24 & @xmath25 & @xmath26 & @xmath27 + + @xmath17&@xmath28&@xmath29 & & @xmath30 & @xmath31 & + @xmath18&@xmath32&@xmath33&@xmath34&@xmath35&@xmath36 & + @xmath19&@xmath37&@xmath38 & & @xmath30 & @xmath39 & @xmath40 + + outstanding @xmath17 issues : @xmath41 can the @xmath15 uncertainty be significantly reduced before future long - baseline projects @xcite ?
@xmath42 is @xmath43 mixing exactly maximal @xmath44 or not ?
@xmath45 besides monotonic @xmath46 disappearance , can we also observe one @xmath46 disappearance+appearance cycle @xcite ?
in atmospheric @xmath2 phenomenology , the small squared mass difference @xmath47 is often neglected . in this case , @xmath48 mixing ( see table 1 ) .
detailed analyses of sk data show that the bounds in eq .
( 2 ) are basically preserved for unconstrained @xmath12 @xcite , and that an upper limit can be placed on @xmath49 .
such limit is significantly strengthened by the chooz reactor results : @xmath50 the bounds ( 2)(4 ) are compatible with all typical solutions to the solar @xmath2 problem .
figure 1 presents the current solutions in the @xmath51 mass - mixing plane @xmath52 , at fixed values of @xmath12 @xmath53 .
unfortunately , there is still an embarrassing multiplicity of allowed @xmath54 regions @xcite .
however , such regions typically favor relatively small values of @xmath12 , thus providing a nontrivial , independent consistency check of the bounds ( 3,4 ) in @xmath18 scenarios .
outstanding @xmath18 issues : @xmath41 can sk atmospheric data reveal @xmath55 @xcite ?
@xmath42 if @xmath56 , how far can we push the experimental sensitivity to earth matter effects @xcite and to sign(@xmath57 ) @xcite ?
@xmath45 are there subleading effects ( e.g. , cp violation @xcite ) induced by @xmath58 on atmospheric @xmath2 oscillations ( or , conversely , by @xmath15 on solar @xmath2 oscillations ) ?
@xmath59 can the multiplicity of solar @xmath2 solutions be reduced in the near future @xcite ?
attempts to reconcile all the solar , atmospheric , and lsnd pieces of evidence for oscillations require @xmath19 scenarios with one sterile neutrino @xmath7 , and mass spectra of the kind `` 2 + 2 '' @xcite or `` 1 + 3 '' @xcite . in the 2 + 2 case
, the atmospheric @xmath2 mixing can be approximately described as in the third line of table 1 , involving the usual @xmath60 parameters plus an additional mixing @xmath61 @xcite . for @xmath62 ( @xmath63 ) ,
the atmospheric oscillation channel is purely @xmath64 ( @xmath65 ) ; intermediate values of @xmath61 correspond instead to mixed active+sterile oscillations .
a recent analysis of sk atmospheric results @xcite in the 2 + 2 scenario shows that the bounds in eq .
( 2 ) are not significantly modified , and that there is room for relatively large values of @xmath61 : @xmath66 this result shows that , although sk @xcite ( as well as macro @xcite ) disfavors pure sterile oscillations @xmath67 , one could have sizable sterile mixing besides the standard @xmath68 mixing @xcite . independently , one can study how the usual @xmath54 solar @xmath2 solutions are modified for generic @xmath61 in @xmath19 models @xcite .
it turns out that solar @xmath2 data prefer relatively large values of @xmath61 @xcite .
one can roughly say that , at present , @xmath69 up to variations related to the chosen solution .
therefore , unlike the @xmath18 case , the @xmath19 case shows some `` tension '' between atmospheric ( 5 ) and solar ( 6 ) results not strong enough , however , to prevent their combination . from ( 5,6 ) one might get a possible indication for nonzero @xmath7 mixing , @xmath70 which favors maximal amplitude of both active and sterile channels in atmospheric and solar @xmath19 oscillations ( `` fourfold maximal mixing '' @xcite ) .
outstanding @xmath19 issues : @xmath41 @xmath41 will the solar - atmospheric data `` tension '' be weakened or not by future data ?
@xmath42 can we detect matter effects associated to @xmath7 mixing @xcite ? @xmath45
how can we discriminate the 2 + 2 and 1 + 3 spectra @xcite ?
the bounds from atmospheric @xmath17 oscillations in the @xmath64 channel [ eq .
( 2 ) ] are robust , and are preserved by @xmath18 and @xmath19 extensions . in @xmath18 scenarios , the sk+chooz data provide a bound [ eq . ( 4 ) ] on the amplitude of @xmath71 mixing . in @xmath19 scenarios , the solar+atm data are still compatible with nonzero @xmath7 mixing within present uncertainties [ eq .
further theoretical and experimental work is needed to nail down the atmospheric @xmath2 mixing amplitudes in table 1 , as well as to reduce the multiplicity of solar @xmath2 solutions in fig . 1 . 99 t. kajita , these proceedings .
f. ronga , these proceedings .
y. suzuki , these proceedings .
spentzouris , these proceedings .
fogli _ et al .
_ , prd 54 , 2048 ( 1996 ) .
fogli _ et al .
_ , prd 59 , 033001 ( 1999 ) .
fogli _ et al .
_ , hep - ph/0009299 .
j. hill , these proceedings .
p. antonioli , these proceedings .
d. montanino , these proceedings .
m. chizhov , these proceedings .
b. gavela , these proceedings .
m. lindner , these proceedings .
minakata , these proceedings .
palazzo , these proceedings .
noble , these proceedings .
e. meroni , these proceedings . c. giunti , these proceedings .
o. peres , these proceedings .
marrone , these proceedings . | some aspects of the current neutrino oscillation phenomenology are briefly discussed , with emphasis on the status of mass - mixing parameters relevant for scenarios with three active neutrinos @xmath0 , and with an additional fourth sterile neutrino @xmath1 . |
in recent years , large multi - wavelength photometric surveys conducted with the @xmath17 and @xmath0 @xmath18 @xmath19 have enabled us to study the properties of galaxies over cosmic time .
synthetic stellar population modeling of broadband spectral energy distributions ( seds ) has led to the determination of various physical properties ( e.g. , stellar mass , star formation rate , age , dust extinction ) of these galaxies ( ; ; ; ; , ; finkelstein et al .
2012 ; ; ; ) . of all derived quantities , stellar masses are particularly robust in stellar population fits .
small changes in the age of the stellar populations , metallicity or other parameters have no significant effect on the estimated masses ( e.g. , ; ) .
considerable progress has been made in refining current characterization of galaxies from the observations .
even so , the measurements of the specific star formation rate ( ssfr , the star formation rate divided by the stellar mass ) have presented a puzzle to the theoretical understanding of the build - up of mass in galaxies .
( e.g. , ; ) .
several past studies had indicated that the ssfr of sources with fixed stellar mass shows no evidence for significant evolution between @xmath1 @xmath20 2 and @xmath1 @xmath20 7 ( ; ) .
this result was in apparent disagreement with semi - analytic models predicting a strong increase in the specific inflow rate ( i.e. inflow rate divided by halo mass ) of baryons with redshift ( ) .
subsequent work has strongly suggested that the ssfr of galaxies is somewhat higher at @xmath10 than it is at @xmath14 ( e.g. , schaerer & de barros 2010 ; stark et al . 2013 ; gonzalez et al .
2014 ; salmon et al . 2015 ) .
the change in the inferred ssfr evolution with cosmic time was the result of improved observational constraints and a more sophisticated treatment of those constraints .
one example of this is in a consideration of dust extinction in computing the ssfrs at @xmath10 .
while it was not possible to properly account for the impact of dust extinction in initial work ( e.g. , ) due to large uncertainties on the @xmath21 colors of @xmath10 galaxies , later work ( ( ) was able to make use of new measurements of the @xmath21 continuum slopes to account for the impact of dust extinction , finding @xmath22 larger ssfrs at @xmath23 .
this suggested a @xmath24 evolution relative to the @xmath14 value ( ) , but still leaving the ssfr at @xmath1 @xmath25 4 approximately constant .
even more important has been the increasing awareness of the impact of rest - frame optical nebular emission lines ( e.g. , h@xmath5 , [ oiii ] , [ oii ] ) on the broadband fluxes ( e.g. , , , ) . at high redshifts these emission lines
are shifted into the infrared , contaminating the irac measurements of the stellar continuum .
inferred stellar masses from fitting of stellar population models will then be overestimated , resulting in an underestimate of the ssfr .
since these strong rest - frame optical lines are inaccessible to spectroscopy beyond @xmath14 - 3 , the strength of the nebular emission lines has been estimated from the contamination of the @xmath0/irac 3.6@xmath4 m and 4.5@xmath4 m bands for galaxies at @xmath1 @xmath25 3 .
( ) show that galaxies in the range 3.8 @xmath15 @xmath1 @xmath15 5.0 are considerably brighter at 3.6 @xmath4 m than expected from the stellar continuum alone and argue that this excess is due to strong h@xmath5 line emission .
( ) derive an h@xmath5 equivalent width ( ew ) distribution by comparing the [ 3.6 ] @xmath3 [ 4.5 ] color of spectroscopically confirmed galaxies in the redshift range 3.8 @xmath15 @xmath1 @xmath15 5.0 , where the h@xmath5 line lies in the 3.6 @xmath4 m band , with an uncontaminated control sample at 3.1 @xmath15 @xmath1 @xmath15 3.6 .
the results indicate a possible trend towards higher h@xmath5 ews at higher redshifts , which is extremely important for estimating the ssfr at @xmath1 @xmath25 5 ( see also ; ; ; ) .
independent evidence for high - ew nebular lines having a large impact on the broadband fluxes of @xmath26 galaxies was obtained from early near - infrared , multi - object spectroscopic campaigns ( e.g. , ; ) .
the highest - redshift window providing us a largely clean measurement of the h@xmath5 ew is the redshift range @xmath27 - 5.40 . here
the flux excess due to the redshifted nebular emission lines gives rise to significantly redder [ 3.6 ] @xmath3 [ 4.5 ] colors over this range , making it possible to quantify the ew of h@xmath5 at @xmath28 in a similar way to that possible in the redshift range @xmath29 to 5.0 . by examining the [ 3.6 ] @xmath3 [ 4.5 ] color of galaxies from the ( ) catalog ( see 3.1 ) over the great observatories
origins deep survey ( goods)-north and south fields ( ) in this highest redshift window @xmath1 @xmath2 5.1 - 5.4 where h@xmath5 can be cleanly measured , , constraints on the @xmath30 ews are also possible from spitzer / irac observations , but would need to rely on the stacking the fluxes of @xmath31 galaxies in the much less sensitive 5.8 @xmath4 m band . ]
we can derive approximate constraints on the h@xmath5 flux and ew at the highest redshift currently accessible at reasonable s / n with current facilities .
this allows us to obtain better constraints on the evolution of the mean h@xmath5 ew and ssfr as a function of redshift .
this paper is structured as follows . in 2 , we describe the observational data sets and our photometric selection of sources in the narrow redshift range . in 3 , we examine our selected sample of galaxies .
we describe the assumptions made in deriving the ews and ssfrs .
finally , in 4 , we discuss our results and give a summary .
we refer to the hst f435w , f606w , f775w , f814w , f850lp , f105w , f125w , and f160w bands as @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 and @xmath39 , respectively . for consistency with previous work
, we adopt the concordance model with @xmath40 = 0.3 , @xmath41 = 0.7 and @xmath42 = 70 km s@xmath12 mpc@xmath12 . throughout
we assume a ( ) initial mass function ( imf ) between 0.1@xmath3100 @xmath43 .
all magnitudes are quoted in the ab photometric system ( ) .
in order to select sources in the redshift range @xmath1 @xmath2 5.1 - 5.4 , we make use of the deep optical / acs and near - infrared / wfc3/ir observations over the goods - north and goods - south fields from three significant @xmath17 @xmath18 @xmath44 ( hst ) programs : goods , ers ( early release science : ) , and candels ( cosmic assembly near - infrared deep extragalactic legacy survey : ; ) .
the moderately deep regions over the candels goods - north and south reach a 5@xmath16 depth of @xmath2 27.5 mag in the @xmath37 , @xmath38 and @xmath39 filters with the hst and cover @xmath2 100 arcmin@xmath45 .
the deep regions over the candels goods - north and south reach a 5@xmath16 depth of @xmath2 28.5 in the @xmath37 , @xmath38 and @xmath39 bands , covering @xmath2 125 arcmin@xmath45 ( ) .
hst observations with the acs ( advanced camera for surveys ) are available in the @xmath32 , @xmath33 , @xmath34 , @xmath35 and @xmath36 bands , up to @xmath2 29 mag at 5@xmath16 in @xmath35 ( ) . over the northern @xmath240 arcmin@xmath46 section of goods south ( ) ,
deep near - ir observations are available ( @xmath228 mag at @xmath47 ) in the @xmath48 , @xmath38 , and @xmath39 bands and also in the @xmath32 , @xmath33 , @xmath34 and @xmath36 bands with acs .
the observations are psf matched to the @xmath39 band before measuring the colors in scalable ( ) apertures .
essential to the analysis we perform is the 3.6 @xmath4 m and 4.5 @xmath4 m irac observations from the @xmath0 @xmath18 @xmath44 . @xmath0/irac
data is from the original goods program , the spitzer extented deep survey ( seds : ) , the spitzer very deep survey ( s - candels : ) exploration science project , the iudf10 program ( ) , and other programs ( such as pid10076 , pi : oesch ) .
the spitzer / irac reductions we utilize were generated by and feature a 1.8@xmath49-diameter fwhm for the psf .
deblending neighboring galaxies in the irac observations and psf correctons are performed using the mophongo software ( , , , ) .
hst f160w images are used as a high resolution prior to construct a model for the contaminating sources , while leaving the normalization of the sources as a free parameter .
the fluxes of all sources in a radius of 13 are then simultaneously fit to best match the irac image .
photometry is then performed within a 2.0 diameter aperture .
the deep irac imaging from s - candels reaches @xmath226.8 mag at 5@xmath16 in the 3.6 @xmath4 m band . for @xmath50 - 5.4
sources not included in the photometric catalogs of bouwens et al .
( 2015 : due to their being located in areas of the goods fields without @xmath32 or @xmath48/@xmath37-band observations ) , we made use of the hst photometry from the 3d - hst goods - north or goods - south catalogs ( ) . this is relevant for 5 @xmath1 = 5.1 - 5.4 galaxies in our final sample .
we refer to ( ) for a detailed description .
nebular emission line contamination in the 3.6 @xmath4 m and 4.5 @xmath4 m photometric filters .
@xmath51 @xmath52 the redshift ranges over which the dominant nebular emission lines , h@xmath5 , h@xmath53 , [ oiii ] , [ nii ] , and [ sii ] , contribute to the 3.6 @xmath4 m and 4.5 @xmath4 m flux measurements .
@xmath54 @xmath52 the predicted [ 3.6 ] @xmath3 [ 4.5 ] color due to various nebular emission lines as a function of redshift .
the solid black circles indicate sources that are selected in the redshift range @xmath6 = 5.1 - 5.4 , where h@xmath5 lies in the 4.5 @xmath4 m filter , while 3.6 @xmath4 m is devoid of strong nebular emission .
their observed [ 3.6 ] @xmath3 [ 4.5 ] colors are primarily very red .
the open circles are the colors for the spectroscopic sample found in ( ) , which we use to estimate a stellar continuum color of @xmath2 0.00 mag .
the dotted line indicates the expected color for a strong evolution of the rest - frame equivalent width , ew@xmath55(h@xmath5 ) @xmath56 ( 1+@xmath1)@xmath57 , according to ( ) .
our selection of sources in the redshift range @xmath50 - 5.4 has a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.31 @xmath8 0.05 mag and 0.35 @xmath8 0.07 mag for our photometric - redshift and spectroscopic samples , respectively , implying a mean ew(h@xmath5+[nii]+[sii ] ) of 665 @xmath8 53 and 707 @xmath8 74 , respectively , for sources in these samples ( and 638 @xmath8 118 and 855 @xmath8 179 , respectively , using direct sed fits ) .
four selected sources show even far redder colors than predicted with the strong evolution model .
the four reddest sources have a mean color of 0.66 @xmath8 0.06 mag , with a notably high ew of 1743 @xmath8 221 . ]
the lyman - break selection criteria applied for sources at @xmath1 @xmath2 5 are as follows : @xmath58 where @xmath59 denotes the and symbol . for non - detections , the 1@xmath16 upper limit is taken as the flux in the dropout band .
the aforementioned criteria enable to select sources in the range @xmath60 @xmath2 4.5 @xmath3 5.5 .
therefore , a high - redshift boundary is set by excluding sources which satisfy the selection criteria for @xmath1 @xmath2 6 selection ( e.g. , bouwens et al .
2015a ) . contamination from sources at lower redshifts is reduced by requiring that the @xmath1 @xmath2 5 sources have a non - detection ( @xmath15 2@xmath16 ) in the @xmath32 band .
furthermore , we exclude point sources by requiring the sextractor stellarity index to be less that 0.9 ( where 0 and 1 correspond to extended and point sources , respectively ) . utilizing these selection criteria results in an initial @xmath61-drop sample of 1567 sources ( ) . -[4.5]$ ] color distribution for sources in our @xmath50 - 5.4 primary selection ( _ red histogram _ ) and @xmath62 - 5.0 control sample ( _ blue histogram _ : appendix a ) .
sources where h@xmath5+[nii ] is in the [ 3.6 ] band uniformly have moderately blue colors , whereas sources with h@xmath5+[nii ] in the [ 4.5 ] band uniformly have moderately red colors .
the impact of the h@xmath5+[nii ] emission lines on the irac colors is quite clear . deriving the h@xmath5+[nii]+[sii ]
ew by contrasting the observed colors of sources in the two samples should produce a very robust result . ] in order to be able to measure the h@xmath5 ew for @xmath63 galaxies , it is easiest to only make use of galaxies in the narrow redshift window @xmath64 - 5.40 .
we therefore use photometric redshifts to identify a sub - sample of galaxies in this window .
the photometric redshifts for our sample are determined using the eazy photometric redshift code ( ) , which compares photometric data with synthetic photometry of galaxies for various template spectra and redshift ranges .
the best - fit redshift is then derived from a statistical analysis of the differences between both data sets .
the aforementioned 8 hst bands ( @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 ) are used to derive the best - fit photmetric redshifts .
the irac photometry is excluded from this fitting to avoid introducing any bias in the measured [ 3.6 ] @xmath3 [ 4.5 ] color .
this reduces the sample size to 393 sources .
figure [ sed ] shows an example output of eazy for a source at @xmath1 @xmath2 5.2 and a comparison of our estimated eazy photometric redshifts for a sample of @xmath1 = 4 @xmath3 6 spectroscopically confirmed sources from stark et al .
( 2010 : see d. stark et al .
2015 , in prep ) and vanzella et al .
the scatter around the one - to - one relation is @xmath65 = 0.036 , which provides confidence that we can select sources in the narrow redshift range @xmath1 = 5.1 - 5.4 .
we select very bright sources with the requirements that s / n(@xmath39 ) @xmath25 7 @xmath59 s(@xmath39)/n(3.6 @xmath4 m ) @xmath25 3 @xmath59 s(@xmath39)/n(4.5 @xmath4 m ) @xmath25 3 , where s / n is the signal to noise ratio .
our selection of sources based on their measured flux in the @xmath39 band and measured noise in the spitzer / irac bands allows us to include sources in our analysis which we would expect to show up prominently in the spitzer / irac bands .
basing the selection on the measured flux in the spitzer / irac bands would bias our measurement of the colors .
we also discard 41 sources for which the contamination by nearby objects is higher than 200 njy or sources with a poor irac deblending ( @xmath66 parameter less than 0.2 ) , reducing the sample to 101 bright sources .
finally , we require that the eazy redshift probability distribution prefers a redshift in the range @xmath50 - 5.4 at @xmath67% , providing a sample with 11 sources in the redshift range @xmath1 @xmath2 5.1 - 5.4 presented in table [ table:1 ] .
in addition to making use of sources very likely to lie in the redshift range @xmath1 = 5.1 - 5.4 using our photometry for the sources and the redshift likelihood distributions we derive , we can also make use of sources known to lie in the redshift range @xmath1 = 5.1 - 5.4 from available spectroscopy ( d. stark et al .
2015 , in prep ) . using spectroscopic redshifts
, we can be even more certain that the sources we are using lie in the narrow redshift range @xmath1 = 5.10 - 5.40 required for our desired measurement of the h@xmath5 flux .
one potential drawback to the inclusion of such sources in the present study is that we might be working with a biased sample , given that essentially all of the spectroscopic redshift measurements we utilize come from ly@xmath5 , and it is not clear a priori that the study of such a sample might bias the mean h@xmath5 ew we measure to higher values .
fortunately , as we show in a separate study ( smit et al .
2015 , in preparation ) , the mean h@xmath5+[nii]+[sii ] ew measured for both photometric - redshift and spectroscopic samples are essentially identical .
cross - correlating the source catalogs of ( ) and ( ) with the spectroscopic catalog of d. stark et al .
( 2015 , in prep ) and ( ) , we identified 13 @xmath1 = 5.10 @xmath3 5.40 galaxies that we can use for our study , with one source overlapping with our photometric selection ( see table [ table:2 ] ) .
as in our photometric redshift selection ( 2.2 ) , we exclude sources where flux from neighboring sources significantly contaminate the photometric apertures for our @xmath64 - 5.40 sample ( 3 sources ) . 10 of the sources in the desired redshift range were from d. stark et al . ( 2015 , in prep ) redshift compilation , while 3 came from the vanzella et al . (
2009 ) compilation .
we will make use of sources from both our high - quality photometric redshift sample and spectroscopic sample for the analyses that follow .
l|c|c|c|c|c|c|c|c & & & & [ 3.6 ] @xmath3 [ 4.5 ] & h@xmath5+[nii]+[sii ] & ssfr & log@xmath68@xmath69 & + i d & ra & dec & @xmath6 & ( mag ) & ew [ @xmath70@xmath71 & [ gyr@xmath12 ] & [ @xmath69@xmath73 & @xmath74 + gndv-7133823953 & 12:37:13.38 & 62:12:39.5 & 5.2 @xmath8 0.1 & 0.5 @xmath8 0.3 & 672 @xmath8 505&30@xmath75&8.84@xmath76 & @xmath321.3 @xmath8 0.1 + gndv-7128013231 & 12:37:12.80 & 62:11:32.3 & 5.3 @xmath8 0.2 & 0.1 @xmath8 0.2 & 190 @xmath8 254 & 17@xmath77&9.24@xmath78 & @xmath320.8 @xmath8 0.1 + gndv-7033233179@xmath79 & 12:37:03.32 & 62:13:31.8 & 5.3 @xmath8 0.1 & 0.4 @xmath8 0.2&719 @xmath8 368&24@xmath80&8.93@xmath81&@xmath320.8 @xmath8 0.1 + gndv-6302234526 & 12:36:30.22 & 62:13:45.3 & 5.2 @xmath8 0.1 & 0.3 @xmath8 0.1&385 @xmath8 170 & 11@xmath82&9.38@xmath83 & @xmath320.6 @xmath8 0.1 + gndv-6285841077 & 12:36:28.58 & 62:14:10.8 & 5.3 @xmath8 0.1 & 0.2 @xmath8 0.2&557 @xmath8 449&21@xmath84&8.76@xmath85 & @xmath320.7 @xmath8 0.1 + gnwv-6514085687 & 12:36:51.40 & 62:08:56.9 & 5.3 @xmath8 0.1 & 0.0 @xmath8 0.1&25 @xmath8 167&1@xmath86&10.12@xmath87 & @xmath320.6 @xmath8 0.1 + gnwv-6121502518 & 12:36:12.15 & 62:10:25.2 & 5.3 @xmath8 0.1 & 0.4 @xmath8 0.3&917 @xmath8 573 & 6@xmath88&9.29@xmath89 & @xmath320.8 @xmath8 0.1 + gnwv-6095211615 & 12:36:09.52 & 62:11:16.2 & 5.2 @xmath8 0.1 & 0.4 @xmath8
0.1&913 @xmath8 274 & 19@xmath90&9.24@xmath91 & @xmath321.6 @xmath8 0.1 + gndv-3756634257 & 12:37:05.66 & 62:13:42.6 & 5.3 @xmath8 0.1 & 0.1 @xmath8 0.1&274 @xmath8 180 & 26@xmath92&9.55@xmath93 & @xmath321.2 @xmath8 0.1
+ gndv-6325033158 & 12:36:32.50 & 62:13:31.6 & 5.3 @xmath8 0.1 & 0.4 @xmath8 0.1&882 @xmath8 262 & 9@xmath94&8.79@xmath95 & @xmath320.9 @xmath8 0.1 + gsdv-2332672480 & 03:32:33.26 & @xmath327:47:24.8 & 5.3 @xmath8 0.1 & 0.6 @xmath8 0.1&1480 @xmath8 110&1@xmath86&10.06@xmath96 & @xmath320.6 @xmath8 0.1 + [ table:1 ] l c c c c c c c c c & & & & & [ 3.6 ] @xmath3 [ 4.5 ] & h@xmath5+[nii]+[sii ] & ssfr & log@xmath68@xmath69 & + i d & ra & dec & @xmath97 & @xmath6 & ( mag ) & ew [ @xmath70@xmath98 & [ gyr@xmath12 ] & [ @xmath69@xmath73 & @xmath99 + gndv-6554953313 & 12:36:55.49 & 62:15:33.1 & 5.19 & 4.9 @xmath8 0.1 & 0.6 @xmath8 0.2 & 1579 @xmath8 359 & 10@xmath100 & 9.09@xmath101 & @xmath321.3 @xmath8 0.1 + gndv-7033233179 & 12:37:03.32 & 62:13:31.8 & 5.21 & 5.3 @xmath8 0.1 & 0.4 @xmath8 0.2 & 719 @xmath8 368 & 24@xmath80 & 8.93@xmath81 & @xmath320.8 @xmath8 0.1 + gndv-7027322916 & 12:37:02.73 & 62:12:29.2 & 5.23 & 5.5 @xmath8 0.1 & 0.5 @xmath8 0.1 & 1000 @xmath8 257 & 37@xmath102 & 9.21@xmath103 & @xmath320.6 @xmath8 0.1 + gndv-6375223629 & 12:36:37.52 & 62:12:36.3 & 5.18 & 5.4 @xmath8 0.1 & 0.0 @xmath8 0.3 & @xmath15419@xmath104 & 21@xmath105 & 8.62@xmath106 & @xmath320.0 @xmath8 0.1 + gndv-6553954912 & 12:36:55.39 & 62:15:49.1 & 5.19 & 5.1 @xmath8 0.2 & 0.8 @xmath8 0.4 & 2568 @xmath8 1561 & 7@xmath107 & 8.29@xmath108 & @xmath319.8 @xmath8 0.1 + gnwv-7347782930 & 12:37:34.77 & 62:18:29.3 & 5.32 & 5.2 @xmath8 0.2 & 0.6 @xmath8 0.3 & 1344 @xmath8 666 & 3@xmath109 & 8.90@xmath110 & @xmath319.9 @xmath8 0.1 + ersv-2213040511 & 03:32:21.30 & @xmath327:40:51.2&5.29 & 5.3 @xmath8 0.1 & 0.5 @xmath8 0.2 & 1075 @xmath8 394 & 18@xmath111 & 9.11@xmath112 & @xmath320.9 @xmath8 0.1 + gswv-2454254386 & 03:32:45.43 & @xmath327:54:38.6&5.38 & 5.4 @xmath8 0.1 & 0.2 @xmath8 0.1 & 289 @xmath8 75 & 54@xmath113 & 8.96@xmath114 & @xmath321.6 @xmath8 0.1 + gnd6418@xmath79 & 12:36:18.19 & 62:10:21.9 & 5.28 & 5.6 @xmath8 0.1 & 0.3 @xmath8 0.1 & 309 @xmath8 88 & 24@xmath115 & 9.99@xmath116 & @xmath321.6 @xmath8 0.1 + gnd33928 & 12:37:36.87 & 62:18:55.9 & 5.35 & 5.2 @xmath8 0.1 & 0.1 @xmath8 0.2 & @xmath15222@xmath104 & 19@xmath117 & 9.09@xmath118 & @xmath320.4 @xmath8 0.1 + gnd29175 & 12:37:31.45 & 62:17:08.3 & 5.25 & 5.3 @xmath8 0.2 & 0.3 @xmath8 0.4 & 666 @xmath8 929 & 3@xmath119 & 9.28@xmath120 & @xmath319.9 @xmath8 0.1 + gnd12038 & 12:36:26.49 & 62:12:07.4 & 5.20 & 5.2 @xmath8 0.4 & 0.0 @xmath8 0.2 & 254 @xmath8 384 & 10@xmath121 & 9.11@xmath122 & @xmath320.3 @xmath8 0.1 + gs48361@xmath79 & 03:32:16.55 & @xmath327:41:03.2&5.25 & 5.5 @xmath8 0.2 & 0.3 @xmath8 0.2 & 668 @xmath8 370 & 6@xmath123 & 9.32@xmath124 & @xmath320.7 @xmath8 0.1 + + [ table:2 ]
using the selection discussed in 2.2 and 2.3 we have isolated a sample of 21 galaxies that have redshifts in the narrow redshift range @xmath1 @xmath2 5.1 @xmath3 5.4 at high confidence . in this section
we will discuss the irac [ 3.6 ] @xmath3 [ 4.5 ] colors of these galaxies and compare them with the irac colors of a spectroscopically confirmed sample of @xmath1 @xmath2 4.5 galaxies .
our selection of galaxies in the redshift range @xmath1 @xmath2 5.1 @xmath3 5.4 allows us to solve for the rest - frame ew of the nebular emission lines using the [ 3.6 ] @xmath3 [ 4.5 ] color . assuming @xmath126 @xmath127 constant
( i.e. @xmath128 @xmath56 @xmath129 ) we approximate the observed flux @xmath130 in the irac filters , for a given ew , by @xmath131 where @xmath132 is the stellar continuum flux and @xmath133 is the observed wavelength of the nebular emission lines ( ) .
@xmath134 denotes the response curve of the filter .
we use all nebular emission lines tabulated in ( ) and the hydrogen balmer lines for the modeling of the [ 3.6]@xmath3 [ 4.5 ] color .
we fix the line intensities relative to h@xmath53 according to the values in ( ) for sub - solar metallicity 0.2 z@xmath135 and assuming case b recombination . the observed [ 3.6 ] @xmath3 [ 4.5 ] color
can then be modeled by @xmath136-[4.5 ] = ( [ 3.6]-[4.5])_{continuum } - 2.5 \log_{10 } \left ( \frac{x_{3.6}}{x_{4.5 } } \right ) .\ ] ] figure [ fig : nebularlines ] illustrates the contamination by nebular emission lines in the photometric filters as a function of redshift .
the top panel shows the redshift ranges where the dominant emission lines , h@xmath5 , h@xmath53 , [ oiii]@xmath1374959 , 5007 , [ nii ] , and [ sii ] fall in the irac 3.6 @xmath4 m and 4.5 @xmath4 m passbands .
the lower panel shows the expected [ 3.6 ] @xmath3 [ 4.5 ] color in the presence of nebular line emission .
the dotted line illustrates the case when the rest frame ew evolves strongly with redshift .
consistent with ( ) who find that ew@xmath55(h@xmath5 ) @xmath56 ( 1+@xmath1)@xmath57 for star forming galaxies at 0 @xmath138 @xmath1 @xmath138 2 .
galaxies are expected to become quite red in the redshift range @xmath1 = 5.1 @xmath3 5.4 , due to contamination of the 4.5 @xmath4 m flux by the h@xmath5 line and no strong nebular emission line contamination in the 3.6 @xmath4 m band .
the selected galaxies in the redshift range @xmath1 @xmath2 5.1 @xmath3 5.4 allow us to deduce the ew of the nebular emission lines in @xmath1 @xmath139 5 objects from the fluxes in the 3.6 @xmath4 m and 4.5 @xmath4 m bands .
we observe a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.31 @xmath8 0.05 mag for the selected sources in our photometric - redshift sample and a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.35 @xmath8 0.07 mag for sources in our spectroscopic sample . in both cases
, we estimate the uncertainty here through bootstrap resampling .
we estimate the [ 3.6 ] @xmath3 [ 4.5 ] continuum color by contrasting the mean [ 3.6 ] @xmath3 [ 4.5 ] colors observed for @xmath62 - 5.0 spectroscopically - confirmed sources from ( ) , ( ) , and ( ) , i.e. , @xmath140 mag , with the mean [ 3.6]@xmath3[4.5 ] colors observed for our selected @xmath1 @xmath2 5.1 @xmath3 5.4 sample and attribute any differences in the [ 3.6]@xmath3 [ 4.5 ] colors to the impact of flux from the h@xmath5 + [ nii ] + [ sii ] lines .
we estimate a color for the stellar continuum of [ 3.6 ] @xmath3 [ 4.5 ] of @xmath141 mag .
see figure [ fig : col36 ] for an illustration of the differences between the two samples .
four sources in our selection show considerably redder [ 3.6 ] @xmath3 [ 4.5 ] colors than we can infer from our model sed with strong line emission , i.e. , 0.66 @xmath8 0.06 mag .
the very red [ 3.6 ] @xmath3 [ 4.5 ] colors observed for many sources in our @xmath1 = 5.1 @xmath3 5.4 sample suggests that the irac [ 3.6 ] @xmath3 [ 4.5 ] color may provide significant leverage in terms of identifying galaxies that specifically lie in the narrow redshift range @xmath1 @xmath2 5.1 @xmath3 5.4 .
this method has previously been explored at @xmath142 by ( ) and ( ) based on the impact of [ oiii ] on the irac fluxes at these high redshifts . by comparing the mean [ 3.6 ] @xmath3 [ 4.5 ] color of our @xmath64 - 5.40 sample with the mean [ 3.6 ] @xmath3 [ 4.5 ] color of our @xmath62 - 5.0 sample
, we find an overall color difference of [email protected] mag relative to our @xmath50 - 5.4 photometric sample and [email protected] mag relative to our @xmath50 - 5.4 spectroscopic sample .
comparing the color difference we observe with that predicted based on simple model spectra with an h@xmath5 ew of 595 @xmath144 and line ratios set by the ( ) model ( _ dotted line _ in figure 2 ) , we infer an approximate h@xmath5 ew of 557 @xmath8 44 @xmath145 for our @xmath50 - 5.4 photometric sample and 592 @xmath8 62 @xmath145 for our @xmath50 - 5.4 spectroscopic sample .
given that flux in the [ nii ] and [ sii ] lines would add to the color difference observed between our @xmath62 - 5.0 and @xmath50 - 5.4 samples and can not be determined separately , it is best to quote a constraint on the mean ew of h@xmath5+[nii]+[sii ] .
( ) predict a contribution of 6.8% from [ nii ] and 9.5% from [ sii ] .
this is in good agreement with an observed ratio of [ nii]/h@xmath5 of 0.05 - 0.09 in @xmath146 galaxies with stellar masses in the range @xmath147 by ( ) . if we correct for this , the mean ew in h@xmath5+[nii]+[sii ] is 665 @xmath8 53 for our photometric sample and 707 @xmath8 74 for our spectroscopic sample .
we can also derive h@xmath5+[nii]+[sii ] ews for individual sources in our photometric and spectroscopic @xmath50 - 5.4 samples .
in computing the ews for individual sources , we use fast to fit the observed seds of individual sources excluding the 4.5 @xmath148 band which is contaminated by h@xmath5 emission .
then , by comparing the observed 4.5 @xmath148 flux with the expected 4.5 @xmath148 flux ( without including emission lines in the fast modeling ) , we derive ews for individual sources .
the results are presented in tables 1 and 2 .
the mean h@xmath5+[nii]+[sii ] ew we derive for our photometric sample is 638 @xmath8 118 , while we find 855 @xmath8 179 for our spectroscopic sample .
if we follow the model results from ( ) and suppose that 16.3% of the 4.5 @xmath148 excess derives from [ nii ] and [ sii ] , the excesses we derive suggest h@xmath5 ews of 534 @xmath8 99 and 715 @xmath8 150 , respectively .
the four reddest sources have a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.66 @xmath8 0.06 mag , consistent with an ew@xmath55(h@xmath5+[nii]+[sii ] ) of 1743 @xmath8 221 , equivalent to ew@xmath55(h@xmath5 ) @xmath149 1458 @xmath8 185 for the above line ratios .
figure [ fig : ew ] shows several values for the h@xmath5 + [ nii ] + [ sii ] ews with redshift from the literature .
the black line gives the evolution of the h@xmath5 + [ nii ] + [ sii ] derived by ( ) for galaxies with masses @xmath69 @xmath2 10@xmath150 @xmath3 10@xmath151 @xmath43 , which we extrapolate to higher redshifts and lower masses .
keeping in mind that the ew(h@xmath5 + [ nii ] + [ sii ] ) as function of the redshift is higher for sources with lower steller masses our result is consistent with the extrapolation from ( ) and the high ews derived by ( ) , ( ) and ( ) .
uncertainties in the photometric redshifts for our @xmath64 - 5.40 sample can lead to a systematic underestimate of the h@xmath5+[nii]+[sii ] flux , if it causes us to include sources which lie outside the desired range . for @xmath152 sources , the 3.6 @xmath4 m band will be contaminated by h@xmath5 + [ nii ] + [ sii ] emission . meanwhile , for @xmath153 sources , h@xmath53 + [ oiii ] emission will contribute to the @xmath154 m band .
in both cases , the [ 3.6 ] @xmath3 [ 4.5 ] color will be much bluer , causing us to infer a substantially lower ew for that source , than is truly present .
a clean measurement of the stellar continuum emission is essential for deriving the ssfr . in our target redshift range
the flux in the 3.6 @xmath4 m band can be used for this purpose , while the 4.5 @xmath4 m band is contaminated by emission lines and should be left out in stellar population modeling .
we derive the mean ssfr for our selected sources using similar method as described in ( ) . we derive stellar masses using the modeling code fast ( ) , which fits stellar population synthesis templates from ( ) to broadband photometry . for consistency , we assume a ( ) imf with 0.1@xmath3100 @xmath43 , a sub - solar metallicity @xmath155 = 0.2 @xmath156 and dust attenuation from ( ) . the ages range from 10 myr to the age of the universe at the redshift of the source . the star formation history is assumed to be constant and the dust content is varied between a@xmath157 = 0 @xmath3 3 mag . in order to avoid the degeneracy between the ages of the galaxies and their dust content in the fast modeling of the ssfr , we derive star formation rates directly from the uv continuum , using the ( ) relation and fixing the dust extinction using the relationship from ( ) between a@xmath157 and the uv - continuum slope @xmath53 . from this method ,
we derive a median ssfr of @xmath11 gyr@xmath12 ( individual ssfr estimates are listed in table [ table:1 ] and table [ table:2 ] ) .
we also explore the use of a smc dust - law , which might seem more appropriate for lyman - break galaxies at @xmath158 based on recent alma results by ( ) .
we fit fast models as above with a ( ) dust law with parameters @xmath159 and @xmath160 and find a ssfr of @xmath161 gyr@xmath12 .
it is useful to compare the median ssfr estimates we find here , @xmath11 gyr@xmath12 and 13 gyr@xmath12 , with previous estimates .
this represents a @xmath162 @xmath163 increase in ssfr compared to the 2.4 gyr@xmath12 value found at @xmath14 ( ) , supporting a significant evolution in the ssfr .
the typical stellar mass for galaxies in our @xmath50 - 5.4 selection is @xmath210@xmath164 @xmath43 , so we will make our comparison with previous measures at this stellar mass .
both stark et al . ( 2013 ) and gonzalez et al .
( 2014 ) find that the typical galaxy with this stellar mass has a @xmath21 luminosity @xmath99 of @xmath320 mag .
accounting for a factor of @xmath22 mean dust attenuation at this luminosity ( as implied by the bouwens et al .
2014 @xmath165 results ) , the equivalent ssfr is @xmath211 gyr@xmath12 . somewhat similarly , salmon et al .
( 2015 ) derive a ssfr of @xmath28@xmath166 gyr@xmath12 , which is again somewhat lower than we find here .
all things being equal , we would expect the ssfr estimates we derive here to be more accurate than previous estimates , given our precise knowledge of the redshifts and hence the position of nebular emission lines within galaxies
. the impact of the lines on the mass and ssfr estimates could be as large as a factor of @xmath21.5 , allowing for an approximate reconciliation of the present ssfr estimates with that from previous work .
however , the present sample of @xmath167 - 5.4 galaxies is still quite small , and therefore expansion of the present sample would certainly be helpful for improving our ssfr estimate . by fitting to the photometry of all passbands uncontaminated by the strong h@xmath5+[nii]+[sii ] nebular emission lines , we can estimate stellar masses for sources in our @xmath50 - 5.4 samples , as described in the previous section .
as these sources are distributed over a wide range of stellar mass , i.e. , @xmath168 @xmath43 to @xmath169 @xmath43 , we can go beyond a simple determination of the mean h@xmath5+[nii]+[sii ] ew for @xmath143 galaxies and look at whether there is a dependence on stellar mass .
any significant dependence on stellar mass would be noteworthy , as it could point to a significant mass or scale dependence to the star formation histories of galaxies .
while such a scale dependence could be expected if there are significant feedback effects at early times ( e.g. , bowler et al .
2014 ; bouwens et al .
2015 ) , many simulations ( e.g. , finlator et al .
2011 ) predict that galaxies build up their stellar mass in a relatively self - similar manner , independent on the overall stellar mass .
we examine the evidence for such a correlation between the h@xmath5+[nii]+[sii ] ew and the stellar mass in figure [ fig : ewmass ] , plotting the h@xmath5+[nii]+[sii ] ews we estimate for individual sources against the stellar masses we estimate for the same sources . also shown on this figure
is the average h@xmath5 ew we measure for galaxies with estimated stellar masses @xmath1510@xmath170 @xmath43
( 773 @xmath8 136 ) and for those with estimated stellar masses @xmath2510@xmath170 @xmath43 ( 522 @xmath8 279 ) . at face value , these results do not provide any strong evidence ( @xmath171 ) for a correlation between the h@xmath5 ew and the stellar mass .
while it is possible that a slight correlation might be expected ( i.e. , since massive galaxies would be the first to experience a slowing in their growth rate due to feedback - type effects ) , no strong trend is evident .
a similar conclusion can be drawn by fitting h@xmath5+[nii]+[sii ] ews vs. stellar mass relation to a straight line .
in this paper , we derive the h@xmath5+[nii]+[sii ] ew and the ssfr by selecting galaxies at @xmath1 @xmath2 5.1 - 5.4 . in doing so , we make use of the highest redshift window allowing for a clean measurement of both the h@xmath5 flux and the stellar continuum .
reliable estimates of the stellar mass and ssfr require a clean measurement of the stellar continuum . at @xmath172 - 6.6 ,
rest - frame optical flux information on galaxies as derived from the spitzer / irac data can be quite ambiguous to interpret , due to a significant contribution from nebular emission lines ( h@xmath5 , h@xmath53 , [ oiii ] , [ nii ] , [ sii ] ) in both sensitive spitzer / irac channels ( i.e. , [ 3.6 ] and [ 4.5 ] ) , and at @xmath173 , the h@xmath5 line redshifts out of the [ 4.5 ] channel .
observing galaxies in the @xmath50 - 5.4 redshift interval is useful as a contrast and for interpreting the observations of galaxies in the redshift interval @xmath29 - 5.0 . in both intervals , the spitzer /
irac fluxes are expected to be dominated by a stellar continuum contribution and a contribution from h@xmath5 .
however , the h@xmath5 line will contribute to the measured flux in a different spitzer / irac band at @xmath50 - 5.4 ( [ 4.5 ] ) than at @xmath29 - 5.0 ( [ 3.6 ] ) , so the colors of galaxies in the two samples can be contrasted and used to set strong constraints on the rest - frame ew of h@xmath5 . in our utilization of the @xmath50 - 5.4 redshift interval to study the h@xmath5 ew and ssfr , we have selected 11 bright sources over the candles goods - north and goods - south fields satisfying the criteria s(@xmath39)/n(3.6 @xmath4 m ) @xmath25 3 , s(@xmath39)/n(4.5 @xmath4 m ) @xmath25 3 , ( @xmath174 @xmath25 1.2 ) , ( @xmath175 @xmath15 1.3 ) , ( @xmath174 @xmath25 0.8 ( @xmath175 ) + 1.2 ) , and s / n(@xmath32 ) @xmath15 2 .
the candidates are required to have @xmath25 85% probability of lying in the redshift range @xmath1 @xmath2 5.1 @xmath3 5.4 .
we have supplemented this sample with 13 @xmath1 = 5.10 @xmath3 5.40 sources from the spectroscopic redshift sample of vanzella et al .
( 2009 ) and d. stark et al .
( 2015 , in prep ) .
we find a mean rest - frame ew(h@xmath5+[nii]+[sii ] ) of 665 @xmath8 53 for our photometric sample and 707 @xmath8 74 for our spectroscopic sample based on the mean [ 3.6 ] @xmath3 [ 4.5 ] colors of these samples . assuming that 84% of the h@xmath5+[nii]+[sii ] line flux is h@xmath5
, we further derive a h@xmath5 ew of @xmath2 557@xmath844 and 592 @xmath862 for our photometric and spectroscopic sample , respectively .
our estimate is consistent with the ( 1+@xmath1)@xmath57 power law derived for a strong line - emitter model by ( ) .
four sources have an even higher ew than predicted by the strong line - emitter model in ( ) .
these reddest sources in our selection have a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.66 @xmath8 0.06 mag corresponding to an average ew of 1743 @xmath8 221 .
our selection at @xmath1 @xmath2 5 has a median ssfr of @xmath2 17@xmath176 gyr@xmath12 .
this represents a @xmath162 @xmath163 increase in ssfr compared to the 2.4 gyr@xmath12 value found at @xmath14 ( ) , supporting a significant evolution in the ssfr .
our estimate is in agreement with the theoretical model of ( ) , matching the increasing specific inflow rate of baryonic particles .
we emphasize the h@xmath5+[nii]+[sii ] ews and ssfr we derive here for our photometric sample is effectively a lower limit on the true value , as the inclusion of any sources in our photometric selection outside the desired redshift range due to photometric redshift uncertainties would result in a bluer mean [ 3.6 ] @xmath3 [ 4.5 ] color and higher redshift optical flux for the average source .
we also take advantage of the stellar population modeling we do of individual sources in our sample and the range of estimated masses to look at a possible correlation between the h@xmath5+[nii]+[sii ] ew of sources in our selection and their stellar masses .
we find no strong evidence ( @xmath152@xmath16 ) for there being a correlation between the ew of h@xmath5+[nii]+[sii ] ews and the stellar mass .
however , we caution that our sample sizes and dynamic range are limited , and so a correlation may be evident when examining large sample sizes or a wider dynamic range . more accurate results would require spectroscopic redshifts for a larger number of sources .
though our target redshift range allows for a relatively clean measurement , constructing large samples of @xmath177 5.1 @xmath3 5.4 galaxies is challenging due to the narrow redshift window we are considering .
we expect significant progress in the future as a result of future samples with the muse spectrograph ( bacon et al .
furthermore , ultra - deep spitzer / irac data over the goods - n and goods - s fields will become available through the goods re - ionization era wide - area treasury from spitzer ( greats , pi : labb ) program ( 2014 ) .
we acknowledge the support of nasa grant nag5 - 7697 , nasa / stsci grant hst - go-11563 , and a nwo vrij - competitie grant 600.065.140.11n211 .
rs acknowledges the support of the leverhulme trust .
this research has made use of the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , un- der contract with the national aeronautics and space administration .
in order to estimate the mean stellar continuum color @xmath179-[4.5]$ ] for @xmath143 ( 3.1 ) , we make use of a spectroscopic selection of sources at @xmath62 - 5.0 from vanzella et al .
( 2009 ) , shim et al .
( 2011 ) , and stark et al .
we obtain a sample of 30 @xmath180 - 5.0 sources by cross - correlating the source catalogs of ( ) and ( ) with the spectroscopic catalog of shim et al .
( 2011 ) , stark et al .
( 2013 ) and ( ) .
as in section [ speczsample ] we exclude any sources which are reported to have detected x - ray counterparts or agn emission lines by ( ) , ( ) and d. stark et al . ( 2013 ) .
the excluded sources including those with the following coordinates ( 03:32:29.29 , @xmath327:56:19.5 ; 03:32:44.11 , @xmath327:54:52.5 ; 12:36:42.05 , 62:13:31.7 ; 03:32:33.77 , @xmath327:52:23.7 ) . we tabulate the measured @xmath179-[4.5]$ ] colors and coordinates of the sources we utilize in table [ table:3 ] . | recent @xmath0/infrared array camera ( irac ) photometric observations have revealed that rest - frame optical emission lines contribute signficantly to the broadband fluxes of high - redshift galaxies .
specifically , in the narrow redshift range @xmath1 @xmath2 5.1 - 5.4 the [ 3.6 ] @xmath3 [ 4.5 ] color is expected to be very red , due to contamination of the 4.5 @xmath4 m band by the dominant h@xmath5 line , while the 3.6 @xmath4 m filter is free of nebular emission lines .
we take advantage of new reductions of deep @xmath0/irac imaging over the goods - north+south fields ( ) to obtain a clean measurement of the mean h@xmath5 equivalent width from the [ 3.6 ] @xmath3 [ 4.5 ] color in the redshift range @xmath1 = 5.1 - 5.4 .
the selected sources either have measured spectroscopic redshifts ( 13 sources ) or lie very confidently in the redshift range @xmath1 = 5.1 - 5.4 based on the photometric redshift likelihood intervals ( 11 sources ) .
our @xmath6 = 5.1 - 5.4 sample and @xmath7 = 5.10 - 5.40 spectroscopic sample have a mean [ 3.6 ] @xmath3 [ 4.5 ] color of 0.31 @xmath8 0.05 mag and 0.35 @xmath8 0.07 mag , implying a rest - frame equivalent width ew(h@xmath5+[nii]+[sii ] ) of 665 @xmath8 53 and 707 @xmath8 74 , respectively , for sources in these samples .
these values are consistent albeit slightly higher than derived by stark et al .
2013 at @xmath9 , suggesting an evolution to higher values of the h@xmath5+[nii]+[sii ] ew at @xmath10 . using the 3.6 @xmath4 m band ,
which is free of emission line contamination , we perform robust sed fitting and find a median specific star formation rate of ssfr = @xmath11 gyr@xmath12 , @xmath13 higher than at @xmath14 .
we find no strong correlation ( @xmath152@xmath16 ) between the h@xmath5+[nii]+[sii ] ew and the stellar mass of sources . before the advent of jwst , improvements in these results
will come through an expansion of current spectroscopic samples and deeper @xmath0/irac measurements .
[ firstpage ] galaxies : high - redshift galaxies : formation galaxies : evolution |
inspired from fredholm integral equations , fredholm learning algorithms are designed recently for density ratio estimation @xcite and semi - supervised learning @xcite .
fredholm learning can be considered as a kernel method with data - dependent kernel .
this kernel usually is called as fredholm kernel , and can naturally incorporate the data information . although its empirical performance has been well demonstrated in the previous works , there is no learning theory analysis on generalization bound and learning rate .
it is well known that generalization ability and learning rate are important measures to evaluate the learning algorithm @xcite . in this paper , we focus on this theoretical theme for regularized least square regression with fredholm kernel . in learning theory literature , extensive studies have been established for least square regression with regularized kernel methods , e.g. , @xcite .
although the fredholm learning in @xcite also can be considered as a regularized kernel method , there are two key features : one is that fredholm kernel is associated with the inner " kernel and the outer " kernel simultaneously , the other is that for the prediction function is double data - dependent .
these characteristics induce the additional difficulty on learning theory analysis . to overcome the difficulty of generalization analysis
, we introduce novel stepping - stone functions and establish the decomposition on excess generalization error .
the generalization bound is estimated in terms of the capacity conditions on the hypothesis spaces associated with the inner " kernel and the outer " kernel , respectively .
in particular , the derived result implies that fast learning rate with @xmath1 can be reached with proper parameter selection , where @xmath2 is the number of labeled data .
to best of our knowledge , this is the first discussion on generalization error analysis for learning with fredholm kernel .
the rest of this paper is organized as follows .
regression algorithm with fredholm kernel is introduced in section [ section2 ] and its generalization analysis is presented in section [ section3 ] .
the proofs of main results are listed in section [ section4 ] .
simulated examples are provided in section [ section5 ] and a brief conclusion is summarized in section [ section6 ] .
let @xmath3 be a compact input space and @xmath4 $ ] for some constant @xmath5 .
the labeled data @xmath6 are drawn independently from a distribution @xmath7 on @xmath8 and the unlabeled data @xmath9 are derived random independently according to the marginal distribution @xmath10 on @xmath11 .
given @xmath12 , the main purpose of semi - supervised regression is to find a good approximation of the regression function @xmath13 in learning theory , @xmath14 and its discrete version @xmath15 are called as the expected risk and the empirical risk of function @xmath16 , respectively .
let @xmath17 be a continuous bounded function on @xmath18 with @xmath19 .
define the integral operator @xmath20 as @xmath21 where @xmath22 is the space of square - integrable functions .
let @xmath23 be a reproducing kernel hilbert space ( rkhs ) associated with mercer kernel @xmath24 .
denote @xmath25 as the corresponding norm of @xmath23 and assume the upper bound @xmath26 .
if choose @xmath27 as the hypothesis space , the learning problem can be considered as to solve the fredhom integral equation @xmath28 .
sine the distribution @xmath7 is unknown , we consider the empirical version of @xmath29 associated with @xmath30 , which is defined as @xmath31 in the fredholm learning framework , the prediction function is constructed from the data dependent hypothesis space @xmath32 given @xmath33 , least - square regression with fredholm kernel ( lfk ) can be formulated as the following optimization @xmath34 where @xmath35 is a regularization parameter .
equation ( [ algorithm1 ] ) can be considered as a discrete and regularized version of the fredholm integral equation @xmath36 .
when @xmath37 is the @xmath38-function , ( [ algorithm1 ] ) becomes the regularized least square regression in rkhs @xmath39 when @xmath40 and replacing @xmath41 with @xmath42 , ( [ algorithm1 ] ) is equivalent to the data - dependent coefficient regularization @xmath43 where @xmath44 it is well known that ( [ algorithm2 ] ) and ( [ algorithm3 ] ) have been studied extensively in learning literatures , see , e.g. @xcite .
these results relied on error analysis techniques for data independent hypothesis space @xcite and data dependent hypothesis space @xcite , respectively .
therefore , the fredholm learning provides a novel framework for regression related with the data independent space @xmath23 and the data dependent hypothesis space @xmath45 simultaneously .
equation ( [ algorithm1 ] ) involves the inner " kernel @xmath46 and the outer " kernel @xmath37 .
denote @xmath47 @xmath48 , and @xmath49 .
it has been demonstrated in @xcite that @xmath50 where @xmath51 .
therefore , fredholm regression in ( [ algorithm1 ] ) can be implemented efficiently and the data - dependent kernel @xmath52 is called fredholm kernel in @xcite .
to provide the estimation on the excess risk , we introduce some conditions on the hypothesis space capacity and the approximation ability of fredholm learning framework . for @xmath53 , denote @xmath54 and @xmath55 for any @xmath56 and function space @xmath57 , denote @xmath58 as the covering number with @xmath59-metric .
( capacity condition ) for the inner " kernel @xmath46 and the outer " kernel @xmath37 , there exists positive constants @xmath60 and @xmath61 such that for any @xmath56 , @xmath62 and @xmath63 , where @xmath64 are constants independent of @xmath65 .
[ condition1 ] it is worthy notice that the capacity condition has been well studied in @xcite . in particular
, this condition holds true when setting the inner " and outer " kernels as gaussian kernel . for a function @xmath66 and @xmath67 ,
denote the @xmath68-norm on @xmath69 as @xmath70 define the data independent regularized function @xmath71 the predictor associated with @xmath72 is @xmath73 and the approximation ability of fredholm scheme in @xmath23 is characterized by @xmath74 ( approximation condition ) there exists a constant @xmath75 $ ] such that @xmath76 where @xmath77 is a positive constant independent of @xmath78.[condition2 ] this approximation condition relies on the regularity of @xmath79 , and has been investigated extensively in @xcite . to get tight estimation
, we introduce the projection operator @xmath80 it is a position to present the generalization bound . under assumptions [ condition1 ] and [ condition2 ] , there exists @xmath81 where @xmath82 is a positive constant independent of @xmath83 [ theorem1 ] the generalization bound in theorem [ theorem1 ] depends on the capacity condition , the approximation condition , the regularization parameter @xmath78 , and the number of labeled data . in particular , the labeled data is the key factor on the excess risk without the additional assumption on the marginal distribution .
this observation is consistent with the previous analysis for semi - supervised learning @xcite . to understand the learning rate of fredholm regression , we present the following result where @xmath78 is chosen properly .
under assumptions [ condition1 ] and [ condition2 ] , for any @xmath84 , with confidence @xmath85 , there exists some positive constant @xmath86 such that @xmath87 where @xmath88 [ theorem2 ] theorem [ theorem2 ] tells us that fredholm regression has the learning rate with polynomial decay .
when @xmath89 , there exists some constant @xmath90 such that @xmath91 with confidence @xmath85 , where @xmath92 ; \\ \frac{2\beta}{s+2\beta+s\beta } , & \beta\in(\frac{2}{2+s},+\infty ] .
\\ \end{array } \right.\end{aligned}\ ] ] and the rate is derived by setting @xmath93 ; \\
l^{-\frac{2}{s+2\beta+s\beta } } , & \beta\in(\frac{2}{2+s},+\infty ] .
\\ \end{array } \right.\end{aligned}\ ] ] this learning rate can be arbitrarily close to @xmath94 as @xmath60 tends to zero , which is regarded as the fastest learning rate for regularized regression in the learning theory literature .
this result verifies the lfk in ( [ algorithm1 ] ) inherits the theoretical characteristics of least square regularized regression in rkhs @xcite and in data dependent hypothesis spaces @xcite .
we first present the decomposition on the excess risk @xmath95 , and then establish the upper bounds of different error terms . according to the definitions of @xmath96
, we can get the following error decomposition .
[ proposition1 ] for @xmath97 defined in ( [ algorithm1 ] ) , there holds @xmath98 where @xmath99 and @xmath100 * proof * : by introducing the middle function @xmath101 , we get @xmath102\\ & & + \mathcal e_{\mathbf z}(l_{w,\mathbf{x}}f_{\lambda})-\mathcal e(l_{w,\mathbf{x}}f_{\lambda})+\mathcal e(l_{w,\mathbf{x}}f_{\lambda})-\mathcal e(l_{w}f_{\lambda } ) + \mathcal e(l_{w}f_{\lambda})-\mathcal e(f_\rho)+\lambda\|f_{\lambda}\|_k^2 \\ & \leq & e_1+e_2+e_3+d(\lambda)\end{aligned}\ ] ] where the last inequality follows from the definition @xmath103 . this completes the proof.@xmath104 in learning theory , @xmath105 are called the sample error , which describe the difference between the empirical risk and the expected risk .
@xmath106 is called the hypothesis error which reflects the divergence of expected risks between the data independent function @xmath107 and data dependent function @xmath108 .
we introduce the concentration inequality in @xcite to measure the divergence between the empirical risk and the expected risk .
let @xmath109 be a measurable function set on @xmath110 .
assume that , for any @xmath111 , @xmath112 and @xmath113 for some positive constants @xmath114 .
if for some @xmath115 and @xmath116 , @xmath117 for any @xmath56 , then there exists a constant @xmath118 such that for any @xmath119 , @xmath120 with confidence at least @xmath121
. [ lemma1 ] to estimate @xmath122 , we consider the function set containing @xmath103 for any @xmath123 , @xmath124 .
the definition @xmath103 in ( [ algorithm1 ] ) tells us that @xmath125 .
hence , @xmath126 with @xmath127 and @xmath128 .
[ proposition2 ] under assumption [ condition1 ] , for any @xmath84 , @xmath129 with confidence @xmath85 .
* proof * : for @xmath130 , denote @xmath131 for any @xmath132 , @xmath133 moreover , @xmath134 for any @xmath135 , there exists @xmath136 this relation implies that @xmath137 where the last inequality from assumption [ condition1 ] . applying the above estimates to lemma [ lemma1 ]
, we derive that @xmath138 with confidence @xmath85 . considering @xmath139 with @xmath127 , we obtain the desired result .
@xmath104 [ proposition3 ] under assumption 1 , with confidence @xmath140 , there holds @xmath141 where @xmath142 is a positive constant independent of @xmath143 . * proof * : denote @xmath144 from the definition @xmath145
, we can deduce that @xmath146 with @xmath147 .
for @xmath148 , define @xmath149 it is easy to check that for any @xmath132 @xmath150 then , @xmath151 for any @xmath152 , there exists @xmath153 then from assumption [ condition1 ] , @xmath154 combining ( [ p11])-([p33 ] ) with lemma [ lemma1 ] , we get with confidence @xmath85 @xmath155 considering @xmath156 , we get the desired result.@xmath104 the following concentration inequality with values in hilbert space can be found in @xcite , which is used in our analysis .
let @xmath157 be a hilbert space and @xmath158 be independent random variable on @xmath159 with values in @xmath157 .
assume that @xmath160 almost surely .
let @xmath161 be independent random samples from @xmath7 .
then , for any @xmath162 , @xmath163 holds true with confidence @xmath85 .
now we turn to estimate @xmath106 , which reflects the affect of inputs @xmath30 to the regularization function @xmath145 .
[ proposition ] for any @xmath84 , with confidence @xmath85 , there holds @xmath164 * proof * : note that @xmath165 denote @xmath166 , which is continuous and bounded function on @xmath69 .
then @xmath167 and @xmath168 we can deduce that @xmath169 and @xmath170 . from lemma [ lemma2 ] , for any @xmath84 , there holds with confidence @xmath85 @xmath171 combining ( [ p111 ] ) and ( [ p222 ] ) , we get with confidence @xmath85 , @xmath172 then , the desired result follows from @xmath173 .
@xmath104 * proof of theorem 1 : * combining the estimations in propositions 1 - 4 , we get with confidence @xmath174 , @xmath175 considering @xmath176 , for @xmath84 , we have with confidence @xmath174 @xmath177,\end{aligned}\ ] ] where @xmath82 is a constant independent of @xmath83 .
* proof of theorem 2 : * when setting @xmath178 , we obtain @xmath179 .
then , theorem [ theorem1 ] implies that @xmath180 when setting @xmath181 , we get @xmath182 .
then , with confidence @xmath174 @xmath183 this complete the proof of theorem 2 .
to verify the effectiveness of lfk in ( [ algorithm1 ] ) , we present some simulated examples for the regression problem . the competing method is support vector machine regression ( svm ) , which has been used extensively used in machine learning community ( https://www.csie.ntu.edu.tw/ cjlin / libsvm/ ) .
the gaussian kernel @xmath184 is used for svm .
for lfk in ( [ algorithm1 ] ) , we consider the following `` inner '' and `` outer '' kernels : * lfk1 : @xmath185 and @xmath186 .
* lfk2 : @xmath187 and @xmath188 . * lfk3 : @xmath187 and @xmath186 .
here the scale parameter @xmath189 belongs to @xmath190 $ ] and the regularization parameter belongs to @xmath191 $ ] for lfk and svm .
these parameters are selected by 4-fold cross validation in this section .
the following functions are used to generate the simulated data : @xmath192\\ f_2(x)&=&xcos(x),~~x\in[0,10]\\ f_3(x)&=&\min(2|x|-1,1),~~x\in[-2,2]\\ f_4(x)&=&sign(x),~~x\in[-3,3].\\\end{aligned}\ ] ] note that @xmath193 is highly oscillatory , @xmath194 is smooth , @xmath195 is continuous not smooth , and @xmath196 is not even continuous .
these functions have been used to evaluate regression algorithms in @xcite .
c|ccccc function & number & svm & lfk1 & lfk2 & lfk3 + @xmath193 & 50 & @xmath197 & @xmath198 & @xmath199 & @xmath200 + & 300 & @xmath201 & @xmath202 & @xmath203 & @xmath204 + @xmath194 & 50 & @xmath205 & @xmath206 & @xmath207 & @xmath208 + & 300 & @xmath209 & @xmath210 & @xmath211 & @xmath212 + @xmath195 & 50 & @xmath213 & @xmath214 & @xmath215 & @xmath216 + & 300 & @xmath217 & @xmath218 & @xmath219 & @xmath220 + @xmath196 & 50 & @xmath221 & @xmath222 & @xmath223 & @xmath224 + & 300 & @xmath225 & @xmath226 & @xmath227 & @xmath228 + [ tab1 ] in our experiment , gaussian noise @xmath229 is added to the data respectively . in each test , we first draw randomly 1000 samples according to the function and noise distribution , and then obtain a training set randomly with sizes @xmath230 respectively .
three hundred samples are selected randomly as the test set .
mean squared error _ ( mse ) is used to evaluate the regression results on synthetic data . to make the results more convincing
, each test is repeated 10 times .
table [ tab1 ] reports the average mse and _ standard deviation _ ( std ) with 50 training samples and 300 training samples respectively .
furthermore , we study the impact of the number of training samples on the final regression performance .
figure 1 shows the mse for learning @xmath231 with numbers of training samples .
these results illustrate that lfk has competitive performance compared with svm .
this paper investigated the generalization performance of regularized least square regression with fredholm kernel .
generalization bound is presented for the fredholm learning model , which shows that the fast learning rate with @xmath0 can be reached . in the future , it is interesting to investigate the leaning performance of ranking @xcite with fredholm kernel . the authors would like to thank prof.dr.l.q .
li for his valuable suggestions .
this work was supported by the national natural science foundation of china(grant nos .
11671161 ) and the fundamental research funds for the central universities ( program nos .
2662015py046 , 2014py025 ) .
99 l. shi , y. feng , and d.x .
zhou , `` concentration estimates for learning with @xmath232-regularizer and data dependent hypothesis spaces , '' _ appl .
2 , pp . 286302 , 2011 .
b. zou , r. chen , and z.b .
xu , `` learning performance of tikhonov regularization algorithm with geometrically beta - mixing observations , '' _ journal of statistical planning and inference _ , vol .
10771087 , 2011 . | learning with fredholm kernel has attracted increasing attention recently since it can effectively utilize the data information to improve the prediction performance . despite rapid progress on theoretical and experimental evaluations ,
its generalization analysis has not been explored in learning theory literature . in this paper
, we establish the generalization bound of least square regularized regression with fredholm kernel , which implies that the fast learning rate @xmath0 can be reached under mild capacity conditions .
simulated examples show that this fredholm regression algorithm can achieve the satisfactory prediction performance .
fredholm learning , generalization bound , learning rate , data dependent hypothesis spaces |
cooperation plays an important role in real world , ranging from biological systems to economic and social systems @xcite .
scientists from many different fields of natural and social sciences often resort to evolutionary game theory @xcite as a common mathematical framework and the prisoner s dilemma game ( pdg ) as a metaphor for studying cooperation between unrelated individuals @xcite .
the original pdg describes the pairwise interactions of individuals with two behavioral options : the two players must simultaneously decide whether to cooperate or to defect . for mutual cooperation
both players receive the rewards @xmath9 , but only the punishment @xmath10 for mutual defection . a defector exploiting a cooperator
gets an amount @xmath11 ( temptation to defect ) and the exploited cooperator receives @xmath12 ( sucker s payoff ) .
these elements satisfy the following two conditions : @xmath13 and @xmath14 .
it is easy to see that defection is the better choice irrespective of the opponent s decision .
thus , the undesired outcome of mutual defection emerges in well - mixed populations @xcite , which has inspired numerous investigations of suitable extensions that enable cooperative behavior to emerge and persist .
some previous works have suggested several mechanisms ( e.g. , kin selection @xcite , the introduction of tit - for - tat@xcite strategy , and voluntary participation @xcite ) to facilitate the emergence and persistence of cooperation in the populations .
the spatial versions @xcite of the evolutionary pdgs can explain the maintenance of cooperation for the iterated games with a limited range of interaction if the players follow one of the two simplest strategies ( defection ( @xmath15 ) and cooperation ( @xmath16 ) ) . recently , the effect of heterogeneous influence of different individuals on the maintenance of cooperative behavior has been studied on regular small - world networks @xcite .
et al . _ have studied the evolutionary pdg and the snowdrift game with preferential learning mechanism on the barabsi - albert networks @xcite . in the evolutionary games the players wish to maximize their total payoffs , coming from pdgs with the neighbors , by
adopting either the deterministic rule introduced by nowak and may @xcite or the stochastic evolutionary rule by szab and tke @xcite . in the present work ,
we make further studies of the evolutionary pdg on square lattice mainly according to the stochastic update rule .
it is natural to consider that different individuals may have different attractiveness in social systems , so when updating their strategies , the individuals may not completely randomly choose a neighbor to refer to .
here , we introduce the nonlinear attractive effect into the game ( see the model below ) .
interestingly , we find that the introduction of this effect can remarkably promote cooperative behavior in the pdg in comparison with the random choice case on square lattice .
[ 0.5 ] ( color online ) .
monte carlo results ( a ) and theoretical analysis ( b ) for cooperator density @xmath4 in the steady state as a function of the temptation to defect @xmath17 for several different values of @xmath2 ( see the plot ) .
the pair - approximation correctly predicts the trends that @xmath4 changes with @xmath17 and @xmath2 , but significantly overestimates the extinction thresholds in contrast to the simulation results ( see the text for details).,title="fig : " ] ( the value just below the extinction threshold @xmath18 when @xmath19 ) for several different values of @xmath2 : ( a ) @xmath20 , ( b ) @xmath19 , ( c ) @xmath21 , ( d ) @xmath22 , ( e ) @xmath23 , ( f ) @xmath24.,width=472 ] as a function of the distance to the extinction threshold @xmath25 for several values of @xmath2 : ( a ) @xmath19 , @xmath18 , ( b ) @xmath21 , @xmath26 , ( c ) @xmath27 , @xmath28 , ( d ) @xmath22 , @xmath29 , ( e ) @xmath23 , @xmath30 , ( f ) @xmath24 , @xmath31 . the solid lines , the power laws @xmath32 fit the data correspondence with the exponents of @xmath33 ( the detailed values are given in the plots , where the figures between parentheses indicate the statistical uncertainties of the last digit).,width=472 ] [ 0.7 ] .
the extinction threshold @xmath34 changes with @xmath2 ( from zero to ten ) .
inset shows that the increment @xmath35 of the extinction threshold @xmath34 changes with @xmath2 ( see the text).,title="fig : " ]
we consider the evolutionary pdg on square lattice with periodic boundary conditions .
each player interacts only with its four nearest neighbors ( self - interaction is excluded ) @xcite , and collects payoffs dependent on the payoff - matrix parameters .
the total payoff of a certain player is the sum over all its interactions .
we have inspected that if every player interacts with its first and second nearest neighbors or the self - interactions are included , the qualitative results are unchanged . following common practices @xcite , we start by rescaling the game to make @xmath36 , @xmath37 , and @xmath38 , where @xmath17 represents the advantage of defectors over cooperators @xcite , being typically constrained to the interval @xmath39 , such that it depends on a single parameter @xmath17 .
we have checked that the qualitative results do not change if we make @xmath40 in order to strictly enforce a pd setting . during the evolutionary process
, each player is allowed to select one of its neighbors as a reference with a probability proportional to the neighbors attractiveness , and then decides whether to change its strategy or not dependent on their payoff difference .
we define the selection probability @xmath41 of @xmath42 selecting a neighbor @xmath43 as @xmath44 where the numerator denotes the attractiveness of the neighbor @xmath43 , and @xmath2 is a tunable parameter describing the extent of the nonlinear effect , and @xmath45 is the total payoff of that neighbor .
the denominator is the sum of attractiveness that runs over all neighbors of @xmath42 .
the basic ingredient which determines the choice of one neighbor is the _ selection kernel _ @xmath0 . on general grounds ,
this _ selection kernel _ should be a nondecreasing function of @xmath1 , namely individuals with better performance may have much stronger attractiveness than the average individual .
we note that the selection probability depends only on the extent of a nonlinear effect @xmath2 since the total payoff collected by any player satisfies @xmath46 in the present model .
thus we will consider the model with @xmath47 .
for @xmath20 , the neighbor is randomly selected so that the game is reduced to the original one in refs .
the case @xmath19 leads to the proportional selection rule ( exclude the player itself ) @xcite . while in the limit of @xmath48 , the neighbor whose payoff is the highest among the neighbors is selected , which resembles to the deterministic selection rule @xcite . for other values of @xmath2 ,
the attractiveness of the neighbors is a nonlinear function of their total payoffs . in this way
, we consider the general situations of the nonlinear attractive effect on the dynamical behavior of the game .
the player @xmath42 adopts the selected @xmath43 neighbor s strategy in the next round with a probability depending on their total payoff difference presented in ref . @xcite as @xmath49 } , \label{rule2}\ ] ] where @xmath50 , @xmath45 denote the total payoffs of individuals @xmath42 and @xmath43 respectively , and @xmath51 characterizes the noise effects , including fluctuations in payoffs , errors in decision , individual trials , etc .
the effect of noise has been reported by szab _
_ @xcite . in this paper
, we make @xmath52 .
qualitatively , the results remain unaffected when changing the parameter @xmath51 .
simulations were carried out for a population of @xmath53 individuals .
we study the key quantity of cooperator density @xmath4 in the steady state .
initially , the two strategies of @xmath16 and @xmath15 are randomly distributed among the individuals with equal probability @xmath54 .
the above model was simulated with synchronous updating @xcite .
no qualitative changes occur if we adopt an asynchronous updating @xcite . eventually , the system reaches a dynamic equilibrium state .
the simulation results were obtained by averaging over the last @xmath55 monte carlo time steps of the total @xmath56 .
[ fig1 ] shows the results of both simulations and theoretical analysis of @xmath4 when increasing @xmath17 for several different values of @xmath2 ( see the plot ) .
we can find that , compared with the well - mixed situation , if @xmath17 is sufficiently small , cooperators can persist in spatial settings in the case of @xmath20 , which indicates that spatial structure can promote cooperation @xcite .
for @xmath57 ( @xmath58 ) , where @xmath59 is the extinction threshold of cooperators when @xmath20 , the benefits of spatial clustering are no longer sufficient to offset the losses along the boundary , hence the cooperators vanish @xcite . for each positive value of @xmath2
, the system evolves to the absorbing state of all defectors at certain values of @xmath17 .
the extinction threshold of cooperators @xmath34 clearly increases with @xmath2 , which indicates that the emergence of cooperation is enhanced .
we know that cooperators survive by forming compact clusters and thus cooperators along the boundary can outweigh their losses against defectors by gains from interactions within the cluster @xcite .
the payoffs collected by the inner cooperators are , in most cases , larger than the boundary defectors .
when considering the attractiveness of the individuals , near the extinction threshold , for cooperator - clusters which are surrounded by defectors , the cooperators along the boundary can keep their cooperative states more easily under stronger preferential selection effect according to the dynamic updating eq .
( [ rule1 ] ) and eq .
( [ rule2 ] ) .
thus the strong attractiveness of individuals can favor the spreading of cooperators , hence promote the persistence of cooperation .
we can also see that , for very large @xmath2 , the homogeneous cooperation state ( @xmath60 ) emerges when @xmath17 is very small . since both the small temptation to defect and the strong nonlinear attractive effect are advantageous to the persistence of cooperation , it is not surprising that @xmath4 approaches the maximal fraction @xmath61 .
moreover , for the same value of @xmath17 , @xmath4 obviously increases with @xmath2 . the existence of strong nonlinear effect can facilitate the formation of cooperator clusters , hence enhance the persistence of cooperators ( see fig .
[ fig2 ] ) .
in addition , the cooperator density @xmath4 and the extinction thresholds @xmath34 change more slowly with increasing @xmath2 .
we will consider this point in the following .
the pair approximation method , which models the frequency of strategy pairs rather than that of strategies , is usually regarded as an analytical approximation of the spatial dynamics @xcite .
it is worth noting that the results obtained by pair approximation are directly associated with the local topological structure of the players and the strategy updating dynamics .
whether the strategy updating is implemented synchronously or asynchronously is inessential @xcite .
therefore we can apply the pair approximation method to predict qualitatively the evolving behavior of @xmath4 . here
we modify the original method by introducing the turnable parameter @xmath2 , i.e. , we substitute the new transition probability @xmath62 for the original transition probability @xmath63 in eq .
( a1 ) in ref.[18 ] , where @xmath64 and the denominator of the fraction denotes the sum of all possible values of @xmath65 when @xmath2 is certain .
then we adjust the eq .
( a2a ) and eq .
( a2b ) in ref.@xcite by substituting @xmath66 for @xmath67 .
the equilibrium values are obtained by numerical integration .
we can see from fig .
[ fig1 ] that the pair approximation correctly predicts the trends , that is , the changes of cooperation for @xmath17 and @xmath2 .
however , it is unable to estimate exactly the extinction thresholds of cooperator density , namely , it overestimates the extinction thresholds ( see the plot ) .
a series of snapshots for several different values of @xmath2 are shown in fig .
[ fig2 ] for the same value of @xmath17.these snapshots are a @xmath68 portion of the full @xmath69 lattice .
we can find that , for random selection case @xmath20 , cooperators are doomed and defectors reign because the value @xmath70 is larger than @xmath71 ( fig .
[ fig2](a ) ) .
but for @xmath19 , when @xmath17 is just below @xmath72 , the cooperators can survive by forming compact clusters which minimize the exploitation by defectors ( see fig .
[ fig2](b ) ) . for larger @xmath2 ,
i.e. , stronger nonlinear attractive effect , more cooperator clusters emerge , which illustrates that cooperators can survive against the invading of defectors more easily . on the contrary ,
the defector clusters decrease with the value of @xmath2 .
interestingly , the spatial patterns adopted by defectors are completely different from that of cooperators when they are the minority in the populations . as shown in fig .
[ fig2 ] ( f ) , defectors exist in the fashion of zigzag pattern ( or step - like ) . because if defector clusters are surrounded by cooperators , in the subsequent generations , the defectors along the boundary would transform probably to cooperators according to the selection rule with nonlinear attractive effects eq .
( [ rule1 ] ) and updating rule eq .
( [ rule2 ] ) . eventually , defectors exist in zigzag pattern from which they can benefit maximumly when interacting with their cooperator neighbors ( even can do better than a cooperator surrounded by four cooperators , so that when updating , they alwaysask their defector neighbors for strategy transformation ) , which in return makes the zigzag pattern stably in the evolution of the game .
we also investigate the divergence behaviors of cooperators near the extinction thresholds @xmath34 for several values of @xmath2 ( see fig .
[ fig3 ] ) .
we constrained the system size as @xmath73 individuals and the results were obtained by averaging over the last @xmath74 time steps of the total @xmath75 . according to our simulations , near @xmath34
the average fraction of cooperators vanishes as power - law like behavior @xmath76 , where @xmath33 are a set of exponents corresponding with the value of @xmath2 ( see the plots for detailed values ) . in physics , such thresholds
are usually associated with phase transitions , and indeed , the transitions from persistent levels of cooperation ( @xmath77 ) to absorbing states of defection ( @xmath78 ) bear the hallmarks of critical phase transitions .
these values of @xmath33 are nearly consistent with the critical exponent ( @xmath8 ) of the two - dimensional directed percolation @xcite and depend weakly on @xmath2
. the estimated errors of @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 are @xmath85 respectively .
we consider that the large errors are due to the limits of computational conditions .
the variation of extinction threshold @xmath34 with the value of @xmath2 ( from zero to ten ) is shown in fig .
we can see that the extinction threshold clearly increases with the value of @xmath2 .
now we define a quantity characterizing the increment of the extinction threshold as @xmath86 .
we note that this quantity decreases with the value of @xmath2 ( inset in fig .
[ fig4 ] ) .
this indicates that , as @xmath48 , the increment @xmath35 will approach to the minimal value @xmath87 . in other words , the extinction threshold @xmath34 will tend towards the maximal value @xmath88 , where @xmath89 is the extinction threshold when @xmath48 , i.e. , when the neighbor whose payoff is the highest of the neighboring is selected to refer to .
in summary , we have investigated the promotion of cooperation in the context of evolutionary pdg resulting from the nonlinear attractive effect of the neighbors on square lattice . a nonlinear function @xmath90 , in terms of the performance of the players , is used as an estimator of their attractiveness .
we have considered the general situations and shown that , compared with the random selection case , the introduction of the nonlinear attractive effect can remarkably promote the cooperative behavior over a wide range of @xmath17 .
particularly , the stronger the extent of the nonlinear effect is , the more prominent the cooperative behavior will be , and for some large @xmath2 values , a homogeneous state of all cooperators can emerge .
interestingly , the spatial patterns adopted by cooperators and defectors are completely different when they are the minority in the populations : cooperators can survive by forming compact clusters , and along the boundary , cooperators can outweigh their losses against defectors by gains from interactions within the cluster ; whereas defectors exist in the way of zigzag pattern ( or step - like ) , from which defectors can benefit maximumly when interacting with their cooperator neighbors . the extinction of cooperators under harsh conditions when @xmath91 displays a power law - like behavior @xmath92
. the introduction of the nonlinear attractive effect can partially resolve the dilemma of cooperation and may shed new lights on the evolution of cooperation in the society . | we introduce nonlinear attractive effects into a spatial prisoner s dilemma game where the players located on a square lattice can either cooperate with their nearest neighbors or defect . in every generation ,
each player updates its strategy by firstly choosing one of the neighbors with a probability proportional to @xmath0 denoting the attractiveness of the neighbor , where @xmath1 is the payoff collected by it and @xmath2 ( @xmath30 ) is a free parameter characterizing the extent of the nonlinear effect ; and then adopting its strategy with a probability dependent on their payoff difference . using monte carlo simulations
, we investigate the density @xmath4 of cooperators in the stationary state for different values of @xmath2 .
it is shown that the introduction of such attractive effect remarkably promotes the emergence and persistence of cooperation over a wide range of the temptation to defect .
in particular , for large values of @xmath2 , i.e. , strong nonlinear attractive effects , the system exhibits two absorbing states ( all cooperators or all defectors ) separated by an active state ( coexistence of cooperators and defectors ) when varying the temptation to defect . in the critical region where @xmath4 goes to zero ,
the extinction behavior is power law - like @xmath4 @xmath5 @xmath6 , where the exponent @xmath7 accords approximatively with the critical exponent ( @xmath8 ) of the two - dimensional directed percolation and depends weakly on the value of @xmath2 . |
the evolution of giant molecular clouds ( gmcs ) is a key ingredient in our understanding of the star formation process . in particular , the low observed star formation efficiency ( sfe ) at the scale of whole gmcs ( @xmath12% ; * ? ? ?
* ) remains a topic of strong debate , with there being two main competing scenarios that attempt to explain it .
these scenarios refer essentially to the effect of the stellar feedback ( mainly from massive stars ) on the star - forming clouds .
one is the scenario that the stars quickly disrupt their parent clouds by dispersal and/or photoionization , before the gaseous mass of the cloud is completely converted to stars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in this case , the sfr of active star - forming sites may be large for brief periods , and then halted by the very stars that have just been formed . in the other scenario ,
the role of stellar feedback is to drive turbulent motions within the gmc , which oppose its self - gravity , allowing it to remain in near hydrostatic equilibrium for times significantly longer than its free - fall time ( @xmath2 ) @xcite . in this case
, the low efficiency of star formation would be due to the dual role of supersonic turbulence in self - gravitating clouds , of opposing global collapse of the cloud while promoting local collapse of turbulent density enhancements , which involve small fractions of the total cloud mass ( * ? ? ?
* ; * ? ? ?
* see also the reviews by mac low & klessen 2004 ; ballesteros - paredes et al .
2007 ) . another controversy , related to the control of the sfe , refers to the nature of the motions originating the linewidths observed in gmcs and their substructure .
the latter were initially proposed to correspond to gravitational contraction by @xcite , but this suggestion was quickly deemed untenable by @xcite , who noted that it would imply total galactic sfrs of the order of the total molecular gas mass in the galaxy ( @xmath3 ) divided by the typical free - fall time for a gmc ( @xmath4 myr ) , or @xmath5 yr@xmath6 , an estimate roughly two orders of magnitude larger than the observed galactic sfr .
@xcite then suggested that the observed linewidths could correspond instead to random , small - scale turbulent motions , a notion that has prevailed until the present .
however , a number of workers have recently advocated a return to the gravitational contraction picture , noting that various observational properties of clouds and clumps can be well matched by models dominated by gravitational contraction ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
moreover , the notion of completely random , small - scale turbulent motions appears difficult to reconcile with the recent realization that the principal component of the velocity differences within clouds and clumps at all scales appears to be `` dipolar '' , indicative of coherent motions at the scale of the whole cloud or clump @xcite . in general ,
several studies comparing simulations and observations have concluded that the motions in molecular clouds are consistent with large- , rather than small - scale driving @xcite .
if a return to the collapsing scenario is to be considered , it is necessary to somehow avoid the @xcite criticism of it .
this is actually not so difficult because that criticism neglects the internal structure of the gmcs .
recent numerical studies of cloud formation by converging streams of warm neutral gas in the interstellar medium show that the clouds are born turbulent , due to one or more of the thermal , thin - shell , and kelvin - helmholz instabilities @xcite .
the turbulence is subsonic with respect to the warm gas , but supersonic with respect to the cold phase , implying large density fluctuations in the latter .
in fact , it has recently been proposed that molecular clouds may actually contain a warmer , atomic substrate in which colder molecular clumps are embedded @xcite . in either case ,
the molecular cloud contains large , _ nonlinear _ density enhancements in which the local free - fall time is significantly shorter than the cloud s average .
thus , once the global collapse begins , the local clumps may complete their collapses earlier than the bulk of the cloud .
they can thus begin forming stars that can begin their feedback action on the gmc before it completes the bulk collapse . in this paper , we present numerical simulations aimed at investigating this scenario , in which we use the same cloud - formation setup of previous papers @xcite , but including a prescription for stellar feedback mimicking ionization heating from massive stars . with this tool
, we investigate the effect of the feedback on the global sfe of the evolving gmc , as well as the nature of the motions in the cloud , in a first effort to shed light on these issues .
as we shall see , it turns out that the physical conditions in the clouds differ significantly from the `` normal '' picture , since accretion of gas from the warm diffuse medium is an integral part of the clouds dynamics and evolution , and thus the clouds can not be considered as isolated .
the plan of the paper is as follows . in [ sec : model ] we describe the numerical code , and the implementation of our star formation and stellar feedback prescriptions . in [ sec : simulations ] we describe the simulations , and in [ sec : results ] we describe the results concerning the control of the sfe by stellar feedback and the nature of the `` clouds '' themselves . finally , in
[ sec : conclusions ] we present a summary and some conclusions .
the numerical simulations used in this work were performed using the hydrodynamics + n - body adaptive refinement tree code art @xcite . among the physical processes implemented in art , relevant for our physical problem ,
are the radiative heating and cooling of the gas , its conversion into stars , ionization - like heating from stellar feedback , and self - gravity , from both the stars and the gas .
we use heating ( @xmath7 ) and cooling ( @xmath8 ) functions of the form @xmath9 these functions are fits to the various heating and cooling processes considered by @xcite , as given by equation ( 4 ) of @xcite . as noted in ( * ? ? ?
* hereafter paper i ) , eq .
( 4 ) in @xcite contains two typographical errors .
the form used here incorporates the necesary corrections , kindly provided by h. koyama ( 2007 , private communication ) . with these heating and cooling laws ,
the gas is thermally unstable in the density range @xmath10 ( cf .
paper i ) . in art ,
star formation is modeled as taking place in the coldest and densest regions , defined by @xmath11 and @xmath12 , where @xmath13 and @xmath14 are the local temperature and number density of the gas , respectively , and @xmath15 and @xmath16 are respectively a density and a temperature threshold .
we set @xmath17 k , which is easily satisfied by all cells with density @xmath15 , so in practice our sf condition depends on density only .
a stellar particle of mass @xmath18 is placed in a grid cell where these conditions are simultaneously satisfied , and this mass is removed from the gas mass in the cell .
thereafter , the particle is treated as non - collisional , and follows n - body dynamics .
no other criteria are imposed . in each gas cell
that satisfies the above criteria a stellar particle is formed with a mass equal to 50% of the gas mass contained in the cell . since the stellar particles are more massive than a single star , each stellar particle should be considered as a small cluster , within which the individual stellar masses are distributed according to some stellar initial mass function ( imf ) .
stellar particles inject thermal energy at a rate @xmath19 erg myr@xmath6 per star with mass greater than @xmath20 contained in the stellar particle .
we assume a @xcite imf , implying that each stellar particle of mass @xmath21 produces one 8-@xmath22 star .
the energy is deposited in the cell in which the stellar particle is instantaneously located , over a typical ob stellar lifetime , which we assume to be 10 myr .
it is important to note that , although initially we experimented with realistic values of @xmath19 based on the lyman continuum fluxes of stars with masses between 10 and 20 @xmath22 ( e.g. , * ? ? ?
* ) , we found that , because all the energy is deposited in a single cell , and the neighboring cells are heated exclusively by conduction , rather than by radiative heating , the resulting hii regions were not so realistic .
thus , we opted instead for taking @xmath19 as a free parameter , and adjusting it until we obtained realistic hii regions , with temperatures @xmath23 k , diameters of a few parsecs , and expansion velocities of a few tens of@xmath24 .
note also that we resort to the common strategy of turning off the cooling in the cell where a stellar particle is located , so that the cell can reach realistically high temperatures . otherwise , the cooling can dissipate most of the thermal energy deposited in very dense cells . in the real ism
this does not occur because the stellar heating is applied through ionization , so that the temperature reached in the star s immediate environment is independent of the medium s local density . instead , in the simulations , the cooling depends on the density , and the temperature resulting from the balance between the stellar heating and the cooling
does depend on the density .
this problem is avoided by turning off the cooling in the cell where the stellar particle is located .
note that this contradicts claims that the need to turn off the cooling can be alleviated simply by increasing the resolution ( e.g. , * ? ? ?
we argue that this problem can only be alleviated by performing radiation - hydrodynamics simulations . in their absence
, we consider that turning off the cooling is actually a _ better _ model of the effect of stellar feedback , because it allows mimicking the fact that the gas temperature in the hii regions is independent of the local density .
finally , in order to further constrain the physical conditions in the hii regions , we also impose a `` ceiling '' to the temperature in the cell containing the stellar particle because otherwise , with the cooling off , the temperature in the cell might diverge .
we set this `` ceiling '' to @xmath25 k. although this procedure is mostly one of trial - and - error , we consider it to be the most adequate one for our purposes , since it is the hii regions that drive the turbulent motions in the dense gas in our simulations , and so it is them that must have realistic properties , even at the expense of a somewhat ad - hoc sf prescription . we show a typical hii region in our simulation in fig .
[ fig : hii_region ] .
the numerical box is initially covered by a grid of @xmath26 ( zeroth level ) cells .
the mesh is subsequently refined as the matter distribution evolves .
the maximum allowed refinement level was set to four , so that high - density regions have an effective resolution of @xmath27 cells , with a minimum cell size of 0.125 pc .
cells are refined ( halved in linear size ) if the gas mass within the cell is greater than 0.32 .
that is , the cell is refined by a factor of 2 when the density increases by a factor of 8 , so that , while refinement is active , the grid cell size @xmath28 scales with density @xmath14 as @xmath29 .
once the maximum refinement level is reached , no further refinement is performed , and the cell s mass can reach much larger values .
in particular , a stellar particle is formed when a fourth - level cell reaches a density @xmath30 , or a mass of @xmath31 , again assuming @xmath32 ( we use this value because we do not follow the actual chemistry , and thus we assume the entire box to be filled with atomic hydrogen . ) thus , a stellar particle typically has a mass @xmath33 .
note that , because we use only four levels of refinement , the largest densities arising in the simulation are by far not sufficiently resolved according to the `` jeans criterion '' proposed by @xcite for adaptive - mesh codes . specifically , at our stellar - particle formation threshold density of @xmath34 , and assuming a gas temperature of @xmath35 k at that density , we find that the jeans length ( using the adiabatic sound speed ) is @xmath36 pc , while the minimum cell size , at 0.125 pc , is roughly 4 times larger .
thus , according to those authors , one should expect artificial fragmentation to occur in our simulations .
however , we do not consider this to be a problem because we are not concerned here with the numbers and masses of the stellar particles formed in the simulation , but simply with the _ total _ amount of mass that goes into stars .
we consider four simulations using the same setup as in paper i , which represents the evolution of a region of 256 pc per side , initially filled with warm gas at a uniform density of @xmath37 and a temperature @xmath38 k , implying an adiabatic sound speed @xmath39 ( assuming a mean particle mass @xmath40 ) .
the whole numerical box thus contains @xmath41 . in this medium
, we set up two streams moving with the same speed @xmath42 ( corresponding to a mach number of 0.8 with respect to the unperturbed medium ) in opposite senses along the @xmath43-direction .
the streams have a radius of 32 pc and a length of 112 pc each , so that the total mass in the two inflows is @xmath44 .
the flows collide head on at the box s center ( see fig . 1 of paper i ) . to the inflow velocity field
we superpose a field of initial low - amplitude turbulent velocity fluctuations , in order to trigger the instabilities in the compressed layer .
we create this initial velocity fluctuation field with a new version of the spectral code used in @xcite and @xcite , modified to run in parallel in shared - memory architectures .
the simulations are evolved for about 40 myr .
the collision nonlinearly triggers a transition to the cold phase , forming a turbulent , cold , dense cloud @xcite , consisting of a complex network of sheets , filaments , and clumps of cold gas embedded in a warm diffuse substrate @xcite .
the largest cold structures may become gravitationally unstable and begin to collapse .
eventually , they proceed to forming stars , which then heat their environment , forming expanding `` hii regions '' that tend to disperse the clumps . in the simulations reported here
, we vary only two parameters : the amplitude of the initial velocity fluctuations and whether the stellar feedback is on or off .
we consider a `` large - amplitude '' ( la ) and a `` small - amplitude '' ( sa ) case for the initial velocity fluctuations , for which the three - dimensional velocity dispersions are @xmath45 and @xmath46 , respectively .
we thus employ a mnemonic nomenclature for the runs using the acronyms la or sa , followed by f0 or f1 , indicating that feedback is off or on , respectively . table [ tab : run_parameters ] summarizes the runs considered in the paper .
ccc + laf0 & 1.7 & off + laf1 & 1.7 & on + saf0 & 0.1 & off + saf1 & 0.1 & on + [ tab : run_parameters ]
the simulations performed here behave very similarly to previous simulations with similar setups , performed with other codes . in particular , our sa runs are very similar to run [email protected] in paper i and the run presented by @xcite .
the main feature of these runs is that , because the initial fluctuations are very mild , the flow collision creates a large , coherent `` pancake '' of cold , dense gas , which is able to undergo gravitational collapse as a whole .
this results in the formation of a dense , massive , and turbulent region at the site where the global collapse finally converges , with physical properties similar to those of high - mass star forming regions @xcite .
however , a recent study varying the parameters of the flow collision @xcite shows that the coherence of the collapse may be lost in the presence of stronger initial velocity fluctuations , and the sfe is decreased . in such cases , smaller clouds appeared to be less strongly gravitationally bound , with the effect of decreasing the sfe .
this feature also happens in our la simulations , in which the cloud formed by the initial flow was much more fragmented and scattered over the simulation volume . as a result , sf also occurs in a much more scattered manner , and the sfes are in general smaller in the la runs than in their sa counterparts .
however , in general a common pattern is followed by all simulations : the transonic converging flows in the diffuse gas induce a phase transition to the cold phase of the atomic gas , which is highly prone to gravitational instability .
this can be seen as follows .
the thermal pressure at our initial conditions is 5000 k@xmath48 . from fig . 2 of paper i
, it can be seen that the thermal balance conditions of the cold medium at that pressure are @xmath49 , @xmath50 k. at these values , the jeans length and mass are @xmath51 pc and @xmath52 , respectively .
these sizes and masses are easily achievable by a large fraction of the cold gas structures , which can then proceed to gravitational collapse and form stars . moreover , the ensemble of these clumps may also be gravitationally unstable as a whole , the likelihood of this being larger for greater coherence of the large - scale pattern
. regions of active star formation form in both sets of simulations by the gravitational merging of pre - existing smaller - scale clumps , which , altogether , form a larger - scale gmc . figure [ fig : laf1_whole_proj ] shows a whole - box image of the density field in run laf1 , in projection . in the electronic version of the paper
, this figure shows an animation of this run from @xmath53 to @xmath54 myr , illustrating the entire evolution of the simulation , from the assembly of the cloud , to its advanced star - forming epochs . in the animation ,
subsequent `` records '' are separated by time intervals @xmath55 myr . in the sa runs , the largest star - forming region forms in the center of the simulation , due to the coherent collapse of the entire sheet - like cloud formed by the collision .
this was the region shown in @xcite to exhibit physical conditions typical of actual high - mass star - forming regions .
we refer to this region as `` the central cloud '' .
figure [ fig : sa_cc_t265_zoom ] shows a view of this region at @xmath56 myr , a time at which the central cloud has grown to a mass of nearly @xmath57 in run saf1 ( cf .
fig.[fig : sa_gas_star_mass_evol ] ) . in the case of
the la runs , because star formation occurs in a much more scattered fashion , we study two of the regions exhibiting the strongest star formation activity , neither of which is located at the center of the numerical box .
these are shown in fig .
[ fig : la_t248_zoom ] , and we refer to them as cloud 1 and cloud 2 .
our main interest in this contribution is the effect of the feedback on the efficiency of the star formation process , and the identification of the mechanism through which this effect is accomplished .
figure [ fig : sa_gas_star_mass_evol ] shows the evolution of the dense gas mass and the stellar mass in these simulations , with ( _ right panels _ ) and without ( _ left panels _ ) feedback .
the _ solid _ lines refer to the total masses in the computational box , while the _ dotted _ lines refer to the masses in the central cloud .
figures [ fig : la_cl1_gas_star_mass_evol ] and [ fig : la_cl2_gas_star_mass_evol ] show the corresponding plots for cloud 1 and cloud 2 . here , the _ solid lines _
represent the masses for the full simulation box , and the _ dotted , short - dashed , and long - dashed lines _
represent the masses contained in cylinders of length and diameter 10 , 20 , and 30 pc , respectively , enclosing the clouds .
we do this because the clouds have very complicated morphologies , with filaments that extend out over tens of parsecs and connecting with other clouds ( fig .
[ fig : la_t248_zoom ] ) , thus making it virtually impossible to fully enclose the `` clouds '' in any given cylindrical volume .
-direction , and which contains the central cloud .
, scaledwidth=50.0% ] it is seen from figs.[fig : sa_gas_star_mass_evol]-[fig : la_cl2_gas_star_mass_evol ] that the inclusion of feedback ( _ right panels _ ) causes the dense gas mass to be _ larger _ and the stellar mass to be smaller than in the case without feedback in general , even though the total cloud mass ( dense gas + stars ) in the simulations is nearly the same in both the cases with and without feedback ( fig .
[ fig : total_mass_evol ] ) . as a consequence ,
the instantaneous sfe , defined as @xmath58 where @xmath59 is the mass of the dense ( @xmath60 ) gas in the simulation , and @xmath61 is the stellar mass , naturally decreases upon the inclusion of feedback in both sets of simulations ( fig.[fig : sfe_global ] ) .
note that in eq .
( [ eq : sfe_def ] ) we have explicitly written out the time dependence of the quantities involved . the sfe is seen to be reduced by a larger factor ( @xmath62 ) in the case of the la runs , in which the collapse is less focused and less massive , than in the case of the sa ones ( @xmath63 ) , in which the opposite is true .
in addition , in fig . [
fig : sfe_local ] we show the evolution of the sfe at the level of the clouds .
the _ left panel _ shows the evolution of the sfe for the central cloud in the sa runs .
the _ middle panels _ show the corresponding plots for cloud 1 and cloud 2 in the laf0 run ( without feedback ) , and the _ right panels _ show the sfes in the laf1 run ( with feedback ) .
again we see a trend for the less massive cloud ( cloud 1 ) to suffer a greater reduction of its sfe ( by a factor of @xmath64 , from @xmath65 - 90% to @xmath66 - 4% ) than the more massive one ( cloud 2 , by a factor of @xmath0 , from @xmath67 - 80% to @xmath51 - 8% ) .
we discuss the possible causes for this mass - dependent effect of the feedback in sec .
[ sec : accr_vs_feed ]
. the factor by which the sfe is reduced upon the inclusion of the feedback in the simulations is plotted versus the system s mass in fig.[fig : sfe_vs_mass ] , both for the full amount of dense gas in the simulations , and for each of the clouds we have been considering .
we see that two sets of points are clearly defined in this plot , one for the clouds , and one for the simulations . in both cases , however , the trend of larger reduction factor at smaller dense gas mass is clearly observed , although , at the level of simulations , we see that their masses are not very different .
thus , in this case the different reduction factors must include a contribution from the larger degree of fragmentation occurring in the la simulations due to the larger amplitude of the initial turbulent fluctuations .
beginfigure in order to compare the sfes in our simulated clouds with those of real molecular clouds , it is convenient to note that regions of low - mass star formation , such as taurus @xcite or the chamaeleon ii dark cloud @xcite generally have low sfes ( @xmath68 - 5% ) , while cluster forming cores have sfe @xmath69 - 50% @xcite .
thus , we can check whether the sfes and sfrs of our three clouds follow the same trend .
figure [ fig : sfrnubes ] shows the evolution of the sfrs for our three clouds . from these plots , we find that the average sfr of the central cloud starting from @xmath70 myr ( the time at which a large , roughly stationary sfr sets in ) is @xmath71 myr@xmath6 , while for clouds 1 and 2 , we find @xmath72 and 60 @xmath22 myr@xmath6 during their entire star - forming stages , respectively . from here , and using the instantaneous sfe at @xmath73 myr from fig . [ fig : sfe_local ] , we can then plot the sfe _ versus _ the sfr .
this is shown in fig.[fig : sfe_vs_sfr ] , in which both the sfr and the sfe have been multiplied by an extra factor of 0.5 , to respresent the fact that the efficiency within the stellar particles , which are created at a density of @xmath74 in our simulations , is still smaller than unity .
we take 0.5 as a representative value . ) , in the full simulation box in the four runs . ,
scaledwidth=50.0% ] we see that the central cloud has values of the sfr and the sfe comparable to those of cluster - forming cores @xcite . specifically , @xcite estimated an sfr @xmath75 myr@xmath6 for the orion nebula cluster ( onc ) .
this calculation used an estimated age spread of the onc of @xmath76 myr @xcite , and @xcite s result of there being 1613 stars in the onc . taking this number as a proxy for the total stellar production of this region , and a mean stellar mass of @xmath77 @xcite , this implied a total stellar mass of @xmath78 . on the other hand , @xcite quote a total stellar mass in the onc of @xmath79 @xcite and an age spread of @xmath66 myr @xcite , implying an sfr @xmath80 myr@xmath6 .
these values bracket the sfr we measured for our central cloud . concerning the sfe , fig . [
fig : sfe_vs_sfr ] shows that the central cloud has sfe @xmath81 10% , which is comparable to that of the orion a cloud @xcite , in which the omc-1 clump is contained .
thus , our central cloud may be compared to the orion a cloud , and its dense core , discussed in @xcite , is comparable to the omc-1 clump . on the other hand , clouds 1 and 2
are seen in fig .
[ fig : sfe_vs_sfr ] to have sfrs comparable to those of low - mass star - forming clouds , @xmath82 - 100 @xmath22 myr@xmath6 , such as taurus , cha ii or lupus , in which the sfe is known to be lower , @xmath68 - 4% @xcite .
this compares very well with the sfes we report for our clouds 1 and 2 in the same figure .
we conclude that clouds 1 and 2 are good models of low - mass star - forming regions , while the central cloud is a good model of a massive sf region , as discussed in @xcite .
-[fig : la_cl2_gas_star_mass_evol ] ) .
the error bars in the reduction factor for clouds 1 and 2 denote the maximum and minimum values of the ratio sfe(w / o feedback)/sfe(w / feedback ) , using the sfe data for the three cylinder sizes of 10 , 20 and 30 pc .
, scaledwidth=49.0% ] up to the 40-myr time to which we have evolved our simulations , the three large clouds ( the central cloud in run saf1 and clouds 1 and 2 in run laf1 ) do not show any instances of the dense gas mass reversing its increasing trend and beginning to decrease due to the feedback , as can be seen in figs . [ fig : sa_gas_star_mass_evol ] through [ fig : la_cl2_gas_star_mass_evol ] .
apparently , the hii region - like feedback we use is unable to overwhelm the large gravitational potential well of these clouds and their enveloping `` atomic '' gas reservoirs .
, actual star formation rates should be a factor of 2 - 3 lower .
gaps indicate periods over which no new stars are formed.,scaledwidth=70.0% ] instead , complete destruction seems to be able to occur in small clumps .
this can be seen in the animation corresponding to fig.[fig : sa_cc_t265_zoom ] , in which various small clumps are seen to be destroyed by their stellar products .
three particularly conspicuous ones are , first , the one that forms a stellar particle at the very starting frame ( record 179 ) of the animation , slightly above and to the right of the screen s center .
next , another stellar particle appears in the clump almost at the left border of the frame , about 2/3 of the way from bottom to top , at record 187 .
finally , a third particle appears at record 189 , slightly below and to the right of the screen s center , as the result of the collision of two clumps . in all of these cases ,
a single stellar particle is formed ( @xmath83 ) , and the clump is destroyed .
it is worth noting that actually , the expansion of the hii regions formed produces new clumps from the material collected around it , but these new clumps either disperse , or simply do not form new stars . and [ fig : sfrnubes ] .
the values of both the sfr and the sfe have been multiplied by a factor of 0.5 , representative of the still - lower - than - unity efficiency within our stellar particles . , scaledwidth=49.0% ] thus , we conclude , similarly to @xcite , that small clouds ( `` clumps '' ) are rapidly destroyed , while large clouds may survive for longer times . however , our clouds exhibit a fundamental difference with respect to the model considered by those authors , namely that the clouds are accreting in general . in the next section
we now discuss this feature .
one crucial feature in all our simulations is that _ the clouds are accreting material from the surrounding diffuse medium .
_ this is fundamentally different from models in which the clouds are isolated entities , in rough balance between their self - gravity and the turbulent pressure , possibly driven by the stellar feedback .
the accretion competes with sf and stellar feedback in regulating the cloud s mass and coherence , with important consequences .
first of all , this implies that simple observational estimates of the sfe in gmcs based on measuring the stellar mass and dividing it by the cloud s mass may be failing to take into account the additional `` raw material '' for sf contained in the part ( or the whole ) of the atomic envelope of the clouds that will eventually be incorporated into the gmc .
second , the competition between feedback and accretion may explain our observation from sec .
[ sec : sfe ] that cases with feedback are characterized by _
larger _ dense gas masses and smaller stellar masses than their counterparts without feedback .
the smaller stellar mass is not surprising , as the obvious effect of stellar feedback is to reheat the cold , collapsing star - forming gas , thus reducing the sfr .
however , the larger cold gas mass in the presence of feedback is indeed surprising , since both gas consumption by sf and the `` ionization '' by stellar feedback act to reduce the dense gas mass .
our result implies that the rate of dense gas consumption by sf far outweighs its rate of destruction by stellar feedback , so that the net effect of reducing the sfr is to allow a larger amount of dense gas to be collected by the accretion .
this scenario is supported by fig.[fig : total_mass_evol ] , which shows the total ( dense gas + stars ) in the clouds in the two sets of simulations .
it is seen that the total cloud mass is nearly the same with and without feedback , suggesting that _ the total cloud mass is mainly determined by the accretion , while the ratio of dense gas to stellar mass seems to be mainly determined by the feedback .
_ these results moreover support the suggestions by @xcite and @xcite that the majority of the gas in a gmc is not participating of the sf process at any given time .
the latter authors suggested that this is how local regions of sf may have a very high specific sfr ( @xmath84 ) , while that of their whole parent gmcs may be much smaller ( @xmath85 ; * ? ? ? * ) . because it is injected by the newly formed stars , the stellar feedback acts preferentially on gas that is about to form stars next .
this allows an efficient suppression of sf , through only a modest fraction of the total available dense gas being destroyed .
that is , if sf is a highly localized process , then it is possible to achieve significant reductions of the sfr with only a modest reduction of the total amount of dense gas , by targeting the destruction precisely at the star - forming gas .
this mechanism may also explain the trend observed in sec.[sec : sfe ] , that the sfe is more strongly reduced by the feedback in cases where the collapsing gas mass is smaller .
this may be understood as a consequence of the fact that stellar feedback is localized , while the accretion is extended , and more so for greater mass of the globally gravitationally unstable region that will form the cloud .
thus , as we observe in the animation corresponding to fig .
[ fig : sa_cc_t265_zoom ] , the low - mass fragments that are undergoing sf while on their way to the central site of the global collapse in run saf1 are easily destroyed by their stellar activity . instead , the massive central cloud is not destroyed , as it continues to accrete mass from large distances at a pace that outweighs the local destruction by stellar feedback .
clouds 1 and 2 in run laf1 , which do not involve such an extended collapse , are intermediate cases , in which the clouds are not destroyed by the feedback , but the latter is more efficient in reducing the sfe .
thus , we can conclude that the efficacy of the feedback in destroying the cloud is maximal when its region of influence is comparable to the spatial extent of the infall , and is progressively reduced as the latter involves progressively larger coherence lengths .
here we consider the global physical properties of the clouds .
we postpone a discussion of the properties of individual clumps within the clouds for a future study , to be performed at higher resolution .
one such study , including magnetic fields , although not including stellar feedback , has recently been presented by @xcite .
it is important to determine the physical conditions in our clouds in order to assess their degree of realism .
one basic diagnostic is the probability density function ( pdf ) of the density field .
although it is well established that in isothermal flows the density pdf takes a lognormal form @xcite , for non - isothermal flows the expectation is in general different , with a near - power - law tail developing at high densities for flows softer than isothermal ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* although see , e.g. , wada & norman 2001 , 2007 ) , and a bimodal form arising for thermally bistable flows ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* see also the review by vzquez - semadeni 2009 ) .
figure [ fig : whole_pdfs ] shows the density pdfs for the whole simulation box of the sa runs _ top panels _ and of the la runs _ bottom panels_. the _ right panels _ show the entire density range , while the _ left panels _ show the pdf for the dense gas ( @xmath86 ) only .
the whole - range pdfs exhibit the bimodality typical of thermally bistable flows , although the high - density tail is seen to exhibit an excess over the power - law in both the cases with and without feedback .
this is probably due to the action of self - gravity @xcite .
in addition , the cases with feedback show a slight excess over the cases without it at densities @xmath87 , probably due to the formation of compressed regions by the expanding `` hii regions '' .
the high - density pdfs , being just the @xmath88 tail of the whole - range pdfs , only show more detail of the very - high density gas , up to the star - forming density of @xmath34 , but with the same shape of the curves as in the whole - range pdfs . ) .
the _ red , dotted _ lines refer to the simulations without feedback , while the _ black , solid _ lines show represent the simulations with feedback . , scaledwidth=50.0% ] in order to see the pdfs at the locations of the three main clouds we have studied ( the central cloud in the sa runs and clouds 1 and 2 in the la runs ) , we show in fig .
[ fig : cloud_pdfs ] their corresponding density pdfs .
it is noteworthy that the pdfs again show a roughly power - law shape at high densities over three to four orders of magnitude in density , in agreement with the expectation for softer - than - isothermal flows @xcite .
this is at odds with results from numerical simulations of turbulent isothermal gas in closed boxes , but then again our clouds are _ not _ isothermal .
instead , they are characterized by an effective polytropic equation of state @xmath89 , with @xmath90 varying from 0.8 to nearly unity for @xmath91 , as can be seen in fig .
[ fig : phase_space ] .
real interstellar molecular gas is not expected to be exactly isothermal , either @xcite . moreover ,
our clouds are immersed in a diffuse , warmer medium which , as proposed , e.g. , by @xcite and @xcite , infiltrates the clouds to some degree . thus , the density pdfs for the volumes containing our clouds also exhibit the bimodal shape characteristic of thermally bistable flows , and have low - density tails that extend down to the wnm regime , contrary to what happens in isothermal simulations . finally , we investigate the global velocity dispersion ( @xmath92 ) in the central cloud and clouds 1 and 2 .
this is shown in fig.[fig : veldisp ] .
the _ solid _ lines show the density - weighted value , while the _ dotted _ lines show the volume weighted value .
_ black _ lines denote cases with feedback , and _ red _ lines correspond to cases without it . and 10 k for @xmath93 .
the exceptions are locations recently heated by new stars .
, scaledwidth=50.0% ] for the central cloud and cloud 2 , which are the two most massive ones , the density - weighted velocity dispersion , which highlights the dense gas , _ decreases _ upon the inclusion of feedback . without feedback , in the central cloud
, it reaches very high values at the end of the run ( @xmath94 , corresponding to mach numbers @xmath95 - 75 ) , which in fact are significantly larger than typical values for cloud complexes of comparable mass ( @xmath96 ; e.g. , * ? ? ?
* ; * ? ? ?
* ) . instead , in the run including feedback , @xmath92 reaches values @xmath97-@xmath98 ( rms mach number , @xmath99 ) , in much better agreement with typically observed values . on the other hand , the volume - weighted velocity dispersion , which highlights the less dense gas ,
is seen in fig .
[ fig : veldisp ] to _ increase _ upon the inclusion of feedback . in clouds 1 and 2 , which are less massive and more scattered than the central cloud
, @xmath92 does not achieve exceedingly large values in any case .
these results can be understood as a consequence of the fact that the dense gas acquires its largest velocities in the case of free - fall collapse .
however , the collapse flow is dismantled in its final ( fastest ) stages by the stellar feedback , so that it is the high - velocity dense gas that is preferentially destroyed by the feedback . on the other hand ,
this gas becomes warm , diffuse , high - velocity _ expanding _ gas , which is the one highlighted by the volume weighting .
these results again reinforce the notion that the feedback is mostly applied on the gas that is closest ( but not quite there yet ) to forming stars . in the three clouds .
the _ solid _ lines show the density - weighted value , while the _ dotted _ lines show the volume weighted value .
red _ lines correspond to cases without feedback , and _ black _ lines correspond to cases with feedback.,scaledwidth=70.0% ] once we have determinations of the velocity dispersion in the cylinders containing the clouds , it is natural to measure the virial mass ( @xmath100 ) of the clouds and compare it with their real mass @xmath101 , in order to check whether they look `` virialized '' .
we compute the virial mass through the standard formula @xmath102 ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for the real cloud mass we consider the sum of the gas ( dense + diffuse ) and stellar masses in the volume being considered .
it is important to note that , with this prescription , variations in the value of the ratio come exclusively from the estimate of the virial mass , because it _ does _ depend on the weighting used , while the actual cloud mass is independent of the method of weighting .
figure [ fig : m_to_mvir ] shows the evolution of the ratio @xmath103 for the three clouds , showing the cases with and without feedback , and using volume- and density - weighting for the calculation of @xmath92 . in all cases we observe that , at advanced stages of the evolution , when the clouds have been fully assembled ( @xmath104 myr ) , this ratio is closest to unity when feedback is included and density - weighting is used in the calculation of @xmath92 .
the cases with no feedback and density weighting are systematically lower than unity by factors @xmath105 - 10 . the cases with volume weighting are more strongly fluctuating , but it is noteworthy that the ratio often takes values close to unity anyway . from these results , we conclude that , in the case with feedback , the clouds appear to be in a pseudo - virialized state with respect to the velocity dispersion of the dense gas ( highlighted by the density weighting ) . in this state , there is an approximate force balance between self - gravity and feedback driving
. however , it differs from true virialization because of the presence of mass sources ( the accretion from the diffuse environment ) and sinks ( the consumption by sf and the destruction by stellar feedback ) in the system .
these seem to self - regulate , so as to be capable of maintaining an approximately constant cloud mass while accretion persists . on the other hand , when the velocity dispersion of the warm gas is taken into account ( still in the case with feedback )
, the clouds appear to be slightly sub - virial , suggesting that the newly formed warm gas in the hii regions is capable of escaping the cloud . in the cases without feedback
, the clouds also appear to be sub - virial even with respect to the density - weighted virial mass estimate . in this case
this appears to be a consequence of the dense gas being in free - fall collapse , and into a potential well produced not only by its own self - gravity , but also by the previously - formed stars that have fallen there too ( see * ? ? ?
of course , our simulations are not free of caveats and limitations .
two that stand out are the resolution and the nature of the feedback that we have included . concerning the resolution , in this paper we have used a relatively limited one , in order to speed up the simulations , at the expense of not fulfilling the jeans criterion proposed by @xcite .
however , as explained in sec.[sec : refinement ] , we do not consider this a problem for our study , since we have restricted it to the global properties of the clouds , rather than their detailed structure , and we have avoided addressing issues related with the fragmentation of the final stages of collapse in the clouds , such as the mass distribution of the cores and stellar products .
we plan to perform a higher - resolution study in a forthcoming paper , in which we can address these issues .
the second limitation is the nature of the stellar feedback that we have considered , which in this study has been restricted to local heating representing the ionizing radiation from massive stars , similarly to the approach used by , e.g. , @xcite , neglecting other sources such as outflows from stars of all masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we do this in part for the numerical simplicity of the approach , and in part because here we have been mainly interested in the sfe and and cloud evolution at the scale of gmcs , for which the expansion of hii regions is likely the main driver @xcite . however , the neglect of bipolar outflows may introduce a non - negligible bias in our finding that smaller - scale clumps are destroyed by the feedback , and this should be confirmed by future simulations that can better resolve these objects and include bipolar outflows .
in this paper we have presented a numerical investigation of the evolution of dense ( `` molecular '' ) clouds , starting from their formation by transonic compressions in the warm neutral medium , followed by a phase transition to the cold neutral medium , the onset of gravitational collapse , star formation , and , finally , energy feedback from massive - star ionizing radiation and the formation of expanding hii regions .
a crucial difference between the results from our simulations and other models and theories of the self - regulation of star formation and its efficiency is that the clouds in our simulations are in general accreting material at high rates from the surrounding diffuse medium , rather than having a fixed mass .
this implies that _ the material making up a cloud is constantly being renewed over time _ because , on the one hand it is accreting fresh gas , and on the other it is losing mass to sf and stellar ionization .
the star - forming regions can be considered to be not objects , but rather the _ loci _ where the gas is just `` passing through '' , from the diffuse - gas state to the `` star '' state , similarly to the nature of a flame , which is the locus of the gas undergoing combustion in a candle , with fresh air entering at its base , burning while it transits through the flame , and the exhaust gases leaving at the top .
witin this framework , we have found that the sfe in the clouds is readily decreased by feedback to levels consistent with observational determinations _ at all times during the clouds evolution _ up to the maximum integration time of 40 myr that we have considered .
this is a significant improvement over our previous non - magnetic studies of the sfe in the context of cloud evolution without feedback , in which the sfe at late times is often found to be excessive ( e.g. , * ? ? ?
* ; * ? ? ?
however , we have found that the reduction factor upon the inclusion of feedback is an inverse function of the dense gas mass in the system , and of the degree of coherence of the global collapse , as illustrated in fig .
[ fig : sfe_vs_mass ] .
this result is complemented by the additional observation that low - mass clumps are fully destroyed by the feedback .
together they imply that at some point , below a certain critical collapsing mass , the appearance of a stationary state may not occur anymore .
this is in qualitative agreement with the results of @xcite , although their model does not include mass accretion onto the cloud .
we plan to present one such model in a future paper .
in general , our results may provide an explanation for the observational fact that regions of massive - star formation , which are themselves more massive than regions of low mass - star formation , appear to have higher sfes @xcite than regions forming low- and intermediate - mass stars @xcite .
essentially , our results imply that _ the more massive the region , the less effective the feedback is in reducing the sfe . _
specifically , these results indicate that the role of the feedback is not the same in all clouds , but rather depends on the initial conditions of the large - scale collapse that produces each individual region .
regions in which the amount of mass involved in the collapse overwhelms the destructive action of the feedback may reach a stationary state that appears virialized , but in which there is actually a countinuous processing of material , since on the one hand the cloud is accreting mass from the outer infalling material , and on the other hand it is losing mass through consumption by sf and evaporation by the feedback . in this case ,
termination of the sf episode probably requires the termination of the gas reservoir involved in the global collapse , and is independent of the feedback .
however , this need not be in contradiction with the low observed global efficiency of sf , if only a small fraction of a gmc s mass is involved in the collapse at any moment in time , with the rest being either supported ( by , e.g. , the magnetic field ) , or in the process of dispersal @xcite .
alternatively , the sf episode may be terminated upon the initiation of supernova events , which we have not included in the present study .
an additional consequence of the mechanisms described in this paper is that the most massive clouds appear to actually contain more mass when feedback is included than when it is not .
this suggests that , first , unimpeded sf is more efficient at removing mass from the dense gas phase to deposit into stars , than the evaporation of dense gas by the stellar feedback , so that , when the latter inhibits sf , the accretion onto the cloud accumulates larger amounts of dense gas than otherwise .
second , this indicates that the deposition of the feedback energy into the clouds is accurately targeted to gas that is already on the verge of forming stars .
if this gas is a minor fraction of the total dense gas in a cloud ( i.e , sf is a highly spatially intermittent phenomenon in molecular clouds ; * ? ? ?
* ) , then its destruction is a highly efficient way to reduce the sfe without destroying a large fraction of the cloud .
finally , we have also investigated the density pdf in the volumes containing the clouds in our simulations , finding that in general they retain the bimodal shape characteristic of thermally bistable flows .
the pdf of the dense gas exclusively shows no turnover at low densities , indicating that low densities are most abundant .
the high - density tails of the pdfs have power - law shapes , as expected for softer - than - isothermal flows , although this last result may be biased by our usage of a single cooling law appropriate for atomic gas , and extrapolating it to molecular - gas densities , rather than using a specific molecular cooling law .
we expect to address this shortcoming in future works .
we are grateful to a. kravtsov for providing us with the numerical code , and to n. gnedin for providing us with the useful analysis and graphics package ifrit .
the calculations were performed on the 8-core server at crya - unam , acquired through unam - papiit grant in112806 to p.c .
this work has received partial financial support from conacyt grants u47366-f to e.v .- s . and j50402-f to g.c.g . , and unam - papiit in106809 to g.c.g . | we numerically investigate the effect of feedback from the ionizing radiation heating from massive stars on the evolution of giant molecular clouds ( gmcs ) and their star formation efficiency ( sfe ) , which we treat as a time - dependent quantity .
we consider the gmcs evolution from their formation by colliding warm neutral medium ( wnm ) streams up to advanced star - forming stages .
we find , in agreement with our previous studies , that the star - forming regions ( `` clouds '' ) within the gmcs are invariably formed by gravitational contraction , so that their internal non - thermal motions must contain a significant component of global convergence , which does not oppose gravity .
after an initial period of contraction , the collapsing clouds begin forming stars , whose feedback evaporates part of the clouds mass , opposing the continuing accretion from the infalling gas .
the competition of accretion against dense gas consumption by star formation ( sf ) and evaporation by the feedback , regulates the clouds mass and energy balance , as well as their sfe .
we find that , in the presence of feedback , the clouds attain levels of the sfe that are consistent _ at all times _ with observational determinations for regions of comparable sf rates ( sfrs ) .
however , we observe that the dense gas mass is _ larger _ in general in the presence of feedback , and that the total ( dense gas + stars ) is nearly insensitive to the presence of feedback , suggesting that the total mass is determined by the accretion , while the feedback inhibits mainly the conversion of dense gas to stars , because it acts directly to reheat and disperse the gas that is directly on its way to forming stars .
we find that the factor by which the sfe is reduced upon the inclusion of feedback is a decreasing function of the cloud s mass , for clouds of size @xmath0 pc .
this naturally explains the larger observed sfes of massive - star forming regions .
we also find that the clouds may attain a pseudo - virialized state , with a value of the virial mass very similar to the actual cloud mass .
however , this state differs from true virialization in that the clouds are the center of a large - scale collapse , continuously accreting mass , rather than being equilibrium entities .
finally , we also calculate the density probability density functions of the clouds , finding that they in general exhibit the bimodal shape characteristic of thermally bistable flows , rather than a lognormal form , which is characteristic of isothermal flows .
this supports suggestions that low density , atomic gas pervades molecular clouds .
we conclude that the general state of star - forming regions is likely to be one of gravitational collapse , although most of the mass in a gmc may be not participating of the instantaneous star - forming activity , as recently suggested by @xcite ; that is , that sf is a spatially and temporally intermittent phenomenon , with strong , localized bursts interspersed within much more quiescent gas .
[ firstpage ] |
* gaas membrane fabrication . *
the gaas membrane is fabricated from a gaas / algaas heterostructure wafer comprised of a ( 100)-oriented gaas substrate ( 350 @xmath0 m ) , an al@xmath49ga@xmath50as etch stop layer ( 1 @xmath0 m ) , and a gaas capping layer ( 160 nm ) as shown in fig .
the backside of the wafer is patterned by photolithography and a selective wet etching process using a citric - acid - based solution is employed to remove the gaas substrate in the area selected by the photolithographic pattern .
finally , another selective wet etching process using hydrofluoric acid is used to remove the sacrificial layer , thus resulting in a 160-nm - thick suspended gaas membrane as shown in figs .
1b and c. * ringdown measurements . * a ti : sapphire laser operating in the range from 810 nm to 884 nm is mode - matched to the cavity inside a vacuum chamber and is used for both cooling and inducing mechanical oscillations by modulating the intensity at frequency @xmath51 around a specific offset level .
the cavity length ( @xmath52 29 mm ) can be stabilized to a variable position on the cavity resonant slope .
the rf signal from the two - segment photodiode and a reference signal at frequency @xmath51 from the signal generator are fed into a 2-phase lock - in amplifier as shown in fig .
the produced mixed - down signals @xmath53 and @xmath54 correspond to the @xmath55 and @xmath56 components of the membrane displacement within the bandwidth of @xmath57 500 hz centred around @xmath51 , and their squared sum corresponds to @xmath58 . for the ringdown measurements
a digital oscilloscope is used to track the signal @xmath58 triggered by a ttl signal , by which the modulation of the cavity field is also turned off .
the ringdown time of the mechanical oscillations is measured starting from the moment when the modulation is turned off while keeping the offset power level of the cavity input for cooling
. * cold damping .
* in the cold damping scenario @xcite the equation of motion for the membrane displacement in the fourier space is modified into @xmath59 due to the photo - induced force @xmath22 , where @xmath60 is the thermal langevin force and @xmath61 is the motional mass of the membrane @xcite . here
the modified resonant frequency @xmath41 and the effective damping rate @xmath20 can be expressed as @xmath62 and @xmath63 , respectively .
@xmath21 is the spatial derivative of @xmath22 introduced by the cavity and @xmath42 is the relevant time lag in the physical mechanism responsible for these modifications @xcite .
the effective temperature can then be defined as @xmath12 @xcite .
* time - domain and frequency - domain measurements . * since the mechanical oscillations arise merely out of the thermal langevin force @xmath60 for the frequency - domain measurements the resultant signal - to - noise ratio is generally worse than the ringdown measurements , for which the intensity - modulated cavity field can be used to amplify the oscillations . note also that sub - hertz mechanical resonances are measured with a spectrum analyzer with the resolution bandwidth of 1 hz , which may contribute to the uncertainty of the determination of the area and certainly affects the width and height of the resonances . because of these deficiencies of the frequency - domain measurements , the majority of our measurements are done in the time - domain method , i.e. , the ringdown . 99 kippenberg , t. j. & vahala , k. j. cavity optomechanics : back - action at the mesoscale .
_ science _ * 321 , * 11721176 ( 2008 ) .
favero , i. & karrai , k. optomechanics of deformable optical cavities .
_ nature photon . _ * 3 , * 201205 ( 2009 ) .
marquardt , f. & girvin , s. m. optomechanics .
_ physics _ * 2 , * 40 ( 2009 ) .
hohberger metzger , c. & karrai , k. cavity cooling of a microlever .
_ nature _ * 432 , * 10021005 ( 2004 ) .
gigan , s. _ et al .
_ self - cooling of a micromirror by radiation - pressure .
_ nature _ * 444 , * 6770 ( 2006 ) .
arcizet , o. , cohadon , p. f. , briant , t. , pinard , m. & heidmann , a. radiation pressure cooling and optomechanical instability of a micromirror .
_ nature _ * 444 , * 7174 ( 2006 ) .
schliesser , a. , delhaye , p. , nooshi , n. , vahala , k. j. & kippenberg , t. j. radiation pressure cooling of a micromechanical oscillator using dynamical backaction .
lett . _ * 97 , * 243905 ( 2006 ) .
metzger , c. , favero , i. , ortlieb , a. & karrai , k. optical self - cooling of a deformable fabry - perot cavity in the classical limit .
b _ * 78 * , 035309 ( 2008 ) .
grblacher , s. _ et al .
_ demonstration of an ultracold micro - optomechanical oscillator in a cryogenic cavity . _ nature phys . _ * 5 , * 485488 ( 2009 ) .
thompson , j. d. _ et al .
_ strong dispersive coupling of a high - finesse cavity to a micromechanical membrane
. _ nature _ * 452 , * 7275 ( 2008 ) .
wilson , d. j. , regal , c. a. , papp , s. b. & kimble , h. j. cavity optomechanics with stoichiometric sin films .
lett . _ * 103 * , 207204 ( 2009 ) .
hennessy , k. _ et al .
_ quantum nature of a strongly coupled single quantum dot - cavity system .
_ nature _ * 445 , * 896899 ( 2007 ) .
lund - hansen , t. _ et al .
_ experimental realization of highly efficient broadband coupling of single quantum dots to a photonic crystal waveguide .
* 101 , * 113903 ( 2008 ) .
ohno , h. making nonmagnetic semiconductors ferromagnetic .
_ science _ * 281 , * 951956 ( 1998 ) .
kroutvar , m. _ et al .
_ optically programmable electron spin memory using semiconductor quantum dots
. _ nature _ * 432 , * 8184 ( 2004 ) .
gerardot , b. d. _ et al .
_ optical pumping of a single hole spin in a quantum dot .
_ nature _ * 451 , * 441444 ( 2008 ) .
brunner , d. _ et al .
_ a coherent single - hole spin in a semiconductor
. _ science _ * 325 , * 7072 ( 2009 ) .
wilson - rae , i. , zoller , p. & imamoglu , a. laser cooling of a nanomechanical resonator mode to its quantum ground state .
_ * 92 , * 075507 ( 2004 ) .
sheik - bahae , m. & epstein , r. i. optical refrigeration .
_ nature photon . _ * 1 , * 693699 ( 2007 ) .
okamoto , h. , ito , d. , onomitsu , k. , sogawa , t. & yamaguchi , h. controlling quality factor in micromechanical resonators by carrier excitation .
express _ * 2 , * 035001 ( 2009 ) .
cole , g. d. et al .
megahertz monocrystalline optomechanical resonators with minimal dissipation . _ in proc .
ieee micro electro mechanical syst . , _ 847850 ( 2010 ) .
23rd int .
hjort , k. , sderkvist , j. & schweitz , j. - .
gallium arsenide as a mechanical material .
_ j. micromech .
microeng . _ * 4 , * 113 ( 1994 ) .
thomsen , c. , grahn , h. t. , maris , h. j. & tauc , j. surface generation and detection of phonon by picosecond light pulses .
b _ * 34 , * 41294138 ( 1986 ) .
matsuda , o. , tachizaki , t. , fukui , t. , baumberg , j. j. & wright , o. b. acoustic phonon generation and detection in gaas / al@xmath64ga@xmath65as quantum wells with picosecond laser pulses .
b _ * 71 , * 115330 ( 2005 ) . rudin , s. , reinecke , t. l. & segall , b. temperature - dependent exciton linewidths in semiconductors .
b _ * 42 , * 1121811231 ( 1990 ) . halliday , d. p. , eggleston , j. m. , lee , k. y. , frost , j. e. f. & beaumont , s. p. optical properties of ultrathin 50 nm gaas membranes .
_ solid state comm . _ * 96 , * 359365 ( 1995 ) .
nahory , r. e. oscillatory photoconductivity of epitaxial gaas .
_ * 178 , * 12931295 ( 1969 ) .
t hooft , g. w. , van der poel , w. a. j. a. , molenkamp , l. w. & foxon , c. t. giant oscillator strength of free excitons in gaas .
b _ * 35 , * 82818284 ( 1987 ) .
* figure 1| experimental setup and structure of the fabricated gaas membrane . * * a * , sketch of the experimental setup . a standard dielectric concave mirror with the reflectivity of 85% and a fabricated 160-nm - thick gaas membrane ( see methods ) with the reflectivity of 75% form a hemispherical cavity with the measured finesse of 10 in the range of the laser wavelength from 810 nm to 884 nm .
the cavity is placed in a vacuum chamber kept at 10@xmath66 pa .
the cavity length ( @xmath52 29 mm ) can be varied with a piezoelectric transducer ( pzt ) attached to the end mirror .
a ti : sapphire laser is used as a cavity input for both cooling and inducing mechanical oscillations ( the beam spot size at the membrane is about 80 @xmath0 m in radius ) by modulating the intensity with a signal generator and an acousto - optic modulator ( aom ) . a diode laser ( 975 nm )
is used to probe the membrane oscillations via a beam deflection method with a two - segment photodiode . for ringdown measurements a 2-phase lock - in amplifier
is used to produce a signal @xmath58 and a digital oscilloscope is used to track the signal ( see methods ) . *
b * , cross section of the fabricated 160-nm - thick suspended gaas membrane ( see methods ) . * c * , dimensions of the membrane .
* figure 2| ringdown results . * * a * , * b * , images of mode shapes for the fundamental mechanical mode ( the ( 1,1)-mode ) and the ( 3,1)-mode , respectively . *
c * , * d * , ringdown results with the cavity field at the wavelength of 810 nm for the ( 1,1)-mode and the ( 3,1)-mode , respectively . in * b
* and * d * red and blue diamonds respectively represent the measured damping rates and the frequencies used to induce the mechanical oscillations .
each point is the average of 5 identical measurements and the error bars correspond to one standard deviation .
lines are produced by the least - square linear fits .
* figure 3| evidence of cooling . * * a * , power spectra of the mechanical resonances ( the ( 1,1)-mode ) for different cavity input powers ( the cavity field at the wavelength of 810 nm ) . *
b * , cooling factors obtained from the areas ( purple circles ) and the widths ( orange circles ) of the resonances .
the cooling factors obtained from the ringdown measurements ( shown in fig .
2b ) are also displayed with the error bars ( red diamonds ) .
each line is produced by the least - square linear fit . for fitting the cooling factors obtained from the widths ( orange circles ) the first 4 points
are not used since they are limited by the resolution bandwidth of the spectrum analyzer .
* c * , dependence of the cooling factor on the cavity detuning for the cavity input power of 30 @xmath0w .
the cavity detuning is normalized to the cavity bandwidth @xmath67 .
red solid line indicates @xmath21 ( see methods ) . in * b * and * c * black dotted horizontal lines represent a cooling factor equal to one , i.e. , no - cooling .
* figure 4| dependence of the cooling factor on the photon energy of the cavity field . *
each point corresponds to the cooling factor for the ( 1,1)-mode with a given intracavity photon number ( equal to the number of 1.53-ev ( 810-nm ) photons inside the cavity for the input power of 50 @xmath0w ) .
black dotted vertical line indicates the gaas band gap energy @xmath26 = 1.43 ev ( 865 nm ) , black dotted horizontal line represents a cooling factor equal to one , i.e. , no - cooling .
conceptual illustrations of a electron - hole pair generation by a photon of the energy of @xmath68 ( left ) and @xmath69 ( right ) are shown on the bottom with the band structure .
we thank j. appel , a. grodecka - grad , j. h. mller and d. j. wilson for discussions , s. schmid for his technical support .
this work was supported by the japan science and technology agency ( jst ) , the japan society for the promotion of science ( jsps ) , the eu project q - essence , and the danish council for independent research ( technology and production science and natural science ) .
k.u . , b.m.n . and e.p .
designed the experiment . k.u .
, a.n . and t.b
. worked on data collection and analysis .
j.l . and s.s .
fabricated the gaas membranes .
p.l . and e.p .
planned and supervised the study .
, s.s . and e.p .
wrote the manuscript .
all authors discussed the results and commented on the manuscript . | optical cavity cooling of mechanical resonators has recently become a research frontier @xcite .
the cooling has been realized with a metal - coated silicon microlever via photo - thermal force @xcite and subsequently with dielectric objects via radiation pressure @xcite .
here we report cavity cooling with a crystalline semiconductor membrane via a new mechanism , in which the cooling force arises from the interaction between the photo - induced electron - hole pairs and the mechanical modes through the deformation potential coupling .
the optoelectronic mechanism is so efficient as to cool a mode down to 4 k from room temperature with just 50 @xmath0w of light and a cavity with a finesse of 10 consisting of a standard mirror and the sub - wavelength - thick semiconductor membrane itself .
the laser - cooled narrow - band phonon bath realized with semiconductor mechanical resonators may open up a new avenue for photonics and spintronics devices .
the first experimental demonstration of the cavity back - action cooling utilized the photo - thermal ( bolometric ) force to cool a gold - coated silicon microlever @xcite .
the force results from the differential thermal expansion between the silicon lever and the gold film and the time lag is provided by the time needed to diffuse the thermal energy along the lever @xcite .
another cooling mechanism based on the radiation pressure force with dielectric objects @xcite has recently become the mainstream approach with a potentail to reach the oscillator ground state @xcite . with this approach
the vibrational mode temperature reduction by up to a factor of 4,000 has been demonstrated @xcite .
cooling down to the ground state via radiation pressure requires a high - finesse cavity and its experimental realization has been hindered by the difficulty to achieve good mechanical qualities and low - loss optical properties on a micro- or nanometer scale simultaneously . a way to mitigate this dilemma by separating optical and mechanical functionalities in the so - called membrane - in - the - middle approach
has been demonstrated @xcite . applying the cavity cooling method to semiconductors
offers interesting prospects . in optoelectronics semiconductor devices with engineered electronic and photonic band - structures have enabled , for example , strongly enhanced light - matter interactions for quantum photonics applications @xcite .
semiconductors are also important in spintronics since they can be made ferromagnetic by doping with magnetic ions @xcite and embrace localized long - lived electron and hole spins @xcite . studying the interaction between the laser - cooled mechanical modes and these rich internal degrees of freedom of semiconductors may lead to a variety of new device conceptions in photonics and spintronics .
a theoretical proposal has been put forward to take advantage of the strong deformation potential coupling in semiconductors for cooling mechanical resonators with an embedded quantum dot instead of using a cavity @xcite .
there is also the so - called optical refrigeration scheme @xcite , which has been predicted to work efficiently for semiconductors and to reduce the entire lattice temperature down to 10 k , though no net cooling has yet been observed . driven by these attractive features of semiconductors
we have manufactured a @xmath1-nm - thick crystalline gallium arsenide ( gaas ) membrane ( see methods ) and investigated its potential for cavity optomechanics .
we have found an extraordinarily large optomechanical coupling and photo - induced mechanical damping in a simple setting where a standard mirror and the membrane form a cavity with a finesse of only 10 .
here we provide evidence of the first experimental realization of cavity cooling of mechanical modes using internal electronic degrees of freedom of semiconductors .
the experimental setup and the structure of the fabricated gaas membrane are summarized in fig . 1 .
a 1-mw diode laser ( 975 nm ) is used to probe the membrane oscillations via a beam deflection method with a two - segment photodiode . by feeding the rf signal from the photodiode to a spectrum analyzer
the mechanical resonant frequencies can be identified .
the frequency of the fundamental mechanical mode ( the ( 1,1)-mode ) is 23.4 khz ( fig .
2a ) .
the other resonant frequencies observed conform within roughly 10% to a simple model of the membrane being a taut rectangular drumhead and the mode shapes are identified by using a laser doppler vibrometer with a similar membrane .
a ti : sapphire laser providing the opto - mechanical coupling is mode - matched to the cavity .
the ringdown of the mechanical modes is observed by inducing the oscillations with an intensity - modulated cavity field ( see methods ) .
figure 2b shows the results of the mechanical ringdown indicating that the damping rate is strongly enhanced by the cavity field at the wavelength of 810 nm and is linearly proportional to the cavity input power offset as is expected from the damping model ( see methods ) .
also shown is the frequency shift as a result of the cavity field .
note that the damping of the mechanical oscillations can only be observed when the cavity resonant frequency @xmath2 is detuned to the red from the laser frequency @xmath3 , i.e. , @xmath4 , as opposed to the case of the cavity cooling via the radiation pressure @xcite .
optomechanical instabilities ( mechanical parametric oscillations ) @xcite set in when @xmath5 ( even for the cavity input power as low as 5 @xmath0w ) or when the cavity input power exceeds 50 @xmath0w ( regardless of the cavity detuning ) preventing us from using these conditions for the ringdown measurements .
the instabilities with such a small photon number imply a substantial optomechanical coupling . to assess the mechanical properties of the membrane the intrinsic mechanical damping rate @xmath6 and the quality factor @xmath7
are inferred by extrapolating the ringdown data down to zero cavity field ( where @xmath8 is the extrapolated bare resonant frequency ) . a markedly high quality factor @xmath9
is found for the ( 3,1)-mode at the frequency of 59.5 khz ( fig . 2c and 2d ) .
the product @xmath10 for this particular mode reaches @xmath11 hz at room temperature . for gaas resonators such high @xmath10 products
have been only observed to - date at a cryogenic temperature @xcite ; although two orders of magnitude higher value at room temperature has been reported for stressed sin nanomembranes @xcite .
this observation strongly encourages ongoing efforts to use crystalline semiconductors for nano- and micro - electromechanics ( nems and mems ) @xcite and optomechanics @xcite .
if the damping is not accompanied by any extra fluctuations , i.e. , is the cold damping @xcite , the enhanced damping also implies cooling .
the effective temperature can then be defined as @xmath12 ( see methods ) .
the ratio of the damping rate at 50 @xmath0w ( just below the instability - onset power ) to the extrapolated intrinsic one thus gives the maximum cooling factor @xmath13 in our setting . from fig .
2c the cooling factor for the ( 1,1)-mode is found to be about 11 .
for the ( 3,1)-mode , holding the highest @xmath14 , the cooling factor is about 75 as shown in fig .
2d , meaning that the effective temperature of the mode is reduced down to as low as 4 k from room temperature . to prove that the observed damping really means cooling frequency - domain measurements have been also performed .
figure 3a shows the power spectra of the mechanical resonances ( the ( 1,1)-mode ) for different cavity input powers .
the slight discrepancy between the resonant frequencies of ringdown measurements in fig . 2c and those in the spectrum in fig .
3a is attributed to day - by - day temperature fluctuation ( about @xmath15 3@xmath16c ) .
the area of the resonance , which is proportional to the effective temperature @xmath17 @xcite , clearly becomes smaller as the cavity input power increases .
the original area with no cavity light divided by the reduced area gives the ratio @xmath18 , i.e. , the cooling factor .
these ratios are displayed in fig .
3b for different cavity inputs ( purple circles ) .
the width of the resonance , on the other hand , reflects the damping rate since it is equal to @xmath19 .
the cooling factors obtained from the width measurements @xmath13 are shown in fig . 3b ( orange circles ) along with the corresponding cooling factors obtained from the ringdown measurements ( red diamonds ) .
the three methods agree with each other within a factor of 3 ( see methods ) .
note that the similar discrepancy between area and width measurements has been reported @xcite .
further evidence of the cavity cooling is provided by the dependence of the cooling factor on the cavity detuning ( fig .
3c ) .
in the cold damping scenario the effective damping rate @xmath20 is proportional to @xmath21 ( see methods ) . here
the slope of the cavity resonance corresponds to @xmath21 since @xmath22 is presumably proportional to the intra - cavity photon number . the red solid line in fig .
3c indicates the slope of the cavity resonance , i.e. , @xmath21 .
although the data from the ringdown measurements differ in detail from the expected behavior , it is evident that the enhancement of damping reflects @xmath21 .
we attribute the observed very efficient cooling to the effect of above - band - gap photons .
photo - induced electrons in the conduction band ( holes in the valence band ) leads to an electronic stress in the lattice through the deformation potential coupling @xcite , because the conduction band electrons bind the lattice weaker than the valence band electrons .
growth of the cavity field due to the membrane motion leads to the photo - induced force on the membrane in the direction of motion , which is opposite to the case of radiation pressure .
this explains why the cavity detuning corresponding to cooling in the present case is opposite to that for radiation pressure cooling .
the electronic stress @xmath23 per single electron - hole pair generated by a photon with the energy @xmath24 1.53 ev ( 810 nm ) can be 100 times larger than the corresponding thermal stress @xmath25 , where the band gap of gaas at room temperature is @xmath26 = 1.43 ev ( 865 nm ) @xcite . here
, @xmath27 is the bulk modulus , @xmath28 is the hydrostatic deformation potential , and @xmath29 and @xmath30 denote the thermal expansion coefficient and the heat capacity , respectively @xcite .
there is also the piezoelectric coupling in gaas , but compared to the deformation potential coupling its contribution to the cooling is negligible @xcite .
the presented optoelectronic cooling mechanism is supported by the observed dependence of the cooling factor on the photon energy ( wavelength ) of the cavity field in the range from @xmath24 1.53 ev ( 810 nm ) down to 1.40 ev ( 884 nm ) across the band gap @xmath26 .
the results are displayed in fig .
4 .
firstly , the cooling factor is essentially constant from @xmath31 to @xmath32 .
this observation rules out the possibility of the photo - thermal force being responsible for the cooling since it explicitly depends on @xmath33 .
the fact that the expected stress ratio @xmath34 is , as mentioned above , around 0.01 for @xmath24 1.53 ev ( 810 nm ) and it drops further as @xmath35 approaches @xmath26 also suggests that the contribution of @xmath36 to the cooling is marginal .
secondly , the cooling gradually reduces when @xmath37 and ceases to effect when @xmath38 ( @xmath39 is the boltzmann constant and @xmath40 26 mev at room temperature ) , supporting the assertion that the carrier excitation is responsible for the cooling .
last , but not least , we find a striking similarity between the trajectory of the cooling factor in fig . 4 and the photoconductivity response of a similar gaas membrane with an embedded field - effect transistor @xcite . the slight decline of the cooling factor starting around 1.49 ev could be due to the interaction of the photo - excited carriers with the optical phonons as seen in the photoconductivity measurement @xcite .
the observed dependencies of @xmath41 and @xmath20 on optical power ( fig.2 ) can be used to infer @xmath42 , the delay time between the photo - induced force and displacement @xcite , which is specific for every cooling mechanism @xcite . using equations @xmath41 and @xmath20 given in the methods section we determine @xmath43 and @xmath44 10 ns .
this fairly long delay time @xmath42 can be backed by the long ( @xmath45 ns ) radiative lifetime @xmath46 of free excitons in gaas @xcite , though more studies are required to decisively determine the delay time @xmath42 .
an outstanding issue is the onset of the instability just above 50 @xmath0w which is unexpected from the simple model that we use .
there are at least three distinctive time scales involved in the cooling mechanism ; the mechanical oscillation time @xmath47 , the cavity decay time @xmath48 , and the delay time @xmath42 introduced by the internal degrees of freedom .
this clearly calls for additional theoretical and experimental work to investigate these effects in the context of cavity optomechanics .
the demonstrated optomechanical coupling is very efficient .
indeed significant changes in the membrane dynamics are observed already with a few microwatts of the input power corresponding to just 10,000 intracavity photons for our 3-cm cavity . for a 300-@xmath0 m cavity
this number is down to 100 photons and even this number can be probably further reduced by a suitable design . using a higher finesse cavity with the membrane in the middle
, the optical power can also be considerably reduced . |
it has recently come to light that the growth of galaxies may be less erratic than previously thought .
the existence of a tight relation between the stellar mass ( @xmath6 ) and star formation rate ( sfr ) of star - forming galaxies , termed the ` main sequence ' ( ms , @xcite ) is indicative of a quasi - equilibrium being maintained between gas supply , sfr , and gas expulsion from galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
this sfr - m@xmath7 relation evolves strongly with look - back time , paralleling the global evolution of the cosmic sfr density from the present to @xmath8 @xcite .
it is becoming clear that such a decline in sfr is in response to diminishing gas reservoirs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the efficiency of star formation ( @xmath9 ) , i.e. , the efficiency at which gas is being converted to stars , is remarkably similar across a wide range of cosmic time within the global star - forming population , usually with gas mass @xmath10 inferred from the co molecular line luminosity @xmath11 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , outliers from the ms are known to exist , with very high specific sfr ( ssfr ) compared to normal ms galaxies , such as local ultra - luminous infrared galaxies ( ulirg , @xcite ) . at @xmath12 ,
these ms outliers represent @xmath13 of the star - forming population and contribute only a moderate fraction ( @xmath14 ) of the global co - moving sfr - density @xcite .
yet , they are still likely to play an important role in galaxy evolution , as a high fraction of galaxies may experience such a star - bursting event .
therefore , it is important to understand the physical conditions that can lead to such a surge in sfr .
local starbursts have a higher sfe compared to ms galaxies .
estimates of their molecular gas content from their co luminosity indicate that a given gas mass is capable of producing higher sfrs @xcite , and a bi - modal distribution of sfes has been proposed for the general ms and outlier population @xcite . while there are uncertainties with respect to the appropriate conversion of @xmath11 to h@xmath15 gas mass ( @xmath16 ) , an offset with respect to ms galaxies is already apparent when considering the observed quantities @xmath11 and total infrared luminosity ( @xmath17 ) .
the physical mechanism responsible for this enhanced sfe is not entirely clear and one can not exclude that sample selection may be responsible for the lack of galaxies with intermediate sfes . while co detections have been obtained for high - redshift submm - selected galaxies ( smgs
; e.g. , * ? ? ?
* ; * ? ? ?
* ) and star - forming radio galaxies ( sfrg ; * ? ? ?
* ) , these samples do not cleanly distinguish starburst outliers from the ms population and submm selected samples could be biased towards galaxies with lower dust temperatures @xcite .
therefore , it is imperative to study the co emission of starbursts securely elevated above the ms at their respective stellar mass and near the peak epoch of star formation ( @xmath8 ) .
@xcite achieve such a clear distinction between ms and ms - outlier galaxies using large photometric samples with multi - wavelength coverage , including the bulk of the fir emission .
our targets then result in a homogenous sample that does not have contaminants such as those inherent with other selections .
to address this issue , we report on interferometric observations of the molecular transition @xmath18co 2 - 1 and @xmath18co 3 - 2 for a sample of seven ms outliers using both the atacama large millimeter array ( alma ) and iram plateau de bure interferometer .
our sample is extracted from the 246 ms outliers ( with ssfrs @xmath5 above the mean of the ms ; figure [ fig : sample ] ) at @xmath19 identified with herschel observations over the cosmos field @xcite that effectively cover the peak fir emission . among these herschel - detected galaxies , those with photometric redshifts between 1.4 and 1.7
are observed through our subaru intensive project @xcite with the fmos near - ir multi - object spectrograph .
these spectra provide a detection of key diagnostic emission lines ( i.e. , h@xmath20 , h@xmath21 , [ nii]@xmath22 , [ oiii]@xmath235008 ) used to measure accurate spectroscopic redshifts , sfrs , and metallicities .
six targets , labeled in figure [ fig : sample ] , are extracted from this sample with an additional galaxy at a slightly higher redshift ( pacs-282 ; @xmath24 ) where the redshift comes from the zcosmos - deep program @xcite .
our confidence in these galaxies being ` outliers ' in the sfr@xmath25 plane is based on the exquisite multi - wavelength coverage of the cosmos field .
stellar masses are derived from sed - fitting using hyperz with stellar population synthesis models @xcite at the respective spectroscopic redshift .
the de - blending of detections in herschel ( or 70@xmath26 _ spitzer _ ) images relies upon @xmath27 mips 24@xmath26 priors .
sfrs are derived from l@xmath28 , an integral of the best - fit model sed @xcite from 8 - 1000
@xmath29 m using the spitzer ( 24 @xmath29 m ) , herschel pacs ( 100 and 160 @xmath29 m ) and spire ( 250 , 350 and 500 @xmath29 m ) bands , thus accounting for the obscured star formation as in @xcite .
moreover , four of the seven galaxies have 1.4 ghz radio detections at @xmath30 level @xcite and radio - based sfrs are consistent with being above the ms .
the presence of strong agn within our sample is ruled out by the lack of individual x - ray detections by @xmath31 .
even if an obscured agn were present , the contribution to l@xmath28 would be negligible @xcite . throughout this work ,
we assume @xmath32 km s@xmath4 mpc@xmath4 , @xmath33 , @xmath34 , and use a salpeter imf for sfrs and stellar masses .
llllllllllll 299&09:59:41.31&02:14:42.91&1.6483&1.6467&10.44&554@xmath35&[email protected]&10.44&590&@xmath372.4&0.52 + 325&10:00:05.47&02:19:42.61&&1.6557 & 10.29&139@xmath38&[email protected]&10.07&764&&0.32 + 819&09:59:55.55&02:15:11.70&1.4451&1.4449&10.61&783@xmath39&[email protected]&10.55&592&[email protected]&0.34 + 830&10:00:08.75&02:19:01.90&1.4631&1.4610&10.86&517@xmath40&1.18@xmath41&10.59&436&[email protected]&0.46 + 867 & 09:59:38.12&02:28:56.56&1.5656&1.5673&10.67&358@xmath42&0.46@xmath43&10.24&472&@xmath371.5&0.29 + 282&10:00:01.54&02:11:24.27&2.1869&2.1924&10.88 & 581@xmath44&[email protected]&10.44&660&@xmath373.4&0.28 + 164&10:01:30.53&01:54:12.96&1.6481&1.6489&@xmath4510.28&358@xmath46&[email protected]&10.40&894&@xmath374.8&0.52 alma observations of five galaxies were carried out in cycle 1 ( project 2012.1.00952.s ) using 25 - 30 12 m antennas and the band 3 receiver tuned to intermediate frequencies between 86.395 - 94.888 ghz with a resolution of 488 khz and spectral bandpass of 1.875 ghz .
two ` scheduling blocks ' ( sb1 and sb2 ) were defined to accommodate the requirements to detect co 2 - 1 within a single sideband for each of our targets ( sb1 : pacs-819 , 830 ; sb2 : pacs-325 , 299 and 867 ) . on -
source exposure times were 32 minutes ( 819 and 830 ) , 2.3 hours ( 299 and 325 ) and 3.8 hours ( 867 ) that met our request based on predictions of co emission from their ir luminosity and an assumed ratio l@xmath47/l@xmath48 ( 3@xmath49 lower than ms galaxies ) .
the antenna configurations had baselines up to @xmath2400 m ( @xmath2200 m ) for sb1 ( sb2 ) thus resulting in a beam size ( fwhm ) of @xmath50 ( @xmath51 ; @xmath52 for pacs 325 only ) .
the images of pacs-819 ( 830 ) have a flux sensitivity level of 0.12 ( 0.10 ) mjy beam@xmath4 over a bandwidth of 400 km s@xmath4 while the deeper observations reach 0.073 - 0.14 mjy beam@xmath4 for the same spectral window .
standard targets were used for flux ( mars , ganymede ) , bandpass , and phase calibration .
we measure the co flux by fitting the data in uv space with the gildas task ` uvfit ' available in the mapping package .
the fit was performed using a point source model or a circular gaussian for resolved sources that returns the centroid , de - convolved source size , and integrated flux .
we find good agreement between fluxes returned from gildas with those using ` imfit ' in the image plane with casa .
additionally , we resolve the emission for pacs- 819 and 830 ; therefore , a measure of the source extent is reported . to increase the sample size ,
interferometric measurements were obtained with iram / pdbi for pacs-282 ( co 3 - 2 ) and pacs-164 ( co 2 - 1 ) . for pacs-282 ( pacs-164 ) , an integration time of 6.2 ( 12.9 ) hours , using the 3 mm band in configuration d ,
was achieved that resulted in a limiting 1@xmath53 sensitivity level of 0.14 ( 0.11 ) mjy beam@xmath4 for a bandwidth of 240 ( 180 ) mhz .
co fluxes are measured with gildas . in table
[ tab : sample ] , we list the source properties and co measurements .
we detect co emission in all seven targeted galaxies with integrated flux densities ( @xmath54 ) ranging from 0.28 to 1.18 jy km s@xmath4 .
all but one have a high level of significance ( @xmath30 ) and those resulting from alma observations are above the 8@xmath53 level with the exception of pacs-325 ( @xmath55 ) . in figure
[ fig : maps ] , we display the optical hst / acs f814w image cutouts @xcite with co and 3.6@xmath29 m emission overlaid as contours . using a large beam for the majority of the sample ,
the co emission is essentially unresolved with the exception of two observations taken at higher resolution ( @xmath56 ) .
pacs-819 has half of the emission coming from a region of [email protected] kpc , while pacs-830 is more extended ( [email protected] kpc ) .
this may suggest that there is diversity in the size distribution of molecular gas at high redshift ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , dissimilar to local starbursts ( e.g. , * ? ? ? * ) . upon close inspection of the maps in figure [ fig : maps ] ,
it is evident that the centroid of the co emission is not always coincident with the brightest regions of uv emission as seen in the hst / acs image ( i.e. , pacs-164 , pacs-830 , pacs-867 ) .
this is likely evidence that a fair fraction of the star formation is obscured as supported by the improved alignment of the co emission with the peak infrared emission detected by @xmath27 in both the irac and mips channels , and clear association with highly significant radio emission at 1.4 ghz @xcite .
we note that the herschel far - ir ( and possibly the co ) emission of pacs-164 is likely the sum of two separate components seen in the irac image .
the peak of the co emission , detected by iram , is located somewhat between two ir sources , as made particularly clear by the @xmath57-band image also shown in figure [ fig : maps ] , and slightly elongated possibly indicating a contribution from both sources . the fmos spectrum that provides the redshift ( @xmath58 ) refers to the western ir component that is co - spatial with the uv - bright source seen in the hst image . with a redshift of the co emission ( @xmath59 ) very close to that of the fmos source
, this system appears to be in the early stages of a merger given the projected separation of 18.7 kpc .
the lower limit for the stellar mass of the system as reported in table 1 refers to the mass of the object observed with fmos , and encircled in figure 2 .
our primary interest is to determine whether high - redshift starbursts convert gas into stars with a higher efficiency than ms galaxies at these epochs . to begin with , we use the observed co to ir luminosity ratio @xmath60 rather than derived quantities , e.g. the gas mass , that depends on the co - to - h@xmath15 conversion factor ( @xmath16 ) , or gas mass surface density , which would further require co - size information for which we only have for 2/7 galaxies in our sample . however , we note that @xmath61sfr ( modulo @xmath62 ) , hence this luminosity ratio is a fair proxy for the gas depletion time which in turn is the inverse of the sfe . in figure
[ fig : sf_eff]@xmath63 , we plot @xmath11 as a function of @xmath64 , indicative of the obscured sfr , and include both low- and high - redshift galaxies with measurements available in the literature and compiled in @xcite . all line luminosities are converted to co ( 1 - 0 ) using values of 0.85 and 0.7 , respectively for co ( 2 - 1 ) and co ( 3 - 2 ) ( daddi et al .
different excitation corrections than those applied here would not impact our results since changes would either be insignificant ( 5 - 10@xmath65 if co ( 2 - 1 ) transitions were generally more excited than assumed here ) or affect only one of our sources ( pacs 282 , for which applying the co(3 - 2)/co(1 - 0 ) ratio of @xcite would result in a 35@xmath65 difference ) . all galaxies in our sample have @xmath60 below the well - defined correlation for ms galaxies .
these observations indicate the existence of an offset in the @xmath60 ratio for high - redshift starbursts as found by @xcite for local ulirgs . in figure
[ fig : sf_eff]@xmath66 , we plot these quantities , normalized to the mean value of galaxies on the ms . in addition , on the abscissa we replace @xmath17 with its implied specific sfr ( ssfr = sfr/@xmath6 ) , also normalized to the ms value .
we measure a median value of @xmath67 for our sample .
this represents an offset ( @xmath68 ) from the ms that is smaller than in local samples of starbursts ( @xmath69 ; dashed line in fig .
[ fig : sf_eff]_a _ ) . along these lines ,
the starbursts in our sample have higher @xmath70 than expected by @xmath71 dex ( given their excess ssfr relative to ms galaxies ) based on the empirical model of @xcite , shown in figure [ fig : sf_eff]@xmath66 .
these results may be indicative of a more continuous range in sfe , following the trend for ms galaxies ( fig .
[ fig : sf_eff]@xmath66 ) and opposed to the notion of a second mode of star - formation operating at higher efficiency distinct from ms galaxies @xcite .
however , any statement on differences in sfe requires a factor @xmath16 as further addressed below . even with continuity in these parameters , there could still exist an underlying bi - modal physical framework for star formation as demonstrated with empirical models shown in figure [ fig : sf_eff]@xmath66 and fully explained in @xcite . with pacs-819 and pacs-830
having marginally resolved co emission , we can estimate the total gas mass using the dynamical mass method ( e.g. , * ? ? ?
the dynamical mass ( @xmath72 ) within the half - light radius ( @xmath73 ) is given for the spherically symmetric case by : @xmath74 where @xmath53= @xmath75/2.35 is the velocity dispersion , @xmath75 is the line width - fwhm , and @xmath76 is the gravitational constant .
the gas mass is then derived by subtracting from the dynamical mass half of the stellar mass and a dark matter component ( @xmath77 @xmath78 ; @xcite ) as follows .
@xmath79 with this method , the gas mass for pacs-819 ( pacs-830 ) is estimated to be @xmath80 m@xmath3 , based on their half - light radii , respectively @xmath81 ( @xmath82 ) and velocity dispersion @xmath53 , respectively @xmath83 km s@xmath4 ( @xmath84 km s@xmath4 ) , that corresponds to a gas fraction @xmath85 ) of 0.34 ( 0.46 ) . based on these values ,
we estimate @xmath86 for pacs-819 and @xmath87 for pacs-830 , both not far from those found for the high - redshift submillimeter galaxies in the proto - cluster gn20 @xcite and consistent with lower conversion factors for galaxies in mergers @xcite as compared to isolated disk galaxies ( see also @xcite ) .
we roughly estimate the h@xmath15 gas masses from the co line luminosity for the galaxies with unresolved co emission using the average ( @xmath88 m@xmath3 / k km s@xmath4
pc@xmath89 ) value of the two estimates of @xmath16 given above which is in very close agreement to that found for local nuclear starbursts ( @xmath90 ; * ? ? ?
based on this conversion factor and keeping in mind the uncertainty , the gas masses are in the range @xmath91 m@xmath3 . this results in gas fractions @xmath92 ) between 28 and 52% ( see table [ tab : sample ] for individual estimates ) , similar to those of ms galaxies at these redshifts @xcite and in line with cosmological galaxy formation models @xcite .
based on these results , an increase in the sfe is likely responsible for their elevation above the ms . in figure
[ fig : sf_eff]@xmath93 , we illustrate this scenario by plotting the gas depletion timescale ( @xmath94 ; units gyr ) as a function of sfr . as described above ,
the lower co luminosities at a given l@xmath28 , compared to ms galaxies , are indicative of a lower gas mass hence shorter gas depletion timescales and higher sfes , in agreement with previous studies @xcite . despite the remaining uncertainties on the assumptions used to derive h@xmath15 masses , we see evidence that starbursts in our sample have a smaller contrast ( i.e. , less extreme ) in sfe , relative to ms galaxies , as compared to local ulirgs and the strongest starbursts at high - z . as a final check , we can assess whether the values of @xmath16 derived above are appropriate for our starbursts by using the metallicity of the ism as a proxy , given that it is well established that @xmath16 is anti - correlated with metallicity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in figure
[ fig : emlines]@xmath63 , we show the mass - metallicity relation at @xmath1 @xcite based on [ nii]@xmath22/h@xmath20 , with our sample shown in red .
we find that 4 out of 6 galaxies ( having errors @xmath95 on the ratio ) have high metallicities , similar to those of local massive galaxies .
therefore , our derived values of @xmath16 are consistent with local galaxies of a similar mass and metallicity , although with the caveat that we are not certain whether the metallicity of the line - emitting gas is representative of the molecular gas producing the co emission .
we also examine whether agn photo - ionization is contributing to the high [ nii]/h@xmath20 ratios seen in some of our galaxies .
figure [ fig : emlines]@xmath66 shows the fmos - cosmos version of the bpt diagram at @xmath1 .
four galaxies with co measurements have all key diagnostic lines detected by fmos .
none of our galaxies fall in the region where strong agn contribution is expected , above the line separating agn from star - forming galaxies at @xmath1 @xcite .
while pacs-819 approaches this line , it is not detected in x - rays with @xmath31 .
nevertheless , we can not rule out the presence of a low - to - moderate luminosity agn , especially because its strong [ nii ] emission also drives this galaxy above the local mz relation . for reference , in figure
[ fig : emlines]@xmath66 , we plot the data for the larger star - forming galaxy population from fmos - cosmos ( grey data points ) .
it is worth highlighting that both pacs-830 and pacs-299 fall within the locus of the star - forming population , clearly offset towards a higher ionization state of the ism compared to local galaxies @xcite .
our observations of the molecular co gas content of seven galaxies well - above the ms at @xmath1 with alma in cycle 1 and iram / pdbi establish an offset in the @xmath96 relation for starbursts at high redshift , compared to ms galaxies , although smaller than expected from previous studies of low - z starburst outliers .
these results may be indicative of a continuous distribution in sfe at high redshift , as a function of distance from the star - forming ms , as opposed to a bi - modal distribution .
an appealing physical explanation for a decrease in the l@xmath97/l@xmath48 ratio for galaxies well - above the ms is galaxy mergers that can lead to rapid gas compression hence effectively boosting star formation resulting in shorter gas depletion timescales .
several galaxies well - above the ms are indeed in a merger phase @xcite . within the present sample , hints of merging
come from the presence of multiple uv / optical emitting regions ( e.g. , in pacs-819 , pacs-830 , pacs-867 ) , even being in a kinematically - linked system ( pacs-164 ) .
we are grateful for the support from the regional alma arcs , in particular akiko kawamura , kazuya saigo , gaelle dumas and sergio martin .
jds was supported by the alma japan research grant of naoj chile observatory , naoj - alma-0127 .
this work was supported by world premier international research center initiative ( wpi initiative ) , mext , japan .
cm and ar acknowledge support from an inaf prin 2012 grant .
this paper makes use of the following alma data : ads / jao.alma#2012.1.00952.s .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) , nsc and asiaa ( taiwan ) , and kasi ( republic of korea ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao and naoj .
arimoto , n. , sofue , y. , tsujimoto , t. 1996 , , 48 , 275 bthermin , m. , daddi , e. , magdis , g. , et al .
2015 , , 573 , a113 bothwell , m.s .
, smail , i. , chapman , s.c .
2013 , , 429 , 3047 bouch , n. , dekel , a. , genzel , r. , et al .
2010 , , 718 , 1001 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 casey , c. m. , chapman , s. c. , neri , r. et al . 2011 , , 415 , 2723 daddi , e. , dickinson , m. , morrison , g. et al .
2007 , , 670 , 156 daddi , e. , bournaud , f. , walter , f. et al .
2010a , , 713 , 686 daddi , e. , elbaz , d. , walter , f. et al .
2010b , , 714 , l118 downes , d. , & solomon , p. m. 1998 , , 507 , 615 draine , b. t. , li , a. 2007 , .. 657 , 810 elbaz , d. , daddi , e. , le borgne , d. et al .
2007 , , 468 , 33 genzel , r. , tacconi , l. j. , gracia - carpio , j. , et al .
2010 , , 407 , 2091 genzel , r. , tacconi , l. j. , combes , f. , et al .
2012 , , 746 , 69 genzel , r. , tacconi , l. j. , lutz , d. , et al .
2015 , , 800 , 20 greve , t. r. , bertoldi , f. , smail , i. et al .
2005 , , 359 , 1165 hodge , j. a. , carilli , c. l. , wlater , f. et al .
2012 , , 760 , 11 ilbert , o. , arnouts , s. , le floch et al .
2015 , , 579 , 2 ivison , r. j. , papadoppoulos , p. p. , smail , i. et al .
2011 , , 412 , 1913 kartaltepe , j. sanders , d. b. , silverman , j. d. et al .
, , 806 , l35 kashino , d. , silverman , j. , rodighiero , g. et al . , 777 , l8 kewley , l. , dopita , m. , leitherer , c. , et al .
2013 , , 774 , 100 koekemoer , a. et al .
2007 , , 172 , 196 lilly , s. j. , carollo , c. m. , pipino , a. , 2013 , , 772 , 119 lilly , s. j. , le fvre , o. , renzini , a. , et al .
2013 , , 772 , 119 madau , p. , & dickinson , m. 2014 , , 52 , 415 magdis , g. e. , daddi , e. , bthermin , m. 2012 , , 760 , 6 magnelli , b. , lutz , d. , saintonge , a. et al .
2014 , , 561 , 86 narayanan , d. , krumholz , m. , ostriker , e. c. , & hernquist , l. 2011 , , 418 , 664 narayanan , d. , bothwell , m. , & dav , r. 2012 , , 426 , 1178 noeske , k. , weiner , b. j. , faber , s. m. et al .
2007 , , 660 , l43 pozzi , f. , vignali , c. , gruppioni , c. et al .
2012 , , 423 , 1909 riechers , d. a. , hodge , j. , walter , f. , carilli , c. l. , bertoldi , f. 2011 , , 739 rodighiero , g. , daddi , e. , baronchelli , i. et al .
2011 , , 739 , l40 sanders , d. b. & mirabel , i. f. 1996 , , 34 , 749 santini , p. , maiolino , r. , magnelli , b. 2014 , , 562 , a30 sargent , m. t. , daddi , e. , bthermin , m. et al .
2014 , , 793 , 19 schinnerer , e. , sargent , m. t. , bondi , m. et al .
2010 , , 188 , 384 schruba , a. , leroy , a. k. , walter , f. et al .
2012 , , 143 , 138 scoville , n. z. , sanders , d. b. , sargent , a. i. et al .
1989 , , 345 , l25 scoville , n. z. , aussel , h. , sheth , k. et al .
2014 , , 783 , 84 silverman , j. d. , kashino , d. , arimoto , n. et al .
2015 , , 220 , 12 solomon , p. m. , downes , d. , radford , s. j. e. , & barrett , j. w. 1997 , , 478 , 144 speagle , j. s. , steinhardt , c. l. , capak , p. l. , silverman , j. d. 2014 , , 214 , 15 steidel , c. c. , rudie , g. c. , strom , a. l. et al .
2014 , , 795 , 165 swinbank , a. m. , simpson , j. m. , smail , i. et al .
2014 , , 438 , 1267 tacconi , l. , genzel , r. , neri , r. , et al .
2010 , , 463 , 781 tacconi , l. , neri , r. , genzel , r. , et al .
2013 , , 768 , 74 tan , q. , daddi , e. , magdis , g. et al .
2014 , , 569 , a98 whitaker , k. e. , van dokkum , p. g. , brammer , g. et al .
2012 , , 754 , l29 wuyts , s. , frster schreiber , n. m. , van der wel , a. et al .
2011 , , 742 , 96 zahid , h. j. , kashino , d. , silverman , j. d. , et al .
2014 , , 792 , 75 | local starbursts have a higher efficiency of converting gas into stars , as compared to typical star - forming galaxies at a given stellar mass , possibly indicative of different modes of star formation . with the peak epoch of galaxy formation occurring at @xmath0 ,
it remains to be established whether such an efficient mode of star formation is occurring at high - redshift . to address this issue
, we measure the molecular gas content of seven high - redshift ( @xmath1 ) starburst galaxies with the atacama large ( sub-)millimeter array and iram / plateau de bure interferometer .
our targets are selected from the sample of herschel far - infrared detected galaxies having star formation rates ( @xmath2300 - 800 m@xmath3 yr@xmath4 ) elevated ( @xmath5 ) above the star - forming ` main sequence ' , and included in the fmos - cosmos near - infrared spectroscopic survey of star - forming galaxies at @xmath1 with subaru .
we detect co emission in all cases at high levels of significance , indicative of high gas fractions ( @xmath230 - 50% ) .
even more compelling , we firmly establish with a clean and systematic selection that starbursts , identified as main - sequence outliers , at high redshift generally have a lower ratio of co to total infrared luminosity as compared to typical main - sequence star - forming galaxies , although with a smaller offset than expected based on past studies of local starbursts .
we put forward a hypothesis that there exists a continuous increase in star formation efficiency with elevation from the main sequence with galaxy mergers as a possible physical driver .
along with a heightened star formation efficiency , our high - redshift sample is similar in other respects to local starbursts such as being metal rich and having a higher ionization state of the interstellar medium . |
searching for possible variations in the fundamental constants of nature allows us to test whether the laws of physics vary over space and time .
there are many complementary probes that , taken together , span much of the universe s history ( see , e.g. , the review * ? ? ? * ) .
optical quasar absorption spectra provide a probe of variations in the fine - structure constant , @xmath2 , along a past - light cone centred on present - day telescopes .
numerous works employed comparison of alkali - doublets in space and in the laboratory to place limits on @xmath0-variation ; these are easily interpreted since the energy separations are simply proportional to @xmath3 .
@xcite developed a different approach : the many - multiplet method , which exploits larger sensitivities to @xmath0-variation available in heavier ions and in transitions other than alkali - doublets .
moreover , the method allows for better control of systematics because there exist some negative shifters ( with energy intervals that reduce with increasing @xmath0 ) as well as positive shifters and `` anchor lines '' ( which have very weak sensitivity to @xmath0-variation ) . in some cases both positive and negative shifters may be observed in the same ion .
additionally , the many - multiplet method sees a statistical gain because many more observed transitions may be used in the analysis . during the last several years many absorption systems ,
observed using the keck telescope in hawaii @xcite and the very large telescope in chile @xcite , have been analysed . taken together ,
these provide measurements of @xmath1 in @xmath4 absorption systems covering most of the sky . here
@xmath5 is the relative variation in @xmath0 at a particular position @xmath6 in the universe where the absorption occurs . in @xcite ,
the combined data sample is interpreted as providing evidence for variation in @xmath0 throughout the universe with an angular - dependence .
this `` dipole '' model is found to be preferred to a monopole ( constant offset ) model of the variation at the @xmath7 level .
the values of @xmath1 for each absorber ( presented in * ? ? ?
* ; * ? ? ?
* ) are modelled in @xcite(see also * ? ? ?
* ) using several variants of a formula which can be written as @xmath8 where @xmath9 is the angle between the direction of the measurement and the axis of the dipole , @xmath10 is a constant ( a `` monopole '' term ) and @xmath11 is the magnitude of the dipole term .
the model variation comes from the form of the factor @xmath11 , which in @xcite was variously * @xmath12 , i.e. eq . ( [ eq : dipolemonopole ] ) represents pure angular - dependence without any distance - dependence ; * @xmath13 , where @xmath14 is the lookback time in giga - lightyears ( glyr ) ; * @xmath15 , here @xmath16 is the redshift and @xmath17 is another fitting parameter .
the model of spatial variation that does not include a distance dependence is difficult to understand theoretically , while fitting @xmath18 simply adds a poorly - constrained parameter to the data .
therefore in this paper we test the dipole interpretation @xmath19 . in this form ( [ eq : dipolemonopole ] ) represents a gradient in the value @xmath0 throughout the universe , and @xmath20 is the distance to a quasar absorption system along that gradient . of course , @xmath21 itself is model dependent at large redshifts . in this work we use the standard @xmath22 cosmology parametrized by wmap5 @xcite to determine @xmath21 from the redshift @xmath16 : @xmath23 again , this reflects the fact that we are taking measurements along a past light cone centred on present - day earth .
an alternative approach is to use the comoving distance rather than lookback time : @xmath24 in this case the dipole model ( [ eq : dipolemonopole ] ) is modified to use .
this parameterisation makes physical sense if one imagines that any spatial @xmath0-variation is `` fixed '' to the cmb frame and follows the same scale factor , @xmath25 .
this paper is organised as follows . in section [ sec : aicc ] we develop criteria to assess the relative likelihood of different models of @xmath0-variation which we use throughout this work . after testing the normalcy of errors using a quantile - quantile plot ,
we use the modified akaike information criterion to confirm that the preferred model of cosmological @xmath0-variation is a dipole model with a monopole term . in section [ sec : dipole_significance ] we perform some additional statistical tests to determine the significance of the dipole : an @xmath26-test and a modified error - ellipsoid method , described in detail in appendix [ sec : eem ] . to test the robustness of the data we perform biased and unbiased clippings of the data . in section
[ sec : dipole_robustness ] it is shown that _ at least _
40% of the quasar absorption systems would need to be removed from the sample to reduce the significance of the dipole to @xmath27 .
the monopole term of our preferred model of @xmath0-variation is somewhat difficult to reconcile with the requirement that @xmath28 should be zero locally , therefore in section [ sec : magnesium ] we test a hypothesis that changes in the relative isotope abundances of magnesium could lead to the observed monopole term .
we find that a relatively small increase in the abundance of @xmath29 mg relative to @xmath30 mg in the absorbers could account for the monopole term .
finally , in section [ sec : wall ] we test whether the data supports a different model : that a `` domain wall '' separates regions of the universe with different values of @xmath0 ( as proposed by * ? ? ?
finding a best fit for this model requires a genetic algorithm , presented in appendix [ sec : genetic ] .
the data used in this paper are taken from @xcite ( keck data ) and @xcite ( vlt ) , kindly supplied by the j. a. king in a usable text format including the location of each absorber ( redshift and direction ) and its measured value of @xmath1 . in total
there are 293 points in our data file including 140 from keck and 153 from vlt of which seven absorbers are seen in both the keck and vlt samples .
the data is described fully in @xcite , here we just note some points of particular interest to the analysis presented in this paper .
both samples account for unknown sources of scatter in the data by including extra systematic errors , @xmath31 , that are added in quadrature with the underlying statistical error .
we will take each sample in turn .
the keck data is taken from the ` fiducial sample ' of @xcite .
this includes several samples of keck / hires spectra observed independently by different groups ( see * ? ? ?
* ; * ? ? ?
* ) for details ) .
the combined data is divided into high-@xmath16 ( @xmath32 ) and low-@xmath16 subsamples , and the high-@xmath16 subsample is further divided into ` low - contrast ' and ` high - contrast ' , defined as systems where both strong and weak lines are fitted .
the 27 systems that constitute the high - contrast sample show large scatter in @xmath1 values , which is attributed to weak components not being fitted when they re near the high optical depth edges of the strong transitions profile . to this high - contrast sample
is added an additional error @xmath33 which serves to reduce @xmath34 per degree of freedom @xmath35 to unity .
the low-@xmath16 and low - contrast high-@xmath16 keck samples do not require additional @xmath31 .
the vlt / uves data presented in @xcite also shows too much scatter in @xmath1 . quantifying this scatter depends on the underlying model assumptions , for example if one assumes that the dipole model is correct ( i.e. that there is a physical dipole in @xmath0 that is being observed ) and the data supports this , then @xmath31 should be smaller than it would be for a monopole model .
the data we use includes @xmath36 .
this is a conservative ( large ) value which will tend to reduce the significance of the dipole relative to a monopole model . in practice
all models give values of @xmath31 in the same vicinity ( see table 2 . of * ? ? ? * ) and the results of this paper are relatively insensitive to the exact @xmath31 used .
in this work we use @xmath34 as a likelihood measure for the various models . for measured values @xmath37 and a model which takes values @xmath38 at the @xmath39th quasar , this is defined as the sum of squared residuals @xmath40 : @xmath41 where @xmath42 is the observational uncertainty .
@xmath34 corresponds to the negative of the log - likelihood function for the normally - distributed random variable @xmath43 ; smaller values of @xmath34 correspond to better model fits .
the normalcy of the residuals is an implicit assumption in using a @xmath34 statistic .
the validity of such an assumption may be ascertained by way of a quantile - quantile , or q - q , plot , which compares the residuals of a data set against the quantiles of a given distribution , or another sample .
in addition to determining whether a data sample is consistent with a given distribution , a q - q plot is a useful graphical method of comparing the qualitative features , such as skew , of samples of data against statistical distributions .
we compare the residuals from the best fit dipole direction ( light - travel distance ) against a normal distribution in figure [ fig : qqplot ] .
if our residuals are normal , then we expect the plot to lie approximately across the diagonal .
our comparison , figure [ fig : qqplot ] , suggests that the residuals are near - consistent with a normal distribution , though it should be observed that there is some ` arching ' in the curve , with the curve being below the diagonal near the tails and above near the centre .
this indicates a negative skewness , that is , a long left - hand tail , in the residual distribution .
however , in light of the relatively small size of our sample , the slight skew we observe does not necessarily indicate that the residuals are not normal .
we conclude that @xmath34 is a reasonable statistic for our data .
quantile - quantile plot for all @xmath44 data points .
coordinate @xmath45 : ordered residuals , @xmath40 , of the data .
coordinate @xmath46 : empirical estimate of the location of the @xmath47 @xmath48-tile of the normal distribution , i.e. @xmath49 where @xmath50 is the cumulative distribution function of the normal distribution .
, width=302 ] the addition of free model parameters on any data set can only reduce @xmath34 , therefore one must seek further justification for their inclusion .
the modified akaike information criterion ( @xmath51 ; the modification is that it has been corrected for small sample sizes ) penalises model parameters , allowing us to compare the goodness of a @xmath34 fit for different models .
the @xmath51 is defined as ( @xcite ; see also @xcite ) @xmath52 where @xmath53 is the number of free parameters in our model and @xmath54 is the number of data points in the sample .
the preferred model is the one with the lower @xmath51 .
a suggested interpretation scale for the @xmath51 is that introduced by @xcite , where a relative difference in the @xmath51 of two models of @xmath55 indicates that the two models are almost equivalent in terms of significance . in table
[ tab : aicc ] we present the results of our @xmath51 goodness - of - fit test of various @xmath0-variation models to the data . on the basis of these results we find : the keck sample exhibits a preference for the monopole fit ;
the vlt data is better fit by a dipole model ; and the combined sample gives lowest @xmath34 for the dipole + monopole model and the two - value model . in the next section we present results from different statistical tests that suggest that the dipole model is preferred even for the keck data subset .
lcdld model & @xmath53 & & & + + null ( @xmath56 ) & 0 & 310.439 & 1.0595 & 310.439 + monopole & 1 & 303.77 & 1.0403 & 305.784 + dipole @xmath21 & 3 & 284.628 & 0.9815 & 290.711 + dipole @xmath57 & 3 & 285.185 & 0.9834 & 291.268 + dipole @xmath21 + monopole & 4 & 279.66 & 0.9677 & 287.799 + dipole @xmath57 + monopole & 4 & 279.845 & 0.9683 & 287.984 + two - value & 2 & 282.585 & 0.9711 & 286.626 + _ keck ( 140 data points ) : _ & & & & + null & 0 & 157.898 & 1.1278 & 157.898 + monopole & 1 & 132.799 & 0.9554 & 134.828 + dipole @xmath21 & 3 & 142.439 & 1.0397 & 148.615 + dipole @xmath57 & 3 & 143.611 & 1.0482 & 149.787 + dipole @xmath21 + monopole & 4 & 131.28 & 0.9653 & 139.576 + dipole @xmath57 + monopole & 4 & 131.309 & 0.9655 & 139.605 + _ vlt ( 153 data points ) : _ & & & & + null & 0 & 152.541 & 0.9970 & 152.541 + monopole & 1 & 149.786 & 0.9854 & 151.812 + dipole @xmath21 & 3 & 139.908 & 0.9327 & 146.069 + dipole @xmath57 & 3 & 138.839 & 0.9256 & 145.000 + dipole @xmath21 + monopole & 4 & 139.288 & 0.9348 & 147.558 + dipole @xmath57 + monopole & 4 & 138.167 & 0.9273 & 146.437 + for the full ( combined ) sample , the best @xmath51 result ( though not the best @xmath34 ) comes from the two - value model when @xmath58 and @xmath59 . according to the jeffreys criterion , it is therefore of similar significance to the dipole+monopole model .
the two - value model is expected if there exist unknown systematics that cause apparent shifts in @xmath0 , and these ` intra - telescope ' systematics are different for the two telescopes .
however we will argue that the two - value result is not in conflict with the dipole result ; it is actually expected given the distribution of quasars seen with keck and vlt .
let us first discuss the idea that there are unknown intra - telescope systematics that cause spurious shifts in the observed value of @xmath1 .
what kind of systematic is required ?
any linear transform ( offset or scaling ) of the frequency scale will not mimic @xmath0-variation .
the many - multiplet method uses many lines for the analysis : some are positive shifters , some are negative shifters , and some are anchors . in order to mimic @xmath0-variation it is therefore necessary for a systematic to shift lines in different directions .
one suggestion is that intra - order systematic shifts in calibration of the keck spectrograph could mimic @xmath0-variation in any one absorber @xcite ; it is less clear that this would have a non - zero mean when averaged over many absorbers .
similar intra - order shifts have also been found in the vlt @xcite , but it s worth noting that these are smaller in amplitude than the keck shifts .
the proposition that intra - telescope systematics can mimic @xmath0-variation can also be tested using the subset of quasar absorption systems that have data from both vlt and keck telescopes .
there are seven such systems , and a sophisticated analysis of them shows no systematic velocity offsets between the two telescopes @xcite .
furthermore , the dipoles fit separately from the two telescopes point in the same direction ( within @xmath27 errors ) .
this is entirely unexpected if systematics accounted for monopole offsets for the telescopes : the probability that the dipoles would point in the same direction by chance was estimated at 6% .
a similar coincidence in the dipole directions was found when the sample was split into high ( @xmath60 ) and low redshifts @xcite ; these subsamples use very different transitions in different ions with different responses to @xmath1 ( see , e.g. * ? ? ?
an alternative explanation to intra - telescope systematics is that the two - value result is simply an artefact due to the two telescopes sampling data from significantly different portions of the sky .
if each telescope had full coverage of the sky , then one would expect that averaging over all absorbers would give back the monopole term of @xmath1 .
however , the telescopes do not each have full sky coverage . figure [
fig : dipolehistogram ] shows the number of quasar absorption systems seen by each telescope against the distance to each absorber projected along the dipolar axis .
the two telescopes gather data from different parts of the sky ( this is simply due to their locations on the earth ) .
if the fitted dipole+monopole model represents the true distribution of @xmath0 in the universe , then it is to be expected that a weighted mean over the observed values yields different values for the two telescopes .
a histogram of the distances , @xmath14 , to each absorber projected along the dipolar axis .
the bars with positive - sloped lines on the left are from the keck sample , while the bars with negative - sloped lines on the right represent the vlt data .
it is clear that the two telescopes sample different portions of the sky.,width=302 ]
additional statistical tests are available for distinguishing between alternate `` nested '' models .
two models with @xmath61 and @xmath62 parameters ( @xmath63 ) are nested if the entire parameter space of model 1 is contained within that of model 2 ; this is obviously the case for the 4-parameter dipole+monopole model and the single - parameter monopole model .
@xmath34 must improve when the three additional dipole parameters are included .
the @xmath26-test provides us a method to find the probability that this improvement is due to chance .
we define the @xmath26 statistic by @xmath64 where @xmath54 is the number of data points . under the null hypothesis
this statistic has an @xmath26-distribution ; we can infer the probability that the improvement in @xmath34 is due to chance by @xmath65 we express this as a standard deviation @xmath66 in the usual way , @xmath67 where @xmath68 is the gauss error function .
the significance of the dipole for the keck , vlt , and combined data samples is presented in table [ tab : sigres ] .
we again find that the keck sample shows no preference for a dipole model , the vlt sample shows a significant preference for a dipole , and the combined sample shows a strong preference for a dipole model .
.[tab : sigres ] significance of the dipole+monopole model using lookback time @xmath21 as the distance measure for the keck sample , the vlt sample , and the combined data .
significance is obtained using both the @xmath26-test and the error - ellipsoid method ( ) . [ cols="^,^,^,^",options="header " , ] another method of assessing the significance of the dipole model is to simply calculate the usual statistical significance of the dipole parameters relative to zero .
we use a maximum likelihood analysis to find the preferred direction and magnitude of the dipole .
an error ellipsoid is generated representing the uncertainty in the three - dimensional parameter space of the dipole , and the proportion of the best - fit dipole that is contained within this error ellipsoid indicates the significance of the dipole model .
this method was used , for example , by @xcite to assess evidence for a dipole in the accelerated expansion of the universe . in appendix
[ sec : eem ] we present an extension of this method , which we refer to as the error - ellipsoid method ( ) , to allow for distance - dependent dipole models of the type ( [ eq : dipolemonopole ] ) .
table [ tab : sigres ] shows that the gives values for the significance of the dipole+monopole model relative to the monopole - only model which are consistent with the @xmath26-test method for all data sets .
we wish to test whether the dipole result is due to a small proportion of the data sample . to do so we perform biased and unbiased clippings of the data .
we start with a biased iterative clipping method , where at each increment we remove a point that lends support to the dipole model .
one option is to remove the absorber producing the smallest residual value @xmath43 .
however , the data points with small residuals may be absorbers lying near the plane orthogonal to the fitted dipole axis , and these would not contribute to the dipole effect .
therefore we introduce an angular weighting that further biases the clipping towards removal of points that lie on the dipole axis : @xmath69 where the @xmath70 is the dipole model value for that absorber and @xmath71 is the angle from the dipole axis . after each clipping of the absorber with smallest weighted residual ,
we calculate the significance of the dipole using the @xmath51 , @xmath26-test and .
we present the result of these tests in figure [ fig : smallclip ] .
the solid black line gives the significance of preference for the dipole+monopole model over the pure monopole model as calculated via the f - test .
this test indicates that the dipole model is preferred over the monopole ( at @xmath27 significance or more ) until approximately 120 of the strongest points , or 40@xmath72 of the data , is removed .
the significance as calculated using is given by the dashed line of figure [ fig : smallclip ] : we see that the consistency between significance assessed using @xmath26-tests and established in section [ sec : dipole_significance ] extends to clipped subsets of the data until @xmath73 points have been removed and the significance is smaller than @xmath27 .
the vertical dashed line indicates the point at which the dipole+monopole model is no longer preferred over the monopole model as determined by the @xmath51 .
again , this occurs after @xmath73 points have been removed .
significance of the preference for a dipole+monopole model over the monopole model versus number of data points clipped . at each step
we have removed the absorber with smallest weighted residual @xmath74 , where @xmath9 is the angle between the location of the absorber and the dipole axis .
solid line : significance assessed using @xmath26-test ; dashed line : significance assessed using .
the vertical dashed line indicates the point at which the dipole model is not preferred according to the @xmath51.,width=302 ] the results of our biased clipping suggest that approximately 40% of the data that best supports the dipole must be removed before the significance of the dipole is reduced to @xmath27 .
therefore , we expect that if we randomly remove 120 data points from the sample and re - fit our data , the dipole+monopole model would still be preferred ( with at least @xmath27 significance ) over the monopole model . in figure [ fig:40pcremovals ]
we present the probability density function for this unbiased clipping performed 10 000 times .
each time , the dipole significance is calculated using an f - test .
we find that , as expected , the significance is at least @xmath27 in almost all 60% subsets of the data . taken together with our biased clipping , this shows that the dipole result is not due to a small number of outliers in the sample . probability density function for the significance of the dipole after random removal of 40% of data , assessed using the @xmath26-test .
we see that in almost all cases the dipole+monopole model is preferred to the pure monopole model at more than @xmath27 significance.,width=302 ] it is useful to contrast our biased iterative clipping to that presented in @xcite . in that work ,
the point with _ largest _ residual was removed at each step of the procedure .
this is useful for showing that the dipole is not caused by a few outlying data points , but it can also remove absorbers that do not support the dipole model .
for example , in @xcite there is no weighting by @xmath75 , and deviant points that lie in the perpendicular plane may be removed .
therefore it is unsurprising that the significance of the dipole increases at first ( fig . 4 in * ?
however , there is a trade - off : the significance eventually falls as more points are removed because there are simply fewer data .
@xcite found that approximately 60% of data with the largest residuals had to be discarded before the significance dropped below @xmath76 .
we have seen from our @xmath51 analysis ( section [ sec : aicc ] ) that there is some justification for including the monopole term of ( [ eq : dipolemonopole ] ) as well as the dipole term .
one natural interpretation of such a term is time - variation of the fine - structure constant , of the kind that was indicated in previous keck studies @xcite .
one would then expect the monopole term to have an explicit time - dependence ( or , equivalently , redshift - dependence @xmath77 ) , although the form may not be resolved by the data . another possible explanation for a monopole could be chemical evolution of the universe , which changes the isotope abundance ratios .
since terrestrial abundances are assumed , any deviation from these ratios in the absorber will shift the centroid of the line profile , and this might mimic a change in @xmath0 . accounting for a systematic such as
this is difficult , especially since the isotope shift is unknown for many of the lines used in the analysis .
the transition with the largest known shift used in the analysis are the @xmath782796 and 2803 lines in ii , and calculations suggest that the unknown isotope shifts are smaller and less important @xcite .
we have applied a simple test to see whether the isotope abundance ratios of magnesium could cause the monopole .
we have simply removed all data points taken from absorbers where the ii lines are present and used in the analysis .
113 absorbers remain from our initial sample of 293 : our statistical significance is hugely reduced .
the removal of points with ii is by no means a random sampling : these ii lines are seen in low redshift systems since at @xmath79 they are redshifted outside the range of optical telescopes . the results of our fitting are shown in table [ tab : mg_removed ] . for models that include a dipole we use the light - travel time as a measure of distance @xmath80 .
we see that the best - fit dipole parameters for the ii - removed data are consistent with those of the complete data set , but we no longer have a statistically significant monopole . using an @xmath26-test we find that the dipole+monopole model is preferred over the monopole model at @xmath81 significance for the ii - removed set .
llldl & parameter values & & + + null & @xmath56 & & 126.565 & 1.1200 + monopole & @xmath82 & @xmath83 & 125.421 & 1.1198 + dipole & @xmath84 & @xmath85 & 117.491 & 1.0681 + & & @xmath86 , @xmath87 + dipole + monopole & @xmath88 & @xmath89 & 116.863 & 1.0721 + & & @xmath90 + & & @xmath91 , @xmath92 + + dipole + monopole & @xmath88 & @xmath93 & 279.66 & 0.9677 + & & @xmath94 + & & @xmath95 , @xmath96 + we may try to estimate the magnitude of the change in ii abundances that could mimic the observed monopole term in the original data set .
the two ii lines ( at @xmath97 ) are often observed simultaneously with ii lines ( @xmath98 ) which in these systems will provide most of the sensitivity to @xmath0-variation ( @xmath99 for positive shifters in ii ) . to obtain our order - of - magnitude estimate
we assume here that only ii and positive - shifting ii lines are present in the system .
the procedure of @xcite is to simultaneously fit the redshift @xmath16 and @xmath1 ( along with column densities , doppler widths , etc . ) for the entire quasar spectrum . the monopole term in the full data set
, @xmath100 , corresponds to a shift in the ii lines of @xmath101 this shift manifests as an effective change in the redshift of the quasar absorption system as measured by the ii lines , but not as measured by the ii lines .
it is possible to obtain consistency in measured @xmath16 if the relative isotope abundances of ii are assumed to vary .
the isotope shift of the ii lines is @xmath102 @xcite . in order to compensate the monopole term of the observed @xmath0-variation
, the relative abundance of @xmath103 would have to change by @xmath104 that is , a roughly 5% increase in the relative abundance of @xmath105 could remove the observed monopole in @xmath1 . alternatively , a @xmath106 increase in the @xmath107 abundance relative to @xmath108 , would also work or any combination of the two ( e.g. the absorber may have the terrestrial @xmath109 ratio , and a reduction in @xmath108 relative to the heavier @xmath110 isotopes ) .
of course our assumptions are quite rough .
however a previous study of the effect of isotope abundance in the keck data came to much the same conclusion @xcite . in their method
, the quasar absorption spectra were refitted using a different value of magnesium heavy - isotope abundance and the @xmath1 extracted were averaged assuming a monopole model .
this found a linear relationship between their assumed @xmath111 ratio and the extracted @xmath1 . a change in @xmath1 of @xmath112 required an increase in the abundance of heavy mg isotopes relative to @xmath108 of around 10% ( see fig .
6 of * ? ? ?
a similar analysis applied to the @xmath113 subsample of the keck+vlt data also found that the monopole could be removed by a change in the @xmath111 abundance ratio from the terrestrial value of 0.21 to a heavy - isotope enhanced value of @xmath114 @xcite .
generally other species are present in our quasar absorption spectra , and this could remove the degeneracy between variation of magnesium isotope abundance and variation of @xmath0 . a more complete analysis should allow mg isotope abundances to vary in the fitting procedure , which may be possible if this is restricted to only one additional parameter in the entire sample of @xmath115 absorption systems .
recently a different model for spatial variation of @xmath0 has been proposed by @xcite where the universe is divided into two domains , each with a different value of @xmath0 . on the earth - side of the domain wall
@xmath0 takes the terrestrial value , while on the other side it takes a different value .
we can parametrize this model by the equation @xmath116 where @xmath117 is the shortest distance to the wall ( we use comoving distance , eq .
( [ eq : comoving_distance ] ) , in this section ) and @xmath9 is the angle between the direction of the shortest distance to the wall and a quasar .
again , values for @xmath117 and the direction of the wall are chosen to minimise @xmath34 , however in this case @xmath34 is discontinuous with respect to changes in the parameters : any given quasar must either be on this side of the wall ( model value @xmath56 ) or the other ( model value @xmath82 ) .
this necessitates the use of a genetic algorithm to determine the minimum values of @xmath34 ; our algorithm is presented in appendix [ sec : genetic ] . the best fit model for the entire data sample was found to have @xmath118 , @xmath119 gly , @xmath120 hr , @xmath121 . the model has 4-parameters with @xmath122 ; this can be compared with the 4-parameter dipole+monopole model with comoving distance , for which @xmath123 .
however it is worth noting that this minimum @xmath34 is obtained for an extremely narrow range of parameters , and even a small deviation in any of them increases it dramatically . for example , in figure [ fig : bestfitwall ] we present @xmath34 in the direction of best - fit ( solid line ) as a function of @xmath117 .
@xmath34 for the wall model as a function of comoving distance to the wall , @xmath117 .
solid line : in the direction of the best fit for the wall ( the best fit occurs at @xmath119 gly ) ; dashed line : in the direction of the dipole axis ( decreasing @xmath0 direction ) . by comparison , @xmath34 for the spatial gradient ( dipole ) model is 279.66.,width=302 ] the best - fit wall is located at an angle @xmath124 to the dipole axis ( towards negative values of @xmath1 , i.e. @xmath125 from the directions presented in table [ tab : mg_removed ] ) .
this is not consistent : if the universe really does have a domain wall in values of @xmath0 one would expect a dipole fit to the data to have the same direction .
if we calculate @xmath34 as a function of wall distance parameter @xmath117 along the direction that is given by the best - fit dipole we obtain the dashed line in figure [ fig : bestfitwall ] .
it is seen that the behaviour is much as would be expected if the dipole model is correct : as @xmath117 increases , fewer absorbers are found on the other side of the wall , the model value @xmath126 for more absorbers , and @xmath34 increases .
we have developed a number of statistical tests to use quasar absorption data to differentiate between possible models of cosmological @xmath0-variation .
we confirm that a model of variation including both dipole and monopole terms is preferred , however the monopole may be an artefact due to non - terrestrial magnesium isotope ratios in the absorbers . a small change ( approximately 5% ) in the ratio of @xmath105 to @xmath108 could account for the monopole , leaving the data to be explained by a purely spatial gradient in values of @xmath0 across the universe .
our robustness tests on the data showed remarkable consistency between various statistical tests : at least 40% of the data must be deleted in a heavily biased way to reduce the significance of the @xmath0-dipole to @xmath27 .
an unbiased removal of 40% of the data leaves the dipole significance at around @xmath127 .
this confirms that the original result , that there is evidence for a spatial gradient in values of @xmath0 at the @xmath7 level , is not due to only a few deviant points .
finally we developed a genetic algorithm to find best - fit parameters for the domain wall model of @xmath0-variation suggested by @xcite .
we find that the best fit direction of the wall is not consistent with the dipole direction .
this is because @xmath34 in the best - fit region is anomalously low with respect to its local neighbourhood of parameter space . on the other hand
if we force the domain wall to be perpendicular to the best - fit dipole direction , the resulting @xmath34 are significantly larger than those of the dipole model .
we thank c. angstmann , m. t. murphy , j. a. king , j. k. webb , and f. e. koch for useful discussions .
this work is supported by the australian research council .
in order to present a comparison of the significance of the models described above , we employ a similar strategy to that taken by @xcite , generalised for our distance - dependent dipole models . the probability that we observe data @xmath128 given
our model value at the @xmath39th quasar , @xmath129 , is @xmath130 where @xmath42 is the observational error and @xmath66 represents an intrinsic scatter of the true values from our model values @xmath131 .
the best fit model parameters can then be found so that they maximise the log - likelihood function @xmath132 \nonumber \\ & = \sum_i \ln \left ( \frac{1}{\sqrt{2\pi(\sigma^2+\sigma_i^2 ) } } \right ) - \frac{(x_i-\bar x_{i})^2}{2 ( \sigma^2+\sigma_i^2 ) } \ , .
\label{eq : loglikelihood}\end{aligned}\ ] ] for a dipole model , @xmath129 is some function on the dipole direction and magnitude , which we represent as * l*. maximising the likelihood with respect to the model parameter gives the expression @xmath133 the covariance matrix for @xmath134 is given by @xmath135 , which will depend on the various models as follows : @xmath136 @xmath137 @xmath138 @xmath139 where @xmath140 is the monopole magnitude , @xmath11 is the dipole magnitude , @xmath21 is the distance to the absorber and @xmath9 is the angle between direction of the dipole and the absorber .
we have taken a direct computational approach to determining the best fit model parameters . the probability that a dipole is preferred over the monopole , and hence the significance of the dipole result ,
is then determined as with the method of cooke and lynden - bell .
given the covariance matrix * a@xmath141 * , we have that the semiprincipal axes of the error ellipsoid for * l * are @xmath142 , where @xmath143 and @xmath144 are the eigenvalues and eigenvectors of the covariance matrix , respectively . the probability that the dipole should be included in the model , pr(*l * ) , is then found by integrating over the volume of the parameter space obtained by expanding or contracting the error ellipsoid such that it just encloses * l*. then @xmath145 where @xmath146 is the fraction of * l * enclosed by the error ellipsoid @xmath147 ^{1/2}\ ] ] and @xmath68 is the error function .
the standard procedure of minimising @xmath34 to find best - fit model parameters is numerically difficult for the wall model ( [ eq : wall_model ] ) because @xmath34 is discontinuous with respect to the model parameters : @xmath34 changes in discrete steps depending on whether any given quasar absorption system is on the earth - side or far - side of the wall .
this necessitates a genetic algorithm with simulated annealing to perform the fit . in a genetic algorithm ,
sets of model parameters are produced in ` generations ' with those producing the best fit ( in our cases the smallest @xmath34 ) in a particular generation retained to ` breed ' , producing the next brood of parameters . in each generation , ` mutations ' are introduced in order to produce the quasi - random variations of parameters necessary to cover the parameter space . the magnitude of such ` mutations ' are incrementally reduced in a process of simulated annealing so that the mutations become finer as parameters converge to their best fit values . as is usual when using genetic algorithms to determine best fits , several annealings are required .
unusually , in this case due to the discontinuous nature of the parameter space , we not only require several annealing steps but also several cycles of sets of annealing with new ` isolated ' populations of parameters in order to avoid a situation where the entire population inhabits local minima in the parameter space . in each cycle , we perform 5 annealing steps , as it was found that subsequent steps produced limited exploration of the parameter space due to the rapid convergence of the parameters , and that the re - initialising of genetic material in each cycle produced more efficient coverage of the parameter space . for this reason
, only a single survivor is retained per cycle , to be bred with nine new randomly created populations . in each annealing step ,
500 generations of model parameters are produced , with the variance scaled down with each generation .
we retain 10 sets of model parameters @xmath148 per generation , with which to produce 500 offspring .
we can think of these as `` vectors '' in the 4-dimensional parameter space of the wall model .
the offspring have genes taken to be the average of the parent values @xmath148 and @xmath149 with an additional random term : @xmath150 where the four terms in @xmath151 are drawn randomly from a gaussian distribution with standard deviation @xmath152 .
the random scatter is reduced as the parameter values converge during the simulated annealing : @xmath153 where @xmath154 is the number of generations that have been run on a given annealing . the new generation of parameter values ( parents and offspring )
are then sorted by the fitness parameter @xmath155 such that the 10 parameter sets producing the largest fitness parameters are retained .
after sufficient cycles ( 5 was found to be adequate ) the result is tested for convergence .
the result is deemed to be converged if the algorithm has not produced a better result for four consecutive cycles .
otherwise , the cycles continue .
this sacrifices speed of convergence for robustness of the result .
19ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop _ _ , ed .
( ) ( ) , @noop ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , ( ) , @noop * * , | we statistically analyse a recent sample of data points measuring the fine - structure constant @xmath0 ( relative to the terrestrial value ) in quasar absorption systems . using different statistical techniques
, we find general agreement with previous authors that a dipole model is a well - justified fit to the data .
we determine the significance of the dipole fit relative to that of a simple monopole fit , discuss the consistency of the interpretation , and test alternate models for potential variation of @xmath0 against the data . using a simple analysis we find that the monopole term ( the constant offset in @xmath1 ) may be caused by non - terrestrial magnesium isotope abundances in the absorbers .
finally we test the domain - wall model against the data . |
let @xmath0 be an algebraically closed field of arbitrary characteristic . an _ algebraic monoid _ is an algebraic variety @xmath1 over @xmath0 with an associative product @xmath2 which is a morphism of varieties , and such that there exists an element @xmath3 with @xmath4 for all @xmath5 .
we write @xmath6 for all @xmath7 .
let @xmath8 be the _ unit group _ of @xmath1 : @xmath9 we have the following picture : @xmath10 it is known that if @xmath1 is an affine algebraic monoid , then @xmath8 is an algebraic group , open in @xmath1 ( @xcite ) .
we extend this result to irreducible algebraic monoids .
moreover , the natural action of @xmath11 on @xmath1 by left and right multiplication has only one closed orbit : @xmath1 is a _ simple embedding _ of @xmath8 .
furthermore , if we require @xmath8 to be a reductive group , then the monoid @xmath1 is affine .
the proof of this result is based on the following remark : if @xmath1 is a reductive monoid , then by the bruhat decomposition there exist opposite borel subgroups @xmath12 of @xmath8 such that @xmath13 is open in @xmath1 .
so @xmath1 is a spherical @xmath14-variety .
it follows that if @xmath1 has a zero , then @xmath1 is affine .
we reduce the general case to that one by using techniques of the theory of spherical varieties . moreover , using the dictionary between simple spherical varieties and a family of combinatorial objects , the _ colored cones _ ( @xcite ) , we classify reductive monoids in arbitrary characteristic , in a way dual to that one of @xcite ( in characteristic zero ) .
i wish to thank m. brion , l. renner and e. vinberg for many useful remarks and sugestions .
we say that an algebraic monoid @xmath1 is _ irreducible _ ( resp .
_ normal _ ) if its underlying variety is irreducible ( resp .
normal ) .
an element @xmath15 is an _ idempotent ( element ) _ of @xmath1 if @xmath16 .
we note @xmath17 the set of idempotents of @xmath1 .
we say that a monoid @xmath1 has a _ zero _ if there exists an element @xmath18 such that @xmath19 for all @xmath5 .
an _ ideal _ of @xmath1 is a non - empty subset @xmath20 such that @xmath21 .
if there exists an ideal @xmath22 contained in all the ideals of @xmath1 , we say that @xmath1 has _ kernel _
@xmath23 .
if the unit group of a monoid is a reductive algebraic group , we say that the monoid is _ reductive .
_ a morphism @xmath24 between two algebraic monoids is a _ morphism of algebraic monoids _ if @xmath25 , and @xmath26 .
it is clear that @xmath27 .
if @xmath1 and @xmath28 have a zero , we impose that @xmath29 .
these conditions are not automatically satisfied : let @xmath30 be the monoid of @xmath31 matrices with coefficients in @xmath0 .
we can identify @xmath30 with the monoid of endomorphisms of a @xmath0-vector space @xmath32 of dimension @xmath33 .
its unit group is @xmath34 , the group of invertible matrices ; it is isomorphic to the group of automorphisms of @xmath32 .
the monoid @xmath30 is an affine , smooth , irreducible , reductive monoid with zero .
the morphism @xmath35 , @xmath36 , commutes with the multiplication , but it is not a morphism of algebraic monoids .
the morphism @xmath37 , @xmath38 , is not a morphism of algebraic monoids .
indeed , the image of the zero matrix is not the zero of @xmath39 . from now on
we suppose that all the algebraic monoids are irreducible , unless otherwise mentioned .
let @xmath40 be an algebraic group .
consider the action of @xmath41 on @xmath40 given by @xmath42 this action is transitive , and the isotropy group of @xmath43 is @xmath44 , the diagonal of @xmath40 .
moreover , the action satisfies the equation : @xmath45 we say that a @xmath46-variety @xmath47 is an _ embedding of @xmath40 _ if there exists an element @xmath48 such that the orbit @xmath49 is open in @xmath47 and isomorphic to @xmath50 .
if there exists only one closed orbit , we say that @xmath47 is a _ simple embedding_. [ teo : group - open ] let @xmath1 be an algebraic monoid .
then @xmath8 is a connected algebraic group , open in @xmath1 , and @xmath1 is a simple embedding of @xmath8 .
moreover , the unique closed @xmath14-orbit in @xmath1 is the kernel of @xmath1 .
_ let @xmath2 be the product in @xmath1 , and @xmath51 the projections .
then @xmath52 and @xmath40 is constructible in @xmath1 .
if we prove that @xmath40 is open and dense in @xmath1 , it will follow in particular that @xmath40 is connected , because @xmath1 is irreducible . from equation ( [ inversibles_constructibles ] )
, we have @xmath53 where @xmath54 is an irreducible component of @xmath55 that contains @xmath56 .
the product @xmath57 is surjective , and @xmath1 is irreducible , so ( ( * ? ? ?
6.4.5 ) ) @xmath58 let @xmath59 , @xmath60 .
in order to show that @xmath40 is dense in @xmath1 , it is sufficient to prove that the morphisms @xmath61 are dominant : then there exists a non - empty open subset of @xmath1 contained in @xmath62 .
the fibre @xmath63 is composed of the couples @xmath64 such that @xmath65 and @xmath66 , so @xmath67 .
because of the upper - semicontinuity of the function @xmath68 , there exists a non - empty open subset @xmath69 such that @xmath70 for all @xmath71 .
therefore , @xmath72 we have then equality , and @xmath73 , and thus @xmath74 is a dominant morphism . by symmetry , @xmath75 is a dominant morphism too .
the left multiplication @xmath76 by an element @xmath77 is an automorphism of @xmath1 , with inverse @xmath78 , such that @xmath79 .
it follows immediately that @xmath40 is open in @xmath1 , and that every point in @xmath40 is smooth in @xmath1 .
next , we show that @xmath80 is a morphism . before doing that
, we prove that if @xmath5 has a left inverse , then @xmath81 is a unit .
let @xmath82 be such that @xmath83 .
then @xmath84 has @xmath85 as a left inverse , so @xmath84 is injective .
the group @xmath40 being open in @xmath1 , @xmath86 is constructible in @xmath1 and @xmath87 .
then @xmath86 contains a non - empty open subset @xmath88 of @xmath1 .
but @xmath89 , so @xmath90 . if @xmath91 , there exists @xmath92 such that @xmath93 , and thus @xmath94 . therefore , we have : @xmath95 and @xmath96 is irreducible , so that @xmath97 is bijective .
if @xmath74 is separable , then it follows from zariski s main theorem that @xmath74 is an isomorphism .
analogously , @xmath75 is an isomorphism , and the inverse map @xmath80 is obtained by the composition @xmath98 , so it is an isomorphism .
so we must prove that @xmath74 is separable .
every point of @xmath40 is smooth in @xmath1 , so we can consider the differential of the product @xmath57 at the point @xmath99 : @xmath100 indeed , @xmath101 and @xmath102 for all @xmath5 ; the restriction of @xmath103 to the subspaces @xmath104 and @xmath105 is then the identity .
in particular , @xmath103 is surjective , and the subvariety @xmath106 is smooth at @xmath56 , with tangent space equal to @xmath107 the differential of @xmath74 at the point @xmath56 is the restriction to @xmath108 of the differential of the projection .
it follows that @xmath109 is surjective , and @xmath74 is separable .
we now show that @xmath1 is a simple embedding of @xmath110 .
we consider the action of @xmath41 over @xmath1 by left and right multiplication : @xmath111 the @xmath46-orbit passing by @xmath43 is @xmath40 , so @xmath1 is an embedding of @xmath40 .
if @xmath23 is a closed @xmath112-orbit , then @xmath113 it follows that every closed orbit is an ideal .
if @xmath23 and @xmath114 are closed orbits , then @xmath115 , and thus @xmath116 , because @xmath23 and @xmath114 are orbits .
let @xmath23 be the unique closed orbit . if @xmath20 is an ideal , then @xmath117 , and it follows that @xmath118 , so @xmath23 is the kernel of @xmath1 .
we remark that in the proof of the preceding theorem we have proved : let @xmath1 be an algebraic monoid . if @xmath5 has a left ( or right ) inverse , then @xmath81 is an unit .
+ on the other side , we have a partial converse to theorem [ teo : group - open ] ( wich was proved by vinberg in characteristic @xmath119
see @xcite ) : let @xmath40 be an algebraic group and @xmath1 be an affine embedding of @xmath40 .
then the product on @xmath40 extends to a product @xmath120 , in such a way that @xmath121 .
_ if the product @xmath122 extends to a morphism @xmath123 , the @xmath124 is a product , because @xmath125 is associative . for convenience ,
we note @xmath126 .
it is clear that the identity @xmath127 satisfies @xmath128 for all @xmath5 .
indeed , this holds for the elements of the open subset @xmath129 . consider the right and left actions of @xmath40 given by : @xmath130 by construction and equation ( [ eq : product - action ] ) , these actions coincide on @xmath131 with the product in @xmath40 .
then there exists a morphism @xmath132 which extends the product of @xmath40 .
therefore , it suffices to prove that every regular function defined in @xmath88 can be extended to @xmath133 .
if @xmath32 is a vector space and @xmath134 a subspace , an easy calculation shows that @xmath135 . applying this to @xmath136\subset k[g]$ ]
, we get : @xmath137\cong k[s]\otimes k[s ] = ( k[s]\otimes k[g])\cap ( k[g]\otimes k[s])\ .\ ] ] on the other hand , a function @xmath138\otimes k[g]$ ] belongs to @xmath136\otimes k[g]$ ] if and only if it can be extendend to a function defined in @xmath139 .
the same happens for @xmath140\otimes k[s]$ ] .
it follows that every function defined in @xmath88 can be extended to a function defined in @xmath133 . finally , we prove that @xmath121 .
the inclusion @xmath141 is trivial .
let @xmath142 be a unit . then @xmath143 , and there exists @xmath77 such that @xmath144 , because @xmath40 is open in @xmath1 .
it follows that @xmath145 . in order to study the geometry of monoids
it is useful to suppose that they are normal .
the following result allows us to doing so without loss of generality : [ lem : normalization ] _ ( @xcite , @xcite ) _ let @xmath1 be an algebraic monoid , and @xmath146 its normalization .
then there exists an unique associative product @xmath147 such that @xmath148 .
moreover , the following proprieties are verified : _ i ) _ @xmath20 is an ideal if and only if @xmath149 is an ideal .
@xmath1 has a zero @xmath150 if and only if @xmath151 , where @xmath152 is a zero of @xmath153 .
@xmath153 is an algebraic monoid , with @xmath154 , and @xmath155 is an isomorphism of groups . _
iv ) _ the restriction of @xmath156 to @xmath157 gives a bijection between @xmath157 and @xmath17 .
it follows that @xmath156 is a morphism of ( algebraic ) monoids .
in particular , @xmath156 is equivariant for the actions of @xmath158 and @xmath159 . _ proof .
_ we define the multiplication @xmath160 by means of the universal propriety of normalizations : @xmath124 is the unique morphism such that the diagram ( 70,74)(0,-10 ) ( 0,50)(0,0)@xmath161 ( 0,0)(0,0)@xmath133 ( 50,50)(0,0)@xmath153 ( 50,0)(0,0)@xmath1 ( -18,25)(0,0)@xmath162 ( 58,25)(0,0)@xmath163 ( 30,-8)(0,0)@xmath164 ( 30,58)(0,0)@xmath165 ( 20,50)(1,0)20 ( 20,0)(1,0)20 ( 0,40)(0,-1)30 ( 50,40)(0,-1)30 is commutative . by the universal property of normalizations , we can easily prove that @xmath124 is associative , and by construction , @xmath156 is then a morphism of algebraic semigroups . we prove only iv ) . for the others assertions ,
we refer the reader to @xcite .
it is clear that @xmath166 .
let @xmath167 , and @xmath168 .
the morphisms @xmath169 and @xmath170 verify the following commutative diagram : ( 150,80)(0,-15 ) ( 0,50)(0,0)@xmath153 ( 120,50)(0,0)@xmath153 ( 0,0)(0,0)@xmath1 ( 120,0)(0,0)@xmath1 ( 60,50)(0,0)@xmath153 ( 60,0)(0,0)@xmath1 ( -8,25)(0,0)@xmath171 ( 128,25)(0,0)@xmath171 ( 68,25)(0,0)@xmath163 ( 30,-8)(0,0)@xmath172 ( 95,-8)(0,0)@xmath173 ( 30,58)(0,0)@xmath174 ( 95,58)(0,0)@xmath175 ( 10,50)(1,0)40 ( 110,50)(-1,0)40 ( 10,0)(1,0)40 ( 110,0)(-1,0)40 ( 0,40)(0,-1)30 ( 120,40)(0,-1)30 ( 60,40)(0,-1)30 again by the universal property of normalizations , @xmath176 . in particular , @xmath177 ,
so @xmath178 .
let @xmath179 . if we replace @xmath180 by @xmath181 in the preceding diagram , we get again a commutative diagram , and it follows that @xmath182 .
analogously , @xmath183 , and thus @xmath184 .
we restrict ourselves to the study of reductive ( irreducible ) monoids .
all the reductive groups are supposed connected .
if @xmath40 is a reductive group , a homogeneous space @xmath185 is _ spherical _ if there exists a borel subgroup @xmath186 of @xmath40 such that @xmath187 is open . if @xmath47 is a @xmath40-variety with an open orbit isomorphic to @xmath185 , then @xmath47 is called an _ embedding of @xmath185 . _
we say that @xmath47 is _ simple _ if the action of @xmath40 over @xmath47 has only one closed orbit . if @xmath47 is normal , we say that @xmath47 is a _
spherical variety . _
if @xmath40 is a reductive group , it follows from the bruhat decomposition that the normal @xmath40-embeddings are spherical varieties , so , by theorem [ teo : group - open ] , normal reductive monoids are simple spherical varieties .
the aim of this section is to study the geometry of reductive monoids from the point of view of the theory of spherical varieties . in particular
, we will show that reductive monoids are affine . in the litterature
, it is imposed to @xmath40-embeddings to be normal .
we do nt do this in order to be able to state results valid for all reductive monoids .
we begin by proving the following let @xmath1 be a reductive monoid with zero . then @xmath1 is affine .
_ more generally , if @xmath47 is a simple spherical variety , with closed orbit a point , then @xmath47 is affine ( ( * ? ? ?
7.6 and 7.7 ) ) .
if @xmath1 is not normal , it suffices to take the normalization @xmath146 .
it follows from lemma [ lem : normalization ] that @xmath153 is a normal monoid with zero .
so @xmath153 is affine , and if follows that @xmath1 is .
indeed , the morphism of the normalization @xmath156 is a finite surjective morphism , so @xmath153 is affine if and only if @xmath1 is ( @xcite ) . in order to continue our study
, we must give _ a priori _ proofs of results concerning idempotents in reductive monoids , which are known in the affine case .
we recall that an _ 1 parameter subgroup _ ( 1-ps ) of a group @xmath40 is a multiplicative morphism @xmath188 .
we note @xmath189 the group of 1-ps of @xmath40 .
[ lem : idempotent ] let @xmath1 be a reductive monoid with unit group @xmath40 , and let @xmath190 be a maximal torus
. then every @xmath46-orbit contains an idempotent element belonging to @xmath191 . _ proof . _
first , we show that we can approach every @xmath46-orbit of @xmath1 by an 1-ps of @xmath40 .
if @xmath192 , there exists a curve germ @xmath193 , where @xmath194 denotes the ring of laurent series in one variable with coefficients in @xmath0 , such that @xmath195 . by iwahori s theorem ( @xcite ) , there exists an 1-ps @xmath196 of @xmath40 such that @xmath197 , where @xmath198 and @xmath199 belong to @xmath200}=\operatorname{hom}(\operatorname{spec}k[[t]],g)$ ] .
let @xmath201 and @xmath202 .
then , @xmath203 the 1-ps @xmath196 has its values in a maximal torus @xmath204 . as all maximal tori of @xmath40
are conjugated , there exists @xmath205 such that @xmath206 for all @xmath207 .
then @xmath208 has a limit in @xmath209 .
next , we observe that if @xmath210 , then @xmath211 so @xmath212 is the idempotent we are looking for . [
lem : orbits - monoids ] let @xmath1 be a reductive monoid with unit group @xmath40 .
let @xmath167 , and @xmath213 .
if @xmath23 is a monoid with identity @xmath212 , then @xmath212 is a central element of @xmath1 , and @xmath23 is an algebraic group .
the isotropy group of @xmath212 satisfies : @xmath214 where @xmath215 is the isotropy group of @xmath212 for the action of @xmath40 by left multiplication .
moreover , @xmath215 is a normal subgroup of @xmath40 . in particular
, @xmath216 is a reductive group .
_ as @xmath212 is the identity of @xmath23 , it follows that @xmath217 for all @xmath77 .
then @xmath218 , and @xmath219 .
moreover , @xmath129 is dense in @xmath1 , so @xmath220 for all @xmath5 .
thus , @xmath212 is a central element of @xmath1 . in particular @xmath221 , and @xmath222 is a bijective morphism . observe that the multiplication in @xmath23 has an inverse , namely @xmath223 .
if @xmath224 and @xmath225 , then @xmath226 it follows that @xmath215 is normal in @xmath40 .
thus , @xmath227 is compatible with products .
on the other hand , it is clear that @xmath228 .
if @xmath229 , then @xmath230 it follows that @xmath231 , and @xmath232 .
the multiplication in @xmath23 may not be induced by that one of @xmath41 through @xmath233 .
indeed , @xmath216 is not always normal in @xmath41 . the following result of waterhouse shows some properties of unit groups of affine algebraic monoids .
( @xcite , ( * ? ? ?
3.3.6 ) ) let @xmath40 be an algebraic group ( not necessarily reductive ) .
then the following proprerties are equivalent : \i ) the character group of @xmath40 is non trivial .
\ii ) the rank of the radical of @xmath40 is not equal to zero .
\iii ) there exists an affine algebraic monoid wich unit group is @xmath40 and such that @xmath234 . if moreover @xmath40 is reductive , we show that the _ connected center _ of @xmath1 ( i.e. the closure in @xmath1 of @xmath235 , the connected center of @xmath40 ) meets the kernel of @xmath1 : [ prop : central - idempotent ] let @xmath1 be a reductive monoid with unit group @xmath234 ; let @xmath235 be the connected center of @xmath40 , and @xmath23 be the kernel of @xmath1 . then @xmath236 .
in particular , the connected center of @xmath40 is non trivial .
_ let @xmath237 be a maximal torus of @xmath40 .
by lemma [ lem : idempotent ] , there exists an 1-ps @xmath238 whose limit exists and is an idempotent @xmath239 .
if @xmath240 is the weyl group of @xmath40 , then @xmath241 acts on @xmath237 by conjugation .
we obtain then an action of @xmath241 over @xmath242 as follows : @xmath243 from @xmath244 we obtain @xmath245 .
let @xmath246 ; it is a @xmath241-invariant 1-ps , with limit @xmath247 . in this way
, we have constructed a 1-ps @xmath248 of the center of @xmath40 ( because it is @xmath241-invariant ) whose limit @xmath212 belongs to @xmath23 . in particular , @xmath248 is non trivial , and @xmath249 : the proposition is proved .
it is known that every affine embedding of a semi - simple group is trivial ( @xcite ) .
this result follows immediately from the proposition above : if the unit group of a monoid @xmath1 is a semi - simple group @xmath250 , then @xmath251 .
+ [ teo : reductif - affine ] let @xmath1 be a reductive monoid .
then @xmath1 is affine .
_ first , we consider the case where @xmath1 is normal .
consider the kernel @xmath23 of @xmath1 , @xmath110 , @xmath252 a maximal torus and @xmath253 the connected center .
let @xmath254 as in proposition [ prop : central - idempotent ] .
then @xmath212 is a central element of @xmath1 , so @xmath255 for all @xmath5 .
an element @xmath256 can be written as a product @xmath257 .
in particular , @xmath23 is an algebraic monoid with identity @xmath258 .
consider the morphism @xmath259 , @xmath260 .
we have : @xmath261 it follows that @xmath262 is an @xmath46-equivariant morphism of monoids .
the fibre @xmath263 is then stable under multiplication and under the action of @xmath216 .
the identity @xmath43 belongs to @xmath264 : the fibre @xmath264 is a monoid with zero ( equal to @xmath212 ) . if @xmath265 , from lemma [ lem : orbits - monoids ] we deduce that @xmath266 .
moreover , @xmath267 , and it follows that @xmath215 is an open subset of @xmath264 , so @xmath268 by theorem [ teo : group - open ] . then @xmath264 is a reductive monoid with zero , so it is affine .
we consider the affine variety @xmath269 .
the reductive group @xmath216 acts in @xmath270 : @xmath271 where @xmath229 , @xmath272 , @xmath273 .
the quotient @xmath274 is then an affine variety .
if we prove that this quotient is isomorphic to @xmath1 , it will follow that @xmath1 is affine .
we consider the morphism @xmath275 as @xmath276 is constant over the @xmath216-orbits of @xmath270 , then @xmath276 induces a morphism @xmath277 .
we show that @xmath278 is bijective and birational , then it follows from zariski s main theorem that @xmath278 is an isomorphism . _
@xmath278 is injective : _ let @xmath279 such that @xmath280
. then @xmath281 , and @xmath282 .
it follows that @xmath283 then @xmath284 , and thus @xmath285 .
so @xmath286 _ @xmath278 is surjective : _ if @xmath5 , then @xmath287 for some @xmath77 .
then @xmath288 , and @xmath289 . _
@xmath278 is birational : _
the group @xmath41 acts on @xmath290 by left multiplication in the left factor , in such a way that @xmath278 is @xmath46-equivariant .
moreover , the orbit @xmath291 passing by @xmath292 is open in @xmath290 , and isomorphic to the quotient @xmath293 .
the restriction of @xmath278 gives then an isomorphism from @xmath291 to @xmath40 , with inverse @xmath294 .
for the general case , observe that if @xmath1 reductive , non - normal monoid , its normalization @xmath153 is a reductive normal monoid by lemma [ lem : normalization ] .
it follows that @xmath153 , and therefore @xmath1 , is affine .
indeed , the normalization @xmath146 is an finite surjective morphism .
the kernel of a normal reductive monoid is a reductive group .
_ we have seen that if @xmath1 is a normal reductive monoid and @xmath23 its kernel , then @xmath23 is a group , with identity @xmath295 .
moreover , if we keep the notation of the proof of theorem [ teo : reductif - affine ] , then @xmath296 .
it follows that @xmath297 but @xmath298 , and @xmath23 is a reductive group . _
i ) _ taking irreducible components , it is easy to prove that if @xmath1 is a ( reducible ) reductive normal monoid , then @xmath299 is affine , and each unit belongs to an unique irreducible component ( @xcite ) .
moreover , any two irreducible components of @xmath299 are isomorphic as varieties .
_ ii ) _ it is known that any irreducible affine monoid can be embedded as a closed submonoid of the monoid of endomorphisms of some vector space @xmath32 of finite dimension ( @xcite ) .
it follows from theorem [ teo : reductif - affine ] that any reductive monoid @xmath1 can be seen as the closure in @xmath300 of a closed reductive subgroup of @xmath301 , which is then isomorphic to @xmath8 .
we can sumarize our results in the following the reductive monoids are exactly the affine embeddings of reductive groups .
the commutative reductive monoids are exactly the affine embeddings of tori .
_ we have only to prove the last statement .
it is clear that if @xmath1 is a affine embedding of a torus then @xmath1 is commutative .
conversely , if @xmath1 is a reductive commutative monoid , then its unit group is a reductive commutative connected group , i.e. a torus .
let @xmath1 be a reductive monoid , and @xmath40 its unit group .
if @xmath237 is a maximal torus of @xmath40 , then the properties of @xmath191 reflect in some sense those of @xmath1 .
we state first a known result on irreducible monoids whose unit group is a torus ( see for example ( @xcite ) .
[ lem : ideals - toric ] let @xmath1 be a monoid whose unit group is a torus @xmath237 .
then the ideals of @xmath1 are exactly the @xmath237-stable closed subsets of @xmath1 .
in particular , every ideal of @xmath1 is closed . moreover , if @xmath302 then @xmath303 if and only if @xmath304 .
+ this result generalizes to reductive monoids as follows : [ prop : ideals - idempotents ] let @xmath1 be a reductive monoid , with unit group @xmath40 .
then the ideals of @xmath1 are exactly the @xmath46-stable closed subsets of @xmath1 .
in particular all ideals are closed .
two idempotents @xmath302 are in the same @xmath46-orbit if and only if they are @xmath40-conjugate . in particular ,
if @xmath252 is a maximal torus , then @xmath305 _ proof .
_ we prove the first part .
for the second one , see for example ( * ? ? ?
* cor . 6.10 ) .
if @xmath20 is a @xmath46-stable closed subset , then @xmath306 that is , @xmath307 is an ideal of @xmath1 .
let @xmath20 be an ideal of @xmath1 . by definition
, @xmath307 is @xmath46-stable .
let us prove that @xmath307 is closed . from lemma [ lem : ideals - toric ]
, we get that @xmath308 then , its closure in @xmath1 is equal to @xmath309 .
we need only to prove that if @xmath310 , then @xmath311 .
note that @xmath312 , with @xmath313 an idempotent element .
moreover , @xmath314 is a @xmath46-stable closed subset of @xmath1 , so it is an ideal of @xmath1 .
the intersection @xmath315 is a non - empty @xmath237-stable subset of @xmath316 , so @xmath317 and the closure of @xmath315 is an ideal of @xmath191 .
let us call this ideal @xmath318 .
as @xmath319 for all @xmath320 , we get @xmath321 , and there exists @xmath322 such that @xmath323 .
it follows from lemma [ lem : ideals - toric ] that @xmath324 and the proposition is proved .
in @xcite , vinberg classified all reductive normal monoids in characteristic zero in terms of the decomposition of their algebra of regular functions for the action of the unit group .
this classification is dual of that of affine embeddings of a reductive group ( see @xcite ) . from the work of knop ( @xcite ) , we know how to classify affine embeddings of reductive group in arbitrary characteristic . in order to do that
, we begin by recalling some results from the theory of spherical embeddings and in particular _ reductive embeddings _ ( i.e. embeddings of reductive groups ) , which proofs can be found in @xcite .
next , we classify ( normal ) reductive monoids . in view of lemma [ lem :
normalization ] , we suppose that all varieties considered are normal .
+ + let @xmath47 be an algebraic variety .
we denote by @xmath325 the field of rational functions over @xmath47 .
let @xmath40 be a reductive group , @xmath185 a spherical homogenous space , and @xmath47 a simple spherical embedding of @xmath185 , with unique closed orbit @xmath23 .
we note @xmath326 the _ set of colours , _ i.e. the set of @xmath186-stable irreducible divisors of @xmath185 .
we call the _ ( set of ) colours _ of @xmath47 , and we denote by @xmath327 , the subset of @xmath328 consisting of the irreducible divisors such that their closure contains @xmath23 .
we denote by @xmath329 the set of irreducible @xmath40-stable divisors of @xmath47 .
let @xmath330 be the set of weights of @xmath331 , we define @xmath332 we call the _ rank _ of @xmath185 the dimension of the @xmath333-vector space @xmath334 .
let @xmath335 be an algebraic variety .
valuation _ of the field @xmath325 is a function @xmath336 with the following properties : _ i ) _
@xmath337 for all @xmath338 .
@xmath339 for all @xmath338 _ iii ) _
@xmath340 , @xmath341 .
it is easy to prove that @xmath340 , @xmath341 .
we say that the valuation @xmath342 is _ normalized _ if @xmath343 .
if @xmath47 is a @xmath40-variety , a valuation @xmath342 of @xmath325 is called @xmath40-invariant if @xmath344 for all @xmath77 , @xmath345 , where the @xmath40-action over @xmath325 is given by @xmath346 for all @xmath48 .
let @xmath185 be an spherical homogeous space , @xmath342 a valuation of @xmath347 and @xmath348 a weight .
if @xmath349 are two rational functions of weight @xmath350 , then the function @xmath351 is constant over the open @xmath186-orbit , and thus it is constant over @xmath185 .
it follows that @xmath352 .
we set @xmath353 .
if @xmath354 and @xmath355 are two rational functions of weight @xmath350 and @xmath356 respectively , then @xmath357 is a rational function of weight @xmath358 , and @xmath359 .
thus , every valuation @xmath342 of @xmath347 defines an element @xmath360 .
if @xmath361 is the set of @xmath40-invariant valuations of @xmath347 , the restriction of @xmath362 to @xmath363 gives an injection @xmath364 ( @xcite ) .
we will identify @xmath363 with its image in @xmath365 .
the set @xmath366 is a polyhedral cone ( i.e. a cone generated by a finite number of elements of the lattice @xmath367 ) , we call it the _ valuation cone
. _ let @xmath368 be a @xmath40-stable irreducible divisor of @xmath47 , and @xmath369 the ( normalized ) @xmath40-invariant valuation associated to @xmath370 .
as @xmath370 is determined by @xmath369 , the map @xmath371 is injective .
we will identify @xmath372 with its image by @xmath362 .
on the other hand , there is a natural mapping @xmath373 : to each color @xmath370 we associate @xmath374 , where @xmath369 is the valuation of @xmath347 associated to @xmath370 .
this mapping , noted @xmath362 , is not necessarily injective ( @xcite ) .
let @xmath185 be a spherical homogenous space .
a _ colored cone _ of @xmath365 is a pair @xmath375 , where @xmath376 is a subset such that : _ i ) _ @xmath377 does not contain @xmath119 , _ ii ) _ @xmath378 is a polyhedral cone generated by @xmath379 , and a finite number of elements of @xmath363 , _ iii ) _
@xmath380 , where @xmath381 denotes the relative interior of @xmath382 .
a colored cone is _ strictly convex _ if @xmath383 is .
a _ colored face _ of the colored cone @xmath384 is a colored cone @xmath385 , with @xmath386 a face of @xmath383 , and @xmath387 .
[ bijection_orbites_faces](see @xcite ) there exists a bijection between the simple embeddings of @xmath185 and the strictly convex colored cones of @xmath388 .
this bijection associates to each variety @xmath47 the colored cone @xmath389 , where @xmath390 is the cone generated by @xmath391 .
we call @xmath392 the _ colored cone associated to @xmath47 .
_ we note @xmath393 .
_ ii ) _ the @xmath40-orbits of a simple spherical variety @xmath47 are in bijection with the _ colored faces _ of its associated colored cone .
this bijection is constructed as follows : let @xmath394 be a @xmath40-orbit .
we denote by @xmath395 the cone generated by the image by @xmath362 of the @xmath186-stable irreducible divisors which contain @xmath23 .
if @xmath396 is the set of colours such that their closure contains @xmath23 , then @xmath397 is the colored face we are looking for .
_ iii ) _ if @xmath398 is the parabolic subgroup given by the intersection of the stabilizers of the colours in @xmath399 , then @xmath400 in particular , the unique closed orbit is a point if and only if the dimension of @xmath401 is equal to the rank of @xmath185 , and the set of colours of the embedding is all @xmath402 . _
iv ) _ a spherical variety @xmath47 is affine if and only if @xmath47 is simple and there exists @xmath403 such that * @xmath404 * @xmath405 * @xmath406 . + in @xcite , vust established the combinatorial data associated to symmetric spaces , in particular to the homogenous space @xmath407 , where @xmath40 is a reductive connected group , and @xmath408 .
in order to carry out the classification of reductive monoids , we must generalize his results to arbitrary characteristic .
this is essentially done in the literature , so we will state the results without proving them . for a proof ,
we refer the reader to @xcite .
let us fix the notation : we note @xmath250 the commutator of @xmath40 .
thus , @xmath409 , where @xmath410 is the connected center of @xmath40 .
moreover , @xmath411 is finite ; @xmath40 is then isomorphic to @xmath412 : @xmath413\cong k[(g_0\times z)/z_0]\cong ( k[g_0]\otimes_k k[z])^{z_0}\,.\ ] ] we fix @xmath414 , a maximal torus of @xmath250 , and @xmath415 , a borel subgroup that contains @xmath414 .
let @xmath416 and @xmath417 be the simple roots and fundamental weights associated to @xmath418 respectively .
we set @xmath419 , a maximal torus of @xmath40 .
we note @xmath241 the weyl group associated to @xmath420 , and @xmath421 the weyl chamber of @xmath40 associated to @xmath186 .
if @xmath422 , the weight lattice of @xmath40 associated to @xmath423 , satisfies : @xmath424 it is a sub - lattice of finite index of @xmath425 .
the reflection associated to a simple root @xmath426 is noted as usual @xmath427 . from the bruhat decomposition , it is easy to prove the following [ q(g)](@xcite ) let @xmath40 be a reductive group . if we keep the preceding notations and identify @xmath428 with @xmath242 , then @xmath429 where the lattice considered is @xmath430 .
we identify @xmath242 with @xmath431 by means of a @xmath241-invariant form @xmath432 , in such a way that @xmath433 . under this identification ,
@xmath434 , @xmath435 , and the valuation cone @xmath436 is opposite to @xmath437 . moreover , the colours of @xmath40 are the divisors @xmath438 , @xmath439 , with associated valuations @xmath440
. + it follows immediately that the behaviour under quotients by a finite central subgroup of the combinatorial data associated to reductive monoids is `` good '' : [ quotient_par_groupe_central ] ( @xcite,@xcite ) let @xmath40 be a reductive group and @xmath441 a finite central subgroup .
then + * @xmath442 , where we consider the lattice @xmath443 . *
if @xmath444 is the canonical projection , then @xmath156 induces a bijection between @xmath445 and @xmath446 .
* under the identification i ) , @xmath447 , and the valuation cones @xmath448 and @xmath449 coincide . in particular , @xmath450 is the identity . *
if @xmath1 is a monoid with unit group @xmath40 , then @xmath451 is a reductive monoid , with unit group @xmath452 .
moreover , if we denote @xmath453 the quotient , then @xmath454 is the identity , and under this identification we have @xmath455 .
+ in the case of reductive embeddings , the application @xmath456 is injective .
we identify then @xmath328 with its image by @xmath362 . from this
, we can recover the fact that a reductive monoid is a quasi - direct product of a reductive group and a reductive monoid with zero ( @xcite,@xcite ) : [ prop : quasi - direct_product ] let @xmath1 be a reductive monoid with unit group @xmath40 .
then there exist a reductive group @xmath215 and a reductive monoid with zero @xmath457 such that @xmath458 where @xmath459 is a finite central subgroup of @xmath40 .
in particular , if @xmath460 , then @xmath461 .
+ it follows that , as in the case of toric varieties , we can restrict ourselves to the study of reductive monoids with zero : [ passage_monoides_zero ] let @xmath1 be a reductive monoid with unit group @xmath40 .
let @xmath462 and @xmath459 be as above . then
, under the identification @xmath463 , we have : @xmath464 _ proof .
_ by propositions [ prop : quasi - direct_product ] and [ quotient_par_groupe_central ] , we can suppose without loss of generality that @xmath465 , and @xmath466 , @xmath457 a monoid with zero and with unit group @xmath467 .
in particular , @xmath468 , and @xmath469 .
if @xmath470 is a borel subgroup of @xmath471 , @xmath60 , then @xmath472 is a borel subgroup of @xmath40 .
if @xmath473 are the simple roots associated to @xmath474 , and @xmath475 those associated to @xmath476 , then @xmath477 are the simple roots associated to @xmath186 . denote @xmath478 and @xmath479 the colours of @xmath215 and @xmath467 respectively . it is immediate that the colours of @xmath40 are : @xmath480 as @xmath140=k[g_1]\otimes k[g_2]$ ] , the valuations associated to @xmath481 and @xmath482 are @xmath483 and @xmath484 respectively . in the same way , we see that @xmath485 .
a divisor @xmath486 is @xmath46-stable if and only if @xmath487 , where @xmath488 is a divisor @xmath489-stable of @xmath457 .
it follows that the valuation associated to @xmath370 satisfies @xmath490 .
on the other hand , the kernel of @xmath1 ( i.e. the unique @xmath491-closed orbit ) is equal to @xmath492 . if follows that the set of colours @xmath493 is @xmath494 ( we recall that the set of colours of a monoid with zero @xmath457 is @xmath495 ) , and the proposition is proved . [ plongaffine ]
let @xmath1 be a reductive monoid with unit group @xmath40 .
let @xmath496 be the colored cone associated to @xmath1 .
then @xmath497 is a strictly convex cone .
+ the converse to this corollary is not true , as the following example shows : consider @xmath498 .
we identify @xmath499 to @xmath500 , and @xmath189 to the lattice @xmath501 .
the affine embeddings of @xmath40 are exactly those where the closed orbit is a point ( with colored cone of the form @xmath502 ) , and the _ elementary embeddings _ associated to the half lines @xmath503 and @xmath504 .
indeed , let @xmath505 be the colored cone associated to an affine embedding without colors .
then there exists @xmath506 such that @xmath507 , @xmath508 for all @xmath509 .
it follows that @xmath510 if and only if @xmath342 belongs to the @xmath511-axis .
let @xmath512 such that the cone @xmath513 generated by @xmath342 and @xmath514 is strictly convex .
then the embedding @xmath1 associated to @xmath515 is affine ( just take @xmath516 ) .
in particular , @xmath1 is a reductive monoid ( with zero : the set of colors of @xmath1 is all @xmath328 ) .
( 0,0 ) # 1#2#3#4#5 @font ( 2484,2489)(289,-1783 ) ( 2051,-609)(0,0)[lb ] ( 985,314)(0,0)[lb ] on the other hand , if we take @xmath517 , then @xmath518 is strictly convex , but the embedding associated to the colored cone @xmath519 is not affine .
an immediate consequence of the preceding proposition is that the colored cone of a reductive monoid @xmath1 is determined by @xmath520 , or even @xmath521 .
moreover , we can find the decomposition of @xmath1 as a quasi - direct product of a reductive monoid with zero and a reductive group ( cf .
proposition [ prop : quasi - direct_product ] ) as follows : if we keep the notation of proposition [ prop : quasi - direct_product ] , the space @xmath522 is @xmath523 , the sub - space of @xmath524 generated by @xmath525 .
the lattice considered contains @xmath526 as a subgroup of finite index .
a simply connected covering of the commutator of the normal subgroup @xmath215 ( resp .
@xmath467 ) is the semi - simple simply connected group whose root system is given by the dynkin diagram of @xmath40 restricted to @xmath493 ( resp .
@xmath527 ) . in terms of the combinatorics of the colored cones of @xmath528 ,
this remark gives the following result : [ construction_monoides ] let @xmath40 be a reductive group .
given a polyhedral cone @xmath386 contained in @xmath529 , such that @xmath530 is strictly convex , there exists an unique subset of colours @xmath531 such that @xmath532 is the colored cone associated to an affine embedding of @xmath40 .
_ let @xmath533 be a cone such that @xmath530 is strictly convex .
if @xmath534 is a subset of @xmath524 , denote @xmath535 the sub - space of @xmath524 generated by @xmath534 .
consider the projection @xmath536 . by the preceding remark , @xmath537 is the natural candidate for @xmath538 to be a colored cone correponding to an affine embedding of @xmath40 .
let @xmath539 be generators of the cone @xmath386 .
then @xmath540 is generated by @xmath541 .
the cone @xmath530 is strictly convex , and thus the same holds for @xmath540 .
let us show that @xmath542 is non - void .
then @xmath543 will be a colored cone .
let @xmath544 .
we have : @xmath545 if @xmath546 is sufficiently small , as the set @xmath547 is contained in @xmath548 , we have @xmath549 it follows that @xmath550 and thus @xmath551 is a colored cone . in order to prove that @xmath551 corresponds to an affine embedding , we consider a basis @xmath552 of @xmath553 , where @xmath554 belongs to @xmath555 for all @xmath556 . by construction , @xmath557 .
we can then add the elements of @xmath558 to @xmath559 and obtain in this way a linearly independent set .
we complete this set in a basis @xmath560 of @xmath561 : @xmath562 define @xmath563 over @xmath564 as follows : @xmath565 and extend @xmath350 by linearity .
it is easy to see that @xmath350 satisfies the conditions a)c ) of proposition [ bijection_orbites_faces ] ( iv ) , so the embedding associated to @xmath566 is affine .
it remains to prove the unicity of the chosen set of colours .
let @xmath567 be a subset of colours such that @xmath568 is a colored cone corresponding to an affine embedding of @xmath40 .
let us show that @xmath569 .
let @xmath570 .
the vector @xmath571 can be written as @xmath572 let @xmath506 be as above . as @xmath573 for all @xmath574 , and @xmath575 over @xmath576 , it follows that @xmath577 .
the only possibility is that @xmath578 , and @xmath579 . by symmetry
, we get @xmath580 .
[ corolario ] let @xmath40 be a reductive group .
let @xmath1 and @xmath28 be two monoids with unit group @xmath40 , such that @xmath581
. then @xmath1 and @xmath28 are isomorphic .
in particular , if @xmath582 , then @xmath1 and @xmath28 are isomorphic .
+ let us give a geometric interpretation of the corollary above .
if @xmath1 is a reductive monoid with unit group @xmath40 and @xmath583 its colored cone , there exists a proper @xmath40-morphism @xmath584 , where @xmath153 is the embedding associated to @xmath585 ( @xcite ) .
moreover , the embedding @xmath153 is minimal for this property ; it is called the _ decoloration _ of @xmath1 .
corollary [ corolario ] says then that two monoids are isomorphic if and only if their decolorations are . in conclusion ,
we have proved the following classification : let @xmath40 be a reductive group .
the isomorphism classes of algebraic monoids with unit group @xmath40 are in bijection with the strictly convex polyhedral cones of @xmath524 generated by all the colours and a finite set of elements of @xmath448 .
we end up with some applications of the classification obtained above .
[ qwert ] let @xmath1 be a monoid with unit group @xmath40 .
then , @xmath587^{(b\times b^-)}=\left\ { f\in k(g)^{(b\times b^-)}\ : \ \chi_f\in { \mathcal c}(s)^\vee \right\}\ , \quad and\ ] ] @xmath588^{(b\times b^-)}=\left\ { f\in k(g)^{(b\times b^-)}\ : \ \chi_f\in { \mathcal c}(s)^\vee \cap \xi_+(g)\right\}\ .\ ] ] in particular , if @xmath408 , the decomposition of @xmath136 $ ] into simple @xmath46-modules is the following : @xmath588=\bigoplus _ { \chi\in { \mathcal c}(s)^\vee\cap \xi_+(g ) } v_\chi\otimes v_\chi^*\ .\ ] ] _ proof . _
the first equality follows from ( * ? ? ?
3.5 ) . in order to prove the second one ,
observe that a function @xmath591^{(b\times b^-)}$ ] extends to a regular function on @xmath1 if and only if @xmath592 for all @xmath593 . by proposition [ q(g ) ] , this means that @xmath594 , i.e. @xmath595 is a dominant weight .
let @xmath596 , where @xmath597 $ ] .
the @xmath46-module @xmath598 $ ] generated by @xmath599 is of finite dimension .
let @xmath600 , @xmath601 , be a system of generators of @xmath32 .
then , there exists a @xmath602-eigenvector @xmath603 , @xmath604 .
consider the @xmath602-eigenvector @xmath605 $ ] ; its weight is @xmath606 .
we have @xmath607 , where @xmath608^{(b\times b^-)}$ ] , and @xmath609 .
it follows that @xmath610 generates @xmath431 . by proposition [ prop : ideals - idempotents ] ,
the ideals of a reductive monoid @xmath1 are exactly the closed @xmath46-stables subvarieties of @xmath1 .
the dimension of an irreducible component of an ideal is then given by the dimension of the open orbit in this component . by proposition [ bijection_orbites_faces ] ( iii ) , to calculate those dimensions it suffices to calculate the stabilizers of the colours .
but the bruhat decomposition gives us this information immediately : [ lem : stabilisateurs - couleurs ] let @xmath40 be a reductive group .
if we keep the preceding notations , the stabilizer in @xmath41 of the colour @xmath438 , @xmath611 , is generated by the parabolic subgroups @xmath612 , @xmath613 , where @xmath614 denotes the minimal parabolic associated to the simple root @xmath426 .
+ apply proposition [ bijection_orbites_faces ] ( iii ) and then lemma [ lem : stabilisateurs - couleurs ] in order to calculate the intersection of the parabolic subgroups which stabilize the colours belonging to @xmath619 .
m. brion , _ sur la gomtrie des varits sphriques .
_ comment .
66 ( 1991 ) , 237262 .
r. hartshorne .
_ ample subvarieties of algebraic varieties .
notes in math 157 .
springer verlag , new york , 1970 .
j. humpreys , _ linear algebraic groups . _ gtm 21 , springer verlag , new york , 1970 . g. kempf , _ algebraic varieties .
_ london math .
notes 172 , cambridge univ . press , 1993 .
f. knop , _ the luna - vust theory of spherical embeddings .
_ in s. ramanan _ et al _ , editors , procedings of the hydebarad conference on algebraic groups , pages 225249 . national board for higher mathematics , manoj , 1991 .
d. mumford , j. fogarty and f. kirwan , _ geometric invariant theory .
_ modern surveys in math .
34 , springer verlag , new york , 3rd enlarged ed . , 1994 .
t. oda , _ torus embeddings and applications .
_ tata press , bombay , 1978 .
putcha , _ linear algebraic monoids . _ london math .
lect . notes 133 , cambridge univ .
press , 1988 .
notes 133 , l. renner , _ algebraic monoids .
thesis , univ . of british columbia ,
a. rittatore , _ monodes algbriques et varits sphriques .
thesis , institut fourier , grenoble , france , 1997 .
http://www-fourier.ujf-grenoble.fr e.b .
vinberg , _ on reductive algebraic semigroups .
, serie 2 , 169 ( 1994 ) , 145 182 .
lie groups and lie algebras .
dynkin seminar .
t. vust , _ plongements despaces symtriques algbriques : une classification . _
ann . scoula norm .
pisa , xvii , 2 ( 1990 ) , 165194 .
w. waterhouse , _ the unit group of affine algebraic monoids . _ proc .
85 ( 1982 ) 506 508 . | we study the geometry of algebraic monoids .
we prove that the group of invertible elements of an irreducible algebraic monoid is an algebraic group , open in the monoid .
moreover , if this group is reductive , then the monoid is affine .
we then give a combinatorial classification of reductive monoids by means of the theory of spherical varieties . |
with the atmospheres of exoplanets now accessible to astronomical scrutiny , there is motivation to understand the basic physics governing their structure .
since highly - irradiated exoplanets are most amenable to atmospheric characterization , a growing body of work has focused on hot earths / neptunes / jupiters , ranging from analytical models to simulations of atmospheric circulation ( e.g. , @xcite ) , in one , two and three dimensions ( 1d , 2d and 3d ) . the path towards full understanding requires the construction of a hierarchy of theoretical models of varying sophistication @xcite . in this context ,
analytical models have a vital role to play , since they provide crisp physical insight and are immune to numerical issues ( e.g. , numerical viscosity , sub - grid physics , spin - up ) .
atmospheres behave like heat engines .
sources of forcing ( e.g. , stellar irradiation ) induce atmospheric motion , which are eventually damped out by sources of friction ( e.g. , viscosity , magnetic drag ) .
it is essential to understand atmospheric dynamics , as it sets the background state of velocity , temperature , density and pressure that determines the spectral and temporal appearance of an atmosphere .
it also determines whether an atmosphere attains or is driven away from chemical , radiative and thermodynamic equilibrium .
even if an atmosphere is not in equilibrium , it must be in a global state of equipoise sources of forcing and friction negate one another ( e.g. , @xcite ) . in the present study , this is our over - arching physical goal : to analytically derive the global , steady state of an exoplanetary atmosphere ( the exoclime " ) in the presence of forcing , friction , rotation and magnetic fields .
-plane ( pseudo - spherical geometry ) and in full spherical geometry .
the key quantity to solve for is the meridional ( north - south ) velocity , from which the zonal ( east - west ) velocity , shallow water height perturbation and magnetic field perturbations straightforwardly follow . ]
[ fig : schematic ] shallow water models are a decisive way of studying exoclimes .
the term shallow water " comes from their traditional use in meteorology and oceanography @xcite and refers to the approximation that the horizontal extent modeled far exceeds the vertical / radial one .
they have been used to understand the solar tacocline @xcite , the atmospheres of neutron stars @xcite and exoplanetary atmospheres @xcite .
our over - arching technical goal is to perform a comprehensive theoretical survey across dimensionality ( 1d and 2d ) , geometry ( cartesian , pseudo - spherical and spherical ) and sources of friction ( molecular viscosity , rayleigh drag and magnetic drag ) for forced , damped , rotating , magnetized systems . to retain algebraic tractability , we study the limits of purely vertical / radial or horizontal / toroidal background magnetic fields .
our main finding is that the global structure of exoplanetary atmospheres is largely controlled by a single , key parameter at least in the shallow water approximation which we denote by @xmath0 . in the hydrodynamic limit , @xmath0
is directly related to the rossby and prandtl numbers . in forced ,
damped atmospheres with non - ideal mhd , @xmath0 encapsulates the effects of molecular viscosity , rayleigh drag , forcing , magnetic tension and magnetic drag . due to the technical nature of the present study
, we find it useful to concisely summarize a set of terminology that we will use throughout the paper . from a set of perturbed equations
, we will obtain wave solutions for the velocity , height and magnetic field perturbations .
we assume that the waves have a temporal component of the form @xmath1 , where @xmath2 is the wave frequency .
generally , the wave frequency has real and imaginary components ( @xmath3 ) , which describe the oscillatory and growing or decaying parts of the wave , respectively . for each model
, we will obtain a pair of equations describing @xmath4 and @xmath5 , which we call the oscillatory dispersion relation " and the growth / decay dispersion relation " , respectively
. we will refer to them collectively as dispersion relations " .
a balanced flow " corresponds to the situation when @xmath6 .
the steady state " of an atmosphere has @xmath7 .
we will refer to molecular viscosity , rayleigh drag and magnetic drag collectively as friction " .
if we are only considering molecular viscosity and rayleigh drag , we will use the term hydrodynamic friction " .
we will use the terms friction " and damping " interchangeably , but not qualify the latter as being hydrodynamic or mhd in nature .
we will refer to systems as being free " if forcing and friction are absent .
similarly , we will use the terms fast " and rapid " interchangeably when referring to rotation .
the zeroth order effect of including a magnetic field introduces a term each into the momentum and magnetic induction equations we will refer to the effects of these ideal - mhd terms as magnetic tension " .
when non - ideal mhd is considered , we include a resistive term in the induction equation that mathematically resembles diffusion we will refer to its influence as magnetic drag " . we will examine an approximation for including the effects of non - constant rotation , across latitude , on a cartesian grid , known traditionally as the @xmath8-plane approximation " ( e.g. , @xcite and references therein ) .
note that this is _ not _ the same as a departure from solid body rotation .
rather , it is an approximation to include the dynamical effects of sphericity .
there are two flavors of this approximation : the simpler version solves for waves that are oscillatory in both spatial dimensions ( simply called @xmath8-plane " ) , while a more sophisticated version allows for an arbitrary functional dependence in latitude ( equatorial @xmath8-plane " ) .
since the equatorial @xmath8-plane treatment more closely approximates the situation on a sphere , we will refer to it as being pseudo - spherical " . in constructing the mathematical machinery in this paper
, we often have to evaluate long , complex expressions . to this end
, we find it convenient to separate out the real and imaginary components using a series of separation functions " , which we denote by @xmath9 , @xmath10 , @xmath11 , @xmath12 , @xmath13 and @xmath14 .
the definitions of these dimensionless quantities vary from model to model .
we note that @xmath14 also functions like a generalized friction that includes magnetic tension .
in [ sect : equations ] , we state the governing equations and derive their linearized , perturbed forms . in [ sect:1d ] , we review and extend the 1d models .
we extend our models to 2d cartesian geometry in [ sect:2d ] and begin to consider the effects of sphericity in [ sect:2d_beta ] . in [ sect : spherical ] , we present results for 2d models in spherical geometry .
applications to exoplanetary atmospheres are described in [ sect : apps ] .
we summarize our findings in
[ sect : discussion ] .
table 1 lists the most commonly used quantities and the symbols used to denote them .
table 2 compares our study with previous analytical work .
table 3 summarizes the salient lessons learnt from studying each shallow water model .
figure [ fig : schematic ] provides a graphical summary of our technical achievements .
[ tab : symbols ] .commonly used symbols [ cols="<,^,^",options="header " , ] + note : all 2d models include rotation .
+ hd : hydrodynamic .
mhd : magnetohydrodynamic .
some useful , commonly used expressions include @xmath15 the expression involving @xmath16 makes the approximation that @xmath17 . | within the context of exoplanetary atmospheres , we present a comprehensive linear analysis of forced , damped , magnetized shallow water systems , exploring the effects of dimensionality , geometry ( cartesian , pseudo - spherical and spherical ) , rotation , magnetic tension and hydrodynamic and magnetic sources of friction . across a broad range of conditions , we find that the key governing equation for atmospheres and quantum harmonic oscillators are identical , even when forcing ( stellar irradiation ) , sources of friction ( molecular viscosity , rayleigh drag and magnetic drag ) and magnetic tension are included .
the global atmospheric structure is largely controlled by a single , key parameter that involves the rossby and prandtl numbers .
this near - universality breaks down when either molecular viscosity or magnetic drag acts non - uniformly across latitude or a poloidal magnetic field is present , suggesting that these effects will introduce qualitative changes to the familiar chevron - shaped feature witnessed in simulations of atmospheric circulation .
we also find that hydrodynamic and magnetic sources of friction have dissimilar phase signatures and affect the flow in fundamentally different ways , implying that using rayleigh drag to mimic magnetic drag is inaccurate .
we exhaustively lay down the theoretical formalism ( dispersion relations , governing equations and time - dependent wave solutions ) for a broad suite of models . in all situations ,
we derive the steady state of an atmosphere , which is relevant to interpreting infrared phase and eclipse maps of exoplanetary atmospheres .
we elucidate a pinching effect that confines the atmospheric structure to be near the equator .
our suite of analytical models may be used to decisively develop physical intuition and as a reference point for three - dimensional , magnetohydrodynamic ( mhd ) simulations of atmospheric circulation . |
we study the cosmic microwave background ( cmb ) anisotropies measured by wmap @xcite from the point of view of the 3-point angular correlation @xmath17 .
current models for structure formation predict that the initial seeds , that later give rise to the observed structure in the universe , were gaussian . if so , one would expect the angular 3-point cmb temperature correlations to be almost zero . even for a gaussian field ,
any non - linear processing of the primordial perturbations would lead to small non - gaussianities . for an excellent review see @xcite and references therein .
angular correlations have some advantages over spherical harmonic analyses .
the main disadvantage is that different angular bins are highly correlated , which means that we need to use the full covariance matrix to assess the significance of any departures between measurements and models .
the advantage of using the n - point function is that correlations can be easily sampled from any region of the sky , even when a foreground mask is required , provided the region is large enough . in comparison , harmonic decomposition breaks the angular symmetry in the sky into orthogonal bases which have an arbitrary phase orientation .
this is not a problem in the case of full sky coverage , but the results will be coordinate - dependent if a mask is used
. moreover , higher order correlations , such as the bispectrum , have to focus on some small portion of the mutidimensional space of triangle configurations ( see @xcite ) . the formal equivalence between the bispectrum and the 3-point function is hardly realized in practice and as such it is of interest to study both separately .
our analysis procedure is similar to that presented in @xcite , to which the reader can refer for further details .
we use the same simulations where we ve generated maps using the standard model and a low - q model , a fiducial variation of the model with low values of the quadrupole and octopole .
we concentrate our analysis on the v - band wmap maps ( fwhm = 21 at 61ghz ) generated using healpix @xcite with pixel resolution n=512 ( 7 pixels ) and to which we apply the conservative wmap kp0 foreground mask @xcite , the resulting maps have approximately 2.4 million pixels .
the 2-point angular correlation function is defined as the expectation value or mean cross - correlation of temperature fluctuations : ( ) at two positions @xmath18 and @xmath19 in the sky : w_2 ( ) , where @xmath20 , assuming that the distribution is statistically isotropic .
we normalize fluctuations to be dimensionless , @xmath21k , or dimensional with @xmath22 . to estimate @xmath23 from the pixel maps we use : w_2 ( ) = _ i , j _ i _ j w_i w_j , where @xmath24 and @xmath25 are the observed temperature fluctuations on the sky and the sum extends to all pairs @xmath26 separated by @xmath27 .
the mean temperature fluctuation is subtracted so that @xmath28 .
the weights @xmath29 can be used to minimize the variance when the pixel noise is not uniform , however this introduces larger cosmic variance . here
we follow the wmap team and use uniform weights ( i.e. @xmath30 ) . at the largest scales we use bins @xmath31 whose size is proportional to the square root of the angle @xmath32 and use linear bins of @xmath33 degrees at the smallest scales . in a similar fashion , the 3-point angular correlation function is defined as the cross - correlation of temperature fluctuations at 3 positions @xmath18 , @xmath19 and @xmath34 on the sky : w_3(_12 , _ 13 , _ 23 ) , where @xmath35 .
consider a triplet of pixels with labels 1,2,3 on the sky .
let @xmath36 and @xmath37 be the angular separations between the corresponding pairs of pixels and @xmath38 the interior angle between these two sides of the triangle .
assuming that the distribution is statistically isotropic , @xmath17 only depends on the sides of the triangles that define the 3 positions on the sky .
one can characterize the configuration - dependence of the three - point function by studying the behavior of @xmath39 for fixed @xmath36 and @xmath37 . to estimate @xmath40 from the pixel maps we use : w_3(_12 , _ 13 , _ 23 ) = _ i , j , k _ i _ j _ k
w_i w_j w_k , where @xmath24 , @xmath25 and @xmath41 are the temperature differences in the map and the sum extends to all triplets separated by @xmath42 , @xmath43 and @xmath44 .
we use same bins and weights as in the 2-point function above .
the 3-point function can either be normalized to a dimensionless amplitude : d_3 where @xmath45 , or a hierarchical amplitude : q_3 .
the covariance matrix between different angular bins can be calculated from the simulations : c_ij & & w_2(_i ) w_2(_j ) + & = & 1 _ l=1^n w_2^l(_i ) w_2^l(_j ) , where @xmath46 . here
@xmath47 is the 2-point or 3-point function measured in the @xmath48-th realization ( @xmath49 ) and @xmath50 is the mean value for the @xmath51 realizations .
the case @xmath52 gives the error variance : @xmath53 .
an estimate of the errors for correlations measured on the real sky can be obtained using a variation of the jackknife error scheme .
this has the potential advantage of producing an error estimate which is model independent .
the accuracy of the jack - knife covariance have been tested for both wmap @xcite and the apm and sdss survey @xcite , @xcite , @xcite . in the estimation ,
the sample is first divided into @xmath54 separate regions on the sky , each of equal area . the analysis is then performed @xmath54 times , on each occasion removing a different region .
these are called the ( jackknife ) subsamples , which we label @xmath55 .
the estimated statistical covariance for @xmath56 at scales @xmath57 and @xmath58 is then given by : c_ij & = & m-1 _ k=1^m w_2^k(_i ) w_2^k(_j ) + & & w_2^k(_i ) w_2^k(_i ) - ( _ i ) where @xmath59 is the measure 2 or 3-point function in the @xmath60-th subsample ( @xmath61 ) and @xmath62 is the mean value for the @xmath54 subsamples .
the case @xmath52 gives the error variance .
note how , if we increase the number of regions @xmath54 , the jackknife subsamples are larger and each term in the sum is smaller .
we typically take @xmath63 corresponding to a division of the sphere into 8 octants .
note that the 3-point correlation function is in general a function of 3 variables .
we will fix two of these variables and show variations of @xmath17 or @xmath6 as a function of a single variable , for which we estimate the covariance with the above prescription .
estimation of correlations in configuration space is quite time consuming .
a brute force algorithm to estimate all pair separations in a map with @xmath51 pixels takes @xmath64 operations .
to do all triplets in eq.[[w3est ] ] we need @xmath65 operations . in the healpix @xmath66 wmap maps ( 7 arcmin pixel s )
we have @xmath67 ( with the kp0 mask galactic cut ) so a brute force analysis requires @xmath68 operations , well beyond current computing power .
we overcome this difficulty in two steps . for large scales , @xmath69 degree , we reduce the resolution of the maps to @xmath70 ( 1 degree resolution ) by averaging fluctuations over nearby pixels .
the resulting correlation function contains the same information at scales larger than the new resolution scale .
now the number of pixels is reduced to @xmath71 and the number of operations to @xmath72 , which lies within current computer power ( teraflops ) . for scales smaller than a few degrees
we can reduce the number of operations by precalculating the distance between neighboring pixels .
the number of operations is then reduced to @xmath73 , where @xmath74 is the number of neighbors . in our case
we have @xmath75 for pixels around @xmath76 degrees , and thus @xmath77 as in the case for the larger angles . in our implementation
we have an overlap between the large and small scale estimations at around @xmath78 degrees .
the agreement between both estimates here is quite good for both @xmath79 and @xmath17 . in the case of the 2-point function
we have compared both estimates in the whole range @xmath80 degrees and find excellent agreement .
3.3 in 3.3 in results are presented in figures [ w3co ] and [ q3al ] . at large scales fig.[w3co ] shows the results of @xmath17 in units of @xmath81 ( eg @xmath82 in eq.[[delta ] ] ) .
note that @xmath83 at small scales , so that @xmath84 in these units ( see eq.[[d3 ] ] ) .
errorbars in both cases are from jackknife subsamples and agree well with errors estimated from the simulations ( see gaztaaga etal 2003 for more details ) .
the results for the equilateral and collapsed triangle configurations can be compared to the cobe analysis presented in fig.1 of @xcite .
the pseudo - collapsed case in @xcite would correspond here to @xmath85 here .
we can see how the cobe and wmap results are in good agreement where the two curves even see the very same bumps and valleys .
this is not totally surprising given the very good match between cobe and wmap temperatures maps , pixel by pixel , when smoothed on large scales .
the top panel in figure [ w3co ] also shows two new triangle configurations : one that is elongated , @xmath86 ( short dashed line ) , and a wide triangle , @xmath87 ( continuous line ) .
the results for intermediate angular configurations are similar . in all cases
the agreement with the gaussian prediction @xmath88 is excellent , given the strong covariance . at small scales
we show the 3-point function in terms of the reduced amplitude @xmath6 in eq.[[q3 ] ] where the best constraints are found on the smallest scales . as can be seen in the figure ,
as we approach triangles with side @xmath89degree , the cosmic errors become large @xmath90 .
we have fit the data to a constant value of @xmath6 , independent of scale using a @xmath91-test : ^2 = _ i , j=1^n _ i c_ij^-1 _ j , where @xmath92 is the difference between the `` observation '' @xmath93 and the @xmath94 value . in order to eliminate the degeneracies in the covariance matrix , we perform a singular value decomposition ( svd ) of the covariance matrix ( see gaztaaga 2003 for more details ) .
a joint fit with the jackknife covariance matrix for collapsed configurations with @xmath95 degree , yields : q_3^collapsed = 19 141 .
the best signal - to - noise is found for the non - collapsed triangle configurations with @xmath96 degree we find : q_3 = 336 218 ; = 0.5 deg providing a marginal detection at about 90% confidence ( shown as dashed line in fig.[q3al ] ) .
this fit corresponds to the jackknife covariance matrix .
a similar result is found for the covariance matrix in the low - q simulations .
the errors increase by almost a factor of two when using the covariance of the standard simulations .
this increase is due to the larger cosmic variance introduced by the quadrupole and octopole which are larger than in the wmap data ( see @xcite , citepgaz03 for a discussion ) .
our bounds on @xmath13 seem marginally better than the ones presented in @xcite from the wmap bispectrum , but a direct comparison is not straight forward .
first of all , our modelling provides a measurement of non - linearities in the final @xmath107 while @xcite provide a modelling for non - linear primordial curvature perturbations @xmath14 , before radiation transfer .
thus while limits on @xmath6 are model independent , direct limits on @xmath13 depend on the assumed model for radiation transfer . secondly , the bispectrum configurations used by @xcite are a subset of all possible triangle configurations so that both approaches are neither equivalent nor optimized in the same way . in both cases ,
errors are dominated by cosmic variance , and they begin to be small enough to constrain the leading order non - linear effects with coupling @xmath13 which are of order unity .
eg acknowledged support from inaoe , the spanish ministerio de ciencia y tecnologia , project aya2002 - 00850 , ec - feder funding and the supercomputing center in barcelona ( cepba and cesca / c4 ) , where part of these calculations were done .
jw thanks eric hivon for providing assistance during the installation of healpix and acknowledges support from a conacyt grant ( 39953-f ) and a scholarship from inaoe .
grski , k. m. , hivon , e. , & wandelt , b. d. , 1999 , in proceedings of the mpa / eso cosmology conference `` evolution of large - scale structure '' , eds .
banday , r.s .
sheth and l. da costa , printpartners ipskamp , nl , pp .
37 - 42 ( also astro - ph/9812350 ) . | we present a study of the 3-pt angular correlation function @xmath0 of ( adimensional ) temperature anisotropies measured by the wilkinson microwave anisotropy probe ( wmap ) .
results can be normalized to the 2-point function @xmath1 in terms of the hierarchical : @xmath2 or dimensionless : @xmath3 amplitudes .
strongly non - gaussian models are generically expected to show @xmath4 or @xmath5 .
unfortunately , this is comparable to the cosmic variance on large angular scales . for gaussian primordial models
, @xmath6 gives a direct measure of the non - linear corrections to temperature anisotropies in the sky : @xmath7 with @xmath8 for the leading order term in @xmath9 .
we find good agreement with the gaussian hypothesis @xmath10 within the cosmic variance of simulations ( with or without a low quadrupole ) .
the strongest constraints on @xmath6 come from scales smaller than 1 degree .
we find @xmath11 for ( pseudo ) collapsed configurations and an average of @xmath12 for non - collapsed triangles .
the corresponding non - linear coupling parameter , @xmath13 , for curvature perturbations @xmath14 , in the sachs - wolfe ( sw ) regime is : @xmath15 , while on degree scales , the extra power in acoustic oscillations produces @xmath16 in the model .
errors are dominated by cosmic variance , but for the first time they begin to be small enough to constrain the leading order non - linear effects with coupling of order unity . |
the current status of neutrino oscillation parameter estimations can be very briefly summarized @xcite as follows : atmospheric ( solar ) neutrinos oscillate with @xmath1 ev@xmath2 and @xmath3 @xcite ( @xmath4 ev@xmath2 , @xmath5 @xcite ) ( @xmath6 ) is the solar ( atmospheric ) mass - squared difference and @xmath7 , @xmath0 and @xmath8 are the mixing angles conventionally denoted by @xmath9 , @xmath10 and @xmath11 , respectively @xcite . ] .
all we presently know about @xmath0 is that @xmath12 at the 2@xmath13 c. l. @xcite .
a long - standing hope is that neutrinos from a core - collapse supernova ( sn ) may shed light on two of the unknown oscillation parameters , sgn(@xmath6 ) and @xmath0 .
only a handful of neutrinos from a type ii sn have ever been detected .
the detection of 11 neutrinos from sn 1987a in kamiokande ii @xcite and 8 neutrinos in the irvine michigan brookhaven experiment @xcite have been of great importance for understanding core - collapse @xcite .
it is evident that the physics potential offered by a future galactic sn event is immense .
with cognizance of this potential , experiments dedicated to sn neutrino detection have been proposed @xcite even though only a few galactic sn are expected per century .
attempts have been made to extract neutrino oscillation parameters from the 19 sn 1987a events .
however , conclusions drawn from such analyses are highly dependent on the neutrino flux model adopted and are far from robust . for example ( and within the context of this paper )
, it was claimed that the data favor the normal hierarchy ( @xmath14 ) over the inverted hierarchy ( @xmath15 ) provided @xmath16 @xcite , but this conclusion was contradicted in ref .
@xcite .
neutrinos from a galactic sn could in principle provide a wealth of information on neutrino oscillations .
a determination of @xmath17 and the neutrino mass hierarchy from sn neutrinos is unique in that ambiguities @xcite arising from the unknown @xmath18 phase @xmath19 and the deviation of atmospheric neutrino mixing from maximality do not corrupt it .
the absence of the eight - fold parameter degeneracies that are inherent in long baseline experiments @xcite results because ( i ) nonelectron neutrino fluxes and @xmath20 interactions , which can not distinguish between the different nonelectron neutrino species ( that we denote by @xmath21 with @xmath22 ) . ] do not depend on @xmath19 independently of neutrino conversion @xcite , and so sn neutrinos directly probe @xmath17 , and ( ii ) whether atmospheric mixing is maximal or not is immaterial since @xmath8 does not affect the oscillation dynamics .
investigations of the effect of neutrino oscillations on sn neutrinos in the context of a static density profile ( _ i.e. , _ neglecting shock effects ) have been made in refs .
@xcite . whether or not the mass hierarchy can be determined and @xmath0 be constrained depends sensitively on the strength of the hierarchy between @xmath23 and @xmath24 .
the higher @xmath25 is above unity , the better the possible determinations @xcite .
unfortunately , modern sn models that include all relevant neutrino interaction effects like nuclear recoil and nucleon bremsstrahlung indicate that the hierarchy of average energies is likely smaller than expected from traditional predictions ; @xmath25 is expected to be about 1.1 , and no larger than 1.2 @xcite as opposed to ratios above 1.5 from older sn codes @xcite .
another relevant uncertainty is that different sn models predict different degrees to which equipartitioning of energy between @xmath26 , @xmath20 and @xmath21 is violated .
for example , in ref .
@xcite an almost perfect equipartitioning is obtained while according to refs .
@xcite , equipartitioning holds only to within a factor of 2 .
given these uncertainties , it is not a simple task to determine @xmath0 and the mass hierarchy simultaneously from sn data @xcite . a significant improvement would be a determination of the mass hierarchy independently of predictions for @xmath25 and equipartitioning from sn models . in this paper
we propose a new method using sn neutrinos to determine the mass hierarchy that exploits recent advances in the understanding of shock propagation in sn . at densities
@xmath27 g/@xmath28 , neutrino oscillations are governed by @xmath29 and @xmath30 @xcite .
neutrinos ( antineutrinos ) pass through a resonance if @xmath31 ( @xmath32 ) . as the shock traverses the resonance , adiabaticity
is severely affected causing oscillations to be temporarily suppressed , as first pointed out in ref . @xcite .
after the shock moves beyond the resonance , oscillations are restored .
then one expects a dip in the time evolution of the average neutrino energy and the number of events .
this modulation is visible in the neutrino ( antineutrino ) channel for a normal ( inverted ) mass hierarchy and only if @xmath33 _
i.e. _ , only for oscillations that would occur adiabatically for a static density profile .
previous work exploiting the dip to obtain information about oscillation parameters can be found in ref .
@xcite . within the first 4 seconds or so ,
the violation of adiabaticity caused by the shock is felt only by neutrinos with energy less than about 20 mev . at these energies
some models predict the @xmath21 flux to be larger than the @xmath26 and @xmath20 fluxes @xcite and others predict the converse @xcite . at later times
, the shock affects higher energy neutrinos for which all models predict the @xmath21 flux to be dominant .
the dip is observable even for @xmath34 because the fluxes are flavor - dependent @xcite .
thus , a signature in high energy neutrinos a few seconds after bounce is quite model - independent . through our analysis
we show that the signal is so robust that a restriction to high - energy events is unnecessary .
we investigate the significance with which the mass hierarchy can be determined by measurement of the @xmath26 spectrum at a 100 kton liquid argon detector and the @xmath20 spectrum at a 1 mton water cherenkov detector from a galactic sn at a distance of 10 kpc with binding energy @xmath35 ergs .
the number of unoscillated events in the liquid ar ( water cherenkov ) detector is expected to be @xmath36 ( @xmath37 ) . (
although a liquid argon detector can measure both the @xmath26 and @xmath20 spectra , it is an order of magnitude more sensitive to the @xmath26 flux than to the @xmath20 flux @xcite ) .
we are interested in the detectability of a dip in the time evolution of either the @xmath26 or the @xmath20 spectrum , but not both . by correlating the two spectra
, it should be possible to establish the mass hierarchy and that @xmath33 .
the presence of a dip will also provide further evidence that sn simulations correctly depict shock propagation .
if a dip is not found in either channel , it would suggest that @xmath38 , since the salient features of the theory of core - collapse sn have already received strong support from sn 1987a @xcite .
realistic time - dependent density profiles of sn are obtained from detailed numerical siumulations . as one moves in towards the neutron star , the profile at a given instant
has a sharp density rise followed by a rarefaction region where the density can drop significantly below that at the outer edge of the shock .
the authors of ref .
@xcite have provided a generic time - dependent density profile that mimics those of supernova simulations .
the shock front is steepened artificially to reintroduce the physical requirement of a density discontinuity which is lost in hydrodynamic simulations due to the limited ( few 100 km ) spatial resolution .
we adopt the empirical parameterization of this profile ( which is continuous in the supernova radius and time ) developed in ref .
@xcite .
because of the dip in density in the rarefaction region , neutrinos may hop between mass eigenstates up to 3 times before leaving the sn envelope . under the assumptions that the transitions factorize and
the neutrino phases can be averaged away , a simple analytic expression for the overall hopping probability @xmath39 was obtained in ref .
@xcite , which agrees remarkably well with phase - averaged results of runge - kutta evolution of the neutrino flavor propagation equations .
we employ the analytic expression for @xmath40 to calculate the @xmath26 and @xmath41 survival probabilities .
these survival probabilities do not depend on @xmath42 since we are neglecting earth matter effects @xcite : @xmath43 here , the @xmath44 and @xmath45 subscripts denote normal and inverted hierarchy , respectively .
we note that the factorization of the 3-neutrino dynamics into two 2-neutrino subsystems continues to hold with the subsystem governed by @xmath42 and @xmath7 remaining adiabatic ( for the now well - established large mixing angle solution @xcite ) as in the case of a static density profile @xcite .
we use the parameterization of ref .
@xcite for the primary unoscillated neutrino spectra given by @xmath46 where @xmath47 , @xmath41 , @xmath21 , and @xmath48 ( @xmath49 ) is the average energy ( dimensionless shape parameter that quantifies the width of the spectrum ) of species @xmath50 . in principle , both @xmath48 and @xmath49 can be time - dependent . throughout ,
we assume @xmath51 , @xmath52 and @xmath53 @xcite . here
, @xmath54 is the neutrino emission rate or luminosity of species @xmath50 .
the only available simulation that tracks neutrino emission for times long enough to enable studies of shock effects on neutrino oscillations is that of the livermore group .
we adopt the luminosities of ref .
@xcite .
our goal is to demonstrate the model - independence of our new method . to this end
, we consider models that span a wide range of predictions for the energy spectra .
we conservatively consider the following parameter ranges for the initial spectra ( indicated by the `` 0 '' superscript ) in our analyses ( with all energies in mev ) : @xmath55 @xmath56 this is in line with the hierarchy of average energies @xmath57 , expected from the well - known interaction strengths of neutrinos with matter ; the species which interacts less decouples earlier and has a higher temperature .
we will occasionally refer to a specific model which appears to be quite pessimistic for determining oscillation parameters because the different neutrino species have very small spectral differences .
it is a model obtained by the garching group by accounting for all relevant neutrino interaction effects .
the initial spectra predicted by the garching model have @xcite @xmath58
for the 1 mton hyper - kamiokande detector @xcite we assume an energy threshold of 7 mev and a fiducial volume for sn neutrinos of 685 kton which is consistent with the fiducial to total volume ratio expected for the uno detector @xcite .
we only consider events from inverse @xmath59-decay and use the higher - order cross section of ref .
we assume that the energy resolution is that of the super - kamiokande detector @xcite . for the 100 kton liquid ar detector @xcite with near - perfect efficiency , we consider the absorption process , @xmath60 , whose cross section is calculated in ref .
the effective threshold is 6 mev .
since the energy resolution is expected to be extremely good , we do not include resolution effects . for each mass hierarchy and several values of @xmath0 , we randomly choose sn models defined by the parameter ranges of eqs .
( [ energyranges],[lumranges ] ) and simulate @xmath61 possible @xmath26 and @xmath41 spectra at the two detectors ( including statistical fluctuations ) .
we fix @xmath62 @xcite and @xmath63 ev@xmath2 @xcite .
we ignore earth matter effects ( thereby not committing ourselves to specific zenith angles for the sn ) which only become comparable to shock effects at very small @xmath0 .
we illustrate the effect of the shock on the neutrino spectra in fig .
[ fig : spectra ] .
( solid green curves ) and @xmath41 spectra ( dashed red curves ) for the normal hierarchy and inverted hierarchy , respectively with @xmath64 . the thick ( thin ) lines correspond to @xmath65 ( @xmath48 that fall linearly 5 secs after the core - bounce ) .
the curves are based on the ( pessimistic ) garching model of eq.([garching ] ) .
note the pronounced dips in the @xmath26 ( @xmath41 ) spectra for the normal ( inverted ) hierarchy .
[ fig : spectra],width=576 ]
we consider two cases ( a ) @xmath65 and ( b ) with @xmath48 falling linearly with time 5 secs post - bounce according to @xmath66 .
we have chosen this parametrization to agree with that in ref . @xcite .
although this fall - off in energy with time is arbitrary , it serves to demonstrate the robustness of our analysis to specific assumptions about @xmath67 .
after all , in a real sn , nature will make a unique selection for @xmath67 which is as yet unknown to us .
so long as @xmath68 , we expect a dip in the @xmath26 ( @xmath41 ) spectrum if the hierarchy is normal ( inverted ) as in fig .
[ fig : spectra ] . a way to establish the channel in which the dip occurs exploits the fact that the dip is expected to occur 4 seconds after the shock forms ( _ i.e. _ after the core rebounds ) .
we split each spectrum into 2 time bins .
all the events occuring in the first 4 seconds constitute the `` early time bin '' , with mean energy @xmath69 and the events occuring between 4 and 10 seconds constitute the `` late time bin '' , with mean energy @xmath70 .
their ratio is @xmath71 in the case that @xmath65 , for the normal hierarchy we expect @xmath72 to be smaller than unity and @xmath73 to be close to unity and vice - versa for the inverted hierarchy . if @xmath74 , we expect both ratios to be close to unity . for falling @xmath48 ,
we expect both @xmath72 and @xmath73 to be slightly below unity even if @xmath74 .
this is because @xmath48 starts falling 5 seconds after bounce , which lowers @xmath75 relative to @xmath69 .
this shock - independent suppression can not be large since even at @xmath76 secs , @xmath77 .
( clearly the size of the effect depends on how fast @xmath48 falls ) .
then , the hierarchy is deduced by comparing the relative deviations of @xmath78 from unity . if @xmath79 , the hierarchy is inverted . if @xmath80 , the hierarchy is normal . in fig .
[ fig : rat]a , we plot @xmath72 vs. @xmath73 for the normal hierarchy ( red squares ) and for the inverted hierarchy ( green triangles ) with @xmath64 assuming @xmath65 .
the blue dots are for @xmath81 .
there is little overlap between the clusters , and the inverted hierarchy can be established at more than 3@xmath13 . as can be seen in fig .
[ fig : rat]b , the case in which @xmath48 fall linearly also yields little overlap between the clusters , although the clusters are now systematically displaced to lower values of @xmath72 and @xmath73 . vs. @xmath73 for sn models spanned by eqs .
( [ energyranges],[lumranges ] ) and @xmath64 assuming @xmath65 .
@xmath78 is the ratio of the mean energy of the events occuring between 4 and 10 seconds to the mean energy of the events occuring in the first 4 seconds .
the red squares ( green triangles ) correspond to the normal ( inverted ) hierarchy .
the blue dots are for @xmath81 .
each contour contains 95% of the @xmath82 @xmath26 and @xmath41 spectra generated for the corresponding case .
we have plotted only @xmath83 triangles , squares and dots so as to not overwhelm the figure .
( b ) is the analog of ( a ) but for the case in which @xmath48 falls linearly with time 5 secs after bounce .
the separation of the clusters is a measure of the significance with which the hierarchy can be determined .
[ fig : rat],width=576 ] we now focus on the @xmath65 case .
it is no surprise that it is more difficult to identify the normal hierarchy considering the lower statistics in the ar detector .
however , it is possible to establish the normal hierarchy reasonably well . in fig .
[ fig : sig]a the dashed red curve shows the fraction of spectra corresponding to the normal hierarchy and @xmath64 with @xmath72 below a given value .
for example , 88% ( 98.7% ) of the spectra corresponding to the normal hierarchy and @xmath64 have @xmath84 ( 0.99 ) .
the solid blue curve shows the number of spectra with @xmath81 and @xmath72 below a given value divided by the number for the normal hierarchy and @xmath64 with @xmath72 below the same value .
thus from fig . [ fig : sig]a , the fraction of spectra with @xmath81 that mimic the spectra involving shocks , and have @xmath84 ( 0.99 ) , is 0.33% ( 7.6% ) . said differently , 88% ( 98.7% ) of the spectra indicate the normal hierarchy correctly with a probability of 99.67% ( 92.4% ) .
we emphasize that these results have included the garching model of eq .
( [ garching ] ) .
as expected , the results of a similar analysis of @xmath85 spectra simulated from the garching model alone are somewhat worse .
specifically , we find that 87% ( 99% ) of the spectra simulated with the normal hierarchy and @xmath86 have @xmath84 ( 0.99 ) , and identify the normal hierarchy with a confidence of 99% ( 87% ) .
it is remarkable that even for initial spectra with tiny differences , such as in the garching model , it is possible to establish the normal hierarchy .
figure [ fig : sig]b is similar to [ fig : sig]a , but for the inverted hierarchy .
clearly , a determination of the inverted hierarchy is easier .
we see that 99.5% of the spectra simulated with an inverted hierarchy and @xmath86 have @xmath87 , and indicate the inverted hierarchy at the 99.99% c. l. the case in which @xmath48 fall linearly with time 5 seconds after bounce gives similar results as those for the constant @xmath48 case .
we have also investigated if the hierarchy can be determined for @xmath88 .
it is still possible to establish the the inverted hierarchy with high - confidence . for the normal hierarchy
we find that while shock effects remain visible , it is no longer possible to establish the normal hierarchy with high - significance .
our analysis has established that detections of shock modulations in either the @xmath26 spectrum in a kton - class liquid ar detector or in the @xmath41 spectrum in a mton - class water cherenkov detector will provide definitive evidence for the true neutrino mass hierarchy and that @xmath68 . we emphasize that the modulations should be observed in one and only one of the two channels .
a similar analysis can be performed supposing that a reverse shock also develops in the sn envelope @xcite .
then , one expects to see a double - dip signature in either the @xmath26 or @xmath41 spectrum if @xmath0 is not too small @xcite .
in addition to probing oscillation parameters , a presence of a double dip will provide confirmation that sn simulations correctly depict shock propagation .
we have supposed that both a large water cherenkov and large liquid argon detector are available . a possible alternative to a large water cherenkov detector
is the icecube detector since good energy resolution is not a requirement of our analysis .
then @xmath89 can be compared with @xmath72 @xcite .
an aspect that bears further consideration is whether our analysis is robust under variations of the shock density profile .
we are currently investigating this issue in ref .
we thank h .- t .
janka , g. raffelt and r. tomas for useful discussions and for providing us with data on profiles and luminosities .
this research was supported by the u.s .
department of energy under grant no .
de - fg02 - 95er40896 , by the nsf under grant no .
eps-0236913 , by the state of kansas through kansas technology enterprise corporation and by the wisconsin alumni research foundation .
v. barger , d. marfatia and k. whisnant , int .
j. mod .
e * 12 * , 569 ( 2003 ) [ arxiv : hep - ph/0308123 ] .
m. ishitsuka [ super - kamiokande collaboration ] , arxiv : hep - ex/0406076
. t. araki _ et al .
_ , [ kamland collaboration ] , arxiv : hep - ex/0406035 . m. apollonio _ et al . _
[ chooz collaboration ] , phys .
b * 466 * , 415 ( 1999 ) [ arxiv : hep - ex/9907037 ] .
k. hirata _ et al . _
[ kamiokande - ii collaboration ] , phys .
* 58 * , 1490 ( 1987 ) ; k. s. hirata _ et al .
_ , phys . rev .
d * 38 * , 448 ( 1988 ) .
r. m. bionta _ et al .
_ , phys .
lett . * 58 * , 1494 ( 1987 ) ; c. b. bratton _ et al . _
[ imb collaboration ] , phys .
d * 37 * , 3361 ( 1988 ) .
t. j. loredo and d. q. lamb , phys .
d * 65 * , 063002 ( 2002 ) [ arxiv : astro - ph/0107260 ] . c. k. jung , arxiv : hep - ex/0005046 . h. minakata and h. nunokawa , phys .
b * 504 * , 301 ( 2001 ) [ arxiv : hep - ph/0010240 ] .
v. barger , d. marfatia and b. p. wood , phys .
b * 532 * , 19 ( 2002 ) [ arxiv : hep - ph/0202158 ] .
m. koike , t. ota and j. sato , phys .
d * 65 * , 053015 ( 2002 ) [ arxiv : hep - ph/0011387 ] ; j. burguet - castell , m. b. gavela , j. j. gomez - cadenas , p. hernandez and o. mena , nucl .
b * 608 * , 301 ( 2001 ) [ arxiv : hep - ph/0103258 ] ; h. minakata and h. nunokawa , jhep * 0110 * , 001 ( 2001 ) [ arxiv : hep - ph/0108085 ] ; g. l. fogli and e. lisi , phys . rev .
d * 54 * , 3667 ( 1996 ) [ arxiv : hep - ph/9604415 ] .
v. barger , d. marfatia and k. whisnant , phys .
d * 65 * , 073023 ( 2002 ) [ arxiv : hep - ph/0112119 ] .
e. k. akhmedov , c. lunardini and a. y. smirnov , nucl .
b * 643 * , 339 ( 2002 ) [ arxiv : hep - ph/0204091 ] .
a. s. dighe and a. y. smirnov , phys .
d * 62 * , 033007 ( 2000 ) [ arxiv : hep - ph/9907423 ]
. g. dutta , d. indumathi , m. v. murthy and g. rajasekaran , phys .
d * 64 * , 073011 ( 2001 ) [ arxiv : hep - ph/0101093 ] .
k. takahashi , m. watanabe , k. sato and t. totani , phys .
d * 64 * , 093004 ( 2001 ) [ arxiv : hep - ph/0105204 ] ; c. lunardini and a. y. smirnov , nucl . phys .
b * 616 * , 307 ( 2001 ) [ arxiv : hep - ph/0106149 ] ; k. takahashi and k. sato , nucl .
a * 718 * , 455 ( 2003 ) [ arxiv : hep - ph/0205070 ] ; c. lunardini and a. y. smirnov , arxiv : hep - ph/0302033
. v. barger , d. marfatia and b. p. wood , phys .
b * 547 * , 37 ( 2002 ) [ arxiv : hep - ph/0112125 ] .
m. t. keil , g. g. raffelt and h. t. janka , arxiv : astro - ph/0208035 .
h. t. janka , proceedings of _ frontier objects in astrophysics and particle physics _ , vulcano , italy , 1992 , edited by f. giovannelli and g. mannocchi .
t. totani , k. sato , h. e. dalhed and j. r. wilson , astrophys .
j. * 496 * , 216 ( 1998 ) [ arxiv : astro - ph/9710203 ] .
r. buras , h. t. janka , m. t. keil , g. g. raffelt and m. rampp , arxiv : astro - ph/0205006 .
a. mezzacappa , m. liebendorfer , o. e. messer , w. r. hix , f. k. thielemann and s. w. bruenn , phys .
* 86 * , 1935 ( 2001 ) [ arxiv : astro - ph/0005366 ] .
r. c. schirato and g. m. fuller , arxiv : astro - ph/0205390 .
r. tomas , m. kachelriess , g. raffelt , a. dighe , h. t. janka and l. scheck , arxiv : astro - ph/0407132 . g. l. fogli , e. lisi , a. mirizzi and d. montanino , arxiv : hep - ph/0412046 ; m. kachelriess and r. tomas , arxiv : hep - ph/0412100 .
a. bueno , i. gil - botella and a. rubbia , arxiv : hep - ph/0307222 . g. l. fogli , e. lisi , d. montanino and a. mirizzi , phys . rev .
d * 68 * , 033005 ( 2003 ) [ arxiv : hep - ph/0304056 ]
. k. nakamura , int .
j. mod .
a * 18 * , 4053 ( 2003 ) .
p. vogel and j. f. beacom , phys .
d * 60 * , 053003 ( 1999 ) [ arxiv : hep - ph/9903554 ] .
s. fukuda _ et al .
_ , [ super - kamiokande collaboration ] phys
. lett .
b * 539 * , 179 ( 2002 ) [ arxiv : hep - ex/0205075 ] ; m. b. smy _ et al .
_ , phys .
d * 69 * , 011104 ( 2004 ) [ arxiv : hep - ex/0309011 ] .
d. b. cline , f. sergiampietri , j. g. learned and k. mcdonald , nucl .
instrum .
a * 503 * , 136 ( 2003 ) [ arxiv : astro - ph/0105442 ] .
e. kolbe , k. langanke , g. martinez - pinedo and p. vogel , j. phys .
g * 29 * , 2569 ( 2003 ) [ arxiv : nucl - th/0311022 ] .
v. barger , p. huber and d. marfatia , work in progress . | we demonstrate that the detection of shock modulations of the neutrino spectra from a galactic core - collapse supernova is sufficient to obtain a high significance determination of the neutrino mass hierarchy if the supernova event is observed in both a mton - class water cherenkov detector and a 100 kton - class liquid argon detector .
neither detailed supernova neutrino flux modelling nor observation of earth matter effects is needed for this determination . as a corollary
, a nonzero value of @xmath0 will be established . |
dwarf galaxies ( with absolute bolometric magnitudes , @xmath9 ) are the most abundant in the universe , and their properties differ significantly depending on the local density : unlike the field , where most of dwarfs are star forming , the red , passive population dominates in the clusters , suggesting that the environment preferentially acts in quenching the star formation .
the general consensus is that it is a complex mixture of both
_ mass quenching _ quenching processes internal to the galaxy and _ environmental quenching _
externally driven quenching processes that drive the evolution .
multiple studies suggest that different processes dominate the quenching at distinct stellar masses , and that there is a characteristic stellar mass scale between the two types of quenching , with the former dominating in bright galaxy transformation and the latter in the low - mass galaxy evolution ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , the _ environmental quenching _ of dwarfs is far from ubiquitous ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the relevant physical mechanisms that can be considered _ environmental quenching _ are the loss of galaxy gas reservoirs @xcite , the gas cooling time @xcite and the suppression of dwarfs due to a combination of feedback , photoionization or / and dynamical processes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , the interplay of the different physical processes and their time - scales are not clear ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
deep spectroscopic surveys to study dwarf galaxies in high - density environments of different physical properties are needed to constrain the physical mechanisms influencing dwarf evolution and to better understand the environmental effects on their evolution ( e.g. , * ? ? ?
we have undertaken a campaign of deep spectroscopic observations of nearby clusters down to the dwarf regime ( @xmath10 ) in order to infer accurate cluster membership and study the properties of low - mass galaxies , such as star formation histories and orbital parameters .
the richness class 1 cluster abell 85 ( a85 hereafter ) is a nearby ( @xmath11 ) and massive cluster ( @xmath12mpc and @xmath13 * ? ? ?
it is one of the brightest clusters in the x - ray sky @xcite with a smoothed gas distribution .
it has been extensively studied in the whole electromagnetic spectrum : in the radio @xcite , in the optical @xcite and in the x - ray ( e.g. * ? ? ?
* ; * ? ? ?
the main cluster hosts an x - ray bright , metal - rich cool core and it is well known that a85 is not completely virialized .
its substructure was analysed in the literature at different wavelengths ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite found that a85 shows a complex x - ray temperature structure , and @xcite detected an extended x - ray peak in the south - west of the cluster .
moreover , other two abell clusters , a87 and a89 , are close in projection to a85 @xcite .
the former is also the nearest and is more likely to be a series of groups moving towards the main body of a85 @xcite .
both the x - ray emission and the optical galaxy distribution trace an elongation of the cluster orientation towards the a87 direction .
therefore , a85 is a perfect starting point for our project due to its physical characteristics together with a few practical reasons : its sky region is covered by the sloan digital sky survey ( sdss ) , so bright galaxies ( @xmath14 ) already have spectroscopic information , allowing us to focus on the faint ones , and its proximity favours deep spectroscopic and large area coverage at the same time .
the galaxy luminosity function ( hereafter lf ) gives the number density of galaxies of a given luminosity .
it is a powerful observable to study the characteristics of galaxies and to compare environments of different densities ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
in particular , the faint end seems to be shaped by the environment ( e.g. , * ? ? ? * ; * ? ? ? * ) , making deep spectroscopic study of dwarf galaxies and the comparison between those in field and clusters fundamental .
the schechter function @xcite is commonly used to fit the lf of galaxies .
this function is exponential at the bright end and a power law at the faint end .
however , its universality is still a matter of debate . while it is a good fit to some ( spectroscopic ) observations of clusters ( e.g. , * ? ? ?
* ) , a linear combination of two schechter functions is needed in other cases ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
previous studies suggested that the most remarkable difference between the global and cluster lfs is in the faint - end slope ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) .
photometric studies show that the lf in clusters has a steeper faint end .
however , this is not the case for some nearby galaxy clusters with deep spectroscopic lfs ( e.g. * ? ? ?
@xcite showed that the lf of a85 measured in three decades in luminosities has an upturn at @xmath15 and a steep faint end .
although the faint end is less steep compared to the photometric results of @xcite .
a85 lf shares common features with the spectroscopic field lf by @xcite , in particular a similar faint - end slope .
the main difference between the a85 and the field lf is the nature of the low - mass galaxies .
thus , red dwarfs dominate the cluster faint end of the lf while blue faint galaxies are more abundant in the field .
these results suggest that the environment plays a major role just in their evolution , transforming field blue dwarfs into cluster red ones , but not in the determination of the lf faint - end slope
i.e. in the relative abundances of low - mass galaxies . the aim of this paper is to study the upturn of the lf of a85 : its dependence on clustercentric distance , on galaxy type , and on the dominant physical processes that shape its profile .
the paper is organized as follows : the data set is presented in section 2 . in section 3 ,
we present the classification of the members in different populations .
section 4 describes and analyses the lf of the cluster , its radial dependence and the influence of the different types of galaxies on the upturn .
the results are discussed in section 5 and the conclusions are presented in section 6 . throughout this work ,
we have used the cosmological parameters @xmath16 , @xmath17 and @xmath18 .
we carried out an extensive spectroscopic study of the galaxies along the a85 line of sight .
our parent photometric catalogue contains all galaxies from the soan digital sky survey data release 6 ( sdss - dr6 ) @xcite , brighter than @xmath19 mag - band magnitudes .
note that these magnitudes have been corrected for extinction . ] and within 3.6 @xmath0 3.6 mpc@xmath1 . from this catalogue , we selected those galaxies with no measured spectroscopic redshifts and with @xmath20 colour bluer than @xmath21 ( see fig .
[ cmd ] ) identifying 3082 targets .
@xmath22 is the typical colour of a 12 gyr old stellar population with [ fe / h ] = + 0.25 supersolar metallicity @xcite .
the fraction of cluster members lost due to this colour selection is expected to be very small , and we match at the same time the colour distribution of the nearby universe ( see * ? ? ? * ; * ? ? ?
the brightest cluster galaxy ( bcg , @xmath23(j2000 ) : @xmath24 @xmath25(j2000 ) : @xmath26 ) is assumed to lie at the centre of the cluster .
this is a sensible choice given that the peak of the x - ray emission is offset only 7 kpc from the bcg centre @xcite .
figure [ cmd ] shows the colour magnitude diagram ( cmd ) of the full sample , with the colour cut applied . , blue diamonds ) and cluster member fraction ( @xmath27 , black triangles ) as a function of @xmath28-band magnitude .
the dashed vertical line represents our limit magnitude for the spectroscopic lf . ]
the majority of the observations were performed with the visible multi - object spectrograph at the very large telescope ( vimos@vlt ) , in combination with the lr - blue+os - blue grisms and filters ( program 083.a-0962(b ) , pi r. snchez - janssen , 2009 august ) . to maximize the number of targets , and avoid the gaps between the instrument ccds
, we designed 25 masks with large overlaps covering an area of 3.0 @xmath0 2.6 mpc@xmath1 around the central galaxy of a85 i.e. extending out to more than 1@xmath29 .
this strategy allowed us to obtain 2861 low - resolution spectra ( r=180 ) .
integrations of 1000 s were sufficient to obtain signal - to - noise ( @xmath30 ) ratios in the range 610 down to the limiting magnitude .
three more pointings have been added using autofib2/wyffos , a fibre spectrograph at the william herschel telescope ( af2@wht , service program sw2013b24 , 2014 september ) covering the central area of the cluster .
these observations gave us 176 low - resolution spectra ( r=280 and grism r158b ) reaching @xmath31 .
the data were collected during three exposures of 1800s per pointing in order to reach @xmath32 .
we used the provided pipelines of the two instruments ( version 2.25 for af2 data ) to perform the standard data reduction : bias , flat - field , sky subtraction and wavelength calibration @xcite . for the latter calibration we used an hene lamp , which yield a wavelength accuracy of @xmath33 pixel@xmath34 in the full spectral range ( @xmath35 ) , for the vimos data , and an he lamp for the wyffos spectra .
we used the _
iraf task @xcite to determine the redshifts of the observed galaxies .
this task cross - correlates a template spectrum library ( in this work , * ? ? ?
* ) with the observed galaxy spectrum and applies the technique described in @xcite .
we determine 2114 velocities between the two observing runs .
the remaining spectra had too low @xmath30 to estimate reliable redshifts .
formal errors of this task are smaller than true intrinsic ones ( e.g. * ? ? ?
this is because the error on the redshift estimate depends on several factors .
first of all , the resolution of the grating used in the observations , which , in this case , gives a sigma around 650700 kms@xmath34 at 5000 @xmath36 with slits of 1 arcsec .
note that this is the expected value for the resolution of vimos , which is worse than the resolution of af2 .
other sources of uncertainty are the wavelength calibration , errors in the cross - correlation and errors in the centring of the targets .
the convolution of these can be evaluated only by comparison of different measurements of the same objects .
thanks to our observational strategy we obtained 676 repeated spectra with vimos and nine with wht pointings , resulting in a @xmath37 velocity uncertainty of @xmath38 km
s@xmath34 and @xmath39 km s@xmath34 , respectively .
although these uncertainties are large , they do not significantly affect the determination of the cluster membership .
we found that in the range v@xmath40 , where @xmath41 is the cluster recessional velocity and @xmath42 its velocity dispersion , only 14 galaxies are discarded by the caustic method ( see section [ memb ] ) .
given that the number of non - member galaxies is relatively low for such a large velocity interval , we are confident that the bias in the cluster membership determination is negligible . combining our data with redshifts from the sdss - dr6 and nasa / ipac extragaalctic database ( ned ) catalogues results in a total number of 1603 redshifts in the magnitude interval @xmath43 , within 1.4@xmath29 and for galaxies bluer than @xmath44 .
the completeness of the data is shown in the upper panel of fig .
it is defined as @xmath45 , where @xmath46 is the number of measured redshifts and @xmath47 the number of photometric targets .
@xmath48 is larger than 90% for galaxies with @xmath49 and decreases around 40% at @xmath50 .
we have used the caustic technique to determine the cluster membership @xcite . when plotted on the plane of the line - of - sight velocity projected clustercentric distance ,
the cluster galaxies occupy a specific and well - defined region with a trumpet shape : a decreasing amplitude , @xmath51 , with increasing clustercentric distance , @xmath28 .
@xcite shoe that @xmath52 is a function of the cluster escape velocity modulated by a function depending on the galaxy orbit anisotropy parameter , @xmath53 .
one of the advantages of the caustic technique is that it measures the escape velocity and the mass profile of the cluster in both its virial and infall regions , where the assumption of dynamical equilibrium does not necessarily hold . as a byproducts ,
tho technique identifies the cluster members , namely those galaxies whose radial velocity is smaller than the escape velocity at a given clustercentric radius .
numerical simulations show that the cluster membership obtained using this technique has an interloper contamination of only 2% within @xmath29 @xcite . in order to estimate @xmath52 for a85
, we used our spectroscopic data set within the vimos field of view , completed with sdss - dr6 spectroscopic velocities up to 1.7@xmath29 .
we adopted this strategy to better constrain the trumpet shape of the caustics and consequently the profiles of the escape velocity and the mass of the cluster .
we obtained a sample of @xmath54 cluster members within 1.7 @xmath29 ; for further analysis , we use only the @xmath55 members that are within the observed area .
the members of a85 are shown in fig .
[ caust_memb ] where they are plotted on the line - of - sight velocity projected clustercentric distance plane . the member fraction defined as @xmath56 , where @xmath57 is the number of members ,
is plotted in the upper panel of fig .
it is a function of galaxy luminosity and is larger than 60 @xmath58 for @xmath59 and then rapidly decreases to @xmath60 20 @xmath58 at @xmath61 .
the completeness as function of both apparent magnitude and mean surface brightness is presented in fig .
[ binned_compl ] .
it shows that the target population is well represented by the galaxies with spectroscopic information up to @xmath62 mag and @xmath4 mag arcsec@xmath5 .
in this work , we aim to understand the type of galaxies producing the main features of the lf of a85 and their location in the cluster .
therefore , we separated the cluster members into different populations according to their colour , their dynamics and their star formation . in the following sections we give the details of our classifications .
we separated the members into red and blue populations by applying a colour cut .
we define as blue the galaxies bluer than @xmath63 , where @xmath64 and @xmath65 are the colour and the dispersion of the red sequence of the cluster .
the red population is formed by the remaining cluster members .
we estimate the red sequence by using the members with apparent @xmath66band magnitude brighter than 19 and @xmath20 colour redder than 0.6 . this strategy was adopted in order to have smaller uncertainties on the red sequence and a better determination of the two populations . indeed , sdss - dr6 errors on the magnitude increase with their values and are larger for low - mass galaxies .
the resulting fractions of red and blue galaxies turned out to be 74% and 26% , respectively . by using the equivalent width ( ew ) of the @xmath67 $ ] emission line
, we have separated the galaxies into passive and star forming .
we measured the ew(@xmath67 $ ] ) using vimos and sdss - dr6 spectra .
no ew(@xmath67 $ ] ) measurements could be obtained for those galaxies with only ned or af2@wht velocity information .
this reduced the number of cluster members by 10% for this classification .
galaxies with ew(@xmath67 $ ] ) @xmath68 are classified as star forming .
this upper limit was confirmed by visual inspection of the spectra .
only 16% of the analysed cluster members turned out to be active star - forming galaxies , as expected for nearby clusters ( see * ? ? ?
our catalogue contains the largest number of spectroscopically confirmed members of a85 presented so far in the literature , including 234 dwarf galaxies that were not taken into account in previous studies .
several works showed that the results on the substructure analysis in clusters strongly depends on the population of galaxies used ( e.g. , * ? ? ? * ) . to avoid biases
, we ran several statistical tests in order to detect substructure in a85 .
we used tests by taking into account only the recessional velocity of the galaxies ( 1d tests , eg .
kolmogorov - smirnof and 1d dedica * ? ? ?
* ) , their positions on the sky ( 2d tests , eg .
* ) , or both their positions and velocities ( 3d tests , eg . * ? ? ?
1d tests returned no significant substructures , while 2d and 3d tests gave evidence for substructure . in order
to confirm which galaxies belong to substructures , we applied the caustic method as described in @xcite .
this algorithm gave 18@xmath58 of a85 members in substructures .
for a more detailed analysis of a85 substructures , their properties and their comparison with x - ray observations see yu et al .
( in preparation ) according to @xcite , the members of a cluster can be classified depending on their infall time .
indeed , the phase space can be divided using the line : @xmath69 where @xmath70 and @xmath71 are the recessional velocities of the galaxies and of the cluster , respectively , @xmath72 is the cluster velocity dispersion , r is the projected clustercentric distance and r@xmath73 is the cluster virial radius .
their simulations show that at a given distance from the cluster centre , a significant fraction of the galaxies above this line fell into the cluster during the last 1gyr . applying this analysis to a85
, we found that 31@xmath58 of the members have been recently accreted .
figures [ ra_dec_pop ] and [ caustic_pop ] show the spatial and phase - space distribution of the different types of galaxies according to our classification . fig .
[ acum_distr_func ] shows the cumulative distribution function of r@xmath74 and @xmath75 for the same subsets .
red , passive and early - infall galaxies are generally located at smaller clustercentric distances than blue , star - forming and recent accreted members . moreover , the latter group of populations have higher velocity dispersion when compared to the global cluster velocity dispersion .
these trends give a radial and infall - time dependence of the star formation .
galaxies in substructure present a very narrow range in velocities , and are mostly located relatively close to the centre of the cluster .
the global properties of the galaxy populations in a85 are similar to those observed in other nearby galaxy clusters ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the lf is a basic statistics that provides information about the nature of the galaxy populations in different environments .
most of the results on the properties of cluster galaxies have been obtained with the extensive analysis of photometric lfs of nearby clusters .
however , few studies have been done based on deep spectroscopic lfs .
we computed the total spectroscopic lf of a85 following the prescription in @xcite .
the lf is given by : @xmath76 , where @xmath77 is the number of photometric targets ( see section [ tg ] ) , @xmath78 the fraction of cluster members defined in section [ memb ] , @xmath51 the surveyed area and @xmath79 the magnitude bin size used .
[ lf_rad ] shows the spectroscopic lf of a85 and the best - fitting double schechter function .
indeed , according to the pearson test , a single schechter function is not a good representation of the observed lf . in order to reduce the degeneracy of the fit
, we used a double schechter function with only one @xmath80 , similarly to previous studies ( e.g. , * ? ? ?
therefore , the free parameters of the fit are the characteristic luminosity of the whole cluster , two different slopes for the bright , @xmath81 , and the faint , @xmath82 components , and their relative normalization .
we use this fit function for the whole analysis carried out in this paper .
the parameters of the global lf are reported in table [ tab_rad_fit ] . due to the larger number of galaxies in the sample and the different double schechter fit used , they are slightly different from those of @xcite . however , the overall shapes of the fitted function are very similar , including the position of the upturn and the value of the faint - end slope .
we studied the radial dependence of the lf of a85 to analyse the influence of the environment on the galaxy population .
we divided the cluster into three regions , an inner circle of radius @xmath83 , an intermediate ring with @xmath84 and the outer area with @xmath85 .
this choice has no specific physical reason , but each bin contains roughly the same number of members giving similar statistical significance .
the resulting lfs are shown in fig .
[ lf_rad ] . according to the pearson test
, a single schechter function is not a good description of the data .
we have then fitted a double schechter function ( see table [ tab_rad_fit ] ) .
the relative contribution of the two components of the fit is radial dependent .
indeed , the second component becomes more prominent in the outer lf , and the relative contribution of the first one decreases with clustercentric distance .
the former component is the one strongly contributing to the upturn in the external regions of the cluster .
a dip is also present highlighting the change in the slope .
it is located at different magnitudes , moving from @xmath15 in the intermediate lf to @xmath86 in the outer radial bin .
we also studied the influence of different apertures on the lf .
in particular , we analysed the cumulative lfs within apertures of 0.4 , 0.7 and 1.4 @xmath29 .
the resulting lfs are shown in fig .
[ lf_aper ] . as in the previous analysis ,
a double schechter function is used to properly model the data and the fitted parameters are listed in table [ tab_rad_fit ] . here
, we do not observe any dip and the second component of the fit dominates only when the galaxies from the most external region of the cluster are considered . both the visual inspection of figs [ lf_rad ] and [ lf_aper ] and the analysis of the fitted parameters give a variation of the faint - end slope , @xmath82 , with radius .
in particular , the faint end is less steep in the outer region of the cluster .
there is also an evolution of @xmath87 from the cluster centre to the outskirts : it becomes fainter at larger radii . in the outskirts of the cluster
the lf is similar to the total and the field spectroscopic lf ( see * ? ? ?
in order to understand the meaning of these variations , we studied the normalized lfs .
the normalization factor used is the integral of each lf i.e. within @xmath88 $ ] .
[ lf_norm ] shows the normalized fitted lfs . for sake of clarity
we plot only the fitted functions to highlight the differences . with a visual inspection of the normalized lfs ( see fig .
[ lf_norm ] ) , we divided the galaxy population of a85 into three groups : bright ( @xmath89 ) , intermediate ( @xmath90 ) and dwarf ( @xmath91 ) galaxies .
the former and the latter subsets do not present significant dependence on the environment .
furthermore , they are invariant with respect to the field ( see lf by * ? ? ?
these results are corroborated by the kolmogolov
smirnov test with a probability always larger than 10% for the two galaxy groups .
therefore , the null hypothesis can not be rejected , and we can not exclude that they come from a single population . however , this is not true for the intermediate galaxies that show a clear environmental dependence .
the morphology density relation proposed by @xcite suggests a radial dependence of the lf . in this section
, we first analyse the global lf quantifying the relative contribution of the galaxy subsets .
we then look to the radial dependence for the same subsets . in the first column of fig .
[ lf_pop_rad ] , we show the global lfs .
the most abundant galaxies in the whole magnitude range are red , passive , virialized and early infall in the cluster history .
these populations contribute to the 75 -80% of the total luminosity density of the cluster ( see table [ tab_pop_fit ] ) .
their lfs present an upturn at the faint end similar to the one showed by the global lf and are modelled by a double schechter function .
blue , star - forming , located in substructures and recent infall members contribute to only [ 15 - 25 % ] of the total luminosity density .
their lfs are well represented by a single schechter function , and the values of the fit parameters are given in table [ tab_pop_fit ] .
the radial lfs are shown in the other columns of fig .
[ lf_pop_rad ] and the fit parameters are given in table [ tab_pop_rad_fit1 ] .
note that the fits are double schechter functions and were performed only on the dominant populations due to the poor statistics of the other ones .
furthermore , we applied the same aperture analysis obtaining results similar to the global lfs .
focusing on fig .
[ lf_pop_rad ] , the general trends of the red , passive , virialized and early infall lfs are the same of the global cluster . in particular , they present the two components with similar relative contributions , the upturn in the outskirts and the dip moving towards brighter magnitude from the intermediate to the outer region .
moreover , the external lfs present a mixture of the two subsets of galaxies , especially at the faint end . consequently
, the upturn of the outer lf of a85 is due to both red and blue , passive and star - forming , and early and late infall galaxies . in fig .
[ lf_pop_norm ] , we show the radial normalized lfs of the dominant populations : red , passive , virialized and early infall . * the normalization factor is the integral of each lf .
[ lf_pop_norm ] * shows features like the global lf of a85 .
indeed , we can identify the three groups of galaxies bright , intermediate and dwarfs and their similar trends with the environment ( compare with fig .
[ lf_norm ] ) .
ccccc @xmath92 ( mag ) & @xmath93 & @xmath82 & @xmath94 & @xmath95 + + @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 + + @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 + + @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 + + @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 + + @xmath116 & @xmath117 & @xmath118 & @xmath109 & @xmath110 + cccc @xmath92 [ mag ] & @xmath119 & @xmath120 & @xmath121 + + @xmath122 & @xmath123 & @xmath124 & 87% + + @xmath125 & @xmath126 & & 13% + + @xmath127 & @xmath128 & @xmath129 & 89% + + @xmath130 & @xmath131 & & 11% + + @xmath132 & @xmath133 & @xmath134 & 86% + + @xmath135 & @xmath136 & & 14% + + @xmath137 & @xmath138 & @xmath139 & 75% + + @xmath140 & @xmath141 & & 25% +
@xcite studied the photometric lf of a85 and compared it to the lf of other nearby clusters . a dip in the lfs
was found at similar magnitudes in these clusters .
they concluded then that this similarity is due to the same composition of morphological types of galaxies in all clusters .
our lf of a85 shows that the dip has a radial dependence : no dip is observed in the inner lf , while in the intermediate and in the outer radial bins is located at @xmath142 and @xmath14319.0 , respectively .
the dip has similar radial trends in the lfs of red , passive , virialized and early - infall galaxies .
it is clearly visible in the cluster outskirts , while it is not present in the inner radial bin ( compare second , third and fourth columns of fig .
[ lf_pop_rad ] ) .
the relative contribution of the galaxy subsets changes with clustercentric distance .
blue , star - forming and recent arrival galaxy densities are smaller in the central region and similar in the external one .
this feature indicates that the presence of the dip can be the result of the evolution of galaxies in the cluster environment .
indeed , it is not present where a small fraction of galaxies is still star forming or blue . in other words , cluster members in the outer region
have not yet fully experienced the effects of cluster environment .
@xcite studied the evolution of the mass function with redshift and the effects of the environment on it .
they analysed different sets of galaxies and , in particular , they traced the blue and star - forming galaxies and the red and passive ones .
they proposed a model where two types of quenching occur and depend on the mass of the galaxy itself .
one of them affects massive galaxies and is called _ mass quenching _ , while the other is effective at low - mass ranges and it is called _
environmental quenching_. in the case of passive and red objects , they needed a double schechter function to model the data and concluded that the two mathematical components of the fit shape the two processes . applying this interpretation to a85
, we have two components that depend on the clustercentric distance and we called bright and faint components .
we have the faint one with increasing weight towards the outskirts .
its relative contribution to the total luminosity in the radial bin ( whole cluster ) changes from the 2% in the inner radial bin to the 22% in the outer one .
similar trend is observed with respect to the total luminosity of the cluster , even if weakened : the faint population passes from 1 to 7% .
consequently , the bright component shows a decreasing teen inside - out .
the radial variation of the @xmath80 parameter highlight these different contributions .
in particular , the trends of the red , blue , passive and star - forming subsets agree with the results of @xcite .
moreover , the lfs of recent - arrival and not - virialized populations are fitted by a single schechter , while the lfs of the respective dominant populations e.g. virialized and early - infall
are better represented by a double schechter .
this suggests that the latter population has spent enough time within the cluster potential to have produced their two components .
normalizing the lf is useful in order to compare its shape among systems of different richness .
a variety of normalization factors has been used in literature , and we analyse here their effects on our study .
we first normalized the lf by scaling it with total luminosity integrated over the full range of magnitudes .
we note that it is common practice to restrict the integration over few magnitudes at the bright end of the lf .
this choice is supported by the environmental invariance of the bright end ( * ? ? ?
* ; * ? ? ?
* among others ) .
we normalized the radial lfs of a85 using factors from the literature .
in particular , we used the ranges @xmath144 , @xmath145 , and @xmath146 ( e.g. , * ? ? ? * ; * ? ? ? * ) .
[ lf_cfr_norm ] shows , we obtained similar faint - end slopes , but different relative number densities of dwarfs .
note that if the magnitude range used for the normalization includes intermediate luminosity galaxies , they are equalized by construction , so no studies on their radial properties can be made .
therefore , we strongly suggest to use the total luminosity as normalization factor and perform a complete and unbiased analysis of the environmental dependence of the lf .
the results on the bright end of the normalized lf of a85 confirm previous outcomes from the literature : it is independent of environment ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this independence is likely a result of dynamical friction .
massive / luminous galaxies suffer of large dynamical friction entering the cluster potential , falling into the central region of the cluster on short time - scales .
there , they merge with the bcg , giving a small fraction of luminous members ( @xmath147 ) in the inner radial bin .
note that dynamical friction can also explain other properties of this galaxy population observed and analysed in the literature ( e.g. * ? ? ?
dwarf galaxies are the most affected by environmental processes .
their shallow potential well impedes them from retaining their gas once in the cluster .
for the same reason , tidal stripping efficiently remove stellar mass .
the main physical processes acting on the gas component are starvation @xcite and ram pressure @xcite , while the change in mass and/or morphology are mostly ascribed to harassment @xcite and strong tidal interactions @xcite .
what the dominant processes in clusters are is still an open question , and the analysis of the faint end of the lf helps to put some constrains .
several literature outcomes show a dependence on the environment of the faint - end slope on photometric lfs . indeed ,
early - type lfs are steeper in the outer regions of a cluster ( see * ? ? ?
* ; * ? ? ?
this result was recently confirmed with the spectroscopic stellar mass function of a medium redshift cluster ( see * ? ? ?
photometric cluster lfs also show steeper slopes than the field ( see e.g. * ? ? ?
* ) indicating an excess of dwarf galaxies in high - density environments .
this result is considered a probe supporting the evolutionary link between dwarf galaxies and clusters .
there are only a handful of deep , spectroscopic observations of nearby clusters , in which cluster - to - cluster variations of the shape of the lf are observed : virgo and abell 2199 lfs do not show any upturn at the faint end @xcite , unlike a85 @xcite .
moreover , the radial analysis of @xcite does not present a significant dependence of the lf on clustercentric radius , but opposite trends in the radial variation of @xmath80 between virgo and abell 2199 . indeed
, the characteristic luminosity seems to be fainter at small radii for abell 2199 , while virgo shows a trend similar to a85 ( see section [ lf_comp ] ) .
these observed cluster - to - cluster variations may depend on the formation history , the evolution of the large - scale environment or on the cluster mass . indeed , ( * ? ?
* see in particular their fig.1 ) hint a relation between the shape of the lf and the mass : more massive clusters show steeper slopes and more complex shapes .
being a85 the most massive and its lf the steepest among the three clusters , this seems to indicate that we spectroscopically observed this relation .
however , a wide spectroscopic survey of nearby clusters spanning a large range in masses will be necessary in order to put constrains on this and on the physical processes that give rise to these variations .
@xcite also studied the red and blue populations and compared their lfs to the field spectroscopic ones by @xcite .
the two subsets show common features , including compatible slopes and the upturn for the red lfs . however , the dominant population at the faint end is different in the two environments : blue dwarfs are dominant in the field while red low - mass galaxies are dominant in the cluster .
they concluded that the processes quenching star formation are playing a key role in this cluster .
the present analysis on the lf of a85 finds similar results : red and passive lfs show the upturn , and the presence of the dip requires the modelling by a double schechter function , while blue and star - forming galaxies are well represented by a single schechter function .
however , they have compatible faint - end slopes , suggesting again that the dominant mechanisms do not act on the size and/or the mass of the dwarfs .
indeed , a recent orbit analysis on simulated virgo - like cluster found that only a small fraction ( 20 - 25% ) of low - mass galaxies suffers from substantial stellar mass loss due to harassment @xcite .
moreover , our radial study found that the number density of late - type dwarfs is barely changing with radius , while the density of early - type galaxies is decreasing inside - out .
these results indicate a transformation of the nature of dwarf galaxies due to the quenching of their star formation .
in addition , the dominant mechanisms are more effective in the centre of the cluster where the hot intracluster medium is denser .
the evolution of dwarfs is then led by starvation and ram - pressure stripping .
photometric studies of the lf in radial bins indicate a variation of the faint end for the red population @xcite .
however , our analysis does not confirm this result .
the normalized lfs shown in fig .
[ lf_pop_norm ] highlight that the variation of the lf is at intermediate luminosities , whereas it is statistically insignificant at the faint end .
the result that finds spectroscopical confirmation is the variation of @xmath80 .
indeed , the characteristic luminosity of the red lf moves towards brighter values in the external cluster region ( compare the top left panel of fig .
[ lf_pop_norm ] with fig .
11 of @xcite ) .
however , due to cluster - to - cluster variations , a wider spectroscopic sample of nearby clusters is needed to quantitatively discuss the comparison between photometric and spectroscopic results .
the results presented in this work show that the lf of a85 changes with the clustercentric distance at the intermediate magnitude range @xmath148 . in addition , cluster and field lfs differ within the same range , giving a clear environmental dependence for the evolution of intermediate luminosity galaxies
. this dependence could be the consequence of the orbital distribution of these galaxies due to dynamical friction .
moreover , the large velocity dispersion of a85 inhibits mergers .
therefore , this type of galaxies will survive for a long time in eccentric orbits
i.e. with large apocentres ( see * ? ? ?
the lf of a85 is dominated by red , passive and early - infall galaxies also at the magnitude range analysed here .
these galaxies have passed near the cluster centre at least once ( see e.g. * ? ? ?
* ) , indicating that the intermediate galaxy population has suffered quenching of star formation due to the long time spent within the cluster potential .
we have identified 460 cluster members of the nearby ( @xmath149=0.055 ) cluster a85 down to @xmath150 , thanks to deep spectroscopic observations .
this data set allowed us to study the total and radial spectroscopic lfs of this cluster .
the global lf and the lf dominant galaxy population
i.e. red , passive , virialized and early - infall galaxies were fitted by a double schechter function and they present a dip . these results point towards the presence of two galaxy populations in the cluster : a bright and a faint one .
their fraction depends on the clustercentric distance . indeed , the faint component is more important in the outskirts of the cluster . in the central region of a85 ,
the lfs do not show a dip and the faint galaxy populations only represent @xmath151 of the total galaxy density .
also blue and star - forming galaxies are a small fraction in this region .
this result can indicate that the quenching of the star formation fills the dip observed in the external lfs .
the normalized lfs of a85 show that the bright and faint ends are independent of the environment .
moreover , they are similar to the field ones . in contrast
, the intermediate luminosity galaxies ( @xmath8 ) present an important dependences on the environment . considering these results together , we suggest that hydrodynamical mechanisms are the main drivers of the dwarf evolution in a85 .
these processes transform blue low - mass galaxies into red ones with no significant changes in their mass / luminosity or size .
the distribution of the bright and intermediate luminosity galaxies is due to dynamical friction .
therefore , the former suffer the largest dynamical friction , fall to the central region and eventually merge with the bcg on a short time - scale .
moreover , the bright end is independent of the environment .
in contrast , the intermediate luminosity galaxies present a clear environmental dependence .
this difference suggests that they survive in the cluster for several gyr on eccentric orbits around the centre thanks to small dynamical friction .
we acknowledge the anonymous referee for the useful comments that helped us to improve the paper .
this work has been partially funded by the mineco ( grant aya2013 - 43188-p ) .
cdv acknowledges support from the ministry of economy and competitiveness ( mineco ) through grants aya2013 - 46886 and aya2014 - 58308 .
cdv and rb acknowledge financial support from the spanish ministry of economy and competitiveness ( mineco ) under the 2011 severo ochoa program mineco sev-2011 - 0187 .
ia and ad acknowledge partial support from the infn grant indark and the grant prin 2012 `` fisica astroparticellare teorica '' of the italian ministry of university and research .
this research has made use of the sixth data release of sdss , and of the nasa / ipac extragalactic database which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
the wht and its service programme are operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
99 abell g. o. , corwin h. g. , jr . , olowin r. p. , 1989 ,
apjs , 70 , 1 adelman - mccarthy , j. k. , ageros , m. a. , allam , s. s. , et al . , 2008 , apjs , 175 , 297 aguerri j. a. l. , snchez - janssen r. , muoz - tun c. , 2007 , a&a , 471 , 17 aguerri j. a. l. , gonzlez - garca a. c. , 2009 , a&a , 494 , 891 aguerri j. a. l. , snchez - janssen r. , 2010 , a&a , 521 , a28 agulli i. , aguerri j. a. l. , snchez - janssen r. , barrena r. , diaferio a. , serra a. l. , mndez - abreu j. , 2014 , mnras , 444 , l34 andreon s. , 1998 , a&a , 336 , 98 annunziatella m. , et al . , 2016 , a&a , 585 , a160 baldry i. k. , balogh m. l. , bower r. g. , glazebrook k. , nichol r. c. , bamford s. p. , budavari t. , 2006 , mnras , 373 , 469 balogh m. l. , baldry i. k. , nichol r. , miller c. , bower r. , glazebrook k. , 2004 , apj , 615 , l101 bardelli s. , zucca e. , vettolani g. , zamorani g. , scaramella r. , collins c. a. , macgillivray h. t. , 1994 , mnras , 267 , 665 barkhouse w. a. , yee h. k. c. , lpez - cruz o. , 2007 , apj , 671 , 1471 barkhouse w. a. , yee h. k. c. , lpez - cruz o. , 2009 , apj , 703 , 2024 bekki k. , couch w. j. , shioya y. , 2002 , apj , 577 , 651 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 biviano a. , katgert p. , 2004 ,
a&a , 424 , 779 blanton m. r. , et al . , 2003 , apj , 592 , 819 blanton m. r. , lupton r. h. , schlegel d. j. , strauss m. a. , brinkmann j. , fukugita m. , loveday j. , 2005 , apj , 631 , 208 bou g. , durret f. , adami c. , mamon g. a. , ilbert o. , cayatte v. , 2008 , a&a , 489 , 11 bravo - alfaro h. , caretta c. a. , lobo c. , durret f. , scott t. , 2009 , a&a , 495 , 379 brooks a. m. , kuhlen m. , zolotov a. , hooper d. , 2013 , apj , 765 , 22 davies l. j. m. , et al . , 2016 , mnras , 455 , 4013 diaferio a. , geller m. j. , 1997 , apj , 481 , 633 diaferio a. , 1999 , mnras , 309 , 610 domnquez palmero l. , jackson r. , molaeinezhad a. , faria c. , balcells m. , benn c. r. , 2014 , spie , 9149 , 91492j dressler a. , shectman s. a. , 1988 , aj , 95 , 985 dressler a. , 1980 , apj , 236 , 351 durret f. , forman w. , gerbal d. , jones c. , vikhlinin a. , 1998 , a&a , 335 , 41 durret f. , felenbok p. , lobo c. , slezak e. , 1999 , a&as , 139 , 525 durret f. , lima neto g. b. , forman w. , churazov e. , 2003 , a&a , 403 , l29 edge a. c. , stewart g. c. , fabian a. c. , arnaud k. a. , 1990 , mnras , 245 , 559 fillingham s. p. , cooper m. c. , wheeler c. , garrison - kimmel s. , boylan - kolchin m. , bullock j. s. , 2015 , mnras , 454 , 2039 + goto t. , 2005 , mnras , 359 , 1415 gunn j. e. , gott j. r. , iii , 1972 , apj , 176 , 1 hogg d. w. , et al .
, 2004 , apj , 601 , l29 ichinohe y. , werner n. , simionescu a. , allen s. w. , canning r. e. a. , ehlert s. , mernier f. , takahashi t. , 2015 , mnras , 448 , 2971 infante l. , mieske s. , hilker m. , 2003 , ap&ss , 285 , 87 izzo c. , kornweibel n. , mckay d. , palsa r. , peron m. , taylor m. , 2004 , msngr , 117 , 33 kennicutt r. c. , jr .
, 1992 , apj , 388 , 310 kurtz m. j. , mink d. j. , wyatt w. f. , fabricant d. g. , torres g. , kriss g. a. , tonry j. l. , 1992 , aspc , 25 , 432 larson r. b. , tinsley b. m. , caldwell c. n. , 1980 , apj , 237 , 692 moore b. , katz n. , lake g. , dressler a. , oemler a. , 1996 , natur , 379 , 613 muriel h. , coenda v. , 2014 , a&a , 564 , a85 oman k. a. , hudson m. j. , behroozi p. s. , 2013 , mnras , 431 , 2307 peng y .- j . ,
et al . , 2010 , apj , 721 , 193 peng y. , maiolino r. , cochrane r. , 2015 , natur , 521 , 192 phillips j. i. , wheeler c. , boylan - kolchin m. , bullock j. s. , cooper m. c. , tollerud e. j. , 2014 , mnras , 437 , 1930 phillips j. i. , wheeler c. , cooper m. c. , boylan - kolchin m. , bullock j. s. , tollerud e. , 2015 , mnras , 447 , 698 pisani a. , 1993 , mnras , 265 , 706 popesso p. , bhringer h. , brinkmann j. , voges w. , york d. g. , 2004 , a&a , 423 , 449 popesso p. , biviano a. , bhringer h. , romaniello m. , 2006 , a&a , 445 , 29 pracy m. b. , couch w. j. , blake c. , bekki k. , harrison c. , colless m. , kuntschner h. , de propris r. , 2005 , mnras , 359 , 1421 quilis v. , moore b. , bower r. , 2000 , sci , 288 , 1617 ramella m. , et al . , 2007 , a&a , 470 , 39 rines k. , diaferio a. , 2006 , aj , 132 , 1275 rines k. , geller m. j. , 2008 , aj , 135 , 1837 schechter p. , 1976 ,
apj , 203 , 297 snchez - janssen r. , aguerri j. a. l. , muoz - tun c. , 2008 , apj , 679 , l77 schenck d. e. , datta a. , burns j. o. , skillman s. , 2014 , aj , 148 , 23 serra a. l. , diaferio a. , murante g. , borgani s. , 2011 , mnras , 412 , 800 serra a. l. , diaferio a. , 2013 , apj , 768 , 116 slee o. b. , roy a. l. , murgia m. , andernach h. , ehle m. , 2001 , aj , 122 , 1172 smith r. , et al .
, 2015 , mnras , 454 , 2502 tonry j. , davis m. , 1979 , aj , 84 , 1511 trentham n. , hodgkin s. , 2002 , mnras , 333 , 423 trentham n. , sampson l. , banerji m. , 2005 , mnras , 357 , 783 tully r. b. , somerville r. s. , trentham n. , verheijen m. a. w. , 2002 , apj , 569 , 573 weisz d. r. , dolphin a. e. , skillman e. d. , holtzman j. , gilbert k. m. , dalcanton j. j. , williams b. f. , 2015 , apj , 804 , 136 wetzel a. r. , tinker j. l. , conroy c. , van den bosch f. c. , 2013 , mnras , 432 , 336 wheeler c. , phillips j. i. , cooper m. c. , boylan - kolchin m. , bullock j. s. , 2014 , mnras , 442 , 1396 white s. d. m. , rees m. j. , 1978 , mnras , 183 , 341 wijesinghe d. b. , et al . , 2012 , mnras , 423 , 3679 worthey g. , 1994 , apjs , 95 , 107 yagi m. , kashikawa n. , sekiguchi m. , doi m. , yasuda n. , shimasaku k. , okamura s. , 2002 , aj , 123 , 87 yu h. , serra a. l. , diaferio a. , baldi m. , 2015 , apj , 810 , 37
ccc @xmath92 ( mag ) & @xmath93 & @xmath82 + + + @xmath152 & @xmath128 & @xmath153 + + @xmath154 & @xmath155 & @xmath156 + + @xmath157 & @xmath158 & @xmath159 + + @xmath160 & @xmath161 & @xmath162 + + + @xmath163 & @xmath164 & @xmath165 + + @xmath166 & @xmath167 & @xmath168 + + @xmath169 & @xmath158 & @xmath170 + + @xmath171 & @xmath172 & @xmath173 + + + @xmath174 & @xmath175 & @xmath176 + + @xmath177 & @xmath178 & @xmath179 + + @xmath180 & @xmath181 & @xmath182 + + @xmath183 & @xmath184 & @xmath185 + + + @xmath186 & @xmath187 & @xmath188 + + @xmath189 & @xmath190 & @xmath191 + + @xmath192 & @xmath193 & @xmath194 + + @xmath195 & @xmath196 & @xmath197 + | we present a new deep spectroscopic catalogue for abell 85 , within 3.0 @xmath0 2.6 mpc@xmath1 and down to @xmath2 . using the visible multi - object spectrograph at the very large telescope and the autofiber 2 at the william herschel telescope
, we obtained almost 1430 new redshifts for galaxies with @xmath3 mag and @xmath4 mag arcsec@xmath5 .
these redshifts , together with sdss - dr6 and ned spectroscopic information , result in 460 confirmed cluster members .
this data set allows the study of the luminosity function ( lf ) of the cluster galaxies covering three orders of magnitudes in luminosities .
the total and radial lfs are best modelled by a double schechter function .
the normalized lfs show that their bright ( @xmath6 ) and faint ( @xmath7 ) ends are independent of clustercentric distance and similar to the field lfs unlike the intermediate luminosity range ( @xmath8 ) .
similar results are found for the lfs of the dominant types of galaxies : red , passive , virialized and early - infall members . on the contrary , the lfs of blue , star - forming , non - virialized and recent - infall galaxies are well described by a single schechter function .
these populations contribute to a small fraction of the galaxy density in the innermost cluster region .
however , in the outskirts of the cluster , they have similar densities to red , passive , virialized and early - infall members at the lf faint end .
these results confirm a clear dependence of the colour and star formation of abell 85 members in the cluster centric distance .
[ firstpage ] galaxies : cluster : individual a85 , galaxies : luminosity function , mass function , galaxies : dwarf |
1 . magnetization of bcc fe as a function of temperature obtained for the simulation cell containing 128 atoms with dynamic spin - spiral boundary conditions .
black stars and white circles correspond to orbital and the rigid spin approximation spin dynamics , respectively .
black squares show experimental data .
the arrow indicates the finite size scaling result for the curie temperature of fe .
2 . total energy as a function of angle between the magnetic moment of a single isolated impurity and the global bulk magnetization .
black circles correspond to a ni impurity in fcc ni ; triangles , to a fe impurity in bcc fe ; white circles , to a ni atom with the relaxation of its nearest neighbors included . a function @xmath39 is also shown by squares . the energy is normalized by the stiffness of the corresponding problem .
arrows indicate the ` critical ' angles where the local moment on the given impurity disappears . | we report the first results of time - dependent density functional simulations of magnetic properties of fe and ni at finite temperatures .
they reveal the existence of local moments in ni above the curie temperature , coupled to strong short range order . in fe the short range order is also present to a lesser extent but spin wave like excitations persist in both materials well above the curie temperature in the paramagnetic state . the prevailing view of magnetic order should be reconsidered in light of these findings .
a proper description of the microscopic behavior of magnetic systems at finite temperature represents one of a few fundamental problems still existing in the theory of magnetism . while numerous model theories with different adjustable parameters lead to some description of magnets at finite temperatures , the attendant approximations significantly influence the final results .
_ ab initio _ methods have been used since the early 1980 s ; however they are limited to zero temperature and largely reproduce the heisenberg model .
as might be expected , such methods provide a good description of magnets with localized moments , but the results for itinerant magnets are rather unsatisfactory . in particular , the temperature dependence of longitudinal fluctuations and magnetic short - range order ( msro ) have always been neglected .
msro in itinerant magnets is the subject of a long - standing controversy .
early reports on inelastic neutron scattering experiments indicated the persistence of spin wave - like modes with relatively small wave vectors above the curie temperature @xmath0 in both ni and fe@xcite .
based on these results , the local band theory was developed@xcite , but its assumptions are incompatible with the observed curie - weiss susceptibility and specific heat anomaly@xcite . since then , various experimentssro confirm the presence of some degree of msro in many ferromagnets , and the present belief is that it can be very strong .
however , up to now theory has failed to estimate the degree of msro in different magnets , owing to the formidable challenges in implementing a tractable theory for real materials .
several quantitative theoretical methods based on the use of the calculated band structure have been developed . in many cases the _ ab initio _ version of the single - site coherent potential approximation ( cpa )
was employed .
the first calculation@xcite evaluates the msro in fe by calculating the non - local exchange coupling .
however , the single - site cpa presupposes that msro is absent ; hence the calculated msro is small .
further cpa calculations@xcite also obtained a good description of finite - temperature properties of bcc fe , but do not find any local magnetic moment ( lmm ) at high temperatures in fcc ni . using onsager s reaction - field approach@xcite ,
the description of ni can be improved , which indicates that some interatomic correlations beyond the mean - field approximation ( mfa ) should be included , at least for ni .
different tight - binding methods not based on the cpa have been used to estimate the msro and the effective coupling in fe and ni @xcite .
most did not find strong msro and failed to explain the corresponding experimental data . in ref .
@xcite it was claimed that the effective coupling is much stronger at the curie temperature than at @xmath1 , which makes the temperature - dependence of the exchange coupling a very important problem .
however , as we will discuss below , the tight - binding approximation leads to a strong underestimation of msro , and hence to different physics .
the first band - structure calculations of the effective exchange coupling@xcite may be regarded as being in the limit which is opposite to the cpa , since the local band theory assumes strong msro . however , these calculations@xcite yielded a much smaller degree of msro in contradiction with the initial assumption .
this contradiction has yet to be resolved .
thus , it seems that no numerical calculations so far have explicitly shown the presence of strong msro in fe or ni based on their band structures , while the experiments clearly indicate its presence .
the second key point concerns the proper inclusion of longitudinal fluctuations in itinerant magnets at finite - temperature , which are missing in any rpa theory . if it were not for serious challenges limiting its tractability , _ ab initio _ spin dynamics would be a very powerful tool to address all these issues . in ref.@xcite a combination of first principles spin dynamics with different stochastic or deterministic types of finite - temperature description was proposed . in the present paper we present an implementation that makes this technique practicable , and apply it for the first time to studies of finite - temperature behavior of ferromagnetic fe and ni .
this approach is based on time - dependent spin density functional theory tddft .
the main equation in this theory is @xmath2where @xmath3 is a non magnetic part of the total potential , @xmath4 is the magnetic part of the exchange correlation potential and @xmath5 is the external magnetic field . instead of considering the evolution of the non - local density matrix
, one can introduce the closed system of coupled equations for the three local one electron quantities : the charge density @xmath6 , the magnetization density @xmath7 , and the velocity density @xmath8 . separating the magnetization dynamics adiabatically
, one can obtain the main one - electron equation of spin dynamics @xmath9 \label{kss}\]]where @xmath10 , @xmath11 , and @xmath12 is the velocity field for the electron liquid .
the basic assumption of the local spin density approximation ( lsda ) is that the exchange - correlation potential of the homogeneous electron gas with fixed charge density @xmath13 and magnetization density @xmath14 can be applied to real external potentials . in this approximation @xmath15 @xmath16 , and hence assumes very strong exchange correlation field @xmath17 compared to the ` kinetic ' field @xmath18 and the smallness of dynamics associated with the exchange correlation term .
this important approximation assumes that correlation effects do not contribute to the spin dynamics of magnets , but attributes all of the dynamics to the kinetic term .
many static calculations of spin excitations have generally confirmed that this approximation is valid for many materials . in ref .
saga we estimated the strength of correlation effects by calculating the contribution of non - local coulomb correlations to the spin wave stiffness in the gw approximation .
this contribution happened to be very small compared to the kinetic contribution .
below we will confirm that this approximation is quite reasonable for dynamic simulations as well .
in this paper we solve eq.([kss ] ) by using the classical langevin approach for finite temperatures , and employ the so - called orbital dynamics approach@xcite so that each orbital can precess with its own frequency , independent of the other orbitals . in this treatment we avoid the standard adiabaticity which is assumed in the dynamic rigid spin approximation , and account for all decoherence effects properly within lsda .
this approach is similar to the linear response technique when the external field is small .
however , our approach does not necessarily assume that the _ perturbation _ is small .
all calculations were performed within lsda using the tight - binding linear muffin - tin orbital method .
in typical calculations with 100 - 120 atoms per cell we used 20 - 30 @xmath19-points in the irreducible part of the brillouin zone .
the correct description of possible long - range magnetic order ( especially at low temperatures ) requires that the simulation cell contains many atoms .
to avoid this numerical difficulty , we added temperature - dependent spin - spiral boundary conditions @xcite that is dynamical , with its corresponding equation of motion .
similar to the static case , such combined real / reciprocal space spin dynamics allowed us to significantly reduce the number of atoms per cell in our simulations and made numerous statistical calculations possible . from the physical point of view
, the simultaneous inclusion of real - space short - range modes ( inside the supercell ) and long - wavelength modes ( between supercells ) enables the description of magnetic excitations developing at different length scales , as well as their interactions , on an equal footing . as we now show , the ability to include multiple - length scales was essential to this work .
first , let us discuss static properties . in fig.1
we show the total magnetization of fe as a function of temperature .
the results for 128-atom simulations ( using dynamical spin spiral boundary conditions ) for fe are shown together with the finite - size scaling result .
the finite - size scaling results were obtained using supercells of four different sizes for both ni and fe .
the effectiveness of the dynamical spiral boundary conditions can be seen in the 128-atom simulation .
the low magnetization tail above @xmath0 was reproduced in standard real space calculations ( monte carlo simulations for the heisenberg model ) only when 7000 - 8000 atoms were used@xcite .
@xmath20 resulting from the simulation is quite satisfactory . for fe ,
nearly perfect agreement with experiment is obtained : 1070k ( 1023k ) ; in ni @xmath0 is about 25% smaller then experiment : 470 k ( 623k ) .
however , this discrepancy may be attributed to the lsda itself : the total energy difference between fully ferromagnetic and nonmagnetic ni is about 550k ( with all parameters identical to the dynamical simulation ) . while this upper limit to the @xmath0 in ni is somewhat higher when gradient corrections are added to lsda , we do not at present have a sufficiently accurate density functional to disentangle the lsda contribution to the error .
we only suggest that this error is connected with some on - site coulomb correlations which are missing in lsda at @xmath1 k. the proper acccount of quantum corrections could significantly affect @xmath0 as well .
the effective paramagnetic moment @xmath21 the was extracted from the calculations of high - temperature susceptibility .
these moments compared to experimental values@xcite are 3.3 ( 3.13 ) and 1.75 ( 1.6 ) bohr magnetons for fe and ni respectively . to interpret these large moments
we will now discuss the dynamical properties .
a very unusual picture of magnetic order near @xmath0 emerges . the crucial physical quantity which reveals a degree of short- or long - range order in magnets is the spin - spin correlation function@xmath22where @xmath23 indicates thermodynamic average .
the short - range part of @xmath24 is strongly affected by the presence of itineracy in the magnetic system .
the analysis of @xmath24 shows that the average angle between nearest neighbors lmm is 67@xmath25 in bcc fe near @xmath0 , however in fcc ni it is just 28@xmath25 .
these numbers are very stable as a function of temperature : within a temperature range of 0.95@xmath0 to 1.05@xmath0 their overal changes do not exceed 3 - 4% .
the deviation from 90@xmath25 indicates that in fe the msro is relatively small ( but noticeable ) , while in ni it is enormous .
a real - space analysis of the correlation function shows that in ni the dynamic ordering is of the spin - spiral or domain - wall type , with a period of @xmath2614 a.u . on the other hand , in fe
nearest neighbor correlations dominate and a relatively weak superparamagnetic type msro is realized at @xmath0 .
overall , the dynamic correlation function has very well defined peaks well above @xmath0 and clearly demonstrates the persistence of the spin - wave - like perturbations in the paramagnetic state of fe and ni .
the relatively high energy of short - wave ferromagnons in these materials is primarily responsible for this feature .
the presence of such msro at @xmath0 indicates that the nearest neighbor interactions in these materials must be larger then @xmath0 .
however , the exchange couplings in these materials are well known ( see ref.wang,prb1 ) and corresponding numbers are somewhat lower then @xmath0 making the existence of the strong msro in these materials impossible . on the other hand , starting from pioneering calculations in ref.@xcite all calculations of @xmath27 assumed a long wave approximation ( lwa ) which has smallness parameter @xmath28 , where @xmath29 is the stoner parameter@xcite . in general , this approximation is inapplicable to systems with strong interatomic interactions ( as in the case of itinerant magnets ) .
we found that the usage of the more general definition @xmath30 leads to a much better interpretation of our results .
first of all , the lwa is suitable for a relatively ` localized ' system such as fe , and the corresponding change in the nearest neighbor exchange is relatively small ( @xmath2615% ) but still somewhat higher than @xmath0 .
ni represents a rather itinerant system , and any local approach ( especially lwa ) is expected to produce a large error . in our case , @xmath31 for ni is increased by more than 3 times relative to the lwa , clearly that the effective coupling is much larger than @xmath0 .
these results explain why the msro was missed in previous calculations of @xmath27 and indicate that the traditional mf approach ( or any other approach which is based on the assumption of no short - range order ) is inapplicable to itinerant systems in general , predicting very high @xmath0 ( above 1000 k in ni ) .
this interpretation has a qualitative character due to known violation of adiabaticity criteria in the itinerant magnets .
the msro allows us to provide a different interpretation of the susceptibility results as described above .
the increase of the effective magnetic moment @xmath21 in the curie - weiss law compared to the localized model is directly related to clustering ( for example , of the superparamagnetic type ) when the effective moment of the cluster is several times larger then original atomic moment .
the ratio of the effective moment to the atomic moment is larger in ni than in fe .
our calculations also indicate that in ni a significant contribution ( 15% ) comes from longitudinal fluctuations
. the high temperature paramagnetic state of ni has both tangential and longitudinal modes as soft modes which are much closer to each other than at @xmath1 k. we performed several model calculations for ni assuming static disorder with 90@xmath25 average angle between nearest lmm ( no msro ) .
for all eight random configurations studied the value of @xmath21 was much lower ( about 50 - 60% of our theoretical value ) , clearly supporting msro contribution to the @xmath21 .
in addition , all final self consistent states in ni for these random configurations appeared to be non magnetic .
so , we believe that strong msro is a necessary requirement for fcc ni to be ferromagnetic and a similar conclusion should apply to all itinerant magnets .
such strong msro can not appear only at finite temperatures ; it must exist already at @xmath1 k ( as indicated by our calculations of the effective coupling ) .
to study msro in more detail we performed the following model calculations at t=0 k. first , we studied the degree of localization by rotating the lmm at one site by an arbitrary angle ( similar to ref.@xcite but with lwa removed ) . in case of the fully localized heisenberg type of interaction
one can expect that such perturbation is perfectly stable and the total energy is described by @xmath32 type of behavior .
our calculation revealed completely different picture ( fig.2 ) . in case of single site perturbation ,
the atomic magnetic moment in fe disappears at approximately 120@xmath25 .
longitudinal fluctuations at this point become so important that the moment is inverted .
as we mentioned above , such effects can not be described in any tight - binding type of approach . in ni
the situation is even more complicated and at @xmath2650@xmath25 the local moment is completely destroyed .
these calculations immediately suggest that both fe and ni do not have a well defined lmm at finite temperature .
we also found , that the lmm with opposite spin configuration in bcc fe can be stabilized if the perturbation is extended to the first shell of neighboring atoms , providing a clear range of minimal msro to support such strong local perturbation . in the itinerant ni the complete flip of lmm
can not be stabilized even if the perturbation is extended to the first neigboring atoms ( fig.2 ) .
the perturbation is stabilized only when it is extended to three shells of neighbors .
these calculations strongly support the idea of giant msro in fcc ni and relatively small ( but essential ) msro in bcc fe .
another indication of itineracy ( and strong msro ) only was obtained from studies of non - magnetic configurations .
we estimated the range of existence of lmm in non magnetic fcc ni . in case
when lmm exist only at one atom such fluctuation is unstable and lmm simply does not exist .
however , if the lmm is fixed at its equilibrium value ( 0.6 @xmath33 ) on nearest atoms , the atomic lmm is stable . according to our lsda calculations fcc ni at its experimental density has stable lmm of @xmath260.6 @xmath33 only if it has ferromagnetic interaction with two nearest shells of atoms with the same moments .
this result clearly indicates that magnetism in ni has itinerant character .
its nature in this material is not so much in intra - atomic interactions but rather in interatomic ferromagnetic interactions with nearest neighbors .
this also indicates that the standard adiabatic criterion is not applicable in ni at finite temperatures .
such strong msro also allow us to resolve a very old issue of the importance of quantum corrections for @xmath0 and high - temperature susceptibility calculations .
these corrections are traditionally small in the localized systems ( @xmath34 magnets ) , but always considered to be significant in the itinerant systems .
msro described above also strongly suppresses quantum corrections in the case of the itinerant magnets by increasing the effective moment . in summary
, we performed _
ab - initio _ spin dynamic simulations at finite temperatures using time - dependent spin density functional theory .
our calculations revealed the existence of magnetic short range order at @xmath0 in @xmath35 ferromagnetic magnets fe and ni .
this short - range order is relatively small in bcc fe ( the average angle between local magnetic moments @xmath36 ) , while in fcc ni it is rather large ( @xmath37 ) .
the existence of distorted spin waves above @xmath0 significantly reduces the range of applicability of the traditional models of magnetism which are based on the idea of the spin wave disappearing at @xmath0 , and demonstrates that more reliable models of magnetism are needed .
this msro strongly affects all observable properties and establishes that the thermodynamics of fe and ni can be described in the framework of the density functional approach .
we believe that msro is a natural property of any magnet over a very wide temperature range and is the _ sine qua non _ in low moment ( @xmath38 ) itinerant magnets where msro dramatically influence thermodynamical properties .
this msro is unique in each system and has a certain memory about the ground state of that system . from the methodical point of view
the successful usage of our approach clearly indicates that the kinetic energy produces the largest contribution to the dynamics of magnets at finite temperature while the exchange - correlation contribution is small .
the author would like to thank k. belashchenko , m. van schilfgaarde and n. zein for helpful discussions .
this work was carried out at the ames laboratory , which is operated for the u.s.department of energy by iowa state university under contract no .
w-7405 - 82 .
this work was supported by the director for energy research , office of basic energy sciences of the u.s.department of energy .
99 h. a. mook , j. w. lynn and r. m. nicklow .
phys .
rev .
lett . * 30 * , 556 ( 1973 ) ; g. shirane , y.j .
uemura , j. p. wicksted , y. endoh , y. ishikawa .
phys .
rev .
b * 31 * , 1227 ( 1985 ) ; h. a. mook , j. w. lynn . j. appl
. phys . * 57 * , 3006 ( 1985 ) . v. korenman , j. l. murray , r. e. prange , phys .
rev .
b * 16 * , 4032 , 4048 , 4058 ( 1977 ) .
d. m. edwards and m. a. rahman , j. phys .
f * 8 * , 1501 ( 1978 ) . c. j. maetz , u. gerhardt , e. dietz , a. ziegler and r. j. jelitto .
phys .
rev .
lett .
* 48 * ( 1982 ) 1686 ; e. haines , r. glauberg , r. feder .
phys .
rev .
lett . * 54 * , 932 ( 1985 ) .
t. oguchi , k. terakura and n. hamada .
j. phys .
f * 13 , * 145 ( 1983 ) . m. cyrot . j. magn
. magn . mater . *
45 * , 9 ( 1984 ) ; j. staunton and b. l. gyorffy .
phys .
rev .
lett . * 69 * , 371 ( 1992 ) .
m. v. you and v. heine . j. phys .
f : met .
phys .
* 12 * , 177 ( 1982 ) ; j. h. samson .
j. phys .
cond . mat . * 1 * , 6717 ( 1989 ) . v. heine , a. i. liechtenstein and o. mryasov .
euro .
lett . ,
* 12 * , 245 ( 1990 ) . c. s. wang , r. e. prange , v. korenman .
phys .
rev .
b * 25 , * 5766 ( 1982 ) . v. p. antropov , m. i. katsnelson , m. van schilfgaarde and b. n. harmon , phys .
rev .
lett . * 75 * , 729 ( 1995 ) ; v. p. antropov .
j. appl
. phys . * 79 * , 5409 ( 1996 ) ; v. p. antropov , m. i. katsnelson , m. van schilfgaarde , b. n. harmon , and d. kusnezov , phys .
rev .
b * 54 * , 1019 ( 1996 ) ; v. p. antropov . in computer simulation studies in condensed - matter physics xvi . ed . by d. p. landau .
v.*95*. ( berlin , heidenberg , springer - verlag , 2004 ) p.41 .
k. l. liu and s. h. vosko .
canad .
j. phys .
* 67 * , 1015 ( 1989 ) .
m. i. katsnelson and v. p. antropov .
phys . rev . *
67 * , 140406 ( 2003 ) .
l. m. sandratskii , p. g. guletskii , j. phys.f * 16 * , l43 ( 1986 ) ; v. p. antropov and m. i. katsnelson ( unpublished ) . m. e. fisher .
rep .
. phys . * 39 * , 615 ( 1967 ) .
k. wohlfart . , in _
ferromagnetic materials_. pergamon press .
north holland .
v.1 ( 1982 ) .
v. p. antropov , b. n. harmon and a. v. smirnov .
j. magn .
magn .
mater . * 200 * , 148 ( 1999 ) . v.
p. antropov .
j. magn .
magn . mater . *
262 * , l192 ( 2003 ) .
s. v. turzgevskii , a. i. liechtenstein and m. i. katsnelson .
sov . phys
. solid state .
* 32 * ( 1990 ) 1138 . |
extensive theory has been established in recent years for causal inference based on directed acyclic graph ( dag ) models .
a popular way for estimating a dag model from observational data employs partial correlation testing to infer the conditional independence relations in the model . in this paper
, we apply algebraic geometry and singularity theory to analyze partial correlations in the gaussian case .
the objects of our study are algebraic hypersurfaces in the parameter space of a given graph that encode conditional independence statements .
we begin with definitions for graphical models in statistics .
a dag is a pair @xmath3 consisting of a set @xmath4 of nodes and a set @xmath5 of directed edges with no directed cycle .
we usually take @xmath6 and we associate random variables @xmath7 with the nodes .
directed edges are denoted by @xmath8 or @xmath9 .
the _ skeleton _ of a dag @xmath10 is the underlying undirected graph obtained by removing the arrowheads .
a node @xmath11 is an _ ancestor _ of @xmath12 if there is a directed path @xmath13 , and a configuration @xmath14 , @xmath15 is a _ collider _ at @xmath16 .
finally , we assume that the vertices are topologically ordered , that is @xmath17 implies @xmath18 .
every dag @xmath10 specifies a _
gaussian graphical model _ as follows .
the adjacency matrix @xmath19 is the strictly upper triangular matrix whose entry in row @xmath11 and column @xmath12 is a parameter @xmath20 if @xmath17 and it is zero if @xmath21 .
the gaussian graphical model is defined by the structural equation model @xmath22 , where @xmath23 .
we assume that @xmath24 , where @xmath25 is the @xmath26-identity matrix .
then the _ concentration matrix _ of this model equals @xmath27 since @xmath28 , the covariance matrix @xmath29 is equal to the adjoint of @xmath30 .
the entries of the symmetric matrices @xmath30 and @xmath31 are polynomials in the parameters @xmath20
. our parameter space for this dag model will always be a full - dimensional subset @xmath32 of @xmath33 . for any subset @xmath34 and distinct elements @xmath35 in @xmath36
, we represent the conditional independence statement @xmath37 by an _ almost - principal minor _ of either @xmath30 or @xmath31 . by this
we mean a square submatrix whose sets of row and column indices differ in exactly one element . to be precise
, @xmath38 holds for the multivariate normal distribution with concentration matrix @xmath30 if and only if the submatrix @xmath39 is singular , where @xmath40 and @xmath41 .
the determinant @xmath42 is a polynomial in @xmath43 .
we are interested in the hypersurface in @xmath33 defined by the vanishing of this polynomial . indeed ,
the _ partial correlation _
@xmath44 is up to sign equal to the algebraic expression @xmath45 since the principal minors under the square root sign are strictly positive , @xmath46 if and only if @xmath47 .
if this holds for all @xmath48 then @xmath38 for @xmath10 and we say that @xmath11 is _ d - separated _ from @xmath12 _ given _ @xmath49 .
this translates into a combinatorial condition on the graph @xmath10 as follows @xcite .
an undirected path @xmath50 from @xmath11 to @xmath12 _ d - connects _ @xmath11 and @xmath12 _ given
_ @xmath49 if 1 .
every non - collider on @xmath50 is not in @xmath49 , 2 .
every collider on @xmath50 is in @xmath49 or an ancestor of a node in @xmath49 . if @xmath10 has no path that d - connects @xmath11 and @xmath12 given @xmath49 , then @xmath11 and @xmath12 are _ d - separated given _
@xmath49 , and @xmath51 as a function of @xmath52 .
the _ weight _ of a path @xmath50 is the product of all edge weights @xmath53 along this path .
it was shown in ( * ? ? ? * equation ( 11 ) ) that the numerator @xmath54 in ( [ eq : parcor ] ) is a linear combination , as in ( [ eq : fsosf ] ) , of the weights of all paths that d - connect @xmath11 to @xmath12 given @xmath49 .
our primary objects of study are the following subsets of the parameter space : @xmath55 here @xmath44 is a function of the parameter @xmath56 ( denoted @xmath57 above ) in the space @xmath58 , @xmath59 is a parameter in @xmath60 $ ] , and @xmath61 is a triple where @xmath11 and @xmath12 are d - connected given @xmath49 .
these `` tubes '' can be seen as hypersurfaces which have been fattened up by a factor which depends on @xmath59 and the position on the hypersurface ( see figure [ fig : tubes ] ) .
the volume of @xmath62 with respect to a given measure , or _ prior _
, @xmath63 on @xmath64 is represented by the integral @xmath65 in this paper we study the asymptotics of this integral when the parameter @xmath59 is close to @xmath66 .
two applications in statistics are our motivation .
the first concerns the strong - faithfulness assumption for algorithms that learn markov equivalence classes of dag models by inferring conditional independence relations .
the pc - algorithm @xcite is a prominent instance .
our set - up is exactly as in @xcite .
the gaussian distribution with concentration matrix @xmath30 is @xmath59-_strong - faithful _ to a dag @xmath10 if , for any @xmath34 and @xmath67 , we have @xmath68 if and only if @xmath11 is d - separated from @xmath12 given @xmath49 .
we write @xmath69 for the volume of the region in @xmath32 representing distributions that are not @xmath59-strong - faithful . in other words
, @xmath69 is the volume of the union of all tubes in @xmath32 that correspond to non d - separated triples @xmath70 .
zhang and spirtes @xcite proved uniform consistency of the pc - algorithm under the strong - faithfulness assumption with @xmath71 , provided the number of nodes @xmath72 is fixed and sample size @xmath73 . in a high - dimensional , sparse setting , kalisch and bhlmann @xcite require strong - faithfulness with @xmath74 , where @xmath75 denotes the maximal degree ( i.e. , sum of indegree and outdegree ) of nodes in @xmath10 . in order to understand the properties of the pc algorithm for a large sample size @xmath76 , it is essential to determine the asymptotic behavior of the unfaithfulness volume @xmath69 when @xmath59 tends to @xmath66
. given a prior @xmath77 over the parameter space , @xmath69 is the prior probability that the true parameter values violate @xmath59-strong faithfulness .
thus @xmath78 for @xmath71 describes the prior probability that the pc algorithm is able to recover the true graph .
we shall see in example [ ex : specialcases ] that @xmath69 depends on the choice of the parameter space @xmath32 and the prior @xmath77 .
we shall address the issue of computing @xmath69 as @xmath79 .
this will be done using the concept of _ real log canonical thresholds _ @xcite .
our section [ sec : real_log_threshold ] establishes the existence of positive constants @xmath80 ( which depend on @xmath10 and @xmath77 ) such that , asymptotically for @xmath81 , @xmath82 ( see ( [ eq : curly_equal ] ) for an exact definition of @xmath83 . )
this refines the results in @xcite on the growth of @xmath69 via the geometry of the _ correlation hypersurfaces _
while @xcite focused on developing bounds on @xmath69 for the low - dimensional as well as the high - dimensional case and showed the importance of the number and degrees of these hypersurfaces , we here analyze the exact asymptotic behavior of @xmath69 for @xmath79 and @xmath10 fixed and demonstrate the importance of the singularities of these hypersurfaces .
singularities get fattened up much more than smooth parts of the hypersurface , and this increases the volumes ( [ eq : volume2 ] ) substantially .
our second application concerns _ stratification bias in causal inference _ ( see e.g. @xcite ) . here
, the volume @xmath85 being large is not a problematic feature , but is in fact desired .
suppose we want to study the effect of an exposure @xmath5 on a disease outcome @xmath86 .
if there is an additional variable @xmath87 such that @xmath88 , then stratifying ( i.e. conditioning ) on @xmath87 tends to change the association between @xmath5 and @xmath86 .
this can lead to biases in effect estimation .
this is known as _ collider - bias_. on the other hand , if @xmath89 holds , then @xmath87 is a _ confounder _ and stratifying on @xmath87 corresponds to bias removal . in certain larger graphs , such as greenland s _ bow - tie
example @xcite , stratifying on @xmath87 removes confounder - bias but at the same time introduces collider - bias . in order to decide whether one should stratify on such a variable @xmath87 ,
it is important to understand the partial correlations involved . in this application
, the volume @xmath85 can be viewed as the cumulative distribution function of the prior distribution of the partial correlation @xmath44 implied by the prior distribution on the parameter space , and we compare the two cumulative distribution functions @xmath90 and @xmath91 . in this paper we examine @xmath85 from a geometric perspective , and we demonstrate how this volume can be calculated using tools from singular learning theory . to derive the asymptotics ( [ eq : volume2 ] ) ,
the main player is the correlation hypersurface , which is the locus in @xmath32 where @xmath92 vanishes .
the first question is whether this hypersurface is smooth , and , if not , one needs to analyze the nature of its singularities .
we study these questions for various classes of interesting causal models , using methods from computational algebraic geometry .
the remainder of this paper is organized as follows : in section [ sec : graphs ] we introduce the families of dags which we will be working with throughout .
example [ ex : tripartite6 ] illustrates the algebraic computations that are involved in our analysis .
we also discuss some simulation results , which indicate the importance of singularities when studying the volume @xmath69 of strong - unfaithful distributions .
presents the connection to singular learning theory @xcite and explains how this theory can be used to compute the volumes of the tubes @xmath93 .
example [ ex_tubes ] illustrates our theoretical results for some very simple polynomials in two variables . in
we develop algebraic algorithms for analyzing the singularities of the correlation hypersurfaces .
we show that , for the polynomials @xmath94 of interest , the real singular locus is often much simpler than the complex singular locus .
for instance , theorem [ thm : uptosix ] states that these hypersurfaces are always smooth for complete dags with up to six nodes . in we
study the singularities and the volumes ( [ eq : volume1 ] ) for trees without colliders .
section [ sec : bias ] focuses on our second application , namely bias reduction in causal inference .
problems [ prob : tripart ] and [ prob : bowtie ] offer precise versions of conjectures by greenland @xcite , in terms of comparing different volumes @xmath85 for fixed @xmath10 .
we establish some instances of these conjectures . in section [ sec : resolution ]
we introduce more advanced methods , based on the resolution of singularities @xcite , for finding the exponents @xmath95 and @xmath96 in ( [ eq : volume2 ] ) .
finally , in section [ sec : integrals ] we present some new results on computing the constants @xmath87 and @xmath97 in our asymptotics ( [ eq : volume2 ] ) for tube volumes .
in this article we will be primarily working with four classes of dags : 1 . _ complete graphs _ : we denote the complete dag on @xmath72 nodes by @xmath98 .
the corresponding matrix @xmath99 is strictly upper triangular and all @xmath100 parameters @xmath20 are present .
2 . _ trees _ : we call a dag @xmath10 a tree graph if the skeleton of @xmath10 is a rooted tree and all edges point away from the root ( i.e. @xmath10 has no colliders ) .
we are particularly interested in the most extreme trees , namely star and chain - like graphs .
we denote the star graph shown in figure [ fig : star ] by @xmath101 and the chain - like graph shown in figure [ fig : chain ] by @xmath102 .
complete tripartite graphs _ : let @xmath103 with @xmath104 .
then we denote by @xmath105 the complete bipartite graph where @xmath106 for all @xmath107 and @xmath108 . a complete tripartite graph is denoted by @xmath109 with @xmath110 .
it corresponds to the dag @xmath111 and is shown in figure [ fig : tripart ] .
bow - ties _ : we define a bow - tie to be a complete tripartite graph @xmath112 with two additional edges , namely @xmath113 and @xmath114 .
a bow - tie is denoted by @xmath115 and is shown in figure [ fig : bowtie ] .
bow - ties with @xmath116 feature prominently in greenland s the following example serves as a preview to the topics covered in this paper .
[ ex : tripartite6 ] we illustrate our objects of study for the tripartite graph @xmath117 . since the error variances are assumed to be fixed at @xmath118 , this dag model has eight free parameters , namely the unknowns in the matrix @xmath119 the covariance matrix @xmath31 equals the inverse ( or the adjoint ) of the concentration matrix @xmath120 one conditional independence statement of interest is @xmath121 .
its correlation hypersurface in @xmath122 is defined by the almost - principal minor in @xmath31 with rows @xmath123 and columns @xmath124 , or the almost - principal minor in @xmath30 with rows @xmath125 and columns @xmath126 .
that determinant equals @xmath127 this is a weighted sum of all paths which d - connect nodes @xmath118 and @xmath128 given @xmath129 .
the first term in the formula ( [ eq : fsosf ] ) for @xmath130 corresponds to the path @xmath131 in @xmath132 , and the last term corresponds to the path @xmath133 .
let @xmath77 denote a prior on the parameter space .
for this example we take @xmath77 to be the lebesgue probability measure on the cube @xmath134 ^ 8 $ ] .
the expression @xmath135 defined in ( [ eq : volume1 ] ) is the volume of the region of parameters @xmath136 that satisfy @xmath137 as a function in @xmath59 , the volume @xmath135 is a cumulative distribution function on @xmath138 .
our aim in this article is to determine the asymptotics of such a function for @xmath81 . in section [ sec : real_log_threshold ]
we shall explain the form of the asymptotics that is promised in ( [ eq : volume2 ] ) . in order to find the exponents @xmath95 and @xmath96 , the first step is to run the algebraic algorithm in section [ sec : sing_locus ] .
this answers the question whether the hypersurface in @xmath32 defined by @xmath139 has any singular points .
the set of such points , known as the _ singular locus _ , is the zero set in @xmath32 of the ideal @xmath140 the tools of section [ sec : sing_locus ] reveal that its _ real radical _ @xcite is the intersection of three prime ideals : @xmath141{j } & \!\ ! = \!\ !
& \bigl\langle\ , \hbox{entries of } \ , \begin{pmatrix } a_{13 } & a_{14 } \\ a_{23 } & a_{24 } \end{pmatrix } \cdot \begin{pmatrix } a_{35 } & a_{36 } \\
a_{45 } & a_{46 } \end{pmatrix}\ , \bigr\rangle \\ & \!\ ! = \!\ !
& \langle a_{13 } , a_{14 } , a_{23 } , a_{24 } \rangle \,\cap \ , \langle a_{35 } , a_{36 } , a_{45 } , a_{46 } \rangle \,\cap \ , \langle \hbox{$2{\times}2$-minors of } \ , \begin{pmatrix } a_{13 } & a_{23 } & a_{45 } & a_{46 } \\
a_{14 } & a_{24 } & \ ! -a_{35 } & \ ! -a_{36 } \\ \end{pmatrix } \rangle.\end{aligned}\ ] ] thus the hypersurface @xmath142 is singular .
its singular locus decomposes into three irreducible varieties , namely two linear spaces of dimension @xmath143 and one determinantal variety of dimension @xmath144 . in section [ sec : bias ]
we return to this example , with focus on a statistical application of the cumulative distribution function @xmath135 .
we will then show that @xmath145 equals @xmath146 .
this paper extends the work of uhler , raskutti , bhlmann and yu in @xcite on the geometry of the strong - faithfulness assumption in the pc - algorithm .
upper and lower bounds on the volume @xmath69 of the unfaithful region @xmath147 for the low- as well as the high - dimensional setting were derived in @xcite .
these bounds involved only the number @xmath148 of parameters and the degrees of the correlation hypersurfaces @xmath149 .
the new insight in the current paper is that singularities are essential for the asymptotic behavior of @xmath69 for @xmath81 .
what led us to this insight was taking a closer look at the simulation results for trees .
in @xcite trees were still treated as one single class .
we subsequently examined the difference between stars and chains , depicted in figure [ fig : star ] and [ fig : chain ] .
our simulation results for @xmath150 and @xmath151 are shown in figure [ fig : graphs_lambda ] .
we shall now explain the curves in these diagrams .
the left diagram in figure [ fig : graphs_lambda ] is for @xmath152 nodes and the right diagram is for @xmath153 .
each curve is the graph of the cumulative distribution function @xmath69 but with the x - axis transformed to a logarithmic scale ( with base 10 ) .
thus we depict the graph of the function @xmath154\ , \rightarrow \,[0,1],\,\ , x\ , \mapsto \,v_g(10^x ) .\ ] ] the red curve is for @xmath155 and the blue curve is for @xmath156 .
these curves were computed by simulation : we sampled the parameter @xmath52 from the uniform distribution on @xmath157^{p-1}$ ] and we recorded the proportion of trials that landed in @xmath147 for various values of @xmath59 .
the diagrams show clearly that @xmath69 is smaller for star graphs than for chain - like graphs .
a theoretical explanation for these experimental results will be given in section [ sec : trees ] .
our asymptotic theory predicts the behavior of these curves as @xmath158 tends to @xmath159 .
the point is that the correlation hypersurfaces for chain - like graphs have deeper singularities than those for star graphs .
the equation of any such hypersurface for a tree is the product of a monomial and a strictly positive polynomial .
this enables us to apply proposition [ prop : rlctmonomial ] . in theorem
[ thm : treev ] and corollary [ cor__const_trees ] we shall determine the constants @xmath95 , @xmath96 and @xmath87 of ( [ eq : volume2 ] ) exactly when the graph @xmath10 is a tree . we shall also address the question of how to obtain @xmath95 , @xmath96 and @xmath87 from simulations .
before we get to graphical models , however , we first need to develop the mathematics needed to analyze @xmath69
. this will be done , in a self - contained manner , in the next section .
we now introduce the basics regarding the computation of integrals like the one in ( [ eq : volume1 ] ) , and we explain why asymptotic formulas like ( [ eq : volume2 ] ) can be expected . while this section is foundational for what is to follow , no reference to any statistical application is made until theorem [ lemma_unfaithful ] .
it can be read from first principle and might be of independent interest to a wider audience .
let @xmath160 be a compact , full - dimensional , semianalytic subset and consider a probability measure @xmath161 on @xmath32 where @xmath162 is the standard lebesgue measure and @xmath163 is a real - analytic function . also , fix an analytic function @xmath164 whose hypersurface @xmath165 has non - empty intersection with the interior of @xmath32 .
we are interested in the volume @xmath166 with respect to the measure @xmath77 of the region @xmath167 here @xmath168 is a parameter that is assumed to be small . in later sections , we often take @xmath32 to be the cube @xmath169^d$ ] , with @xmath77 its lebesgue probability measure , and @xmath170 is usually a polynomial .
the asymptotics of the volume function @xmath166 depends on the singularities of the hypersurface @xmath142 .
this phenomenon is illustrated in figure [ fig : tubes ] .
our measure for the complexity of the singularities of @xmath170 is a pair @xmath145 of non - negative real numbers .
that pair is the _ real log canonical threshold _ of @xmath170 .
it is related to the volume @xmath166 for small values of @xmath59 by the formula @xmath171 here @xmath87 is a positive real constant whose study we shall defer until section [ sec : integrals ] .
[ ex_tubes ] let @xmath172 and @xmath77 the lebesgue probability measure on the square @xmath173 ^ 2 $ ] .
our problem is to compute the area of the tube @xmath174 . here
@xmath175 is one of the four simple polynomials below whose tubes are shown in figure [ fig : tubes ] .
1 . @xmath176 : the corresponding tube is a rectangle and its area equals @xmath177 so , in this example , we have @xmath178 and @xmath179 . for other lines , the value of @xmath87 will change .
proposition [ thm : smoothfunctionrlct ] below shows that @xmath178 for smooth hypersurfaces .
@xmath180 : the tube in figure [ fig : tubes_2 ] consists of four copies of a region that is the union of a small rectangle and a certain area under a hyperbola . using calculus , we find @xmath181 the logarithm function appears in this case .
we have @xmath182 and @xmath179 .
@xmath183 : the corresponding tube is shown in figure [ fig : tubes_3 ] . its area equals @xmath184 so , the real log canonical threshold equals @xmath185 , and we have @xmath186 . see proposition [ prop : rlctmonomial ] for a formula for @xmath145 when @xmath170 is a monomial in any number of variables .
@xmath187 : the corresponding tube is shown in figure [ fig : tubes_4 ] .
this example is a slight generalization of ( b ) . as in ( b ) there is just one singularity at the origin , given by the intersection of lines .
computing the area @xmath166 is more challenging .
in example [ ex : fourlines ] we shall see that the real log canonical threshold equals @xmath188 . for general bivariate polynomials @xmath175 we are facing a hard calculus problem , namely integrating the function @xmath189 that is defined implicitly by @xmath190 .
we can approach this by expanding @xmath191 as a puiseux series in @xmath59 whose coefficients depend on @xmath192 . integrating these coefficients
leads to asymptotic formulas in @xmath59 .
these are consistent with what is to follow .
we now return to the general setting defined at the beginning of this section .
let @xmath193 be a random variable taking values in @xmath32 with distribution @xmath77 .
the volume @xmath166 with respect to the measure @xmath77 can then be viewed as the cumulative distribution function of the random variable @xmath194 .
the corresponding probability distribution function @xmath195 is called the _ state density function_. its mellin transform is known as the _ zeta function _ of @xmath170 .
it is denoted by @xmath196 according to asymptotic theory @xcite , our volume has the asymptotic series expansion @xmath197 here the index @xmath95 runs over some arithmetic progression of positive rational numbers and @xmath198 is the dimension of the parameter space @xmath32 .
the equation ( [ eq : asymp ] ) is valid for sufficiently small @xmath168 . to be precise , writing @xmath199 , where @xmath200 for small @xmath59 , means that @xmath201 using the little - o notation , this is equivalent to @xmath202 as @xmath81 for each positive integer @xmath16 .
it is a common misconception to think that the infinite series converges to @xmath166 for each fixed @xmath59 when @xmath59 is small .
rather , it means that for each fixed @xmath16 , the @xmath16-term approximation for @xmath166 gets better as @xmath81 .
we will primarily be interested in the first term approximation ( [ eq : firstorder ] ) .
[ thm : defrlct ] we here define the _ real log canonical threshold _ @xmath203 of @xmath170 over @xmath32 with respect to @xmath77 .
this is a pair in @xmath204 which we denote by @xmath205 .
it measures the complexity of the singularities of the hypersurface defined by @xmath206 .
the following four definitions of @xmath207 are known to be equivalent : 1 . for large @xmath208 , the _ laplace integral _
@xmath209 is asymptotically @xmath210 for some constant @xmath87 .
the _ zeta function _
@xmath211 has its smallest pole at @xmath212 and that pole has multiplicity @xmath96 .
3 . for small @xmath168 , the _ volume function _ @xmath213 is asymptotically @xmath214 for some constant @xmath87 .
4 . for small @xmath168 , the _ state density function _ @xmath215 is asymptotically @xmath216 for some constant @xmath87 . if the real analytic hypersurface @xmath217 is empty , we set @xmath218 and we leave @xmath96 undefined .
we say that @xmath219 if @xmath220 or if @xmath221 and @xmath222 .
hence , the pairs are ordered reversely by the size of @xmath223 for sufficiently small @xmath168 .
let us provide some intuition for the ordering of the pairs @xmath145 .
the real log canonical threshold is a measure of complexity for singularities .
analytic varieties can be stratified into subsets where this measure is constant .
the highest stratum contains the smooth points of the variety . as we go deeper , to strata with lower real log canonical thresholds , we encounter singularities of increasing complexity .
the volumes of @xmath59-fattenings of deeper singularities will , asymptotically as @xmath59 goes to zero , also be larger than those of their less complex counterparts .
for instance , in both figures [ fig : tubes_2 ] and [ fig : tubes_3 ] the singular locus consists of the origin , but the @xmath59-fattening of the origin in figure [ fig : tubes_3 ] is larger than in figure [ fig : tubes_2 ] .
see also example [ ex : mono_s_t ] .
[ ex : d - ball ] let @xmath224 and @xmath77 the lebesgue probability measure on @xmath134^d$ ] .
then @xmath225 is the _ standard ball _ of radius @xmath226 , whose @xmath77-volume is @xmath227 by definition [ thm : defrlct ] ( iii ) , the real log canonical threshold equals @xmath228 .
we now list some formulas for computing the real log canonical threshold .
a first useful fact is that @xmath205 is independent of the underlying measure @xmath77 as long as it is positive everywhere .
we can thus assume that @xmath77 is the uniform distribution on @xmath32 .
[ prop : phi ] if @xmath163 is strictly positive and @xmath118 denotes the constant unit function on @xmath32 , then @xmath229 see ( * ? ? ?
* lemma 3.8 ) .
[ prop : rlctmonomial ] suppose that @xmath32 is a neighborhood of the origin . if @xmath230 where the @xmath231 are nonnegative integers and the function @xmath232 does not have any zeros , then @xmath233 where @xmath234 this is a special case of theorem [ thm : rlctmonomial2 ] which will be proved later .
recall that an analytic hypersurface @xmath235 is _ singular _ at a point @xmath236 if @xmath56 satisfies @xmath237 if the hypersurface is not singular at any point @xmath236 , then it is said to be @xmath238 .
[ thm : smoothfunctionrlct ] if the hypersurface @xmath239 } is smooth then @xmath240 .
this is also a special case of theorem [ thm : rlctmonomial2 ] .
[ ex : mono_s_t ] following up on example [ ex_tubes ] , we now consider an arbitrary monomial function @xmath241 on the square @xmath242 ^ 2 $ ] .
the tube looks as in figure [ fig : tubes_3 ] .
its area satisfies @xmath243 this formula for the asymptotics ( [ eq : firstorder ] ) follows from definition [ thm : defrlct ] ( iii ) and proposition [ prop : rlctmonomial ] . for the statistical applications in this paper , the relevant functions @xmath170 are polynomials .
they are determinants @xmath244 , where @xmath245 as in section [ sec : intro ] .
let @xmath246 denote the corresponding real log canonical threshold over @xmath242^e$ ] with respect to a positive density @xmath77 .
the theory developed so far says that the real log canonical threshold of the correlation hypersurface gives an asymptotic volume formula for @xmath85 .
[ lemma_unfaithful ] if @xmath247 satisfies the assumptions in proposition [ prop : phi ] , then as @xmath59 tends to zero , the volume of the region @xmath248 ( see ( [ eq : tube_ijs ] ) ) is asymptotically @xmath249 for some constant @xmath250 ( which only depends on @xmath10 ) and @xmath251 . by part ( iii ) in definition
[ thm : defrlct ] , the desired pair @xmath203 is the real log canonical threshold of the partial correlation @xmath252 .
this is the algebraic ( and hence analytic ) function in ( [ eq : parcor ] ) .
this function differs from the polynomial @xmath253 by a denominator that does not vanish over @xmath32 .
that denominator is a unit in the ring of real analytic functions over @xmath32 , and multiplying by a unit does not change the rlct of an analytic function @xcite .
we close this section by relating our results directly to the study of unfaithfulness in @xcite .
[ prop_unfaithful ] under the assumptions in theorem [ lemma_unfaithful ] , as @xmath59 tends to zero , the volume of @xmath59-strong - unfaithful distributions satisfies @xmath254 for some constant @xmath255 . here
@xmath145 is the minimum of the pairs @xmath246 , where @xmath70 runs over all triples in the dag @xmath256 such that @xmath11 is not d - separated from @xmath12 given @xmath49 .
the function @xmath69 is the volume of the union of the regions @xmath93 .
thus , @xmath257 asymptotically , for small positive values of @xmath59 , both the lower and upper bounds vary like a constant multiple of @xmath258 where @xmath145 is the minimum over all pairs @xmath246 . in this minimum , @xmath70 runs over all triples such that @xmath11 and @xmath12 are d - connected given @xmath49 .
the asymptotic integration theory in section [ sec : real_log_threshold ] requires us to analyze the singular locus @xmath259 of the real algebraic hypersurface determined by a given polynomial @xmath170 .
if @xmath259 is empty then the hypersurface is smooth and proposition [ thm : smoothfunctionrlct ] characterizes the asymptotics of the integral . in this section
we return to gaussian graphical models , we develop tools for computing the relevant singular loci , and we show that they are empty in many cases . in many of the remaining cases ,
the singularities are of the monomial type featured in proposition [ prop : rlctmonomial ] .
consider any almost - principal minor @xmath260 of the concentration matrix @xmath30 of a dag @xmath10 .
this is a polynomial function on the parameter space @xmath261 .
this polynomial and its partial derivatives are elements in the polynomial ring @xmath262 $ ] .
the _ jacobian ideal _ of @xmath170 is the ideal in this polynomial ring generated by @xmath170 and its partials .
we denote it by @xmath263 the singular locus @xmath259 is the subvariety of real affine space @xmath264 defined by the jacobian ideal @xmath265 .
the structure of the real variety @xmath259 governs the volume @xmath85 of the set @xmath93 of unfaithful parameters . if @xmath266 then proposition [ thm : smoothfunctionrlct ] tells us that @xmath85 asymptotically equals @xmath267 for some constant @xmath268 . if the singular locus is not empty then understanding @xmath259 is essential for computing its real log canonical threshold @xmath145 .
we conducted a comprehensive study of all dags with few nodes by computing the singular locus for every almost - principal minor in their concentration matrix @xmath30 .
our first result concerns the special case of complete graphs .
non - complete graphs will be studied later .
[ thm : uptosix ] suppose that @xmath247 satisfies the assumptions in proposition [ prop : phi ] .
for any conditional independence statement on the complete directed graph @xmath98 with @xmath269 nodes , we have @xmath266 , and hence @xmath270 for all triples @xmath70 .
it is tempting to conjecture that the hypothesis @xmath269 can be removed in this theorem .
presently we do not know how to approach this problem other than by direct calculation . applying corollary [ prop_unfaithful ]
, this means that the volume of @xmath59-strong - unfaithful distributions for the complete graph satisfies @xmath271 for @xmath79 , which is the best possible behavior regarding strong - faithfulness .
this may be counter - intuitive , but is confirmed in simulations . in figure
[ fig : all_classes ] we plot ( via ( [ funct_plots ] ) ) the proportion of strong - unfaithful distributions @xmath69 for the five graphs in section [ sec : graphs ] for varying values of @xmath59 . especially in the plot for @xmath272 it becomes apparent that the behavior for @xmath79 is very different than , say , for @xmath273 .
for @xmath79 we have @xmath274 , although the chain - like graph is much sparser than the complete graph .
note also that the complete graph @xmath275 has @xmath276 relevant triples @xmath70 , whereas for @xmath277 there are only @xmath278 such triples . in what follows
we explain the algebraic computations that led to theorem [ thm : uptosix ] .
we used ideal - theoretic methods from @xcite in their implementation in the grbner - based software packages macaulay 2 @xcite and singular @xcite . an important point to note at the outset is that the ideal @xmath265 is almost never the unit ideal . by hilbert s nullstellensatz
, this means that the hypersurfaces @xmath279 do have plenty of singular points over the field @xmath280 of complex numbers .
what theorem [ thm : uptosix ] asserts is that , in many of the cases of interest to us here , none of those singular points have their coordinates in the field @xmath281 of real numbers . in order to study the real variety of an ideal ,
techniques from real algebraic geometry are needed .
a key technique is to identify sums of squares ( sos ) .
indeed , the _ real nullstellensatz _ @xcite states that the real variety is empty if and only if the given ideal contains a certain type of sos . to apply this to directed gaussian graphical models
, we shall use the fact that every principal minor of the covariance matrix or the concentration matrix furnishes such an sos .
[ lemma : eins ] every principal minor @xmath282 of the concentration matrix @xmath30 of a dag is equal to @xmath118 plus a sum of squares in @xmath262 $ ] . in particular
, its real variety is empty .
we can write the principal submatrix @xmath283 as the product @xmath284 , where @xmath285 refers to the submatrix with row indices @xmath286 .
thus @xmath283 is the product of an @xmath287-matrix and its transpose . by the cauchy - binet formula
, @xmath282 equals the sum of squares of all maximal minors of the @xmath287-matrix @xmath288 .
one of these maximal minors is the identity matrix .
hence the polynomial @xmath282 has the form @xmath289 .
in particular , the matrix @xmath283 is invertible for all parameter values in @xmath261 .
we note that lemma [ lemma : eins ] holds more generally also in the case of unequal noise variances . in the context of commutative algebra
, it now makes sense to introduce the _ saturations _ @xmath290 these are also ideals in @xmath291 $ ] . by definition ,
@xmath292 consists of all polynomials that get multiplied into the jacobian ideal by some power of the determinant of @xmath283 , and @xmath293 consists of polynomials that get multiplied into @xmath292 by some monomial .
by @xcite , the variety of @xmath294 is the zariski closure of the set - theoretic difference of the variety of @xmath295 and the hypersurface @xmath296 .
we saw in lemma [ lemma : eins ] that the latter hypersurface has no real points .
the ideal @xmath297 represents singularities in @xmath298 .
the singular locus of the real algebraic hypersurface @xmath299 in @xmath261 coincides with the set of real zeros of the ideal @xmath300 .
the set of real zeros of @xmath301 is the zariski closure of the subset of all singular points whose coordinates are non - zero .
we computed the ideals @xmath265 and @xmath294 for every almost - principal minor @xmath302 in the concentration matrices of the graphs @xmath303 . in all cases the ideal @xmath294 was found to equal the unit ideal @xmath304 .
these exhaustive computations were carried out using the software singular @xcite .
this establishes theorem [ thm : uptosix ] .
we briefly discuss our computations for the complete directed graph on six nodes .
[ ex : k6 ] fix the complete directed graph @xmath305 .
we tested all 240 conditional independence statements and computed the corresponding ideal @xmath294 .
we discuss one interesting instance , namely @xmath306 .
the almost - principal minor @xmath307 @xmath308 contains all @xmath309 parameters except @xmath310 .
its determinant is a polynomial of degree @xmath311 . of its @xmath312 partial derivatives ,
@xmath313 have degree @xmath144 .
the derivative with respect to @xmath314 has degree @xmath143 .
thus @xmath315 is generated by @xmath309 polynomials of degrees @xmath316 .
the matrix @xmath317 is the upper left @xmath318-block in the matrix above .
the square of its determinant is a polynomial of degree @xmath319 that happens to lie in the ideal @xmath315 .
this proves @xmath320 . for graphs
@xmath10 that are not complete , @xmath294 may not be the unit ideal .
we already saw one non - obvious instance of this for the tripartite graph in example [ ex : tripartite6 ] .
here is an even smaller example where the jacobian ideal and its saturations are equal , and not the unit ideal .
[ ex : sec4graph1 ] let @xmath321 and take @xmath10 to be the almost - complete graph with adjacency matrix @xmath322 the conditional independence statement @xmath323 is represented by the almost - principal minor @xmath324 of the concentration matrix .
the determinant of this minor factors into two binomial factors : @xmath325 the jacobian ideal is the prime ideal generated by these factors : @xmath326 the left equality holds because @xmath327 is a non - zerodivisor modulo @xmath328 .
the singular locus of ( [ eq : itfactors ] ) is the three - dimensional real variety defined by this binomial ideal in the parameter space @xmath329 .
its real log canonical threshold is found to be @xmath330 .
.rlct for all dags with three nodes [ cols="<,^,^,^",options="header " , ] [ table2 ] this example inspired us to analyze the partial correlations of all small dags with @xmath331 nodes . in our experiments
, we found that @xmath332 is frequently the product of a monomial with a strictly positive sum of squares .
this is the case when there is a unique path which d - connects nodes @xmath11 and @xmath12 given @xmath49 .
for instance , this holds for trees .
such cases are denoted as `` monomial '' in tables [ table1 ] and [ table2 ] . for these
, the rlct is read off directly from proposition [ prop : rlctmonomial ] .
the rows labeled `` smooth '' cover cases that are not monomial but where @xmath333 is the unit ideal , so proposition [ thm : smoothfunctionrlct ] gives us the rlct .
the next theorem summarizes the complete results .
the trivial case @xmath334 is excluded because there is only one graph @xmath335 , with @xmath336 .
here and in tables 1 and 2 we enumerate unlabeled dags .
[ thm : smalldags ] under the assumptions in theorem [ lemma_unfaithful ] , for all dags with @xmath331 nodes and all triples @xmath70 , the value @xmath246 is given in tables [ table1 ] and [ table2 ] . in all cases but one , we have @xmath337 where @xmath338 . to establish theorem [ thm : smalldags ]
, we listed every dag @xmath10 and every triple @xmath70 that is not d - separated in @xmath10 .
the rows `` monomial '' and `` smooth '' were discussed above . on three nodes
there are only 3 partial correlations that correspond to the weighted sum of more than one d - connecting path , namely the partial correlations @xmath339 , @xmath340 , @xmath341 in the complete dag @xmath342 , @xmath343 , @xmath344 .
these are the 3 cases of smooth rlcts in table 1 .
the row `` normal crossing '' refers to cases covered by theorem [ thm : rlctmonomial2 ] .
the `` special '' cases are treated in examples [ ex : sec4graph1 ] and [ ex : specialcases ] .
lastly , the row `` blowup '' represents instances where the real singular locus is a linear space .
our computation of @xmath345 for such instances uses the method in example [ ex:4charts ] .
we now examine the unique exceptional case where @xmath346 .
[ ex : sec4graph2 ] let @xmath321 and @xmath347 .
its concentration matrix may be obtained from example [ ex : sec4graph1 ] by setting @xmath348 . the partial correlation for @xmath323
is now given by @xmath349 for this monomial , proposition [ prop : rlctmonomial ] tells us that @xmath350 . here is an interesting case where the rlct depends in a subtle way on the choice of @xmath32 .
[ ex : specialcases ] consider the conditional independence statement @xmath351 for the dag in figure [ fig:4node ] .
the partial correlation is represented by the almost - principal minor @xmath352 the component @xmath353 is smooth in @xmath354 .
however , it is disjoint from the cube @xmath242 ^ 4 $ ] . to see this , note that @xmath355 in @xmath32 . with this
, @xmath356 would imply @xmath357 and hence @xmath358 , a contradiction .
consequently , if @xmath32 is the cube @xmath157 ^ 4 $ ] then the correlation hypersurface is simply @xmath359 , and the rlct equals @xmath146 by proposition [ prop : rlctmonomial ] .
the other special case with @xmath360 in table 2 comes from swapping the labels of nodes @xmath118 and @xmath128 .
now , if we enlarge the parameter space @xmath32 then the situation changes .
for instance , suppose @xmath361 is in the interior of @xmath32 .
this is a singular point of @xmath362 .
the rlct can be computed by applying theorem [ thm : rlctmonomial2 ] .
it is now @xmath363 instead of @xmath146 .
this example shows that the asymptotics of @xmath364 depends on @xmath32 .
however , it is possible to choose @xmath32 in such a way that further enlargement will not cause the asymptotics of @xmath69 to change .
such a choice could be used as a worst - case analysis for @xmath69 , but to avoid complicating the paper , we will not explore this any further .
we briefly return to the issue of faithfulness in the pc - algorithm .
zhang and spirtes @xcite introduced a variant known as the _ conservative pc - algorithm_. as the name suggests , this algorithm is more conservative and may decide not to orient certain edges .
the conservative pc - algorithm only requires _
adjacency - faithfulness _ for correct inference , which is simply strong - faithfulness restricted to the edges of @xmath10 : @xmath365 if @xmath366 is not adjacent to @xmath286 then the relevant minor equals @xmath367 , where @xmath170 is a polynomial in @xmath368 , the correlation hypersurface is smooth , and @xmath369 . if @xmath366 is adjacent to @xmath286 then the behavior can be more complicated , as seen in example [ ex : specialcases ] .
in @xcite trees were treated as one class . however , as noted when discussing figure [ fig : graphs_lambda ] , there is a striking difference between the volume @xmath69 for chain - like graphs compared to stars . in this section
we give an explanation for this difference based on real log canonical thresholds .
we use the notation @xmath370 for any polynomial that is a sum of squares of polynomials in the model parameters @xmath43 .
suppose that @xmath10 is a tree on @xmath371 and let @xmath96 be the longest length of an undirected path in @xmath10 .
it was shown in ( * ? ? ?
* corollary 4.3(a ) ) that any non - zero almost - principal minor of the concentration matrix @xmath30 has the form @xmath372 where @xmath373 is the monomial of degree @xmath374 formed by multiplying the parameters @xmath53 along the unique path between @xmath11 and @xmath12 .
specifically , for the two trees in figure [ fig : graphs ] we have @xmath375 in both cases , the term @xmath370 disappears when @xmath11 and @xmath12 are leaves of the tree @xmath10 ; cf .
( [ eq : chaincor]),([eq : starcor ] ) .
since the correlation hypersurfaces for trees are essentially given by monomials , we can apply proposition [ prop : rlctmonomial ] .
the minimal real log canonical threshold is @xmath376 where @xmath96 is the largest degree of any of the monomials in ( [ eq : sosmono ] ) .
corollary [ prop_unfaithful ] implies the following result : [ thm : treev ] under the assumptions in theorem [ lemma_unfaithful ] , if @xmath10 is a tree then the volume of @xmath59-strong - unfaithful distributions satisfies @xmath377 where @xmath96 is the length of the longest path in the tree @xmath10 , and @xmath87 is a suitable constant . for the case of stars we have @xmath378 , whereas for chain - like graphs we have @xmath379 .
[ for : treecor ] under the assumptions in theorem [ thm : treev ] , the volume @xmath69 of strong - unfaithful distributions satisfies @xmath380 where @xmath381 and @xmath382 are suitable positive constants . as a consequence ,
the volume @xmath69 is asymptotically larger for chains compared to stars , and the difference increases with increasing number of nodes @xmath72 .
this furnishes an explanation for figure [ fig : graphs_lambda ] , at least for small values of @xmath59 . in that figure
we saw the curve for the chain lying clearly above the curve for the star tree .
however , one subtle issue is the size of the constants @xmath381 and @xmath382 .
these need to be understood in order to make accurate comparisons . in section [ sec : integrals ] , we develop new theoretical results regarding the computation of the constant @xmath87 in ( [ eq : firstorder ] ) .
theorem [ thm_constants ] gives an integral representation for @xmath87 when the partial correlation hypersurface is essentially defined by a monomial .
in example [ ex : chainstarconstants ] we shall then derive : [ cor__const_trees ] the two constants in ( [ eq : treev ] ) are @xmath383 this result surprised us at first .
it establishes the counterintuitive fact that , as @xmath72 grows , the constant for the lower curve in figure [ fig : graphs_lambda ] is exponentially larger than that for the upper curve .
therefore , in order to fully explain the relative position of the two curves for a wider range of values of @xmath168 , it does not suffice to just consider the first order asymptotics ( [ eq : firstorder ] ) .
instead , we need to consider some of the higher order terms in the series expansion ( [ eq : asymp ] )
. as we shall see in section [ sec : integrals ] , it is difficult to determine the constants @xmath384 in ( [ eq : asymp ] ) analytically . in the remainder of this section , we propose a procedure based on simulation and linear regression for estimating the constants @xmath384 in the asymptotic explanations of the volumes @xmath69 or @xmath85 . for simplicity
we focus on the latter case and we take @xmath385 .
suppose that @xmath10 is a dag for which the real log canonical thresholds @xmath145 in theorem [ lemma_unfaithful ] and corollary [ prop_unfaithful ] are known .
this is the case for all trees by theorem [ thm : treev ] .
our procedure goes as follows .
we first sample @xmath76 points uniformly from @xmath32 and compute the proportion of points @xmath56 that lie in @xmath386 for different values of @xmath59 .
we then fit a linear model to @xmath387 where @xmath203 is the known real log canonical threshold . in the following ,
we illustrate this procedure for chains and stars .
we analyze two specific examples of partial correlation volumes , namely the ones corresponding to the longest paths in each graph , that is @xmath388 for @xmath151 and @xmath389 for @xmath150 . for chain - like graphs , @xmath390 whereas for star graphs , @xmath391 we first approximate @xmath388 for chain - like graphs and @xmath389 for star graphs by simulation for various values of @xmath59 .
this means that we sample @xmath76 points uniformly in the @xmath392-dimensional parameter space @xmath32 , and we count how many of them are @xmath393 .
the results for @xmath394 and @xmath272 are shown in figure [ rplot_chains_stars ] .
these are based on a sample size of @xmath395 .
we then fit a linear model @xmath396 for chain - like graphs .
the curve resulting from the regression estimates is shown in black in figure [ rplot_chains_stars ] .
the curve resulting from the first - order approximation with the constants computed using corollary [ cor__const_trees ] is shown in grey in figure [ rplot_chains_stars ] .
we note that especially for chain - like graphs , where the true constant in corollary [ cor__const_trees ] is small , the first order approximation is very bad .
the approximation by regression on the other hand is a fast way to get pretty accurate estimates of all constants .
the same was done with star graphs , but with the linear model @xmath397 figure [ rplot_chains_stars ] shows that the first - order approximation is more accurate for stars than for chains .
we now discuss the problem of quantifying bias in causal models .
our point of departure is greenland s paper @xcite , where the problem of quantifying bias has been discussed for binary variables .
in contrast to the previous sections , in the situation discussed here , a large tube volume is in fact desired since it corresponds to small bias . in this section
we use the notation @xmath398 for the almost - principal minor @xmath302 of the concentration matrix .
we are interested in estimating the direct effect of an exposure @xmath5 on a disease outcome @xmath86 ( i.e. the coefficient on the edge @xmath399 ) from the partial correlation @xmath400 , where @xmath49 is a subset of the measurable variables .
this partial correlation is a weighted sum over all _ open paths _ ( i.e paths which d - connect @xmath5 to @xmath86 ) given @xmath49 ( the direct path @xmath401 being just one of them ) . for estimating the direct effect @xmath401 from @xmath400 ,
all open paths other than the direct path are thus considered as bias .
we shall analyze two forms of bias which are of particular interest in practice , namely confounding bias and collider - stratification bias .
we start by defining collider - stratification bias .
suppose we are given a dag @xmath10 with @xmath402 and there is another node @xmath87 such that @xmath403 this says that @xmath87 is a collider on a path from @xmath86 to @xmath5 .
stratifying ( i.e. conditioning ) on @xmath87 opens a path between @xmath5 and @xmath86 leading to bias when estimating @xmath401 .
the partial correlation corresponding to the opened path between @xmath5 and @xmath86 is known as _ collider - stratification bias_. collider - stratification bias arises for example in the context of discrete variables , where instead of obtaining a random sample from the full population , a random sample is obtained from the subpopulation of individuals with a particular level of c. [ ex : tripartitecollider ] we illustrate collider - stratification bias for the tripartite graph @xmath404 shown in figure [ fig : tripart_51 ] .
let node @xmath118 represent the exposure @xmath5 and node @xmath128 the disease outcome @xmath86 .
in this example , node @xmath144 is a collider @xmath87 for multiple paths between @xmath5 and @xmath86 .
when stratifying on @xmath405 , node @xmath406 is d - connected to node @xmath407 via the following paths : @xmath408 the bias introduced for estimating the direct effect of @xmath5 on @xmath86 when conditioning on @xmath87 is @xmath409 the numerator @xmath410 is the weighted sum of all open paths between @xmath5 and @xmath86 .
similarly , nodes @xmath411 and @xmath143 are colliders for multiple paths .
the bias when conditioning on these is @xmath412 problem [ prob : tripart ] is about comparing the tube volume for ( [ eq : bias1 ] ) with the tube volume for ( [ eq : bias2 ] ) .
a question of practical interest in causal inference is to understand the situations in which stratifying on a collider leads to a particularly large bias .
it is widely believed that collider - stratification bias tends to attenuate when it arises from more extended paths ( see @xcite ) .
what follows is our interpretation of this statement as a precise mathematical conjecture .
[ prob : tripart ] let @xmath413 and @xmath414 .
we introduce a partial order on the _ collider set _ @xmath415 by setting @xmath416 if all paths on which @xmath87 is a collider also go through @xmath97 .
given subsets @xmath417 we set @xmath418 if for all @xmath419 there exists @xmath420 such that @xmath421 .
if this holds , then the bias introduced when conditioning on @xmath49 should be smaller than when conditioning on @xmath422 . to make this precise , we conjecture : @xmath423.\ ] ] we now study this conjecture for the tripartite graphs @xmath109 . here
, it says that the collider - stratification bias introduced when conditioning on the third level @xmath424 is in general smaller than when conditioning on the second level of nodes @xmath425 , i.e. @xmath426 this inequality is confirmed by the simulations shown in figure [ rplot_tripartite ] .
here @xmath116 , @xmath427 is shown in red and @xmath272 , @xmath427 is shown in blue .
the solid lines correspond to the volume @xmath428 , whereas the dashed lines correspond to the volume @xmath429 .
going beyond simulations , we now present an algebraic proof of our conjecture when @xmath59 is small , for the tripartite graphs in figure [ fig : tripart_p ] where the second level has only one node .
[ ex : tripart_1 ] for @xmath430 the left hand side of ( [ conj_tripart ] ) is given by @xmath431 depending on the values of @xmath72 , the corresponding real log canonical threshold is given by @xmath432 for @xmath433 this was example [ ex : sec4graph2 ] .
to prove ( [ rlct_tripart_1 ] ) for @xmath434 , we need two ingredients .
firstly , if the polynomial is a product of factors with disjoint variables , then the rlct is the minimum of the rlct of the factors , taken with multiplicity ( e.g. if the rlcts are @xmath435 and @xmath436 , then the combined rlct is @xmath437 , just like in the case of a monomial ) . secondly , the rlct of a sum of squares of @xmath198 unknowns is equal to @xmath438 .
we saw this in example [ ex : d - ball ] .
for the right hand side of ( [ conj_tripart ] ) , we condition on node @xmath411 .
now , the defining polynomial is @xmath439 by proposition [ prop : rlctmonomial ] , this has rlct @xmath363 , which is larger or equal to all values of @xmath203 in ( [ rlct_tripart_1 ] ) . to compare the behavior of @xmath440 and @xmath441 for small @xmath59 , we will need to derive the constant @xmath87 in ( [ eq : firstorder ] ) .
in example [ ex : bowtieconstant ] , we will show that if @xmath442 and the parameter space is @xmath443 then the asymptotic constants are given by @xmath444 and @xmath445 .
we conclude that @xmath446 for small values of @xmath447 , as conjectured in problem [ prob : tripart ] .
a slight twist to example [ ex : tripart_1 ] is the almost - chain graph shown in figure [ fig : almost_chain ] , with edges @xmath448 . for such graphs , problem
[ prob : tripart ] asks whether @xmath449 this holds for small @xmath59 because @xmath450 . by proposition [ prop : rlctmonomial ] ,
@xmath451 so @xmath452 for @xmath453 .
in example [ ex : tripart_1 ] we resolved problem [ prob : tripart ] for tripartite graphs whose middle level consists of one node .
we next consider the case @xmath454 where the third level has one node .
[ ex : tripart_1third ] the graph @xmath455 shown in figure [ fig : tripart_51 ] was discussed in example [ ex : tripartitecollider ] .
we focus on the numerators in ( [ eq : bias1 ] ) and in ( [ eq : bias2 ] ) .
the polynomial ( [ eq : bias1 ] ) will be studied in example [ ex : tripartite_blowup ] where we prove that @xmath456 .
using the same method for @xmath457 gives @xmath458 thus , we conclude that @xmath459 for small values of @xmath460 .
[ ex : tripart_62 ] for the graph @xmath461 in figure [ fig : tripart_62 ] we check if @xmath462 for small @xmath59 . as before ,
@xmath463 , but a hard computation using the tools of section [ sec : resolution ] reveals that now @xmath464 .
thus , knowledge of the rlct is not sufficient to establish ( [ eq : conjecture1 ] ) .
what is needed is a finer analysis along the lines of section [ sec : integrals ] .
the second form of bias studied by greenland @xcite is confounder bias . in the context of a directed graphical model @xmath10 ,
a _ confounder _ for the effect of @xmath5 on @xmath86 is a node @xmath87 such that @xmath465 the partial correlation introduced by the path from @xmath5 to @xmath86 passing through @xmath87 is referred to as _
confounder bias_. in such situations , stratifying on @xmath87 blocks the path between @xmath5 and @xmath86 ( i.e. @xmath87 d - separates @xmath5 from @xmath86 ) and therefore corresponds to bias removal . in certain graphs , such as the bow - tie example in @xcite
, there are variables where stratifying removes confounder bias but at the same time introduces collider - stratification bias . for instance , consider the graph @xmath466 , where node @xmath143 corresponds to exposure @xmath5 and node @xmath144 corresponds to disease outcome @xmath86 . then conditioning on node @xmath411 blocks the paths @xmath467 and therefore reduces confounder bias , but opens the path @xmath468 and therefore introduces collider - stratification bias .
this trade - off is of particular interest in situations where one can not condition on @xmath118 and @xmath128 , for example because these variables were unmeasured .
it is believed that in such examples the bias removed by conditioning on the confounders is larger than the collider - stratification bias introduced and one should therefore stratify .
we translate this statement into the following mathematical problem : [ prob : bowtie ] let @xmath413 and we denote by @xmath469 the confounder - collider subset , i.e. @xmath470 we conjecture the following inequality for the relevant tube volumes : @xmath471.\ ] ] this conjectural inequality is interesting for the bow - tie graphs @xmath472 .
it means that conditioning on the nodes in the second level reduces bias since the bias removed by conditioning on the confounders is larger than the collider - stratification bias introduced by conditioning : @xmath473 this is confirmed by our simulations in figure [ rplot_bowties ] , for @xmath116 in red and @xmath272 in blue .
the solid line corresponds to the volume @xmath474 and the dashed line corresponds to @xmath475 . in the following example
we prove the inequality ( [ conj : bow ] ) for @xmath116 and small @xmath168 .
let @xmath476 as in figure [ fig : bowtie ] . the left hand side ( [ conj : bow ] )
is represented by @xmath477 this monomial is the path in ( [ paths2 ] ) .
the corresponding real log canonical threshold is @xmath478 .
the polynomial representing the right hand side ( [ conj : bow ] ) is a weighted sum of the paths in ( [ paths_1 ] ) : @xmath479 we derive its real log canonical threshold using the blowups described in section [ sec : resolution ] .
we find that it is @xmath146 .
since @xmath480 , we conclude @xmath481 .
in this section we develop more refined techniques for computing real log canonical thresholds .
the following theorem combines the monomial case of proposition [ prop : rlctmonomial ] with the smooth case of proposition [ thm : smoothfunctionrlct ] . as promised in section [ sec : real_log_threshold ] , this furnishes the proofs for these two propositions .
[ thm : rlctmonomial2 ] suppose @xmath482 and @xmath483 where @xmath484 are nonnegative integers , @xmath485 are positive integers , and the hypersurface @xmath486 is either empty or smooth and normal crossing ( see definition below ) with @xmath487 .
we write @xmath488 , @xmath489 and @xmath490 , and we define @xmath491 where @xmath492 is the set of all indices @xmath493 such that @xmath494 has a zero in @xmath32 .
then we have @xmath495 provided the equations @xmath496 for @xmath497 have a solution in the interior of @xmath32 .
the normal crossing hypothesis in theorem [ thm : rlctmonomial2 ] means that the system @xmath498 does not have any solutions in @xmath32 .
see @xcite to learn more about normal crossing singularities .
we begin with a technical lemma establishing that the rlct can be computed locally .
[ lemma : rlctlocalization ] for every @xmath499 , there exists a neighborhood @xmath500 of @xmath192 such that @xmath501 for all neighborhoods @xmath502 of @xmath192 .
moreover , @xmath503 where we take the minimum over all @xmath192 in the real analytic hypersurface @xmath504 .
this comes from ( * ? ? ?
* lemma 3.8 , proposition 3.9 ) .
lemma [ lemma : rlctlocalization ] states that @xmath505 is the minimum of @xmath506 as @xmath192 varies over @xmath32 .
writing each subset @xmath507 as @xmath508 where @xmath509 is a sufficiently small neighborbood of @xmath192 in @xmath510 , we claim that @xmath511 if this minimum is attained in the interior of @xmath32 . indeed , for @xmath192 in the interior of @xmath32 ,
we get @xmath512 . otherwise , the volume of @xmath513 is less than that of @xmath514 for all @xmath59 .
hence @xmath515 , and the claim follows .
now to prove theorem [ thm : rlctmonomial2 ] , it suffices to show that for each @xmath516 we have @xmath517 where @xmath518 is the set of all indices @xmath493 that satisfy @xmath519 . without loss of generality ,
suppose @xmath520 where @xmath521 and @xmath522 are nonzero .
if @xmath523 , we may divide @xmath524 by @xmath525 without changing the rlct in a sufficiently small neighborhood @xmath509 of @xmath192 .
the rlct of the remaining monomial is determined by ( * ? ? ?
* proposition 3.7 ) .
now , let us suppose @xmath526 . because @xmath525 is normal crossing with @xmath527 , one of the derivatives @xmath528 must be nonzero at @xmath192 for some @xmath529 .
we assume @xmath509 is sufficiently small so that this derivative and @xmath530 do not vanish .
consider the map @xmath531 given by @xmath532 the jacobian matrix of @xmath533 is nonsingular , so this map is an isomorphism onto its image .
set @xmath534 and @xmath535 .
then , for all @xmath536 , we have @xmath537 where the factors @xmath538 and @xmath539 do not vanish in @xmath540 . by using the chain rule (
* proposition 4.6 ) , we get @xmath541 . the latter rlct can be computed once again by dividing out the nonvanishing factors and applying ( * ? ? ?
* proposition 3.7 ) .
the hypersurface @xmath542 may not satisfy the hypothesis in theorem [ thm : rlctmonomial2 ] . in that case
, we can try to simplify its singularities via a change of variables @xmath543 . with some luck
, the transformed hypersurface @xmath544 will be described locally by monomials and the rlct can be computed using theorem [ thm : rlctmonomial2 ] .
more precisely , let @xmath540 be a @xmath198-dimensional real analytic manifold and @xmath545 a real analytic map that is _ proper _ , i.e. the preimage of any compact set is compact
. then @xmath546 _ desingularizes _
@xmath524 if it satisfies the following conditions : 1 .
the map @xmath546 is an isomorphism outside the variety @xmath547 .
2 . given any @xmath548
, there exists a local chart with coordinates @xmath549 such that @xmath550 where @xmath551 is the jacobian determinant , the exponents @xmath552 are nonnegative integers and the real analytic functions @xmath553 do not vanish at @xmath191 .
if such a desingularization exists , then we may apply @xmath546 to the volume function to calculate the rlct .
care must be taken to multiply the measure @xmath77 with the jacobian determinant @xmath554 in accordance with the change - of - variables formula for integrals .
hironaka s celebrated theorem on the resolution of singularities @xcite guarantees that such a desingularization exists for all real analytic functions @xmath524 .
the proof employs transformations known as _ blowups _ to simplify the singularities .
we now describe the blowup @xmath555 of the origin in @xmath510 .
the manifold @xmath540 can be covered by local charts @xmath556 such that each chart is isomorphic to @xmath510 and each restriction @xmath557 is the monomial map @xmath558 here , the coordinate hypersurface @xmath559 , also called the _ exceptional divisor _ , runs through all the charts .
if the origin is locally the intersection of many smooth hypersurfaces with distinct tangent hyperplanes , then these hypersurfaces can be separated by blowing up the origin @xcite .
[ ex : fourlines ] consider the curve @xmath560 in figure [ fig : tubes_4 ] . to resolve this singularity , we blow up the origin . in the first chart ,
the map is @xmath561 , so @xmath562 the lines @xmath563 , @xmath564 and @xmath565 are transformed to @xmath566 , @xmath567 and @xmath568 respectively in this chart , thereby separating them .
the line @xmath569 does not show up here , but it appears as @xmath570 in the second chart , where @xmath571 and @xmath572 since the curve @xmath573 is normal crossing with @xmath574 in the first chart , we can now apply theorem [ thm : rlctmonomial2 ] . the chain rule ( * ? ? ?
* proposition 4.6 ) shows that @xmath575 is the minimum of @xmath576 for @xmath577 . in both charts ,
this rlct equals @xmath578 .
[ ex:4charts ] let @xmath321 and @xmath10 be the almost complete dag with @xmath579 .
we consider the conditional independence statement @xmath580 .
the correlation hypersurface is defined by @xmath581 the real singular locus is a line in the parameter space @xmath582 , since @xmath583 .
blowing up this line in @xmath582 creates four charts @xmath584 .
for instance , the first chart has @xmath585 then @xmath170 transforms to @xmath586 where @xmath587 .
the hypersurface @xmath588 has no real singularities , so it is smooth in @xmath589
. we can thus apply theorem [ thm : rlctmonomial2 ] with @xmath590 to find @xmath591 .
the behavior is the same on @xmath592 and @xmath593 , and we conclude that @xmath594 .
this example is one of the @xmath595 cases that were labeled as `` blowup '' in table [ table2 ] .
the other @xmath596 cases are similar .
[ ex : tripartite_blowup ] in example [ ex : tripart_1third ] we claimed that @xmath597 for @xmath404 .
we now prove this claim by using the blowing up method .
the polynomial in question is @xmath598 the singular locus of the hypersurface @xmath142 is given by @xmath599 we blow up the linear subspace @xmath600 in @xmath601 .
this creates two charts .
the map for the first chart is @xmath602 .
this map gives @xmath603 now , by setting @xmath604 and @xmath605 , the transformed function @xmath606 is the monomial @xmath607 .
then theorem [ thm : rlctmonomial2 ] can be employed to evaluate @xmath608 .
the calculation in the second chart is completely analogous .
the same approach as in example [ ex : tripartite_blowup ] can be applied to the polynomial @xmath609 in example [ ex : tripartite6 ] .
a lengthy calculation , involving many charts and multiple blowups , eventually reveals that @xmath117 satisfies @xmath610 .
this was stated in example [ ex : tripart_62 ] .
we now describe a method for finding the constant @xmath87 in the formula @xmath611 in ( [ eq : firstorder ] ) .
the two theorems in this section are new and they extend the results of greenblatt @xcite and lasserre @xcite on the volumes of sublevel sets . unless stated otherwise , all measures used in this section are the standard lebesgue measures .
we begin by showing that the constant @xmath87 is a function of the highest order term in the laurent expansion of the zeta function of @xmath170 .
[ thm : volumeconstant ] given real analytic functions @xmath612 , consider the laurent expansion of @xmath613 where @xmath95 is the smallest pole and @xmath96 its multiplicity .
then , asymptotically as @xmath59 tends to zero , @xmath614 according to the proof of ( * ? ? ?
* theorem 7.1 ) , the volume function @xmath166 equals @xmath615 where @xmath616 is the state density function and @xmath617 is the delta function . now , using the proof of ( * ? ? ?
* theorem 3.16 ) , we obtain @xmath618 here we used the little - o notation . finally , using integration by parts , we find that @xmath619 in example [ ex : d - ball ] , we saw that the volume of the @xmath198-dimensional ball defined by @xmath620 is equal to @xmath621 for some positive constant @xmath87 .
we here show how to compute that constant using asymptotic methods . by lemma [ thm : volumeconstant ]
, @xmath622 where @xmath623 is the coefficient of @xmath624 in the laurent expansion of the zeta function @xmath625 computing this laurent coefficient from first principles is not easy .
instead , we derive @xmath623 using the asymptotic theory of laplace integrals .
the connection between such integrals and volume functions was alluded to in definition [ thm : defrlct ] . by (
* proposition 5.2 ) , the laplace integral @xmath626 is asymptotically @xmath627 for large @xmath628 . but
this laplace integral also decomposes as @xmath629 where each factor is the classical gaussian integral .
solving for @xmath623 leads to the formula @xmath630 -0.5 cm in section [ sec : real_log_threshold ] , we saw how the rlcts of smooth hypersurfaces and of hypersurfaces defined by monomial functions can be computed .
the following two theorems and their accompanying examples demonstrate how the asymptotic constant @xmath87 can also be evaluated in those instances . here
, we say that two hypersurfaces _ intersect transversally _ in @xmath510 if the points of intersection are smooth on the hypersurfaces and if the corresponding tangent spaces at each intersection point generate the tangent space of @xmath510 at that point .
[ thm : smoothconstant ] let @xmath142 be a smooth hypersurface and let @xmath163 be positive .
suppose @xmath631 is nonvanishing in @xmath32 .
let @xmath193 be the projection of the hypersurface @xmath632 onto the subspace @xmath633 and let @xmath634 be the map @xmath635 .
if the boundary of @xmath32 intersects transversally with the hypersurface @xmath636 , then @xmath637 asymptotically ( as @xmath79 ) , where @xmath638 the asymptotics of the volume @xmath166 depends only on the region @xmath639 .
so we may assume that @xmath32 is a small neighborhood of the hypersurface @xmath542 .
as we saw in the proof of theorem [ thm : rlctmonomial2 ] , the map @xmath546 is an isomorphism onto its image .
thus after changing variables , the zeta function associated to @xmath166 becomes @xmath640 here , the lower and upper limits @xmath641 straddle zero because the boundary of @xmath32 is transversal to the hypersurface . by substituting the taylor series @xmath642 and the exponential series @xmath643
, we get the laurent expansion @xmath644_0^{\varepsilon_2 } + \cdots \\ & = \frac{1}{1-z}\cdot \frac{\varphi}{|\partial_{\omega_1 } f| } \circ \rho^{-1}(0 , \omega_2 , \ldots , \omega_d ) + \cdots.\end{aligned}\ ] ] the same is true for the integral from @xmath645 to @xmath66 .
the result now follows from lemma [ thm : volumeconstant ] .
by theorem [ thm : uptosix ] , all conditional independence statements in small complete graphs lead to smooth hypersurfaces . here
we analyze the statement @xmath646 in the complete 3-node dag .
this example was studied in @xcite .
the corresponding partial correlation is @xmath647 this partial correlation hypersurface lives in @xmath648 and it is depicted in ( * ? ? ?
* figure 2(b ) ) .
we apply theorem [ thm : smoothconstant ] by setting @xmath649 ^ 3 $ ] , @xmath650 and @xmath651 , the uniform distribution on @xmath32 .
we choose @xmath652 to be @xmath314 .
then @xmath653 .
the projection @xmath193 of the surface @xmath654 onto @xmath655 ^ 2 \}\,$ ] is the whole square @xmath157 ^ 2 $ ] .
the formula for the constant @xmath87 now simplifies to @xmath656 this two - dimensional integral was evaluated numerically using ` mathematica ` .
we now come to the monomial case that was discussed in theorem [ thm : rlctmonomial2 ] .
[ thm_constants ] let @xmath232 and @xmath163 be positive , and let @xmath657 where the @xmath231 are nonnegative integers .
suppose that @xmath658 and that the boundary of @xmath32 is transversal to the subspace @xmath659 defined by @xmath660 .
let @xmath661 and @xmath662 denote the vectors @xmath663 and @xmath664 respectively .
then @xmath665 asymptotically as @xmath59 tends to zero where @xmath666 let us suppose for now that @xmath32 is the hypercube @xmath667^d$ ] .
our goal is to apply lemma [ thm : volumeconstant ] by computing the laurent coefficient @xmath668 of the zeta function @xmath669 .
we first study the taylor series expansion of the integrand about @xmath670 .
this gives @xmath671 the higher order terms in this expansion contribute larger poles to @xmath669 , so we only need to compute the coefficient of @xmath672 in the laurent expansion of @xmath673 because @xmath674 is positive , the last integral in the above expression does not have any poles near @xmath212 , so the constant term in its laurent expansion comes from substituting @xmath212 . hence , @xmath675 now suppose @xmath32 is not the hypercube .
since the boundary of @xmath32 is transversal to the subspace @xmath659 , we decompose @xmath32 into small neighborhoods which are isomorphic to orthants .
summing up the contributions from these orthants gives the desired result .
we revisit the planar tubes shown in figures [ fig : tubes_1][fig : tubes_3 ] . using the formula ( [ eq : constant ] ) in theorem [ thm_constants ]
, one can easily check the constants @xmath87 we saw in example [ ex_tubes ] , namely @xmath676 [ ex : chainstarconstants ] we apply theorem [ thm_constants ] to find the constants in corollary [ cor__const_trees ] for chains and stars . in both cases we set @xmath242^{p-1}$ ] and @xmath677 . for chains we have @xmath678 and
@xmath659 is the subspace @xmath679 .
then the integral in ( [ eq : constant ] ) is the evaluation of the denominator of ( [ eq : chaincor ] ) at the origin multiplied by @xmath77 , so @xmath680 as claimed . for stars ,
@xmath182 is achieved by @xmath681 and @xmath682 with @xmath683 since @xmath684 , the quantity @xmath69 is the volume of the union of the tubes @xmath685 over all @xmath686 . by applying formula ( [ eq : constant ] ) , the asymptotic volume of each tube computes to @xmath687 .
meanwhile , the volumes of the intersections of these tubes become negligible as @xmath688 . after summing over all @xmath686 , we get @xmath689 . [ ex : bowtieconstant ] we compute the constant @xmath87 of the volume @xmath690 for @xmath691 as in example [ ex : tripart_1 ]
. let @xmath692 and @xmath32 be given by ( [ eq : newparamspace ] ) .
we are interested in the tube @xmath693 the measure on @xmath32 is @xmath694 where @xmath695 and @xmath696 is the lebesgue probability measure on the ball @xmath697 .
according to theorem [ thm_constants ] , @xmath698 by substituting spherical coordinates for the integration , this expression simplifies to @xmath699 yielding the constant @xmath700 needed for the bias reduction analysis in example [ ex : tripart_1 ] .
in this paper we examined the volume of regions in the parameter space of a directed gaussian graphical model that are given by bounding partial correlations .
we established a connection to singular learning theory , and we showed that these volumes can be computed by evaluating the real log canonical threshold of the partial correlation hypersurfaces . throughout the paper
we have made the simplifying assumption of equal noise , i.e. @xmath24 .
ideally , one would like to allow for different noise variances .
this would increase the dimension of the parameter space @xmath32 .
it would be very interesting to study this more difficult situation and understand how the asymptotic volumes change , or more generally , how the asymptotics depends on our choice of the parameter space @xmath32 .
this issue was discussed briefly in example [ ex : specialcases ] .
this paper can be seen as a first step towards developing a theory which would allow to compute the complete asymptotic expansion of particular volumes .
we have concentrated on computing only the leading coefficients of these expansions , and even this question is still open in many cases ( e.g. example 6.6 ) .
an interesting extension would be to better understand how to use properties of the graph to compute the coefficients @xmath701 in the asymptotic expansion . finally , another interesting problem for future research would be to ascertain all values of @xmath702 for which @xmath701 is non - zero , in terms of intrinsic properties of the underlying graph .
this project began at the workshop `` algebraic statistics in the alleghenies '' , which took place at penn state in june 2012 .
we are grateful to the organizers for a very inspiring meeting .
we thank thomas richardson for pointing out the connection to the problem of bias reduction in causal inference .
we also thank all reviewers for thoughtful comments and one of the reviewers in particular for numerous detailed comments and for spotting a mistake in a proof .
this work was supported in part by the us national science foundation ( dms-0968882 ) and the darpa deep learning program ( fa8650 - 10-c-7020 ) .
s. chaudhuri and t.s .
richardson : using the structure of d - connecting paths as a qualitative measure of the strength of dependence , _ proceedings of the nineteenth conference on uncertainty in artificial intelligence _ , pages 116123 , 2003 . | an asymptotic theory is developed for computing volumes of regions in the parameter space of a directed gaussian graphical model that are obtained by bounding partial correlations .
we study these volumes using the method of real log canonical thresholds from algebraic geometry .
our analysis involves the computation of the singular loci of correlation hypersurfaces .
statistical applications include the strong - faithfulness assumption for the pc - algorithm , and the quantification of confounder bias in causal inference .
a detailed analysis is presented for trees , bow - ties , tripartite graphs , and complete graphs . * * c shaowei lin@xmath0 , caroline uhler@xmath1 , bernd sturmfels@xmath0 and peter bhlmann@xmath2 + + @xmath0department of mathematics , uc berkeley + @xmath1ist austria + @xmath2seminar for statistics , eth zrich + |
throughout this paper , all fields have characteristic @xmath2 .
first we recall some definitions and notations .
let @xmath5 be the valuation ring associated with @xmath6 .
it is a local ring , and @xmath7 is its maximal ideal .
let @xmath8 be the residue field ( or @xmath9 when there is no danger of confusion ) .
we let @xmath10 denote the _ compositum _ of all finite galois extensions of @xmath11 of degree a power of @xmath0 .
a valued field @xmath1 is @xmath0-henselian if @xmath6 extends uniquely to @xmath10 .
equivalently ( see @xcite , thm 4.3.2 ) , if it satisfies a restricted version of hensel s lemma ( which we call @xmath0-hensel lemma ) : @xmath11 is @xmath0-henselian if and only if every polynomial @xmath12 $ ] which splits in @xmath10 and with residual image in @xmath13 $ ] having a simple root @xmath14 in @xmath15 , has a root @xmath16 in @xmath17 with @xmath18 .
furthermore , another result ( see @xcite , thm 4.2.2 ) shows that @xmath1 is @xmath0-henselian if and only if @xmath6 extends uniquely to every galois extension of degree @xmath0 .
+ the aim of this note is to give a complete proof of the following result : + * theorem*. let @xmath1 be a valued field .
@xmath1 is @xmath0-henselian _ if and only if _ @xmath19 .
+ this result was announced in @xcite , proposition 1.4 , however the proof was not complete .
the notion of @xmath0-henselianity is important in the study of fields with definable valuations , and in particular it is important to show that the property of @xmath0-henselianity is an elementary property of valued fields .
the proof we give is elementary , and uses extensively pseudo - convergent sequences and their properties . recall that a sequence @xmath20 indexed by an ordinal @xmath21 is said to be _ pseudo - convergent _ if for all @xmath22 : @xmath23 a pseudo - convergent sequence @xmath24 is called _ algebraic _ if there is a polynomial @xmath25 in @xmath26 $ ] such that @xmath27 ultimately for all @xmath28 , _ i.e _ : @xmath29 otherwise , it is called _
transcendental_. we assume familiarity with the properties of pseudo - convergent sequences , see @xcite for more details , and in particular theorem 3 , lemmas 4 and 8 .
first , we prove a lemma in order to restrict our study to immediate extensions : + * observation*. _ let @xmath1 be a valued field and @xmath30 be a galois extension of degree a prime @xmath31 .
then , if @xmath32 is residual or ramified , @xmath33 is the unique extension of @xmath6 to @xmath34 . _ the fundamental equality of valuation theory ( see @xcite , thm 3.3.3 ) tells us that if @xmath34 is a galois extension of @xmath11 , then @xmath35 = e(l / k)f(l / k)gd\ ] ] where @xmath36 is the ramification index , @xmath37 the residue index , @xmath38 the number of extensions of @xmath6 to @xmath34 and @xmath39 , the defect , is a power of @xmath0 .
the forward direction is an immediate application of the @xmath0-hensel lemma.conversely , assume that @xmath44 .
every galois extension of @xmath11 of degree @xmath0 is an artin - schreier extension , _
i.e _ is generated over @xmath11 by a root @xmath16 of a polynomial @xmath45 , with @xmath46 .
the previous observation gives us the result when @xmath47 is not immediate .
let @xmath34 be an immediate galois extension of degree @xmath0 and @xmath48 an extension of @xmath6 to @xmath34 ( hence with the same value group @xmath49 and residue field @xmath50 as @xmath11 ) .
we can write @xmath51 where @xmath52 .
+ step 1 : ( claim ) the set @xmath53 is contained in @xmath54 and has no last element .
first observe that @xmath55 : if @xmath56 , then the equation @xmath57 has a root in @xmath11 , so that @xmath58 : contradiction .
let @xmath59 , @xmath60 such that @xmath61 . as @xmath62 is immediate
there is @xmath63 such that @xmath64 . if @xmath65 then @xmath66 and @xmath67 in @xmath68 , which as above give a contradiction .
hence @xmath69 , and from @xmath70 , we deduce that @xmath71 , and @xmath72 .
this shows the claim .
+ step 2 : we extract a strictly well - ordered increasing and cofinal sequence from @xmath73 .
if we write @xmath74 , we get a sequence @xmath24 in @xmath11 such that the sequence @xmath75 is stricly increasing and cofinal in @xmath73 .
thus , the sequence @xmath76 is pseudo - convergent ( with @xmath77 one of its limits ) . as @xmath78
, we have @xmath79 for @xmath80 .
thus , the sequence @xmath81 is also pseudo - convergent .
furthermore , @xmath24 has no limit in @xmath11 : if @xmath82 is a limit of @xmath24 then @xmath83 is a limit of @xmath76 . as @xmath84 is cofinal in @xmath73
, @xmath85 would be a maximal element of @xmath73 : contradiction .
+ step 3 : ( claim ) let @xmath86 $ ] , and assume that @xmath87 is strictly increasing ultimately .
then @xmath88 .
+ we take such a @xmath89 of minimal degree , assume this degree is @xmath90 , and will derive a contradiction .
one consequence of lemma 8 in @xcite is that : @xmath91 where @xmath92 is the ultimate valuation of @xmath93 and @xmath94 is the valuation of @xmath95 for @xmath96 ( which does not depend on @xmath97 as @xmath24 is pseudo - convergent ) .
we write @xmath98 with @xmath99 . then , @xmath100 is ultimately of constant valuation , and we let @xmath101 be this valuation . as @xmath24 has no limit in @xmath11 , it is easy to see that @xmath102 , so that @xmath103 . by lemma 4 in @xcite
, there is an integer @xmath104 such that we have ultimately : @xmath105 then , ultimately : @xmath106 thus , we have ultimately @xmath107 . as @xmath108 ,
the left hand side of the equation increases strictly monotonically with @xmath109 .
but the right hand side is constant : it has no dependence in @xmath109 !
we have a contradiction , thus @xmath110 . + step 4 : clearly , @xmath24 is of algebraic type . by theorem 3 in @xcite , if @xmath111 is a root of @xmath25 , we get an immediate extension @xmath112 .
let @xmath113 , we have @xmath114 isomorphic to @xmath115 .
thus : @xmath116 \quad \tilde{v}\big(q(a)\big ) = v'\big(q(a)\big ) = v\big(q(a_\rho)\big ) \text { ultimately}\ ] ] this shows the uniqueness of @xmath48 and concludes the proof of the theorem . | let @xmath0 be a prime . in this paper we give a proof of the following result : a valued field @xmath1 of characteristic @xmath2 is @xmath0-henselian if and only if every element of strictly positive valuation if of the form @xmath3 for some @xmath4 . |
a neutron - star low - mass x - ray binary ( ns - lmxb ) consists of a roche - lobe - filling low - mass star and a mass - accreting neutron star with a weak magnetic field of @xmath13 g. these x - ray sources generally have two spectral states , namely the soft and hard states ( e.g. mitsuda et al .
1984 ; mitsuda et al . 1989
; white & mason 1985 ; white et al . 1988 ) . in the soft state ,
the x - ray spectrum is relatively soft , and comprises two major optically - thick emission components .
the softer of them is a disk blackbody with a temperature of @xmath14 kev , and the harder one is a blackbody of temperature @xmath15 kev , often weakly comptonized as evidenced by data excess towards higher energies @xcite .
in addition , a harder emission component , possibly of non - thermal nature , may underlie the optically - thick spectrum @xcite . in the hard state
, the spectrum has a hard power - law - like shape that also consists of two emission components . a soft thermal component and a hard one with strong comptonization .
there have been a degeneracy between two modelings of the emission .
one invokes a soft disk blackbody plus a comptonized blackbody , while the other a soft blackbody plus a hard comptonized disk blackbody @xcite .
@xcite , hereafter paper 1 , showed that the former modeling is preferred by broad - band suzaku data of the ns - lmxb aquila x-1 ( hereafter aql x-1 ) : the blackbody emission from the neutron - star surface is strongly comptonized by a hot spherical corona , resulting in a hard continuum approximated by a power - law photon index of @xmath16 . [ cols="^,^,^,^,^,^,^ " , ]
[ table_6thand7th ] among the seven observations , obs@xmath17 6 and 7 covered the faintest phase , wherein the source intensity had already decreased rapidly .
the 0.810 kev intensity ( table [ table_obsproperty ] ) on these two occasions are no higher than @xmath18% of that in obs@xmath17 5 , and the spectra on these occasions are much more clearly concave at @xmath3 kev than that of obs@xmath17 5
. nevertheless , the spectra are still quantitatively similar in shape to that of obs@xmath17 5 .
therefore , we proceeded to fit these spectra with a blackbody plus its comptonization , namely , the same two alternative models as used in obs@xmath17 5 which in turn are both natural extensions to that for obs@xmath17 24 .
as shown in figure [ fig_6thand7th ] ( panels a and b ) and table [ table_6thand7th ] , the obs@xmath17 6 spectrum was successfully reproduced by either model like in section [ ss_obs5 ] . according to model 1 ,
the coronal temperature remained approximately the same as in obs@xmath17 5 , but the coronal optical depth decreased from @xmath19 to @xmath20 .
this low optical depth enables more blackbody photons to pass through the corona without any scattering , to make the zero - scattered component stand out at @xmath21 kev and make the spectrum look more concave . in model 2
, the coronal parameters turned out to be consistent within errors with those of obs@xmath17 5 .
what makes the spectrum more concave is the change of coronal covering fraction , namely , ratio between the seed photon area to the total ( seed plus direct ) blackbody area .
in fact , from obs@xmath17 5 to obs@xmath17 6 , the ratio changed from @xmath22 to @xmath23 ( though the errors are large ) , so that the direct blackbody component became stronger . as shown in figure [ fig_6thand7th ] ( panels c and d ) , the obs@xmath17 7 spectrum was also well reproduced by either model 1 or model 2 .
the more prominent concaveness of the obs@xmath17 7 spectrum , than that of obs@xmath17 6 , can be explained as a further change of the model parameters , from obs@xmath17 5 through obs@xmath17 6 , and to obs@xmath17 7 . if , e.g. , model 1 is employed , the optical depth further decreased to @xmath24 ( table [ table_6thand7th ] ) . if , instead , model 2 is employed , the spectral change from obs@xmath17 6 to obs@xmath17 7 is mainly attributed to the increase ( though with large errors ) in the coronal electron temperature , which made the comptonization continuum harder . by extrapolating the best - fit two models , like in section [ ss_obs5 ] , we calculated the luminosity in the 0.8100 kev band .
the results are shown in table [ table_6thand7th ] .
in the present study , we analyzed five wide - band suzaku spectra of aql x-1 ( the 2nd , 4th , 5th , 6th , and 7th ) among the seven observations conducted over decay phase of the 2007 september - october outburst .
if the individual 0.8100 kev luminosities ( section [ s : spectral_analysis ] ; either directly measured or extrapolated ) are normalized to the eddington luminosity of @xmath25 erg s@xmath1 for an object of @xmath26 , we obtain the eddington ratios of 0.077 for obs@xmath17 1 , @xmath27 for obs@xmath17 2 , 3 , and 4 , @xmath28 for obs@xmath17 5 , and @xmath29 for the last two observations .
thus , during these observations , the luminosity changed by @xmath3 orders of magnitude , and the spectra ( figure [ fig_allobs ] ) changed drastically meanwhile .
like in obs@xmath17 3 already reported in paper 1 , obs@xmath17 2 and obs@xmath17 4 caught the source in the typical hard state , wherein the spectra were reproduced by a strongly comptonized blackbody plus a weaker disk blackbody .
the innermost disk radii , @xmath30 km ( obs@xmath17 2 ) and @xmath31 km ( obs@xmath17 4 ) , and the blackbody radii , @xmath32 km ( obs@xmath17 2 ) and @xmath33 km ( obs@xmath17 4 ) , are all consistent with those of obs@xmath17 3 .
thus , at these luminosities , the standard disk is truncated at a radius of @xmath34 km , and the accreting matter then transforms into an optically - thin , geometrically - thick coronal flow . as discussed in paper 1 , this flow is considered to accrete almost spherically onto the neutron star and make its entire surface emit the blackbody photons with a temperature of @xmath35 kev and a radius of @xmath36 km ( table [ table_2ndand4th ] ) .
these photons are then compton scattered by the succeeding coronal flows . the spectra of obs@xmath17 5@xmath377 are reproduced by a blackbody and its comptonization , without the disk component which is probably below the xis energy band . in these three faintest data sets , it is ambiguous , however , whether the blackbody source is covered entirely by a relatively thin ( @xmath38 ) and very hot ( @xmath39 kev ) corona ( model 1 ) , or partially by a thicker ( @xmath40 ) and cooler ( @xmath41 kev ) one ( model 2 ) .
the characteristic hump below @xmath3 kev seen in these spectra can be explained either by those blackbody photons which escaped through the thin corona without scattering ( model 1 ) , or emission from those regions on the neutron star which are not covered by the thick corona ( model 2 ) . given the model ambiguity in the dimmest three data sets , we may ask ourselves whether the obs@xmath17 2@xmath374 data , which have so far been analyzed with model 1 ( plus diskbb ) , can also be described with model 2 ( plus diskbb ) . to see this
, we fitted the obs@xmath17 2@xmath374 spectra with a model diskbb + compps + bbody + gauss , wherein the additional blackbody component , representing surface emission without coronal obscuration , was constrained to have the same temperature as the compton seed photons .
then , the directly visible blackbody was found to little contribute , with the radius of @xmath42 km , @xmath43 km , and @xmath44 km from obs@xmath17 2 , obs@xmath17 3 and obs@xmath17 4 , respectively . in none of these cases ,
the fit improvement was significant .
therefore , these data sets prefer model 1 , in agreement with paper 1 in which the corona was concluded to be approximately spherical .
an important source property revealed by analyzing the seven data sets is luminosity - dependent changes in @xmath45 , which hold regardless of the model 1/model 2 ambiguity .
this is evident in figure [ fig : rbbfig ] , where @xmath45 is plotted against the eddington - normalized 0.8100 kev luminosity . as discussed in paper 1 , the small value of @xmath45 in obs@xmath17 1 is considered to mean that the accretion in the soft state occurs only onto an equatorial region of the neutron star . as @xmath46 decreased to @xmath47 erg s@xmath1 ( obs@xmath17 2@xmath374 ) and the source made a transition into the hard state , the emission radius increased to @xmath36 km , which is consistent with the neutron - star radius @xmath48 .
this is thought to be a consequence of the nearly spherical coronal flow ( paper 1 ) .
as the source became still dimmer , @xmath45 started decreasing again , down to @xmath3 km at a luminosity of @xmath2 erg s@xmath1 .
as shown in figure [ fig : rbbfig ] by a solid line , the dependence of @xmath45 on @xmath46 ( in obs@xmath17 2 to obs@xmath17 7 ) is approximated as @xmath49 .
( 88mm,120mm)figure6.eps the decrease of @xmath45 towards lower luminosity is presumably due to some collimation of the optically - thin accretion flow .
this , in turn , is likely to be caused by the emergence of a neutron - star magnetosphere which is usually negligible when the source is luminous .
to evaluate this idea , let us recall that the alfven radius @xmath50 ( for spherical accretion ) depends on the mass accretion rate @xmath51 as @xmath52 ( elsner & lamb 1977 ) , where @xmath53 is the neutron - star magnetic field ; here we assumed the neutron - star mass and radius of @xmath54 and @xmath55 km , respectively , and the magnetic field of dipolar configuration .
we may consider that @xmath45 starts decreasing in figure [ fig : rbbfig ] at an eddington ratio of @xmath56 , or @xmath57 erg s@xmath1 , because the magnetic field starts to affect the accretion flow , or equivalently , @xmath58 becomes comparable to or larger than @xmath48 .
this , together with equation [ eq : alfvenradius ] , yields @xmath59 g. since the rotation frequency of aql x-1 is considered as @xmath60550 hz ( zhang et al@xmath17 1998 ; casella et al@xmath17 2008 ) , the corotation radius is calculated as @xmath61 employing the field strength obtained above , @xmath58 is then inferred to exceed @xmath62 at @xmath63 erg s@xmath1 . at such low luminosities ( like in obs@xmath17 6 and obs@xmath17 7 )
, the accreting matter will be strongly affected by the magnetic pressure , and will start extracting the angular momentum from the neutron star before it reaches @xmath64 .
this effect , often called propeller effect " @xcite , can explain the large luminosity drop from obs@xmath17 5 to obs@xmath17 6 and obs@xmath17 7 .
although @xmath65 was found to reduce as @xmath51 decreases , the effect could be some artifact ; e.g. due to radiative transfer in the ns atmosphere ( e.g. heinke et al@xmath17 2006 ) . as an independent examination ,
let us consider here the luminosity dependence of the coronal optical depth @xmath66 .
after equation ( 6 ) of paper i , the mass continuity condition along the radius @xmath67 can be written as @xmath68 where @xmath69 is the mean molecular weight of ions , @xmath70 is the proton mass , @xmath71 is the flow density , @xmath72 is the flow cross section ( @xmath73 if spherical ) , @xmath74 is the free fall velocity , and @xmath75 denotes possible deviation of the actual inflow velocity from @xmath76 ( paper 1 ) . assuming that @xmath51 is constant , that @xmath77 approximately scales as @xmath78 ( nearly radial flow ) , and that @xmath79 depends only weakly on @xmath67 , equation ( [ eq : mass_continuity ] ) yields @xmath80 where subscript 0 specifies values on the neutron - star surface , @xmath81 .
like equation ( 7 ) in paper i , we may integrate this from @xmath82 to an outer coronal radius @xmath83 , to express the optical depth as @xmath84 further assuming @xmath85 and a negligible dependence of @xmath75 on @xmath51 , we finally obtain @xmath86 where @xmath51 was replaced with the 0.8100 kev luminosity @xmath46 . to see the behavior of this @xmath87 , the values of @xmath88 for model 1 were calculated from table 2 , table 3 , and table 4 , and are shown in figure [ fig : rbb - l_tau ] against @xmath65 .
the four data points indicate a very tight relation as @xmath89 .
combined with equation ( [ eq : flow_cross_sec ] ) , this implies @xmath90 , which is close to the @xmath91 scaling expected if @xmath65 is correctly estimated . further incorporating the scaling of figure [ fig : rbbfig ] , we find @xmath92 , which reconfirms that the flow becomes indeed constrained onto a limited surface area as @xmath46 decreases . in other words
, @xmath66 decreased only by less than an order of magnitude as the luminosity became lower by more than 2 orders of magnitude ( from obs@xmath17 2@xmath374 to obs@xmath17 5@xmath377 ) , because the flow became more strongly funneled onto limited regions ( most likely the magnetic poles ) .
( 75mm,120mm)figure7.eps if , instead , applying the same method to model 2 ( in obs@xmath17 5@xmath377 ) by substituting @xmath93 for the saturated @xmath66 , the dependence becomes @xmath94 , of which the power index is too much larger than the normal value 2 ( @xmath95 ) .
through these arguments , we consider that model 1 is more preferable than model 2 , in agreement with the inference made at the end of subsection 4.1 .
the configuration of model 1 is still possible even when the coronal flow is funneled onto the magnetic poles , because the flow would be rather spherical at @xmath96 .
this means that model 1 ( sometimes with an addition of diskbb ) can describe the broad - band spectrum of aql x-1 in the hard state from obs@xmath17 2 through obs@xmath17 7 , over @xmath15 orders of magnitude in luminosity .
however , the reality could be an intermediate condition between model 1 and model 2 . summarizing the results obtained so far
, the @xmath51-dependent changes in the accretion geometry of an ns - lmxb may be described in the following manner , in terms of the four radii , @xmath50 , @xmath62 , @xmath97 , and @xmath82 ( matsuoka & asai 2012 ) . 1 .
when @xmath51 is high and hence @xmath98 , the gravity is dominant in the accretion ( corresponding to obs@xmath17 1 ) .
the optically - thick disk continues down to @xmath99 .
as @xmath51 decreases , the system makes a transition into the hard state . since @xmath100 and @xmath101 , the gravity is still dominant , but the disk becomes truncated at @xmath102 , and the matter accretes nearly spherically onto the neutron star ( obs@xmath17 2@xmath374 ) .
3 . when @xmath51 decreases to make @xmath103 , the effect of magnetic field begins to stand out .
the flow becomes anisotropic , and funneled onto the magnetic poles ( obs@xmath17 5 ) .
as @xmath51 further decreases , the magnetic field becomes more dominant , resulting in a condition of @xmath104 .
the propeller effect ( lamb et al@xmath17 1973 ) hampers the accretion , thus causing a rapid luminosity decrease .
however , a small fraction can still fall along the magnetic field lines onto a small area at the magnetic poles ( obs@xmath17 6 and obs@xmath17 7 ) .
through an analysis of the seven suzaku data sets of aquila x-1 tracing an outburst decay , we obtained spectra at various luminosities ( from @xmath0 erg s@xmath1 to @xmath2 erg s@xmath1 ) , and carried out model fitting to each of them .
consequently we obtained the results as below . 1 .
like obs@xmath17 3 which we reported in paper 1 , obs@xmath17 2 and obs@xmath17 4 had luminosities of @xmath105 erg s@xmath1 , and showed typical characteristics of the hard state .
the spectra are composed of a comptonized blackbody emission from the whole neutron - star surface and a softer emission from a truncated disk .
the spectra of obs@xmath17 5 , obs@xmath17 6 , and obs@xmath17 7 were reproduced either by a comptonized blackbody ( model 1 ) or that with directly - seen blackbody ( model 2 ) .
the obtained radius of the blackbody emission was significantly smaller than 10 km and decreased along with the luminosity , regardless of the model ambiguity .
3 . we clarified luminosity dependence of @xmath45 , which is approximated as @xmath49 .
this can be considered as an evidence the emergence of a neutron - star magnetosphere , which confines the accretion flow onto the magnetic poles .
the decrease of @xmath45 toward the lower luminosity was reconfirmed independently based on the coronal optical depth .
5 . as to obs@xmath17 5 , obs@xmath17 6 and obs@xmath17 7 , model 1
is considered to be more appropriate than model 2 , although the reality could be in between . | the neutron - star low - mass x - ray binary aquila x-1 was observed by suzaku for seven times , from 2007 september 28 to october 30 .
the observations successfully traced an outburst decay in which the source luminosity decreased almost monotonically from @xmath0 erg s@xmath1 to @xmath2 erg s@xmath1 , by @xmath3 orders of magnitude .
to investigate luminosity - dependent changes in the accretion geometry , five of the seven data sets with a typical exposure of @xmath4 ks each were analyzed ; the other two were utilized in a previous work @xcite .
the source was detected up to 100 kev in the 2nd to the 4th observations , to 40 kev in the 5th , and to 10 kev on the last two occasions .
all spectra were reproduced successfully by comptonized blackbody model with relatively high ( @xmath5 ) optical depths , plus an additional softer optically - thick component .
the faintest three spectra were reproduced alternatively by a single comptonized blackbody model with a relatively low ( @xmath6 ) optical depth .
the estimated radius of the blackbody emission , including seed photons for the comptonization , was @xmath7 km at a 0.8100 kev luminosity of @xmath8 erg s@xmath1 ( the 2nd to the 4th observations ) .
in contrast , it decreased to @xmath9 km and further to @xmath10 km , at a luminosity of @xmath11 erg s@xmath1 ( the 5th observation ) and @xmath12 erg s@xmath1 ( the 6th and 7th ) , respectively , regardless of the above model ambiguity .
this can be taken as evidence for the emergence of a weak magnetosphere of the neutron star . |
we consider a short two - dimensional quantum wire incorporating two parallel quantum dots which is placed in a photon cavity . we will call this device the central system .
the short quantum wire , and indirectly the entire central system , is weakly coupled to two external leads , the left ( l ) and the right ( r ) leads , acting as electron reservoirs .
the coupling opens up the central system to electron transport through it at time @xmath0 .
below we first establish details of the central system of strongly coupled electrons and photons and review its properties .
then we describe the master equation formalism used to model the time - dependent electron transport through the open central system .
we consider a short two - dimensional quantum wire with two parallel quantum dots placed in a photon cavity .
the potential energy landscape can be described by the formula @xmath1\nonumber\\
\times&\theta\left(\frac{l_x}{2}-x\right)\theta\left(\frac{l_x}{2}+x\right ) , \label{potential}\end{aligned}\ ] ] with @xmath2 mev , @xmath3 mev , @xmath4 nm@xmath5 , @xmath6 nm , @xmath7 nm , @xmath8 nm , and @xmath9 is the heaviside step function .
the potential is shown in fig . [ fig00 ] . the first line in eq .
( [ potential ] ) describes the parabolic confinement of the short quantum wire in the @xmath10-direction , perpendicular to the transport direction .
mev , and the small perpendicular external magnetic field @xmath11 t define a characteristic length @xmath12 nm.,scaledwidth=45.0% ] the central system can contain several photons ( 0 - 16 ) and few electrons ( 0 - 4 ) and is best described by a many - body formalism .
the interactions of its constituents are accounted for via a configuration interaction ( ci ) approach ( also known as exact numerical diagonalization).@xcite we assume gaas parameters , @xmath13 , @xmath14 , and @xmath15 .
the hamilton operator for the central system in terms of the field operators is @xmath16 with @xmath17 where @xmath18 is a classical vector potential generating an external homogeneous small magnetic field @xmath11 t along the @xmath19-axis perpendicular to the plane of the two - dimensional semiconductor inserted in order to break spin and orbital degeneracies to enhance accuracy of the results .
@xmath20 is the hamiltonian for the single cavity photon mode with energy @xmath21 mev , and the electron - electron static coulomb repulsion is represented by @xmath22 written in terms of four field operators and a spatially dependent coulomb kernel.@xcite the quantized vector potential of the cavity photon field is @xmath23 .
the last two terms of the hamiltonian ( [ hclosed ] ) stand for the para- and the diamagnetic electron - photon interactions , respectively , necessary since we will consider photon energy that might , or might not be , close to a transition resonance between particular electron states.@xcite the charge and the charge - current density operators are @xmath24 the external magnetic field , @xmath11 t , and the parabolic confinement energy of the leads and the central system @xmath25 mev , lead together with the cyclotron frequency @xmath26 to an effective characteristic frequency @xmath27 and an effective magnetic length @xmath28 .
we consider a rectangular photon cavity @xmath29 \times [ -a_\mathrm{c}/2,a_\mathrm{c}/2]\times [ -d_\mathrm{c}/2,d_\mathrm{c}/2]\}$ ] with the short two - dimensional quantum wire centered in the @xmath30 plane .
using the coulomb gauge the polarization of the electric field is parallel to the transport in the @xmath31-direction by selecting the te@xmath32 mode , or perpendicular by selecting the te@xmath33 mode . in terms of the creation and annihilation operators , @xmath34 and @xmath35 , for cavity photons
the quantized vector potential is @xmath36 for the te@xmath32 and te@xmath33 modes , respectively .
the strength of the vector potential , @xmath37 , determines the coupling constant @xmath38 , here set to 0.05 mev , leaving a dimensionless polarization tensor @xmath39 where @xmath40 and @xmath41 are single - electron states of the short two - dimensional quantum wire .
latin indices are used for the single - electron states , and greek for the many body states to be described below . in order to find the energy spectrum and the states of the closed central system we use a stepwise scheme of exact numerical diagonalizations and truncations:@xcite first
, the single - electron states of the system are used to build a fock - space for 0 - 4 noninteracting electrons @xmath42 .
this space is truncated well above the bias window .
second , the hamiltonian of the coulomb interacting electrons is diagonalized in the fock basis creating a new fock - space of coulomb interacting electrons @xmath43 .
third , the fock - space of the coulomb - interacting electrons is used to build a many - body basis for coulomb interacting electrons and cavity photons by a tensor product of @xmath43 and the eigenstates @xmath44 of the photon number operator .
this new basis is used to diagonalize the full hamiltonian of the central system ( [ hclosed ] ) obtaining a new fock - space of interacting electrons and photons @xmath45 , with pure photon states and cavity - photon dressed electron states .
this construction is reminiscent and parallel to the possible construction of a green function for the system ( not done here ) , where one starts with a noninteracting green function , and in a stepwise fashion dresses it with the coulomb and the photon interactions through the dyson equation .
the construction here is entirely carried out in functional spaces in a grid - free manner .
the many - body energy spectra versus the plunger gate voltage @xmath46 for the central system are presented in fig .
[ fig - erof - x ] for @xmath31-polarization of the cavity photon field , -polarized cavity photons versus plunger gate voltage @xmath46 .
electron number ( left panel ) , and the mean photon number ( right panel ) is color coded into the states .
@xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] and in fig .
[ fig - erof - y ] for @xmath10-polarization .
-polarized cavity photons versus plunger gate voltage @xmath46 .
electron number ( left panel ) , and the mean photon number ( right panel ) is color coded into the states .
@xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] in both cases we indicate with a color coding the electron content ( an integer in the closed system ) in the left panel of the figure , and the photon number in the right panel . without the electron - photon interaction the photon number is an integer , but the interaction does not conserve the number of photons . we will come back to this fact below . the dressed electron states do thus not contain an exact photon number and we can not refer to one- or two - photon replicas of an electron state , but as the replica concept is useful we refer to the first or second replica of a certain electron state . for our parameters here we use 16 photon states . in our experience the convergence with respect to the photon basis is not critical , but instead the number of electron states is , as photon dressing in the strong coupling regime leads to increased polarization or spreading of the charge of a state.@xcite the central system is weakly coupled to external leads and opened up to electron transport through it at time @xmath0 .
the semi - infinite leads are parabolically confined and are in the same perpendicular classic weak magnetic field as the central system . in our calculations
we include 4 of their subbands .
the weak coupling is described by the hamiltonian@xcite @xmath49 where @xmath50 is the annihilation operator for an electron in the single - electron state of the the central system labeled with @xmath51 , and @xmath52 is the creation operator for an electron in lead @xmath53 with momentum and subband index labeled by the composite index @xmath54 .
the coupling tensor @xmath55 describes the coupling between these single - electron states of lead @xmath53 and the central system , and depends on the geometrical form of the corresponding wave functions in the contact area extending approximately @xmath56 into each subsystem.@xcite the remaning overall coupling constant to the leads is @xmath57 mev .
the switching of the coupling is determined by the heaviside unit step function of time , at @xmath0 . before the coupling there exist no correlations between the leads , or the leads and the central system . the time evolution after the coupling of the leads and
the central system can be described by the liouville - von neumann equation for the density operator of the system @xmath58 were , @xmath59 is the liouville operator , defined by the commutator @xmath60,\ ] ] with @xmath61 the density operator of the total system , describing the dynamic state of both the leads and the central system . as the energy spectra of the leads ( the electron reservoirs )
are dense , we have in earlier publications resorted to applying a formalism of nakajima@xcite and zwanzig@xcite in which the dynamics of the whole system is projected on the central system leading to a generalized master equation ( gme)@xcite @xmath62 -\frac{1}{\hbar}\int_0^t dt ' k[t , t - t';\rho_\mathrm{s}(t ' ) ] \label{gme}\ ] ] for the reduced density operator @xmath63 describing properties of the central system under influence of the external leads , and defined by tracing out variables of the leads @xmath64 .
the dissipative integral kernel , @xmath65 , is constructed using terms of the coupling hamiltonian ( [ h_t ] ) up to second order , but the integro - differential form of the gme guarantees higher order terms in the solution of the elemental type already present in @xmath65 .
the solution to eq .
( [ gme ] ) can be found by numerical integration and iterations.@xcite however , for the @xmath66 states needed here for the transport calculations in the photon dressed basis @xmath67 , it is not feasible to integrate the gme in time much farther than 1000 ps . to go beyond
that we have made a markovian approximation to the gme ( [ gme ] ) avoiding any further approximations and using the kronecker tensor product and vectorization of matrices to map the equation from the fock space of states to liouville space of transitions.@xcite in the @xmath68-dimensional liouville space@xcite the linear equation of motion is@xcite @xmath69 with @xmath70 a general non - hermitian operator with complex eigenvalues .
the exact solution of eq .
( [ eq : finalpde ] ) is@xcite @xmath71\mathfrak{b}\right\}\operatorname{vec}(\rho_\mathrm{s}(0)),\ ] ] in terms of the left , @xmath72 , and right , @xmath73 , eigenvectors of @xmath70 , @xmath74 as the terms in @xmath70 can be cleanly traced back to the left or the right lead with no mixed terms we can , as before,@xcite calculate the current from the left lead into the central system , @xmath75 , and the current from the central system into the right lead , @xmath76 , using the time - derivative of @xmath77 in the equation of motion ( [ eq : finalpde ] ) and its solution ( [ eq : solution ] ) .
we explore the time dependent electron transport into and through the system for three different values of the plunger gate voltage , @xmath46 .
we start with @xmath78 mv , when only the two spin components of the one - electron ground state are in the bias window , set by the chemical potentials of the left and right leads , @xmath79 mev and @xmath80 mev , respectively .
we expect resonant transport through the system . in figure
[ fig_neg - vg20 ] we show results for the mean electron and photon number in the central system for two initial states , totally empty system ( 0 g ) , and only one photon in the central system ( 0g@xmath81 ) . ) .
@xmath31-polarized photons , @xmath78 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] in both cases the mean electron number rises quickly and remains constant just over 0.7 . in the latter case ( 0g@xmath81 )
the mean photon number vanishes on a longer time scale , and for the former case the mean photon number never gains a value visible in fig .
[ fig_neg - vg20 ] .
not surprisingly the current for the two cases , shown , in fig .
[ fig_curr - vg20 ] , is approximately the same . ) .
@xmath31-polarized photons , @xmath78 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] for a short time the current is towards the central system from both leads ( the right current , @xmath76 , is negative in the beginning ) .
after 10 ns the system seems to be in a steady state with the current flowing through the central system .
this view is not in agreement with the occupation of the many - body states of the central system shown in the upper panel of fig .
[ fig_occ - vg20 ] . for initially the vacuum state ( 0 g )
( upper panel ) , and one photon in the system ( 0g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath78 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] for initially the vacuum state ( 0 g ) ( upper panel ) , and one photon in the system ( 0g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath78 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] there , we see for the former case ( 0 g ) that indeed the system starting in the vacuum state @xmath82 reaches quickly a steady state formed by a mixture of @xmath82 and the partial population of the two spin components of the one - electron ground state , @xmath83 and @xmath84 . the lower panel of fig .
[ fig_occ - vg20 ] shows that even when we see no changes in the current there are changes in the occupation of states .
we see that due to the presence of one photon initially in the system charge is promoted temporarily to the intermediate states @xmath85 and @xmath86 , that are one - electron states with approximately one photon .
they are photon replicas of the two spin components of the one - electron ground state , and as such have energy 1.98 mev corresponding to one photon energy above the one - electron ground state in the bias window .
we see thus , as noted earlier , photon assisted tunneling into the system.@xcite but , photons are not supplied to the system and the initial photon vanishes and the population transfers smoothly to the states @xmath83 , @xmath84 , and @xmath82 as before with no change seen in the current through the central system . in the far - infrared ( fir ) regime the electromagnetic transitions in the system are slow so the steady state is only reached after 10 @xmath87s .
in addition , it is important to have in mind that electromagnetic transitions are only active due to the coupling to the leads , which is weak here . without this coupling
the many - body states of the central system are its eigenstates .
we now turn to more complex situations with the plunger gate voltage set at @xmath88 and @xmath89 mv shown in fig .
[ fig_erofneg ] .
-axis ) , the mean electron , the mean photon content , and the spin of the 64 lowest many - body states for @xmath88 mv ( upper panel ) , and @xmath90 mv ( lower panel ) .
@xmath31-polarized photons .
mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] -axis ) , the mean electron , the mean photon content , and the spin of the 64 lowest many - body states for @xmath88 mv ( upper panel ) , and @xmath90 mv ( lower panel ) .
@xmath31-polarized photons .
@xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] for the case of @xmath88 mv , the upper panel of fig .
[ fig_erofneg ] , we see a photon replica of the singlet two - electron ground state together with the three spin components of the lowest two - electron triplet state . just below the bias window are two one - electron states with vanishing photon components . in the lower panel of fig .
[ fig_erofneg ] we have just changed the plunger gate voltage to @xmath89 mv in order to raise the one - electron states into the bias window still keeping the four two - electron states there .
this we do to shed light on the roles assumed by the one- and two - electrons states in our transport picture , built on sequential tunneling . here , we have more obvious choices for the initial state of the system . in fig .
[ fig_neg - vg01 - 0gx ] we see for the case of an initially empty system ( 0 g ) , or with only one photon initially ( 0g@xmath81 ) , that the photons seem not to play any major role . ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] the system is charged quickly to one electron , and further delayed charging seems to go in hand with the vanishing of the initial photon in the system , or if there was none in the system initially , with the appearance of a very small photon component , that again vanishes .
as there are several states available below the bias window we might expect charge to get trapped in the system suppressing possible current through it in the steady state . before looking at the corresponding current and occupation we present in fig .
[ fig_neg - vg01 - 2gx ] the evolution of the mean electron and photon numbers for the two electrons initially in the central system , in the two - electron ground state ( 2 g ) , and the first photon replica ( 2g@xmath81 ) thereof . ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] here , all changes are slow , occurring for time between 100 ns and 10 @xmath87s .
for the first time we see here that the discharging of the system depends on the initial photon number .
it is faster for the first photon replica of the two - electron ground state than for the corresponding ground state .
we should have in mind that the @xmath31-polarized photon field in the system stretches the charge density into the contact area of the central system facilitating the charge to leave or enter the system.@xcite in fig .
[ occ - vg01 - 0gx ] we present the time dependent occupation of states for the case of no electron initially in the system , but either no ( 0 g ) ( upper panel ) , or one photon ( 0g@xmath81 ) ( lower panel ) , corresponding to fig .
[ fig_neg - vg01 - 0gx ] . for initially the vacuum state ( 0 g )
( upper panel ) , and one photon in the system ( 0g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] for initially the vacuum state ( 0 g ) ( upper panel ) , and one photon in the system ( 0g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] for the ( 0 g ) case in the upper panel we identify the one - electron states @xmath91 and @xmath92 with energy @xmath93 mev just below the bias window that are active in charging the system .
the steady state is a combination of the two spin components of the one - electron ground state , and with a higher probability the two - electron ground state .
the intermediate states active around 3 @xmath87s are the two - electron triplet states @xmath94 , @xmath95 , and @xmath96 together with the one - electron state @xmath97 .
all states with odd spatial parity ( measured along the @xmath10-axis ) , that is slightly broken by the external magnetic field .
as we use a state dependent coupling to the leads that depends on the spatial properties of the wave functions , the coupling of these states to the lowest subband in the leads with even parity is small .
for the ( 0g@xmath81 ) case displayed in the lower panel of fig .
[ fig_neg - vg01 - 0gx ] we see the initial one - photon state @xmath98 rapidly depopulated with the one - electron states @xmath99 and @xmath100 containing one photon and having energy 1.88 mev taking over .
these states are the first photon replicas of the one - electron states we noticed just below the bias window .
again , we are observing a photon assisted charging of the system , and before the system reaches the same steady state as in the ( 0 g ) case we see intermediate states of two types ; low energy one - electron states with no photon content like @xmath84 and @xmath97 , and one - electron states with approximately one photon @xmath101 and @xmath102 , and last we see small occupation of the triplet states in the bias window .
so , differently from the ( 0 g ) case we see radiative transitions on the way to the steady state . not surprisingly , we see in fig . [ current - vg01 - 0gx ] for @xmath103 ps high charging current into the system .
this is followed by a considerable current through the system up to 1 @xmath87s during the period it takes the system to relax on the way to the steady state .
but , the final and real steady state is not fully reached until after approximately 20 @xmath87s . ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] the logarithmic time scale in the figures can be a bit deceiving when estimating the charge flowing into and through the system . as can be confirmed by fig .
[ fig_neg - vg01 - 0gx ] approximately one electron enters the central system in the charging phase lasting to 200 - 500 ps , but in the intermediate time regime extending to 1 @xmath87s the number of electrons passing through is of the order of 10@xmath104 , although on the average only 0.6 are staying in ! on the current scale used in fig . [ current - vg01 - 0gx
] one is tempted to conclude that there is no steady state current through the system . in order to explore this situation
we now analyze in more detail what happens when initially there are two electrons in the system .
we started with this in fig.[fig_neg - vg01 - 2gx ] , but now turn to the occupation ( population ) for the ( 2 g ) and ( 2g@xmath81 ) cases shown in fig .
[ occ - vg01 - 2gx ] . for initially the two - electron ground state ( 2 g ) ( upper panel ) , and the first photon replica of the two - electron ground state of the system ( 2g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] for initially the two - electron ground state ( 2 g ) ( upper panel ) , and the first photon replica of the two - electron ground state of the system ( 2g@xmath81 ) ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] the ( 2 g ) case with initially 2 electrons in the ground state is a simple case with no electromagnetic transitions active . very slowly the two - electron ground state @xmath85 ( a singlet ) looses charge to the two spin components of the one - electron ground state , @xmath82 and @xmath105 .
this transition is slow as the single - electron wave functions associated with these states have a small weight in the contact area of the short wire in the central system , and all transitions are activated by the coupling to the leads , that depends on the wave functions in that region . the relaxation of the charge from the first photon replica of the two - electron ground state @xmath106 ( the 2g@xmath81 case seen in the lower panel of fig .
[ occ - vg01 - 2gx ] ) in the bias window is faster with of course the same steady state reached in the end , but now the active intermediate states are the one - electron states @xmath101 and @xmath102 , a radiative transition terminating in the two spin components of the one - electron ground state @xmath82 and @xmath105 .
now , we are prepared for viewing the current into and through the system for both the cases with initially 2 electrons in the system ( 2 g and 2g@xmath81 ) displayed in fig .
[ current - vg01 - 2gx ] . for the long transient regime
up to 1 @xmath87s we see the system is slowly loosing charge as @xmath75 is negative , indicating a current from the central system into the left lead , and @xmath76 is positive indicating current from the central system into the right lead . ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev .
, scaledwidth=48.0% ] but , we notice a _ nonzero _
steady state current through the system , albeit small , that we have to analyze further .
before , we have to ask the question whether even in the case of an initially empty system ( see fig .
[ current - vg01 - 0gx ] ) we also had this finite but small steady state current without noticing it .
indeed , fig .
[ current - compare - x ] confirms that there is the same steady state current through the system for the different initial cases , independent of whether we start with none , or two electrons in the system .
mv ( upper panel ) , and @xmath90 mv ( lower panel ) , @xmath31-polarized photons , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] mv ( upper panel ) , and @xmath90 mv ( lower panel ) , @xmath31-polarized photons , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] even , when we displace slightly the plunger gate voltage to @xmath90 mv and thus lift the two one - electron states @xmath91 and @xmath92 into the bias window we continue to have a slight steady state current through the system .
where does it come from ?
these two one - electron states are not populated in the steady state .
the answer can be found by analyzing the partial current through each state in the system , a feature of the model that can not be repeated in experiments , but can give a valuable insight into the underling active processes in the system . in fig .
[ irlmu - x ] we display in the upper panel the contribution of the active one - electron states in the system , and in the lower panel the active two - electron states . from the left lead ( l ) , and into the right lead ( r ) for the most active one - electron states ( upper panel ) , and the most active two - electron states ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] from the left lead ( l ) , and into the right lead ( r ) for the most active one - electron states ( upper panel ) , and the most active two - electron states ( lower panel ) .
@xmath31-polarized photons , @xmath88 mv , @xmath11 t , @xmath47 mev , and @xmath48 mev.,title="fig:",scaledwidth=48.0% ] in addition , we have selected states in the bias window that could be expected to carry current . in the upper panel
we see that the one - electron states contribute to the charge loss from the system as we noticed earlier .
the steady state current of these one - electron state seems to vanish .
we have confirmed that this is indeed true . in the lower panel of fig .
[ irlmu - x ] we notice that the steady state current is only carried by the two - electron ground state @xmath85 , even though it is not in the bias window .
how is this possible ? as we saw earlier ( see fig .
[ occ - vg01 - 2gx ] ) the two - electron ground state is the state with the highest energy contributing significantly to the steady state . in the steady state
the charge has all relaxed to states below the bias window .
the two - electron ground state ( a spin singlet ) with even parity ( with respect to the @xmath10-direction ) has higher coupling to the lowest subband of the leads than the spin triplet states in the bias window .
we are thus seeing an off - resonance inelastic ( with respect to the system lead tunneling ) contribution to the steady state current by the two - electron ground state .
the current is low , as it is inelastic and the coupling to the two electron state mostly confined to the parallel quantum dots is low on grounds of the geometry . even though we can talk about a strong electron - photon coupling here
it should be stated that for the @xmath31-polarized cavity field we are well off - resonance with the photon number close to an integer for most states . for the @xmath10-polarization
there is a weak rabi - splitting for the low lying one - electron states , but not for the two - electron states , and here we used two - electron states as the initial state or started with no electrons in the central system .
the fact that electron - photon system is only in a weak resonance here can be confirmed by inspecting figures [ fig - erof - x ] , [ fig - erof - y ] , and [ fig_erofneg ] .
all the same , it is necessary to have in mind that even for the @xmath31-polarization here , we have dressed electron states with a mixed photon content .
this is shown in fig .
[ figres ] for the one- and two - electron ground states , ( @xmath82 and @xmath85 ) , and the first replica of the two - electron ground state ( @xmath106 ) .
( 1 g ) ( a ) , the two - electron ground state @xmath107 ( 2 g ) ( b ) , and the first photon replica of the two - electron ground state @xmath106 ( 2g@xmath81 ) ( c ) , in terms of a tensor product of coulomb interacting electron states @xmath108 and eigenstates @xmath109 of the photon number operator . the photon sectors are marked by their photon number and indicated with vertical thin green lines . the unit probability is indicated with a thin yellow horizontal line .
@xmath47 mev , and @xmath48 mev.,scaledwidth=48.0% ] in case of a stronger resonance between the electrons and the cavity photons we have observed in an earlier publication stronger influence of radiative transitions on the evolution of the system and stronger variations in the mean photon number.@xcite
we have used our model of electron transport through a central system with interacting electrons and photons to describe the time evolution of the system after it has been started / prepared in a certain eigenstate of the closed system .
time resolved experiments in this field have not yet been performed , but transport experiments on electron systems in photon pumped cavities are gaining strength.@xcite the importance of time dependent modeling of the systems is twofold : first , they allow us to gain insight into the underlying processes in the system .
second , with the knowledge of these active fundamental processes and their time evolution we can research the states of the system we would like to maintain with an external photon pumping and their relations , even in a system with a complex structure .
accuracy in numerical calculations is essential and error can cumulate in different parts of a model evaluation .
we have taken care designing all functional bases and necessary truncations thereof balancing ram memory requirements and exactness of the results for the closed system within reasonable bounds , not shying away from heavy ram usage .
this is only feasible with heavy parallelization of the code .
as we use no time - integration of the equation of motion for the long times we are interested in , we escape errors seeping in through finite time steps in an integration .
instead , we have to pay attention to the quality of the complex valued eigenvalues of the nonsymmetric liouvillian @xmath110 in eq .
[ eq : solution ] .
this is a nontrivial task and has been studied within the qutip python - framework.@xcite in our approach the accuracy of the eigenvalues and eigenvectors measured by the trace of the reduced density operator is such that the trace is always equal to 1 with at least 8 accurate figures for all the cases described here , evaluated with double precision in fortran .
we only notice a deviation form this level of accuracy if we attempt a calculation with no states in the bias window defined by the chemical potentials of the left and right external leads .
only 84 points , exponentially distributed on the time - axis , are used in the figures to conserve computational time
. a careful study of the details of the relaxation of the spin components forming a braided pattern , for example in figure [ fig_occ - vg20 ] in the range @xmath111 ps , would require more points .
as expected , resonant tunneling through the lowest states of the central system quickly bring it into a steady state .
the situation becomes more complex if the bias window defined by the external leads is located higher in the many - body energy spectrum .
then the character of the states in the window and below it , shaped by their geometry and the coupling between the constituents of the system , are of importance . in these cases
we find small but important off - resonance current through the system .
we observe that even when the system , judged from the transport current through it , seems to have reached a steady state there still can be slow internal processes at work bringing it into the real ( final ) steady state .
( see , for example , figures [ fig_curr - vg20 ] and [ fig_occ - vg20 ] ) from our modeling efforts it is clear that the character of the transport into and through the system is highly tunable with , or strongly depends on , the plunger gate voltage and the photon frequency of the cavity .
this fact underlines the importance of a computationally effective approach to explore the dynamic properties of the system,@xcite and the possibility to enhance further the description of the details of the underlying physical processes beyond the presently implemented sequential tunneling to high order .
this work was financially supported by the research fund of the university of iceland , the icelandic research fund , grant no .
163082 - 051 , and the icelandic instruments fund .
we also acknowledge support from the computational facilities of the nordic high performance computing ( nhpc ) , and the nordic network nanocontrol , project no .
: p-13053 .
hsg acknowledges support from most , taiwan , under grant no.103 - 2112-m-002 - 003-my3 .
31ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevlett.113.036801 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.108.046807 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.066802 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.126804 [ * * , ( ) ] @noop ( ) link:\doibase 10.1103/physrevlett.107.256804 [ * * , ( ) ] link:\doibase 10.1103/physrevx.6.021014 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.73.3443 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.155303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.195427 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.085416 [ * * , ( ) ] link:\doibase 10.1021/acsphotonics.5b00115 [ * * , ( ) ] , @noop ( ) , @noop ( ) , link:\doibase 10.1103/physrevb.94.085150 [ * * , ( ) ] link:\doibase 10.1103/physreve.86.046701 [ * * , ( ) ] link:\doibase 10.1002/prop.201200053 [ * * , ( ) ] http://stacks.iop.org/1367-2630/14/i=1/a=013036 [ * * , ( ) ] link:\doibase 10.1002/andp.201500298 [ * * , ( ) ] http://stacks.iop.org/1367-2630/11/073019 [ * * , ( ) ] http://stacks.iop.org/1367-2630/11/113007 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) http://link.springer.com/article/10.1007%2fbf01395429 [ * * , ( ) ] link:\doibase 10.1007/s10773 - 010 - 0606 - 9 [ * * , ( ) ] link:\doibase 10.1143/ptp.123.395 [ * * , ( ) ] link:\doibase 10.1016/j.physe.2014.07.030 [ * * , ( ) ] http://stacks.iop.org/0953-8984/28/i=37/a=375301 [ * * , ( ) ] link:\doibase 10.1103/physreve.91.013307 [ * * , ( ) ] link:\doibase 10.1016/j.cpc.2012.11.019 [ * * , ( ) ] | we analyze theoretically the charging current into , and the transport current through , a nanoscale two - dimensional electron system with two parallel quantum dots placed in a photon cavity .
a plunger gate is used to place specific many - body states of the interacting system in the bias window defined by the external leads .
we show how the transport phenomena active in the many - level complex central system strongly depend on the gate voltage .
we identify a resonant transport through the central system as the two spin components of the one - electron ground state are in the bias window . with two - electron states in the bias window
we observe a more complex situation with intermediate radiative and nonradiative relaxation channels leading to a steady state with a weak nonresonant current caused by inelastic tunneling through the two - electron ground state of the system .
various properties of nanoscale electron and spin systems in microwave cavities are presently the focus point of many researchers .
just to mention some ; photon emission from a cavity - coupled double quantum dot caused by an electron transport through it has been reported,@xcite and the manipulation of spin qubits in cavities has gained paramount interest.@xcite investigations of transport of electrons through solid state electronic systems placed in photon cavities are gaining attention .
partially , due to the obvious connection to efforts to achieve quantum computation in a solid state system , and partially due to the interest to study fundamental light - matter interactions in a system expected to be highly tunable and offer increased sensitivity of measurements . in several cases
the electronic systems have been single or multiple quantum dots created with inas,@xcite gaas,@xcite carbon nanotubes,@xcite or graphene,@xcite and very recently in sige heterostructures.@xcite experiments have been reported on carbon nanotube quantum dots in a planar microwave cavity coupled to external fermionic or superconducting leads .
the sensitivity of the measurements due to the cavity allows for the detection of a photon assisted current of 0.3 pa corresponding to the mean photon number in the cavity being 120,@xcite a current much lower than is common in measurements of photon assisted tunneling through quantum dots , when not placed in a cavity.@xcite numerous models have been presented for transport and processes in cavity - quantum electrodynamics systems,@xcite and time dependent electron transport through nano electron systems in a photon cavity in the transient,@xcite or in the long time regime on the way towards the steady state.@xcite here , we present results concentrating on the charging current of , or the transport current through , a system of two multilevel parallel quantum dots embedded in a short two - dimensional quantum wire placed in a photon cavity with one mode .
we use a recently developed markovian version of the model to analyze the transport current in the long time limit in liouville space.@xcite alternative scheme has been employed by marino and diehl that transform a markovian master equation into a dynamical field theoretical model that they subject to a keldysh functional renormalization to investigate nonequilibrium phase transitions in driven open systems.@xcite as we are describing transport through a multilevel system including both the para- and diamagnetic part of the electron - photon interaction@xcite we will not be limited to couple any particular two levels of the electron system resonantly with the cavity photon mode , but we are rather interested in analyzing all underlying relaxation channels .
experiments on transport through electron systems in a photon cavity are usually not performed with time resolution , but in externally photon pumped cavities.@xcite calculational results from time resolved models of the underlying processes can serve as a basis for understanding and interpreting experimental results with growing complexity of the systems , as will be discussed below .
our goal is to describe complex nanoscale electron systems placed in the photon cavity , and that requires a special attention to the geometry of the system and all interactions and couplings .
we use a model with state dependent coupling tensor between the central electron system and the semi - infinite external leads and a stepwise introduction of complexity ( interactions and components ) together with appropriate truncation of the ensuing many - body spaces at each step.@xcite |
the pure leptonic decays of the heavy - light meson @xmath12 are very interesting . in principle
, the pure - leptonic decay @xmath13(see , fig.1 ) can be used to determine the decay constant @xmath5 if the fundamental cabibbo - kobayashi - maskawa matrix element @xmath4 of standard model ( sm ) is known .
conversely if we know the value of decay constant @xmath5 from other method@xcite , these process also can be used to extract the matrix element @xmath4 .
but there are the well known effect of helicity suppression and there are only one detected finial state , the measurement of pure leptonic decays of such pseudoscalar meson is very difficult .
the pure leptonic weak decays of the pseudoscalar meson corresponding to fig.1 are helicity suppressed by factor of @xmath14 : @xmath15 of them only the process @xmath16 is special , it does not suffer so much from the helicity suppression , and its branching ratio may reach to @xmath17 .
however the produced @xmath18 will decay promptly and one more neutrino is generated in the cascade decay at least , thus it makes the decay channel difficult to be observed .
for the channels when the lepton is @xmath19 or @xmath20 and their neutrinos , besides the small branching ratios , there are only one detected finial state , the measurement of such channels are very difficult .
fortunately , having an extra real photon emitted in the leptonic decays , the radiative pure leptonic decays can escape from the suppression@xcite , furthermore , as pointed out in ref.@xcite , with the extra photon to identify the decaying pseudoscalar meson @xmath12 in experiment from the backgrounds has advantages , since one more particle can be detected in the detector . although the radiative corrections are suppressed by an extra electromagnetic coupling constant @xmath21 , it will not be suppressed by the helicity suppression .
therefore , the radiative decay may be comparable , even larger than the corresponding pure leptonic decays@xcite .
so , the problem to increase the accuracy of the theoretical calculation , at least up to the first order radiative corrections , emerges .
the radiative pure leptonic decays , theoretically , have infrared divergences and will be canceled with those from loop corrections of the pure leptonic decays . in all the existing calculation of radiative decays@xcite ,
this part is ignored , since they do not include the radiative corrections of the pure leptonic decays . in this paper , we are interested in considering the radiative decays and the pure leptonic decays with one - loop radiative corrections together .
since the process @xmath16 dose not suffer the helicity supression and has a large branching ratio , the corresponding loop correction(virtual photon ) to these process should has a considerable larger branching ratio , at least comparing with the radiative decay , and can not be ignored . the paper is organized as follow : in section ii , we present the calculations of radiative pure leptonic contributions : _
i ) _ besides the leptonic decays shown as fig.1 , the radiative decays into three family leptons with a real photon appearing in the final state as shown in fig.2 ; _ ii ) _ the virtual photon corrections to the pure leptonic decays , i.e. , a photon in loop as shown in figs.3.1 , 3.2 , 3.3 , 3.4 . in section iii
, we evaluate the values of the decays . as uncertainties the dependence of the results on the parameters appearing in the considered model
is discussed .
finally , preliminary conclusion is obtained .
let us start with the radiative decay .
these contributions are corresponding to the four diagrams in fig.2 . according to the constituent quark model which is formulated by bethe - salpeter ( b .- s . )
equation , the amplitude turns out to be the four terms @xmath22 : @xmath23\times\ ] ] @xmath24 { \epsilon}^{\lambda}\times\ ] ] @xmath25 @xmath26{\frac { i \left(-g^{\mu\nu}+ \frac{p^{\mu}p^{\nu}}{m_{w}^{2}}\right)}{p^{2}-m_{w}^{2}}}\ ] ] @xmath27 @xmath28\ ] ] @xmath29 @xmath30\ ] ] @xmath31 where @xmath32 is bethe - salpeter wave function of the meson @xmath12 ; @xmath33 is the momentum of @xmath12 ; @xmath34 are the polarization vector and momentum of the emitted photon . in the quark model ,
the momenta of @xmath35-quarks inside the bound state , i.e. , the @xmath12 meson , are : @xmath36 where @xmath37 is the relative momentum of the two quarks inside the @xmath12 meson . as the @xmath12 meson is a nonrelativistic bound state in nature ,
so the higher order relativistic corrections may be computed precisely , but being an approximation for a @xmath38-wave state , and focusing the light on the radiative decay corrections only at this moment now , we ignore @xmath37 dependence , i.e. , we may still have the non - relativistic spin structure for the wave function of the meson @xmath12 ( a @xmath39 state ) correspondingly : @xmath40 here @xmath41 is the wave function at origin , and by definitions it connects to the decay constant @xmath5 : @xmath42 where m is the mass of @xmath12 meson .
moreover we note that for convenience we take unitary gauge for weak bosons to do the calculations throughout the paper . with a straightforward computation
, the amplitude can be simplified as : @xmath43 @xmath44(1-\gamma_{5})\nu_{\ell},\ ] ] @xmath45 @xmath46 & + & \left .
is_{1}\varepsilon^{\alpha\mu\beta\nu}p_{\alpha } \epsilon_{\mu}k_{\beta}\gamma_{\nu}\right]+ \frac{(\not\ !
{ p}-\not\ ! { k})}{m_{w}^{2}}p\cdot{\epsilon}(m^{2}-p\cdot{k } )
\right\}(1-\gamma_{5})\nu_{\ell } , \end{array}\ ] ] where @xmath47 is the momentum of the charged lepton , and @xmath48 @xmath49 as a matter of fact , there is infrared infinity when performing phase space integral about the square of matrix element at the soft photon limit .
it is known that the infrared infinity can be cancelled completely by that of the loop corrections to the corresponding pure leptonic decay @xmath50 . in eqs.(7 ) and ( 8) , the infrared terms can be read out : @xmath51(1-\gamma_{5})\nu_{\ell}.\ ] ] as the diagrams ( g ) , ( h ) , ( i ) , ( j ) in figs.3.3 , 3.4 always have a further suppression factor @xmath52 to compare with the other loop diagrams , and there is no infrared infinity in these four loop diagrams , we may ignore the contributions from these four diagrams safely . furthermore we should note that in our calculations throughout the paper , the dimensional regularization to regularize both infrared and ultraviolet divergences is adopted , while the on - mass - shell renormalization for the ultraviolet divergence is used .
if feynman gauge for photon is taken ( we always do so in the paper ) , the amplitude corresponding to the diagrams ( a ) , ( b ) of fig.3.1 can be written as : @xmath53}\right]\ ] ] @xmath54 ( -\gamma_{\nu})(1-\gamma_{5})\nu_{\ell},\ ] ] @xmath55}\right]\ ] ] @xmath56 ( -\gamma_{\nu})(1-\gamma_{5})\nu_{\ell},\ ] ] where the @xmath57 , @xmath58 denote the momenta of the loop and the neutrino respectively .
these two terms also have infrared infinity when integrating out the loop momentum @xmath57 . after doing the on - mass - shell subtraction , the terms corresponding to vertex and self - energy diagrams ( c ) , ( d ) , ( e )
, ( f ) can be written as : @xmath59 @xmath60@xmath61.\ ] ] now let us see the cancellation of the infrared divergencies precisely .
the infrared parts of the decay widths which are from the interference of the self - energy and vertex correction diagrams with the tree diagrams : @xmath62 @xmath63 @xmath64 \ ; ; \ ] ] the infrared part of the decay widths from the interference of the `` box '' correction diagrams with the tree diagrams : @xmath65\ ; ; \ ] ] the infrared part of the decay width from the real photon emission : @xmath66 @xmath67 \;.\ ] ] here @xmath68 is a small energy , which corresponds to the experimental resolution of a soft photon so that the phase space of the emitting photon in fact is divided into a soft and a hard part by @xmath68 . where @xmath20 is the dimensional parameter appearing in the dimensional regularization .
it is easy to check that when adding up all the parts : the real photon emission @xmath69 and the virtual photon corrections @xmath70 , the infrared divergences @xmath71 are canceled totally and the @xmath20 dependence is also cancelled .
hence we may be sure that we finally obtain the pure leptonic widths for the @xmath12 meson decays to the three families of leptons which are accurate up - to the ` next'-order corrections and ` infrared finite ' but depend on the experimental resolution @xmath68 .
first of all , let us show the ` whole ' leptonic decay widths , i.e. , the sum of the corresponding radiative decay widths and the corresponding pure leptonic decay widths with radiative corrections , and put them in table ( 1 ) . why we put the radiative decay and the pure leptonic decay with radiative corrections together here
is to make the width not to depend on the experimental resolution for a soft photon .
the values for the parameters @xmath72 , @xmath73@xcite , @xmath74 gev , @xmath75 gev , @xmath76 gev and @xmath77 gev@xcite . + considering the possible experimental resolutions , our prediction of the radiative decay widths @xmath78 and @xmath79 are close to the values in ref@xcite , but our prediction of @xmath80 is much smaller than the one in ref@xcite . in our model , if we using a smaller cut @xmath81 , the obtained @xmath82 will become larger , but since the decay widths depend on @xmath83 , the change of branching ratios will be not so much on the selecting of @xmath81 , we can see this in table ( 5 ) and ( 6 ) , and for another example , if @xmath84 , the @xmath85 , @xmath86 , @xmath87 , but it is very difficult to obtain so small a @xmath81 in experiment .
we can conclude that the best radiative decay channel is easy to search is @xmath88 . for the convenience to compare with experiments , we present the photon spectrum of the radiative decays in fig.4 and fig.5 .
in addition , we should note that the widths are quite sensitive to the decay constant @xmath5 , and are sensitive to the values of the quark masses @xmath89 and @xmath90 .
s2 as.abada et al .
phys . b*376 * ( 1992 ) 172 .
c.r.allton et al . , ape collaboration , nucl .
b(proc . suppl.)*34 * ( 1993 ) 465 .
c.w.bernard et al . , phys .
rev . d*49 * ( 1994 ) 2536 .
gustavo burdman , t.goldman and daniel wyler , phys .
rev . d*51 * ( 1995 ) 111 .
d.arwood , g.eilam and a.soni , mod .
a11 ( 1996 ) 1061 .
c.q.geng,c.c.lih and wei - min zhang,*hep - ph*/0012066 .
chang , c - d l , g .-
wang and h .- s .
zong , phys .
rev . d*60 * ( 1999 ) 114013 .
d.e.groom et al . , the european physical journal c*15 * ( 2000 ) 1 .
c.r.allton et al . ,
ape collaboration , nucl .
* 34 * ( 1993 ) 456 . | the radiative corrections to the pure leptonic decay @xmath0 up - to one - loop order is presented .
we find the virtual photon loop corrections to @xmath1 is negative and the corresponding branching ratio is larger than @xmath2 .
considering the possible experimental resolutions , our prediction of the radiative decay @xmath3 is not so large as others , and the best channel to determine the @xmath4 or @xmath5 is @xmath6 . how to cancel the infrared divergences appearing in the loop calculations , and the radiative decay @xmath7 is shown precisely .
it is emphasized that the radiative decay may be separated properly and may compare with measurements directly as long as the theoretical ` softness ' of the photon corresponds to the experimental resolutions .
9.in 6.2 in @xmath8 , department of physics , fujian normal university , fuzhou 350007 , china + @xmath9 , institute of theoretical physics , academia sinica , p.o.box 2735 , beijing 100080,china + @xmath10 , department of physics , nankai university , tianjin 300070 , china + @xmath11 , ccast ( world laboratory ) p.o .
box 8730 , beijing 100080 , china . |
high controllability of ultracold fermi gases enables us to study various many - body phenomena in a systematic manner@xcite . using a tunable pairing interaction associated with a feshbach resonance
, we can study properties of fermi superfluids from the weak - coupling regime to the strong - coupling limit in a unified manner@xcite .
introducing population imbalance to a superfluid fermi gas@xcite , one can simulate fermi superfluids discussed in various research fields , such as metallic superconductivity under an external magnetic field , and color superconductivity in a dense quark matter@xcite .
the population imbalance naturally induces mismatch of fermi surfaces between ( pseudo)spin-@xmath2 and ( pseudo)spin-@xmath3 components , which suppresses the conventional @xmath4-wave superfluid state@xcite . in this case ,
other pairing states , such as the fulde - ferrell - larkin - ovchinnikov ( fflo ) state@xcite and the sarma state , have been predicted@xcite
. when we deal with both the tunable interaction and population imbalance at the same time , pairing fluctuations are crucial in determining the phase diagram in terms of the interaction strength , polarization , and temperature . even in the weak - coupling regime , it has been pointed out@xcite that the fflo phase is unstable against pairing fluctuations in the absence of a lattice potential . however , despite the importance of strong - coupling effects , polarized fermi gases have been so far mainly examined within the mean - field level , so that roles of pairing fluctuations in this system have not been fully understood yet .
one of the crucial reasons for this current situation is the breakdown of the strong - coupling theory developed by nozires and schmitt - rink ( nsr)@xcite in the presence of population imbalance@xcite .
while the nsr theory has succeeded in explaining the bcs ( bardeen - cooper - schrieffer)-bec ( bose - einstein condensation ) crossover physics in unpolarized fermi gases , it gives negative spin susceptibility in the presence of population imbalance@xcite , which is , however , forbidden thermodynamically@xcite . to resolve the above mentioned problem that the nsr theory possesses , we investigate a polarized fermi gas in the bcs - bec crossover region .
we point out that one must carefully treat the pseudogap phenomenon associated with pairing fluctuations , as well as many - body corrections to the spin fluctuations in the presence of population imbalance . to overcome this problem , the ordinary @xmath0-matrix theory ( which has been
also frequently applied to the unpolarized case ) is found to be still not enough . further extending the @xmath0-matrix theory to include higher order fluctuation effects , we obtain the expected positive spin susceptibility in the entire bcs - bec crossover region .
the calculated spin susceptibility agrees well with the recent experiment on a @xmath1li fermi gas . using this extended @xmath0-matrix theory
, we also determine the superfluid phase transition temperature @xmath5 in a polarized fermi gas .
we consider a two - component fermi gas with population imbalance ( @xmath6 , where @xmath7 is the number of atoms with pseudospin @xmath8 ) .
since recent experiments are all using a broad feshbach resonance , we employ the single - channel bcs model , described by the hamiltonian , @xmath9 here , @xmath10 is the creation operator of a fermi atom with momentum @xmath11 and pseudospin @xmath12 .
@xmath13 is the kinetic energy , measured from the chemical potential @xmath14 ( where @xmath15 is an atomic mass ) . when we write the chemical potential as @xmath16 , the hamiltonian in eq .
( [ h ] ) effectively describes a system under an external magnetic field @xmath17 .
the pairing interaction @xmath18 ( @xmath19 ) is assumed to be tunable by a feshbach resonance , which is related to the @xmath4-wave scattering length @xmath20 as @xmath21 $ ] ( where @xmath22 is a cutoff energy ) . in this scale , the weak - coupling bcs regime and the strong - coupling bec regime
are , respectively , characterized by @xmath23 and @xmath24 ( where @xmath25 is the fermi momentum ) .
the region @xmath26 is called the crossover region .
strong - coupling effects are conveniently described by the self - energy @xmath27 in the single - particle thermal green s function , @xmath28^{-1 } , \label{gf}\ ] ] where @xmath29 is the fermion matsubara frequency , and @xmath30^{-1}$ ] is the non - interacting green s function . in this paper , we take the self - energy as @xmath31 here , @xmath32 is the boson matsubara frequency .
the particle - particle scattering matrix @xmath33 $ ] involves fluctuation contributions described by the sum of ladder - type diagrams@xcite , where @xmath34 is the lowest - order pair - propagator .
th self - energy in the ordinary @xmath0-matrix theory is given by eq .
( [ se ] ) where @xmath35 is replaced by @xmath36 , as @xmath37 the nsr theory also uses eq .
( [ set ] ) , where eq .
( [ gf ] ) is expanded to o(@xmath38 ) , as @xmath39 for clarify , we call the strong - coupling theory using the self - energy in eq .
( [ se ] ) the _ extended _
@xmath0-matrix theory in this paper .
the superfluid phase transition temperature @xmath5 is determined from the thouless criterion @xmath40 when the highest @xmath5 is obtained at @xmath41 , the ordinary uniform superfluid phase is realized .
the fflo state , on the other hand , corresponds to the case of @xmath42 .
however , since the latter state is known to be unstable against pairing fluctuations in the absence of a background lattice potential@xcite , we only consider the case of @xmath41 in this paper . as usual , we solve eq .
( [ thoulessc ] ) , together with the number equations @xmath43
the spin susceptibility @xmath44 is given by @xmath45 substituting the green s function given by eq .
( [ gf ] ) into eq .
( [ sus ] ) , we automatically obtain , not only self - energy corrections , but also vertex corrections to the spin susceptibility in a consistent manner . in the extended @xmath0-matrix theory ( where the self - energy in eq .
( [ se ] ) is used ) , many - body corrections to the spin susceptibility ( @xmath46 ) are diagrammatically given by the first line in fig.[fig1 ] . here
, the first term ( a1 ) involves many - body corrections to @xmath44 through the density of states @xmath47 .
that is , noting that the spin susceptibility @xmath44 is deeply related to @xmath47 at the fermi level ( @xmath48 ) , when pairing fluctuations induce the pseudogap in @xmath47 above @xmath5 , @xmath44 is also suppressed due to the suppression of @xmath47 around @xmath48 . to understand the role of ( a2 ) , ( a3 ) and higher order diagrams in fig.[fig1 ] , it is convenient to approximate @xmath49 to the lowest - order contribution @xmath18 .
then , the sum of these diagrams are found to be similar to the random - phase approximation ( rpa ) , as @xmath50 where @xmath51 is given by the diagram ( a1 ) in fig.[fig1 ] . when one uses the nsr green s function @xmath52 in eq .
( [ sus ] ) , the resulting spin susceptibility ( @xmath53 ) is diagrammatically described by the third line in fig.[fig1 ] . reflecting that the self - energy is only taken into account to @xmath54 , the diagram ( a1 ) is now approximated to ( c1 ) and ( c2 ) . in this regard
, we note that the nsr green s function @xmath55 is known to overestimate the magnitude of the pseudogap@xcite , leading to the unphysical _ negative _ density of states around the fermi level in the bcs - bec crossover region . because of this
, the suppression of the spin susceptibility by the pseudogap effect is also overestimated in @xmath56 , leading to the negative spin susceptibility , as shown in fig.[fig2 ] . in the strong - coupling bec limit , one finds@xcite @xmath57 where @xmath58 is the susceptibility of a free fermi gas .
@xmath59^{-1 } $ ] and @xmath60 is the superfluid phase transition temperature in the bec limit ( where @xmath61 is the fermi temperature ) .
the second term in eq . ( [ afx ] ) just arises from the diagram ( c2 ) in fig .
[ fig1 ] , which is related to fluctuation corrections to the density of states@xcite . as shown in fig.[fig2 ] , this term is the origin of the negative spin susceptibility in the bec regime .
the problem of the negative density of states is absent in the ordinary @xmath0-matrix theory ( where eq .
( [ set ] ) is used)@xcite . in this case , the ( a1)-type diagram in fig.[fig1 ] is correctly included as ( b1 ) in the second line of fig.[fig2 ] .
however , when we evaluate @xmath44 based on this theory ( @xmath62 ) , although the situation becomes much better than the nsr result , @xmath63 still becomes negative for @xmath64 .
( see the inset in fig.[fig2 ] ) .
this is because the rpa - type series in the extended @xmath0-matrix theory is truncated at the first order ( ( b2 ) in fig.[fig2 ] ) in the ordinary @xmath0-matrix theory .
evaluating @xmath63 by approximating @xmath49 to @xmath18 , one obtains @xmath65 while eq.([susetma ] ) is always positive , eq.([sustma ] ) becomes negative when @xmath66 .
the above analyses clearly indicate that one needs to carefully treat the pseudogap effect and the rpa - type many - body corrections to the spin susceptibility in the presence of population imbalance .
since they are both correctly treated in our extended @xmath0-matrix theory , the calculated spin susceptibility @xmath67 is positive in the entire bcs - bec crossover . as shown in fig.[fig2 ] , @xmath67 gradually decreases with increasing the interaction strength , and almost vanishes when @xmath68 , because the spin degrees of freedom almost dies out due to the formation of tightly binding singlet pairs in the bec regime .
we compare our theoretical result with the recent experiment on @xmath1li@xcite in fig.[fig3 ] .
the calculated spin susceptibility @xmath67 well agrees with the observed susceptibility in the normal state ( the left side of the vertical dotted line ) .
we emphasize that there is no fitting parameter in obtaining fig.[fig3 ] .
so far , we have discussed spin susceptibility , which is obtained in the zero polarization limit ( @xmath69 ) . here , we briefly consider the case of finite polarization .
figure [ fig4 ] shows the phase diagram of a polarized fermi gas in terms of the interaction strength , polarization @xmath70 , and the temperature . in the weak - coupling bcs regime
, @xmath5 vanishes at a critical polarization @xmath71 . the magnitude of @xmath71 increases , as one passes through the bcs - bec crossover region . in the bec regime
, one finds @xmath72 .
we briefly note that , although we have only considered the second order phase transition by using the thouless criterion in fig.[fig4 ] , the first order transition may exist below @xmath71@xcite . to examine the first order phase transition of a polarized fermi gas
, one needs to take into account the possibility of phase separation between the paired molecules and fully polarized fermi atoms , which remains as our future problem .
to summarize , we have investigated an ultracold fermi gas with population imbalance .
we clarified that the breakdown of the nsr and @xmath0-matrix theories in the presence of population imbalance can be eliminated by carefully treating the pseudogap effects and the rpa - type many - body corrections to the spin susceptibility .
the resulting extended @xmath0-matrix theory gives the expected positive spin susceptibility in the entire bcs - bec crossover region .
the calculated spin susceptibility well agrees with the recent experiment on a @xmath1li fermi gas . since polarized fermi gases are deeply related to metallic superconductivity under an external magnetic field
, our results would be useful for the further development of the physics of polarized fermi superfluids .
we would like to thank s. watabe , y. endo , d. inotani , and r. hanai for useful discussions .
y. o. was supported by grant - in - aid for scientific research from mext in japan ( 22540412 , 23104723 , 23500056 ) .
99 s. giorgini , l. pitaevskii , and s. stringari , _ rev .
phys . _ * 80 * , 1215 ( 2008 ) .
i. bloch , j. dalibard , and w. zwerger , _ rev .
_ * 80 * , 885 ( 2008 ) . c. chin , r. grimm , p. julienne , and e. tiesinga , _ rev .
* 82 * , 1225 ( 2010 ) .
m. w. zwierlein , a. schirotzek , c. h. schunck , and w. ketterle , _ science _ * 311 * , 492 ( 2006 ) . g. b. partridge , w. li , r. i. kamar , y .- a .
liao , and r. g. hulet , _ science _ * 311 * , 503 ( 2006 ) .
w. v. liu and f. wilczek , _ phys .
* 90 * , 047002 ( 2003 ) .
a. m. clogston , _ phys .
* 9 * , 266 ( 1962 ) .
p. fulde and r. a. ferrell , _ phys .
rev . _ * 135 * , a550 ( 1964 ) .
a. i. larkin and y. n. ovchinnikov , _ sov .
. jetp _ * 20 * , 762 ( 1965 ) . for review ,
see d. e. sheehy and l. radzihovsky , _ ann .
_ * 322 * , 1790 ( 2007 ) . h. shimahara , _ j. phys
jpn . _ * 67 * 1872 ( 1998 ) .
y. ohashi , _
jpn . _ * 71 * , 2625 ( 2002 ) .
p. nozires and s. schmitt - rink , _ j. low temp . phys .
_ * 59 * , 195 ( 1985 ) .
liu and h. hu , _ europhys .
_ , * 75 * , 364 ( 2006 ) . m. m. parish , f. m. marchetti , a. lamacraft , b. d. simons , _ nat
. phys . _ * 3 * , 124 ( 2007 ) .
g. l. sewell , _ quantum mechanics and its emergent macrophysics _ , princeton university press , 2002 .
s. tsuchiya , r. watanabe , and y. ohashi , _ phys .
a _ , * 80 * , 033613 ( 2009 ) .
t. kashimura , r. watanabe , and y. ohashi , in preparation .
a. a. varlamov , g. balestrino , e. milani , and d. v. livanov , _ adv .
phys . _ * 48 * , 655 ( 1999 ) . c. sanner
, e. j. su , a. keshet , w. huang , j. gillen , r. gommers , and w. ketterle , _ phys .
_ * 106 * , 010402 ( 2011 ) . | we investigate magnetic properties of an ultracold fermi gas with population imbalance . in the presence of population imbalance , the strong - coupling theory developed by nozires and schmitt - rink ( which is frequently referred to as the nsr theory , or gaussian fluctuation theory ) is known to give unphysical results in the bcs - bec crossover region .
we point out that this problem comes from how to treat pseudogap effects originating from pairing fluctuations and many - body corrections to the spin susceptibility .
we also clarify how to overcome this problem by including higher order fluctuations beyond the ordinary @xmath0-matrix theory .
calculated spin susceptibility based on our extended @xmath0-matrix theory agrees well with the recent experiment on a @xmath1li fermi gas .
pacs numbers : 03.75.hh , 05.30.fk , 67.85.lm . |
in the context of image analysis and image processing a variety of generative models for digitalized picture arrays in the two dimensional plane have been proposed @xcite . out of the different techniques adopted for various models ,
grammar based techniques utilize the rich theory of formal grammars and languages and develop array grammars generating two dimensional languages whose elements are picture arrays .
there are two distinct types of array grammars , isometric array grammars and non - isometric array grammars .
since application of rewriting rule can increase or decrease the length of the rewritten part , the dimension of rewritten sub array can change in the case of non - isometric grammars but application of such a rule is shape preserving in the case of isometric grammars due to the fact that the left and right sides of an array rewriting rule is geometrically identical . in order to handle more context with rewriting systems , a system with several components
is composed and defined a cooperation protocol for these components to generate a common sentential form .
such devices are known as cooperating distributed ( cd)grammar systems @xcite .
components are represented by grammars or other rewriting devices , and the protocol for mutual cooperation modifying the common sentential form according to their own rules . a variety of string grammar system models @xcite have been introduced and studied in the literature .
rudolf freund extended the concept of grammar system to arrays @xcite by introducing array grammar system and further j. dassow , r. freund and gh .
p@xmath0un elaborated the power of cooperation in array grammar system ( cooperating array grammar system ) for various non - context - free sets of arrays which can be generated in a simple way by cooperating array grammar systems and simple picture description @xcite .
they also proved that the cooperation increases the generative capacity even in the case of systems with regular array grammar components .
different kinds of control mechanism that are added to component grammars for regulated rewriting rules have been considered in string grammar systems and such control devices are known to increase the generative power of the grammar in many cases @xcite .
random context grammar is viewed as one of the prototype mechanism in which components grammars that permit or forbid the application of a rule based on the presence or absence of a set of symbols .
hexagonal arrays and hexagonal patterns are found in the literature on picture processing and image analysis . the class of hexagonal kolam array language ( hkal )
was introduced by siromoneys @xcite .
the class of hexagonal array language was introduced by subramanian .
the class of local and recognizable picture languages were introduced by dersanambika et.al .
recently we extended cooperative distributed grammar system to hexagonal arrays and different capabilities of the system are studied @xcite .
in this paper we associate permitting symbols with rules of the grammar in the components of cooperating distributed context - free hexagonal array grammar systems as a control mechanism and investigating the generative power of the resulting systems in the terminal mode .
this feature of associating permitting symbols with rules when extended to patterns in the form of connected arrays also requires checking of symbols , but this is simpler than usual pattern matching .
the benefit of allowing permitting symbols is that it enables us to reduce the number of components required in a cooperating distributed hexagonal array grammar system for generating a set of picture arrays .
let @xmath1 be a finite non - empty set of symbols .
the set of all hexagonal arrays made up of elements of @xmath1 is denoted by @xmath2 .
the size of the hexagonal array is defined by the parameters @xmath3(left upper ) , @xmath4(left lower ) , @xmath5(right upper ) , @xmath6(right lower ) , @xmath7(upper ) , @xmath8(lower ) as shown in figure 1 . for @xmath9 the length of the left upper side of @xmath10
is denoted by @xmath11 .
similarly we define @xmath12,@xmath13,@xmath14,@xmath15 and @xmath16 .
figure 1 an isometric hexagonal array grammar is a construct @xmath17 where @xmath18 and @xmath19 are disjoint alphabets of non terminals and terminals respectively , @xmath20 is the start symbol , @xmath21 is a special symbol called blank symbol and @xmath22 is a finite set of rewriting rules of the form @xmath23 where @xmath24 and @xmath25 are finite subpatterns of a hexagonal pattern over @xmath26 satisfying the following conditions : 1 .
the shape of @xmath24 and @xmath25 are identical .
2 . @xmath24 contains at least one element of @xmath27 the elements of @xmath19 appearing in @xmath24 are not rewritten .
a non @xmath21 symbol in @xmath24 is not replaced by a blank symbol in @xmath28 4 . the application of the production @xmath23 preserves connectivity of the hexagonal array . for a hexagonal array grammar @xmath17
we can define @xmath29 for @xmath30 if there is a rule @xmath31 such that @xmath24 is a subpattern of @xmath32 and @xmath33 is obtained by replacing @xmath24 in @xmath32 by @xmath28 the reflexive closure of @xmath34 is denoted by @xmath35 the hexagonal array language generated by @xmath36 is defined by @xmath37 a hexagonal array grammar is said to be context free if in the rule @xmath23 1 . non @xmath21 symbol in @xmath24 are not replaced by @xmath21 in @xmath25 2 .
@xmath24 contain exactly one non - terminal and some occurrences of blank symbol .
.the family of languages generated by a context free hexagonal array grammar is denoted by @xmath38 .
a context free hexagonal array grammar is said to be regular if rules are of the form @xmath39
@xmath40 @xmath41 the family of languages generated by a regular hexagonal array grammar is denoted by @xmath42 a cooperating hexagonal array grammar system ( of type @xmath43 and degree @xmath44 ) , is a construct @xmath45 where @xmath18 and @xmath19 are non - terminal and terminal alphabets respectively , @xmath46 and @xmath47 are finite sets of regular respectively context free rules over @xmath48 let @xmath49 be a cooperating hexagonal array grammar system .
let @xmath50 then we write @xmath51 if and only if there are words @xmath52 such that 1 . @xmath53 2 .
@xmath54 that is , @xmath55 @xmath56 moreover , we write + @xmath57 if and only if @xmath58 for some @xmath59 + @xmath60 if and only if @xmath61 for some @xmath62 + @xmath63 if and only if @xmath64 for some @xmath65 + @xmath66 if and only if @xmath63 and there is no @xmath67 with @xmath68 by @xmath69 we denote family of hexagonal array language generated by cooperating hexagonal array grammar system consisting of at most @xmath70 components of type @xmath71 in the mode @xmath72 a random context grammar is a quadruple @xmath73 where @xmath18 is the alphabet of non - terminals , @xmath19 is the alphabet of terminals such that @xmath74 is the start symbol , and @xmath22 is a finite set of productions of the form @xmath75 where @xmath76 is a context free production,@xmath77 and @xmath78,and @xmath79 . for @xmath80 and a production @xmath81 ,
the relation @xmath82 holds provided that @xmath83 and @xmath84 .
a permitting ( forbidding)grammar is a random context grammar @xmath73 where for each production + @xmath81 , it holds that @xmath85
the set of all symbols in the labeled cells of the array @xmath86 is denoted by alph@xmath87 a permitting cf hexagonal array rule is an array grammar @xmath36 is of the form @xmath88 where @xmath23 is a context - free hexagonal array rewriting rule and @xmath89 , where @xmath18is the set of non -terminals of the grammar .
if @xmath90 , then we avoid mentioning it in the rule.for any two arrays @xmath91 and a permitting cf hexagonal array rule @xmath92 , the array @xmath93 is derived from @xmath86 by replacing @xmath24 in @xmath86 by @xmath25 provided that @xmath94 .
a permitting cooperating distributed context - free hexagonal array grammar systems ( pcdcfhags ) is @xmath95 where @xmath18 is a finite set of non terminals,@xmath20 is the start symbol,@xmath19 is a finite set of terminals , @xmath96 and each @xmath97 , for @xmath98 is a finite set of permitting cf hexagonal array rewriting rules . for any two hexagonal arrays @xmath91 , we denote @xmath99 an array rewriting step performed by applying a permitting cooperating cf hexagonal array rule in @xmath100 , and by @xmath101 the transitive closure of @xmath102 also we say that the array @xmath86 derives an array @xmath93 in the terminal mode or @xmath103 mode and write @xmath99 , if@xmath104 and there is no array @xmath105 such that @xmath106 the array language generated by @xmath49 in the @xmath103 mode is defined as @xmath107 for @xmath108 note that @xmath109 is any sequence of symbols belonging to @xmath110 where repeated symbols are allowed . also @xmath111 denote the family of array languages generated in the @xmath103 mode by permitting cooperating cf hexagonal array grammar systems with at most @xmath70 components the pcdcfhags @xmath116 where @xmath117 consists of the following rules .
@xmath118 generates ( in the @xmath103 mode ) the set @xmath119 of all arrays over @xmath115 in the shape of a left arrow head with left upper arm and left lower arm are equal in size ( figure 3 ) .
@xmath120 the derivations starts with rule 1 followed by rule 2 which can be applied as the permitting symbol @xmath121 is present in the array .
this grows left upper arm ( lu ) by one cell .
then rule 3 can be applied due to the presence of permitting symbol @xmath122 and this grows left lower arm ( ll ) by one cell .
an application of rule 4 followed by 5 , again noting that the permitting symbols of the respective rules are present changes @xmath122 to @xmath123 and @xmath124 to @xmath121 .
the repeated application of the process growing both the left upper arm and left lower arm equal in size .
rules 6 and 7 are applied changing @xmath123 to @xmath125 and @xmath121 to @xmath126 so that the derivation can be terminated by the application of rules 8 and 9 thus yielding a hexagonal array in the shape of a left arrow head with size of @xmath3 and @xmath4 are equal .
a @xmath127 where the set of permitting symbols in all the components is empty , is simply a cooperating distributed @xmath128 hexagonal array system ( @xmath129)@xcite . the family of array languages generated in the @xmath103 mode by a @xmath129 with at most @xmath70 components is denoted by @xmath130 if the rules in all the components are only in the form of rules of a regular array grammar , then this family is denoted by @xmath131 we now show that the set @xmath132 of all @xmath133 arrays over @xmath115 in the form of hollow hexagonal frame .
figure 3 can be generated ( in the @xmath103 mode ) by a @xmath127 with only two components .
the set @xmath132 is generated ( in the @xmath103 mode ) by the @xmath135 .
the rules in the component@xmath136 are given by @xmath137 @xmath138the rules in the component @xmath139 are given by @xmath140 @xmath141 using @xmath103- mode of derivation , starting with the symbol @xmath142 an application of rule ( 1 ) in the first component followed by rule(2 ) which can be applied as the permitting symbol @xmath121 is present in the array , grows in the @xmath3 arm by one place .
since @xmath122 is the permuting symbol for rule ( 3 ) , rule ( 3 ) can then applied which results the growth of @xmath4 arm by one place .
now the situations are ready for applying rules ( 4 ) and ( 5 ) and at this stage @xmath122 becomes @xmath123 and @xmath124 becomes @xmath121 .
the process can be repeated and this in turn results the growth of @xmath3 and @xmath4 arms equal in length . instead of rule
( 4 ) rule ( 6 ) is applied followed by ( 7),(8),(9),(10),(11 ) allows upper and lower arms to grow in equal length , with permitting symbols in all these rules directing the sequence of applications in the right order .
if rule ( 12 ) is used instead of rule ( 10 ) and this is followed by rule ( 13 ) right upper ( @xmath5 ) and right lower ( @xmath6 ) arms grows equal in size and correct application of rule ( 18 ) and ( 19 ) will result in the symbol @xmath10 in the @xmath5 arm and @xmath143 in the lower right arm where @xmath10 is at the position of right end point of @xmath5 and @xmath6 arm so that further application of productions in @xmath117 is not possible at any non - terminals . at this stage applying productions in @xmath144 and this in turn terminate the derivation yielding a hollow hexagon with its parallel arms are equal in size .
the equality follows from the results from @xcite .
we know that , a @xmath127 with empty set of permitting symbols associated with the rules is same as a cooperating distributed @xmath128 hexagonal array grammar system and so @xmath151 if @xmath90 .
examples ( 1 ) and ( 2 ) illustrated the fact that the set @xmath119 of all hexagonal arrays over @xmath152 in the shape of left arrow head with left upper arm and left lower arm with equal size is generated by the @xmath153 with only one component and working in the @xmath103-mode and hence the inclusion is proper which proves ( 1 ) .
similar arguments for inclusion in statement(2 ) are hold . from the proof of the lemma(1 ) it is very clear that under the strict application of the derivation rules with the respective permitive symbols generate a hollow hexagon with parallel arms equal in size . such a generation is not possible in @xmath154 since @xmath155 and incorrect application of terminating rule will leads to non - completion of the hollow hexagon with parallel arms equal in size .
thus @xmath156 consider the array languages generated by @xmath127 in example ( 1 ) and in the proof of lemma ( 1 ) .
it can be seen that generating the arrow head of patterns of the language , we should require two growing heads at the same time .
but in the @xmath157 with any number of components the array rules contains only one growing head .
so the same language can not be generated by @xmath158 and this proves ( 3 ) . to show the power of the cooperating hexagonal array grammar system with the array rules controlled by permitting symbols , consider the following example of a language of set of all hexagons with parallel arms are equal in length over a one letter alphabet .
starting with @xmath168 repeated application of the first five rules of the component @xmath136 in this order generate @xmath169 having equal arms , with @xmath3 arm having non - terminal @xmath123 and @xmath4 arm having non - terminal @xmath170 once two rules @xmath171 and @xmath172 of component @xmath136 are used , the generations of these two arms end with terminal @xmath173 and then starts the generations of upper and lower arms of the hexagon using rules ( 6 ) to ( 11 ) .
then the right application of rule ( 12 ) to ( 17 ) then ( 18 ) , ( 19 ) , @xmath174 ( 45 ) subjected to the permitting symbols will result to a hexagonal picture and finally by the application of rules in @xmath175 we get the required hexagon over the one letter alphabet @xmath115 as in figure 5 . in the siromoney matrix grammar ( 9 )
rectangular arrays are generated in two phases ; one in horizontal and the other in vertical .
further it was extended by associating a finite set of rules in the second phase of generation with each table having either right linear non - terminal rules of the form @xmath176 or right - linear terminal rules of the form @xmath177 and such array languages are denoted by @xmath178 and @xmath179 and we have a well known result @xmath180 ( refer 11 ) .
correspondingly it can be established for hexagonal arrays .
here we compare @xmath127 with these classes . except the rules @xmath186 in @xmath117
generates the @xmath3 and @xmath4 arm with a middle marker @xmath10 and the symbols @xmath187 above and below @xmath10 in both arms and which are equal in number .
the first two rules of @xmath144 changes the symbols in the left most cell in to @xmath173.then the remaining rules of @xmath144 and the last two rules of @xmath117 ( @xmath188,@xmath189 ) and the rules of the component @xmath190 generate the arms such that each cell in the middle arm in the horizontal direction is made up of @xmath191 and all other cells above and below are made up of @xmath192 except the leftmost arms.the generation of the cells finally terminates , yielding the rightmost cells are rewritten by @xmath191 .
thus a hexagonal array in the shape of a left arrowhead ( describing as figure(6 ) is generated).if we treat @xmath193 as blank , such arrays can not be in @xmath179 and which in turn proves that @xmath194 .
* conclusion .
* in this paper , the picture array generating power of cooperating @xmath128 hexagonal array grammar systems endowed with permitting symbols are studied .
it is seen that the control mechanism which we here used namely the permitting symbols is shape preserving in picture generation and also it reduce size complexity .
h. bordihm , m. holzer , in : c. matin - vide , f. otto , h. fernau ( eds . ) , _ random context in regulated rewriting versus cooperating distributed grammar systems _ , in : lecture notes in computer science , vol . 5159 , springer - verlag , 2008,pp 125136 .
k. s. dersanambika , k. krithivasan , martin - vide , k. g. subramanian , _ local and recognizable hexagonal picture languages _ ,
international journal of pattern recognition and artificial intelligence , 19(7 ) , 2012 , pp 553571 .
k. s. dersanambika , k. krithivasan , h. k. agarwal , j. guptha _ hexagonal contextual array p - systems , formal models , languages and application _ , series in machine perception artificial intelligence 66 , 2006 , pp 7996 . | in this paper we associate permitting symbols with rules of grammars in the components of cooperating distributed context - free hexagonal array grammar systems as a control mechanism and investigating the generative power of the resulting systems in the terminal mode .
this feature of associating permitting symbols with rules when extended to patterns in the form of connected arrays also requires checking of symbols , but this is simpler than usual pattern matching .
the benefit of allowing permitting symbols is that it enables us to reduce the number of components required in a cooperating distributed hexagonal array grammar system for generating a set of picture arrays . *
subject classification : 68rxx * * keywords : hexagonal arrays , cooperating hexagonal array grammar systems , generative power * |
symmetries , like equivariance and reversibility , play an important role in determining the behavior of a dynamical system .
equivariant symmetries have been studied extensively in connection with bifurcation theory ; see @xcite . the general theory for reversible symmetries in dynamical systems had a more recent development as described in the thorough surveys @xcite .
indeed , symmetries are geometric invariants which play an important role through several applications in physics , from the classical @xcite and quantum mechanics @xcite to thermodynamics @xcite .
however , there is still a gap in the literature concerning the area - preserving reversible systems and the question about the validity of classic results in the this setting arises .
the main object of the present paper is twofold : we intend to study dynamical systems which keep invariant a reversing symmetry and a symplectic form at the same time .
as reported in @xcite , the interconnection between these two types of geometrical invariants is much common since there is a large class of hamiltonians which come equipped with a reversing symmetry .
a paradigmatic example relies on the chirikov - taylor standard map @xcite . in the present article , after a scrupulous examination of properly chosen examples , we would like to systematize their features and consider the general family of area - preserving maps which display a reversing symmetry .
the area - preserving maps appear naturally when considering time - one maps of a hamiltonian with @xmath7 degrees of freedom which is the base of celestial and classical mechanics . in what follows
, we characterize the maps that we are going to consider along the paper .
let @xmath2 denote a compact , connected , boundaryless , @xmath8 riemannian two - dimensional manifold , let @xmath9 be a smooth symplectic form on @xmath2 and @xmath10 the measure associated to @xmath9 that we call _ lebesgue measure _ or _
area_. setting by @xmath11 the set of @xmath6 diffeomorphisms on @xmath2 , let @xmath12 stand for the set of @xmath6 diffeomorphisms on @xmath2 such that if @xmath13 , then for every borelian subset @xmath14 of @xmath2 , the following property holds : @xmath15 .
these maps are often called _ area - preserving _
diffeomorphisms on @xmath2 .
a diffeomorphism @xmath16 is anosov if @xmath2 is a uniformly hyperbolic set for @xmath3 , _
i.e. _ there are @xmath17 and @xmath18 such that , for every @xmath19 , there is a @xmath20-invariant continuous splitting @xmath21 such that @xmath22 and @xmath23 .
we are considering the @xmath6-whitney topology where a property is said @xmath6-_generic _ if it holds on a @xmath6-residual set .
a @xmath6-_residual set _ is a countable intersection of @xmath6-open and @xmath6-dense sets .
observe that both sets @xmath24 and @xmath25 endowed with the @xmath6-topology are baire spaces ( * ? ? ?
3.5 ) : so every residual set is dense . in this paper
we address the existence of elliptic periodic orbits far from the anosov maps within the space @xmath25 which exhibit some reversing symmetry . roughly speaking , a reversing symmetry is a diffeomorphism @xmath26 such that for all @xmath19 , @xmath27 and @xmath28 .
a diffeomorphism @xmath29 is called @xmath0-_reversible _ if there exists an isometry @xmath30 , which conjugates @xmath3 to its inverse @xmath31 , _ i.e. _ , such that the following equality holds : @xmath32 the set of diffeomorphisms that are area - preserving and @xmath0-reversible will be denoted by @xmath33 .
the fixed - point subspace of the involution @xmath0 , defined as @xmath34 , is a central object of analysis .
while fixed point subspaces of preserving symmetries are invariant under the dynamics , in general they are not setwise invariant for reversing symmetries . throughout this article , this set is assumed to be a smooth one - dimensional submanifold of @xmath2 .
let @xmath35 .
we say that @xmath36 is an _ elliptic periodic orbit _ of period @xmath37 for the diffeomorphism @xmath3 if the following conditions hold : * @xmath38 ( @xmath39 i.e. @xmath40 is periodic for @xmath3 ) , * @xmath41 , for all @xmath42 and * the map @xmath43 has non - real spectrum with modulus @xmath44 . starting with birkhoff @xcite
, the dynamic structure of symplectic maps near an elliptic periodic orbit has been studied intensively in the context of hamiltonian mechanics .
going back in to the seventies , in the generic theory of area - preserving diffeomorphisms , we find the seminal result of newhouse @xcite that says that @xmath6-generic diffeomorphisms in surfaces are anosov or else the elliptical points are dense . the proof is supported in the symplectic structure and on the finding of homoclinic tangencies associated to periodic orbits .
the extension of the newhouse dichotomy for the reversible context is not so straightforward as we may expect due to the technical difficulties that the reversibility entails : there is no guarantee that the construction of the @xmath0-reversible and conservative perturbation , through homoclinic tangencies , may be done in a reversible way .
this is why our proof does not run along the same lines to that of @xcite but rather to those of @xcite .
the present article gives a proof of newhouse s dichotomy in the reversible and area - preserving setting ; we also obtain the abundance of elliptic closed orbits @xmath6-far from the anosov maps and the proof of the stability conjecture for this class .
a consequence of the latter result is that the only symplectic reversible diffeomorphisms on a compact manifold which are structurally stable ( among symplectic reversible diffeomorphisms ) are the anosov diffeomorphisms .
anosov diffeomorphisms within the class @xmath33 were studied with detail in @xcite . in the dissipative context , using the concept of heteroclinic tangencies , the authors of @xcite establish a new type of newhouse domains , in which there is a dense set of diffeomorphisms having simultaneously sinks , sources and elliptic periodic orbits . in order to extend the dichotomy of @xcite for area - preserving reversible maps , we first state and prove a new version of the @xmath6-closing lemma , one of the classic results in dynamical systems .
this is meant to refer to a bifurcation problem in which there is a recurrent ( or non - wandering ) non - closed orbit .
recall that a point @xmath45 is _ non - wandering _ for @xmath3 if for all neighbourhood @xmath46 of @xmath47 , there exists @xmath37 such that @xmath48 . by perturbing the original system ,
we obtain a @xmath6-near system that has a periodic orbit passing near the non - wandering point .
several papers deal with the classic @xmath6-closing lemma and improvements thereof .
we refer the reader to the survey @xcite that consider numerous kinds of closing lemmas . using similar arguments to those of @xcite combined with @xcite , we state and
prove the following version of the @xmath6-closing lemma : [ closing ] let @xmath0 be an isometric involution on @xmath2 .
there exists a residual set @xmath49 such that if @xmath50 , then for lebesgue almost every point @xmath45 , @xmath51 and @xmath52 , there exists @xmath53 such that @xmath54 is @xmath55-@xmath6-close to @xmath3 and @xmath56 for some @xmath57 .
using some perturbation lemmas that will be developed in section [ perturbation ] , the proof of theorem [ closing ] will be addressed in section [ closinglemma ] . a direct corollary of the reversible @xmath6-closing lemma together and the poincar recurrence theorem is the following : ( _ general density theorem_)[gdt ] there exists a residual subset @xmath58 such that the closure of the set of hyperbolic or elliptic periodic points of any @xmath59 is the whole manifold @xmath2 .
let @xmath60 .
extending the bochi - ma theorem ( @xcite ) , the authors of @xcite proved that there exists a @xmath6- residual subset @xmath61 , say @xmath62 , such that every @xmath63 either is anosov or have zero lyapunov exponents at lebesgue almost every point .
if @xmath3 is non - anosov , the result says nothing with respect to the existence of elliptic periodic points . in this paper , we are able to prove that if @xmath3 is far from being anosov , thus for any open set @xmath64 and @xmath52 , there exists @xmath65 such that @xmath54 is @xmath55-@xmath6-close to @xmath3 and @xmath54 has an elliptic periodic orbit through @xmath66 .
this is the content of the following reversible version of newhouse s dichotomy ( ( * ? ? ?
* theorem 1.1 ) ) : [ newhouse ] if @xmath67 be a map in the @xmath6-interior of the complement of the anosov maps , then for any ( non - empty ) open set @xmath64 and @xmath52 , there exists @xmath65 such that @xmath54 is @xmath55-@xmath6-close to @xmath3 and @xmath54 has an elliptic periodic orbit through @xmath66 .
an interesting consequence of proposition [ newhouse ] relies on the denseness of elliptic periodic orbits : [ newhouse2 ] there exists a @xmath6-residual @xmath68 such that any @xmath69 is anosov or else the elliptic periodic orbits of @xmath3 are dense in @xmath2 . recalling that the only surface that admits anosov diffeomorphisms is the torus @xmath70 ( see @xcite ) , when @xmath71 , the previous result can be stated in a more powerful way : if @xmath71 , then there exists a @xmath6-residual @xmath68 such that any @xmath69 the elliptic periodic orbits of @xmath3 are dense in @xmath2 . we also obtain the proof of the stability conjecture for area - preserving reversible maps .
recall , that @xmath69 is said to be @xmath6-_structurally stable _ if there exists a @xmath6-neighbourhood @xmath72 such that , for any @xmath73 , there exists a homeomorphism @xmath74 ( not necessarily in @xmath33 ) such that @xmath75 .
[ sc ] a map @xmath69 is @xmath6-structurally stable if and only if @xmath3 is anosov .
more precise statements of the results will be given throughout the article . since there is a gap in the classic literature concerning the reversible systems , in the present paper we revisit the construction of perturbation toolboxes , namely local perturbations lemmas @xcite and the franks - type lemma @xcite in section [ perturbation ] , after having provided some preliminaries and notation in section [ preliminaries ] .
the proof of theorem [ closing ] is done is section [ closinglemma ] and those of proposition [ newhouse ] , theorems [ newhouse2 ] and [ sc ] are addressed in section [ mains ] .
we also revisit newhouse s proof @xcite that obtained a dense set of elliptic points via the existence of homoclinic tangencies of the invariant manifolds of periodic orbits . at the end of this section ,
the reader will realize why the proof of proposition [ newhouse ] does not run along the same lines to that of @xcite .
in this section , we describe precisely the ambient space and properties of the maps under consideration in the article .
we introduce some general features and the property of reversibility for a map on the surface @xmath2 .
we will use the canonical norm of a bounded linear map @xmath14 given by @xmath76 . by darboux s theorem ( see e.g. ( * ? ? ? * theorem 1.18 ) ) there exists an atlas @xmath77 , where @xmath78 is an open subset of @xmath2 , satisfying @xmath79 with @xmath80 . since @xmath2 is compact we can always take an atlas with a finite number of charts , say @xmath81 .
we fix once and for all the finite atlas @xmath82 .
given @xmath83 , for some @xmath84 , let @xmath85 stands for the open ball of radius @xmath86 centered in @xmath87 .
clearly , if @xmath86 is small enough we get that @xmath88 . in this case
we will refer to @xmath89 the set @xmath90 .
we notice that all the estimates along the present paper are made in using the darboux coordinate charts ( @xcite ) .
hence , we may assume that @xmath91 , @xmath92 , and the exponential map @xmath93 , which do not preserve necessarily a symplectic structure , is , under these assumptions , the identity . here
, we introduce basic definitions and recall some results on symmetric systems , whose proof can be found in @xcite , where it has been presented a compact survey of the literature on reversible dynamical systems .
we say that the diffeomorphism @xmath5 has @xmath94-degree if @xmath95 , where @xmath96 is the lower positive integer that satisfy the previous equality and @xmath97 the identity map . in the case
@xmath98 we have that @xmath0 is an _ involution _ ,
i.e. , @xmath99 .
we say that @xmath0 is a _ reversing symmetry _ of a diffeomorphism @xmath1 if we have : @xmath100 identity ( [ reversivel1 ] ) implies that if @xmath101 is an orbit of @xmath45 , then @xmath102 is also an orbit where the time is reverted . in other words , a reversing symmetry maps orbits into orbits reversing time direction .
the definition of reversibility is algebraic and does not rely on any differentiability property of the diffeomorphism @xmath3 .
note that saying that @xmath3 is @xmath0-reversible means that @xmath0 conjugates @xmath3 and @xmath31 .
the reversing symmetries that we are going to consider along the paper will be _ isometries _ meaning that given the metric @xmath103 related to the riemannian structure in @xmath2 we have @xmath104 , that is , the pull - back of @xmath103 , via @xmath0 , leave the metric @xmath103 invariant .
since the tangent map of @xmath0 is a linear isometry it is clear that @xmath105 .
in addition , we also assume that the closed subset of fixed points of @xmath0 , that is @xmath106 , has dimension equal to one as is the most common in several examples of the literature see for instance @xcite and references therein .
in the present section , we revisit certain general perturbation statements that are at the basis of most results referred in the article .
these perturbation lemmas have been proved in the @xmath6-topology .
more precisely , we state two useful perturbation lemmas that are the basis of most results mentioned in the text : the first one , adapted from @xcite , gives a way to perform local perturbations among reversible systems .
the second relies on the franks lemma @xcite , allowing to realize locally small abstract perturbations of the derivative along periodic or non - periodic orbits .
the two types of results combine perturbations in the area - preserving setting that leave a reversing symmetry invariant .
the next result shows how to perform a local perturbation in the world of reversible systems in order to keep the perturbations inside this setting .
two general assumptions about the point @xmath45 around where we are doing a @xmath0-reversible perturbation are required : @xmath107 and @xmath108 . later on proposition
[ d ] we will se that these conditions are generic in our context .
( ( * ? ? ?
* lemma 7.1))[local ] given @xmath60 and @xmath109 , there exist @xmath110 and @xmath111 such that , for any point @xmath45 , whose orbit by @xmath3 is not periodic and @xmath112 , and every @xmath6 area - preserving diffeomorphism @xmath74 , coinciding with the identity in @xmath113 and @xmath114-@xmath6-close to the identity , there exists @xmath65 which is @xmath115-@xmath6-close to @xmath3 and such that @xmath116 outside @xmath117 and @xmath118 in @xmath119 . [ local2 ] the local perturbation @xmath120 can be done in such a way that we get @xmath121 satisfying @xmath122 .
in fact , by lemma [ local ] we can build @xmath123 supported in a ball @xmath124 and follow the same idea .
finally , we obtain @xmath125 such that @xmath126 outside @xmath127 and @xmath122 in @xmath124 .
another perturbation lemma which is a key step in order to obtain several results in dynamics is a perturbation which provides an abstract tangent action by a given map @xmath6-close to an initial one , and was first considered in @xcite for dissipative diffeomorphisms as a contribution to the solution of the _ stability conjecture _ ; it will also be used to prove theorem [ sc ] .
we will consider a finite segment of a given orbit and a small perturbation of the derivative along that segment .
then , we ask if there exists a dynamical system close to the initial one such that its derivative equals the perturbations we considered .
the following result is the version of franks lemma ( see ( * ? ? ?
* lemma 1.1 ) ) for reversing symmetric diffeomorphisms .
since @xmath3 is area - preserving , we make use of the conservative franks lemma proved in @xcite . as in @xcite , we should impose some restrictions .
[ free2 ] given a subset @xmath128 of @xmath2 , we say that @xmath128 is @xmath129-free if : @xmath130 it has been proved in ( * ? ? ?
* lemma 3.4 ) that if @xmath131 , @xmath45 and @xmath132 does not belong to the @xmath3-orbit of @xmath47 , then this orbit is @xmath129-free .
thom transversality theorem allows us to conclude that : [ d ] there exists a residual @xmath49 such that for any @xmath133 the set of orbits outside @xmath134 which are not @xmath129-free is countable .
see the proof in @xcite . as a trivial conclusion of the previous result and the fact that @xmath135 we obtain that generically in @xmath136 the set of @xmath129-free orbits has full lebesgue measure .
we will now consider an area - preserving reversible diffeomorphism , a finite set in @xmath2 and an abstract tangent action that performs a small perturbation of the derivative along that set .
then we will search for an area - preserving reversible diffeomorphism , @xmath6 close to the initial one , whose derivative equals the perturbed cocycle on those iterates .
to find such a perturbed diffeomorphism , we will benefit from the argument , suitable for area - preserving systems , presented in @xcite .
* lemma 7.4)[franks ] fix an isometric involution @xmath0 and @xmath137 .
let @xmath138 be a finite set of distinct points in @xmath2 .
asume that @xmath139 is @xmath129-free .
denote by @xmath140 and @xmath141 .
let @xmath142 be a linear area - preserving map .
for every @xmath52 , there exists @xmath143 , such that if @xmath144 , then there exists @xmath121 @xmath55-@xmath6-close to @xmath3 and satisfying @xmath145 for every @xmath146 .
moreover , if @xmath147 is a compact on @xmath2 and @xmath148 then @xmath54 can be chosen such that @xmath126 in @xmath149 .
the following result is the peak of the well known ma dichotomies on periodic orbits - the dominated splitting is the only obstruction to obtain trivial spectrum on periodic orbits .
the abstract general result for dissipative systems was obtained in ( * ? ? ?
* corollary 2.19 ) . here
we consider the area - preserving reversible version .
[ mainbgv ] let @xmath137 .
then for any @xmath52 there are two integers @xmath150 and @xmath94 such that , for any periodic point @xmath47 of period @xmath151 : 1 .
the orbit of @xmath47 is an @xmath129-free set or either ; 2 .
@xmath3 admits an @xmath150-uniform hyperbolic splitting along the orbit of @xmath47 or else ; 3 . for any neighbourhood @xmath46 of the orbit of @xmath47 ,
there exists @xmath121 @xmath55-@xmath6-close to @xmath3 , coinciding with @xmath3 outside @xmath46 and on the orbit of @xmath47 , and such that @xmath47 is a parabolic point of @xmath54 for which the differential @xmath152 has all eigenvalues real and with the same modulus , thus equal to @xmath153 .
the proof of theorem [ mainbgv ] follows closely the steps of the work @xcite , this is why we shall omit it .
however , we point out the key points that we have to be careful about : 1 .
firstly , in @xcite , the authors need not be worried about perturbation obstructions as the ones we are considering due to the reversibility : the periodic orbits in their perturbations need not to be @xmath129-free . in our case , since the lebesgue measure can not `` see '' the set of the orbits that are not @xmath129-free , we may exclude them of the scenario .
secondly , it is worth to stress that the item ( 2 ) of theorem [ mainbgv ] refers to obtaining dominated splitting . by (
* lemma 3.11 ) , in a two - dimensional area - preserving setting , @xmath150-uniform hyperbolicity is equivalent to @xmath150-dominated splitting .
3 . finally , in the absence of uniform hyperbolicity , we will combine the arguments in @xcite with lemma [ franks ] , when perturbations are requested to obtain ( 3 ) .
the perturbations that are used in @xcite are rotations which are clearly area - preserving ( see also ( * ? ? ? * lemme 6.6 ) performed for @xmath154 cocycles ) .
these transformations are used to obtain real spectrum with the same modulus ( _ i.e. _ @xmath153 ) . if @xmath49 is the @xmath6-residual set of proposition [ d ] , the kupka - smale theorem in @xmath136 allows us to conclude that : there exists a @xmath6-residual @xmath49 such that for any @xmath133 the set of periodic orbits satisfying ( 2 ) or ( 3 ) of theorem [ mainbgv ] are dense in @xmath155 . if @xmath156 , where @xmath157 is the @xmath6-residual given in corollary [ gdt ] , we easily obtain : [ good ] there exists a @xmath6-residual @xmath158 such that for any @xmath159 the set of periodic orbits satisfying ( 2 ) or ( 3 ) of theorem [ mainbgv ] are dense in @xmath2 .
a fundamental result on dynamical systems , which goes back to poincar , is the well known _
closing lemma_. in rough terms , we intend to _ close _ , in a sense that we turn it into a periodic orbit , a given recurrent or non - wandering orbit by making a small perturbation on the original system .
until now , there are satisfactory answers to this problem if the small perturbations are with respect to lower topologies ( @xmath160 and @xmath6 ) . despite the fact that the @xmath160-closing lemma is a quite simple exercise , the @xmath6 statement reveals several difficulties .
the @xmath6-closing lemma was first established by pugh @xcite in the late 1960 s and , in the early 1980 s , by pugh and robinson @xcite for a large class of systems as volume - preserving , symplectic diffeomorphisms as also for hamiltonians . in what follows ,
we present the reversing symmetric version of it .
but before that we will introduce two core tools developed in @xcite . 1 .
a fundamental key step to obtain the closing of trajectories is the _ lift lemma _ , proved for area - preserving diffeomorphisms in @xcite ( see 8 b ) or a ) because our area - preserving context is both symplectic and volume - preserving ) .
if very brief terms , _ lifting _ implies that we can push points in a given direction , acting only locally and supported in a given small ball .
the intensity of this perturbation is uniform from point to point and also proportional to the size of the ball . due to
our area - preserving and reversing symmetric restrictions the previous push must be done within this class .
the area - preserving push is guaranteed by pugh and robinson judicious perturbations . to treat the @xmath0-invariance of the perturbations adapted very carefully the ideas of @xcite .
the other tool has more to do with the drawbacks intermediate recurrences and how we can use them in our favor and also to obtain some _ conformal _ behavior .
in fact , given @xmath161 and its iterate @xmath162 , very close to one another , if we intend to push @xmath161 to @xmath163 by a small @xmath6-perturbation , two kind of problem occur : first a @xmath6-perturbation may not allow a sufficiently large shift as we have seen in ( a ) , and second , nothing assures us that we have a clean scenario in their neighbourhood , say free from intermediate recurrences formed by points @xmath164 with @xmath42 .
fortunatly , the ingenious ideas developed in @xcite solve this hard obstructions .
for that we have the _ fundamental lemma _ in @xcite , which assures that we can always find two intermediate isolated recurrences : * @xmath164 and @xmath165 with @xmath166 inside a box @xmath167 and * if we inflate @xmath167 a little bit there are no elements in @xmath168 for @xmath169 .
it has been proved in @xcite that if @xmath131 , @xmath45 and @xmath170 then , the set @xmath101 is @xmath129-free .
recall proposition [ d ] that says that generically in @xmath136 the set of @xmath129-free orbits has full lebesgue measure .
the proof of theorem [ closing ] will be done in several steps , using arguments of @xcite .
we suggest that the reader follows the proof observing figure [ closing lemma 1 ] .
let @xmath0 be an isometric involution and @xmath171 where @xmath172 is the residual set given proposition [ d ] .
following pugh and robinson @xcite , we will perform an @xmath55-perturbation of @xmath3 supported on small disks that will be specified later .
we recall the _ axiom lift lemma _ proved in @xcite that says that , in the pattern given by a basis @xmath173 of @xmath174 , one can shift the orbit through @xmath40 on the orbit through @xmath175 in @xmath176 steps , without changing @xmath3 out of the @xmath176 first iterates of the `` square '' which is @xmath177-times the smallest square containing both @xmath40 and @xmath175 with @xmath178 very small . the proof of the axiom lift is related to the _ selection theorem _
( 3.2 ) ) .
the key to the proof of the @xmath6-closing lemma is the following technical result on linear algebra ( on the tangent bundle ) .
it combines two crucial tools namely the selection theorem and the fundamental lemma ( ( * ? ? ?
* lemma ( 4.1 ) ) ) . [ crucial ] given @xmath52 , @xmath109 and @xmath179
there is @xmath180 such that : for any set of linear maps @xmath181 in @xmath182 satisfying @xmath183 , there is an orthogonal basis @xmath184 of @xmath185 such that , for every pair of points @xmath186 in the unitary square with respect to @xmath173 ( @xmath187_e^2 $ ] ) , there is a sequence @xmath188 , @xmath189 , of area - preserving diffeomorphisms @xmath190 with : 1 .
@xmath191 ; 2 .
@xmath188 is equal to @xmath192 out of the image of @xmath193_e^2})$ ] and 3 .
@xmath194 .
before we continue the proof , it is worth to interpret the meaning of the previous powerful result .
condition ( 2 ) means that , up to a dimensional constant , we can specify the proportions of the square @xmath195 with respect to the basis @xmath173 inside which a `` gradual '' perturbation occurs .
the performed perturbations occur in the tangent bundle .
to `` go down '' to the manifold @xmath2 , we need to realize the perturbation on @xmath2 .
this is the goal of the lift axiom lemma ( ( * ? ? ?
pp . 265 ) ) , which ensures that for each @xmath196 and each @xmath6 neighbourhood @xmath197 of @xmath3 , there exists a uniform @xmath198 such that for all @xmath199 , @xmath200 with @xmath201 , we have an area - preserving perturbation @xmath120 of the identity satisfying the following properties : * @xmath202 ; * @xmath203 and * the topological closure of the set where @xmath120 differs from the identity is contained in the ball centered on @xmath40 and radius @xmath204 .
this means that the perturbation @xmath120 can lift points @xmath40 in a prescribed direction @xmath205 with results @xmath206-proportional to the support . , @xmath207 , in the disk @xmath172 .
_ fundamental lemma _ shows that there exists two of these points @xmath208 and @xmath209 , @xmath210 , and a square @xmath195 containing them such that the homothetic square @xmath211 does not contain any other point @xmath212 where @xmath213.,height=566 ] let us fix the size @xmath52 of the perturbations we allow and pick some small numbers @xmath109 and @xmath179 .
remark [ local2 ] allows to fix , once and for all , an integer @xmath176 such that : @xmath214 where @xmath215 is a uniform bound for the global distortion along @xmath216 , @xmath217 where , for all @xmath218 we have @xmath219 ; the constant @xmath220 is what the authors of @xcite call the _ altitude bolicities _ of @xmath216 for @xmath221 .
let @xmath222 be a non - wandering point with respect to @xmath3 .
by definition of non - wandering point , there exists @xmath223 arbitrarily close to @xmath224 , and @xmath225 such that @xmath226 and the iterate @xmath227 is arbitrarily close to @xmath47 .
the points @xmath161 and @xmath228 are so close to @xmath47 that we may assume that the restrictions of @xmath229 to @xmath230 is governed by the linear map @xmath20 along the @xmath3-iterates of @xmath172 ( if necessary , shrink the neighbourhood of @xmath47 ) , obeying the hypothesis of lemma [ local ] .
lemma [ crucial ] provides local coordinates on the disk @xmath231 given by the basis @xmath173 .
now , we concentrate our attention at the set @xmath232 of all the return times of @xmath161 in the disk @xmath231 : @xmath233 so we get a finite set of points @xmath234 , @xmath235 , in the disk @xmath231 .
lemma [ crucial ] shows that there exists two of these points @xmath208 and @xmath209 , @xmath210 , and a square @xmath195 containing them such that the homothetic square @xmath211 does not contain any other point @xmath212 where @xmath213 .
since the orbit is @xmath129-free , it follows that @xmath107 and thus the hypothesis of lemma [ local ] holds .
using now lemmas [ crucial ] and [ local ] ( acting together ) we may perform two balanced local perturbations in order to obtain a resultant map inside the class of @xmath0-reversible maps . more precisely , both results build a @xmath55-@xmath6-perturbation @xmath236 of @xmath3 which is equal to @xmath3 out of the @xmath176 first @xmath3-iterates of the square @xmath211 , and such that @xmath237 as @xmath3 has not been changed on @xmath238 with @xmath239 , one gets that @xmath209 is a periodic orbit of @xmath54 of period @xmath240 .
twin " perturbation is automatically constructed in the following sense : taking into account lemma [ local ] , we may define additional perturbations in @xmath241 and one obtains @xmath242 @xmath55-@xmath6-close to @xmath54 such that @xmath243 outside @xmath241 and @xmath244 inside @xmath241 .
please note that we did not overlap the two perturbations during both perturbations because we have started with a small neighbourhood @xmath231 of @xmath47 such that for all @xmath245 , @xmath246 does not intersect @xmath132 . by the @xmath6-closing lemma , one knows that very non - wandering point can be made periodic by a small @xmath6-perturbation .
this periodic point can be made hyperbolic or elliptic by a new perturbation , persisting under small perturbations .
the main goal of this section is the proof of proposition [ newhouse ] and theorems [ newhouse2 ] and [ sc ] .
we first revisit the newhouse proof @xcite that obtained a dense set of elliptic points via the existence of homoclinic tangencies of the invariant manifolds of periodic orbits . the approach followed by newhouse in @xcite was to obtain elliptic points near homoclinic tangencies , when the stable and the unstable manifolds associated to a hyperbolic periodic orbit intersect in a non - transversal way .
let us see how the elliptic points are obtained : extending a result of zehnder @xcite , in @xcite , it is first proved that if @xmath3 is non - anosov , then homoclinic tangencies associated to hyperbolic periodic orbits are created for @xmath54 @xmath6-close to @xmath3 more details in @xcite .
the perturbations are not explicit .
the homoclinic tangencies are created near the point where no transversality exists and they are done in such a way that the angle between the stable and the unstable direction of a periodic hyperbolic point has a homoclinic tangency .
we point out that , in the dissipative case , an explicit relationship between homoclinic tangencies and the angle of the stable and unstable subspaces of periodic points has been given in ( * ? ? ?
* lemma 2.2.1 ) , which can not be directly extended to our study . secondly , in ( * ? ? ?
* lemma 4.1 ) , the author shows that a symplectic diffeomorphism with a non - transverse homoclinic point ( for some hyperbolic periodic orbit ) can be perturbed to produce an elliptic periodic orbit nearby . for a small @xmath52 and @xmath180 ,
the idea is to construct an isotopy @xmath247 ( @xmath248 ) of perturbing maps in @xmath249 such that : * @xmath250 * there is a point @xmath251 $ ] for which @xmath252 has a bifurcation ( a fixed point appears ) .
it implies that @xmath253 has 1 as an eigenvalue of multiplicity two ( for the map @xmath254 ) and thus an elliptic periodic point .
the combination of these two results constitutes the proof of the dichotomy . in this section
, we put together the previous information about and we show proposition [ newhouse ] .
let @xmath67 be a map in the @xmath6-interior of the complement of the anosov maps , @xmath64 a non - empty open set and @xmath52 .
we will show that there exists @xmath65 such that @xmath54 is @xmath55-@xmath6-close to @xmath3 and @xmath54 has an elliptic periodic orbit through @xmath66 . before proceeding the proof
, we need to recall a basic result on topological dimension .
there are several different ways of defining the topological dimension of topological space @xmath14 , which we shall denote by @xmath255 . for separable metrizable spaces ,
all these definitions are equivalent .
the topological dimension is a topological invariant .
for full details on this concept see @xcite , where it is proved that : ( szpilrajn theorem @xcite)[sz ] given @xmath256 . if @xmath14 has zero lebesgue measure , then @xmath257 . in order to show proposition [ newhouse ] , we start by proving the following result : [ empty ] given @xmath131 and @xmath258 a uniformly hyperbolic set , then either @xmath259 ( anosov ) or else @xmath260 has empty interior .
let @xmath131 , @xmath261 be a uniformly hyperbolic set and @xmath262 ( _ i.e. _ @xmath3 is not anosov ) .
we want to prove that @xmath260 has empty interior .
we may assume that @xmath3 is not on the boundary of anosov reversible maps on @xmath2 . by contradiction ,
suppose that @xmath260 has positive interior .
this means that there is an open set @xmath263 such that @xmath264 .
any open set has the topological dimension of the ambient space , thus @xmath265 .
hyperbolic sets have a hyperbolic continuation by any small @xmath6 perturbation of @xmath3 , which is also hyperbolic .
denote by @xmath54 any small @xmath6 perturbation of @xmath3 and @xmath266 the hyperbolic continuation of @xmath260 ( by @xmath54 ) .
since the conjugacy maps non - empty open sets into non - empty open sets , then the set @xmath266 should contain a non - empty open set and therefore @xmath267 . by zehnder (
* section 2 ) we may smoothing out the map @xmath3 _ i.e. _ there exists @xmath268 @xmath6-close to @xmath3 .
notice that there is no need @xmath269 being @xmath0-reversible .
if @xmath270 is the hyperbolic continuation of @xmath260 , then it is clear that @xmath271 .
using now ( * ? ? ?
* appendix b ) , either @xmath272 or @xmath273 .
the second case does not hold because @xmath3 is far from the anosov maps
. therefore , @xmath272 . by lemma [ sz ]
we conclude that @xmath274 which is a contradiction because it should contain an open set .
( of proposition [ newhouse ] ) let @xmath131 , let @xmath3 be @xmath6-far from the anosov maps and let @xmath66 be a non - empty open set of @xmath2 . through a @xmath6-small perturbation we may assume that @xmath159 and @xmath3 is not anosov , where @xmath275 is the @xmath6-residual set given in lemma [ good ] .
now , two situations should be taken in consideration : 1 .
if some parabolic periodic orbit of @xmath3 goes through @xmath66 we are over by just using lemma [ franks ] to turn it into an elliptic periodic point ; 2 .
otherwise , if a dense subset of @xmath150-uniformly hyperbolic periodic orbits of @xmath3 go through @xmath66 , then its closure defines a hyperbolic set @xmath276 .
since @xmath3 is @xmath6-far from the anosov maps , by theorem [ empty ] , we get that @xmath277 has empty interior , which is a contradiction . in order to prove theorem [ newhouse2 ]
, we strongly make use of proposition [ newhouse ] .
we are going to exhibit a residual set @xmath278 such that @xmath279 is a residual set of @xmath280 for which the dichotomy of theorem [ newhouse2 ] holds .
denote by @xmath281 the @xmath6-open set defined by the complement of the @xmath6-closure of the set of anosov maps @xmath282 in @xmath280 , where @xmath283 denotes the @xmath6-closure of the set @xmath14 .
let @xmath284 stands for the subset of @xmath285 such that @xmath286 if and only if @xmath3 has an elliptic periodic orbit intersecting the open ball @xmath287 .
observe that if @xmath288 is an open set , then @xmath289 defined by : @xmath290 is an open set for the product topology of @xmath285 .
let @xmath291 be a dense sequence in @xmath2 ( it exists because @xmath2 is a compact set ) and @xmath292 a sequence of positive real numbers converging to zero .
now , for @xmath37 , define @xmath293 and @xmath294 .
the set @xmath295 is the countable union of open sets and if @xmath296 then the elliptic periodic orbits of @xmath3 are dense in @xmath2 . in order to prove theorem [ sc ] , we use theorem [ newhouse2 ] .
( @xmath297 ) if @xmath3 is anosov , it is known that @xmath3 is @xmath6-structurally stable .
( @xmath298 ) suppose , by contradiction , that is @xmath6-structurally stable and non - anosov
. by theorem [ newhouse2 ] there exists a @xmath6-residual set @xmath68 such that any @xmath69 , elliptic periodic orbits of @xmath3 are dense in @xmath2 .
bearing in mind that an elliptic periodic point @xmath299 on the plane is conjugated to a rotation around an open neighbourhood @xmath300 of @xmath299 , by the pasting lemma proved in ( * ? ? ?
* theorem 3.6 ) , for any @xmath301 we may construct two perturbations of @xmath3 , say @xmath302 and @xmath303 , inside @xmath280 , which are in the @xmath6-domain of topological conjugacy of @xmath3 ( note that we are assuming that @xmath3 is @xmath6-structurally stable ) .
therefore , there exist homeomorphisms @xmath304 such that @xmath305 those perturbations can be made such that @xmath302 is a rotation of rational angle centered at @xmath299 and @xmath303 is a rotation of irrational angle centered at @xmath299
. then , there exists @xmath306 conjugating @xmath302 and @xmath303 , i.e. , @xmath307 which is a contradiction .
mb was partially supported by national funds through fct - fundao para a cincia e a tecnologia " , project pest - oe / mat / ui0212/2011 .
cmup is supported by the european regional development fund through the programme compete and by the portuguese government through the fundao para a cincia e a tecnologia ( fct ) under the project pest - c / mat / ui0144/2011 . ar was supported by the grant sfrh / bpd/84709/2012 of fct .
j. moser , e. zehnder , _ notes on dynamical systems _ , courant lecture notes in mathematics , 12 .
new york university , courant institute of mathematical sciences , new york ; american mathematical society , providence , ri , 2005 e. zehnder , _ note on smoothing symplectic and volume - preserving diffeomorphisms _
, in _ proc .
iii latin amer .
school of math .
pura aplicada cnpq , rio de janeiro , 1976 _ , volume 597 _ lecture notes in math .
_ , 828854 , springer , berlin , 1977 | in this paper we study @xmath0-reversible area - preserving maps @xmath1 on a two - dimensional riemannian closed manifold @xmath2 , i.e. diffeomorphisms @xmath3 such that @xmath4 where @xmath5 is an isometric involution .
we obtain a @xmath6-residual subset where any map inside it is anosov or else has a dense set of elliptic periodic orbits . as a consequence
we obtain the proof of the stability conjecture for this class of maps . along the paper
we also derive the @xmath6-closing lemma for reversible maps and other perturbation toolboxes .
_ msc 2000 : _ primary 37d20 ; 37c20 ; secondary 37c27 , 34d30 .
+ _ keywords : _ reversing symmetry , area - preserving map , closing lemma , elliptic point . + |
in treatment planning of radiotherapy with protons and heavier ions , the pencil - beam ( pb ) algorithm is commonly used ( hong 1996 , kanematsu 1998 , 2006 , schaffner 1999 , krmer 2000 ) , where a radiation field is approximately decomposed into two - dimensionally arranged gaussian beams that receive energy loss and multiple scattering in matter . in the presence of heterogeneity
, these beams grow differently to reproduce realistic fluctuation in the superposed dose distribution .
comparisons with measurements and monte carlo ( mc ) simulations , however , revealed difficulty of the pb algorithm at places with severe lateral heterogeneity such as steep areas of a range compensator and lateral interfaces among air , tissue , and bone in a patient body ( goitein 1978 , petti 1992 , kohno 2004 , ciangaru 2005 ) .
one reason for the difficulty is that particles in a pencil beam are assumed to receive the same interactions , whereas they may be spatially overreaching beyond the density interface .
the other reason is that only straight paths radiating from a point source are considered in beam transport , whereas actual particles may detour randomly by multiple scattering .
schneider ( 1998 ) showed that a phase - space analysis could address the overreach and detour effects for a simple lateral structure .
schaffner ( 1999 ) and soukup ( 2005 ) subdivided a physical spot beam virtually into smaller beams to naturally reduce overreaches .
pflugfelder ( 2007 ) quantified lateral heterogeneity , with which subdivision and arrangement could be optimized .
unfortunately , those techniques are ineffective against beam - size growth during transport . for electrons ,
the overreach and detour effects are intrinsically much severer .
shiu and hogstrom ( 1991 ) developed a solution , the pb - redefinition algorithm , where minimal pencil beams are occasionally regenerated , considering electron flows rigorously .
the same idea was in fact partly applied to heavy particles for beam customization ( kanematsu 2008b ) , but the poly - energetic beam model to deal with heterogeneity could be seriously inefficient in high - resolution calculations necessary for bragg peaks . in this study ,
we develop an alternative method to similarly address the overreach and detour effects . in the following sections , we incorporate our findings on the gaussian distribution into the pb algorithm , test the new method in a carbon - ion beam experiment , and discuss the results and practicality for clinical applications .
the pb algorithm in this study basically follows our former works ( kanematsu 1998 , 2006 , 2008b ) .
a pencil beam with index @xmath0 is described by position @xmath1 , direction @xmath2 , number of particles @xmath3 , residual range @xmath4 , angular variance @xmath5 , angular - spatial covariance @xmath6 , and spatial variance @xmath7 of the involved particles . as described in [ sec_appendix ] , these parameters are initialized and modified with transport distance @xmath8 . the resultant beams with variance @xmath9
are superposed to form dose distribution @xmath10 where @xmath11 is the beam-@xmath0 origin , @xmath12 is the distance at the closest approach to point @xmath13 , @xmath14 is its equivalent water depth , and @xmath15 and @xmath16 are the tissue - phantom ratio and the beam range in water .
any normalized gaussian distribution @xmath17 with mean @xmath18 and standard deviation @xmath19 can be represented with the standard normal distribution @xmath20 as @xmath21 incidentally , we have found that binomial gaussian function @xmath22,\end{aligned}\ ] ] reasonably approximates @xmath23 as shown in ( a ) , where we first fixed symmetric displacement @xmath24 for the binomial terms and determined their reduced standard deviation @xmath25 to conserve variance @xmath26 .
similarly , the daughter gaussian terms in @xmath27 splits into grand daughters to form approximate function @xmath28,\end{aligned}\ ] ] and then into grand - grand daughters to form approximate function @xmath29,\end{aligned}\ ] ] as shown in figures [ fig : splitting](b ) and [ fig : splitting](c ) .
summarizes size - reduction , displacement , and share - fraction factors for splitting with @xmath30 ( @xmath31 ) .
further splitting with the same displacement is not possible with valid ( @xmath32 ) gaussian terms .
( gray area ) and its approximate functions ( a ) @xmath27 , ( b ) @xmath33 , and ( c ) @xmath34 ( solid lines ) comprised of multiple , displaced , narrowed , and scaled gaussian distributions ( dashed lines).,width=491 ] lcccc factor name & symbol & @xmath35 & @xmath36 & @xmath37 + size reduction & @xmath38 & @xmath39 & @xmath40 & @xmath41 + displacement & @xmath42 & @xmath43 & @xmath44 & @xmath45 + share fraction & @xmath46 & @xmath47 & @xmath48 & @xmath49 + [ tab : splitting ] an overreaching gaussian beam may split two - dimensionally into @xmath50 smaller beams with these approximations .
because beam multiplication will explosively increase computational amount , it must be applied only when and where necessary with optimum multiplicity @xmath18 for required size reduction . in a grid - voxel patient model with density distribution @xmath51 , we define density gradient vector @xmath52 as @xmath53}{\delta_g}\ , \vec{e}_g,\end{aligned}\ ] ] where @xmath54 and @xmath55 are the grid interval and the basis vector for axis @xmath56 as shown in and operation @xmath57 $ ] equals @xmath58 if @xmath59 or otherwise @xmath0 .
we quantify the lateral heterogeneity by effective lateral density gradient @xmath60 with which we define the distance to an interface of density change @xmath61 as @xmath62 where @xmath63 may be appropriate for interfaces among air ( @xmath64 ) , soft tissues ( @xmath65 ) , and bones ( @xmath66 ) ( kanematsu 2003 ) .
effective lateral grid interval @xmath67 multiplied by @xmath68 is the effective distance to a second laterally adjacent grid , beyond which the distance to the interface can not be estimated from the gradient .
the pencil beams are examined at every transport step in a patient .
ones subject to splitting should be not only overreaching beyond a density interface but also substantially influential to the dose for computational efficiency . we thus require conditions @xmath69 for mother beam @xmath0 to split .
cutoff parameters @xmath70 and @xmath71 set limits on number of particles @xmath3 and residual range @xmath4 with respect to those of the ancestral original beam , @xmath72 and @xmath16 .
condition @xmath73 suppresses splitting into beams narrower than effective grid resolution @xmath74 with @xmath75 .
condition @xmath76 defines the state of overreaching .
the optimum multiplicity is determined as @xmath77 to suppress recursive splitting with @xmath75 and 3 . with the beam-@xmath0 coordinate system @xmath78 shown in and defined as @xmath79 daughter beams @xmath80 ( @xmath81 $ ] ) are initialized as @xmath82 where @xmath83 is the displaced position
, @xmath84 is the radial direction from the focus or the virtual source ( icru-35 1984 ) of the mother beam , @xmath85 is the number of shared particles , @xmath86 is the conserved residual range , @xmath87 is the reduced spatial variance , and @xmath88 and @xmath89 conserve focal distance @xmath90 and local angular variance @xmath91 in splitting .
the mother beam splits into the daughter beams to form different detouring paths .
sets of the initial parameters for daughter beams are sequentially pushed on the stack of computer memory and the last set on the stack will be the first beam to be transported in the same manner , which will be repeated until the stack has been emptied before moving on to the next original beam .
an experiment to assess the present method was carried out with accelerator facility himac at national institute of radiological sciences .
a @xmath92 beam with nucleon kinetic energy @xmath93 mev was broadened to a uniform field of nominal 10-cm diameter by the spiral - wobbling method ( yonai 2008 ) . the horizontal wobbler at @xmath94 cm and the vertical wobbler at @xmath95 cm formed a spiral orbit of maximum 10-cm radius on the isocenter plane .
a 0.8-mm - thick pb ( @xmath96 , @xmath97 cm ) foil was placed at @xmath98 cm as a scatterer , which increased the instantaneous rms beam size from pristine 8.3 mm to 25 mm at the isocenter .
a large - diameter parallel - plate ionization chamber was placed at @xmath99 cm for dose monitoring and beam - extraction control .
an al ( @xmath100 , @xmath101 cm ) ridge filter for semi - gaussian range modulation of @xmath102 cm and @xmath103 cm in water ( schaffner 2000 ) and a 2-mm - thick al base plate were inserted at @xmath104 cm to moderate the bragg peak just to ease dosimetry . as shown in , a water ( @xmath105 , @xmath106 cm ) tank with a 1.9-cm - thick pmma ( @xmath107 , @xmath108 cm ) beam - entrance wall was placed at the irradiation site with the upstream face at @xmath109 cm .
the radiation field was defined by a 8-cm - square 5-cm - thick brass collimator whose downstream face was at 65 cm .
two identical 3-cm - thick pmma plates were inserted .
the downstream plate was attached to the beam - entrance face of the tank covering only the @xmath110 side to form a phantom system with a bump .
the upstream plate was put with its downstream face at @xmath111 cm covering only the @xmath112 side to compensate the bump .
such arrangement is typical for range compensation and sensitive to the detour effects ( kohno 2004 ) .
these beam - customization elements were manually aligned to the nominal central axis at an uncertainty of 1 mm .
a multichannel ionization chamber ( mcic ) with 96 vented sense volumes aligned at intervals of 2 mm along the @xmath113 axis was installed at @xmath114 in the water tank .
the mcic system was electromechanically movable along the @xmath115 axis and the upstream limit at @xmath116 cm was chosen for the reference point with reference depth @xmath117 cm of equivalent water from the tank surface . with a reference open field without the pmma plates or the collimator , we measured reference dose / mu reading @xmath118 at reference height @xmath119 for every channel @xmath120 for a calibration purpose .
every dose / mu reading @xmath121 of channel @xmath120 at height @xmath115 for any field is divided by corresponding reference reading @xmath118 to measure dose @xmath122 at position @xmath123 as @xmath124 where divergence - correction factor @xmath125 is to measure the doses in dose unit @xmath126 that would be the isocenter dose for the reference depth of the reference field .
we then measured reference - field doses @xmath127 in the phantom at varied @xmath115 positions , from which we get tissue - phantom ratio @xmath128 beam range @xmath16 with gaussian modulation was equated to the distal 80%-dose depth ( koehler 1975 ) @xmath129 cm as shown in ( a ) . with indications for the measurement ( @xmath130 ) and the reference and 80%-dose depths ( @xmath131 and @xmath132 ) and ( b ) effective density @xmath133 and ( c ) effective lateral density gradient @xmath134 distributions in gray scale at @xmath114 in the calculation model.,width=491 ] with the collimator and the pmma plates in place , lateral dose profiles were measured in the same manner with particular interest around @xmath135 3.3 cm , 6.8 cm , and 10.3 cm , where the bragg peaks were expected for the primary ions passing through none , either , and both of the pmma plates .
shows range loss @xmath136 and scattering @xmath137 for the beam - line elements , and the resultant contributions to source sizes @xmath138 and @xmath139 estimated by back projection to the sources .
the ridge filter with the base plate was modeled as plain aluminum of average thickness .
the scattering for the scatterer was estimated from measured beam size 25 mm quadratically subtracted by pristine size in the distance of 425 cm .
total range loss 2.10 cm was deduced from range 16.24 cm expected for @xmath93 mev carbon ions ( kanematsu 2008c ) and deficit 0.68 cm for the pristine beam may be attributed to minor materials in the beam line .
lccccc element & @xmath115 & @xmath136 & @xmath140 & @xmath141 & @xmath142 + pristine & & 0.68 cm & & 8.3 mm & 8.3 mm + scatterer & 425 cm & 0.46 cm & 5.5 mrad & 5.6 mm & 2.5 mm + ridge filter & 235 cm & 0.96 cm & 3.2 mrad & 9.3 mm & 7.5 mm + total & & 2.10 cm & & 13.7 mm & 11.5 mm + [ tab : contributions ] as described in [ sec_appendix ] , pencil beams were defined to cover the collimated field at intervals of @xmath143 mm on the isocenter plane , where the open field was assumed to have uniform unit fluence @xmath144 .
exact collimator modeling was omitted because we were interested in the density interface in the middle of the field .
the upstream pmma plate was modeled as a range compensator with range loss @xmath145 cm for @xmath112 or @xmath146 for @xmath147 , where the original beams were generated , followed by the range loss and scattering .
the phantom system comprised of the downstream pmma plate and the water tank was modeled as density voxels at grid intervals of @xmath148 mm for a 2-l volume of @xmath149 , @xmath150 , and @xmath151 .
figures [ fig : model](b ) and [ fig : model](c ) show the density and lateral heterogeneity distributions .
we carried out dose calculations with beam splitting enabled ( splitting calculation ) and disabled ( non - splitting calculation ) . in this geometry ,
the density interface at @xmath152 was almost parallel to the beams and only ones in the two nearest columns would split . to examine effectiveness and efficiency of this method with larger heterogeneity , a 3-cm diameter cylindrical air cavity at @xmath153 and two 1-cm diameter bone rods with density @xmath154 at @xmath155 and @xmath156
were added to the phantom in the calculation model .
we carried out splitting and non - splitting dose calculations of the same carbon - ion radiation to monitor changes in frequencies of splitting modes , number of stopped beams , total path length @xmath157 , total effective volume @xmath158 in the heterogeneous phantom , and computational time with a 2-ghz powerpc g5/970 processor by apple / ibm .
the splitting calculation could be more effective for protons because they generally suffer larger scattering .
we thus carried out equivalent dose calculations for protons with enhanced scattering angle by factor 3.61 in otherwise the same configuration including the tissue - phantom - ratio data .
shows the two - dimensional dose distributions measured in the carbon - ion beam experiment and the corresponding non - splitting and splitting calculations . shows their lateral profiles in the plateau and at depths for sub peak , main peak , and potential sub peak expected for particles that penetrated both , either , and none of the pmma plates .
a dip / bump structure was commonly formed along the @xmath152 line for lateral particle disequilibrium ( goitein 1978 ) .
there was actually a sub peak in the measurement and in the splitting calculation , while it was naturally absent in the non - splitting calculation .
the observed loss of the main - peak component was also reproduced by the splitting calculation .
the potential sub peak was barely noticeable only in the splitting calculation . by ( a ) measurement , ( b ) splitting calculation , and ( c ) non - splitting calculation.,width=491 ] by measurement ( @xmath130 ) , splitting calculation ( solid ) , and non - splitting calculation ( dashed ) at ( a ) @xmath135 14.8 cm ( plateau ) , ( b ) 10.3 cm ( sub peak ) , ( c ) 6.8 cm ( main peak ) , and ( d ) 3.3 cm ( potential sub peak).,width=491 ] shows details of the heterogeneous phantom and the dose distributions by splitting calculation for the carbon - ion and proton radiations .
the larger scattering for protons naturally led to the larger dose blurring .
shows the dose profiles at the main peak and where the heterogeneity effects were large by splitting and non - splitting calculations .
in addition to the loss of the main - peak component at @xmath159 , beam splitting caused some dose enhancement in the shoulders of the profiles especially for the carbon ions . of ( a ) density @xmath160 and ( b ) effective lateral density gradient @xmath161 in the calculation model and doses @xmath162 from ( c ) carbon - ion and ( d ) proton radiations calculated with splitting.,width=491 ] shows the statistical results , where the splitting effectively increased the carbon - ion and proton beams by factors of 27 and 25 in number , 20 and 25 in path length , 6.6 and 12 in volume , and 4.9 and 4.2 in total computation . lcccc projectile & & + beam splitting & no & yes & no & yes + frequency of @xmath75 & 0 & 0.243 & 0 & 3.813 + frequency of @xmath163 & 0 & 0.132 & 0 & 0.714 + frequency of @xmath164 & 0 & 1.636 & 0 & 0.967 + number of stopped beams & 1 & 26.8 & 1 & 25.0 + mean
path length / cm & 20.0 & 394.6 & 20.0 & 499.8 + mean effective volume/@xmath165 & 3.52 & 23.1 & 30.8 & 380.4 + computational time / s & 9.3 & 45.8 & 15.3 & 63.8 + [ tab : statistics ]
subdivision of a radiation field into virtual pencil beams is an arbitrary process in the pb algorithm although the beam sizes and intervals should be limited by lateral heterogeneity of a given system . in the pb - redefinition algorithm ( siu and hogstrom 1991 ) ,
beams are defined in uniform rectilinear grids and hence regeneration in areas with little heterogeneity may be potentially wasteful . in the beam - splitting method
, beams are automatically optimized in accordance with local heterogeneity . in other words ,
the field will be covered by minimum number of beams in a density - modulated manner as a result of individual independent self - similar splitting .
relative errors in similarity , @xmath166 ( @xmath31 ) , are maximum at @xmath152 with values @xmath167 , @xmath168 , and @xmath169 .
the resultant dose errors would be smaller , due to contributions of other beams , and may be tolerable .
effectiveness of the splitting method was demonstrated in the experiment .
the most prominent detour effect was the loss of range - compensated main - peak component in ( c ) , which amounted to about 10% in dose and approximately as large as the distortion due to lateral particle disequilibrium .
the splitting calculation and the measurement generally agreed well , considering that the experimental errors in device alignment could have been 1 mm or more .
the potential sub peak for particles detouring around both pmma plates was not detected , which may be natural because detouring itself requires scattering .
the dose resolution of the mcic system of about 1% of the maximum should have also limited the detectability . in the applications to the heterogeneous phantom model , although we do nt have reference data to compare the results with , it is natural that the splitting calculation with finer beams resulted in finer structures in the dose distributions .
computational time is always a concern in practice . in our example , the slowing factor for beam splitting with respect to non - splitting calculation was almost common to carbon ions and protons and the speed performance , a minute for 2-l volume in 1-mm grids , may be already acceptable for clinical applications . in principle , the total path length determines the computational amount for path integrals and the total effective volume determines that for dose convolution . their influences on the actual computational time will depend on algorithmic implementations ( kanematsu 2008a ) .
in fact , the slowing factor for splitting was less than 5 , which is even better than either estimation .
in addition to common overhead that should have superficially reduced the factor , our code optimization with algorithmic techniques , which will be reported elsewhere , could have contributed to the performance .
accuracy and speed also depend strongly on the cutoff parameters and logical conditions in the implemented algorithm , size and heterogeneity of a patient model , and resolution clinically needed for a dose distribution .
the automatic multiplication of tracking elements resembles a shower process in physical particle interactions usually calculated in mc simulations .
in fact , mc simulations for dose calculation share many things in common .
transport and stacking of the elements are essentially the same and the probability for scattering may be equivalent to the distribution in the gaussian approximation . as far as efficiency is concerned ,
the essential differences from the mc method are that the pb method deals with much less number of elements and that it does not rely on stochastic behavior of random numbers .
the beam - splitting method is based on a simple principle of self - similarity and can be applied to any gaussian beam model of any particle type to fill the gap between monte carlo particle simulations and conventional beam calculations in terms of accuracy and efficiency .
however , it is difficult for beam splitting or any beam model in general to deal with interactions that deteriorate particle uniformity , such as nuclear fragmentation processes ( matsufuji 2005 ) .
in this work , we applied our finding of self - similar nature of gaussian distributions to dose calculation of heavy charged particle radiotherapy .
the self - similarity enables dynamic , individual , and independent splitting of gaussian beams that have been grown larger than the limit from lateral heterogeneity of the medium . as a result
, pencil beams will be arranged with optimum and modulated areal density to minimize overreaching and to address detouring with deflecting daughter beams . in comparison with a conventional calculation and a measurement ,
the splitting calculation was prominently effective in the target region with steep range adjustment by an upstream range compensator .
the detour effect was about 10% for the maximum and of the same order of magnitude with lateral particle disequilibrium effect . in comparison between carbon ions and protons , the effects of splitting were not significantly different because other scattering effects were also larger for protons .
although performances depend strongly on physical beam conditions , clinical requirement , and algorithmic implementation , a typical slowing factor of the order of 10 may be reasonably achievable for involvement of beam splitting .
in fact , factor of 5 has been achieved in our example .
the principle and formulation for beam splitting are general and thus the feature may be added to various implementations of the pb algorithm in a straightforward manner .
on generation of pencil beam @xmath0 on a plane at height @xmath170 as shown in , beam position @xmath1 , residual range @xmath4 , and variances @xmath5 , @xmath6 , and @xmath7 are initialized as @xmath171 where @xmath172 is the beam-@xmath0 origin , @xmath173 is the beam position on the isocenter plane , @xmath138 and @xmath139 are the source sizes at virtual source heights @xmath174 and @xmath175 , @xmath16 is the initial residual range , and @xmath176 is the beam direction radiating from the virtual sources with @xmath177 because nuclear interactions are effectively handled in tissue - phantom ratio @xmath178 in dose calculation , number of particles @xmath3 is modeled as invariant .
the fermi - eyges theory ( eyges 1948 , kanematsu 2008c , 2009 ) gives increments of the pb parameters in step @xmath179 within a density voxel by @xmath180 \delta s , \label{eq : variance}\end{aligned}\ ] ] where @xmath160 and @xmath181 are the effective density ( kanematsu 2003 ) and radiation length of the medium in units of those of water and @xmath115 and @xmath182 are the particle charge and mass in units of those of a proton .
for the last physical step with @xmath183 and diverging @xmath184 , the growth is directly given by @xmath185 and then disabled by @xmath186 in the unphysical @xmath187 region .
kanematsu n , akagi t , futami y , higashi a , kanai t , matsufuji n , tomura h and yamashita h 1998 a proton dose calculation code for treatment planning based on the pencil beam algorithm _ jpn
phys . _ * 18 * 88103 schaffner b , pedroni e and lomax a 1999 dose calculation models for proton treatment planning using a dynamic beam delivery system : an attempt to include density heterogeneity effects in the analytical dose calculation 2741 schaffner b , kanai t , futami y , shimbo m and urakabe e 2000 ridge filter design and optimization for the broad - beam three - dimensional irradiation system for heavy - ion radiotherapy _ med . phys . _
* 27 * 71624 | the pencil - beam model is valid only when elementary gaussian beams are small enough with respect to lateral heterogeneity of a medium , which is not always the case in heavy charged particle radiotherapy .
this work addresses a solution for this problem by applying our discovery of self - similar nature of gaussian distributions . in this method ,
gaussian beams split into narrower and deflecting daughter beams when their size has exceeded the lateral heterogeneity limit .
they will be automatically arranged with modulated areal density for accurate and efficient dose calculations .
the effectiveness was assessed in an carbon - ion beam experiment in presence of steep range compensation , where the splitting calculation reproduced the detour effect of imperfect compensation amounting up to about 10% or as large as the lateral particle disequilibrium effect .
the efficiency was analyzed in calculations for carbon - ion and proton radiations with a heterogeneous phantom model , where the splitting calculations took about a minute and were factor of 5 slower than the non - splitting ones .
the beam - splitting method is reasonably accurate , efficient , and general so that it can be potentially used in various pencil - beam algorithms . |
biological evolution presents a rich array of phenomena that involve nonlinear interactions between large numbers of units . as a consequence ,
problems in evolutionary biology have recently enjoyed increasing popularity among statistical and computational physicists.@xcite however , many of the models used by physicists have unrealistic features that prevent the results from attracting significant attention from biologists . in this paper
we therefore develop and explore individual - based models of coevolution in predator - prey systems based on more realistic population dynamics than some earlier models.@xcite
recently the author , together with r. k. p. zia , introduced a simplified form of the tangled - nature model of biological macroevolution , which was developed by jensen and collaborators.@xcite in these simplified models,@xcite the reproduction rates in an individual - based population dynamics with nonoverlapping generations provide the mechanism for selection between several interacting species .
new species are introduced into the community through point mutations in a haploid , binary genome " of length @xmath1 , as in eigen s model for molecular evolution.@xcite the potential species are identified by the index @xmath2 $ ] .
( typically , only @xmath3 of these species are present in the community at any one time @xmath4 . ) at the end of each generation , each individual of species @xmath5 gives birth to a fixed number @xmath6 of offspring with probability @xmath7 before dying , or dies without offspring with probability @xmath8 .
each offspring may mutate into a different species generally with different properties with a small probability @xmath9 .
mutation consists in flipping a randomly chosen bit in the genome . in these models ,
the reproduction probability for an individual of species @xmath5 is given by the nonlinear function @xmath10 } \ ; , \label{eq : pi}\ ] ] where @xmath11 is an external resource that is renewed at the same level each generation , and @xmath12 is the set of population sizes of all the species resident in the community in generation @xmath4 .
the function @xmath13 is given by @xmath14 here @xmath15 is an energy cost " of reproduction ( always positive ) , and @xmath16 ( positive for primary producers or autotrophs , and zero for consumers or heterotrophs ) is the ability of individuals of species @xmath5 to utilize the external resource @xmath11 , while @xmath17 is an environmental carrying capacity@xcite ( a.k.a .
verhulst factor@xcite ) .
the total population size is @xmath18 .
the main feature of this reproduction probability is the random _ interaction matrix _
@xmath19,@xcite which is constructed at the beginning of a simulation run , and thereafter kept constant ( quenched randomness ) .
if @xmath20 is positive and @xmath21 negative , @xmath5 is a predator and @xmath22 its prey , and vice versa . if both matrix elements are positive , the relationship is a mutualistic one , while both negative indicate an antagonistic relationship .
two versions of this model were studied in earlier work . in the first version ,
which we have called model a , there is no external resource or birth cost , and the off - diagonal elements of @xmath19 are stochastically independent and uniformly distributed over @xmath23 $ ] , while the diagonal elements are zero .
this model evolves toward mutualistic communities , in which all species are connected by mutually positive interactions.@xcite of greater biological interest is a predator - prey version of the model , called model b. in this case a small minority of the potential species ( typically 5% ) are primary producers , while the rest are consumers .
the off - diagonal part of the interaction matrix is antisymmetric , with the additional restriction that a producer can not also prey on a consumer.@xcite in simulations we have taken @xmath15 and the nonzero @xmath16 as independent and uniformly distributed on @xmath24 $ ] .
this model generates simple food webs with up to three trophic levels.@xcite both of these models provide interesting results , which include intermittent dynamics with power spectral densities ( psds ) of diversities and population sizes that exhibit @xmath0-like noise , as well as power - law distributions for the lifetimes of individual species and the duration of quiet periods of relative evolutionary stasis . from a theoretical point of view
they also have the great advantage that the mean - field equation for the steady - state average population sizes in the absence of mutations reduces to a set of linear ( if @xmath25 ) or at most quadratic equations and thus can easily be solved exactly.@xcite the models thus provide useful benchmarks for more realistic , but generally highly nonlinear models . in fact , the population dynamics defined by eqs .
( [ eq : pi ] ) and ( [ eq : delta ] ) are not very realistic .
in particular , by summing over positive and negative terms in @xmath13 , the models enable species with little food to remain near a steady state if they are also not very popular as prey , or have very low birth cost .
a more serious problem is the _ ad - hoc _ nature of the normalization by the total population size @xmath26 in the resource and interaction terms in @xmath13 .
while this is the source of the models analytic solvability , it implies an indiscriminate , universal competition without regard to whether or not two species directly utilize the same resources or share a common predator .
the purpose of the present paper is to develop models with more realistic population dynamics and explore the properties of their evolutionary dynamics .
here we develop a model with more realistic population dynamics that include competition between different predators that prey on the same species , as well as a saturation effect expected to occur for a predator with abundant prey .
in doing so , we retain from the models discussed above the important role of the interaction matrix @xmath19 , as well as the restriction to dynamics with nonoverlapping generations .
we first deal with the competition between predator species by defining the number of individuals of @xmath22 that are available as prey for @xmath5 , corrected for competition from other predator species , as @xmath27 where @xmath28 runs over all @xmath1 such that @xmath29 , i.e. , over all predators of @xmath22 .
thus , @xmath30 , and if @xmath5 is the only predator consuming @xmath22 , then @xmath31 .
analogously , we define the competition - adjusted external resources available to a producer species @xmath5 as @xmath32 as in the case of predators , @xmath33 , and a sole producer species has all of the external resources available to it : @xmath34
. with these definitions , the total , competition - adjusted resources available for the sustenance of species @xmath5 are @xmath35 where @xmath36 runs over all @xmath22 such that @xmath37 , i.e. , over all prey of @xmath5 .
a central concept of the model is the _ functional response _ of species @xmath5 with respect to @xmath22 , @xmath38.@xcite this is the rate at which an individual of species @xmath5 consumes individuals of @xmath22 .
the simplest functional response corresponds to the lotka - volterra model:@xcite @xmath39 if @xmath40 and 0 otherwise .
however , it is reasonable to expect that the consumption rate should saturate in the presence of very abundant prey.@xcite for ecosystems consisting of a single pair of predator and prey , or a simple chain reaching from a bottom - level producer through intermediate species to a top predator , the most common forms of functional response are due to holling.@xcite for more complicated , interconnected food webs , a number of functional forms have been proposed in the recent literature,@xcite but there is as yet no agreement about a standard form . here
we choose a ratio - dependent@xcite holling type ii form,@xcite @xmath41 where @xmath42 $ ] is the metabolic efficiency of converting prey biomass to predator offspring .
analogously , the functional response of a producer species @xmath5 toward the external resource @xmath11 is @xmath43 in both cases , if @xmath44 , then the consumption rate equals the resource ( @xmath45 or @xmath46 ) divided by the number of individuals of @xmath5 , thus expressing intraspecific competition for scarce resources . in the opposite limit , @xmath47 ,
the consumption rate is proportional to the ratio of the specific , competition - adjusted resource to the competition - adjusted total available sustenance , @xmath48 .
the total consumption rate for an individual of @xmath5 is therefore @xmath49 the birth probability is assumed to be proportional to the consumption rate , @xmath50 \ ; , \label{eq : bi}\ ] ] while the probability that an individual of @xmath5 avoids death by predation until attempting to reproduce is @xmath51 the total reproduction probability for an individual of species @xmath5 in this model is thus @xmath52 .
we simulated the functional - response model over @xmath53 generations ( plus @xmath54 generations warm - up " ) for the following parameters : genome length @xmath55 ( @xmath56 potential species ) , external resource @xmath57 , fecundity @xmath58 , mutation rate @xmath59 , proportion of producers @xmath60 , interaction matrix @xmath19 with connectance @xmath61 and nonzero elements with a symmetric , triangular distribution over @xmath23 $ ] , and @xmath62 .
we ran five independent runs , each starting from 100 individuals of a single , randomly chosen producer species . by time
the dashed curves count the total number of species that have gone extinct after attaining a maximum population greater than 1000 .
the black dashed curve refers to all species , the light gray one ( green online ) to producers , and the dark gray one ( red online ) to consumers .
the ratio of approximately 1.89 between the dashed and full black curves indicate that major species recur on average about twice during the evolution .
this is an artifact of the finite genome length .
the inset shows the detailed , intermittent structure of the solid , black curve over @xmath63 generations .
the interval is indicated by a horizontal bar in the main panel .
, title="fig : " ] by time @xmath4 .
the dashed curves count the total number of species that have gone extinct after attaining a maximum population greater than 1000 .
the black dashed curve refers to all species , the light gray one ( green online ) to producers , and the dark gray one ( red online ) to consumers .
the ratio of approximately 1.89 between the dashed and full black curves indicate that major species recur on average about twice during the evolution .
this is an artifact of the finite genome length .
the inset shows the detailed , intermittent structure of the solid , black curve over @xmath63 generations .
the interval is indicated by a horizontal bar in the main panel .
, title="fig : " ] time series of diversities ( effective numbers of species ) and population sizes for one realization are shown in fig .
[ fig : timser ] . to filter out noise from low - population , unsuccessful mutations
, we define the diversity as the exponential shannon - wiener index.@xcite this is the exponential function of the information - theoretical entropy of the population distributions , @xmath64 $ ] , where @xmath65 with @xmath66 for the case of all species , and analogously for the producers and consumers separately .
the time series for both diversities and population sizes show intermittent behavior with quiet periods of varying lengths , separated by periods of high evolutionary activity . in this respect ,
the results are similar to those seen for models a and b in earlier work.@xcite however , diverse communities in this model seem to be less stable than those produced by the linear models . in particular , this model has a tendency to flip randomly between an active phase with a diversity near ten , and a `` garden of eden '' phase of one or a few producers with a very low population of unstable consumers , such as the one seen around 10 million generations in fig . [
fig : timser ] . a common method to obtain information about the intensity of fluctuations in a time series at different time scales
is the power - spectral density ( squared fourier transform ) , or psd .
psds are presented in fig .
[ fig : psd ] for the diversity fluctuations and the fluctuations in the population sizes ( fig .
[ fig : psd](a ) ) and the intensity of extinction events ( fig .
[ fig : psd](b ) ) .
the former two are shown for the total population , as well as separately for the producers and consumers .
all three are similar .
extinction events are recorded as the number of species that have attained a population size greater than one , which go extinct in generation @xmath4 ( marked as species " in the figure ) , while extinction sizes are calculated by adding the maximum populations attained by all species that go extinct in generation @xmath4 ( marked as population " in the figure ) .
the psds for all the quantities shown exhibit approximate @xmath0 behavior .
for the diversities and population sizes , this power law extends over more than five decades in time .
the extinction measures , on the other hand , have a large background of white noise for frequencies above @xmath67 generations@xmath68 , probably due to the high rate of extinction of unsuccessful mutants . for lower frequencies ,
however , the behavior is consistent with @xmath0 noise within the limited accuracy of our results .
( averaged over 16 generations ) falls continuously below some cutoff .
the inset is a histogram of @xmath69 , showing a gaussian center with approximately exponential wings .
the parabola in the foreground is a gaussian fit to this central peak .
the cutoff values for the main figure , between 0.008 and 0.024 , were chosen on the basis of this distribution .
the data in both parts of the figure are averaged over five independent simulation runs .
, title="fig : " ] ( averaged over 16 generations ) falls continuously below some cutoff .
the inset is a histogram of @xmath69 , showing a gaussian center with approximately exponential wings .
the parabola in the foreground is a gaussian fit to this central peak .
the cutoff values for the main figure , between 0.008 and 0.024 , were chosen on the basis of this distribution .
the data in both parts of the figure are averaged over five independent simulation runs .
, title="fig : " ] the evolutionary dynamics can also be characterized by histograms of characteristic time intervals , such as the time from creation till extinction of a species ( species lifetimes ) or the time intervals during which some indicator of evolutionary activity remains continuously below a chosen cutoff ( duration of evolutionarily quiet periods ) .
histograms of species lifetimes are shown in fig .
[ fig : time](a ) . as our indicator of evolutionary activity
we use the magnitude of the logarithmic derivative of the diversity , @xmath70 , and histograms for the resulting durations of quiet periods , calculated with different cutoffs , are shown in fig .
[ fig : time](b ) .
both quantities display approximate power - law behavior with an exponent near @xmath71 , consistent with the @xmath0 behavior observed in the psds.@xcite it is interesting to note that the distributions for these two quantities for this model have approximately the same exponent .
this is consistent with the previously studied , mutualistic model a,@xcite but not with the predator - prey model b.@xcite we believe the linking of the power laws for the species lifetimes and the duration of quiet periods indicate that the communities formed by the model are relatively fragile , so that all member species tend to go extinct together in a mass extinction . "
in contrast , model b produces simple food webs that are much more resilient against the loss of a few species , and as a result the distribution of quiet - period durations decays with an exponent near @xmath72.@xcite
( a ) . * ( b ) * time series of the number of new species that have reached a population greater than 1000 ( lower curve ) and greater than 100 ( upper curve ) .
the inset shows the intermittent structure of the upper curve on a very fine scale of 2000 generations .
see discussion in the text .
, title="fig : " ] ( a ) . *
( b ) * time series of the number of new species that have reached a population greater than 1000 ( lower curve ) and greater than 100 ( upper curve ) .
the inset shows the intermittent structure of the upper curve on a very fine scale of 2000 generations .
see discussion in the text .
, title="fig : " ] the model studied above is one in which species forage indiscriminately over all available resources , with the output only limited by competition . also , there is an implication that an individual s total foraging effort increases proportionally with the number of species to which it is connected by a positive @xmath20 .
a more realistic picture would be that an individual s total foraging effort is constant and can either be divided equally , or concentrated on richer resources .
this is known as adaptive foraging . while one can go to great length devising optimal foraging strategies,@xcite we here only use a simple scheme , in which individuals of @xmath5 show a preference for prey species @xmath22 , based on the interactions and population sizes ( uncorrected for interspecific competition ) and given by @xmath73 and analogously for @xmath11 by @xmath74 the total foraging effort
is thus @xmath75 .
the preference factors are used to modify the reproduction probabilities by replacing all occurrences of @xmath20 by @xmath76 and of @xmath16 by @xmath77 in eqs .
( [ eq : neff][eq : phiir ] ) .
the results of implementing the adaptive foraging are quite striking .
the system appears now to have a metastable low - diversity phase similar to the active phase of the non - adaptive model , from which it switches at a random time to an apparently stable high - diversity phase with much smaller fluctuations . as seen in fig .
[ fig : adap](a ) , the switchover is quite abrupt , and fig . [
fig : adap](b ) shows that it is accompanied by a sudden reduction in the rate of creation of new species . as seen in fig .
[ fig : adappsd ] , the psds for both the diversities and population sizes in both phases show approximate @xmath0 noise for frequencies above @xmath78 generations@xmath68 . for lower frequencies , the metastable phase shows no discernible frequency dependence , while for the stable phase , the frequency dependence continues at least another decade .
it thus appears that long - time correlations are not seen beyond @xmath79 generations for the metastable phase , and probably not beyond about @xmath80 generations for the stable one .
these observations are consistent with species - lifetime distributions for both phases ( not shown ) , which are quite similar to those for the non - adaptive model , but typically with cutoffs in the range of @xmath79 to @xmath80 generations , much shorter than the total simulation times .
noise for frequencies above about @xmath78 , but the psds appear to approach constant levels for the lowest frequencies .
, title="fig : " ] noise for frequencies above about @xmath78 , but the psds appear to approach constant levels for the lowest frequencies .
, title="fig : " ] in fact , the system can also escape from the low - diversity phase to total extinction , which is an absorbing state , and in some of our simulation runs we avoided this by limiting @xmath81 to less than 0.9 .
this restriction does not seem to have any effect on the dynamics in the high - diversity phase .
these results are preliminary , and it is possible that the high - diversity phase corresponds to a mutational meltdown
. more research is clearly needed regarding the effects of adaptive foraging in this model .
in this paper we have shown that very complex and diverse dynamical behavior results , even from highly over - simplified models of biological macroevolution .
in particular , psds that show @xmath0-like noise and power - law lifetime distributions for species as well as evolutionarily quiet states are generally seen .
this is the case , both in the analytically tractable , but somewhat unrealistic tangled - nature type models , and in the nonlinear predator - prey models based on the more realistic holling type ii functional response . particularly intriguing
is the appearance of a new , stable high - diversity phase in the latter type of model when adaptive foraging behavior is included . among the many questions about this new phase that remain to be addressed
is the structure of the resulting community food webs .
supported in part by u.s .
national science foundation grant nos.dmr-0240078 and dmr-0444051 and by florida state university through the school of computational science , the center for materials research and technology , and the national high magnetic field laboratory .
p. a. rikvold and r. k. p. zia , in _
computer simulation studies in condensed matter physics xvi _ , edited by d. p. landau , s. p. lewis , and h .- b .
schttler ( springer - verlag , berlin , 2004 ) , pp . 3437 .
r. k. p. zia and p. a. rikvold , j. phys .
a * 37 * , 5135 ( 2004 ) .
p. a. rikvold , in _ noise in complex systems and stochastic dynamics iii _ , edited by l. b. kish , k. lindenberg , and z. gingl ( spie , the international society for optical engineering , bellingham , wa , 2005 ) , pp .
148155 , e - print arxiv : q - bio.pe/0502046 .
v. sevim and p. a. rikvold , in _
computer simulation studies in condensed matter physics xvii _ , edited by d. p. landau , s. p. lewis , and h .- b .
schttler ( springer - verlag , berlin , 2005 ) , pp .
. v. sevim and p. a. rikvold , j. phys .
a * 38 * , 9475 ( 2005 ) .
n. d. martinez , r. j. williams , and j. a. dunne , in _ ecological networks : linking structure to dynamics in food webs _ , edited by m. pasqual and j. a. dunne ( oxford university press , oxford , 2006 ) , pp . | we explore the complex dynamical behavior of simple predator - prey models of biological coevolution that account for interspecific and intraspecific competition for resources , as well as adaptive foraging behavior . in long kinetic monte carlo simulations of these models
we find quite robust @xmath0-like noise in species diversity and population sizes , as well as power - law distributions for the lifetimes of individual species and the durations of quiet periods of relative evolutionary stasis . in one model , based on the holling type ii functional response , adaptive foraging produces a metastable low - diversity phase and a stable high - diversity phase . |
the _ real bruno functions _ are arithmetical functions @xmath6 , @xmath7 $ ] which characterize numbers by their rate of approximation by rationals . they have been introduced by j.c .
yoccoz @xcite ( cases @xmath8 and @xmath9 ) and then studied in a more general context in @xcite . for their relationship with arithmetical properties of real numbers ,
bruno s functions enter in a huge number of dynamical system problems involving small divisors , for instance in the problem of the _ stability of a fixed point _ of a holomorphic diffeomorphism of a complex variable ( the so called _ schrder siegel _ problem ) @xcite , in the _ schrder siegel _ problem in the _ gevrey _ setting in one complex variable @xcite and in some _ local conjugacy _ problems : _ semistandard map _
@xcite , _ analytic circle diffeomorphisms _ @xcite and some _ analytic area preserving annulus map _ including the _ standard map _ and some of its generalizations @xcite .
let us now concentrate on the @xmath0bruno function . ] .
@xmath10 is @xmath11periodic , even ( for this reason it is also called _ even bruno s function _ ) and verifies the _ functional equation _ :
@xmath12 the set @xmath13 is called the set _ of bruno s numbers _ : by it follows that @xmath14 is invariant under the action of the modular group @xmath15 .
the bruno function can be extended to rational numbers by setting @xmath16 when @xmath17 . using the _ continued fraction _
algorithm one can solve to obtain : @xmath18 where @xmath19 , @xmath20 , @xmath21 , @xmath22 and @xmath23 is the _
nearest integer continued fraction map_. in
[ ssec : contfrac ] we will give a brief account of useful facts concerning continued fractions . in @xcite the _ complex bruno function _ has been introduced complex bruno function_. in fact we will see at the end of section [ sec : cmplxbruno ] , that it is constructed `` following '' the gauss continued fraction algorithm . in this way we could also distinguish it from the _
@xmath0complex bruno function _ that we will introduce in section [ sec : cmplxbruno ] `` following '' the nearest integer continued fraction algorithm . ] , more precisely authors defined an analytic map @xmath24 , where @xmath25 is the _ upper poincar half plane _ , @xmath11periodic , which verifies a functional equation similar to the one for the @xmath26bruno function .
the boundary behavior of @xmath4 is given by ( see theorem 5.19 and 5.2.9 of @xcite ) : 1 .
let @xmath27 , then the imaginary part of @xmath28 tends to @xmath29 when @xmath30 and @xmath31 , whenever @xmath32 ; 2 .
@xmath33 is bounded on @xmath25 , its trace is continuous at irrational points and it has a jump of @xmath34 for @xmath35 .
in section [ sec : cmplxbruno ] we introduce an explicit formula for the @xmath0complex bruno function which corrects a small error in
page 836 , and gives more details than appendix a.4 of @xcite .
we will also give an algorithm to compute it numerically .
we already observed that the function @xmath10 is related to the stability problem of a fixed point of an analytic diffeomorphism of @xmath36 , in the rest of this section we will show this relation by describing the yoccoz result ( @xcite , chapter ii ) .
let @xmath37 and let us consider the _
quadratic polynomial _ @xmath38 .
the origin is a fixed point and we are interested in studying its stability .
if @xmath39 ( hyperbolic case ) then it follows from the results of poincar and koenigs that the origin is stable , whereas if @xmath40 ( parabolic case ) the origin is not stable .
let now consider @xmath41 and let @xmath42 be the conformal map which locally linearizes @xmath43 ( its existence is guaranteed by the poincar
koenigs results ) : @xmath44 where @xmath45 , and let us denote by @xmath46 the radius of convergence of @xmath47 .
one can prove that @xmath47 can be analytically continued to a larger set , the _ basin of attraction of @xmath48 _ : @xmath49 , but not to the whole of @xmath36 , and it has a unique singular point on its circle of convergence @xmath50 , which will be denoted by @xmath51 .
the function @xmath52 is called the _
yoccoz function_. yoccoz proved that @xmath5 has an analytic bounded extension to @xmath53 and moreover it can be obtained as limit of polynomials @xmath54 , uniformly over compact subsets of @xmath53 , where @xmath55 is the critical point of the quadratic polynomial .
being this extension not identically zero , by a classical result of fatou , the yoccoz function has radial limits almost everywhere , and the set @xmath56 for which @xmath57 has zero measure .
this means that the quadratic polynomial is linearizable ( @xmath58 ) for a full measure set of @xmath59 , but the proof does nt give any information on this set . when @xmath60 and @xmath61 is not a _ root of the unity _ , assuming @xmath62 , for some irrational @xmath63 , yoccoz proved ( @xcite , theorem 1.8 chapter ii ) that @xmath64 is linearizable if and only if @xmath65 , moreover there exists a constant @xmath66 , and for all @xmath67 a constant @xmath68 such that for all @xmath69 : @xmath70 we are then interested in studying the function @xmath71 and some `` natural '' questions arise ( @xcite 3.2 page 72 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( yoccoz s conjecture ) .
_ is the function @xmath72 bounded for @xmath73 ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ motivated by numerical results of @xcite and by some analytic properties of the real bruno function ( see the following remark [ rem:12hold ] and @xcite ) it has been conjectured that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( marmi moussa
yoccoz s conjecture ) . _
the function , defined on the set of bruno number , @xmath74 , extends to a @xmath0hlder continuous function on @xmath75 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ rem:12hold ] in @xcite authors proved a `` stability result '' for @xmath10 ( 4 page 285 ) .
let us rewrite the functional equation for the @xmath0bruno function as follow : @xmath76=-\log x \ , , \ ] ] if we add to the r.h.s .
a `` regular term '' @xmath77 , say @xmath78hlder continuous , and we call @xmath79 the solution of : @xmath80=-\log x + f(x ) \ , , \ ] ] then @xmath81 is @xmath0hlder continuous if @xmath77 is at least @xmath0hlder .
hence if we prove to `` reduce the strength of singularities '' is the main idea of the _
modular smoothing_. we refer to @xcite where authors describe the method and apply it to the critical function of the semistandard map . ]
that the function @xmath82-\log \omega$ ] is hlder continuous with exponent @xmath83 , for @xmath84 $ ] , then the conjecture holds . very recently x. buff and a. cheritat @xcite proved the yoccoz conjecture . whereas the marmi moussa
yoccoz conjecture is still open
. we will be interested in the following conjecture , equivalent to the one of marmi moussa yoccoz : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( marmi moussa
yoccoz s conjecture ) _ the analytic function , defined on the upper poincar half plane , @xmath85 , extends to a @xmath0hlder continuous function on @xmath86 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the aim of this paper is twofold
. first give more insight into the @xmath0complex bruno function and second to make a first step toward the understanding of the last conjecture .
our numerical results allows to conclude that @xmath87 is @xmath78hlder continuous and we obtain an estimate of the hlder exponent @xmath88 .
this gives us good numerical evidence that the marmi
yoccoz conjecture should be true .
the paper is organized as follows : in section [ sec : cmplxbruno ] we introduce the @xmath0complex bruno function and some results from number theory ( approximations of rationals by rationals ) , to obtain an algorithm to compute the complex bruno function . in section [ sec : yoccozf ] we explain how to calculate the yoccoz function and then , after a brief introduction of the littlewood paley theory in [ sec : lpm ] , used to test the hlder continuity , we present our results in section [ sec : numrel ] .
appendix [ sec : numcons ] collects some considerations related to technical aspects of our numerical test .
_ acknowledgments .
_ i am grateful to jacques laskar for putting at my disposal a large amount of cpu times on computers of _ astronomie et systmes dynamiques _ team at imcce paris .
the aim of this section is to introduce , starting from appendix a.4 of @xcite , a complex extension of the @xmath0real bruno function and to give an algorithm to compute it numerically .
let us consider @xmath89)$ ] , extended : @xmath26periodic , @xmath90 for all @xmath91 , and even @xmath92 for all @xmath93 $ ] , and then let us introduce the operator @xmath94 acting on such @xmath77 by : @xmath95 we remark that the functional equation can be rewritten as : @xmath96 let @xmath97 be the operators defined by : @xmath98 \text{branch $ m^+$ } \\ x f\left(m-\frac{1}{x}\right ) & x\in \left ( \frac{1}{m},\frac{1}{m-1/2 } \right ] \text{branch $ m^-$}\\ 0 & \text{otherwise , } \end{cases}\ ] ] then using the periodicity and the evenness of @xmath77 , we can rewrite as follows : @xmath99 to introduce the @xmath0complex bruno function we have to extend to complex analytic functions ; this is done @xcite by considering the complex vector space of holomorphic functions in @xmath100 $ ] , vanishing at infinity : @xmath101\right)$ ] ( which is isomorphic to the space of hyperfunctions with support contained in @xmath102 $ ] ) .
so , let @xmath103 be the hilbert transform of @xmath77 : @xmath104 then starting from we define the action of @xmath94 on @xmath103 as follows : @xmath105 where @xmath106 , @xmath107 and @xmath108 acts on @xmath101\right)$ ] by : @xmath109-\frac{ad - bc}{c } \varphi^{\prime}\left ( -\frac{d}{c}\right ) \ , .\ ] ] in the spirit of we want to consider @xmath110 acting on some @xmath111\right)$ ] , and to obtain a @xmath11periodic , `` even function '' . ]
, we will consider : @xmath112\varphi ( z - n ) \ , .\ ] ] let us introduce the operator @xmath113 defined by @xmath114 , then from and the relation : @xmath115 , we can expand @xmath116 in terms of matrices @xmath117 and @xmath118 , to obtain a sum of matrices of the form : @xmath119 , where @xmath120 , @xmath121 and @xmath122 , for @xmath123 .
let us set @xmath124 and for @xmath120 : @xmath125 and finally @xmath126 : the _ @xmath0monoid _ ( we left to
[ ssec:12monoid ] a more detailed discussion of this monoid and the reason of its name ) .
it remains to specify the `` good '' @xmath127\right)$ ] to apply , to have the wanted properties for @xmath4 .
this is done by considering the hilbert transform of the _ even , @xmath26periodic _ logarithm function defined in @xmath128 $ ] , namely : @xmath129 where @xmath130 is the _ dilogarithm _ function @xcite : the analytic continuation of @xmath131 , to @xmath132 .
we are now able to define the _
@xmath0complex bruno function _ to be : @xmath133 \varphi_{1/2}(z - n ) \ , .\ ] ] this formula defines complex bruno function and we omit it referring to @xcite for any details . ] an holomorphic function , defined in @xmath25 , @xmath11periodic , with an _ even imaginary part _ for @xmath134 $ ] , and then complex bruno s function instead of even complex bruno s function .
] an _ odd real part _
, on the same domain .
[ rem : fourcoeffb ] the @xmath0complex bruno function is @xmath26periodic and so we can consider its fourier series : @xmath135
. introducing the variable @xmath136 the bruno function is mapped into an analytic function , @xmath137 , defined in @xmath138 , which can be extended by continuity to @xmath53 .
its taylor series at the origin is @xmath139 , hence fourier coefficients of @xmath140 corresponding to negative modes are all identically zero .
moreover , because of the parity properties of @xmath141 and @xmath142 , its fourier coefficients are all purely imaginary , in fact : @xmath143 \ , dx \ , .\ ] ] the goal of the next sections will be to express in terms of a sum over a class of rational numbers in such a way we could give ( [ ssec : cmplxbruno ] ) an algorithm to compute it .
this will be accomplished thanks to a new characterization ( [ ssec:12monoid ] ) of the _ @xmath0monoid _
@xmath144 , after having introduced some results from number theory :
[ ssec : contfrac ] continued fraction theory and [ ssec : approx ] farey series .
we consider the so called _ nearest integer continued fraction _
algorithm of the parameter , so we will also call it @xmath0continued fraction . ] .
we state here some basic facts we will need in the following and we refer to @xcite for a more complete discussion .
let @xmath145 , then to each @xmath91 we associate a continued fraction as follows : @xmath146 and then inductively for all @xmath147 , as long as @xmath148 : @xmath149 we will use the standard compact notation to denote the continued fraction @xmath150 $ ] . from the definition it follows that @xmath151 and so @xmath152 .
[ rem : conv ] let @xmath153 $ ] be a finite continued fraction of length @xmath154 . then , whenever @xmath155 , we must also have @xmath156 , namely @xmath157 $ ] represents the same rational number that @xmath158 $ ] . moreover
a finite continued fraction can not contain a couple @xmath159 for any @xmath160 .
we recall , without proof some known results : * the continued fraction algorithm stops if and only if @xmath161 , this correspondence in bijective up to the standard convention of remark [ rem : conv ] ; * [ it : pnqn]for any positive integer @xmath162 ( or smaller than the length of the finite continued fraction ) the @xmath163 convergent is defined by : @xmath164 \ , , \ ] ] one can prove that @xmath165 and @xmath166 are recursively defined by : @xmath167 starting with @xmath168 , @xmath169 and @xmath170 . *
[ it : segnopnqn ] for all @xmath162 we have : @xmath171 .
let @xmath172 , the _ farey series _ @xcite of order @xmath162 is the set of irreducible fractions in @xmath173 $ ] whose denominators do not exceed @xmath162 $ ] which can be constructed by induction starting with : @xmath174 and then defining the @xmath175th element of @xmath176 , @xmath177 , by : @xmath178 the farey tree of order @xmath162 is clearly larger than the corresponding farey series and @xmath179 . ] : @xmath180 : ( p , q)=1 \text { and } q\leq n \}.\ ] ] the cardinality of @xmath181 is given by the euler @xmath182 function and so it is asymptotic to @xmath183 for @xmath162 large .
the farey series is characterized by the following two properties @xcite : [ thm:1carat ] let @xmath177 .
if @xmath184 and @xmath185 are two successive elements of @xmath181 , then : @xmath186 .
[ thm : thm2 ] let @xmath177 .
if @xmath185 , @xmath184 and @xmath187 are three successive elements ( in this order ) of @xmath181 , then : @xmath188 using an idea contained in the proof of theorem [ thm : thm2 ] given in @xcite , we construct an algorithm ( easily implementable on a computer ) which allows us to carry out for any @xmath189 the farey series of order @xmath162 . using proposition [ prop : farey2 ]
we will give a second algorithm to compute the farey series up to any given order @xmath162 , using the continued fraction development .
let @xmath190 , then the elements of @xmath181 , @xmath191 , are recursively defined by : @xmath192 where @xmath193 , starting with @xmath194 , @xmath195 and @xmath196 .
let @xmath197 .
because @xmath198 and @xmath199 are relatively prime we can always solve in @xmath200 the linear diophantine equation @xmath201 : let @xmath202 be a particular solution and let @xmath203 be the integer such that : @xmath204 , namely @xmath205 .
let us define @xmath206 and @xmath207 , then the following claims are trivial : @xmath208 is again a solution of the linear diophantine equation , @xmath209 and @xmath210 .
so @xmath211 . clearly @xmath212 and
we claim that it is the immediate successor of @xmath184 in @xmath181 . to obtain a constructive algorithm we must solve the linear diophantine equation ,
this is achieved by considering the element which precedes @xmath184 in @xmath181 : let us denote it by @xmath185 . a particular solution
is then given by @xmath213 , @xmath214 : from the previous result the element following @xmath184 is then given by : @xmath215 , @xmath216 , where @xmath217 .
to finish the algorithm we need two starting elements of @xmath181 apart of @xmath218 , but it is easy to realize that the first three elements of @xmath181 are @xmath218 , @xmath219 and @xmath220 , whenever @xmath190 .
we are now able to give a second algorithm to construct the farey series of order @xmath162 .
here is the idea : given an irreducible fraction @xmath221 , we compute its continued fraction development and then following two rules : _ truncate _ and _ subtract one _ , we obtain two new irreducible fractions in @xmath173 $ ] which will result to be the predecessor and the successor of @xmath184 in @xmath181 with @xmath222 .
[ prop : farey2 ] let @xmath223 and let @xmath224 be three successive elements of @xmath225 .
assume @xmath226 $ ] for some @xmath227 and let us define the rational numbers @xmath228 and @xmath229 as follows , then @xmath156 by remark [ rem : conv ] , and @xmath230 $ ] .
] : @xmath231 \quad \text{(truncate ) } \ , , \ ] ] and @xmath232 \quad \text{(subtract one ) } \ , .\ ] ] then if @xmath233 , we have @xmath234 whereas if @xmath235 we have the symmetric case , namely @xmath236 by , and we have : @xmath237 then : @xmath238 where we used the definition of @xmath184 with its finite continued fraction of length @xmath154 . finally : @xmath239 and similarly @xmath240 from which the proof follows easily . in this paragraph we will study the monoid @xmath144 of @xmath241 , introduced in and used in the construction of the @xmath0complex bruno function .
our aim is to show its relation with the nearest integer continued fraction : for this reason we call it _
@xmath0monoid_. we will prove that given @xmath242 we can `` fill '' the matrix @xmath243 in exactly two ways , such that it belongs to @xmath144 `` following the nearest integer continued fraction development '' .
[ prop : porp2p6 ] let @xmath242 , @xmath227 and assume @xmath244 $ ] to be the finite continued fraction of @xmath184 .
we claim that the matrices @xmath245 and @xmath246 , where the rational @xmath228 and @xmath229 have been defined in proposition [ prop : farey2 ] , are given by : @xmath247 where for @xmath248 , matrices @xmath249 and integer @xmath250 are defined by : @xmath251 before to prove the proposition we make the following : [ rem:2tipes ] it results @xmath252 for all @xmath175 , in fact whenever @xmath253 one has @xmath254 ( see remark [ rem : conv ] ) .
because @xmath255 the first couple @xmath256 can only be one of the following two : @xmath257 if @xmath258 $ ] or @xmath259 if @xmath260 .
let @xmath261 and let us introduce matrices @xmath262 and integers @xmath263 as in according to the continued fraction of @xmath184 .
then we claim that @xmath264 is equal to @xmath265 where @xmath266 $ ] .
this can be proved by induction ( use remark [ rem:2tipes ] to prove the basis of induction ) and then follows by putting @xmath267 . to prove it is enough to calculate : @xmath268 clearly matrices of type t belong to @xmath269 ( because @xmath270 ) , whereas those of type s belong to the monoid if and only if the continued fraction of the rational @xmath184 ends with a couple @xmath271 , in fact in this way the matrix @xmath272 is given by @xmath273 , where we used that @xmath274 ( because @xmath155 ) and @xmath275 for all @xmath276 . remark also that , if @xmath277 is of type t then it can not ends with @xmath278 , in fact this will implies a continued fraction ending with @xmath279 $ ] , but we know that this is impossible and so : or @xmath280 , @xmath281 and @xmath282 , either @xmath156 , @xmath270 and @xmath283 . with the following proposition
we will prove that @xmath269 is the union of matrices of type t and of type s with @xmath271 .
let us denote by @xmath284 the monoid of matrices of type t and @xmath285 those of type s , with @xmath271 .
[ prop:12monoidts ] @xmath286 .
clearly @xmath287 .
let us prove the other inclusion .
let @xmath120 , @xmath288 , @xmath289 , such that @xmath290 .
let us consider two cases , first one : @xmath291 and @xmath292 or @xmath293 and @xmath294 ; second case : @xmath291 and @xmath295 .
in the former case to @xmath277 we associate a continued fraction by introducing , for @xmath296 : @xmath297 @xmath298 $ ] represents some rational @xmath184 , let us define as before @xmath228 and then @xmath299 ( by proposition [ prop : porp2p6 ] , where we also defined @xmath250 s ) .
observe that @xmath300 to conclude @xmath301 .
the second case can be treated similarly .
now to @xmath277 we associate the continued fraction @xmath302 $ ] where , for @xmath303 : @xmath304 let @xmath302 $ ] be some rational @xmath184 , define as before @xmath305 $ ] , then by proposition [ prop : porp2p6 ] : @xmath306 and it belongs to @xmath285 . to end this section we introduce a third characterization of the @xmath0monoid , which corrects a small error in
a.4.4 page 836 of @xcite , and which will be useful to construct the numerical algorithm for the @xmath0complex bruno function .
let @xmath307 .
then @xmath277 belongs to @xmath269 if and only if @xmath308 , @xmath309 and @xmath310 , where @xmath311 .
the proof can be done by direct computation and we omit it .
we end this part with the following : [ rem : gaussmonoid ] in @xcite authors considered the complex bruno function constructed using the monoid @xmath312 : @xmath313 we recall that according to the gauss continued fraction algorithm we always have @xmath314 , we can then prove modified version of propositions [ prop : farey2 ] and [ prop : porp2p6 ] to conclude that @xmath312 is constructed `` following '' the gauss continued fraction algorithm : starting from @xmath223 , we complete the matrix @xmath315 into @xmath272 and @xmath316 , where @xmath229 and @xmath228 are obtained with the truncate and subtract operations acting on the gauss finite continued fraction of @xmath184 . using the results of the previous sections we are now able to give an algorithm to compute the @xmath0complex bruno function .
let us rewrite definition : @xmath317 \varphi_{1/2}(z - n ) \ , , \ ] ] where @xmath318 and the action @xmath319 has been defined in . from the previous sections we know that the sum over @xmath144 can be replaced by a sum over @xmath242 , @xmath320 , in such a way that to each @xmath184 we associate the matrix @xmath316 , and also @xmath272 whenever the continued fraction of @xmath184 ends with @xmath321 .
using the periodicity and the parity properties of @xmath4 we can restrict to @xmath322 $ ] . let us consider the contribution of some @xmath242 to @xmath4 .
because of the form of @xmath323 and of the action @xmath319 we remark that the larger is the denominator of the fraction , the smaller is its contribution to the sum ; moreover different rational numbers with the same denominator give comparable contributions , so we decide to order the rationals w.r.t . increasing denominators :
in other words _ according to the farey series_. a similar statement holds w.r.t .
the sum over @xmath11 : large @xmath162 s give small contribution to the sum .
we then introduce two _ cut off _ to effectively compute : @xmath324 denoting the largest order of the farey series considered and @xmath325 the largest ( in modulus ) @xmath326 which contributes to the sum over integers .
we refer to appendix [ sec : numcons ] to explain the role of this cut off . ]
. then the @xmath0complex bruno function can be numerically approximated by : @xmath327 where @xmath328 and the sum is restricted to fractions such that @xmath329 , ( being always @xmath330 ) .
this approximation can be made as precise as we want , by choosing @xmath324 and @xmath325 large enough , in fact can be obtained as double limit @xmath331 and @xmath332 . in appendix
[ sec : numcons ] we will give numerical results showing the convergence of varying the cut - off values , the convergence of @xmath333 to @xmath334 when @xmath30 and @xmath32 , and the @xmath34jumps of @xmath33 when @xmath335 , as proved in @xcite .
in figure [ fig : imbrunoc ] we show some plots of @xmath28 for fixed ( small ) values of @xmath336 and @xmath322 $ ] , whereas in figure [ fig : polareib ] we show two polar plots of @xmath337 .
the aim of this section is to briefly introduce the algorithm used to compute the yoccoz function , @xmath338 , introduced in
[ ssec : yoccozfunc ] .
let @xmath41 , let @xmath339 be the quadratic polynomial and let us introduce the polynomials : @xmath340 .
then we recall that the yoccoz function is the uniform limit , over compact subsets of @xmath53 , of @xmath341 . from and its original definition , @xmath342
, we get : @xmath343 for all integer @xmath162 . hence
to compute @xmath344 we need to know how close is @xmath345 to the identity , near zero and this can be done using some standard distortion estimates @xcite .
so for any fixed @xmath41 , we can find @xmath346 s.t .
@xmath347 is contained in some fixed disk on which we can apply the distortion estimate and then from compute an approximation to @xmath344 with a prescribed precision @xmath348 .
[ rem : parityfourcoeff ] let us observe the following facts .
assume @xmath349 , with @xmath350 fixed , and @xmath351 varying in @xmath352 and let us introduce @xmath353 , to stress the dependence on @xmath351 only .
then we claim that @xmath354 and @xmath355 .
the proof can be done as follows .
first remark that @xmath61 , as a function of @xmath351 , is mapped into @xmath356 , when @xmath357 ; then is enough to observe that polynomials @xmath358 verify , for @xmath190 , @xmath359 , namely : @xmath360 a similar statement holds for @xmath361 . using the @xmath11periodicity
, we consider the fourier series of @xmath362 and using an argument similar to the one of remark [ rem : fourcoeffb ] we conclude that all the fourier coefficients are real and zero for negative fourier modes .
clearly taylor s coefficients of @xmath344 coincide with fourier coefficients of @xmath362 .
figure [ fig : polaru ] shows some polar plots of @xmath363 , for different values of @xmath364 , whereas in figure [ fig : imreu ] real and imaginary parts of @xmath365 are given .
compare with figures [ fig : polareib ] and [ fig : imbrunoc ] .
let us conclude this section with the following remark .
[ rem : polinomi ] in figure [ fig : polarv ] we show some polar plots of the `` yoccoz function '' used in @xcite ( figure 2 , page 484 ) : they do nt look like our previous pictures . here is the reason .
they take the following quadratic polynomial @xmath366 , which can be conjugate to our choice , @xmath367 , using @xmath368 : @xmath369 let us denote by @xmath370 the yoccoz function for the polynomial @xmath371 , then we claim that : @xmath372 which explain completely the relation between figure [ fig : polaru ] and figure [ fig : polarv ] . because @xmath373 exhibits the same jumps at rationals as the real part of the complex bruno function does , we choose the quadratic polynomial in the form @xmath43 .
the aim of this section is to introduce the basic ideas and results of the _ littlewood paley _ theory , for a more complete discussion we refer to @xcite and also to @xcite where authors apply this theory to study the regularity properties of the conjugating function for _ critical circle maps_. in [ ssec : lpm ] we will present the numerical implementation of this theory to study the hlder regularity of the function @xmath87 and the obtained estimate for the hlder exponent . the decay rate of the coefficients of a _
trigonometric series _
, @xmath374 , does not determine whether this series is the _
fourier series _ of some @xmath375 function if @xmath376 .
more precisely given @xmath377 , @xmath378 and its fourier series @xmath379 , then for `` almost every choice of signs @xmath380 '' , the series @xmath381 is not the fourier series of a @xmath375 function .
this problem has been overcome by littlewood and paley by `` grouping together '' trigonometric coefficients in _
dyadic blocks_. let @xmath382 , @xmath383 and , for @xmath384 , let @xmath385 , be the _ dyadic partial sum _ of @xmath77 . introducing the _ littlewood paley d function _ :
@xmath386^{1/2}$ ] , one can prove @xcite that for all @xmath387 there exist positive constants @xmath388 and @xmath389 such that : @xmath390 the littlewood paley theory is indeed more general , allowing to characterize other functional spaces by property of fourier coefficients , for instance it applies @xcite to sobolev spaces , hardy spaces , hlder spaces and besov spaces . in the case of hlder regularity
one can easily realize that fourier coefficients of an @xmath78hlder continuous function decay according to @xmath391 , the converse is not true but again the littlewood
paley theory can characterize the hlder regularity by the decay rate of the dyadic blocks .
an important tool in the theory of fourier series is the _ poisson kernel _ : @xmath392 , @xmath393 and @xmath394 .
let @xmath395 be the convolution product for @xmath26periodic functions
. then one can prove the following result ( @xcite lemma 5 or @xcite theorem 15.6 ) [ thm : clp ] let @xmath396 , @xmath397 and @xmath77 be a continuous @xmath26periodic function .
then @xmath77 is @xmath78hlder continuous if and only if there exists @xmath398 such that for all @xmath350 : @xmath399 where @xmath400 .
we remark that if the theorem holds for some @xmath401 , then the same is true for any @xmath402 , @xmath403 .
we call this theorem _ continuous littlewood
paley _ to distinguish it from the following result , which is more close to the original idea of dyadic decomposition and we will call it _ discrete littlewood paley _
( see @xcite theorem 5.9 ) [ thm : dlp ] let @xmath404 and let @xmath405 .
then @xmath77 is @xmath78hlder continuous function if and only if for all @xmath382 there exists a positive constant @xmath406 such that for all @xmath407 we have : @xmath408 one usually take @xmath409 , and so the name dyadic decomposition , but the result is independent of the value of @xmath410 . in the numerical implementation of this method we will use a value @xmath410 close to @xmath411 for computational reasons .
this section collects our numerical results about the marmi moussa yoccoz conjecture that we recall here : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the analytic function , defined on @xmath25 : @xmath85 , extends to a @xmath0hlder continuous function on the closure of @xmath25 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let us begin with some consideration concerning @xmath87 .
remark [ rem : fourcoeffb ] and remark [ rem : parityfourcoeff ] imply that @xmath412 has _ even real part _ for @xmath413 $ ] and an _ odd imaginary part _ on the same domain .
moreover @xmath412 is @xmath26periodic and its fourier series has only real coefficients , which correspond to non negative fourier modes : @xmath414 . in figure [
fig : immiblogu ] we plot real and imaginary parts of @xmath412 for some fixed small @xmath336 ; remark that @xmath87 still has a `` structure '' but jumps of @xmath4 and @xmath415 seem to `` compensate '' to give a _
continuous function_. the same fact holds for the `` bubbles '' of @xmath416 and @xmath417 , figure [ fig : polaremibu ] show some polar plots of @xmath418 for fixed small @xmath419 , there are still some `` bubbles '' but they are far from @xmath420 .
the hlder continuity will be proved in the next paragraph , by giving an estimate of the hlder exponent applying the _ littlewood paley theory _ in the _ discrete _ and _ continuous _ versions . to numerically implement the littlewood
paley theory we compute from the numerical values of @xmath87 , a finite but large , number of fourier coefficients . then to apply theorem [ thm : clp ] we remark that the convolution with the poisson kernel and the @xmath203th derivative have the form : @xmath421 where we used the previous remark on the fourier coefficients of @xmath87 .
we numerically compute @xmath422 for several small values of @xmath423 and some @xmath424 , then applying a linear regression over the data : @xmath425 we obtain a numerical value for @xmath426 . from a numerical point of view the continuous version of the littlewood paley method is better than the discrete one , in fact the former has two parameters to vary @xmath423 and @xmath203 . we can vary @xmath203 to control if the computed value of @xmath427 stays constant or not
. moreover we can compute the l.h.s . of for many values of @xmath423 and for each one all the known fourier coefficients
are used , whereas in the dyadic decomposition to `` small '' @xmath428 only `` few '' fourier coefficients give their contribution and only for `` large '' @xmath428 a large number of fourier coefficients enter . in figure
[ fig : clp ] we report data from and the corresponding linear regression values .
the estimated values of @xmath78 obtained for different @xmath203 are : @xmath429 , @xmath430 and @xmath431 ( errors are standard deviation errors of linear regression ) .
they agree in the numerical precision and this gives a good indication of the validity of the results .
there is no reason to prefer one value to the other and so we estimate @xmath432 : the mean value of the interval obtained by the union of the three intervals obtained for @xmath433 .
we also report the numerical results obtained using the discrete littlewood paley theorem .
we fix some @xmath382 is fixed in such a way we can take @xmath428 sufficiently large to have a good asymptotic , even if we have a finite number of fourier coefficients . ] and from the computed fourier coefficients of @xmath87 we construct the dyadic partial sums for some large @xmath434
. then we use a linear regression on the data : @xmath435 to obtain the estimate value of the hlder coefficient : @xmath436 and @xmath437 . in figure
[ fig : dlp ] we report data from and the linear regression applied on `` large @xmath428 '' .
we conclude this paper by summarizing the obtained results .
we introduced the @xmath0complex bruno function and the yoccoz function , both with an algorithm to evaluate them numerically .
we studied the function @xmath438 defined on the upper poincar plane , and we conclude that it can be extended to its closure , with a trace @xmath78hlder continuous .
numerical results based on the littlewood paley theory give us the estimated value for the hlder exponent : @xmath439 with an error of @xmath440 .
we can then conclude , with a good numerical evidence , that the marmi moussa
yoccoz conjecture should hold with the maximal exponent @xmath0 .
the aim of this appendix is to consider in details some technical parts of our numerical calculations .
we will consider the role of the cut off and their relations with the accuracy of the computations .
we will also compare the numerical properties of @xmath4 with the analytical ones proved in @xcite .
let us recall the formula defining the @xmath0complex bruno function : @xmath317 \varphi_{1/2}(z - n ) \
, , \ ] ] as already observed we need to introduce three cut off to compute it : @xmath324 , @xmath325 and @xmath441 . the first one determine the largest farey series involved , namely only fractions @xmath184 s.t .
@xmath442 , @xmath320 and @xmath443 will be considered to compute @xmath4 .
the other two cut off : @xmath444 , are introduced to truncate the sum over @xmath11 . because the larger is @xmath199 , the smaller is its contribution to @xmath4 , to gain cpu times we decide to truncate the sum over @xmath11 at @xmath445 if @xmath199 is `` small '' , and to @xmath446 if @xmath199 is `` large '' .
results showed in section [ sec : numrel ] are obtained with @xmath447 , @xmath448 and @xmath449 .
in the rest of this paragraph we will study the dependence of the computed bruno function on the cut off .
let us fix two cut off but one , call it generically @xmath428 , and let us stress the dependence of @xmath4 on it , by setting @xmath450 .
we are then interested in studying the relative error : @xmath451 , where @xmath452 is fixed and @xmath140 is numerically computed with some fixed large cut off : @xmath453 , @xmath448 and @xmath449 . or we can consider @xmath454 the mean value of @xmath455 for @xmath322 $ ] and some fixed value of @xmath364 . in table [ tab : sup ] we report values of @xmath456 , for @xmath457 . whereas in figure [ fig : cutoff ] we show , @xmath458 , for @xmath459 and @xmath460 and @xmath461
. clearly the larger are the cut off the more accurate are the results , but recall that large cut off implies large cpu times ; in particular the cpu times increases almost linearly w.r.t . @xmath325 and @xmath441 , but quadratically w.r.t . @xmath324 . | we study the @xmath0complex bruno function and we produce an algorithm to evaluate it numerically , giving a characterization of the monoid @xmath1 .
we use this algorithm to test the marmi
moussa
yoccoz conjecture about the hlder continuity of the function @xmath2 on @xmath3 , where @xmath4 is the @xmath0complex bruno function and @xmath5 is the yoccoz function . we give a positive answer to an explicit question of s. marmi et al @xcite . |
the standard model ( sm ) requires a higgs boson to explain the generation of fermion and gauge boson masses .
the precise electroweak ( ew ) measurements at lep suggest that the higgs boson must be relatively light@xcite , with @xmath0 gev .
a light higgs is also required in order to unitarize the longitudinal @xmath1 scattering amplitude@xcite .
in addition , the triviality bound indicates that the sm with a light higgs can be a theory valid all the way up to the planck scale .
this simple assumption of the sm with a single higgs doublet , however , has the theoretical problem that the higgs boson mass is quadratically sensitive to any new physics which may arise at high energy scales . to cancel these quadratic divergences due to the sm particles , a set of new states , which are related to the sm particles by some symmetry , have to appear at the tev scale .
little higgs ( lh ) models are a new approach to stabilizing the mass of the higgs boson .
these models have an expanded gauge structure at the tev scale which contains the standard model @xmath2 electroweak gauge groups .
they are constructed such that there are multiple global symmetries that prohibit the higgs boson from obtaining a quadratically divergent mass .
it is only when _ all _ these global symmetries are broken can a quadratic contribution to the scalar potential arises , which is at least at the two loop order .
there is generally a tension between the solution to the gauge hierarchy problem and the ew precision fit , nevertheless .
the deviations from the sm predictions due to the presence of new interactions above the sm cutoff scale , @xmath3 , can be parameterized by a set of higher dimensional operators which have been classified in ref . .
among these operators , the most stringent bounds are those on the coefficients of the dim-6 operator , @xmath4 , which breaks the custodial su(2 ) symmetry , and @xmath5 , and thus contributes to the s - parameter .
these bounds indicate that the cutoff scale @xmath3 has to be above 5 tev .
thus the form of possible new physics , including the little higgs models , which have to appear at the tev scale to solve the gauge hierarchy problem , is severely constrained .
this review is organized as follows . in sec .
[ intro ] , i introduce the basic idea of the little higgs models , show explicitly how the quadratic divergences are cancelled in various sectors of the littlest higgs model , and briefly review other existing little higgs models .
[ preew ] is devoted to the precision electroweak constraints in these models , followed by sec .
[ other ] in which other issues such as implicit fine - tuning and uv completion are briefly discussed .
[ concl ] concludes this review .
the idea of higgs boson being a pseudo - goldstone boson arising from the breaking of some approximate global symmetry was proposed@xcite in the early 80 s .
because the higgs boson mass is generated radiatively , this therefore provides a natural way to understand why the higgs boson is so light . in its early realizations , the quadratic contributions to the higgs
boson mass arise at one - loop , leading to a higgs mass that is still too heavy because it is only suppressed by the one - loop factor .
the new ingredient in the little higgs models is the so - called collective symmetry breaking .
the idea is to choose the gauge and yukawa interactions in such a way that by turning off some part of these interactions , the model has enhanced global symmetries that forbid a quadratic contribution to the higgs mass . as a result
, the quadratic contributions to the higgs mass can arise only when two or more operators are involved , which can occur only at the two loop level or beyond .
this leads to a higgs mass @xmath6 of the order of @xmath7 .
note that , the logarithmic contributions to the higgs mass can still appear at one - loop .
these models have the following general structure : at the electroweak scale , @xmath8 gev , there are one or two higgs doublets and possibly a few additional scalar fields ; at the scale @xmath9 tev , there exist new gauge bosons and fermions ; above the cut - off scale of the non - linear sigma model , @xmath10 tev , the model becomes strongly interacting .
various realizations of the little higgs idea are described below .
the minimal realization of the little higgs idea is the littlest higgs model@xcite , which is a non - linear sigma model based on @xmath11 .
the @xmath12 global symmetry in the model is broken down to @xmath13 by the vev of the sigma field , @xmath14 , which transforms as an adjoint under @xmath12 , where , @xmath15 the sigma field can be expanded around the vev in terms of the goldstone modes , @xmath16 , as , @xmath17 where @xmath18 correspond to the broken su(5 ) generators .
the gauge subgroup is chosen to be @xmath19_{1 } \times \left [ su(2 ) \times u(1 ) \right]_{2}$ ] .
it is broken down to its diagonal subgroup , @xmath20_{\mbox{\tiny sm}}$ ] , which is identified as the sm gauge group .
the kinetic term of the sigma field can be written as @xmath21 \ ; , \ ] ] where the covariant derivative is , @xmath22 \ ; .\ ] ] the generators of the two @xmath23 gauge groups , @xmath24 and @xmath25 for @xmath26 , are , @xmath27 and the two u(1 ) generators , @xmath28 and @xmath29 , are , @xmath30 when the gauge coupling constants @xmath31 and @xmath32 are turned off , a global symmetry @xmath33 , which acts on the last three indices , is restored , @xmath34 similarly , when @xmath35 and @xmath36 are turned off , there is an enhanced @xmath37 global symmetry acting on the first three indices , @xmath38 each of these two @xmath39 symmetries individually forbids a quadratic contribution to the higgs potential . thus a quadratic contribution to the higgs potential can arise only when both global symmetries are broken , which is possible only at the two loop level .
the 14 goldstone bosons resulting from the breaking of the global symmetry can be decomposed in the following way , @xmath40 where the subscripts in the above equation denote the hypercharges .
the components @xmath41 and @xmath42 are eaten and become the longitudinal degrees of freedom of the heavy gauge bosons , @xmath43 , @xmath44 and @xmath45 . in the low energy spectrum , there are two scalar fields : the component @xmath46 is identified as the complex doublet higgs boson , @xmath47 , of the sm , while @xmath48 is an additional complex @xmath49 triplet higgs , @xmath50 . in terms of these low energy degrees of freedom , the @xmath51 field can be written as , @xmath52 where @xmath53 expanding the kinetic terms of the @xmath54 field in eq .
( [ kin ] ) , one finds , @xmath55 \biggr|^{2 } \\ & \rightarrow & \frac{f^{2}}{8 } \bigg\ { \biggl [ g_{1}^{2 } w_{1 } w_{1 } - 2 g_{1}g_{2 } w_{1 } w_{2 } - 2 g_{1 } g_{2 } w_{1 } w_{2 } + g_{2}^{2 } w_{2 } w_{2 } \biggr ] \nonumber\\ & & \quad + \frac{1}{5 } \biggl [ g_{1}^{\prime 2 } b_{1 } b_{1 } - 2 g_{1}^{\prime } g_{2}^{\prime } b_{1 } b_{2 } + g_{2}^{\prime 2 } b_{2 } b_{2 } \biggr ] \bigg\ } \ ; . \nonumber\end{aligned}\ ] ] this gives the following mass matrices for the gauge bosons , @xmath56 and the mass eigenstates of these mass matrices are @xmath57 where the mixing angle @xmath58 is @xmath59 and @xmath60 while @xmath61 and @xmath62 .
the two massless eigenstates , @xmath63 and @xmath64 are identified as the weak gauge bosons in the sm .
the two massive eigenstates , @xmath45 and @xmath65 , are the additional gauge bosons having masses of the order of @xmath66 .
the gauge coupling constants of the unbroken subgroup , @xmath49 and @xmath67 , are given by , @xmath68 the quartic couplings of the higgs boson to the gauge bosons arise from the next - to - leading order terms in the expansion in eq .
( [ kin ] ) , @xmath69\bigg|^{2 } \\ & \rightarrow & \frac{1}{4 } ( g_{1 } g_{2 } w_{1 } w_{2 } + g_{1}^{\prime } g_{2}^{\prime } b_{1 } b_{2 } ) h^{\dagger } h + ... \nonumber\\ & = & \frac{1}{4 } \biggl [ g^{2 } ( w_{l}w_{l}-w_{h } w_{h } ) + g^{\prime 2 } ( b_{l}b_{l}-b_{h } b_{h})\biggr ] h^{\dagger } h + ..... \nonumber\end{aligned}\ ] ] thus the quartic couplings @xmath70 and @xmath71 are of equal magnitude but opposite signs .
the opposite signs come about because the @xmath63 and @xmath45 gauge bosons are orthogonal to each other .
the cancellation of quadratic divergences among diagrams shown in fig .
[ gaugeloop ] at one loop thus ensues . in the fermion sector , to cancel the top loop contribution to the radiative corrections to the higgs boson mass , one needs to introduce a vector - like pair of the color triplet and iso - singlet heavy tops , @xmath72 and @xmath73 .
the field @xmath72 then form a triplet , together with @xmath74 , under the @xmath33 global symmetry , @xmath75 .
the yukawa interactions take the following form , @xmath76 the first term in this yukawa lagrangian preserves the @xmath33 global symmetry and breaks the @xmath37 global symmetry , while the mass term of the vector - like quarks preserves the @xmath37 and breaks @xmath33 . due to the @xmath33 global symmetry ,
the couplings of @xmath77 to @xmath78 and @xmath79 , and the quartic coupling @xmath80 are related , @xmath81 u_{3}^{\prime } + h.c . \ ; .\ ] ] these relations lead to the cancellation of the quadratic divergences among diagrams shown in fig .
[ fermionloop ] in the fermion sector at one - loop .
because of the non - abelian transformation , @xmath82 , the quadratic contributions to @xmath83 can only arise at the two - loop level , due to the following scalar and fermion interactions , @xmath84 + g_{j}^{\prime 2 } \mbox{tr } \biggl [ \biggl(y_{j } \sigma\biggr ) + \biggl ( y_{j } \sigma\biggr)^{\ast}\biggr]\bigg\ } \\ \mathcal{l}_{f } & = & -\frac{a^{\prime}}{4 } \lambda_{1}^{2 } f^{4 } \epsilon^{wx } \epsilon_{yz}\epsilon^{ijk } \epsilon_{kmn } \sigma_{iw}\sigma_{jx}\sigma^{\ast my}\sigma^{\ast nz } \ ; , \end{aligned}\ ] ] where the parameters @xmath85 and @xmath86 are the coefficients that parametrize the unknown uv physics .
the scalar potential arises by integrating out the heavy top , @xmath87 , as well as the heavy gauge bosons , @xmath88 where @xmath89 + 8 a^{\prime } \lambda_{1}^{2 } \\ \lambda_{h\phi h } & = & -\frac{a}{4 } \biggl [ \frac{g^{2}(c^{2}-s^{2}}{s^{2}c^{2 } } + \frac{g^{\prime 2}(c^{\prime 2}-s^{\prime 2})}{s^{\prime 2 } c^{\prime 2 } } \biggr ] + 4 a^{\prime } \lambda_{1}^{2 } \ ; .\end{aligned}\ ] ] the complete feynman rules of the littlest higgs model has been presented in ref . .
various collider phenomenology of the littlest higgs model has been discussed extensively in ref . , and
the flavor sector in this model has been studied in ref . , including the generation of fermion masses@xcite .
as we will see in the next section , the littlest higgs model is severely constrained by the precision electroweak data .
the most stringent one comes from the tree level contributions to the @xmath90 parameter , due to the presence of the @xmath91 coupling and the tri - linear @xmath92 coupling , which breaks the tree level custodial symmetry explicitly . in ref .
, cheng and low found that these operators can be forbidden by imposing a discrete @xmath93 symmetry , called the t - parity . under the t - parity ,
all new particles , except the heavy top partner , @xmath72 , that are responsible for canceling the sm contributions to the one - loop quadratic divergences in the higgs potential are odd , while all the other particles are t - even . as a result ,
the heavy particle contributions to the observables involving only the sm particles in the external states are forbidden at the tree level .
these contributions of the new particles to precision ew observables can arise only at the loop levels in this model , and thus the constraints in this model is not as stringent as in the littlest higgs model .
due to t - parity , all t - odd particles have to be pair - produced , and the lightest stable particle can be the candidate of the weak scale dark matter@xcite . various collider phenomenology and flavor constraints in this model has been discussed in ref . and , respectively .
it was found that the collider signatures in this model mimic that of the mssm . in ref .
, the global symmetry of the model is chosen to be @xmath95 which is broken down to @xmath96 .
this has the advantage that the pseudo - goldstone multiplet does not contain a su(2 ) triplet component .
this thus weakens the constraints from precision data .
the gauged subgroup of @xmath95 is @xmath97^{2}$ ] , which is broken down to its diagonal subgroup , @xmath97_{\mbox{\tiny sm}}$ ] .
there are fourteen goldstone bosons @xmath98 resulting from the global symmetry breaking .
four of them are eaten due to gauge symmetry breaking .
thus at low energy spectrum , there are two complex doublet scalar fields . at the tev scale , there are one complex singlet scalar , a few pairs of vector - like colored fermions and an extra copy of the @xmath2 gauge bosons . by turning off either one of the su(2 ) gauge coupling constants , there is an enhanced su(4 ) global symmetry which forbids a square mass term for the higgs .
thus the quadratic divergences to the higgs mass can only arise at two loop .
the issue of vacuum stability in the presence of the anti - symmetric condensate of the @xmath54 field in this model has also been investigated@xcite . in the model proposed in ref . , the global symmetry is @xmath99^{4}$ ] which breaks down to @xmath100^{4}$ ] . and
the gauge subgroup is @xmath101 which breaks down to @xmath2 .
there are @xmath102 nbg s : 12 of them are eaten due to gauge symmetry breaking and the remaining 16 real components are decomposed into two complex doublets , three complex @xmath23 singlets and two real singlet scalars . by embedding the electroweak @xmath49 gauge group into a simple group such as @xmath39 or su(4 ) , this model has a nice feature that the cancellation of one - loop quadratic divergences from the gauge and perturbatively coupled fermion loops is automatic .
generation of neutrino masses in this model has been discussed in ref . .
the coset of the model proposed in ref .
is @xmath103 while the electroweak gauge symmetry of this model is expanded to @xmath104 which is embedded into @xmath105 . due to the enlarged gauge group , @xmath104
, there is no mixing induced by the vev of the higgs boson between the light and the heavy gauge bosons in this model , and it has only one additional gauge boson , the @xmath106 . in moose models ,
the electroweak sector of the sm is embedded into a theory with a product global symmetry , @xmath107 , which is broken by a set of condensates transforming as bi - fundamental representations under @xmath108 for pair ( i , j ) .
some subgroup of @xmath107 which contains the sm is being gauged for each site ; it is then broken down to the sm gauge group @xmath2 by the bi - fundamental condensate at the tev scale . the minimal moose model proposed in ref . has @xmath100^{8}$ ] as global symmetry .
the gauge subgroup of the model is @xmath109 $ ] .
the spectrum at low energy contains two complex higgs doublets , a complex triplet and a singlet . by enlarging the global group in the minimal moose model to @xmath110^{8}=[so(5)_{l}]^{4}\times[so(5)_{r}]^{4}$ ] and the gauge group to @xmath111 , it has been shown in ref .
that the model preserves an approximate custodial @xmath23 symmetry . as a generic feature of the moose models , it is a two - higgs doublet model at low energy . at the tev scale , there is a colored dirac fermion , a triplet higgs and extra gauge bosons in the spectrum .
the mass splitting among the triplet components give large contributions to the @xmath90 parameter , which occurs in the @xmath13 moose model .
this problem can be alleviated by expanding the global symmetry to @xmath112 as shown in ref . .
in the @xmath112 moose model , the coset is @xmath113 $ ] , while the embedded gauge symmetry is @xmath114 . at the tev scale , there are three triplet higgses present whose vevs preserve an approximate custodial symmetry .
this is an essential feature that reduces the contributions to the @xmath90 parameter from the scalar sector in this model .
as the electroweak sector of the sm has been tested to a very high accuracy , an important test of the validity of any new models is through the agreement between the predictions of these models with the precision data .
a sm - like renormalization procedure with three input parameters in the gauge sector is valid as long as the models has tree level custodial symmetry and thus @xmath115 at three level .
examples of models with @xmath116 at tree level include the sm augmented by additional higgs doublets or singlets , additional fermion families and mssm . on the other hands , in many extension of the sm , the tree level custodial symmetry is no longer a good symmetry of the model , _
i.e. _ @xmath117 already at the tree level . models of this type include the left - right symmetric model based on @xmath118 , models with additional @xmath49 triplet higgses ( which might be relevant to generation of neutrino masses@xcite ) , and various little higgs models , to name a few . in the gauge sector of the sm , there are only three _ independent _ parameters , the @xmath49 and @xmath67 gauge coupling constants , @xmath119 and @xmath120 , as well as the vev of the higgs doublet , @xmath121 . once these three parameters and their counter terms are fixed by the experimental data , all other physical observables in the gauge sector
can then be predicted in terms of these three input parameters .
a special feature of the sm with the assumption of one higgs doublet is the validity of the tree level relation , @xmath122 due to the tree level custodial symmetry .
there is thus a definite relation between the w - boson mass and the z - boson mass . of course
, one can equivalently choose any three physical observables as the input parameters in the gauge sector .
if we choose @xmath123 and @xmath124 as the three input parameters in the gauge sector , the w - boson mass , @xmath125 , then is predicted in the usual way via muon - decay , @xmath126 \ ; , \ ] ] where @xmath127 summarizes the one - loop radiative corrections , and it is given in terms of the gauge boson self - energy two point functions as , @xmath128 the counter term for the weak mixing angle @xmath129 which is defined through the w- and z - boson mass ratio , @xmath130 , is given by , @xmath131 \ ; .\ ] ] both of the two point functions , @xmath132 and @xmath133 , have identical leading quadratic @xmath134 dependence , @xmath135 , and thus their difference is only logarithmic .
the two - point function , @xmath136 , is also logarithmic in @xmath134 .
however , the difference between @xmath137 and @xmath138 has quadratic dependence in @xmath134 .
thus the prediction for @xmath125 is quadratic in @xmath134 . in the presence of a @xmath49 triplet higgs , on the other hand , a tri - linear coupling between the doublet and the triplet higgs , @xmath139 , is allowed by the gauge symmetry @xmath140 .
so unless one imposes a discrete symmetry to forbid such a tri - linear interaction , the vev of the triplet is non - zero , @xmath141 .
this thus leads to the need for a fourth input parameter in the gauge sector , with the fourth parameter being the vev of the triplet higgs , @xmath142 .
many of the familiar predictions of the standard model are drastically changed by the need for this extra input parameter@xcite .
one can equivalently choose the effective leptonic mixing angle , @xmath129 , as the fourth input parameter , where @xmath129 is defined through the ratio of the vector to axial vector parts of the @xmath143 coupling , @xmath144 . the counter term for @xmath145 is formally related to the wave function renormalizations for @xmath146 and @xmath147 and it is given by , @xmath148\bigg\ } \ ; , \label{cs}\ ] ] where @xmath149 is the axial part of the electron self - energy and @xmath150 are the vector and axial vector parts of the @xmath143 vertex corrections .
contrary to the sm case in which the @xmath134 dependence is quadratic , in models with a triplet higgs the dominant contribution in eq .
[ cs ] , @xmath151 , depends on @xmath134 only logarithmically .
due to this logarithmic dependence , the constraint on the model is weakened . on the other hand ,
the scalar contributions become important as they are quadratic due to the lack of the tree level custodial symmetry , as pointed out in ref . .
in many little higgs models , the @xmath90 parameter also differs from one already at the tree level , due to the presence of the triplet higgs which acquires a non - vanishing vev .
( for analyses based on tree level constraints can be found in ref . and
those including the heavy top effects at one loop can be found in ref . .
) a consistent renormalization scheme thus requires an additional input parameter in the gauge sector . in ref .
, the authors choose the muon decay constant @xmath152 , the physical z - boson mass @xmath153 , the effective lepton mixing angle @xmath145 and the fine - structure constant @xmath154 as the four independent input parameters in the renormalization procedure .
the @xmath90 parameter , defined as , @xmath155 , and the @xmath156-boson mass , which is defined through muon decay , are then derived quantities .
since the loop factor occurring in radiative corrections , @xmath157 , is similar in magnitude to the expansion parameter , @xmath158 , of the chiral perturbation theory for @xmath66 of a few tev , the one - loop radiative corrections can be comparable in size to the next - to - leading order contributions at tree level .
both types of corrections are of the order of a few percent . the effective leptonic mixing angle is defined through the ratio of the vector to axial vector parts of the @xmath159 coupling , @xmath160 which differs from the naive definition of the weinberg angle in the littlest higgs model , @xmath161 , by , @xmath162 \ ; .\ ] ] the w - boson mass is defined through muon decay , @xmath163 \ ; , \ ] ] where @xmath164 summarize the tree level corrections due to the change in definition in the weak mixing angle as well as the contributions from exchange of the heavy gauge bosons , @xmath165 the one - loop radiative corrections are collected in @xmath166 , @xmath167 + \pi^{\gamma\gamma}(0)^{\prime } - \frac{c_{\theta}}{s_{\theta } } \frac{\pi^{\gamma z}(m_{z})}{m_{z}^{2 } } \nonumber\ ; .\end{aligned}\ ] ] the predictions for @xmath168 with and without the one - loop contributions for @xmath169 tev is given in fig .
[ fig0 ] , which demonstrates that a low value of @xmath66 ( @xmath170 tev ) is allowed by the experimental restrictions from the @xmath156 and @xmath147 boson masses,@xcite due to cancellations among the tree - level and one - loop corrections .
this shows the importance of a full one - loop calculation in placing the electroweak precision constraints .
[ ] [ ] @xmath171 ( gev ) [ ] [ ] @xmath172 [ ] [ ] @xmath173(total ) @xmath174 [ ] [ ] @xmath175(total ) @xmath174 [ ] [ ] @xmath173(tree ) @xmath176 [ ] [ ] 1 @xmath177 limit on @xmath173 ( exp ) @xmath178 [ ] [ ] 1 @xmath177 limit on @xmath175 ( exp ) @xmath179 except for the littlest higgs model with t - parity , where @xmath66 can be as low as 500 gev@xcite , all other models discussed in the previous section receive tree level contributions to the ew observables . the precision ew constraint in the @xmath94 model has been analyzed in ref . .
due to the absence of the triplet higgs , the contribution to the @xmath87 parameter is minimal , and the scale for @xmath66 in this model can be 1 tev@xcite . in the simple group model based on @xmath99^{4}/[su(3)]^{4}$ ] , the
precision ew fit has been performed in ref . , and the bound on @xmath66 is found to be 4.2 tev . in the @xmath103 simple group model ,
the bound on @xmath66 is slightly improved to be 3.3 tev @xcite .
the precision ew constrains in the minimal moose model have been investigated in ref . ,
in which @xmath66 is found to be very severely constrained . due to the approximate custodial symmetry in the so(5 ) and
so(9 ) moose models , this bound can be relaxed@xcite .
it has been pointed out that in many new models with physics beyond the standard model , there is generally an implicit fine - tuning among the model parameters which may be over - looked at first glance but show up in a systematic analysis@xcite .
such implicit fine - tuning , which is needed to render the models phenomenologically viable , has been quantified by barbieri and giudice in ref . . in ref .
, casas , espinosa and hidalgo examine this issue in the little higgs models and find that the degree of such implicit fine - tuning is usually much more substantial than the rough estimate .
if we define the amount of fine - tuning in the higgs vev in the model _ a la _ barbieri and giudice@xcite as , @xmath180 , where @xmath181 for @xmath182 are input parameters of the model , then the amount of fine - tuning associated with parameter @xmath181 is given by , @xmath183 and the total amount of fine - tuning in the model is defined as , @xmath184 fig .
[ dvsmh ] shows the fine - tuning that exists in various versions of the little higgs models as well as in the sm and mssm .
such implicit fine - tuning in the little higgs models has been found to be more severe compared to the fine - tuning required in mssm@xcite .
( see also ref . .
) since the little higgs models are effective theories valid only up to the cut - off scale , @xmath185 tev , an important question one has to address is what lies beyond this cut - off scale@xcite .
a few possible uv completions have been speculated . in ref . , the little higgs model is uv completed into a model in which the little higgs is composite at 10 tev and the model has a matter parity , @xmath186 , similar to the r - parity in susy . in ref . and , the little higgs model is completed by another little higgs model
this thus postpones the onset of the strong coupling regime to @xmath187 tev . in ref .
an alternative was proposed in which the littlest higgs model is uv completed by a five - dimensional anti - de sitter space where the global su(5 ) symmetry in 4d littlest higgs model corresponds to an su(5 ) gauge symmetry in the 5d bulk . the issue of vacuum stability in various little higgs models has been investigated in ref . , while their finite temperature effects have been studied in ref . .
an attempt to supersymmetrize the little higgs model has been made in ref . , and it was found that the model can be embedded into an @xmath95 gut .
little higgs models are a new approach to stabilizing the higgs mass . the higgs boson in these models arises as a pseudo - nambu - goldstone boson resulting from the spontaneous breaking of some global symmetry .
no interaction in these models alone can break the complete global symmetries that prohibit the quadratic contribution to the higgs potential .
thus only when these interactions act collectively can the complete global symmetries be broken to give a quadratic mass term to the higgs boson .
the quadratically divergent contributions to the higgs potential can therefore arise only at the two - loop level . in this note ,
i review various little higgs models and the precision ew constraints placed in them . in most of these models , the parameter space is severely constrained by the precise electroweak data , with the exception of the littlest higgs model with t - parity , in which a low cutoff scale is allowed .
the importance of a full one - loop calculation is also emphasized and demonstrated explicitly in the case of the littlest higgs model .
there are other important questions such as the uv completion , vacuum stability and the contributions to the electroweak precision observables from higher dimensional operators at the cutoff scale , which are either out of the scope of this review or have only been briefly mentioned .
more careful studies on these issues are clearly needed .
i thank my collaborators , s. dawson , t. krupovnickas and k.t .
mahanthappa for numerous discussions .
the hospitality of the aspen center for physics , where part of this work was done , is also acknowledged .
fermilab is operated by universities research association inc .
under contract no .
de - ac02 - 76ch03000 with the u.s .
department of energy .
b. w. lee , c. quigg and h. b. thacker , phys . rev . *
d16 * , 1519 ( 1977 ) .
z. chacko , h. s. goh and r. harnik , hep - ph/0506256 .
z. chacko , y. nomura , m. papucci and g. perez , hep - ph/0510273 ; z. chacko , h. s. goh and r. harnik , hep - ph/0512088 .
w. buchmuller and d. wyler , nucl . phys . *
b268 * , 621 ( 1986 ) .
d. b. kaplan and h. georgi , phys . lett . * b136 * , 183 ( 1984 ) ; d. b. kaplan , h. georgi and s. dimopoulos , phys .
b136 * , 187 ( 1984 ) ; h. georgi and d. b. kaplan , phys . lett . *
b145 * , 216 ( 1984 ) ; h. georgi , d. b. kaplan and p. galison , phys . lett . *
b143 * , 152 ( 1984 ) ; m. j. dugan , h. georgi and d. b. kaplan , nucl . phys . *
b254 * , 299 ( 1985 ) .
n. arkani - hamed , a. g. cohen and h. georgi , phys . lett . *
b513 * , 232 ( 2001 ) ; n. arkani - hamed , a. g. cohen , t. gregoire and j. g. wacker , jhep * 0208 * , 020 ( 2002 ) ; n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys . rev .
* d67 * , 095004 ( 2003 ) .
g. burdman , m. perelstein and a. pierce , phys .
lett . * 90 * , 241802 ( 2003 ) [ erratum - ibid . *
92 * , 049903 ( 2004 ) ] ; c. dib , r. rosenfeld and a. zerwekh , hep - ph/0302068 ; s. c. park and j. song , phys . rev .
* d69 * , 115010 ( 2004 ) ; g. azuelos _ et al .
_ , eur . phys .
j. * c39s2 * , 13 ( 2005 ) ; j. j. liu , w. g. ma , g. li , r. y. zhang and h. s. hou , phys .
rev . * d70 * , 015001 ( 2004 )
; g. a. gonzalez - sprinberg , r. martinez and j. a. rodriguez , phys . rev . * d71 * , 035003 ( 2005 ) ; g. c. cho and a. omote , phys
. rev . * d70 * , 057701 ( 2004 ) ; w. kilian , d. rainwater and j. reuter , phys
* d71 * , 015008 ( 2005 ) ; c. x. yue and w. wang , phys
. rev . * d71 * , 015002 ( 2005 ) ; c. x. yue ,
w. wang and f. zhang , hep - ph/0503260 ; c. x. yue , w. wang , z. j. zong and f. zhang , hep - ph/0504253 ; j. a. conley , j. hewett and m. p. le , hep - ph/0507198 ; c. f. berger , m. perelstein and f. petriello , hep - ph/0512053 .
w. j. huo and s. h. zhu , phys .
rev . * d68 * , 097301 ( 2003 )
. s. r. choudhury , n. gaur , a. goyal and n. mahajan , phys .
* b601 * , 164 ( 2004 ) . s. r. choudhury , n. gaur , g. c. joshi and b. h. j. mckellar , hep - ph/0408125 .
a. j. buras , a. poschenrieder and s. uhlig , nucl .
b716 * , 173 ( 2005 ) .
f. bazzocchi , s. bertolini , m. fabbrichesi and m. piai , phys . rev . *
d68 * , 096007 ( 2003 ) ; _ ibid . _ * d69 * , 036002 ( 2004 ) ; f. bazzocchi , phys . rev . *
d70 * , 013002 ( 2004 ) ; j. y. lee , jhep * 0412 * , 065 ( 2004 ) ; o. c. w. kong , phys . rev . *
d70 * , 075021 ( 2004 ) ; t. han , h. e. logan , b. mukhopadhyaya and r. srikanth , phys .
* d72 * , 053007 ( 2005 ) ; m .- c .
chen and k. t. mahanthappa , phys .
rev . * d62 * , 113007 ( 2000 ) ; _ ibid . _ * d65 * , 053010 ( 2002 ) ; _ ibid .
* d68 * , 017301 ( 2003 ) ; hep - ph/0009059 ; hep - ph/0110037 ; aip conf .
proc . * 698 * , 222 ( 2004 ) ; hep - ph/0409165 ; hep - ph/0510296 ; phys .
rev . * d71 * , 035001 ( 2005 ) ; int .
j. mod .
a16 * , 3923 ( 2001 ) ; m .- c .
chen , phys .
d71 * , 113010 ( 2005 ) .
chen and k. t. mahanthappa , int .
* a18 * , 5819 ( 2003 ) ; aip conf . proc . *
721 * , 269 ( 2004 ) .
g. passarino , nucl .
b361 * , 351 ( 1991 ) ; b. w. lynn and e. nardi , nucl . phys . *
b381 * , 467 ( 1992 ) ; t. blank and w. hollik , nucl . phys . *
b514 * , 113 ( 1998 ) ; m. czakon , m. zralek and j. gluza , nucl . phys . *
b573 * , 57 ( 2000 ) ; m. czakon , j. gluza , f. jegerlehner and m. zralek , eur .
j. * c13 * , 275 ( 2000 ) .
d. toussaint , phys . rev .
* d18 * , 1626 ( 1978 ) ; g. senjanovic and a. sokorac , phys . rev . *
d18 * , 2708 ( 1978 ) .
m. b. einhorn and j. wudka , phys . rev .
* d39 * , 2758 ( 1989 ) ; phys . rev . *
d47 * , 5029 ( 1993 ) . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys
. rev . * d67 * , 115002 ( 2003 ) ; m. perelstein , m. e. peskin and a. pierce , phys
. rev . * d69 * , 075002 ( 2004 ) .
w. kilian and j. reuter , phys . rev . * d70 * , 015004 ( 2004 ) ; z. han and w. skiba , hep - ph/0506206 .
chen and s. dawson , phys . rev .
* d70 * , 015003 ( 2004 ) ; proceedings of susy 2004 , _ tsukuba 2004 , supersymmetry and unification of fundamental interactions _ , 921 - 924 , hep - ph/0409163 .
chen , s. dawson and t. krupovnickas , hep - ph/0504286 ; under preparation . | the little higgs idea is an alternative to supersymmetry as a solution to the gauge hierarchy problem . in this note ,
i review various little higgs models and their phenomenology with emphases on the precision electroweak constraints in these models . |
hd 14437 is a poorly studied peculiar a star with a magnitude of @xmath1 , effective temperature @xmath2 ( glagolevskij , 1995 ) .
observations of the 1980s with the 6-m telescope showed a variability of the longitudinal magnetic field with an amplitude of about 2 kg and a constant negative sign .
the description of the observations and data reductions are presented by glagolevskij et al .
( 1985 ) in more details .
in this paper we present new observations of hd 14437 which were carried out in 199697 using the ccd detector ( chuntonov & glagolevskij , 1997 ) . we have observed the spectra of the star on the 6-m telescope with a spectral resolution @xmath3 .
the context nice ( knyazev & shergin , 1995 ) in the midas system was used .
the new and old observational data are presented in table [ observations ] ( jd 24446552445900 : with photographic plates , jd 2449555 and jd 2449556 : with the hydrogen lines magnetometer , from jd 2450000 on : with the ccd detector ) .
we have carried out the line identification using the vienna astrophysical line data base ( vald ) ( piskunov et al . , 1995 ) and moore s tables ( moore , 1945 ) . in view of determining the individual land factor of each spectral line used in our measurements
, we have identified the lines in the following spectral ranges : @xmath459556385 and @xmath444604680 .
si and cr are observed to be overabundant .
we have found an underabundance of o by the lack of the lines oi @xmath46155.97 , 6556.78 and 6158.18 which are usually strong at spectral class a2 .
we have determined the effective magnetic field using babcock s standard formula with the individual land factor @xmath5 for each line .
the measurements are given in table [ observations ] .
lr|lr jd 2400000&@xmath6,g&jd 2400000&@xmath6,g + 44655.208&@xmath7 & 50415.240&@xmath8 + 44656.308&@xmath9 & 50415.269&@xmath10 + 44659.188&@xmath11 & 50499.196&@xmath12 + 44660.196&@xmath13 & 50499.221&@xmath14 + 44860.541&@xmath15 & 50500.161&@xmath16 + 45303.354&@xmath17 & 50500.188&@xmath18 + 45303.362&@xmath19 & 50617.507&@xmath20 + 45476.541&@xmath21 & 50643.521&@xmath22 + 45900.493&@xmath23 & 50705.451&@xmath24 + 49555.472&@xmath25*&50706.410&@xmath26 + 49556.420&@xmath27*&50707.412&@xmath28 + 50413.523&@xmath29 & 50709.602&@xmath30 + 50414.125&@xmath31 & 50710.432&@xmath32 + 50415.217&@xmath33 & & + the effective magnetic field has preserved its sign for about ten years .
the suggestion about the long period of the star contradicts our data , because we have observed the longitudinal component of the magnetic field to undergo significant changes within a few days only .
so , we can draw the conclusion that the axis of rotation , the magnetic axis and the line of sight are inclined at small angles .
having the observational data of the effective magnetic field @xmath34 , we tried to find the rotational period of hd 14437 .
it has been determined by yurkevich s method in the interval of 0100 days .
we have found three groups of probable periods : 13 , 16 and 30 days .
the best fitted variability curve of @xmath34 was derived for the period of @xmath35 .
independently , a search for the period has been performed by mathys et al .
( 1997 ) on the basis of the surface field @xmath36 .
they have found three groups of probable periods : 15 , 30 and 350 days .
later , wade ( 1997 ) proposed another period , close to @xmath37 , which he has calculated using 100 points of hipparcos photometry .
the variability curve of the effective magnetic field for @xmath37 looks somewhat worse than for @xmath35 because of greater dispersion , nethertheless it remains quite acceptable . besides
, determination of the period by photometry , with a much larger number of homogeneous points , undoubtedly gives a more reliable result .
the @xmath34 phase diagram for @xmath38 is presented in figure [ curv26 ] .
we would like to thank v. g. elkin for his help in observations and discussions of the results , and g. wade for the information about the 26 days period of hipparcos photometry .
this work was partly supported by a rfbr grant 970216247 . | the magnetic field of the cp star hd 14437 was discovered by glagolevskij et al .
( 1985 ) using the 6-m telescope of the special astrophysical observatory .
no polarity changes have been found during 2 years of observations .
a very long ( @xmath0 34 years ) period of rotation was proposed to explain the measurements . to check this hypothesis
, we made a new series of magnetic field observations for this star 10 years later with the 6-m telescope .
the polarity of the longitudinal magnetic field is still negative and has not shown any change during more than 10 years of observations .
we found the most probable periods to be several days .
it means that we have observed a star where the magnetic and rotation axes are inclined at a small angle , and the negative pole is not far from the line of sight . |
the number density of massive galaxy clusters is exponentially sensitive to the amplitude of initially gaussian density fluctuations and has long been recognized as a sensitive cosmological probe @xcite . in particular
, the evolution of the number counts above a given mass threshold can determine the properties of the dark energy that accelerates the expansion @xcite .
this potential can only be realized if , in addition to the cluster redshifts , the cluster masses themselves are known , at least statistically .
unfortunately the total mass is not a direct observable and must be inferred through scaling relations with , e.g. the sunyaev - zeldovich ( sz ) flux decrement , the @xmath0-ray flux or temperature , weak lensing shear , or optical velocity dispersion . indeed
the normalization of the mass - temperature relation is currently the leading source of ambiguity in interpreting the local cluster abundance ( e.g. @xcite ) .
it compromises dark energy constraints if the cluster mass selection can not be defined to a few percent in mass @xcite .
the mass - observable relation can potentially be calibrated within a survey itself if its effect on the number counts is not degenerate with the cosmology .
for example , with a single cut on the temperature , the survey itself can calibrate the mass - temperature normalization to better than the tens of percent that span the determinations in the current literature , if it does _ not _ evolve @xcite .
however this method fails if the relationship has uncertain evolution that mimics the cosmology @xcite .
because the number density of clusters as a function of mass has a fixed functional form given by cosmological simulations ( e.g. @xcite ) , cluster number counts as a function of the observable in principle have the ability to self - calibrate even an evolving mass - observable relation .
here we study the potential for internal calibration and consistency checks in idealized future cluster number count surveys .
the cosmological utility of cluster counts stems from the simulation - based prediction of their comoving differential number density as a function of mass @xcite , @xmath1\ , .
\label{eqn : massfun}\ ] ] here @xmath2 is the variance in the linear density field smoothed with a top hat that encloses @xmath3 at the mean matter density today @xmath4 .
the cosmological sensitivity comes from these quantities and the comoving volume element in a redshift slice and solid angle . to assess the sensitivity of counts to various cosmological and mass - observable scaling parameters @xmath5
, we employ the fisher matrix technique @xcite @xmath6 where the covariance matrix is given by @xmath7 the first term represents sample covariance in the volume @xmath8 from large - scale structure , calculated as described in @xcite , and the second term shot variance . here the number density in a bin @xmath9 is defined by the redshift interval @xmath10 around @xmath11 and a selection based on some observable quantity @xmath12 , such as the sz flux decrement .
note that sampling errors for the selections in the same redshift bin completely covary .
the fisher matrix is a local approximation to the covariance matrix of the parameters @xmath13 where @xmath14 is the covariance matrix from prior information .
the fisher matrix is evaluated around a fiducial choice of parameters .
for the cosmology , we take : the dark energy density @xmath15 , equation of state @xmath16 , physical matter density @xmath17 , physical baryon density @xmath18 , tilt @xmath19 , and the initial normalization of the curvature fluctuations @xmath20 at @xmath21 mpc@xmath22 @xcite ( more conventionally : @xmath23 , @xmath24 , or @xmath25 ) .
we will however work in a future context where the cosmic microwave background ( cmb ) has constrained the high redshift universe parameters with 1@xmath26 errors of @xmath27 ( e.g. @xcite ) in a flat universe , leaving the dark energy parameters to be determined by the survey . for the mass - observable relation
, we take the form @xmath28 where @xmath29 and @xmath30 are dimensional constants and the normalization parameter @xmath31 is dimensionless .
we take the scaling parameters as the amplitudes of a piecewise constant form @xmath32 and @xmath33 .
specifically , given a variation @xmath34 and @xmath35 from the fiducial model , the true mass is related to the apparent mass as @xmath36 where @xmath37 is the pivot mass scale that defines a normalization point for possible external information .
this generalizes previous treatments which have assumed constant normalization @xcite and power law evolution @xcite . in the limit of small redshift bins
, these simpler cases can be recovered from our more general treatment by noting that under a re - parameterization of the space to the set @xmath38 , possibly of lower dimension , the covariance matrix transforms as @xmath39 the fisher matrix definition itself can be so viewed . in the usual approach ( e.g. @xcite ) , all of the clusters in a given redshift bin above a given observable threshold are simply binned together . clearly , allowing for arbitrary variations @xmath40
no cosmological information can be extracted .
however the data at a given redshift may be binned into several _
apparent _ mass bins based on the observable @xmath12 .
the additional information supplied by multiple bins allows for a breaking of the degeneracy . likewise ,
even if the normalization is fixed by detailed followup by @xmath0-ray or weak lensing measurements at some mass @xmath41 ( e.g. @xcite ) , an uncertain scaling index @xmath42 would again destroy the cosmological information with a single bin if @xmath43 .
multiple bins again in principle allow the cosmology and the mass - observable relations to be jointly determined .
= 3.25truein this treatment ignores intrinsic scatter in the mass - observable relation as well as measurement error in the observable .
more realistically the selection function is not sharp in mass but assuming its functional form is well characterized this does not substantially compromise the cosmological information @xcite . for definiteness ,
let us take a fiducial cluster survey with specifications similar to the planned south pole telescope ( spt ) survey for clusters with the sz effect : an area of 4000 deg@xmath44 survey and a sensitivity corresponding to a constant @xmath45 ( e.g. @xcite ) .
we divide the number counts into bins of redshift @xmath46 out to @xmath47 and @xmath48 .
this crude binning is sufficient to retain the cosmological information . for illustration purposes ,
we take a pivot mass scale of @xmath49 to reflect potential mass measurement followup on the high mass end .
let us first consider that simulations or mass followup with different observables have placed strong priors on the mass - observable normalization and scaling index @xmath50 . in this case ,
the division into mass bins yields nearly independent measurements of the dark energy for a consistency check . in fig .
[ fig : consistency ] , we illustrate the consistency check in the @xmath51 plane for individual mass bins of @xmath52 from @xmath53 .
although the strongest constraints do come from the lowest mass bin , as expected from the increasing rarity of massive clusters at high redshift , the first 6 bins show significant constraints on the dark energy .
this consistency test demonstrates that there is sufficient information in the mass binning to attempt some degree of self - calibration of the mass - observable relations .
let us next keep the scaling index fixed @xmath54 but allow the normalization @xmath55 to vary ( see fig .
[ fig : plotwl ] ) .
solid lines represent constraints with the 6 mass bins and dashed lines those with a single mass bin or mass threshold .
we first allow only a redshift - independent normalization factor @xmath56 .
as shown in @xcite , even a single mass bin is sufficient to calibrate the relation to good enough accuracy for dark energy constraints ( here @xmath57 ) .
we next employ a power law evolution in the normalization @xmath58 . here
errors on the dark energy parameters degrade substantially with a single mass bin @xcite , but much of the loss is recovered from the multiple binning . finally , we take a fully arbitrary evolution in the redshift bins , i.e. no constraint on the form of @xmath34 .
as expected , there is no constraint on the dark energy with a single mass bin , but with multiple bins interesting constraints on the dark energy can still be extracted .
the key here is that the dark energy evolution is assumed to be smooth and parameterized by two numbers and so can not compensate an arbitrary variation in the normalization evolution .
= 3.25truein figure [ fig : aerror ] ( top ) shows the errors on @xmath34 with roughly 10% constraints on the normalization in each redshift band out to @xmath59 .
note that the covariance between the bands is nearly complete so that the overall normalization is still only known to @xmath60 but that the evolution is highly constrained .
conversely , external information on the mass - observable normalization would have to be substantially better than @xmath61 to improve on the internal results . in the context of our fiducial cosmology ,
the critical prior assumption is that the initial amplitude of the fluctuations is fixed by the cmb .
otherwise what is constrained by the multiple mass bins is a degenerate combination of the initial amplitude and the mass - observable normalization .
we have so far assumed that the scaling index @xmath62 is known a priori but of course that too is part of the uncertain mass - observable relation .
allowing for arbitrary variations @xmath34 _ and _ @xmath35 destroys most of the information on the dark energy .
however in the less drastic cases of adding a constant @xmath63 to a constant @xmath64 for a two parameter model , the additional degradation @xmath65 is a negligible @xmath66 and of the addition of power law variation in @xmath62 , @xmath67 to power law variation in @xmath64 for a 4 parameter model , a factor of @xmath68 for a total of @xmath69 .
another interesting case to consider is if some independent mass calibration , say from weak lensing or @xmath0-ray temperature followup , normalizes the mass - observable relation on the high mass end .
extrapolation down to the survey mass limit can be dangerous due to the uncertain physics of low mass clusters .
for illustrative purposes , let us take this constraint as @xmath70 at the pivot mass of @xmath49 .
despite the constraint on the normalization , if we allow for arbitrary variations in the scaling index and retain only a threshold at @xmath71 , no constraint on the dark energy can be extracted . with multiple bins ,
the errors on @xmath72 are only degraded by a factor of 1.16 to @xmath73 for a completely arbitrary @xmath35 . furthermore the errors on @xmath63 shown in fig .
[ fig : aerror ] ( bottom ) are at the several percent level and would be of interest in studying cluster physics .
as in the case of classical cosmological tests for the dark energy involving standardized candles and rulers , e.g. supernovae and the peaks in the cmb power spectrum , cluster number count tests require a standardized mass based on observable quantities . we have demonstrated that consistency with the well - determined shape of the mass function from cosmological simulations can in principle be used to calibrate the survey internally .
two examples are a normalization that has an arbitrary evolution in redshift and a scaling from the more easily calibrated high mass end that has an arbitrary evolution in the index .
our study involves a number of idealizations that merit future study and so internal calibration is best viewed as a useful check on cross - calibration studies with detailed multi - wavelength followup ( e.g. @xcite ) and cosmological simulations .
the crucial assumptions are that the survey be sufficiently deep to explore a substantial dynamic range in the mass function , the underlying mass function is known given a cosmology , the mass - observable selection function is sharp compared with the binning , the mean mass - observable relation is power law in form , the high redshift normalization of the fluctuations is fixed by future cmb data , and the dark energy equation of state is constant . in the fiducial model , a depth comparable to that planned for the spt sz survey is sufficient to make self - calibration useful but note that the necessary mass limit scales roughly with the non - linear mass scale @xmath74 . that depends on the currently uncertain normalization and so
if the normalization is lower than our fiducial choice it may be beneficial to sacrifice survey width for depth .
while the mass function form is based on simulations and currently its scaling with cosmology is only known to the @xmath75 level @xcite , determination requires only the well - understood gravitational physics of the dark matter and is far more secure than simulation - based mass - observable relations . in any case
, this knowledge is a prerequisite for any cluster number count study of cosmology .
we have assumed a deterministic mass - observable relation that makes the selection function sharp in apparent mass .
gaussian scatter in this relation does not seriously compromise the cosmological information @xcite but long uncharacterized tails to low mass would , given the steep mass function .
likewise the cluster physics of preheating and cooling can enter into the low mass end to make the mass - observable relations deviate from a power law . here
the binning serves as a useful consistency test for the implicitly assumed cluster physics .
the required prior information on the initial normalization is within reach of the upcoming cmb satellite missions if the extent of reionization can be determined ( e.g. @xcite ) .
if not , binning still serves to constrain the evolution of the mass - observable relation but leaves a degeneracy between the overall normalization of the relations and the fluctuations which can be fixed , e.g. by cmb lensing , ly@xmath76 forest clustering or even local cluster abundance studies .
conversely , external calibration of the mass - observable relation at the percent level can determine the fluctuation normalization at a comparable level .
binning can then provide tests of the dark energy model assumptions @xcite or other cosmological priors .
it is especially valuable if redshifts are available only locally ( e.g. @xmath77 from current optical surveys ) . in summary , utilizing the extra information in counting clusters as a function of both the redshift and their observable properties allows for joint solutions to the mass - observable relations and cosmology , if they are both simple .
if not , it offers valuable internal consistency checks against overly simplistic assumptions . _ acknowledgments : _ i thank z. haiman , a.v .
kravtsov , j.j .
mohr , c. pryke and the chicago thunch group for useful conversations and j. valdes for * nix expertise in file recovery .
wh is supported by nasa nag5 - 10840 and the doe oji program .
evrard , astrophys .
j , * 341 * , l71 ( 1989 ) ; c.s .
frenk , s.d.m .
white , g. efstathiou , m. davis , ibid , * 351 * , 10 ( 1990 ) ; j.p .
henry , m. arnaud , ibid , * 372 * , 410 ( 1991 ) ; p.b .
lilje , ibid , * 386 * , l33 ( 1992 ) ; n.a . bahcall , r.y .
cen , ibid , * 565 * , l5 ( 1992 ) . | cluster number counts offer sensitive probes of the dark energy if and only if the _ evolution _ of the cluster mass versus observable relation(s ) is well calibrated .
we investigate the potential for internal calibration by demanding consistency in the counts as a function of the observable . in the context of a constant dark energy equation of state , known initial fluctuation amplitude expected from the cmb , universal underlying mass function , and an idealized selection , we find that the ambiguity from the normalization of the mass - observable relationships , or an extrapolation of external mass - observable determinations from higher masses , can be largely eliminated with a sufficiently deep survey , even allowing for an arbitrary evolution .
more generally , number counts as a function of both the redshift and the observable enable strong consistency tests on assumptions made in modelling the mass - observable relations and cosmology . |
it is well known that transport problems on the line involving convex cost functions have explicit solutions , consisting in a monotone rearrangement . recently
, an efficient method has been introduced to tackle this issue on the circle @xcite . in this note
we introduce an algorithm that enables to tackle optimal transport problems on the line ( but actually also on the circle ) with concave costs .
our algorithm complements the method suggested by mccann @xcite .
mccann considers general real values of supply and demand and shows how the problem can be reduced to convex optimization somewhat similar to the simplex method in linear programming .
our approach as presented here is developed for the case of unit masses and is closer to the purely combinatorial approach of @xcite , but extends it to a general concave cost function .
the extension to integer masses will be presented in @xcite . +
the method we propose is based on a class of local indicators , that allow to detect consecutive points that are matched in an optimal transport plan .
thanks to the low number of evaluations of the cost function required to apply the indicators , we derive an algorithm that finds an optimal transport plan in @xmath2 operations in the worst case . in practice , the computational cost of this method appears to behave linearly with respect to @xmath3 .
+ since the indicators apply locally , the algorithm can be massively parallelized and also allows to treat optimal transport problems on the circle . in this way
, it extends the work of aggarwal _ et al . _
@xcite in which cost functions have a linear dependence in the distance .
for @xmath4 , consider @xmath5 and @xmath6 two sets of points in @xmath1 that represent respectively demand and supply locations .
the problem we consider in this note consists in minimizing the transport cost @xmath7 where @xmath8 is a permutation of @xmath9 .
this permutation forms a _ transport plan_. + we focus on the case where the function @xmath10 involves a concave function as stated in the next definition .
[ def : cost ] the _ cost function _ in ( [ eq:4 ] ) is defined on @xmath1 by @xmath11 with @xmath12 , where @xmath13 is a concave non - decreasing real - valued function of a real positive variable such that @xmath14 . some examples of such costs are given by @xmath15 with @xmath16 , and @xmath17 or @xmath18 is with @xmath19 .
+ finally , we denote by @xmath20 the permutation associated to a given optimal transport plan between @xmath21 and @xmath22 : for all permutation @xmath8 of @xmath9 , @xmath23
in this section , we present a way to build a particular partition of the set @xmath24 .
+ consider two pairs of matched points @xmath25 and @xmath26 , say e.g. @xmath27 , @xmath28 .
it is easy to prove that the following alternative holds : 1 .
@xmath29\cap [ p_{i'},q_{\sigma^\star(i')})]= \emptyset$ ] , 2 .
@xmath29\subset [ p_{i'},q_{\sigma^\star(i')})]$ ] or @xmath30\subset[p_i , q_{\sigma^\star(i)}]$ ] .
this remark is a direct consequence of the concavity of the cost function and is often denominated as `` the non - crossing rule '' @xcite . in the next section ,
we show how it allows decompose the initial situation in sub - problems where supply and demand points are alternated .
+ because of the non - crossing rule in an optimal plan there are as many supply points as demand points between any pair of matched points @xmath31 and @xmath32 . for a given demand point @xmath31 ,
define its _ left neighbor _ @xmath33 as the nearest supply point on the left of @xmath31 such that the numbers of supply and demand points between @xmath33 and @xmath31 are equal ; define the _
right neighbor _ @xmath34 of @xmath31 in a similar way
. then define a _
chain _ as a maximal alternating sequence of supply and demand points ( @xmath35 ) or ( @xmath36 ) such that each @xmath37 is the right neighbor of @xmath38 and the left neighbor of @xmath39 .
an extension of the proof of lemma 3 of @xcite shows that the collection of chains forms a partition of the set @xmath40 .
an simple example of such a partition is shown on fig .
[ exchaine ] .
( 9,0 ) ; ( 1,0 ) node @xmath41 ; ( 4,0 ) node @xmath41 ; ( 5,0 ) node @xmath41 ; ( 8,0 ) node @xmath41 ; ( 2,0 ) node @xmath42 ; ( 3,0 ) node @xmath42 ; ( 6,0 ) node @xmath42 ; ( 7,0 ) node @xmath42 ; ( 2,0 ) arc ( 0:180 : .5 ) ; ( 5,0 ) arc ( 0:180:1.5 ) ; ( 6,0 ) arc ( 0:180 : .5 ) ; ( 4,0 ) arc ( 0:-180 : .5 ) ; ( 7,0 ) arc ( 0:-180:1.5 ) ; ( 8,0 ) arc ( 0:-180 : .5 ) ; note that construction of this collection only depends on relative positions of supply and demand points and does not involve any evaluation of the cost function .
it can be done in @xmath43 operations .
+ the non - crossing rule implies that all matched pairs of points in an optimal transport plan must belong to the same chain .
we therefore restrict ourselves in the sequel , without loss of generality , to the case of a single chain @xmath44 for @xmath45 and keep these last notations throughout the rest of this paper .
thanks to the non - crossing rule , one knows that there exists at least two consecutive points @xmath46 or @xmath47 that are matched in any optimal transport plan . starting from this remark , we take advantage of the structure of a chain to introduce a class of indicators that enable to detect a priori such pairs of points .
[ def : lmi]we define @xmath48 where @xmath49 are such that @xmath50 and @xmath51 , and @xmath52 for @xmath53 , such that @xmath54 and @xmath55 .
the interest of these functions lies in the next result .
[ thlmi]let @xmath56 with @xmath57 and @xmath58 ( resp .
@xmath59 ) , such that @xmath60 ( resp .
@xmath61 ) .
+ assume that 1 .
@xmath62 for @xmath63 , @xmath64,[hyp1 ] 2 .
@xmath65 for @xmath63 , @xmath66 , [ hyp2 ] 3 .
@xmath67 ( resp .
@xmath68 ) .
[ hyp3 ] then any permutation @xmath20 associated to an optimal transport plan satisfies @xmath69 for @xmath70 ( resp .
@xmath71 for @xmath70 ) . in practice
, these indicators allow to find pairs of neighbors that are matched in an optimal transport plan .
this result is illustrated on fig .
[ pic : thlmi ] .
( 7,0 ) ; ( 1,0 ) node @xmath41 ; ( 3,0 ) node @xmath41 ; ( 5,0 ) node @xmath41 ; ( 2,0 ) node @xmath42 ; ( 4,0 ) node @xmath42 ; ( 6,0 ) node @xmath42 ; ( 4,0 ) arc ( 0:180 : .5 ) ; ( 5,0 ) arc ( 0:180:1.5 ) ; ( 0 + 8,0 ) node @xmath72 ; ( 0 + 9,0 ) ( 7 + 9,0 ) ; ( 1 + 9,0 ) node @xmath41 ; ( 3 + 9,0 ) node @xmath41 ; ( 5 + 9,0 ) node @xmath41 ; ( 2 + 9,0 ) node @xmath42 ; ( 4 + 9,0 ) node @xmath42 ; ( 6 + 9,0 ) node @xmath42 ; ( 3 + 9,0 ) arc ( 0:180 : .5 ) ; ( 5 + 9,0 ) arc ( 0:180 : .5 ) ; ( 8,-2 ) node @xmath73 ; ( 0 + 4.5,-4 ) ( 7 + 4.5,-4 ) ; ( 1 + 4.5,-4 ) node @xmath41 ; ( 3 + 4.5,-4 ) node @xmath41 ; ( 5 + 4.5,-4 ) node @xmath41 ; ( 2 + 4.5,-4 ) node @xmath42 ; ( 4 + 4.5,-4 ) node @xmath42 ; ( 6 + 4.5,-4 ) node @xmath42 ; ( 4 + 4.5,-4 ) arc ( 0:180 : .5 ) ; + before giving the proof , we state a basic result . [ lemm : basic ] we keep the previous notations .
define @xmath74 for @xmath53 , such that @xmath50 and @xmath51 , and @xmath75 for @xmath53 , such that @xmath54 and @xmath55 . both functions @xmath76 and @xmath77 are decreasing with respect to each of their two variables .
this lemma is a direct consequence of the concavity of the function @xmath78 .
we are now in the position to give the sketch of the proof of theorem [ thlmi ] .
+ * proof of theorem [ thlmi ] : * we consider the case where @xmath79 .
the case @xmath80 can be treated the same way .
+ the proof consists in proving that assumptions ( [ hyp1][hyp3 ] ) imply that neither demand nor supply points located between @xmath81 and @xmath82 can be matched with points located outside this interval , i.e. that the set @xmath83 is stable by an optimal transport plan . in this case , the result follows from assumption ( [ hyp1][hyp2 ] ) .
+ suppose that @xmath84 is not preserved by an optimal transport plan @xmath20 .
three cases can occur : a. [ c1 ] there exists @xmath85 , such that @xmath86 and @xmath87 and there exists @xmath88 , such that @xmath89 and @xmath90 .
b. [ c2 ] there exists @xmath91 , with @xmath92 such that @xmath93 .
c. [ c3 ] there exists @xmath94 , with @xmath95 such that @xmath96 .
we first prove that case [ c1 ] ) can not occur .
+ in case [ c1 ] ) , one can assume without loss of generality that @xmath97 is the largest index such that @xmath86 , @xmath98 and that @xmath99 is the smallest index such that @xmath89 , @xmath90 .
with such assumptions , the ( possibly empty ) subset @xmath100 is stable by @xmath20 . because of assumptions ( [ hyp1][hyp2 ] )
, no nesting can occur in this subset , and @xmath71 for @xmath101 .
+ on the other hand , since @xmath20 is supposed to be optimal , one has : @xmath102 thanks to lemma [ lemm : basic ] , one deduces from this last inequality that : @xmath103 and then : @xmath104 according to assumption ( [ hyp1 ] ) , @xmath105 and @xmath106 , so that : @xmath107 @xmath108 combining these last inequalities with ( [ ineq:1 ] ) one finds that : @xmath109 which contradicts assumption ( [ hyp3 ] ) .
+ let us now prove that cases [ c2 ] ) and [ c3 ] ) contradict the assumptions .
cases [ c2 ] ) and [ c3 ] ) can be treated in the same way .
consider case [ c2 ] ) . without loss of generality
, one can assume that @xmath110 is the smallest index such that @xmath111 and @xmath112 . because there are necessarily as much demands as supplies between @xmath113 and @xmath114 , there exists one and only one index @xmath115 such that @xmath116 and @xmath117 .
consequently , the ( possibly empty ) subsets @xmath118 and @xmath119 are stable by an optimal transport plan .
because of assumptions ( [ hyp1][hyp2 ] ) , no nesting can occur in these subsets , and @xmath69 for @xmath120 and @xmath71 for @xmath121 .
+ on the other hand , since @xmath20 is supposed to be optimal , one has @xmath122 thanks to lemma [ lemm : basic ] , one deduces from this last inequality that : @xmath123 because the cost is supposed to be increasing with respect to the distance , one finds that @xmath124 , so that ( [ ineq3 ] ) implies : @xmath125 and then : @xmath126 according to assumption ( [ hyp1 ] ) @xmath127 , so that : @xmath128 combining these last inequalities with ( [ ineq4 ] ) one finds that : @xmath109 which contradicts assumption ( [ hyp3 ] ) .
+ we have then shown that neither demand nor supply points located between @xmath81 and @xmath129 can be matched with located outside this interval .
the set @xmath84 is then stable by an optimal transport plan .
according to assumption ( [ hyp1][hyp2 ] ) , no nesting can occur in @xmath84 .
the result follows .
the recursive use of our indicators is on the basis of the next algorithm . + * algorithm : * set @xmath131 , @xmath132 , @xmath133 , and @xmath134 .
+ while @xmath135 and @xmath136 do 1 .
compute @xmath137 and @xmath138 for @xmath139 and @xmath140.[costlystep ] 2 .
define @xmath141 @xmath142 and do [ noncostlystep ] 1 . if @xmath143 and @xmath144 , set @xmath145 .
do [ removingstep ] * for all @xmath146 in @xmath147 and for @xmath148 , do * * define @xmath149 , * * remove @xmath150 from @xmath151 , * * remove @xmath152 and @xmath153 from @xmath154 and @xmath155 respectively .
* for all @xmath156 in @xmath157 and for @xmath158 , do * * define @xmath159 , * * remove @xmath160 from @xmath151 , * * remove @xmath152 and @xmath153 from @xmath154 and @xmath155 respectively . *
set @xmath161 , and rename the points in @xmath151 such that @xmath131 , @xmath162 * set @xmath134 .
if @xmath163 , for @xmath164 set @xmath159 .
+ to test the efficiency of our algorithm , we have applied it to an increasing number @xmath0 of pairs of points . for a fixed value of @xmath0 , 100 samples of points have been chosen randomly in @xmath165 $ ] , and the mean of the number of evaluations of @xmath78 has been computed .
the results are shown on fig .
+ the best case consists in finding a negative indicator at each step , and the worst corresponds to the case where all the indicators are positive . these two cases require respectively @xmath166 and @xmath167 evaluations of @xmath78
. + [ c][t]in - line evaluations of @xmath78 [ c][b]number of pairs of points @xmath0 [ l][c ] : @xmath168 , @xmath169 [ l][c ] : @xmath170 , @xmath171 [ l][c ] : @xmath172 , @xmath173 [ l][c ] : worst case @xmath167 , @xmath173 [ l][c]@xmath174 [ c][c]@xmath175 [ c][c]@xmath176 [ c][c]@xmath177 [ c][c]@xmath178 [ c][c]@xmath179 [ c][c]@xmath180 [ c][c]@xmath181 [ c][c]@xmath182 [ c][c]@xmath183 is the slope of the log - log graphs.,title="fig : " ] 00 a. aggarwal , a. bar - noy , s. khuller , d. kravets , and b. schieber .
efficient minimum cost matching using quadrangle inequality . in _
foundations of computer science , 1992 .
proceedings of 33rd annual symposium _ , pages 583592 , 1992 . | in this note , we introduce a class of indicators that enable to compute efficiently optimal transport plans associated to arbitrary distributions of @xmath0 demands and @xmath0 supplies in @xmath1 in the case where the cost function is concave .
the computational cost of these indicators is small and independent of @xmath0 .
a hierarchical use of them enables to obtain an efficient algorithm . |
pb@xmath5sn@xmath6se and pbte were investigated by powder x - ray diffraction ( xrd ) using a diamond anvil cell at room temperature , performed at beamline x17c of the national synchrotron light source .
a 4:1 methanol - ethanol mixture was used as the pressure transmitting medium and ruby fluoresence for pressure calibration .
the results are shown in fig .
[ figs1 ] . a pressure - induced structural phase transition in both samples
is indicated by new peaks in the powder diffraction pattern .
concomitant with the structural phase transition are a reduction of the infrared reflectance and a dramatic decrease of the infrared absorbance , shown for pbte in fig . [ figs4](a ) and ( e ) .
this was also observed in other materials , such as pbse and pb@xmath53sn@xmath54se ; fig .
[ figs2 ] shows their absorbance close to the structural phase transition . using either xrd or infrared spectroscopy
, we found that at room temperature , the ambient - pressure structure of pb@xmath0sn@xmath1se ( @xmath2 , 0.15 , and 0.23 ) and pbte sustains up to 5.1 , 4.0 , 2.9 , and 5.7 gpa , respectively .
the pressure dependence of the lattice parameter @xmath15 of pb@xmath5sn@xmath6se and pbte was obtained from rietveld refinement on the xrd patterns , shown in fig .
[ figs1 ] ( d ) and ( e ) .
the lattice parameter decreases almost linearly , with @xmath55 and @xmath56 / gpa for pb@xmath5sn@xmath6se and @xmath57 and @xmath58 / gpa for pbte .
the drude weight tensor @xmath59 has the form @xcite @xmath60 where @xmath61 is the planck constant , @xmath62 the electron charge , @xmath63 the free electron mass , @xmath64 the band index , @xmath40 the crystal momentum , @xmath65 the band dispersion , and @xmath24 the fermi energy .
@xmath66 is the @xmath43th component of the momentum matrix element between states in the _ _ n__th band with the crystal momentum @xmath40 , @xmath67 where @xmath68 is the wavefunction of bloch electrons in the @xmath64th band .
it can be shown that @xcite @xmath69 in which @xmath70 is the @xmath43th component of the bloch electron mean velocity in the @xmath64th band @xmath71 substituting eqs . and into eq . ,
we have @xmath72 evaluating the integral using the property of the dirac delta function , the above equation can be further written as @xmath73 where @xmath74 is the @xmath43th component of the fermi velocity in the _ _
n__th band and @xmath75 is the density of states at fermi energy in the _ _ n__th band .
the electronic density of states as a function of @xmath29 , corresponding to the band structure in fig .
1(a b ) of the main text , is shown in fig .
[ figs6 ] .
since the density of states at the fermi energy has a minimum at the gap closing pressure , the maximum in @xmath35 must be attributed to the maximum in the fermi velocity .
the reflectance measured at the sample - diamond interface is approximately its bulk value because we used samples thicker than 10 @xmath76 m .
it is given by @xmath77 where the diamond refractive index @xmath78 is known , and the sample refractive index @xmath64 and extinction coefficient @xmath79 can be obtained from the dielectric function @xmath80 the reflectance data of pb@xmath5sn@xmath6se are fitted to the above equations and the fitting parameters are shown in fig .
[ figs3 ] . the plasma minimum occurs roughly at the screened plasma frequency @xmath47 , which has a maximum according to fig .
2(c ) in the main text .
since fig .
[ figs3](c ) shows a monotonic increase of @xmath33 with pressure , the maximum in the screened plasma frequency must be due to a maximum in @xmath35 .
this is consistent with fig .
[ figs3](a ) .
we also performed high - pressure infrared reflectance and absorbance measurements on pbte .
the results are shown in fig .
[ figs4 ] .
the far - infrared reflectance data are fitted to eqs .
( s7 ) and ( s8 ) . @xmath33
[ figs4](d ) ] increases monotonically but drops at about 5.6 gpa due to the structural phase transition .
the second structure persists even when the pressure is released to below 5.6 gpa , resulting in a hysteresis loop in @xmath33 . although @xmath35 [ fig . [
figs4](b ) ] has a maximum at about 5.3 gpa , it is too close to the structural phase transition to conclude whether a topological phase transition is the origin . the absorption edge in the room - temperature infrared absorbance [ fig .
[ figs4](e ) ] systematically redshifts up to 5.3 gpa and then dramatic changes its shape at 5.7 gpa , associated with the structural phase transition . in the same pressure range , the absorption peak in
the far infared grows and saturates , followed by a precipitous drop due to the structural phase transition . to separate the intraband and interband transitions , we cooled the sample to 30 k and measured the mid - infrared absorbance , shown in fig .
[ figs4](f ) .
the absorption edge first redshifts from 1.1 to 3.2 gpa ; it then tilts towards higher photon energy at 3.9 and 4.2 gpa , followed by a structural phase transition at 4.5 gpa .
this is similar to that observed in pbse , serving as another example of pressure - induced band gap closing and reopening .
the calculated electronic band structure of pbse near the l point of the brillouin zone shows a mexican hat shape at sufficiently high pressure , as shown in fig .
[ figs5 ] at @xmath81 . | the characteristics of topological insulators are manifested in both their surface and bulk properties , but the latter remain to be explored . here
we report bulk signatures of pressure - induced band inversion and topological phase transitions in pb@xmath0sn@xmath1se ( @xmath2 , 0.15 , and 0.23 ) .
the results of infrared measurements as a function of pressure indicate the closing and the reopening of the band gap as well as a maximum in the free carrier spectral weight .
the enhanced density of states near the band gap in the topological phase give rise to a steep interband absorption edge .
the change of density of states also yields a maximum in the pressure dependence of the fermi level .
thus our conclusive results provide a consistent picture of pressure - induced topological phase transitions and highlight the bulk origin of the novel properties in topological insulators .
topological insulators ( tis ) represent a new phase of matter that completely defies landau s spontaneous symmetry breaking framework @xcite .
their insulating bulk is topologically nonequivalent to the vacuum , giving rise to metallic surface states .
such exotic properties come with a conundrum : the topological invariants characterizing the topological property are defined using the bulk electronic wavefunctions , but nevertheless remain elusive to bulk experimental probes . to date , the proof of the existence of tis is based on the detection of the metallic surface states @xcite . exploiting the surface spin texture , spin - resolved angle - resolved photoemission spectroscopy provides the only measure of the topological invariants in 3d tis @xcite . despite the great amount of research activity in this field , the bulk properties of tis that are intimately tied to the nontrivial topology are mostly unexplored .
establishing the bulk properties can complete our understanding of tis and facilitate their identification .
an effective approach for establishing the bulk signatures of tis is to follow the evolution of characteristic features , starting from the trivial insulating state through the topological phase transition ( tpt ) and into the ti phase . applying pressure offers a particularly attractive method for controllably driving a material through such a transition .
generally , a hallmark of a tpt in a non - interacting system is band inversion @xcite : the bulk band gap closes at the phase transition and reopens afterwards , inverting the characters of the bottom conduction band and top valence band .
although the resulting change in the bulk band structure is expected to be dramatic , detecting and understanding the associated experimental signatures are surprisingly challenging .
for example , angle - resolved photoemission spectroscopy is not compatible with pressure tuning nor is it sensitive to the bulk .
previously , two groups @xcite reported investigations for a pressure - induced tpt in bitei @xcite , but reached different and actually contradictory conclusions . in one case @xcite , an observed maximum in the free carrier spectral weight
was interpreted as strong evidence for a tpt . in the other @xcite , a monotonic redshift of the interband absorption edge
was interpreted to indicate the absence of such a transition .
obviously , resolving this contradiction is necessary for our understanding of tis to move forward .
here we clarify this current controversy by demonstrating bulk signatures of a pressure - induced band inversion and thus a tpt in pb@xmath0sn@xmath1se ( @xmath2 , 0.15 , and 0.23 ) .
a maximum in the free carrier spectral weight is reconfirmed in this system and is possibly a generic feature of pressure - induced tpts when bulk free carriers are present .
the absorption edge initially redshifts and then blueshifts under pressure , but only when its overlap with the intraband transition is suppressed .
extra evidence for the tpt is uncovered , including a steeper absorption edge in the topological phase compared to the trivial phase and a maximum in the pressure dependence of the fermi level .
the tpts in pb@xmath0sn@xmath1se imply the creation of 3d dirac semimetals at the critical pressure , serving as a route for pursuing weyl semimetals @xcite .
the robust bulk signatures of tpts identified here are expected to be useful for exploring a variety of candidate pressure - induced tis @xcite and even topological materials with electronic correlation @xcite .
lead chalcogenides are candidate topological crystalline insulators ( tcis ) under pressure @xcite , with the role of time - reversal symmetry in tis replaced by crystal symmetries @xcite .
similar to tis , tcis nontrivial band topology is associated with an inverted band structure , shown in snte @xcite , pb@xmath3sn@xmath4te @xcite , and pb@xmath5sn@xmath6se below 100 k @xcite .
these narrow - gap semiconductors crystallize in the rocksalt structure and share simple band structures ideal for investigating bulk characteristic of tpts . at ambient condition ,
pb@xmath7 ( @xmath7 = s , se , or te ) has a direct band gap at the l point of the brillouin zone , with the l@xmath8 ( l@xmath9 ) character for the bottom conduction band ( top valence band ) @xcite . band inversion is known to be induced in pb@xmath0sn@xmath10 ( @xmath7 = se and te ) by doping @xcite or , in the series of pb - rich alloys , by cooling @xcite , although the connection to band topology was only recently recognized @xcite .
pressure - induced band inversion in pbse and pbte has been proposed on theoretical grounds @xcite especially in the context of tpts @xcite , but experimentally it has not been firmly established @xcite . in this work we present a systematic infrared study of pressure - induced band inversion in pb@xmath0sn@xmath1se .
samples with nominal @xmath2 , 0.15 , and 0.23 were synthesized by a modified floating zone method .
hall effect measurements determined a hole density of roughly 10@xmath11 @xmath12 for pbse at room temperature .
high - pressure experiments were performed using diamond anvil cells at beamline u2a of national synchrotron light source , brookhaven national laboratory .
samples in the form of thin flakes ( @xmath13 m ) were measured in transmission while thicker ( @xmath14 m ) pieces were measured in reflection .
details of the measurement procedures are described elsewhere @xcite . for comparison with the experiment , an _ ab initio _ method based on the wien2k package @xcite
was used to simulate the pressure effects on pbse .
we combined the local spin density approximation and spin - orbit coupling for the self - consistent field calculations , adjusting only the lattice parameter @xmath15 to mimic pressure effects .
the mesh was set to 46@xmath1646@xmath1646 k - points and r@xmath17k@xmath18 , where r@xmath19 is the smallest muffin tin radius and k@xmath20 the plane wave cutoff .
the effect of doping was considered in the self - consistent calculation according to the experimental hole density . while the calculation does not yield the exact lattice parameter at which the tpt occurs
, we found the results to be in qualitative agreement with our experimental observations . along the l-@xmath21 and l - w directions of the brillouin zone .
@xmath22 is the experimental lattice constant in the zero temperature limit and at ambient pressure @xcite .
the tpt occurs at @xmath23 .
the dashed lines indicate the fermi level @xmath24 for a hole density of @xmath25 @xmath12 .
( c ) the direct band gap @xmath26 at the l point , absolute value of the fermi level @xmath27 relative to the top valence band , and @xmath28 [ roughly the energy threshold for direct interband transitions in the presence of free carriers , as indicated by the arrow in ( a ) ] as a function of @xmath29 .
( d ) fermi velocity @xmath30 along the l-@xmath21 and l - w directions as a function of @xmath29 for @xmath25 @xmath12 . ]
lead chalcogenides exhibit structural phase transitions under pressure @xcite .
we determined that , at room temperature , pb@xmath0sn@xmath1se maintains the ambient - pressure structure up to 5.1 , 4.0 , and 2.9 gpa for @xmath2 , 0.15 , and 0.23 , respectively @xcite . in the following we focus on the ambient - pressure structure and investigate pressure - induced tpts .
we began by revealing a maximum in the pressure - dependence of the bulk free carrier spectral weight .
in the same manner as for bitei @xcite , this serves as a signature of band inversion .
though the free carrier response of the bulk has made analysis of the surface behavior challenging , it actually provides a sensitive and convenient probe of band inversion and thus tpts .
for doped semiconductors with simple bands , the low - frequency dielectric function for intraband transitions can be described by @xmath31 , where @xmath32 is the photon frequency , @xmath33 the high - energy dielectric constant , @xmath34 the drude spectral weight ( @xmath35 the bare plasma frequency ) , and @xmath36 the electronic scattering rate .
the drude weight @xmath34 connects with the band dispersion through the fermi velocity @xmath30 , @xmath37 , because @xmath38 ( in general as a tensor ) for a single band is given by @xcite @xmath39 where @xmath40 is the crystal momentum , @xmath41 the band dispersion , @xmath42/\hbar$ ] the @xmath43th component of the bloch electron mean velocity , and @xmath24 the fermi energy .
the measured reflectance has a minimum near the zero crossing in the real part of @xmath44 , called the plasma minimum , located at @xmath45 @xcite .
the plasma minimum is universally observed in pb@xmath0sn@xmath46se for @xmath2 , 0.15 , and 0.23 ( see fig .
[ fig2 ] ) . upon increasing pressure , it initially blueshifts and then redshifts , indicated by the dashed lines in fig .
[ fig2 ] .
since the phase space for intraband transitions reaches a minimum at the gap - closing pressure and @xmath33 increases monotonically under pressure @xcite , the maximum in @xmath47 must be attributed to @xmath30 going through a maximum near the critical pressure of the tpt [ fig.[fig1](d ) ] .
the expression for @xmath44 provides excellent fits to our data , exemplified for pb@xmath5sn@xmath6se in fig .
[ fig2](d ) and quantifying the maximum in the pressure dependence of @xmath35 [ fig .
[ fig2](e ) and ref .
@xcite ] .
such a maximum was also observed in bitei @xcite and is likely generic in pressure - induced tpts when a significant carrier density exists .
having established the tpts , we now turn to the interband transitions to address the controversy over the absorption edge @xcite .
fig .
[ fig3](a ) shows the absorbance [ defined as @xmath48log(transmittance ) ] of pbse at 298 k for pressures up to @xmath495.1 gpa , at which point a structural phase transition occurs @xcite . to assist the discussion ,
we roughly define three photon energy regions , illustrated in fig .
[ fig3](a ) .
region i is dominated by the intraband transition , but is outside the spectral range for the instrument used in the measurement .
region iii hosts the majority of the absorption edge , defined as the steep rising part due to the onset of interband transitions and indicated by the arrow in fig .
[ fig3](a ) , which is expected to redshift and then blueshift across the tpt . the absorption edge shown in region iii of fig .
[ fig3](a ) redshifts monotonically under pressure , indicating band gap closing , but not reopening .
above 1.3 gpa , the initial rising part of the absorption edge which determines the band gap moves into regions ii and i , overlapping significantly with the intraband transition . close inspection of region ii in fig .
[ fig3](a ) reveals band gap reopening . despite of the overlap with the conspicuous tail of the intraband transition ,
the interband absorption edge in region ii shows a clear change of slope : it becomes steeper as the pressure is increased to 5.1 gpa , possibly due to the band gap reopening . for a more conclusive observation of the band gap reopening , we cooled the sample to 70 k in order to reduce the electronic scattering rate @xmath36 , so that the intraband transition peak became narrower and overlapped less with the interband absorption edge . as shown in fig .
[ fig3](b ) , the absorption edge systematically tilts towards higher photon energy from 2.4 to 4.2 gpa , suggesting that the band gap monotonically increases .
finally , a structural phase transition occurs at 4.5 gpa ( at 70 k ) , causing a dramatic overall decrease of absorbance .
pressure - induced band gap closing and reopening were also observed in pbte at low temperature @xcite .
the above discussion illustrates the complexity of analyzing the interband absorption edge to identify gap closure and band inversion at a tpt .
considering the apparent monotonic increase of spectral weight in region iii [ see fig .
[ fig3](a ) ] as a function of pressure , one might conclude that a tpt had not occurred .
but the key signature of gap closure is obscured by overlap with the intraband absorption , as well as by thermal broadening .
this can be circumvented by cooling the material , revealing both the redshift and then blueshift of the absorption edge as the band gap closes and then reopens .
the situation for bitei is even more complicated due to the rashba splitting and the additional optical transitions among the split subbands @xcite .
cooling does not alleviate this complication .
thus , inferring how the band gap changes in that and similar materials from measurements to sense the absorption edge is not practical . in the rest of this letter , we present two more signatures of pressure - induced tpts in pbse , namely a steeper absorption edge in the topological phase and a maximum in the pressure dependence of the fermi level . the absorption edge becomes steeper after the tpt , distinguishing the topological phase from the trivial phase .
such behavior is clearly observed in fig .
[ fig3](b ) and confirmed by the calculated optical conductivity shown in fig . [ fig3](c ) .
the results emphasize the hybridization nature of the band gap in a ti ( or tci ) as illustrated in fig .
[ fig3](d ) , qualitatively different from that in a trivial insulator . as demonstrated by the evolution of the band structure across a tpt close to the direct band gap , shown in fig .
[ fig1](a b ) , before the tpt , pressure suppresses the band gap and transforms the band dispersion from a near - parabolic shape to almost linear . after the tpt ,
the band dispersion briefly recovers the near - parabolic shape and then flattens .
[ at even higher pressure , it develops a mexican - hat feature @xcite similar to that illustrated in fig .
[ fig3](d ) . ]
the flat band makes the joint density of states just above the band gap much greater than that of an ordinary insulator with the same band gap size , yielding a steeper absorption edge .
it also gives rise to van hove singularities that differ from the typical ones @xcite , shown as peaks in the optical conductivity for @xmath50 and 1.00 in fig .
[ fig3](c ) . such a peak feature
was previously observed ( although unexplained ) in the bi@xmath51te@xmath51se ti material with low free carrier density @xcite , but is absent in our infrared absorbance data shown in fig .
[ fig3](a b ) , possibly for two reasons . first
, the peak only appears deep in the topological phase , which requires a high pressure that in reality causes a structural phase transition .
second , the burstein - moss effect ( see the next paragraph ) in our sample precludes optical transitions connecting the states near the top valence band and the bottom conduction band and thus the observation of van hove singularities . the dashed line in fig . [ fig3](c )
demonstrates that holes with a density of 10@xmath11 @xmath12 completely smears the sharp peak .
lastly , our calculation shows a maximum in the pressure - dependence of the fermi level @xmath27 at the gap - closing pressure [ fig .
[ fig1](c ) ] , which can be measured from shubnikov - de haas oscillations @xcite to support the tpt .
this maximum in @xmath27 happens because the density of states near the top valence band diminishes as the band dispersion becomes linear @xcite , pushing the fermi level away from the top valence band to conserve the phase space for the holes .
this effect also manifests in the infrared spectra , although @xmath24 can not be easily determined from them .
when free carriers are present , the band gap associated with the absorption edge is not the true band gap in the electronic band structure . as illustrated in fig .
[ fig1](a ) , the holes shift the fermi level to below the top valence band , making the energy threshold for direct interband transitions approximately @xmath28 , known as the burstein - moss effect .
the absorption edge characterizes @xmath28 instead of @xmath26 .
the combined pressure effects on @xmath26 and @xmath24 retain a minimum in @xmath28 , but the corresponding pressure could be different from the critical pressure for band gap closing , shown in fig .
[ fig1](c ) .
moreover , the absorption edge never redshifts to zero photon energy even when @xmath52 .
the burstein - moss effect adds further difficulty to the identification of tpts using the absorption edge . to summarize ,
we have established bulk signatures of pressure - induced band inversion and thus topological phase transitions in pb@xmath0sn@xmath1se ( @xmath2 , 0.15 , and 0.23 ) .
infrared reflectance shows a maximum in the bulk free carrier spectral weight near the gap - closing pressure .
the interband absorption edge tracks the change of the band gap across the topological phase transition , however the free carriers complicate the picture due to the overlap with the intraband transition and the shift of the fermi level . the absorption edge becomes steeper in the topological phase due to the hybridization nature of the band gap in topological insulators .
a maximum in the pressure dependence of the fermi level is also expected .
these robust bulk features complement the surface - sensitive techniques and serve as a starting point to investigate topological phase transitions in more complicated systems .
we are grateful for helpful discussions with c. c. homes .
this work was supported by the u. s. department of energy through contract de - ac02 - 98ch10886 at bnl .
the use of u2a and x17c beamline was supported by nsf ( dmr-0805056 ; ear 06 - 49658 , compres ) and doe / nnsa ( de - fc03 - 03n00144 , cdac ) .
99 m.z .
hasan and c.l .
kane , http://link.aps.org/doi/10.1103/revmodphys.82.3045[rev .
mod .
phys . * 82 * , 3045 ( 2010 ) ] .
x.l .
qi and s.c .
zhang , http://link.aps.org/doi/10.1103/revmodphys.83.1057[rev .
mod .
phys . * 83 * , ( 2011 ) ] .
d. hsieh , y. xia , l. wray , d. qian , a. pal , j. h. dil , j. osterwalder , f. meier , g. bihlmayer , c. l. kane , y. s. hor , r. j. cava , and m. z. hasan , http://www.sciencemag.org/content/323/5916/919.full[science * 323 * , 919 ( 2009 ) ] .
s .- y .
xu , y. xia , l.a .
wray , s.jia , f. meier , j.h .
dil , j. osterwalder , b. slomski , a. bansil , h. lin , r.j .
cava , and m.z .
hasan , http://dx.doi.org/10.1126/science.1201607[science * 332 * , 560 ( 2011 ) ] .
h. zhang , c .- x .
liu , x .-
l .
qi , x. dai , z. fang , and s .- c .
zhang , http://www.nature.com/nphys/journal/v5/n6/full/nphys1270.html[nat .
phys . 5 , 438 ( 2009 ) ] .
x. xi , c. ma , z. liu , z. chen , w. ku , h. berger , c. martin , d.b .
tanner , and g.l .
carr , http://prl.aps.org/abstract/prl/v111/i15/e155701[phys .
rev .
lett . * 111 * , 155701 ( 2013 ) ] .
m.k .
tran , j. levallois , p. lerch , j. teyssier , a.b .
kuzmenko , g. auts , o.v .
yazyev , a. ubaldini , e. giannini , d. van der marel , and a. akrap , http://arxiv.org/abs/1305.5485[phys .
rev .
lett . * 112 * , 047402 ( 2014 ) ] .
m.s .
bahramy , b .- j .
yang , r. arita , and n. nagaosa , http://www.nature.com/ncomms/journal/v3/n2/full/ncomms1679.html[nat .
commun . * 3 * , 679 ( 2012 ) ] .
s. chadov , x. l. qi , j. kbler , g.h .
fecher , c. felser , and s.c .
zhang , http://www.nature.com/nmat/journal/v9/n7/full/nmat2770.html[nat .
mater .
* 9 * , 541 ( 2010 ) ] .
b. sa , j. zhou , z. song , z. sun , and r. ahuja , http://prb.aps.org/abstract/prb/v84/i8/e085130[phys .
rev .
b * 84 * , 085130 ( 2011 ) ] .
w. li , x .- y .
wei , j .- x .
zhu , c.s .
ting , and y. chen , http://prb.aps.org/abstract/prb/v89/i3/e035101[phys .
rev .
b * 89 * , 035101 ( 2014 ) . ]
b .- j .
yang and y.b .
kim , http://prb.aps.org/abstract/prb/v82/i8/e085111[phys .
rev .
b * 82 * , 085111 ( 2010 ) ] . c.h .
kim , h.s .
kim , h. jeong , h. jin , and j. yu , http://prl.aps.org/abstract/prl/v108/i10/e106401[phys .
rev .
lett . * 108 * , 106401 ( 2012 ) ] .
e.v .
gorbar , v.a .
miransky , and i.a .
shovkovy , http://journals.aps.org/prb/abstract/10.1103/physrevb.89.085126[phys .
rev .
b * 89 * , 085126 ( 2014 ) ] .
p. barone , t. rauch , d. di sante , j. henk , i. mertig , and s. picozzi , http://prb.aps.org/abstract/prb/v88/i4/e045207[phys .
rev .
b * 88 * , 045207 ( 2013 ) ] . l. fu , http://prl.aps.org/abstract/prl/v106/i10/e106802[phys .
rev .
lett . * 106 * , 106802 ( 2011 ) ] .
t.h .
hsieh , h. lin , j.w .
liu , w.h .
duan , a. bansil , and l. fu , http://www.nature.com/ncomms/journal/v3/n7/full/ncomms1969.html[nat . commun . * 3 * , 982 ( 2012 ) ] .
y. tanaka , z. ren , t. sato , k. nakayama , s. souma , t. takahashi , k. segawa , and y. ando , http://www.nature.com/nphys/journal/v8/n11/full/nphys2442.html[nat .
phys .
* 8 * , 800 ( 2012 ) ] .
s .- y .
xu , c. liu , n. alidoust , m. neupane , d. qian , i. belopolski , j.d .
denlinger , y.j .
wang , h. lin , l.a .
wray , g. landolt , b. slomski , j.h .
dil , a. marcinkova , e. morosan , q. gibson , r. sankar , f.c .
chou , r.j .
cava , a. bansil , and m.z .
hasan , http://www.nature.com/ncomms/journal/v3/n11/full/ncomms2191.html[nat .
commun . * 3 * ( 2012 ) ] .
p. dziawa , b.j .
kowalski , k. dybko , r. buczko , a. szczerbakow , m. szot , e. usakowska , t. balasubramanian , b. m. wojek , m. h. berntsen , o. tjernberg , and t. story , http://www.nature.com/nmat/journal/v11/n12/full/nmat3449.html[nat . mater . * 11 * , 1023 ( 2012 ) ] .
j.o .
dimmock , i. melngailis , and a.j .
strauss , http://prl.aps.org/abstract/prl/v16/i26/p1193_1[phys .
rev .
lett .
* 16 * , 1193 ( 1966 ) ] .
a.j .
strauss , http://prola.aps.org/abstract/pr/v157/i3/p608_1[phys . rev . * 157 * , 608 ( 1967 ) ] .
t.c .
harman , a.r .
calawa , i. melngailis , and j.o .
dimmock , http://dx.doi.org/10.1063/1.1652675[appl .
. lett .
* 14 * , 333 ( 1969 ) ] .
a. svane , n.e .
christensen , m. cardona , a.n .
chantis , m. van schilfgaarde , and t. kotani , http://prb.aps.org/abstract/prb/v81/i24/e245120[phys .
rev .
b * 81 * , 245120 ( 2010 ) ] .
g. martinez , http://prb.aps.org/abstract/prb/v8/i10/p4693_1[phys .
rev .
b * 8 * , 4693 ( 1973 ) ] . b.a .
akimov , r.s .
vadkhva , s.m .
chudinov , sov .
. semicond .
* 12 * , 1146 ( 1978 ) .
p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , and j. luitz , _
wien2k , an augmented plane wave _ + _ local orbitals program for calculating crystal properties _ ( technische universitt wien , vienna , austria , 2001 ) . s.v .
ovsyannikov , v.v .
shchennikov , a.y .
manakov , a.y .
likhacheva , i.f .
berger , a.i .
ancharov , and m.a .
sheromov , http://onlinelibrary.wiley.com/doi/10.1002/pssb.200672565/abstract[phys .
status solidi b * 244 * , 279 ( 2007 ) ] .
see supplemental material for details of the structural phase transition , derivation of the drude weight , density of states , parameters from the fitting of reflectance , data for pbte , and additional band structure data .
h. preier , http://link.springer.com/article/10.1007%2fbf00886018[appl . phys . * 20 * , 189 ( 1979 ) . ] p.y .
yu and m. cardona , _ fundamentals of semiconductors _ ( springer , new york , 2010 ) , pp .
306310 . raw data in fig .
[ fig2](a ) below 2.0 gpa and in fig . [ fig3](a b ) contain interference fringes due to the multiple reflections of the infrared beam in the sample and pressure - transmitting medium . for clarity ,
these fringes are removed by fourier transforming the spectra and filtering the associated peaks . the spectral features for absorbance greater than 2.2 are likely artifacts due to stray light leakage ,
but they do not affect the analysis discussed here .
x.g .
he , x. xi , and w. ku ( unpublished ) .
p. di pietro , f.m .
vitucci , d. nicoletti , l. baldassarre , p. calvani , r. cava , y.s .
hor , u. schade , and s. lupi , http://journals.aps.org/prb/abstract/10.1103/physrevb.86.045439[phys .
rev .
b * 86 * , 045439 ( 2012 ) ] .
a. akrap , m. tran , a. ubaldini , j. teyssier , e. giannini , d. van der marel , p lerch , and c.c .
homes http://journals.aps.org/prb/abstract/10.1103/physrevb.86.235207[phys .
rev .
b * 86 * , 235207 ( 2012 ) ] .
d. vangennep , s. maiti , d. graf , s.w .
tozer , c. martin , h. berger , d.l .
maslov , and j.j .
hamlin , http://arxiv.org/abs/1404.7004[arxiv:1404.7004 ( 2014 ) ] .
* supplemental material * |
in recent years the investigation of distant clusters of galaxies became a key issue for the study of cosmological parameters as well as for structure formation .
in addition , the comparison of the physical properties of distant with nearby galaxy clusters can give useful insight into their evolution .
+ the distant cluster 3c 295 , also known as cl 1409 + 526 has been extensively studied in optical wavelength bands : dressler & gunn ( 1992 ) determined its velocity dispersion to be @xmath8km / sec in a study of seven distant clusters of galaxies .
thimm et al .
( 1994 ) in the search for line emission in galaxies observed the cluster with an imaging fabry - perot interferometer and found the percentage of emission line galaxies in 3c 295 to be @xmath9% .
smail et al .
( 1997 ) presented the results of a weak lensing analysis of 12 distant clusters , among them 3c 295 .
the analysis they present is based on data of the _ hubble space telescope_. + henry & henriksen ( 1986 ) presented the analysis of _ einstein _ data of the cluster .
the authors conclude that there is strong evidence for 3c 295 to host a cooling flow in the center .
+ cooling flows generally occur in clusters in which the cooling time of the hot gas , the intracluster medium ( hereafter icm ) , is less than the age of the cluster , so that the energy loss due to radiation can not be neglected and mechanisms for the compensation must occur .
this was first found by cowie & binney ( 1977 ) , fabian & nulsen ( 1977 ) , and mathews & bregman ( 1978 ) , based on the findings of observations by uhuru ( lea et al .
1973 ) for reviews see fabian et al .
( 1984 , 1991 ) and sarazin ( 1986 , 1988 ) .
+ in recent years a handful of other high redshift clusters also show indications of hosting cooling flows in their centers .
most recently schindler et al .
( 1997 ) reported on rxj1347 - 1145 ( z=0.45 ) having an extremely massive cooling flow with a mass accretion rate exceeding 3000 m@xmath3/y .
fabian & crawford ( 1995 ) reported on the detection of a cooling flow in iras p09104 + 4109 ( z=0.442 ) with a mass accretion rates of about 1000 m@xmath3/y , and edge et al .
( 1994 ) found in zwicky 3146 at z=0.29 a cooling flow with similar strength .
+ in this paper we aim to give more insight into the cooling flow and its dependence on the temperature structure of the icm , the gas and dark matter distribution and the overall dynamical state of 3c 295 .
the outline is therefore as follows : sec.2 gives a brief description of the observation and data handling .
in sec.3 we describe our spatial analysis based on the isothermal @xmath1-model , which is followed by a section describing our way of calculating the x - ray luminosity of the cluster .
the cooling flow analysis is presented in sec.5 .
sec.6 describes the details of our employed mass analysis .
discussion and conclusion follow hereafter . + throughout the paper we assume @xmath10km / sec / mpc , @xmath11 , and @xmath12 .
3c 295 was observed with the rosat hri for 29.6 ksec .
fig.[fig : plot ] shows the resulting image of the cluster .
we only use channels 2 to 9 out of the 15 channels of the instrument to optimize the signal to background ratio as the neglected channels mostly have low or negligible sensitivity ( for details see david et al .
1997 ) . with about 830 source photons within a radius of 8.3 arcmin
we obtain a countrate @xmath13s@xmath14 for the cluster .
the cluster is elongated in north - south direction with an axial ratio of 0.78 ( see below and tab.[tab : beta ] ) .
the emission is centrally peaked , which is a strong indication for the existence of a cooling flow ( see below ) .
there is some isophot twist visible in the contours in fig.[fig : plot ] .
the lines are more compressed in the south - east and more stretched in the north - west
. we will discuss this feature and its significance later on in more detail .
+ there are no other clearly extended sources visible in the vicinity of the cluster as all other emission peaks in fig.[fig : plot ] , which have at least two contour lines , are consistent with point sources we checked whether these sources were possibly extended by determining their individual surface brightness profiles and comparing them to the on - axis point spread function of the rosat hri .
the radial surface brightness profile of the cluster can be seen in fig.[fig : surf ] . to explore physical quantities of 3c 295
we make use of the isothermal @xmath1-model ( cavaliere & fusco - femiano 1976 , 1981 ; sarazin 1986 ) .
this model enables us to convert the brightness profile of a cluster analytically into a density profile of the gas . the one - dimensional surface brightness profile of this model has the following form : @xmath15 which translates into a density profile : @xmath16 where @xmath17 ( the so - called core radius ) , @xmath18 ( the central surface brightness ) , @xmath19 ( the background ) and @xmath1 are all fit parameters .
+ there exists also a two - dimensional form of the isothermal @xmath1-model which takes into account a possible elongation of the cluster .
this model has the representation : @xmath20 @xmath21 we apply in the following the one - dimensional and the two - dimensional isothermal @xmath1-model to the data .
the first model has four fit parameters ( @xmath18 , @xmath17 , @xmath1 , @xmath19 ) , the second eight fit parameters ( @xmath18 , @xmath22 , @xmath23 - two core radii , one for the major axis , one for the minor axis , @xmath1 , @xmath19 , @xmath24 , @xmath25 - the position of the center of the cluster in x and y , and the position angle @xmath26 ) . in the one - dimensional fit we predefine the center of the cluster by sorting the photons into concentric annuli centered on the peak of the x - ray cluster emission . in the two - dimensional case , where we look at pixels instead of annuli , this predefintion is not necessary and the center of the cluster emission is simultanously fitted together with the other parameters . +
the two - dimensional fit we apply here takes into account the effects of the psf of the rosat hri by convolving each tested model with the psf before fitting . for the fitting
we exclude serendipitous sources , which are clearly not connected to the x - ray emission of the cluster .
the positions of these excluded sources can be seen in tab.[tab : exclude ] .
for the one - dimensional fit we bin the data in concentric annuli with a width of 10 arcsec . the results for this fit give a core radius of @xmath27 kpc and a slope parameter of @xmath28 ( 2 @xmath29-errors ) .
the reduced @xmath30 of the fit is 1.03 , which gives a very high confidence in the fit results . the fitted profile can be seen in fig.[fig : surf ] , and details of the other fit parameters are given in tab.[tab : beta ] . reducing the width of the annuli , in order to better resolve the central part does not significantly change the fit parameters and the connected uncertainties .
+ the derived quantities are in good agreement with the results obtained by henry & henriksen ( 1986 ) , who analyzed einstein data and derived a core radius of @xmath31 and a @xmath32 ( 1 @xmath29-errors ) . for the two - dimensional fit we use an image with a pixel size of @xmath33 , taking into account all pixels located within a radius of 2.5 arcmin from the pointing position at ra=@xmath34 , dec.=@xmath35 ( j2000 ) . + our fitting procedures for both the one - dimensional and the two - dimensional case are based on @xmath30-statistics .
this is ideal for the one - dimensional case , in which we look at concentric annuli centered on the emission peak of the cluster , as there are always enough counts per bin to assure that the poisson distribution in each bin is similar to a gaussian distribution , which is required for @xmath30-fitting .
the decrease in cluster emission outwards in this case is compensated by the increase in area of the annuli . however , in the two - dimensional case , in which we use image pixels all of the same size , this compensation does not occur . here
the counts per pixel , especially in outer regions , are too low to show gaussian behaviour , but instead , show a classical poisson distribution . to overcome the problem here , we apply a small gaussian filter ( @xmath36 ) to the image before fitting .
this gaussian filter changes the poisson statistics into a gaussian shape . for the error determination in each pixel we can fully account for the gaussian filtering by adding the corresponding errors of all contributing pixels quadratically
( the exact formalism of the error determination can be found in neumann & bhringer ( 1997 ) ) . in this way
we change the errors in countrate per pixel without reducing the number of degrees of freedom .
we do not lower the number of degrees of freedom here , as we would in the case of increasing the pixel size of the used image , since we keep all the available spatial information . + to account for the gaussian filtering , which causes the inferred profile of the cluster to become shallower , we also apply the same gaussian filter to the point spread function image with which we convolve the fitted cluster model . to prove that with this procedure we indeed are able to determine the precise shape of the cluster we constructed a simulated cluster image with known properties , similar to 3c 295 , and with an equal exposure time as our analyzed rosat hri pointing .
we apply poisson noise on this image and convolve it with the on - axis psf of the rosat hri .
subsequently we apply the gaussian filter and fit the two - dimensional @xmath1-model to the simulated cluster image , taking into account the ( gaussian filtered ) psf .
the results can be seen in tab.[tab : simul ] .
the intrinsic cluster model properties and the obtained fit parameters agree very well within the errors , giving clear evidence that our method indeed gives correct results . + as @xmath30-statistic does not allow an error of 0 , we define the error of a pixel with value 0 to have an error of 1 ( before filtering ) .
because of this necessary modification , we are , unfortunately , not able to give a confidence level or a reduced @xmath30 value of the fitted elliptical cluster model , and therefore base our error determination on a monte carlo - technique ( see below ) .
our fit results are the following : for the core radii we get @xmath37 kpc and @xmath38 kpc , for the slope parameter we get @xmath39 ( 2@xmath29-results ) .
the results for the other six fit parameters can be seen in tab.[tab : beta ] and the profile can be seen in fig.[fig : surf ] .
one can see that while @xmath1 is relatively unaffected by the psf of the rosat hri the same is not true for the core radius .
the core radius changes when the psf is taken into account and drops from 7.2 arcsec to 4.2 arcsec or 3.3 arcsec , respectively .
however , one should note that the errors of this parameter ( which are all 2 @xmath29 errors ) show a large overlap .
+ the approach of fitting after the application of a gaussian filter in the two - dimensional case has been already successfully applied to other clusters , such as cl0016 ( neumann & bhringer 1997 ) and a2218 ( neumann & bhringer 1998 ) . for the error determination of the 2d - fit parameters we apply a monte - carlo method for which we construct 100 artificial images by adding random poisson noise on the real cluster image and fitting each individual poisson image in the same way as the original image . for the errors we use two times the standard deviation of each fit parameter from the 100 different fit results .
the comparison of the two fits to the original data , the one taking into account the effects of the psf the other one not , can be seen in fig.[fig : surf ] .
in fact , apart from the central region , which has a more peaked emission when taking into account the psf , the differences in the surface brightness profile are rather small .
therefore we conclude that the psf might affect the cooling flow analysis ( see below ) , but not the overall cluster properties at large radii . in order to see whether the difference in the fit results are partly due to ellipticity
, we perform a one - dimensional fit , which takes into account the effects of the psf .
the results are identical to our fit result of the two - dimensional fit , showing that the differences are only due to psf effects . in the above analysis we neglected possible errors in the aspect resolution of the instrument . in order to see whether the psf is degraded because of effects concerning the wobbling of the telescope or the reaquisition of guide stars we compared the theoretical psf of the telescope and hri with point sources in the field - of - view .
the comparison of theoretical and actual psf gives somewhat different results , depending on the binning of the actual observed point sources - this effect is most likely due to poisson noise as all point sources are with a count rate well below 0.003 cts / sec very faint . -
the brightest source has about 60 source counts ( see tab.[tab : exclude ] ) .
however , allsurface brightness profiles of serendipitous sources are quite consistent with the theoretical psf ( see above ) .
if the aspect resolution is slightly degraded , it is not possible to assess the error it introduces and it is also not possible to correct for it , due to the lack of bright sources in the pointing ( for corrections concerning blurring see for example morse 1994 and harris et al .
this blurring can in principle effect our @xmath1-model fits and the resulting analysis .
however , we do not expect the effects to play an important role . in order to search for possible substructure in the cluster
we construct a cluster model image from the best fit parameters of our two - dimensional fit , which includes the background ( the model can be seen in fig.[fig : sym ] ) .
subsequently we convolve the model with the psf of the rosat hri and subtract the model from the original data .
the image of the so obtained residuals are shown in fig.[fig : resi ] .
+ in this image we find positive residuals only below a significance level of three @xmath29 above background and cluster emission , so that there is no strong evidence for a perturbation in the cluster potential , like a subgroup falling onto the cluster .
there is a decrement in the residual map in the south - east of the cluster , which could in principle be an indication of some disturbance .
however , it could also be just an indication of a region with higher intrinsic galactic absorption .
unfortunately , surveys of the 21-cm line do not ( yet ) provide the necessary spatial resolution to confirm or rule out a strong gradient in galactic absorption in the line - of - sight of the cluster .
+ the fact that this is the only evidence that the cluster deviates from the applied cluster model weakens the case for a perturbation in the potential . to see whether this decrement might be due to an offset between the true cluster center and cluster model ( which is very unlikely as the peak of emission in fig.[fig : plot ] and the cluster center in fig.[fig : sym ]
coincide very well ) , we shift the center of the subtracted cluster model several arcseconds in the direction nw of the center .
the decrement remains after the shift , indicating that this deficit in emission might indeed be real .
+ finally , we conclude that , apart from this decrement , there is no further evidence for the cluster being dynamically young .
it does not seem that the cluster is suffering or has recently suffered from a major merger phase , which is in agreement with the cluster hosting a cooling flow normally a strong indication for a relaxed cluster .
this fact is strengthend by the high confidence level of the one - dimensional fit , which one does not expect in case of a non - relaxed cluster .
however , we want to stress that our 3c 295 exposure has limited signal - to - noise , and that future x - ray telescopes , such as axaf or xmm will give further information due to higher sensitivity and lower background . to calculate the central electron density of the cluster we employ the results of the isothermal @xmath1-model . using the one - dimensional model as input we calculate a central electron density of @xmath40@xmath41
for the two - dimensional @xmath1-model we use as mean core radius the geometrical mean of @xmath22 and @xmath23 , giving @xmath42 .
the derived @xmath43 in this case is @xmath44@xmath41 .
the errors are hardly dependent on the temperature due to the rosat energy band ( see for example jones & forman 1992 ; bhringer 1994 ) but are dependent on the intrinsic shape parameters of the isothermal @xmath1-model .
the errors are larger for the two - dimensional case , as we take into account the errors of the two core radii , the @xmath1 and the @xmath18 , while for the 1-d case we only take into account the error of the one core radius and @xmath1 .
the rosat / hri source countrate of 3c 295 is @xmath45sec@xmath14 . to determine this countrate we subtract from the original data the background rate determined from the spherical symmetric @xmath1-model fit and exclude serendipitous sources not belonging to the cluster .
+ assuming that the cluster temperature lies between kt=@xmath46 kev , the result by mushotzky & scharf ( 1997 ) based on asca data , we obtain an x - ray rest frame luminosity of @xmath47 kev@xmath48erg / sec in the rosat band , corresponding to a bolometric luminosity of @xmath49erg / sec . the error on the luminosity is a combination of both countrate statistics and the uncertainty in temperature . for the model we use thermal bremsstrahlung , and a galactic hydrogen column density of @xmath50@xmath51 from 21cm - line measurements ( dickey & lockman 1990 ) .
our estimate for @xmath52 is in reasonably good agreement with the result of mushotzky & scharf ( 1997 ) , who obtain an x - ray luminosity of @xmath53 erg / sec , when transformed to @xmath11 .
the discrepancy in the luminosity estimate can have several sources .
first of all the presence of colder gas in the cooling flow region might lower the effective temperature in the rosat band , which is softer than the asca band .
secondly , the discrepancy in the luminosity estimates might be due to the large psf of the asca instruments , leading normally to an underestimate of the countrate .
+ our results are also in agreement with the results of lea & henry ( 1988 ) , who determined an x - ray luminosity of @xmath54kev@xmath55erg / sec on the basis of einstein data .
in central parts of clusters of galaxies the time in which the icm cools from several @xmath56k to temperatures several orders of magnitude less can be lower than the age of the cluster . in these regions
the energy loss due to radiation is therefore not negligible and mechanisms for the compensation must occur ( for details see for example fabian et al .
1991 ) . to give an estimate out to
what radius the energy loss is not negligible one often makes use of the so - called cooling time @xmath57 , which is defined as the enthalpy of the icm divided by the energy loss due to radiation in x - rays , _
i.e. _ @xmath58 here , @xmath59 is defined as cooling rate and is only dependent on metallicity and temperature .
typical values for @xmath59 are in the range of @xmath60erg s@xmath14@xmath61 . inserting our results from the isothermal @xmath1-model together with some estimates on @xmath59 from bhringer & hensler ( 1989 ) ( see also tab.[tab : lambda ] ) , we can calculate the cooling time as function of radius .
the cooling flow radius is then defined as the radius at which the cooling time is equal the cluster formation time , which depends on cosmological parameters . assuming the cooling time to be @xmath62yr , which corresponds to the age of the cluster at a redshift of z=0.46 , we get cooling flow radii corresponding to 60100 kpc , depending on the @xmath1-model and on the temperature we apply . for a comparison with nearby clusters ,
if we neglect the fact that we see the cluster in a younger stage than nearby ones , and assume a cooling time of @xmath63 y , we get cooling flow radii in the range of 70120 kpc . if we assume that the only efficient mechanisms which compensate the energy loss due to radiation in the center of the cluster are mass accretion and cooling of the icm , we can write for the energy balance @xmath64 where @xmath65 is the gravitational potential . the left hand side of this equation is the energy loss due to radiation as function of radius , the right hand side shows the mechanisms of compensation : cooling and accretion .
now , we can insert our results of the isothermal @xmath1-model in equ([equ : balance ] ) and calculate the mass accretion . to determine the gravitational potential we can use the hydrostatic equation @xmath66 assuming that the icm has the same properties as an ideal gas we can , in combination with the isothermal @xmath1-model , calculate the gravitational potential of the cluster and with it its total mass as function of radius , _ i.e. _ @xmath67 where @xmath68 is the total mass within radius @xmath69 .
we employ the isothermal @xmath1-model for the gas density distribution even though the cluster might not be isothermal as the emission in the rosat energy band ( 0.1 - 2.4 kev ) depends only weakly on the temperature of the icm ( see above ) .
we would like to stress here that we use the isothermal @xmath1-model only as fit function to deproject the surface brightness profile into a density profile of the icm and that possible temperature gradients or the presence of multiphase gas can be neglected as the measured count rate in the rosat band hardly depends on cluster temperature .
the use of any other analytical function instead of the @xmath1-model for deprojecting would not increase the reliability of the calculated density profile , as we do not have any information on the temperature distribution .
deprojecting numerically by subtracting shell by shell , as is done for nearby clusters can not be undertaken here , as our data have too poor statistics .
+ we also want to note that our analysis presented here describes and assumes idealized circumstances , for example that either the cooling flow region is isothermal or has a constant temperature gradient .
assuming for now the stronlgy idealized case that the cluster is isothermal , and inserting equ([equ : density ] ) , equ([equ : hydro ] ) and equ([equ : masse ] ) in equ([equ : balance ] ) , we can directly calculate the mass accretion rate from @xmath70 with this equation one can see the dependence of @xmath4 on @xmath69 , @xmath17 , and @xmath1 . in case that the cooling flow radius is smaller than the core radius @xmath17 , or in case that the @xmath1 value is very small ( @xmath71 ) , the mass accretion rate is proportional to @xmath69 , since then the term @xmath72 almost stays constant .
this is in agreement with the findings of other authors , e.g. thomas et al .
( 1987 ) , and fabian et al .
( 1991 ) , neumann & bhringer ( 1995 ) .
however , as in the case of 3c295 the core radius is between 25 kpc and 50 kpc and therefore smaller than the cooling flow radius , and as @xmath73 , the term @xmath72 changes significantly and therefore influences the shape of the mass accretion rate as a function of radius .
the results can be seen in fig.[fig : cf1 ] ( for the isothermal @xmath1-model taking into account the psf ) and fig.[fig : cf2 ] ( for the isothermal @xmath1-model neglecting the effects of the psf ) .
the maximal mass accretion rate is in the range 550900 m@xmath3/y , depending on the assumed temperature . in the isothermal case the mass accretion rates rise with radius up to 50 kpc and then
stay more or less constant .
it is generally found that the temperatures in cooling flow regions are lower than the overall cluster temperature .
often a temperature gradient in the cooling flow region can be observed .
if we assume that there exists a constant temperature gradient in the cooling flow region of 3c 295 , we can rewrite equ([equ : balance ] ) as @xmath74 where @xmath75 is the temperature gradient one infers and @xmath76 is the central temperature . in equ([equ : grad ] ) we take into account that a temperature gradient changes the shape of the gravitational potential ( see equ([equ : masse ] ) ) .
we assume that the gas is still in hydrostatic equilibrium and that the hydrostatic equation equ([equ : masse ] ) is still applicable , assuming a single phase icm .
one way to see this : while the gas is cooling down its density distribution changes accordingly to fulfill equ([equ : masse ] ) . for our calculation of mass accretion rates
we assume the temperature at the center to be 2 kev and at a radius of 120 kpc to be either 5.8 , 7.1 , or 9.2 kev the range of temperatures representing the error in the determination of the overall temperature of 3c 295 by mushotzky & scharf ( 1997 ) . for simplification
we assume that @xmath77 is linear in temperature between 2 to 10 kev .
the adopted values for @xmath59 can be seen in tab.[tab : lambda ] . for intermediate temperatures we interpolate linearly .
the results of our calculations are shown in fig.[fig : cf1 ] , and fig.[fig : cf2 ] , respectively . applying a temperature gradient
lowers the mass accretion rates and changes the shape of the mass accretion rate as a function of radius .
the maximum mass accretion rate occurs close to the core radius of either model , being in the range of 400600 m@xmath3/y .
after its maximum the mass accretion rate drops at larger radii the mass accretion rate at 120 kpc varies between 300500 m@xmath3/y .
already henry & henriksen ( 1986 ) analyzing einstein ipc and hri data suggested that 3c 295 hosts a cooling flow in the center .
the authors calculated a mass accretion rate ( @xmath78 ) of about 145 m@xmath3/y .
the determination of @xmath78 was based on pointlike excess emission over the best fit isothermal @xmath1-model in the center .
as one can see in fig.[fig : surf ] , we do not see excess emission in the center .
the isothermal @xmath1-model coincides very well with the cluster emission in the center .
this suggests two possibilities : the first one being that there was an active point source in the center of the cluster , while it was exposed with the einstein hri , and this source was not anymore active when exposed with the rosat hri .
the second possibility is that because henry & henriksen ( 1986 ) simultanously fitted a point source and an isothermal @xmath1-model to the data , this led to an overestimate of the central point source resulting from an underestimate of the emission from the cluster itself .
a weak indication for this is the fact that henry & henriksen obtain a core radius of @xmath31 and a @xmath32 both values being higher than our best fit results .
however , the overall agreement between einstein and rosat data is very encouraging
. the higher fit values for core radius and @xmath1 obtained by henry & henriksen ( 1986 ) cause the estimated cluster profile to be flatter in the center than in our fitted models . + the cluster of galaxies 3c 295 is known to host a luminous radio source in the center ( akujor et al .
1994 and references therein ) .
the source consists of two radio lobes with a separation of about 5 arcsec .
the extent of the lobes is each about 2 arcsec . in principle
the spatial resolution of the hri is sufficient to resolve these lobes , if they emit in x - rays .
however the poisson statistics , which is the factor which really limitats the spatial resolution here , makes a detection of their separation difficult .
it is also not clear whether these lobes lead to an enhancement in x - rays or diminish the x - ray brightness in this area .
an example of radio lobes dimming the x - ray luminosity in the center of a cluster was shown by bhringer et al .
( 1993 ) for the perseus cluster or schindler & prieto ( 1997 ) in the case of a2634 . +
strong evidence that these sources are not very prominent and do not obscure the x - ray surface brightness profile is the fact that we yield a very high confidence level for the isothermal @xmath1-fit with a reduced @xmath30=1.03 . the good agreement between fit and real data is difficult to explain in the presence of additional sources in the center , whether they enhance or diminish the x - ray brightness , as they affect the overall fit .
+ also a single point source in the center can be ruled out by the same argument .
it seems very unlikely that a central point source would have exactly the `` right '' emission to fit the model .
also , in this unlikely scenario , a central source would affect the profile only out to about 35 arcsec the size of the psf .
as our calculated cooling flow radius is larger than this ( @xmath79 ) , a single central point source can not explain the x - ray peak in the center .
+ another test to see whether a central point source can influence the surface brightness profile in the center of the cluster is to perform an isothermal @xmath1-model fit ( one - dimensional ) neglecting the innermost bin of the radial profile ( see fig.[fig : surf ] ) .
this central bin should contain almost all emission from a central point source , if present .
the result of this fit leads to an even more centrally peaked cluster model , with a best fit result for the core radius of about 1.5 arcsec , a @xmath800.54 and a very high central surface brightness of @xmath81sec@xmath14 arcsec@xmath82 .
however , this fitted model has much larger error bars than the original one - dimensional fit , and the uncertainties in core radius and @xmath1 of the original fit are very well encompassed in the error estimates of the modified fit .
+ putting this all together , we conclude that if there are other individual x - ray sources present in the center of the cluster , then their effect on the cooling flow analysis is neglible , and they are unlikely to lead to a dramatic overestimate of the mass accretion rates .
to determine the total and the gas mass of the cluster we utilize the results of the isothermal @xmath1-model fitting for the gas density profile .
we assume that the temperature throughout the whole cluster lies between 5.8 - 9.2 kev , which are the results obtained by mushotzky & scharf ( 1997 ) analyzing asca data . for the mass determination
we assume spherical symmetry and hydrostatic equilibirum and under these assumptions we can use equ([equ : masse ] ) .
these assumptions are justified and lead to reliable estimates for the mass determination of clusters of galaxies as shown by schindler ( 1996 ) and evrard et al .
( 1996 ) , who analyzed hydrodynamic simulations under these assumptions and compared the obtained results with the intrinsic properties . for the determination of the total mass
we perform a monte - carlo analysis ( neumann & bhringer 1995 ) , which calculates possible random temperature variations within the above given limits . for the stepwidth of these temperature variations we use 300 kpc . as the uncertainties in temperature by far exceed the uncertainties in the density profile we neglect here the errors of the isothermal @xmath1-model , which introduce errors in the density distribution of the gas .
+ as a smoothing parameter , to avoid strong and unphysical oscillations of the temperature profiles , we use @xmath83 kev , which means that each temperature at a certain radius has to lie within 0.3 kev of the adjacent step radius . the temperature of the next inner step radius t@xmath84 kev .
the results can be seen in fig.[fig : masse ] and tab.[tab : masse ] .
calculating the total mass of the cluster we get m@xmath85m@xmath3 at a radius of 2.1 mpc .
the result is slightly dependent on the employed @xmath1-model .
for the gas to total mass ratio we get m@xmath7/m@xmath86 for @xmath87 and m@xmath7/m@xmath88 for @xmath89 .
a summary of the results is given in tab.[tab : masse ] .
despite the fact that 3c 295 hosts a cooling flow , we do not see evidence for central excess emission above the fitted isothermal @xmath1-model as seen in almost all other cooling flow clusters .
normally , fitting an isothermal @xmath1-model to the x - ray data of these kind of clusters leads to results which underestimate the central emission .
however , most of these studies are based on nearby clusters , which can be traced out to larger radii than 3c 295 .
we can trace the x - ray emission of 3c 295 only out to a radius of 2.0 - 2.5 arcmin ( see fig.[fig : surf ] ) , which corresponds to 900 - 1100 kpc .
therefore we are unable to see the outer parts of the cluster and our fit is only based on the central parts .
fits of this kind , where one only takes into account the center of a cooling flow cluster , generally lead to smaller core radii and @xmath1 s .
our values for core radius and @xmath1 are also relatively small , when compared to other clusters , which generally have core radii ranging between several 100 kpc and @xmath1 s , which lie between 0.6 and 0.9 for rich and x - ray luminous clusters of galaxies . therefore we conclude that the reason we do not see a central excess lies solely in the fact that we see and fit only the central part of 3c 295 . for an anology ,
we more or less see only the tip of the iceberg .
+ future x - ray instruments such as aboard on axaf or xmm , which have higher sensitivity and lower background than the rosat / hri will enable us to trace distant clusters out to much larger radii .
we determine two different isothermal @xmath1-models in our spatial analysis : one takes into account the effects of the psf and ellipticity , and gives a @xmath89 and a small core radius ( @xmath90 arcsec ) , while in the other one neglects psf effects which gives @xmath87 with a larger value for the core radius ( @xmath91 arcsec ) .
+ comparing the results of the subsequent cooling flow analysis of the two models , one can see that in the @xmath87 case we overestimate the cooling flow radius and the mass accretion rates ( see fig.[fig : cf1 ] and fig.[fig : cf2 ] ) .
the reason why we determine a larger cooling flow radius when using the @xmath87 model instead of the @xmath89 model lies in the fact that the inferred gas density profile shows a less steep drop in the first case .
this comes from the fact that the psf makes the appearance of the profiles more shallow .
+ even though the calculated central electron density is lower in the @xmath87 case , the densities of the gas are equal in both models at a radius of 60 kpc , and at larger radii the @xmath87 model gives higher values for the gas density than the @xmath89 case .
the consequence is that the cooling times in the @xmath87 case at radii larger than 60 kpc are smaller than the ones of the other model , due to the fact that @xmath92 . as the cooling flow region
is determined by its cooling time being less than a certain time interval , the cooling flow radius is , of course , bigger in the @xmath87 case .
the overestimate of the cooling flow radius is of the order of 20% , but does not have a big influence on the resulting maximal mass accretion rate , as at outer regions the mass accretion rate stays almost constant .
+ the overestimate of the mass accretion rates in the @xmath87 case again results from a shallower profile , in this case from the gravitational potential . a shallower potential stands for a smaller potential gradient and this means that more gas must flow into the center to compensate the energy loss due to radiation ( see equ([equ : balance ] ) ) .
the overestimate of the mass accretion rate due to psf effects is of the order of 1025% , which is not very high given all the other uncertainties of the physical properties , which enter the determination of the mass accretion rates .
introducing a temperature gradient in the center of the cluster reduces the resulting mass accretion rates ( see fig.[fig : cf1 ] and fig.[fig : cf2 ] ) .
this shows how sensitive the determination of mass accretion rates is on the temperature structure of the icm . with a temperature gradient
the gas has another mechanism to compensate for the energy loss due to radiation ( we assume here the case of a homogenous cooling flow ) : while the gas is falling into the center it not only looses potential energy but also kinetic energy due to cooling .
however , assuming that the hydrostatic equation is valid in this region , and assuming that multiphases in the gas are negligible , a temperature gradient also causes the gravitational potential to be more shallow ( see equ.([equ : masse ] ) ) . due to this shallower potential
the energy supported by loss of potential energy of the gas falling inwards is not as efficient as in the isothermal case .
however , as the overall mass accretion rates drop , the effect of cooling compensates easily the effect of a shallower gravitational potential .
+ we only want to stress here that our assumed models for the cooling flow are simple , as we , for example , neglect the presence of a multiphase gas in the cooling flow region .
however , as 3c 295 is a very distant cluster , a fact which makes a more detailed analysis of the cooling flow almost impossible at the moment , we only want to give an idea of what mechanisms occur in the cooling flow region of this cluster , and how sensitive the analysis is to the observational uncertainties . comparing the determined mass accretion rate and the calculated cooling flow radius with other clusters , for example the sample of white et al .
( 1997 ) , one can see that 3c 295 is relatively outstanding .
it shows a very high mass accretion rate for its size of cooling flow radius . following the fit results of white et al .
( 1997 ) , the cluster should have a mass acccretion rate of about 60 m@xmath3/y with a cooling flow radius of roughly 100 kpc alternatively , with its mass accretion rate it should have a cooling flow radius of about 200 kpc .
however , the difference between the calculation of white et al . and here is that we neglect inhomogenities in the cooling flow , in which material can cool out completely and provides additional energy resources for radiation .
but it is very unlikely that this effect causes the mass accretion rate to drop by a factor of ten .
the discrepancy lies more likely in the fact that white et al .
look at clusters at a redshift z@xmath93 , while 3c 295 lies at z=0.46 .
it might be that the discrepancy is some sort of evolutionary effect .
this idea is strengthened by another high redshift cooling flow cluster rxj1347 - 1145 ( schindler et al .
1997 ) at z=0.45 with a mass accretion rate of @xmath94m@xmath3/y , and a cooling flow radius of 200 kpc , which also does not fit in the relation of white et al .. also the results of two other distant clusters , zwicky 3146 ( z=0.291 ) and abell a1835 ( z=0.252 ) by edge et al .
( 1994 ) and allen et al .
( 1996a ) show mass accretion rates also somewhat too large for their cooling flow radii to match the result by white et al .. it is therefore important to look at other high redshift cooling flow clusters , to see whether this is indeed an effect depending on redshift , or whether it is just that these clusters are exceptional in their properties .
the determined gas to total mass ratio in 3c 295 is m@xmath7/[email protected] similar to the findings in other clusters . as the gas to total
mass ratio is a limit on the baryon fraction in clusters , which is thought to be representative for the universe in total ( based on numerical simulations , see for example evrard 1990 ; cen & ostriker 1993 ) , this approach can be used to measure the fraction of baryon to dark matter density in the entire universe ( the mass in galaxies in clusters is only a small fraction of the mass of the icm ) . as studies of primordial nucleosynthesis give precise values of @xmath96 ( the ratio of baryon to critical density of the universe ) , with @xmath97 ( see for example walker et al .
1991 ) this can be directly used to determine @xmath98 , the total density of the universe , as shown by briel et al .
( 1992 ) and white et al .
( 1993 ) in calculating @xmath99 .
the determined values of @xmath98 range between 0.20.4 , and give strong evidence for a low density universe .
our values of the gas to total mass ratio of 3c 295 suggest @xmath100 .
the errors in the determination of @xmath101 come almost exclusively from the uncertainties in @xmath102 .
fig.[fig : masse ] shows the total mass profile of 3c 295 .
+ to be able to compare our results for the mass with the weak lensing study by smail et al .
( 1997 ) ( -the lensing approach determines the total mass along the line - of - sight within a certain radius ) , we also show the ( along the line - of - sight ) projected mass profile .
we project out to a radius of 3 mpc .
+ there is some discrepancy between the mass results from the lensing analysis and from the hydrostatic equation used here .
the lensing result gives a higher mass .
the values differ by a factor of less than two .
both estimates have relatively large error bars , which , however , do not overlap .
+ where can this discrepancy come from ? partly it can be due to the fact that the mass inferred from x - rays is biased downward due to the cooling flow .
the cooling flow in the central parts of the cluster could lower the measured overall temperature by asca due to cooled gas .
also the small values for @xmath1 and core radius , typical for a cluster with a central cooling flow might bias the value for the inferred mass downward - the mass calculated from the hydrostatic approach is proportional to @xmath1 . and finally the fact that we do not see the outer regions of the cluster in x - rays could systematically lower the result of our mass determination . on the other hand
the mass determined with the weak lensing approach is dependent on the mean redshift of the lensed galaxies . while this is not a strong effect for nearby clusters it plays an important role for distant clusters such as 3c 295 .
smail et al .
( 1997 ) used an approach with a mean redshift of the background galaxies of z=0.83 . increasing the value to for example
z=1 lowers the inferred mass considerably .
an indication that the mass from the weak lensing might be estimated too high is the fact that the m / l for this cluster found by smail et al.(1997 ) is relatively high in comparison with other clusters in their sample
. therefore it might be that the different results of the mass can be entirely explained by systematic errors in the two approaches .
however , we can not rule out that there might be some physical reason for this apparent discrepancy of the mass results , and that for example 3c 295 has either prolate symmetry along the line - of sight or suffers from projection effect from another massive structure .
in this paper we analyse rosat hri data of the distant cluster 3c 295 .
we fit a one and a two - dimensional isothermal @xmath1-model to the data , the two - dimensional fit taking into account the ellipticity of the cluster and effects of the psf . the one - dimensional model which does not take into account
the psf gives higher values for @xmath1 and the core radius @xmath17 ; however , the results agree within the error bars . to determine the errors of the parameters of the two - dimensional fit we use a monte - carlo approach . + calculating the cooling time of the central region we confirm previous findings by henry & henriksen ( 1986 ) of a central cooling flow .
we estimate the cooling flow radius to lie between 60120 kpc , depending on the model and adopted cooling time for the cluster .
generally we find that the isothermal @xmath1-model , which does not take into account the effects of the psf , gives higher values for the cooling flow radius and the mass accretion rates .
+ for the determination of the mass accretion rates we present a simple model , which is solely based on the isothermal @xmath1-model , neglecting heat conduction , inhomogenities and heat sources .
we show that the inferred mass accretion rates , with values of @xmath2 m@xmath3/y are dependent on the temperature structure in the cooling flow region .
a temperature gradient generally lowers the mass accretion rates .
+ the mass accretion rate in the central part of 3c 295 is somewhat too high for its cooling radius , when compared to the sample of nearby clusters of white et al .
( 1997 ) . as 3c 295 is not the only distant cluster showing this behaviour , this might be an indication that generally this relationship between mass accretion rate and cooling flow radius is evolving with time . + to search for substructure in the cluster
we subtract a two - dimensional cluster model , following the best fit results of the isothermal @xmath1-model from the original data . apart from a decrement in the south - east in the cluster , which might be caused by absorption ,
we do not find strong indication for the cluster to deviate from hydrostatic equilibrium .
the fact that 3c 295 seems to be in complete hydrostatic equilibrium , and has had enough time to form a cooling flow is an indication that the cluster must have had considerably time to relax , which suggests a low density universe , with or without a cosmological constant .
+ calculating the total mass of the cluster based on the hydrostatic equation using a temperature estimate of mushotzky & scharf ( 1997 ) , we obtain results of m@xmath103m@xmath3 at a radius of 2.1 mpc . for the gas to total mass ratio we obtain m@xmath7/m@xmath104 .
this ratio is typical for other clusters , and is again a strong indication for a low density universe .
+ + i like to thank m. arnaud , s. schindler and h. bhringer for very helpful discussions , and jean - luc sauvageot for computer and software support .
i like to especially thank c.a .
collins for a very careful reading of the manuscript and the referee , andy fabian for his very useful comments .
rosat is supported by the bmft .
i thank the cnrs and the mpg for funding .
akujor c.e .
, ldke e. , browne i.w.a . ,
leahy j.p . , garrington s.t . , jackson n. , & thomasson p 1994 , a&as , 105 , 247 allen s.w . , fabian a.c .
, edge a.c . ,
bautz m.w .
, & furuzawa a. 1996a , mnras , 283 , 263 allen s.w . , fabian a.c . , & kneib j.p .
1996b , mnras , 279 , 615 bhringer h. , & hensler g. 1989 , a&a , 215 , 147 bhringer h. , voges w. , fabian a.c . , edge a.c . , &
neumann d.m .
1993 , mnras , 264l , 25 bhringer h. 1994 , in proceedings of the nato advanced study institute on cosmological aspects of x - ray clusters of galaxies , ed .
w. seitter ( boston : kluwer ) , 123 bhringer h. , tanaka y. , mushotzky r.f . , ikebe y. , & hattori m. 1998 , a&a , in press briel u.g .
, henry j.p . , & bhringer h. 1992 , a&a , 259 , l13 cavaliere a. , & fusco - femiano r. 1976 , a&a 49 , 137 cavaliere a. , & fusco - femiano r. 1981 , a&a 100 , 194 cen r. , & ostriker j.p .
1993 , apj , 417 , 404 cowie l.l .
, & binney j. 1977 , apj , 215 , 723 david l.p .
1997 , the rosat high resolution imager ( hri ) calibration report , u.s .
rosat science data center / sao dickey j.m . , & lockman f.j .
1990 , a&ar , 28 , 215 dressler a. , & gunn j.e . , 1992 , apjs , 78 , 1 edge a.c . , fabian a.c . ,
allen s.w .
, crawford c.s . , white d.a . , bhringer h. , & voges w. 1994 , 270 , l1 evrard a.e .
, metzler c.a . , &
navarro j.n . , 1996 , apj , 469 , 494 evrard a.e . 1990 , 363 , 349 fabian a.c . , &
nulsen p.e.j .
1977 , mnras , 180 , 479 fabian a.c .
, nulsen p.e.j .
, & canizares c.r .
1984 , nat .
310 , 733 fabian a.c .
, nulsen p.e.j . , &
canizares c.r .
1991 , a&ar , 2 191 fabian a.c .
crawford 1995 , mnras , 274 , l63 harris d.e .
, silverman j.d . ,
hasinger g. , lehmann i. 1998 , a&as , 133 , 431 henry j.p . , & henriksen m.j . 1986 , apj , 301 , 689 hewitt a. , & burbidge g. 1991 , apjs , 75 , 297 hewitt a. , & burbidge g. 1993 , apjs , 87 , 451 jones c. , & forman w. 1992 , clusters and sperclusters of galaxies , ed .
fabian ( dordrecht : kluwer ) , 49 lea s.m . , silk j. , kellog e. , & murray s. 1973 , apj , 184 , l105 lea s.m . , &
henry j.p .
1988 , apj , 332 , 81 mathews w.g .
, & bregman j.n . 1978 , apj , apj , 244 , 308 morse j.a . 1994 , pasp , 106 , 675 mushotzky r.f . , & scharf c.a . 1997 , apj , 482 , l13 neumann d.m . , &
bhringer h. 1995 , a&a , 301 , 865 neumann d.m . ,
& bhringer h. 1997 , mnras , 289 , 123 neumann d.m . , &
bhringer h. 1998 , apj , submitted pierre m. , le borgne j.f . , soucail g. , & kneib j.p .
1996 a&a , 311 413 sarazin c. 1986 , rev .
mod . phys .
58 , 1 sarazin c. 1988 , x - ray emission from clusters of galaxies , c.u.p .
schindler s. , 1996 , a&a , 305 , 756 schindler s. , hattori m. , neumann d.m . , & bhringer h. 1997 , 317 , 646 schindler s. , & prieto m.a .
1997 , a&a , 327 , 37 smail i. , ellis r.s .
, dressler a. , couch w.j . ,
oemler a.jr .
, sharples r.m . , & butcher h. 1997 , apj , 479 , 70 squires g. , kaiser n. , babul a. , fahlman g. , woods d. , neumann d.m . , &
bhringer h. 1996 , apj , 461 572 squires g. , neumann d.m . , kaiser n. , arnaud m. , babul a. , bhringer h. , fahlman g. , & woods g. 1997 , apj , 482 , 648 thimm g.j . , rser h .- j . , hippelein h. , & meisenheimer k. 1994 , a&a , 285 , 785 thomas p.a .
, fabian a.c . , & nulsen p.e.j .
1987 , mnras , 228 , 973 walker t.p .
, steigman g. , kang h. , schramm d.m . , & olive k.a .
1991 , apj , 376 , 51 white d.a . , jones c. , & forman w. 1997 , mnras , 292 , 419 white s.d.m . , navarro j.f . , evrard a.e . , &
frenk c.s .
1993 , nat .
366 , 429 * fig.*[fig : plot ] + the countrate image of the hri image of 3c 295 .
we use only channel 2 to 9 to optimize the signal to background ratio ( for details see text ) .
the image is gaussian filtered with a @xmath29 of 3.5 arcsec .
the contour levels are spaced logarithmically with @xmath105 .
the lowest contour lies at 7.22 cts / sec / arcmin@xmath106 .
the image is centered on r.a.=@xmath107 and dec.=@xmath108 ( j.2000 ) .
+ + * fig.*[fig : surf ] + surface brightness profile of 3c 295 .
the crosses show the data with 1-@xmath29-errors in y - direction .
the full line corresponds to the best fit of the one - dimensional isothermal @xmath1-fit not taking into account effects of the hri s psf ( @xmath87 ) .
the dotted line shows the results of the two - dimensional model fit correcting for the psf ( with @xmath89 ) . for displaying we
utilize here as core radius the geometrical mean of the two core radii , corresponding to 25 kpc@xmath109 arcsec .
+ + * fig.*[fig : sym ] + the cluster model corresponding to the two - dimensional fit parameters shown in tab.[tab : beta ] .
the lowest contour and the spacing are identical with the ones in fig.[fig : plot ] .
+ + * fig.*[fig : resi ] + the residuals of the cluster after subtracting the isothermal @xmath1-model following the best fit results of the two - dimensional fit including background .
the model is convolved with the psf before subtraction .
the contours show the significance of the residuals above cluster plus background emission in @xmath29 .
the dashed line is the zero @xmath29-line , negative residuals are dotted , positive residuals have a full line .
the step width is 1 @xmath29 from contour to contour .
the shown significance levels take into account the applied gaussian filter of 3.5 arcsec .
+ + * fig.*[fig : cf1 ] + mass accretion rates in the central part of the cluster using the model taking into account the effects of the psf ( @xmath89 , @xmath110 arcsec ) .
shown are the rates as a function of radius for the isothermal case ( upper lines ) and for the case of constant temperature gradient ( lower lines ) .
the dotted lines correspond to a temperature of 5.8 kev , the full line to 7.1 kev , and the dashed line to 9.2 kev .
for the temperature gradient case we use a central temperature of 2 kev . in this case 5.8 , 7.1 , and 9.2 kev are the temperatures at a radius of 120 kpc .
+ + * fig.*[fig : cf2 ] + the same as previous figure , only difference : we use here the isothermal @xmath1-model from the one - dimensional fit , not taking into account the effects of the psf .
+ + * fig.*[fig : masse ] + mass profile of the cluster 3c 295 .
the full lines correspond to the @xmath1-model taking into account psf effects with @xmath89 and a core radius of @xmath111=3.7 arcsec .
the dots show the result for the model with @xmath87 . shown
are the mean values @xmath112-errors .
the open circles with the long dashed lines correspond to the projected mass profile with @xmath1=0.52 ( - here we only show the error limits @xmath112-errors ) .
the dotted line shows the gas mass profile with the @xmath89-model , the dashed line the profile with @xmath87 using a gas temperature of @xmath46 kev ( mushotzky & scharf 1997 ) . the vertical line at 400 kpc shows the result of the weak lensing analysis by smail et al .
( 1996 ) transformed to @xmath10 .
clccc & source & ra . & dec .
& total counts + 1 & galaxy ( z=0.2737)@xmath113 & @xmath114 & @xmath115 & 10 + 2 & galaxy ( z=0.4733 ) ( sy1)@xmath116 & @xmath117 & @xmath118 & 60 + 3 & qso ( z=1.29)@xmath119 & @xmath120 & @xmath121 & 20 + + + + + ccccccccc & @xmath18 & @xmath22 & @xmath23 & @xmath1 & @xmath24 & @xmath25 & @xmath26 & @xmath19 + & [ s@xmath14arcsec@xmath82 ] & [ arcsec ] & [ arcsec ] & & 2000.0 & 2000.0 & & [ s@xmath14arcsec@xmath82 ] + 1-d & 3.8@xmath123 & @xmath124 & & @xmath125 & @xmath126 & @xmath127 & & @xmath128 + 2-d & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath127 & @xmath134 & @xmath135 + lllll & total mass ( @xmath89 ) & total mass ( @xmath87 ) & gas mass ( @xmath89 ) & gas mass ( @xmath87 ) + & [ @xmath139m@xmath3 ] & [ @xmath139m@xmath3 ] & [ @xmath139m@xmath3 ] & [ @xmath139m@xmath3 ] + at 2.1 mpc & @xmath141 & @xmath142 & 2.0 & 1.8 + | we present rosat hri data of the distant and x - ray luminous ( @xmath0 erg / sec ) cluster of galaxies 3c 295 .
we fit both a one - dimensional and a two - dimensional isothermal @xmath1-model to the data , the latter one taking into account the effects of the point spread function ( psf ) . for the error analysis of the parameters of the two - dimensional model
we introduce a monte - carlo technique .
+ applying a substructure analysis , by subtracting a cluster model from the data , we find no evidence for a merger , but we see a decrement in emission south - east of the center of the cluster , which might be due to absorption . + we confirm previous results by henry & henriksen ( 1986 ) that 3c 295 hosts a cooling flow .
the equations for the simple and idealized cooling flow analysis presented here are solely based on the isothermal @xmath1-model , which fits the data very well , including the center of the cluster .
we determine a cooling flow radius of 60120 kpc and mass accretion rate of @xmath2 m@xmath3/y , depending on the applied model and temperature profile .
+ we also investigate the effects of the rosat psf on our estimate of @xmath4 , which tends to lead to a small overestimate of this quantity if not taken into account .
this increase of @xmath4 ( 10 - 25% ) can be explained by a shallower gravitational potential inferred by the broader overall profile caused by the psf , which diminishes the efficiency of mass accretion .
+ we also determine the total mass of the cluster using the hydrostatic approach . at a radius of 2.1 mpc
, we estimate the total mass of the cluster ( m@xmath5 ) to be @xmath6m@xmath3 .
for the gas to total mass ratio we get m@xmath7/m@xmath5=0.170.31 , in very good agreement with the results for other clusters of galaxies , giving strong evidence for a low density universe . |
fourier transforming @xmath49 yields the spin susceptibility in real space @xmath110 , which is relevant , e.g. , for the proposed use of rkky interactions to couple quantum - dot - confined spins as part of quantum - information - processing protocols @xcite .
figure [ fig : chicomps ] shows the density dependence of @xmath110 for a 2d hole system . in the low - density limit , we find @xmath111 , reflecting the strong easy - axis anisotropy arising due to hh - lh splitting @xcite . as the density of 2d holes in the lowest subband increases , in - plane components are observed to become relevant .
for @xmath112 we find @xmath113 to be the dominant component of the spin susceptibility tensor while @xmath114 .
switching of the effective hamiltonian that describes the coupling of two localized spins at a fixed distance is therefore possible by simply adjusting the density of the 2d hole system in which they are embedded .
real - space spin - susceptibility @xmath115 of a 2d hole system , plotted in units of @xmath116 as a function of @xmath117 for fixed @xmath118 and @xmath119 .
[ here @xmath120 denote the polar coordinates of the 2d real - space vector @xmath3 . ] ] | we have studied quantum - well - confined holes based on the luttinger - model description for the valence band of typical semiconductor materials .
even when only the lowest quasi - two - dimensional ( quasi-2d ) subband is populated , the static spin susceptibility turns out to be very different from the universal isotropic lindhard - function lineshape obtained for 2d conduction - electron systems .
the strongly anisotropic and peculiarly density - dependent spin - related response of 2d holes at long wavelengths should make it possible to switch between easy - axis and easy - plane magnetization in dilute magnetic quantum wells .
an effective _ g _ factor for 2d hole systems is proposed .
_ introduction _ in semiconductors , electric current can be carried by conduction - band electrons or valence - band holes . besides being oppositely charged , these two types of quasiparticles exhibit strikingly different magnetic properties .
band electrons are spin-1/2 particles like electrons in vacuum . in contrast
, holes have a spin angular momentum of 3/2 .
size quantization strongly affects the holes spin-3/2 degree of freedom @xcite .
this is believed to be the origin of an unusual paramagnetic response observed for quantum - well - confined holes @xcite , and the same phenomenon is expected to stabilize out - of - plane easy - axis magnetism in a dilute - magnetic - semiconductor ( dms ) @xcite two - dimensional ( 2d ) hole system @xcite .
controlling the confinement of holes thus enables appealing routes toward realizing magnetic semiconducting devices based on strain - induced anisotropies @xcite or wave - function engineering in heterostructures @xcite . here
we present a detailed theoretical study of the static spin susceptibility of 2d hole systems , which reveals unexpectedly rich magnetic properties of p - type quantum wells .
the paramagnetic response is characterized by a strongly anisotropic and density - dependent effective _ g_-factor . in a 2d dms system ,
valence - band mixing drives magnetic transitions that could enable new magneto - electronic device functionalities @xcite based on electric - field manipulation of the magnetization in low - dimensional systems .
_ background & aim _ the magnetic properties of a many - particle system are most comprehensively characterized by the spin - susceptibility tensor @xcite . in a homogeneous electron system with spin - rotational invariance
, it has a universal isotropic form that depends only on the dimensionality of the system @xcite .
this case applies to the 2d electron systems realized by confining carriers from the conduction band in semiconductor heterostructures as long as inversion symmetry is not broken by the crystal lattice or due to structuring of the sample @xcite .
lowest three ( each of them doubly degenerate ) subbands of a two - dimensional hole system realized by a symmetric hard - wall quantum - well confinement of width @xmath0 .
dispersions are calculated based on the four - band luttinger - model description of bulk valence - band states in axial approximation , using band - structure parameters applicable to gaas confined in [ 001 ] direction , and @xmath1 .
gray lines indicate the range of energies and wave vectors for which only the lowest subband is occupied .
this is the regime we focus on in this work . ] here we consider the properties of 2d _ hole _ systems whose charge carriers have a spin-3/2 degree of freedom that is strongly coupled to their orbital motion even when inversion symmetry is intact @xcite .
we adopt the luttinger model @xcite in axial approximation @xcite , which provides a useful description of the upper - most valence band of typical semiconductors in situations where its couplings to the conduction band and split - off valence band are irrelevant .
subband k - dot - p theory @xcite is employed to obtain the lowest quasi-2d hole subbands for a symmetric hard - wall confinement characterized by its spatial width @xmath0 .
see fig .
[ fig : dispersion ] .
we consider the case where only the lowest 2d subband is occupied and calculate the spin susceptibility . based on this result
, we discuss the paramagnetic response of 2d holes and identify various magnetic phases that emerge in dms quantum wells .
_ spin susceptibility of a quasi-2d system _ the spin susceptibility of 2d charge carriers is given by @xcite @xmath2 \rangle \ , .
\label{eq : spinsus}\end{aligned}\ ] ] here @xmath3 and @xmath4 are components of the position vector in the 2d ( @xmath5 ) plane and in the perpendicular ( growth ) direction , respectively . the spin density operator ( in units of @xmath6 )
is defined in terms of field operators @xmath7 , @xmath8 and the cartesian components @xmath9 of the charge carriers intrinsic angular momentum as @xmath10 . the field operators can be expressed in terms of operators associated with general eigenstates [ labelled by band index @xmath11 and in - plane wave vector @xmath12 of the non - interacting hamiltonian as @xmath13 . the ( normalized ) spinors @xmath14 and eigenvalues @xmath15 are obtained by solving the multi - band schrdinger equation for the confinement in growth direction of the 2d heterostructure .
we can then express the spin susceptibility as [ eq : spinsten ] @xmath16 in terms of the 2d fourier - transformed susceptibility @xmath17 where @xmath18 denotes the fermi function , and @xmath19^\dagger \cdot \big [ \hat{j}_i \ , \xi_{l{{\ensuremath{\bm{\mathrm{k}}}}}+{{\ensuremath{\bm{\mathrm{q}}}}}}(z)\big ] \nonumber \\ & & \hspace{3 cm } \times \big [ \xi_{l{{\ensuremath{\bm{\mathrm{k}}}}}+{{\ensuremath{\bm{\mathrm{q}}}}}}(z')\big]^\dagger \cdot \big [ \hat{j}_j \ , \xi_{n{{\ensuremath{\bm{\mathrm{k}}}}}}(z')\big ] \,\ , .\end{aligned}\ ] ] axial symmetry of the 2d system implies @xmath20 ( where @xmath21 and @xmath22 are the polar coordinates of @xmath23 ) and permits the ansatz @xcite @xmath24 simplifying calculation of the matrix elements @xmath25 . in the following , we consider the growth - direction - averaged spin susceptibility @xmath26 calculated at zero temperature . _ luttinger - model description of quasi-2d holes _ using the @xmath27 luttinger hamiltonian in axial approximation for the bulk valence band , the bound states of 2d holes confined by a potential @xmath28 are given in terms of spinor wave functions @xmath29 that satisfy the schrdinger equation @xmath30 \bar\xi_{n k } = e_{n k}\ , \bar\xi_{n k}$ ] , where @xmath31 \frac{d^2}{dz^2 } + v(z ) \,\ , , \\
\mathscr{h}_1 & = & \frac{\hbar^2 k}{m_0 } \ , \sqrt{2 } \,\tilde \gamma_2\ , ( -i ) \left ( \{\hat j_z , \hat j_- \}+ \{\hat j_z , \hat j_+ \ } \right ) \frac{d}{d z } \,\ , , \\
\mathscr{h}_2 & = & -\frac{\hbar^2 k^2}{2 m_0}\left [ \gamma_1\ , \openone + \tilde\gamma_1 \left
( \hat j_z^2 -\frac{5}{4}\,\openone \right ) - \tilde\gamma_3 \left ( \hat j_+^2 + \hat j_-^2 \right)\right ] .
\nonumber \\\end{aligned}\ ] ] here @xmath32 is the electron mass in vacuum , hole energies are counted as negative from the bulk valence - band edge , and we used the abbreviations @xmath33 , @xmath34 .
the constants @xmath35 and @xmath36 are materials - related band - structure parameters @xcite and depend on the quantum - well growth direction .
straightforward application of the subband k - dot - p method @xcite yields the energy dispersions @xmath37 with associated eigenspinors @xmath38 . at @xmath39 ,
the eigenspinors are also eigenstates of @xmath40 with eigenvalues @xmath41 ( heavy holes ) or @xmath42 ( light holes ) , which are split in energy .
this phenomenon is often referred to as _ hh - lh splitting _ @xcite . at finite @xmath43 ,
the spinors @xmath38are not eigenstates of @xmath40 anymore .
this phenomenon of _ hh - lh mixing _ arises because a spin-3/2 degree of freedom has a much richer structure than the more familiar spin-1/2 case @xcite .
figure [ fig : dispersion ] shows the 2d - subband dispersions obtained for a symmetric hard - wall confinement of width @xmath0 , using band - structure parameters for gaas confined in [ 001 ] direction .
the numerically obtained @xmath37 and @xmath38 serve as input for the calculation of the 2d - hole spin susceptibility according to eqs .
( [ eq : spinsten ] ) with eq .
( [ eq : sepansatz ] ) .
@xmath44-dependent spin susceptibility for @xmath45 and values of the 2d - hole fermi wave vector @xmath46 as indicated in the individual panels .
calculations are based on the band structure shown in fig .
[ fig : dispersion].,title="fig : " ] + @xmath44-dependent spin susceptibility for @xmath45 and values of the 2d - hole fermi wave vector @xmath46 as indicated in the individual panels .
calculations are based on the band structure shown in fig .
[ fig : dispersion].,title="fig : " ] + @xmath44-dependent spin susceptibility for @xmath45 and values of the 2d - hole fermi wave vector @xmath46 as indicated in the individual panels .
calculations are based on the band structure shown in fig .
[ fig : dispersion].,title="fig : " ] _ results for the 2d - hole spin susceptibility _ using polar coordinates @xmath47 for the 2d wave vector @xmath44 and introducing the scale @xmath48 , we find that the tensor elements of @xmath49 have the generic form [ eq : genstruc ] @xmath50 , \\
\overline{\chi}_{yy}({{\ensuremath{\bm{\mathrm{q } } } } } ) & = & \chi_0\left[f_\parallel(q ) - g(q ) \ ,
\cos\left(2\phi_{{\ensuremath{\bm{\mathrm{q}}}}}\right ) \right ] , \\
\overline{\chi}_{xy}({{\ensuremath{\bm{\mathrm{q}}}}})&= & \chi_0\ , g(q ) \ , \sin\left ( 2\phi_{{\ensuremath{\bm{\mathrm{q}}}}}\right ) \,\ , , \\ \label{eq : genstruczz } \overline{\chi}_{zz}({{\ensuremath{\bm{\mathrm{q}}}}})&= & \chi_0\ , f_\perp(q ) \,\,.\end{aligned}\ ] ] the functions @xmath51 and @xmath52 depend on materials and morphological parameters of the 2d hole system , especially the hole sheet density @xmath53 , but @xmath54 generally . figure [ fig : chiq ] shows typical results obtained at low , intermediate and high densities where still only states in the lowest quasi-2d subband are occupied .
the density dependence of the 2d - hole spin susceptibility follows a generic trend . in the limit of low density [ fig .
[ fig : chiq](a ) ] , @xmath55 , and the line shape of @xmath56 is similar to the universal 2d - electron ( lindhard ) result @xcite .
the strong easy - axis response is expected @xcite as a result of hh - lh splitting , which favors a spin-3/2 quantization axis perpendicular to the 2d plane @xcite .
interestingly , the behavior of the spin susceptibility changes as density is increased . for intermediate values of hole density ,
results like that shown in fig .
[ fig : chiq](b ) are obtained , exhibiting _ easy - plane _ anisotropy in the long - wave - length limit and a nontrivial structure developing at wave vectors comparable to @xmath46 . the significant deviation from both the universal-2d - electron behavior and also
the easy - axis response expected from hh - lh _ splitting _ arises because , as the 2d - hole density is increased , higher-@xmath43 states get occupied that are more strongly influenced by hh - lh _
mixing_. at the highest values of density where still only the lowest quasi-2d hole subband is populated , easy - axis anisotropy is restored in the long - wave - length limit but the response at finite @xmath57 becomes as important in strength as that for @xmath58 .
_ effective g - factors for quasi-2d holes _ motivated by recent experimental @xcite and theoretical @xcite interest in the paramagnetic response of 2d holes , we apply our results for the spin susceptibility of a noninteracting 2d hole system to define an effective _ g _ factor .
effective @xmath59-factor as a function of the 2d - hole fermi wave vector @xmath60 for a perpendicular ( dashed curve ) and in - plane ( solid curve ) magnetic field . ] a magnetic field parallel to the @xmath61 axis couples to the holes spin via the zeeman term @xmath62 , where @xmath63 is the bohr magneton and @xmath64 the bulk - hole _
g _
factor @xcite . in the low - field limit
, the paramagnetic susceptibility is given by @xmath65 in terms of our calculated spin susceptibility ) at @xmath66 . ] .
comparison of this relation with the expression for the pauli susceptibility of conduction electrons from a parabolic band @xcite suggests defining the effective @xmath59-factor for a many - particle state via @xmath67 , which explicitly yields @xmath68 here @xmath69 is the static 2d - hole lindhard function ; its @xmath66 limit equals the density of states at the fermi level .
we find a density - dependent and anisotropic @xmath59-factor , reflecting the interplay between hh - lh splitting and mixing in confined valence - band states .
figure [ fig : gj ] shows the density dependence of the transverse @xmath70 and in - plane @xmath71 @xmath59-factors for a 2d hole system . in the low - density limit , behavior expected from hh - lh splitting
is found , whereas the paramagnetic response at intermediate and high density is substantially affected by hh - lh mixing . _
2d - hole - mediated magnetism _
magnetism is introduced into intrinsically nonmagnetic semiconducting materials via doping with magnetic ions @xcite such as mn , co , fe or gd .
one way to generate an exchange interaction between any two localized magnetic moments embedded in a conductor is provided by the ruderman - kittel - kasuya - yosida ( rkky ) mechanism @xcite , which gives rise to the effective two - impurity spin hamiltonian @xmath72 here @xmath73 denotes the @xmath74th cartesian component of an impurity spin located at position @xmath75 , and @xmath76 is the exchange constant for the contact interaction between the spin density of delocalized charge carriers with the impurity spins . for a random but
on average homogeneous distribution of magnetic ions , standard mean - field theory @xcite applied to the spin model described by the hamiltonian of eq .
( [ eq : rkkyham ] ) yields the curie temperatures @xmath77 for ferromagnetic order with magnetization direction parallel to the @xmath61 axis .
the temperature scale @xmath78 depends on the impurity - spin magnitude @xmath79 and the average 3d density @xmath80 of magnetic impurities , and its functional form is that obtained for 2d charge carriers in a parabolic band @xcite with effective mass @xmath81 .
mean - field curie temperatures for 2d - hole - mediated easy - axis ( @xmath82 ) and easy - plane ( @xmath83 ) magnetism as a function of the 2d - hole fermi wave vector @xmath46 , obtained for a hard - wall confinement with width @xmath0 and band - structure parameters applicable to gaas . ]
figure [ fig : curiet ] shows the mean - field curie temperatures for perpendicular - to - plane ( @xmath84 ) and in - plane ( @xmath85 ) magnetization directions calculated with the same input parameters used for obtaining the subbands given in fig . [ fig : dispersion ] .
the density dependence of @xmath86 is directly reflected in that of @xmath87 . at the mean - field level , the ordered state associated with the maximum transition temperature will be established .
our results suggest that the type of magnetic ordering can be modified by changing the 2d - hole density , e.g. , by adjusting the gate voltage in accumulation - layer devices @xcite .
easy - axis magnetism prevails at low and high densities , whereas an unexpected easy - plane magnetic order emerges at intermediate values of the density . also for high densities , a local
maximum appears in @xmath88 at @xmath89 , which almost reaches the value of @xmath90 .
see fig .
[ fig : chiq](c ) . even after averaging over the polar angle
, we find that for @xmath91 the in - plane susceptibility can have a maximum at @xmath92 which is as large as @xmath93 .
thus it may be possible that the high - density easy - axis ferromagnetic state must coexist ( or compete ) with helical magnetism @xcite .
_ finite - temperature effects _ thermal excitation of spin waves ( magnons ) suppresses the magnitude of the magnetization below its mean - field value @xmath94 .
this effect is captured by the relation @xcite @xmath95 where @xmath96 is the occupation - number distribution function of magnon modes at temperature @xmath97 .
a spin - wave - related critical temperature is defined by the condition @xmath98 because , for @xmath99 , too many magnon excitations will have been excited to sustain a finite magnetization . in equilibrium
, @xmath96 is given by the bose - einstein distribution function @xmath100 } - 1)$ ] , which depends on the spin - wave energy dispersion @xmath101 .
the latter s expression in terms of the charge carriers
@xmath44-dependent spin susceptibility depends on the type of magnetic order ( heisenberg , ising , or helical ) @xcite , but the parameterisation @xmath102 \label{eq : magnondisansatz}\ ] ] typically holds for the relevant energy range .
the dimensionless quantities @xmath103 , @xmath104 , @xmath105 can be determined from the functional form of the @xmath44-dependent spin susceptibility .
stability of the magnetic order requires both coefficients @xmath103 and @xmath105 to be positive . specializing to our situation of 2d - hole - mediated magnetism , we see from fig . [
fig : chiq](a ) that the easy - axis magnetism expected at low hole - sheet densities is destabilized by magnons because @xmath106 . considering the easy - plane magnetism at intermediate densities , fig .
[ fig : chiq](b ) reveals that the associated magnon dispersion is characterized by @xmath107 and @xmath108 , which again implies destabilization of this magnetic order due to spin - wave excitations . for the easy - axis ( ising ) magnet expected at high densities [ cf .
fig .
[ fig : chiq](c ) ] , we find @xmath107 and @xmath109 . in this case a finite spin - wave - related critical temperature is obtained .
further studies need to explore the effect of coulomb interactions , which can stabilize ferromagnetic order mediated by 2d carriers @xcite .
_ conclusions _ the spin susceptibility of 2d holes is strongly density dependent . in the low - density limit ,
the easy - axis response due to hh - lh _ splitting _ is exhibited . with increasing density , hh - lh _ mixing _ changes the spin - related response of confined holes even more drastically than the density response @xcite .
an effective _ g _ factor for 2d holes is proposed .
we clarify the impact of band - structure effects in 2d dms systems that had previously been only considered for the 3d case @xcite or outside the rkky limit @xcite .
the switching behavior of the magnetization found here should be observable in p - type quantum wells where the 2d - hole density is independently adjustable @xcite and the magnetic doping is sufficiently low to ensure a sizable mean free path .
_ acknowledgments _ the authors benefited from useful discussions with a. h. macdonald , j. splettster , and r. winkler .
43ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _
_
( , , ) @noop * * , ( ) link:\doibase 10.1103/physrevb.73.115320 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1126/science.281.5379.951 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.71.113307 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.63.035314 [ * * , ( ) ] @noop * * , ( ) http://stacks.iop.org/1367-2630/12/i=9/a=093002 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.83.205206 [ * * , ( ) ] @noop * * , ( ) |
this paper deals with the study of the asymptotic behavior of the solution of poisson equation in a bounded domain @xmath2 of @xmath0 ( @xmath3 ) consisting of two sub - domains separated by a thin layer of thickness @xmath4 ( destined to tend to 0 ) .
the mesh of these thin geometries presents numerical instabilities that can severely damage the accuracy of the entire process of resolution . to overcome this difficulty
, we adopt asymptotic methods to model the effect of the thin layer by problems with either appropriate boundary conditions when we consider a domain surrounded by a thin layer ( see for instance @xcite ) or , as in this paper , with suitable transmission conditions on the interface ( see for instance @xcite ) .
although this type of conditions has been widely studied , there is still a lot to be understood concerning the effects of thin shell and their modelisation .
our motivation comes from @xcite , in which the authors have worked on problems of electromagnetic and biological origins .
we cite for example that of poignard ( * ? ? ?
* chapter 2 ) .
he considered a cell immersed in an ambient medium and studied the electric field in the transverse magnetic ( tm ) mode at mid - frequency and from which our problem was inspired .
let us give now precise notations .
let @xmath2 be a bounded domain of @xmath0 ( @xmath5 ) consisting of three smooth sub - domains : an open bounded subset @xmath6 with regular boundary @xmath7 , an exterior domain @xmath8 with disjoint regular boundaries @xmath9 and @xmath10 , and a membrane @xmath11 ( thin layer ) of thickness @xmath4 separating @xmath6 from @xmath12(see fig .
[ fig1 ] ) .
define the piecewise regular function @xmath13 by @xmath14{ll}\alpha_{e } & \text{if } x\in\omega_{e,\delta},\\ \alpha_{\delta } & \text{if } x\in\omega_{\delta},\\ \alpha_{i } & \text{if } x\in\omega_{i,\delta } , \end{array } \right.\ ] ] where @xmath15 @xmath16 and @xmath17 are strictly positive constants satisfying @xmath18 or @xmath19 which correspond to the case of mid - diffusion . for a given @xmath20 in @xmath21 we are interested in the unique solution @xmath22 in @xmath23 of the following diffusion problem [ 1]@xmath24{ll}-div\left ( \alpha\nabla u_{\delta}\right ) = f & \text{in } \omega,\\ u_{\delta|\partial\omega}=0 & \text{on } \partial\omega , \end{array } \right . \label{1.01}\ ] ] with transmission conditions on the interfaces@xmath24{ll}u_{d,\delta|\gamma_{\delta,2}}=u_{e,\delta|\gamma_{\delta,2 } } & \text{on } \gamma_{\delta,2},\\ \alpha_{\delta}\partial_{\mathbf{n}_{\delta,2}}u_{d,\delta|\gamma_{\delta,2}}=\alpha_{e}\partial_{\mathbf{n}_{\delta,2}}u_{e,\delta|\gamma_{\delta,2 } } & \text{on } \gamma_{\delta,2},\\ u_{i,\delta|\gamma_{\delta,1}}=u_{d,\delta|\gamma_{\delta,1 } } & \text{on } \gamma_{\delta,1},\\ \alpha_{i}\partial_{\mathbf{n}_{\delta,1}}u_{i,\delta|\gamma_{\delta,1}}=\alpha_{\delta}\partial_{\mathbf{n}_{\delta,1}}u_{d,\delta|\gamma_{\delta , 1 } } & \text{on } \gamma_{\delta,1 } , \end{array } \right . \label{1.02}\ ] ] where @xmath25 and @xmath26 denote the derivatives in the direction of the unit normal vectors @xmath27 and @xmath28 to @xmath29 and @xmath7 respectively ( see fig .
[ fig1 ] ) . the main result of this paper is to approximate the solution @xmath30 of problem ( [ 1 ] ) by a solution of a problem involving poisson equation in @xmath2 with two sub - domains separated by an arbitrary interface @xmath31 between @xmath7 and @xmath9 ( see fig . [ fig2 ] and fig .
[ fig3 ] ) , with transmission conditions of order two on @xmath31 , modeling the effect of the thin layer .
however , it seems that the existence and uniqueness of the solution of this problem are not obvious@xmath32 therefore , we rewrite the problem into a pseudodifferential equation ( cf .
@xcite ) and show that in the case of mid - diffusion , we can find the appropriate position of the surface @xmath31 to solve this equation .
the cases 3d and 2d are similar .
we treat the three - dimensional case and the two dimensional one comes as a remark .
the present paper is organized as follows . in section 2 ,
we give the statement of the model problem considered . in section 3 ,
we collect basic results of differential geometry of surfaces .
sections 4 and 5 are devoted to the asymptotic analysis of our problem .
we present , in section 4 , hierarchical variational equations suited to the construction of a formal asymptotic expansion up to any order , while section 5 focuses on the convergence of this ansatz . with the help of the asymptotic expansion of the solution @xmath30 , we model , in the last section , the effect of the thin layer by a problem with appropriate transmission conditions .
, width=226 ] , width=132 ] we consider a parallel surface @xmath31 to @xmath7 and @xmath9 dividing @xmath11 into two thin layers @xmath33 and @xmath34 of thickness respectively @xmath35 and @xmath36 where @xmath37 and @xmath38 are nonnegative real numbers satisfying @xmath39 and such that @xmath37 and @xmath38 belong to a small neighborhood of @xmath40 ( see fig .
[ fig2 ] and fig .
[ fig3 ] ) .
the term _ small _ neighborhood means that the constants @xmath41 and @xmath38 are not too close to @xmath42 or @xmath43 , in order to avoid having a layer too thin compared to the other because the following analysis does not lend itself to this case . under the aforementioned assumptions , we investigate in @xmath23 the solution @xmath44 of the following problem [ 1.1]@xmath45 with transmission conditions@xmath46 where @xmath47 denotes the derivative in the direction of the unit normal vector @xmath48 to @xmath31 ( outer for @xmath33 and inner for @xmath34 )
the goal of this section is to define and to collect the main features of differential geometry @xcite ( see also @xcite ) in order to formulate our problem in a fixed domain ( independent of @xmath4 ) which is a key tool to determine the asymptotic expansion of the solution @xmath30 . in the sequel
, greek indice @xmath49 takes the values 1 and 2 .
let @xmath50 and @xmath51 we parameterize the thin shell @xmath52 by the manifold @xmath53 through the mapping @xmath54 defined by @xmath24{rcl}\gamma\times i_{\delta,\beta } & \overset{\psi_{\beta}}{\rightarrow } & \omega_{\delta,\beta}\\ ( m,\eta_{\beta } ) & \rightarrow & x:=m+p_{\beta}\eta_{\beta}\mathbf{n}(m ) .
\end{array } \right.\ ] ] as well - known @xcite , if the thickness of @xmath52 is small enough , @xmath55 is a @xmath56-diffeomorphism of manifolds and it is also known ( * ? ? ?
* remark 2.1 ) that the normal vector @xmath57 to @xmath58 can be identified to @xmath48 . to each function
@xmath59 defined on @xmath52 , we associate the function @xmath60 defined on @xmath61 by@xmath24{rl}\widetilde{v}_{\beta}(m,\eta_{\beta } ) & : = v_{\beta}(x),\\ x & = \psi_{\beta}\left ( m,\eta_{\beta}\right ) , \end{array } \right.\ ] ] then , we have @xmath62 where @xmath63 and @xmath64 are respectively the surfacic gradient of @xmath65 at @xmath66 and the curvature operator @xmath64 of @xmath31 at point @xmath67 the volume element on the thin shell @xmath52 is given by@xmath68 now , we introduce the scaling @xmath69 and the intervals @xmath70 and @xmath71 such that the @xmath56-diffeomorphism @xmath72 defined by @xmath24{rcl}\omega^{\beta}:=\gamma\times i_{\beta } & \overset{\phi_{\beta}}{\rightarrow } & \omega_{\delta,\beta}\\ ( m , s_{\beta } ) & \rightarrow & x:=m+\delta p_{\beta}s_{\beta}\mathbf{n}(m ) , \end{array } \right.\ ] ] parameterizes the thin shell @xmath73 to any function @xmath59 defined on @xmath52 , we associate the function @xmath74}$ ] defined on @xmath75 through@xmath24{rl}v^{[\beta]}(m , s_{\beta } ) & : = v_{\beta}(x),\\ x & = \phi_{\beta}(m , s_{\beta } ) , \end{array } \right.\ ] ] then the gradient takes the form@xmath76}+p_{\beta}^{-1}\delta^{-1}\frac{\partial v^{[\beta]}}{\partial s_{\beta}}\mathbf{n}.\label{2}\ ] ] the volume element on the thin shell @xmath52 becomes @xmath77 where@xmath78 let @xmath79 and @xmath59 be two regular functions defined on @xmath52 . from ( [ 2 ] ) and ( [ 3 ] ) , we get the change of variables formula @xmath80}.\nabla_{\gamma}v^{[\beta]}\det j_{\delta , \beta}\ d\gamma ds_{\beta}\nonumber\\ & + p_{\beta}^{-1}\delta^{-1}\int_{\omega^{\beta}}\partial_{s_{\beta}}u^{[\beta]}\partial_{s_{\beta}}v^{[\beta]}\det j_{\delta,\beta}\ d\gamma ds_{\beta}.\label{4}\ ] ] in the two - dimensional case , if @xmath66 , we parameterize the curve @xmath31 by @xmath81 where @xmath82 is the curvilinear abscissa and @xmath83 is the length of the curve @xmath31 , then formula ( [ 4 ] ) turns into@xmath84}\partial_{s_{\beta}}v^{[\beta]}\ d\gamma ds_{\beta}\nonumber\\ & + p_{\beta}\delta\int_{\omega^{\beta}}\left ( 1+p_{\beta}\delta s_{\beta } \mathcal{r}\right ) ^{-1}\partial_{t}u^{[\beta]}\partial_{t}v^{[\beta ] }
\ d\gamma ds_{\beta}.\label{6}\ ] ] for any function @xmath85 defined in a neighborhood of @xmath86 we denote , for convenience , by @xmath87 the trace of @xmath85 on @xmath31 indifferently in local coordinates or in cartesian coordinates .
this section is devoted to the asymptotic analysis of the solution of problem ( [ 1.1 ] ) .
we show that this latter is equivalent to a variational equation from which we derive the asymptotic expansion of @xmath88 we give a hierarchy of variational equations needed to determine the terms of the expansion and we calculate the first two terms of the expansion .
let @xmath89 be in @xmath90 we denote by @xmath91 its restriction to @xmath52 . multiplying equation @xmath92 by test functions @xmath89 , using ( [ 1.1.7 ] ) , ( [ 1.1.9 ] ) , ( [ 1.1.11 ] ) and green s formula , we get @xmath93 in which @xmath94 denotes the duality pairing between @xmath95 and @xmath96 .
we use the dilation in the thin layer and formula ( [ 4 ] ) , to obtain@xmath97 } \left ( u_{d,\delta}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) \right ] = \int_{\omega_{\delta}}f_{|\omega_{\delta}}v_{d}\ d\omega_{\delta},\label{12}\ ] ] which is the starting point for the asymptotic analysis , where the bilinear form @xmath98 } \left ( .,.\right ) $ ] is defined by@xmath99 } \left ( u^{[\beta]},v^{[\beta]}\right ) & : = p_{\beta}\int_{\omega^{\beta}}j_{\delta,\beta}^{-2}\nabla_{\gamma}u^{[\beta]}.\nabla_{\gamma}v^{[\beta]}\det j_{\delta,\beta}\ d\gamma ds_{\beta}\nonumber\\ & + p_{\beta}^{-1}\delta^{-2}\int_{\omega^{\beta}}\partial_{s_{\beta}}u^{[\beta]}\partial_{s_{\beta}}v^{[\beta]}\det j_{\delta,\beta}\ d\gamma ds_{\beta},\label{10}\ ] ] for every @xmath100}$ ] and @xmath74}$ ] in @xmath101 in the spirit of @xcite , we will consider two asymptotic expansions .
exterior expansions corresponding to the asymptotic expansion of @xmath30 restricted to @xmath8 and to @xmath102 and characterized by the ansatz@xmath103 where the terms @xmath104 and @xmath105 @xmath106 are independent of @xmath4 and defined on @xmath107 and on @xmath108 which are respectively the limits of @xmath8 and @xmath6 for @xmath109 they fulfill@xmath24{ll}-div\left ( \alpha_{i}\nabla u_{i , n}\right ) = \delta_{0,n}f_{|\omega_{i } } & \text{in } \omega_{i}\text{,}\\ -div\left ( \alpha_{e}\nabla u_{e ,
n}\right ) = \delta_{0,n}f_{|\omega_{e } } & \text{in } \omega_{e}\text{,}\\ u_{e , n|\partial\omega}=0 & \text{on}\ \partial\omega , \end{array } \right . \label{15}\ ] ] where @xmath110 indicates the kronecker symbol , and an interior expansion corresponding to the asymptotic expansion of @xmath111 written in a fixed domain and defined by the ansatz @xmath112 } = u_{0}^{\left [ \beta\right ] } + \delta u_{1}^{\left [ \beta\right ] } + \cdots,\text { in } \omega^{\beta},\label{16}\ ] ] where the terms @xmath113 } , \
n\in\mathbb{n}$ ] , are independent of @xmath4 . using a taylor expansion in the normal variable , we infer formally@xmath114 and transmission conditions ( [ 1.1.6 ] ) and ( [ 1.1.10 ] ) become@xmath115 } + \delta u_{1|s_{2}=1}^{\left [ 2\right ] } + \cdots,\label{18}\\ u_{i,0|\gamma}+\delta(u_{i,1|\gamma}-p_{1}\partial_{\mathbf{n}}u_{i,0|\gamma } ) + \cdots & = u_{0|s_{1}=-1}^{\left [ 1\right ] } + \delta u_{1|s_{1}=-1}^{\left [ 1\right ] } + \cdots.\label{17}\ ] ] as @xmath116 thanks to green s formula , we get @xmath117 using the the scaling @xmath118 we obtain @xmath119 } ( m,0)\ d\gamma+\alpha _ { e}\delta a_{\delta}^{\left [ 2\right ] } \left ( \sum_{n\geq0}\delta ^{n}u_{e , n},v^{\left [ 2\right ] } \right ) \nonumber\\ -\left\langle \alpha_{e}\partial_{\mathbf{n}_{\delta,2}}\left ( \sum_{n\geq 0}\delta^{n}u_{e , n|\gamma_{\delta,2}}\right ) \circ\phi_{2}(m,1),v^{\left [ 2\right ] } ( m,1)\right\rangle _ { h^{-1/2}(\gamma\times\left\ { 1\right\ } ) \times h^{1/2}(\gamma\times\left\ { 1\right\ } ) } \nonumber\\ = \int_{\omega_{\delta,2}}f_{|\omega_{\delta,2}}v_{d_{2}}\ d\omega_{\delta , 2}.\label{18.1}\ ] ] in the same way , we obtain the equation for @xmath120 : @xmath121 } \left ( \sum_{n\geq0}\delta^{n}u_{i , n},v^{\left
[ 1\right ] } \right ) -\int_{\gamma}\alpha _ { i}\partial_{\mathbf{n}}\left ( \sum_{n\geq0}\delta^{n}u_{i , n|\gamma}\right ) v^{\left [ 1\right ] } ( m,0)\ d\gamma\nonumber\\ + \left\langle \alpha_{i}\partial_{\mathbf{n}_{\delta,1}}\left ( \sum_{n\geq 0}\delta^{n}u_{i , n|\gamma_{\delta,1}}\right ) \circ\phi_{1}(m,-1),v^{\left [ 1\right ] } ( m,-1)\right\rangle _ { h^{-1/2}(\gamma\times\left\ { -1\right\ } ) \times h^{1/2}(\gamma\times\left\ { -1\right\ } ) } \nonumber\\ = \int_{\omega_{\delta,1}}f_{|\omega_{\delta,1}}v_{d_{1}}\ d\omega_{\delta , 1}.\label{18.2}\ ] ] inserting expansions ( [ 13 ] ) , ( [ 14 ] ) and ( [ 16 ] ) in ( [ 12 ] ) , using ( [ 16.1])-([16.2 ] ) and ( [ 18.1])-([18.2 ] ) , we get@xmath122 } ( m,0)\ d\gamma-\alpha _ { i}\delta a_{\delta}^{\left [ 1\right ] } \left ( \sum_{n\geq0}\delta ^{n}u_{i , n},v^{\left [ 1\right ] } \right ) \nonumber\\ + \displaystyle\sum_{\beta=1}^{2}\left [ \alpha_{\delta}\delta a_{\delta } ^{\left [ \beta\right ] } \left ( \sum_{n\geq0}\delta^{n}u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) \right ] -\alpha_{e}\delta a_{\delta}^{\left [ 2\right ] } \left ( \sum_{n\geq0}\delta^{n}u_{e , n},v^{\left [ 2\right ] } \right ) \nonumber\\ -\int_{\gamma}\alpha_{e}\partial_{\mathbf{n}}\left ( \sum_{n\geq0}\delta ^{n}u_{e , n|\gamma}\right ) v^{\left [ 2\right ] }
( m,0)\ d\gamma=0.\label{18.3}\ ] ] now , we use the identity ( see @xcite)@xmath123 .\end{aligned}\ ] ] since@xmath124 where @xmath125 and @xmath126 are respectively the mean and the gaussian curvatures of the surface @xmath31 , the bilinear form @xmath127 } ( .,.)$ ] admits the expansion @xmath99 } \left ( .,.\right ) & = \delta^{-2}a_{0,2}^{\left [ \beta\right ] } + \delta^{-1}a_{1,2}^{\left [ \beta\right ] } + \left ( a_{2,2}^{\left [ \beta\right ] } + a_{0,1}^{\left [ \beta\right ] } \right ) + \delta a_{1,1}^{\left [ \beta\right ] } + \cdots\nonumber\\ & + \delta^{n-1}a_{n-1,1}^{\left [ \beta\right ] } + \delta^{n}r_{n}^{\left [ \beta\right ] } \left ( \delta;.,.\right ) , \label{21}\ ] ] where the forms @xmath128 } $ ] are independent of @xmath4 and are given by @xmath129 } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}p_{\beta}^{-1}\partial_{s_{\beta}}u^{\left [ \beta\right ] } \partial_{s_{\beta}}v^{\left [ \beta\right ] } \
d\gamma ds_{\beta},\\ a_{1,2}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}2\mathcal{h}s_{\beta } \partial_{s_{\beta}}u^{\left [ \beta\right ] } \partial_{s_{\beta}}v^{\left [ \beta\right ] } \ d\gamma ds_{\beta},\\ a_{2,2}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}p_{\beta}\mathcal{k}s_{\beta}^{2}\partial_{s_{\beta}}u^{\left [ \beta\right ] } \partial_{s_{\beta } } v^{\left [ \beta\right ] }
\ d\gamma ds_{\beta},\\ a_{0,1}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}p_{\beta}\nabla_{\gamma } u^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } \
d\gamma ds_{\beta},\\ a_{1,1}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}2p_{\beta}^{2}s_{\beta } \left ( \mathcal{h}i-\mathcal{r}\right ) \nabla_{\gamma}u^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } \ d\gamma ds_{\beta } , \\ a_{2,1}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}p_{\beta}^{3}\left ( \mathcal{k}i-4\mathcal{hr}+3\mathcal{r}^{2}\right ) s_{\beta}^{2}\nabla_{\gamma}u^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } \ d\gamma ds_{\beta},\\ a_{n-1,1}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) & : = \int_{\omega^{\beta}}p_{\beta}^{n}\left [ \left ( n-2\right ) \mathcal{kr}^{n-3}-\left ( n-1\right ) 2\mathcal{hr}^{n-2}\right . \\ & \left
. + n\mathcal{r}^{n-1}\right ] \left ( -s_{\beta}\right ) ^{n-1}\nabla_{\gamma}u^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } \
d\gamma ds_{\beta}.\end{aligned}\ ] ] the form @xmath130 } ( \delta;.,.)$ ] is the remainder of expansion ( [ 21 ] ) and is expressed by @xmath131 } ( \delta;u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } ) : = \int_{\omega^{\beta}}\left ( b_{n,\delta}+2\mathcal{h}b_{n-1,\delta}+\mathcal{k}b_{n-2,\delta}\right ) s_{\beta}^{n}\nabla_{\gamma } u^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } d\gamma ds_{\beta},\ ] ] with@xmath132{l}\left ( -\mathcal{r}\right ) ^{n}\left ( nj_{\delta,\beta}^{-1}+j_{\delta , \beta}^{-2}\right ) \text { if}\ n\geq0,\\ j_{\delta,\beta}^{-2}\text { otherwise.}\end{array } \right.\ ] ] in the two - dimensional case , with the help of ( [ 6 ] ) , expansion ( [ 21 ] ) turns into @xmath133 } \left ( .,.\right ) = \delta^{-2}a_{0,2}^{\left [ \beta\right ] } + \delta^{-1}a_{1,2}^{\left [ \beta\right ] } + a_{0,1}^{\left [ \beta\right ] } + \delta a_{1,1}^{\left [ \beta\right ] } + \cdots+\delta^{n-1}a_{n-1,1}^{\left [ \beta\right ] } + \delta^{n}r_{n}^{\left [ \beta\right ] } \left ( \delta;.,.\right ) , \ ] ] with@xmath134 } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) : = \int_{\omega^{\beta}}p_{\beta}^{n-1}\left ( s_{\beta}\mathcal{r}\right ) ^{n}\partial_{s_{\beta}}u^{\left [ \beta\right ] } \partial_{s_{\beta}}v^{\left [ \beta\right ] } \ d\gamma ds_{\beta},\\ a_{n,1}^{\left [ \beta\right ] } \left ( u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) : = \int_{\omega^{\beta}}p_{\beta}^{n+1}\left ( -s_{\beta}\mathcal{r}\right ) ^{n}\partial_{t}u^{\left [ \beta\right ] } \partial_{t}v^{\left [ \beta\right ] }
\ d\gamma ds_{\beta},\\ r_{n}^{\left [ \beta\right ] } \left ( \delta;u^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) : = \int_{\omega^{\beta}}j_{\delta,\beta } ^{-1}\left ( -s_{\beta}\mathcal{r}\right ) ^{n}\partial_{t}u^{\left [ \beta\right ] } \partial_{t}v^{\left [ \beta\right ] } \ d\gamma ds_{\beta}.\end{gathered}\ ] ] inserting expansion ( [ 21 ] ) in ( [ 18.3 ] ) and matching the same powers of @xmath135 we obtain the following variational equations , which hold for all @xmath136 } , v^{\left [ 2\right ] } \right ) $ ] in @xmath137 @xmath138 } \left ( \alpha_{\delta}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0},v^{\left [ 1\right ] } \right ) + a_{0,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } \right ) = 0,\label{23}\]]@xmath139 } \left ( \alpha_{\delta}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0},v^{\left [ 1\right ] } \right ) + a_{0,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{1}^{\left [ 1\right ] } -\alpha_{i}u_{i,1},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{1,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } \right ) + a_{0,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{1}^{\left [ 2\right ] } -\alpha_{e}u_{e,1},v^{\left [ 2\right ] } \right ) \nonumber\\ = \alpha_{e}\int_{\gamma}\partial_{\mathbf{n}}u_{e,0|\gamma}v^{\left [ 2\right ] } \left ( m,0\right ) \ d\gamma-\alpha_{i}\int_{\gamma}\partial_{\mathbf{n}}u_{i,0|\gamma}v^{\left [ 1\right ] } \left ( m,0\right ) \
d\gamma,\label{24}\]]@xmath140 } \left ( \alpha_{\delta}u_{2}^{\left [ 1\right ] } -\alpha_{i}u_{i,2},v^{\left [ 1\right ] } \right ) + a_{1,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{1}^{\left [ 1\right ] } -\alpha_{i}u_{i,1},v^{\left [ 1\right ] } \right ) \nonumber\\ + \left ( a_{2,2}^{\left [ 1\right ] } + a_{0,1}^{\left [ 1\right ] } \right ) \left ( \alpha_{\delta}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{0,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{2}^{\left [ 2\right ] } -\alpha_{e}u_{e,2},v^{\left [ 2\right ] } \right ) + a_{1,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{1}^{\left [ 2\right ] } -\alpha_{e}u_{e,1},v^{\left [ 2\right ] } \right ) \nonumber\\ + \left ( a_{2,2}^{\left [ 2\right ] } + a_{0,1}^{\left [ 2\right ] } \right ) \left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } \right ) \nonumber\\ = \alpha_{e}\int_{\gamma}\partial_{\mathbf{n}}u_{e,1|\gamma}v^{\left [ 2\right ] } \left ( m,0\right ) \ d\gamma-\alpha_{i}\int_{\gamma}\partial_{\mathbf{n}}u_{i,1|\gamma}v^{\left [ 1\right ] } \left ( m,0\right ) \
d\gamma,\label{25}\]]@xmath141 } \left ( m,0\right ) \ d\gamma+a_{0,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{3}^{\left [ 1\right ] } -\alpha_{i}u_{i,3},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{1,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{2}^{\left [ 1\right ] } -\alpha_{i}u_{i,2},v^{\left [ 1\right ] } \right ) + \left ( a_{2,2}^{\left [ 1\right ] } + a_{0,1}^{\left [ 1\right ] } \right ) \left ( \alpha_{\delta}u_{1}^{\left [ 1\right ] } -\alpha_{i}u_{i,1},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{1,1}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0},v^{\left [ 1\right ] }
\right ) + a_{0,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{3}^{\left [ 2\right ] } -\alpha_{e}u_{e,3},v^{\left [ 2\right ] } \right ) \nonumber\\ + a_{1,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{2}^{\left [ 2\right ] } -\alpha_{e}u_{e,2},v^{\left [ 2\right ] } \right ) + \left ( a_{2,2}^{\left [ 2\right ] } + a_{0,1}^{\left [ 2\right ] } \right ) \left ( \alpha_{\delta}u_{1}^{\left [ 2\right ] } -\alpha_{e}u_{e,1},v^{\left [ 2\right ] } \right ) \nonumber\\ + a_{1,1}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } \right ) -\alpha_{e}\int_{\gamma } \partial_{\mathbf{n}}u_{e,2|\gamma}v^{\left [ 2\right ] } \left ( m,0\right ) \
d\gamma=0,\label{25.5}\]]@xmath142 } \left ( m,0\right ) \ d\gamma+a_{0,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{n+1}^{\left [ 1\right ] } -\alpha_{i}u_{i , n+1},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{1,2}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{n}^{\left [ 1\right ] } -\alpha_{i}u_{i , n},v^{\left [ 1\right ] } \right ) + a_{0,1}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{n-1}^{\left [ 1\right ] } -\alpha_{i}u_{i , n-1},v^{\left [ 1\right ] } \right ) \nonumber\\ + a_{1,1}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{n-2}^{\left [ 1\right ] } -\alpha_{i}u_{i , n-2},v^{\left [ 1\right ] } \right ) + a_{2,1}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{n-3}^{\left [ 1\right ] } -\alpha _ { i}u_{i , n-3},v^{\left [ 1\right ] } \right ) \nonumber\\ + \sum\limits_{l=3}^{n}a_{l-1,2}^{\left [ 1\right ] } \left ( \alpha_{\delta } u_{n - l}^{\left [ 1\right ] } -\alpha_{i}u_{i , n - l},v^{\left [ 1\right ] } \right ) + a_{0,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{n+1}^{\left [ 2\right ] } -\alpha_{i}u_{e ,
n+1},v^{\left [ 2\right ] } \right ) \nonumber\\ + a_{1,2}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{n}^{\left [ 2\right ] } -\alpha_{i}u_{e , n},v^{\left [ 2\right ] } \right ) + a_{0,1}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{n-1}^{\left [ 2\right ] } -\alpha_{i}u_{e , n-1},v^{\left [ 2\right ] } \right ) \nonumber\\ + a_{1,1}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{n-2}^{\left [ 2\right ] } -\alpha_{i}u_{e , n-2},v^{\left [ 2\right ] } \right ) + a_{2,1}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{n-3}^{\left [ 2\right ] } -\alpha _ { i}u_{e , n-3},v^{\left [ 2\right ] } \right ) \nonumber\\ + \sum\limits_{l=3}^{n}a_{l-1,2}^{\left [ 2\right ] } \left ( \alpha_{\delta } u_{n - l}^{\left [ 2\right ] } -\alpha_{i}u_{e , n - l},v^{\left [ 2\right ] } \right ) -\alpha_{e}\int_{\gamma}\partial_{\mathbf{n}}u_{e , n|\gamma } v^{\left [ 2\right ] } \left ( m,0\right ) d\gamma=0,\ n\geq3.\label{26}\ ] ] in this paragraph , we first recall some theoretical results needed for our calculation . after , we calculate explicitly the first two terms of expansions ( [ 13])-([14 ] ) and ( [ 16 ] ) in order to present a recursive method to define successively the terms of these expansions .
let @xmath143 be a nonnegative real number .
we define @xmath144 ( see @xcite ) the space of functions in @xmath2 , with @xmath145-regularity in @xmath146 and @xmath147 , as follows@xmath148 equipped with the norm@xmath149 we need the following theorem .
its proof @xcite is an application of the reflection principle @xcite .
[ theo1]let @xmath150 belongs to @xmath151 @xmath152 .
then the following problem @xmath24{ll}-div\left ( \alpha_{i}\nabla u_{i}\right ) = 0 & \text{in } \omega_{i},\\ -div\left ( \alpha_{e}\nabla u_{e}\right ) = 0 & \text{in } \omega_{e},\\ u_{i|\gamma}=u_{e|\gamma } & \text{on } \gamma,\\ \alpha_{i}\partial_{\mathbf{n}}u_{i|\gamma}-\alpha_{e}\partial_{\mathbf{n}}u_{e|\gamma}=g & \text{on } \gamma,\\ u_{e|\partial\omega}=0 & \text{on } \partial\omega , \end{array } \right.\ ] ] admits a unique solution @xmath153 in @xmath154 moreover , if @xmath155 is a nonnegative integer , and @xmath156 .
then @xmath157 we also need the following technical lemma .
this construction was motivated by @xcite and its proof is a straightforward verification . [ lem1]for @xmath158 , let @xmath159 } $ ] be a given function in @xmath160 and let @xmath161 } $ ] be a vectorial function in @xmath162 such that the partial application @xmath163 } ( .,s_{\beta})$ ] is valued in the space of vectorial fields tangent to @xmath31 and also @xmath164 } \in l^{2}(\omega^{\beta}).$ ] then the solution @xmath165 } $ ] of the variational equation@xmath166 } v^{\left [ \beta\right ] } : = \int _ { \omega^{\beta}}h^{\left [ \beta\right ] } \partial_{s_{\beta}}v^{\left [ \beta\right ] }
\ d\gamma ds_{\beta}+\int_{\omega^{\beta}}k^{\left [ \beta\right ] } .\nabla_{\gamma}v^{\left [ \beta\right ] } \
d\gamma ds_{\beta } = 0;\\ \forall v^{\left [ \beta\right ] } \in h^{1}(\omega^{\beta}),\ v^{\left [ \beta\right ] } ( .,0)=0,\end{gathered}\ ] ] is explicitly given by@xmath167 } \left ( m , s_{\beta}\right ) = \int_{s_{\beta}}^{(-1)^{\beta}}div_{\gamma}k^{\left [ \beta\right ] } \left ( m,\lambda \right ) \ d\lambda.\ ] ] moreover , if @xmath168 } ( .,0)\neq0,$ ] we have@xmath169 } v^{\left [ \beta\right ] } & = ( -1)^{\beta+1}\int_{\gamma}h^{\left [ \beta\right ] } ( m,0)v^{\left [ \beta\right ] } ( m,0)\ d\gamma\\ & = \int_{\gamma}\left [ ( -1)^{\beta+1}\int_{0}^{(-1)^{\beta}}div_{\gamma } k^{\left [ \beta\right ] } \left ( m , s_{\beta}\right ) \ ds_{\beta}\right ] v^{\left [ \beta\right ] } \left ( m,0\right ) \ d\gamma.\end{aligned}\ ] ] equation ( [ 23 ] ) implies that @xmath170 } = 0.$ ] using ( [ 18 ] ) , ( [ 17 ] ) and ( [ 1.1.8 ] ) , we obtain@xmath171 } ( m , s_{1})=u_{0}^{\left [ 2\right ] } ( m , s_{2})=u_{e,0|\gamma},\ m\in\gamma.\label{27}\ ] ] the choice of @xmath172 such that @xmath173 } = 0 $ ] in ( [ 24 ] ) gives @xmath138 } \left [ \alpha_{\delta}u_{1}^{\left [ 1\right ] } -\alpha_{i}u_{i,1},v^{\left [ 1\right ] } \right ] = 0.\ ] ] an intgration by parts in @xmath174 leads to @xmath175 } = \alpha_{i}\partial_{\mathbf{n}}u_{i,0|\gamma}.\label{28}\ ] ] similarly , the choice of @xmath172 such that @xmath176 } = 0 $ ] in ( [ 24 ] ) gives@xmath177 } ( \alpha_{\delta}u_{1}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } ) = 0.\ ] ] we obtain @xmath178 } = \alpha_{e}\partial_{\mathbf{n}}u_{e,0|\gamma}.\label{29}\ ] ] therefore @xmath179 } ( m,0)d\gamma=\alpha_{e}\int_{\gamma}\partial_{\mathbf{n}}u_{e,0|\gamma } v^{\left [ 2\right ] } ( m,0)d\gamma.\ ] ] as @xmath176 } ( m,0)=v^{\left [ 2\right ] } ( m,0)$ ] , @xmath180 let us define @xmath181 and @xmath182 @xmath183 by @xmath184{ll}\alpha_{e } & \text{if } x\in\omega_{e},\\ \alpha_{i } & \text{if } x\in\omega_{i } , \end{array } \right .
\text { and \ } u_{n}=\left\ { \begin{array } [ c]{ll}u_{e , n } & \text{in } \omega_{e},\\ u_{i , n } & \text{in } \omega_{i}. \end{array } \right.\ ] ] therefore , with ( [ 15 ] ) , ( [ 27 ] ) and ( [ 30 ] ) , @xmath185 satisfies the following problem @xmath24{ll}-div\left ( \alpha_{0}\nabla u_{0}\right ) = f & \text{in } \omega,\\ u_{0|\partial\omega}=0 & \text{on } \partial\omega . \end{array } \right.\ ] ] elliptic regularity results ( see e.g. @xcite ) show that if @xmath20 belongs to @xmath186 , then @xmath187 is a well - defined element of @xmath188 as a consequence the first term is determined . integrating relations ( [ 28 ] ) and ( [ 29 ] ) in @xmath189
yields@xmath190 } ( m , s_{1 } ) & = u_{i,1|\gamma}+p_{1}\left [ ( s_{1}+1)\alpha_{i}\alpha_{\delta}^{-1}-1\right ] \partial_{\mathbf{n}}u_{i,0|\gamma},\ \forall(m , s_{1})\in\omega^{1},\\ u_{1}^{\left [ 2\right ] } ( m , s_{2 } ) & = u_{e,1|\gamma}+p_{2}\left [ ( s_{2}-1)\alpha_{e}\alpha_{\delta}^{-1}+1\right ] \partial_{\mathbf{n}}u_{e,0|\gamma},\ \forall(m , s_{2})\in\omega^{2}.\end{aligned}\ ] ] by identifying terms of order 1 in ( [ 17 ] ) and ( [ 18 ] ) , we obtain the first transmission condition on @xmath31@xmath191 the second one follows the same lines as for order @xmath192 the choice of @xmath172 such that @xmath173 } = 0 $ ] in ( [ 25 ] ) gives@xmath138 } \left ( \alpha_{\delta}u_{2}^{\left [ 1\right ] } -\alpha_{i}u_{i,2},v^{\left [ 1\right ] } \right ) + a_{0,1}^{\left [ 1\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0},v^{\left [ 1\right ] } \right ) = 0.\ ] ] we apply lemma [ lem1 ] with@xmath193 } & = p_{1}^{-1}\alpha_{\delta}\ \partial_{s_{1}}u_{2}^{\left [ 1\right ] } -p_{1}^{-1}\alpha_{i}u_{i,2}=p_{1}^{-1}\alpha_{\delta}\ \partial_{s_{1}}u_{2}^{\left [ 1\right ] } -\alpha_{i}\partial_{\mathbf{n}}u_{i,1|\gamma}-s_{1}p_{1}\partial_{\mathbf{n}}^{2}u_{i,0|\gamma},\\ k^{\left [ 1\right ] } & = p_{1}\nabla_{\gamma}\left ( p_{1}^{-1}u_{0}^{\left [ 1\right ] } -\alpha_{i}u_{i,0}\right ) = p_{1}\left ( \alpha_{\delta}-\alpha_{i}\right ) \nabla_{\gamma}u_{i,0|\gamma},\end{aligned}\ ] ] we find@xmath194 } ( m , s_{1})-\alpha_{i}\partial_{\mathbf{n}}u_{i,1|\gamma}-s_{1}p_{1}\partial_{\mathbf{n}}^{2}u_{i,0|\gamma}=-\left ( s_{1}+1\right ) p_{1}\left ( \alpha_{\delta}-\alpha_{i}\right ) \delta_{\gamma}u_{i,0|\gamma}.\ ] ] morover , for all @xmath176 } $ ] we obtain @xmath195 } v^{\left [ 1\right ] } = -\int_{\gamma}p_{1}\left ( \alpha_{\delta}-\alpha_{1}\right ) \delta_{\gamma}u_{i,0|\gamma } \ v^{\left [ 1\right ] } ( m,0)\ d\gamma.\ ] ] again similarly , the choice of @xmath172 such that @xmath176 } = 0 $ ] in ( [ 25 ] ) gives @xmath177 } \left ( \alpha_{\delta}u_{2}^{\left [ 2\right ] } -\alpha_{e}u_{e,2},v^{\left [ 2\right ] } \right ) + a_{0,1}^{\left [ 2\right ] } \left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0},v^{\left [ 2\right ] } \right ) = 0.\ ] ] we apply lemma [ lem1 ] with @xmath196 } & = p_{2}^{-1}\alpha_{\delta}\ \partial_{s_{2}}u_{2}^{\left [ 2\right ] } -p_{2}^{-1}\alpha_{e}u_{e,2}=p_{2}^{-1}\alpha_{\delta}\ \partial_{s_{2}}u_{2}^{\left [ 2\right ] } -\alpha_{e}\partial_{\mathbf{n}}u_{e,1|\gamma}-s_{2}p_{2}\partial_{\mathbf{n}}^{2}u_{e,0|\gamma},\\ k^{\left [ 2\right ] } & = p_{2}\nabla_{\gamma}\left ( \alpha_{\delta}u_{0}^{\left [ 2\right ] } -\alpha_{e}u_{e,0}\right ) = p_{2}\left ( \alpha_{\delta}-\alpha_{e}\right ) \nabla_{\gamma}u_{e,0|\gamma},\end{aligned}\ ] ] we find@xmath197 } ( m , s_{2})-\alpha_{e}\partial_{\mathbf{n}}u_{e,1|\gamma}-s_{2}p_{2}\partial_{\mathbf{n}}^{2}u_{e,0|\gamma}=\left ( 1-s_{2}\right ) \left ( \alpha_{\delta}-\alpha_{e}\right ) \delta_{\gamma}u_{e,0|\gamma}.\ ] ] morover , for all @xmath173 } $ ] we obtain@xmath198 } v^{\left [ 2\right ] } = -\int_{\gamma}p_{2}\left ( \alpha_{\delta}-\alpha_{e}\right ) \delta_{\gamma}u_{e,0|\gamma } \ v^{\left [ 2\right ] } ( m,0)\ d\gamma.\ ] ] as a consequence , @xmath199 v^{\left [ 1\right ] } \left ( m,0\right ) \ d\gamma\\ & = \int_{\gamma}\left [ \alpha_{e}\partial_{\mathbf{n}}u_{e,1|\gamma}+p_{2}\left ( \alpha_{\delta}-\alpha_{e}\right ) \delta_{\gamma}u_{e,0|\gamma } \right ] v^{\left [ 2\right ] } \left ( m,0\right ) \ d\gamma.\end{aligned}\ ] ] as @xmath176 } ( m,0)=v^{\left [ 2\right ] } ( m,0)$ ] , @xmath200 it follows from ( [ 15 ] ) , ( [ 30a ] ) , ( [ 31 ] ) and theorem [ theo1 ] that @xmath201 is the unique solution of the following problem@xmath24{ll}-div\left ( \alpha_{i}\nabla u_{i,1}\right ) = 0 & \text{in } \omega_{i},\\ -div\left ( \alpha_{e}\nabla u_{e,1}\right ) = 0 & \text{in } \omega_{e},\\ u_{e,1|\partial\omega}=0 & \text{on } \partial\omega , \end{array } \right.\ ] ] with transmission conditions on @xmath31@xmath202 or @xmath203 \partial_{\mathbf{n}}u_{i,0|\gamma},\\ \alpha_{i}\partial_{\mathbf{n}}u_{i,1|\gamma}-\alpha_{e}\partial_{\mathbf{n}}u_{e,1|\gamma } & = \left [ p_{1}(\alpha_{\delta}-\alpha_{i})+p_{2}(\alpha_{\delta}-\alpha_{e})\right ] \delta_{\gamma}u_{i,0|\gamma}.\end{aligned}\ ] ]
the process described in the previous section can be continued up to any order provided that the data are sufficiently regular .
we can also estimate the error made by truncating the series after a finite number of terms .
let @xmath204 be in @xmath205 we set@xmath206{c}u_{d_{1},\delta}^{\left ( n\right ) } : = \sum\limits_{j=0}^{n}\delta^{j}u_{d_{1},j}\text { in } \omega_{\delta,1,}\\ u_{d_{2},\delta}^{\left ( n\right ) } : = \sum\limits_{j=0}^{n}\delta^{j}u_{d_{2},j}\text { in } \omega_{\delta,2 } , \end{array } \right.\ ] ] where @xmath207 } ( m , s_{\beta});$ ] @xmath208 .
[ convergence theorem][theo2]for all integers @xmath204 , there exists a constant @xmath209 independent of @xmath4 such as @xmath210 since @xmath20 is @xmath211 all terms in expansions ( [ 13 ] ) , ( [ 14 ] ) and ( [ 16 ] ) up to order @xmath212 may be obtained from equations ( [ 23])-([26 ] ) .
let us define the remainders @xmath213 and @xmath214 of taylor expansions in the normal variable with respect to @xmath4 up to order @xmath204 of @xmath215 and @xmath216 respectively by @xmath217 } & : = \left ( u_{i,\delta}^{\left ( n\right ) } \right ) ^{\left [ 1\right ] } -\sum\limits_{j=0}^{n}\sum\limits_{l=0}^{n - j}\frac{(s_{1})^{l}\delta^{j+l}}{l!}p_{1}^{l}\partial_{\mathbf{n}}^{l}u_{i , j|\gamma}:=\left ( u_{i,\delta}^{\left ( n\right ) } \right ) ^{\left [ 1\right ] } -\sum\limits_{j=0}^{n}\delta^{j}u_{i , j},\label{35}\\ r_{2,n}^{\left [ 2\right ] } & : = \left ( u_{e,\delta}^{\left ( n\right ) } \right ) ^{\left [ 2\right ] } -\sum\limits_{j=0}^{n}\sum\limits_{l=0}^{n - j}\frac{\left ( s_{2}\right ) ^{l}\delta^{j+l}}{l!}p_{2}^{l}\partial_{\mathbf{n}}^{l}u_{e , j|\gamma}:=\left ( u_{e,\delta}^{\left ( n\right ) } \right ) ^{\left [ 2\right ] } -\sum\limits_{j=0}^{n}\delta ^{j}u_{e , j},\label{36}\ ] ] where @xmath218 we shall rely on the following proposition to show the estimates of the remainders @xmath219 and @xmath220 .
the steps of the proof are very similar to those given in ( * ? ? ?
* section 5 ) .
we refer the reader to this paper .
[ prop1]there exists a constant @xmath221 independent of @xmath135 such as@xmath222 moreover , there exists an extension @xmath223 of @xmath219 into @xmath11 with@xmath224 [ continuation of the proof of theorem [ theo2]]let @xmath225 and @xmath226 be the remainders made by truncating series ( [ 13 ] ) , ( [ 14 ] ) and ( [ 16])@xmath227 and @xmath228 be the linear form defined on @xmath229 @xmath230 in which @xmath223 is the extension function of @xmath219 into @xmath231 using green s formula in @xmath147 and in @xmath146 with the help of ( [ 15 ] ) , we obtain @xmath232 } ( u_{0}^{\left [ \beta\right ] } + \cdots+\delta^{n}u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } ) \right ] \\ & + \alpha_{i}\int_{\omega_{\delta,1}}\nabla u_{i,\delta}^{\left ( n\right ) } .\nabla
v_{1}\ d\omega_{\delta,1}+\alpha_{e}\int_{\omega_{\delta,2}}\nabla u_{e,\delta}^{\left ( n\right ) } .\nabla v_{2}\ d\omega_{\delta,2}\\ & + \alpha_{e}\int_{\gamma}\left ( \partial_{\mathbf{n}}u_{e,0|\gamma}+\cdots+\delta^{n}\partial_{\mathbf{n}}u_{e , n|\gamma}\right ) v_{e|\gamma } \ d\gamma-\alpha_{\delta}\int_{\omega_{\delta}}\nabla\mathcal{p}r.\nabla v_{d}\ d\omega_{\delta}.\end{aligned}\ ] ] it follows , from ( [ 33])-([36 ] ) , that@xmath233 } ( u_{0}^{\left [ \beta\right ] } + \cdots+\delta ^{n}u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } ) \\ & + \alpha_{i}\delta a_{\delta}^{\left [ 1\right ] } ( u_{i,0}+\cdots+\delta ^{n}u_{i , n},v^{\left [ 1\right ] } ) + \alpha_{e}\delta a_{\delta}^{\left [ 2\right ] } ( u_{e,0}+\cdots+\delta^{n}u_{e , n},v^{\left [ 2\right ] } ) \\ & -\alpha_{i}\int_{\gamma}\left ( \partial_{\mathbf{n}_{\delta,1}}u_{i,0|\gamma}+\cdots+\delta^{n}\partial_{\mathbf{n}}u_{i , n|\gamma}\right ) v_{i|\gamma}\ d\gamma\\ & + \alpha_{e}\int_{\gamma}\left ( \partial_{\mathbf{n}}u_{e,0|\gamma}+\cdots+\delta^{n}\partial_{\mathbf{n}}u_{e , n|\gamma}\right ) v_{e|\gamma } \
d\gamma.\end{aligned}\ ] ] now , we use the fact that @xmath234 } , \ldots , u_{n+1}^{\left [ \beta\right ] } , $ ] @xmath235 are solutions of equations ( [ 23])-([26 ] ) , we obtain@xmath236 } \left ( u_{n+1}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) -\left ( a_{2,2}^{\left [ \beta\right ] } + a_{0,1}^{\left [ \beta\right ] } \right ) \left ( u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) \right . \\ & -a_{1,1}^{\left [ \beta\right ] } \left ( u_{n-1}^{\left [ \beta\right ] } + \delta u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) -a_{2,1}^{\left [ \beta\right ] } \left ( u_{n-2}^{\left [ \beta\right ] } + \delta u_{n-1}^{\left [ \beta\right ] } + \delta^{2}u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) -\cdots\\ & \left .
-a_{n-1,1}^{\left [ \beta\right ] } \left ( u_{1}^{\left [ \beta\right ] } + \cdots+\delta^{n-1}u_{n-1}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) -r_{n}^{\left [ \beta\right ] } \left ( \delta ; u_{1}^{\left [ \beta\right ] } + \cdots+\delta^{n}u_{n}^{\left [ \beta\right ] } , v^{\left [ \beta\right ] } \right ) \right\ } \\ & + \alpha_{i}\int_{\omega_{i,\delta}}\nabla r_{i , n}.\nabla v_{i}\ d\omega_{i,\delta}+\alpha_{e}\int_{\omega_{e,\delta}}\nabla r_{e , n}.\nabla v_{e}\ d\omega_{e,\delta}-\alpha_{\delta}\int_{\omega_{\delta}}\nabla \mathcal{p}r.\nabla v_{d}\ d\omega_{\delta}.\end{aligned}\ ] ] by the estimates based on the explicit expressions of the bilinear form @xmath128 } ( .,.)$ ] and those of propositions [ prop1 ] , we have@xmath237 } \right\vert _ { l^{2}(\omega^{\beta})}+\delta^{-1}\left\vert \partial
_ { s_{\beta}}v^{\left [ \beta\right ] } \right\vert _ { l^{2}(\omega^{\beta})}+\left\vert v^{\left [ \beta\right ] } \right\vert _ { l^{2}(\omega^{\beta})}\right ) \\ & + c\delta^{n-1/2}\left ( \left\vert v_{i}\right\vert _ { h^{1}(\omega _ { i,\delta})}+\sum_{\beta=1}^{2}\left\vert v_{\beta}\right\vert _ { h^{1}(\omega_{\delta,\beta})}+\left\vert v_{e}\right\vert _ { h^{1}(\omega_{e,\delta } ) } \right ) .\end{aligned}\ ] ] since @xmath4 is small enough , we have , @xmath238@xmath239 } \right\vert _ { l^{2}(\omega^{\beta})}+\delta ^{\frac{-1}{2}}\left\vert \partial_{s_{\beta}}v^{\left [ \beta\right ] } \right\vert _ { l^{2}(\omega^{\beta})}+\delta^{\frac{1}{2}}\left\vert v^{\left [ \beta\right ] } \right\vert _ { l^{2}(\omega^{\beta})}\right ) \\ & + c\delta^{n-1/2}\left ( \left\vert v_{i}\right\vert
_ { h^{1}(\omega _ { i,\delta})}+\sum_{\beta=1}^{2}\left\vert v_{\beta}\right\vert _ { h^{1}(\omega_{\delta,\beta})}+\left\vert v_{e}\right\vert _ { h^{1}(\omega_{e,\delta } ) } \right ) .\end{aligned}\ ] ] therefore @xmath240 since @xmath241 is in @xmath242 we set in ( [ 36.1 ] ) @xmath243 we obtain@xmath244 thanks to proposition [ prop1 ] , we find @xmath245 moreover , since @xmath246 and @xmath247 are @xmath248 and for every integer @xmath249 , we have @xmath250 @xmath251 and @xmath252 , therefore ( see @xcite)@xmath253 this completes the proof .
this section is devoted to the approximation of @xmath30 by a solution of a problem modelling the effect of the thin layer with a precision of order two in @xmath254 we truncate the series defining the asymptotic expansions , keeping only the first two terms@xmath255 } ( m , s_{1})+\delta u_{1}^{\left [ 1\right ] } ( m , s_{1}),\ \forall x=\phi_{1}(m , s_{1})\in\omega_{\delta,1},\\ u_{d_{2},\delta}(x ) & \simeq u_{d_{2},\delta}^{\left ( 1\right ) } ( m , s_{2}):=u_{0}^{\left [ 2\right ] } ( m , s_{2})+\delta u_{1}^{\left [ 2\right ] } ( m , s_{2}),\ \forall x=\phi_{2}(m , s_{2})\in\omega_{\delta,2},\end{aligned}\ ] ] where @xmath256 is the solution of@xmath24{ll}-div\left ( \alpha_{i}\nabla u_{i,\delta}^{\left ( 1\right ) }
\right ) = f_{|\omega_{i } } & \text{in } \omega_{i},\\ -div\left ( \alpha_{e}\nabla u_{e,\delta}^{\left ( 1\right ) } \right ) = f_{|\omega_{e } } & \text{in } \omega_{e},\\ u_{i,\delta|\gamma}^{\left ( 1\right ) } -u_{e,\delta|\gamma}^{\left ( 1\right ) } = \delta\mathcal{a}\left ( u_{i,\delta}^{\left ( 1\right ) } \right ) -\delta^{2}\xi_{\delta } & \text{on } \gamma,\\ \alpha_{i}\partial_{\mathbf{n}}u_{i,\delta|\gamma}^{\left ( 1\right ) } -\alpha_{e}\partial_{\mathbf{n}}u_{e,\delta|\gamma}^{\left ( 1\right ) } = \delta\mathcal{b}\left ( u_{i,\delta}^{\left ( 1\right ) } \right ) -\delta^{2}\rho_{\delta } & \text{on}\ \gamma,\\ u_{\delta|\partial\omega}^{\left ( 1\right ) } = 0 & \text{on } \partial\omega , \end{array } \right .
\label{41}\ ] ] with@xmath257 \left ( \partial_{\mathbf{n}}u_{|\gamma}\right ) , \\ \mathcal{b}\left ( u\right ) & : = \left [ p_{1}(\alpha_{\delta}-\alpha _ { i})+p_{2}(\alpha_{\delta}-\alpha_{e})\right ] \delta_{\gamma}u_{|\gamma},\\ \xi_{\delta } & : = \left [ p_{1}(1-\alpha_{i}\alpha_{\delta}^{-1})+p_{2}(\alpha_{i}\alpha_{e}^{-1}-\alpha_{i}\alpha_{\delta}^{-1})\right ] \partial_{\mathbf{n}}u_{i,1|\gamma},\\ \rho_{\delta } & : = \left [ p_{1}(\alpha_{\delta}-\alpha_{i})+p_{2}(\alpha_{\delta}-\alpha_{e})\right ] \delta_{\gamma}u_{i,1|\gamma}.\end{aligned}\ ] ] let @xmath258 be the solution of ( [ 41 ] ) with @xmath259 and @xmath260 .
we obtain a problem @xmath261 with transmission conditions of order equal to that of the differential operator .
the new transmission conditions on @xmath31 are defined by @xmath24{l}u_{i,\delta|\gamma}^{ap}-u_{e,\delta|\gamma}^{ap}=\delta\left [ p_{1}(1-\alpha_{i}\alpha_{\delta}^{-1})+p_{2}(\alpha_{i}\alpha_{e}^{-1}-\alpha _ { i}\alpha_{\delta}^{-1})\right ] \partial_{\mathbf{n}}u_{i,\delta|\gamma } ^{ap},\\ \alpha_{i}\partial_{\mathbf{n}}u_{i,\delta|\gamma}^{ap}-\alpha_{e}\partial_{\mathbf{n}}u_{e,\delta|\gamma}^{ap}=\delta\left [ p_{1}(\alpha_{\delta}-\alpha_{i})+p_{2}(\alpha_{\delta}-\alpha_{e})\right ] \delta_{\gamma}u_{i,\delta|\gamma}^{ap}. \end{array } \right .
\label{42}\ ] ] however , the bilinear form associated to problem @xmath262 is neither positive nor negative .
then the existence and uniqueness of the solution are not ensured by the lax milgram lemma .
therefore , we reformulate problem @xmath261 into a nonlocal equation on the interface @xmath31 ( cf .
a direct use of transmission conditions ( [ 42 ] ) leads to an operator which is not self - adjoint .
so , we choose the position of @xmath31 in such a way that the jump of the trace of the solution on @xmath31 is null .
we put @xmath263 we obtain@xmath264 which is valid only when @xmath18 or @xmath265 this corresponds to the case of mid - diffusion . transmission conditions ( [ 42 ] )
become @xmath24{l}u_{i,\delta|\gamma}^{ap}-u_{e,\delta|\gamma}^{ap}=0,\\ \alpha_{i}\partial_{\mathbf{n}}u_{i,\delta|\gamma}^{ap}-\alpha_{e}\partial_{\mathbf{n}}u_{e,\delta|\gamma}^{ap}=\delta\dfrac{\left ( \alpha _ { e}-\alpha_{\delta}\right ) \left ( \alpha_{i}-\alpha_{\delta}\right ) } { \alpha_{\delta}}\delta_{\gamma}u_{i,\delta}^{ap}. \end{array } \right.\ ] ] after , we remove the right - hand side of problem @xmath262 by a standard lift : let @xmath150 be in @xmath266 such that @xmath267 then the function @xmath268 solves the following problem@xmath24{ll}-div\left ( \alpha_{i}\nabla\psi_{i}\right ) = 0 & \text{in } \omega_{i},\\ -div\left ( \alpha_{e}\nabla\psi_{e}\right ) = 0 & \text{in } \omega_{e},\\ \psi_{i|\gamma}-\psi_{e|\gamma}=0 & \text{on } \gamma,\\ \alpha_{i}\partial_{\mathbf{n}}\psi_{i|\gamma}-\alpha_{e}\partial_{\mathbf{n}}\psi_{e|\gamma}-\delta\dfrac{\left ( \alpha_{e}-\alpha_{\delta}\right ) \left ( \alpha_{i}-\alpha_{\delta}\right ) } { \alpha_{\delta}}\delta_{\gamma } \psi_{i|\gamma}=g & \text{on}\ \gamma,\\ \psi_{e|\partial\omega}=0 & \text{on } \partial\omega , \end{array } \right.\ ] ] where @xmath269 we introduce the steklov - poicar operators @xmath270 and @xmath271 ( called also dirichlet - to - neumann operators ) defined from @xmath272 onto @xmath273 by @xmath274 where @xmath275 is the solution of the boundary value problem@xmath24{ll}-\delta u_{i}=0 & \text{in } \omega_{i},\\ u_{i|\gamma}=\varphi & \text{on } \gamma , \end{array } \right.\ ] ] and by @xmath276 where @xmath277 is the solution of the boundary value problem @xmath24{ll}-\delta u_{e}=0 & \text{in } \omega_{e},\\ u_{e|\gamma}=\psi & \text{on } \gamma,\\ u_{e|\partial\omega}=0 & \text{on } \partial\omega . \end{array } \right.\ ] ] then @xmath261 is equivalent to the boundary equation @xmath278 where @xmath279 is the trace of @xmath280 on the surface @xmath281 we are now in position to state the existence and uniqueness theorem , which proof is similar to that of theorem 2.5 in @xcite .
[ theo3]the operator @xmath282 is an elliptic self - adjoint semi - bounded from below pseudodifferential operator of order 2 .
moreover , there exists series @xmath283 growing to infinity such that for any @xmath284 with @xmath285 , we have the following : 1 . if @xmath286 then equation @xmath287 admits a unique solution in @xmath288 which , in addition , belongs to @xmath289 2 . if @xmath290 then there is either no solution or a complete affine finite dimensional space of @xmath291 solutions .
finally , we give an error estimate between the solution @xmath30 of ( [ 1 ] ) and the approximate solution @xmath292 let us define @xmath293 on @xmath52 @xmath294 , ap}(m , s_{1}):=u_{i,\delta|\gamma}^{ap}+\delta\frac{\alpha _ { i}\left ( \alpha_{e}-\alpha_{\delta}\right ) } { \alpha_{\delta}\left ( \alpha_{e}-\alpha_{i}\right ) } \left [ ( s_{1}+1)\alpha_{i}\alpha_{\delta}^{-1}-1\right ] \partial_{\mathbf{n}}u_{i,\delta|\gamma}^{ap},\\ u_{d_{2},\delta}^{ap}(x ) & : = u_{d_{2},\delta}^{\left [ 2\right ] , ap}\left ( m , s_{2}\right ) : = u_{e,\delta|\gamma}^{ap}+\delta\frac{\alpha_{e}\left ( \alpha_{\delta}-\alpha_{i}\right ) } { \alpha_{\delta}\left ( \alpha_{e}-\alpha_{i}\right ) } \left [ ( s_{2}-1)\alpha_{e}\alpha_{\delta}^{-1}+1\right ] \partial_{\mathbf{n}}u_{e,\delta|\gamma}^{ap},\end{aligned}\ ] ] and let us denote by @xmath295 the approximate solution defined on @xmath2 @xmath296{cc}u_{i,\delta}^{ap } & \text{in } \omega_{i,\delta},\\ u_{d_{\beta},\delta}^{ap } & \text{in } \omega_{\delta,\beta},\\ u_{e,\delta}^{ap } & \text{in } \omega_{e,\delta}. \end{array } \right.\ ] ] we can now formulate our main result . according to the convergence theorem
, it is sufficient to estimate the error @xmath298 .
therefore , as in @xcite , we perform an asymptotic expansion for @xmath299 .
the ansatz@xmath300 where @xmath301 and @xmath302 , gives the recurrence relations@xmath24{ll}-div\left
( \alpha_{i}\nabla w_{i , j}\right ) = f_{|\omega_{i}}\delta_{j,0 } & \text{in } \omega_{i},\\ -div\left ( \alpha_{e}\nabla w_{e , j}\right ) = f_{|\omega_{e}}\delta_{j,0 } & \text{in } \omega_{e},\\ w_{i , j|\gamma}-w_{e , j|\gamma}=0 & \text{on } \gamma,\\ \alpha_{i}\partial_{\mathbf{n}}w_{i , j|\gamma}-\alpha_{e}\partial_{\mathbf{n}}w_{e , j|\gamma}=\dfrac{\left ( \alpha_{e}-\alpha_{\delta}\right ) \left ( \alpha_{i}-\alpha_{\delta}\right ) } { \alpha_{\delta}}\delta_{\gamma } w_{i , j-1|\gamma } & \text{on}\ \gamma,\\ w_{e , j|\partial\omega}=0 & \text{on } \partial\omega , \end{array } \right.\ ] ] with the convention that @xmath303 a simple calculation shows that the two first terms @xmath304 and @xmath305 coincide with the two first terms of ( [ 13 ] ) and ( [ 14 ] ) . furthermore , since @xmath247 and @xmath246 are @xmath211 each term of ( [ 46 ] ) is bounded in @xmath306 .
then , by setting @xmath307 there exists @xmath308 , such as @xmath309 which gives the desired result .
99 s. agmon , a. douglis and l. nirenberg , estimates near the boundary for solutions of elliptic partial differential equations satisfying general boundary conditions .
ii , comm . pure appl .
17 ( 1964 ) 35 - 92 .
a. bendali and k. lemrabet , asymptotic analysis of the scattering of a time - harmonic electromagnetic wave by a perfectly conducting metal coated with a thin dielectric shell , asymptot .
57 ( 2008 ) 199 - 227 . c. poignard , about the transmembrane voltage potential of a biological cell in time - harmonic regime , in : mathematical methods for imaging and inverse problems , volume 26 of esaim proc .
edp sci . les ulis ( 2009 ) 162 - 179 .
c. poignard , mthodes asymptotiques pour le calcul des champs lectromagntiques dans des milieux couches minces . application aux cellules biologiques , thse de doctorat , universit claude bernard - lyon 1 , 2006 . | this work consists in the asymptotic analysis of the solution of poisson equation in a bounded domain of @xmath0 @xmath1 with a thin layer .
we use a method based on hierarchical variational equations to derive asymptotic expansion of the solution with respect to the thickness of the thin layer .
we determine the first two terms of the expansion and prove the error estimate made by truncating the expansion after a finite number of terms .
next , using the first two terms of the asymptotic expansion , we show that we can model the effect of the thin layer by a problem with transmission conditions of order two .
* keywords : * asymptotic analysis ; asymptotic expansion ; approximate transmission conditions ; thin layer ; poisson equation . |
the radio sky at 90100ghz remains largely uncharted territory . at frequencies above 5ghz , traditional surveys of even modest areas of sky become impractical because of the small field of view of current telescopes .
for example , the effective field of view of the nrao vla telescope is 0.22deg@xmath3 at 1.4ghz , but less than 0.0002deg@xmath3 at its highest frequency of 50ghz .
most radio telescopes designed to operate at millimetre wavelengths have similarly small fields of view for increasing frequency @xmath4 . ] , so that blind surveys of large areas of sky are only possible by using specially - designed instruments or techniques . figure [ fig.surv ]
summarizes the sensitivity limits of large area radio continuum surveys which cover at least a quarter of the sky .
there are currently only two large area surveys above 5ghz .
the wmap point source catalogue ( bennett et al .
2003 ; hinshaw et al .
2007 ) detects sources down to 0.81.0jy in five frequency bands from 2394ghz . the at20 g survey ( ricci et al .
2004 ; sadler et al .
2006 ) uses a wide band ( 8ghz bandwidth ) analogue correlator to scan the sky rapidly at 20ghz , detecting sources down to a limit of @xmath5mjy , and currently represents the highest radio frequency at which a blind survey of a large fraction of the sky has been carried to a detection limit below 100mjy .
there are several reasons why we would like to learn more about the extragalactic radio source population at frequencies well above those probed by existing large area surveys .
corrections for extragalactic foreground confusion ( e.g. de zotti et al . 2005 ) will be critical for next generation surveys of the cosmic microwave background ( cmb ) like the planck mission , which aims to measure cmb anisotropies on angular scales of 530arcmin in the frequency range 30350ghz .
such corrections require an accurate knowledge of the foreground point
source population at frequencies of 30ghz and above .
observations at 95ghz are particularly relevant to planck , since the planck 100ghz channel and the nearby 143ghz channel are the `` cleanest '' ones for cmb studies in both temperature and polarization on small angular scales , where the main foreground contamination comes from extragalactic sources .
the 100 and 143ghz channels will also be the most sensitive planck channelst / t per pixel ) is slightly better at 30ghz than at 100 and 143ghz , the smaller pixel size ( in arcmin ) at high frequencies means that the 100 and 143 ghz channels are far more sensitive when scaled to the same angular resolution . ]
( lamarre et al .
2003 ) , far exceeding the performances of wmap , so that an accurate subtraction of point source contamination at these frequencies will be very important .
a second motivation arises because current models of radio galaxy evolution imply that very young radio sources ( tens to hundreds of years old ) should have radio spectra which peak above 5ghz ( the so called gigahertz peaked spectrum , or gps sources ; odea 1998 ) , with the spectra of younger sources peaking at progressively higher frequencies .
such objects are rare , since they represent a very short lived evolutionary stage , but are predicted to be extremely luminous and hence easily detectable out to high redshift in high frequency radio surveys . until now
, models for the luminosity evolution of young radio galaxies have been almost totally unconstrained by the available data ; and the need for a large , homogeneously selected sample of gps sources is well - recognized ( e.g. snellen et al .
an analysis of evolutionary properties of gps galaxies has recently been carried out by tinti & de zotti ( 2006 ) .
in addition to these science goals , 95ghz observations are relevant to the calibration strategy for the atacama large millimetre array ( alma ; de breuck 2005 ) , which will need to do rapid switching between object and calibrators to achieve the required phase stability .
this ideally requires a network of calibrators no more than 12 degrees apart and stronger than 50 - 100 mjy at 90 ghz . the existing proposal ( holdaway et al .
2004 ) is to find 90 ghz calibrators by selecting candidates from low - frequency ( 15 ghz ) surveys and using alma itself to measure their 90 ghz flux density .
we have already suggested ( sadler et al .
2006 ) that a more efficient alternative would be to select alma calibrators from the high - frequency australia telescope 20ghz ( at20 g survey ) .
one aim of the present study is therefore to measure the surface density of at20 g sources at 90100ghz .
early work in this area dates back to the 1970s .
owen et al . ( 1978 ) used the nrao 11 m radio telescope at kitt peak to measure the 90ghz flux density of 237 strong sources which were known to have flat or inverted radio spectra at centimetre wavelengths .
they also made near - simultaneous observations of about 100 of these sources at 1.4 , 4.6 , 15 and 22ghz using dishes of the nrao vla .
their main findings were that most of their sources had flat spectra ( @xmath6 ) over the whole frequency range 190ghz , and that the level of radio variability at 90ghz was lower than expected ( most sources varied by less than 30% on a one year timescale ) .
owen , helfand & spangler ( 1981 ) observed 25 strong extragalactic millimetre sources at @xmath7kev with the _ einstein _ x ray obsetvatory , and found a strong correlation between their x ray and 90ghz flux densities .
they noted that this is consistent with a synchrotron self
compton model in which the x ray emission arises from the compton scattering of millimetre radio photons off the relativistic electrons which created them .
steppe et al .
( 1988 ) observed a sample of 294 strong ( @xmath8jy at 5ghz ) extragalactic sources at 90ghz with the iram 30 m telescope , and also measured 2090ghz spectral indices for a subset of these objects .
they found that only about 5% of compact , flat spectrum radio sources had inverted ( @xmath9 ) spectra at 2090ghz , and that most of these were variable sources undergoing an outburst .
more recent continuum work at 90100ghz has mainly focussed on variability studies of targeted samples of objects such as qsos and blazars ( e.g. landau et al .
1980 ; tersranta et al .
2004 ; hovatta et al . 2007 ) .
all these studies have been based on sources pre selected at frequencies of 5ghz or below . the wmap catalogue ( hinshaw et al .
2007 ) provides the first complete sample of radio sources selected in the 90ghz band , but only 121 of the 323 discrete sources listed in the 3year catalogue are detected at 94ghz ( limiting flux density 0.8jy ) . throughout this paper ,
we use @xmath10 = 70 km s@xmath11 mpc@xmath11 , @xmath12 = 0.3 and @xmath13 = 0.7 .
the data analysed in this paper were obtained at the australia telescope compact array ( atca ) in two observing runs during 2005 and 2006 ( see table [ tab.obs ] for details of the observations ) .
.log of our atca observations at 95ghz . [ cols="<,^,^,^ " , ]
in this study , we have shown that the atca can measure robust 95ghz ( 3 mm ) flux densities for relatively faint continuum sources in a few minutes of integration time , provided that careful attention is given to calibrating the telescope pointing .
our atca measurements at 20 and 95ghz appear to be on the same flux scale as the wmap catalogue ( hinshaw et al .
2007 ) to within the errors .
the arrival in 2008 of the new atca broadband correlator , which has a 2ghz bandwidth compared to the current value of 128mhz , will significantly increase the continuum sensitivity at 95ghz and should make it possible to measure 95ghz sources down to @xmath14mjy with integration times similar to those used here .
we have made a new estimate of the 95ghz radio source counts for sources stronger than 80mjy , and show that some previous studies have over - estimated the number of these sources by up to a factor of two . in particular , the confusion level in the planck beam at 95ghz is likely to be significantly lower than previously estimated .
fewer than half the sources in our 95ghz sample currently have measured redshifts , and obtaining good - quality optical spectra for the whole sample would be valuable both to study their rest frame radio spectra and to test whether the brightest 95ghz sources are relativistically beamed in the optical , as suggested in 4.8 . we plan to make new measurements of a complete sample of sources from the at20 g bright source sample ( massardi et al .
2008 ) at 20 , 40 and 95ghz in late 2007 , using the new atca 7 mm receiver system .
these observations , which will also include polarization measurements , should provide further insights into the properties of the strongest 95ghz sources .
we thank the staff at the australia telescope compact array for their support during our observing runs , and the at20 g and 6dfgs teams for making their data available to us in advance of publication .
we also acknowledge the support of the australian research council through the award of an arc australian professorial fellowship to ems and a federation fellowship to rde .
the australia telescope compact array is part of the australia telescope , which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
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 .
this research has also made use of data obtained from the supercosmos science archive , prepared and hosted by the wide field astronomy unit , institute for astronomy , university of edinburgh , which is funded by the uk particle physics and astronomy research council .
we thank the referee , prof .
ian browne , for a prompt and very helpful report .
99 baker j.c .
, hunstead r.w . , 1995 ,
apj , 452 , l95 bennett c.l .
et al . , 2003 , apj , 583 , 1 bock , d.c - j .
large , m.i .
sadler e.m .
1999 , aj , 117 , 1593 bolton r.c .
et al . , 2004 , mnras , 354 , 485 browne i. w.a . , wright a.e .
, 1985 , mnras , 213 , 97 cleary k.a .
et al . , 2005 , mnras , 360 , 340 cohen a.s . , lane w.m . , cotton w.d .
, kassim n.e . ,
lazio t.j.w . , perley r.a .
, condon j.j . , erickson w.c . , 2007 , aj , 134 1245 colless m. et al . , 2001 , mnras , 328 , 1039 condon j.j . , 1984 ,
apj , 287 , 461 condon j.j . , cotton w.d . ,
greisen e.w .
, yin q.f . , perley r.a . ,
taylor g.b .
, broderick j.j . 1998 , aj , 115 , 1693 cornwell t.j . , 1987 , a&a ,
180 , 269 de breuck c. , 2005 , _ scientific requirements of alma , and its capabilities for key - projects : extragalactic _ , in `` the dusty and molecular universe . a prelude to herschel and alma '' , esa publications , p. 27
( astro - ph/0410746 ) de zotti g. , ricci r. , mesa d. , silva l. , mazzotta p. , toffolatti l. , gonzlez - nuevo j. , 2005 , a&a , 431 , 893 giommi p. , colafrancesco s. , cavazzuti e. , perri m. , pittori c. , 2006 , a&a , 445 , 855 gregory p.c .
, scott w.k . , douglas k. , condon j j. , 1996 , apjs , 103 , 427 griffith m.r . , wright a.e . , 1993 ,
aj , 105 , 1666 hambly n.c .
et al . , 2001 ,
mnras , 326 , 1279 hinshaw g. et al . , 2007 , apjs , 170 , 288 holdaway m.a . , owen f.n . , rupen m.p . ,
1994 , millimeter array memo # 123 , _ source counts at 90ghz _ , http://www.alma.nrao.edu/memos holdaway m.a .
, carilli c. , laing r. 2004 . alma memo # 493 , _ finding fast switching calibrators for alma _ , http://www.alma.nrao.edu/memos hovatta t. , tornikoski m. , lainela m. , lehto h.j .
, valtaoja e. , torniainen i. , aller m.f .
, aller h.d . , 2007 ,
a&a , 469 , 899 jones d.h .
et al . , 2004 ,
mnras , 355 , 747 kellermann k.i . , 1964 , apj , 140 , 969 kellermann k.i .
, pauliny - toth i.i.k . ,
davis m.m .
, 1968 , astrophysical letters , 2 , 105 kesteven m. , 1998 , atnf technical memo at/39.3/081 , _ atca upgrade pointing considerations _ , http://www.atnf.csiro.au/projects/mnrf1996/technical_memos lhteenmki a. , valtaoja e. , 1999 , apj , 521 , 493 lamarre j.m .
et al . , 2003 , newar , 47 , 1017 landau r. , epstein e.e . , rather j.d.g . , 1980 ,
aj , 85 , 363 lpez - caniego m. , herranz d. , gonzlez - nuevo j. , sanz j.l . , barreiro r.b . , vielva p. , argeso f. , toffolatti l. , 2006 , mnras , 370 , 2047 lowe s.r . , gawronski m.p . ,
wilkinson p.n . ,
kus a.j . ,
browne i.w.a . ,
padzerski e. , feiler r. , kettle d. , 2007 , a&a , 474 , 1093 massardi m. et al . , 2008 ,
mnras , 384 , 775 middelberg e. , sault r.j .
, kesteven m.j . , 2006 , pasa , 23 , 147 odea c.p . , 1998 ,
pasp , 747 , 493 owen f.n .
, mufson s.l . , 1977 ,
aj , 82 , 776 owen f.n . , porcas r.w . , mufson s.l . , moffett t.j . , 1978 ,
aj , 83 , 685 owen f.n . , helfand d.j .
, spangler s.r .
, 1981 , apj , 250 , l55 patnaik a.r .
, browne i.w.a . ,
wilkinson p.n . ,
wrobel j.m . , 1992 , mnras , 254 , 655 rengelink r.b . , tang y. , de bruyn a.g . ,
miley g.k . ,
bremer m.n .
, rttgering h.j.a . ,
bremer m.a.r .
1997 , a&as , 124 , 259 ricci r. et al .
2004 , mnras , 354 , 305 sault r.j .
, teuben p.j .
, wright m.c h. , 1995 , asp conference series , 77 , 433 sadler e.m .
et al . , 2006 ,
mnras , 371 , 898 scott d. , white m. , 1999 , a&a , 346 , 1 steppe h , salter c.j . , chini r. , kreysa e. , brunswig w. , lobato perez j. , 1988 , a&as , 75 , 317 tersranta h. et al .
, 2004 , a&a 427 , 769 thompson a r. , moran j.m . , swenson g.w . , 2001 , in _ interferometry and synthesis in radio astronomy _ , 2nd edition , new york , wiley tinti s. , de zott , g. , 2006 , a&a , 445 , 889 ) toffolatti l. , argueso gomez f. , de zotti g. , mazzei p. , franceschini a. , danese l. , burigana c. , 1998 , mnras , 297 , 117 snellen i.a.g . ,
schilizzi , r.t . ,
miley , g.k . , de bruyn , a.g . , bremer , m.n .
, rttgering h.j.a . , 2000 , mnras , 319 , 445 voss h. , bertoldi f. , carilli c. , owen f. n. , lutz d. , holdaway m. , 2006 , a&a , 448 , 823 waldram e.m .
, pooley g.g .
, grainge k.j.b .
, jones m.e . , saunders r.d.e . ,
scott p.f . ,
taylor , a.c . , 2003 , mnras , 342 , 915 waldram e.m . , bolton r.c . ,
pooley g.g .
, riley j.m . , 2007 , mnras , 379 , 1442
the 6dfgs spectrum ( jones et al . 2004 ) shows a bright featureless continuum and no redshift could be measured . * 0111@xmath154749 : * the optical i d is a 2dfgrs ( colless et al .
2001 ) galaxy at @xmath16=0.154 whose spectrum shows strong , narrow emission lines .
this is one of the few at20 g sources in our sample which was undetected at 95ghz .
the radio spectrum rises steeply from 0.8 to 20ghz , and so must peak somewhere above 20ghz .
the 843mhz sumss images shows extended emission with a deconvolved largest angular size of 83arcsec at pa 36@xmath17 .
* 0136@xmath154044 : * this appears to be the same object as pmnj0136@xmath154044 , though the catalogued pmn position is significantly offset from the more accurate at20 g radio position .
there is a faint ( stellar ) optical i d at the at20 g position . *
0433@xmath154502 : * one of the few sources in our sample which shows an inverted radio spectrum at all frequencies from 0.8 to 95ghz .
the dss optical i d is a 19th magnitude galaxy whose redshift is currently unknown . * 0440@xmath154731 : * this source is well below the limit of the sumss catalogue at 843mhz , but examination of the original sumss image shows a marginal ( 3@xmath18 ) detection of a 2.5mjy source at the at20 g position .
there is no optical i d , and this is one of the few at20 g sources which does not appear in any previous optical or radio catalogue . * 0523@xmath152955 : * this source is extended ( or perhaps confused ) in the sumss image ( deconvolved major axis 47arcsec at pa141@xmath17 ) .
the nvss image also shows extended emission with largest angular size of 56arcsec . *
0608@xmath153041 : * this source is clearly extended north south in the sumss image ( deconvolved major axis 92arcsec at pa1@xmath17 ) , and an extended structure is also clearly seen in the at20 g 5ghz image even though the image quality is poor in this snapshot observation .
the 1.4ghz nvss resolves the source into two components separated by about 70arcse . at 20ghz
the source is unresolved .
the 6dfgs spectrum shows strong , broad balmer emission lines typical of a seyfert 1 galaxy or qso . at z=0.2237 ,
this is a relatively nearby object .
* 1427@xmath153305 : * the 6dfgs spectrum shows a featureless continuum , and no redshift could be measured .
the optical counterpart is stellar on the blue dss image , but is fuzzy ( and classified as a galaxy ) on the red image , suggesting that this may be a relatively nearby bl lac object . | we have used the australia telescope compact array ( atca ) at 95ghz to carry out continuum observations of 130 extragalactic radio sources selected from the australia telescope 20ghz ( at20 g ) survey .
we use a triple correlation method to measure simultaneous 20 and 95ghz flux densities for these objects , and over 90% of our target sources are detected at 95ghz .
we demonstrate that the atca can robustly measure 95ghz flux densities with an accuracy of @xmath010% in a few minutes for sources stronger than about 50mjy .
we measure the distribution of radio spectral indices in a flux
limited sample of extragalactic sources , and show that the median 2095ghz spectral index does not vary significantly with flux density for s@xmath1mjy .
this finding allows us to estimate the extragalactic radio source counts at 95ghz by combining our observed 2095ghz spectral index distribution with the accurate 20ghz source counts measured in the at20 g survey .
our derived 95ghz source counts at flux densities above 80mjy are significantly lower than those found by several previous studies .
the main reason is that most radio sources with flat or rising spectra in the frequency range 520ghz show a spectral turnover between 20 and 95ghz . as a result ,
there are fewer 95ghz sources ( by almost a factor of two at 0.1jy ) than would be predicted on the basis of extrapolation from the source populations seen in lower frequency surveys .
we also derive the predicted confusion noise in cmb surveys at 95ghz and find a value 2030% lower than previous estimates .
the 95ghz source population at the flux levels probed by this study is dominated by qsos with a median redshift @xmath2 .
we find a correlation between optical magnitude and 95ghz flux density which suggests that many of the brightest 95ghz sources may be relativistically beamed , with both the optical and millimetre continuum significantly brightened by doppler boosting .
= = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] radio continuum : galaxies galaxies : active quasars : general radio continuum : general cosmic microwave background techniques : interferometric |
until recently the @xmath0mno@xmath1 perovskites ( @xmath0=rare earth , y , sc ) have been object of research mainly as parent materials of mixed valence manganites exhibiting colossal magnetoresistivity ( cmr).@xcite in the past few years , however , there is an increased interest in the complex relationships among the lattice distortions , magnetism , dielectric and transport properties of undoped @xmath0mno@xmath1 .
@xcite in particular , it has been found that with decreasing radius ( @xmath2 ) of @xmath0 ( @xmath0=la to eu ) the transition temperature t@xmath10 to a - type antiferromagnetic ( a - afm ) structure also decreases . with further decrease of @xmath2 ( @xmath0=gd to ho ) the magnetic structure below t@xmath10
changes from a - afm to incommensurate antiferromagnetic one(ic - afm ) with sine - wave ordering of the mn moments with temperature - dependent wave vector @xmath11 along the @xmath12-axis ( in @xmath13 notations ) . at t@xmath14@xmath15t@xmath10 an
incommensurate - to - commensurate ( ic - cm ) transition takes place and @xmath16 locks at a value @xmath17.@xcite a transition to commensurate e - type antiferromagnetic structure ( e - afm , @xmath18 ) has been observed below t@xmath19 only for homno@xmath1 .
@xcite for tbmno@xmath1 and dymno@xmath1 it was found that the ic - cm transition is accompanied by a ferroelectric transition , associated with lattice modulation in the cm phase.@xcite .
large magnetodielectric effects have also been reported for orthorhombic homno@xmath1 , and ymno@xmath1.@xcite although the role of lattice distortions in the interplay of magnetic , dielectric and transport properties of @xmath0mno@xmath1 perovskites is widely recognized , there are relatively few studies on the variations of these distortions with @xmath0.@xcite raman spectroscopy can provide significant additional information on the distortions and their variations with both @xmath0 and temperature .
the raman active modes in @xmath0mno@xmath1 ( @xmath20 ) are activated exclusively due to the deviations from ideal perovskite structure and in principle the activation of each raman line can with definite certainty be assigned to the value of one or two types of basic distortion ( rotations of mno@xmath9 octahedra around [ 101 ] or [ 010 ] directions , jahn - teller distortion , or shift of @xmath0 atoms).@xcite the first assignment of the raman lines in the spectra of @xmath0mno@xmath1 to definite atomic vibrations has been done by comparison of experimentally obtained raman phonon frequencies of lamno@xmath1 and ymno@xmath1 with those predicted by lattice dynamical calculations ( ldc ) .
@xcite although for most of the experimentally observed raman lines the frequencies have been in good agreement with those predicted by ldc , some uncertainties about the phonon line assignment remained .
indeed , some raman lines of same symmetry have close frequencies and due to the approximations of the ldc model one reasonably expects some difference between measured and calculated values .
the assignment of the two @xmath21 raman lines between 450 and 520 @xmath22 , where are the @xmath21 modes corresponding to anti - stretching(as ) and bendings(b ) of mno@xmath9 octahedra , has been challenged by martin - carron et al.@xcite in a raman study of several @xmath0mno@xmath1 perovskites ( @xmath0=la , pr , nd , tb , ho , er , y ) . based on the fact that the mn - o distances exhibit only small changes within the @xmath0mno@xmath1 series , they assigned the as band observed near 480 - 490 @xmath22 to the anti - stretching mode , while the b band which shifts from @xmath23 @xmath22 in lamno@xmath1 to @xmath24 @xmath22 in ermno@xmath1 has been assigned to bending mode(s ) .
an obvious problem of this re - assignment is the `` crossing '' of the as and b modes , which are of the same a@xmath25 symmetry . in this work we present results of a detailed study of polarized raman spectra of @xmath0mno@xmath1 ( @xmath0=la , pr , nd , sm , eu , gd , tb , dy , ho , y ) at room temperature .
the comparison of variations with @xmath0 of phonon frequencies and relative intensities of most pronounced @xmath21 and @xmath26 modes shows that some corrections of previous phonon mode assignment are needed .
the strong mixing and spectral weight transfer between @xmath21 anti - stretching ( as ) and bending ( b ) modes in the 500 @xmath22 region , most clearly observed for @xmath0=eu , gd and tb , correlates with changes of magnetic structure .
similar distortion - dependent mode mixing is observed for the rotational @xmath5 and o1(@xmath6 ) [ @xmath7 modes .
prmno@xmath1 , ndmno@xmath1 and smmno@xmath1 samples were grown by the floating zone method described in ref.@xcite .
single crystals of dymno@xmath1 of average sizes @xmath27 mm@xmath28 were prepared following a procedure to be described elsewhere @xcite .
orthorhombic ymno@xmath1 , homno@xmath1 samples were prepared under pressure as described in ref.@xcite . for @xmath0=eu ,
gd , tb the precursor powders were heated at 1130 - 1160@xmath29 in o@xmath30 for 24 hours .
although the samples ( except for dymno@xmath1 ) were polycrystalline , the size of the constituting microcrystals was larger than the laser spot .
it was possible to select microcrystals with crystallographic orientation such that the spectra in parallel and crossed scattering configurations were dominated by @xmath21 or @xmath26 contributions , respectively . for exact separation of the @xmath21 and @xmath26
spectra the spectra subtract program of the grams ai software package was used .
the raman spectra were measured at room temperature in backscattering configuration using two different spectrometers : labram-800 and jobin yvon hr640 , both equipped with microscope and liquid - nitrogen - cooled ccd detector .
the spectra obtained with 514.5 nm ( ar@xmath31 ) and 632.8 nm ( he - ne ) were practically identical .
the @xmath21 and @xmath26 spectra of @xmath0mno@xmath1 as obtained at room temperature are shown in figure 1 . with same scattering configuration
the absolute value of raman intensities were increasing with decreasing r ionic radius ( @xmath2 ) . for better comparison
the @xmath21 and @xmath26 spectra were normalized to the integrated intensities of @xmath32 and @xmath33 lines , respectively .
at least five @xmath21 and four @xmath26 lines are clearly pronounced and the variations of their parameters with @xmath0 can be followed with a good accuracy .
the phonon line frequencies are given in table i. and @xmath26 raman spectra of @xmath0mno@xmath1 at room temperature .
the phonon mode notations correspond to those of refs.@xcite . the @xmath21 and @xmath26 spectra are normalized to the integrated intensity of @xmath32 and @xmath33 lines , respectively.,width=264 ] ccccccccccccc mode & main atomic motions & la&pr & nd & sm & eu & gd & tb & dy & ho & y&basic distortion + @xmath34&o2 anti - stretchng @xmath35 mno@xmath9 bending&490.7 & 491.2 & 491.7 & 496.1 & 501.0 & 506.0 & 509.0 & 513.4 & 520.0 & 520.6&jt,[010]@xmath35[101 ] + @xmath36&mno@xmath9 bending @xmath35 o2 anti - stretching&444.8 & 461.8 & 465.6 & 476.7 & 479.0 & 486.4 & 488.8 & 492.1 & 498.9 & 496.7&[101]@xmath35jt,[010 ] + @xmath8&out - of - phase mno@xmath9 @xmath6-rotations & 283.6 & 323.7 & 331.8 & 352.9 & 360.8 & 370.5 & 378.2 & 386.0 & 395.9 & 398.1 & [ 101 ] + @xmath37&in - phase mno@xmath9
@xmath38-rotations@xmath35 o1(@xmath6 ) & 256.8 & & & & & 309.8 & 314.6 & 319.5 & 323.8 & 325.3&[010]@xmath39-shift + @xmath40&o1(@xmath6 ) @xmath35in - phase mno@xmath9 @xmath38-rotations & 198.1 & 231.8 & 245.0 & 266.1 & & 276.1 & 280.5 & & 287.8 & 289.4 & @xmath0-shift @xmath35[010 ] + + @xmath33&in - plane o2 stretching & 612.3 & 606.7 & 606.2 & 606.6 & 609.5 & 611.7 & 612.2 & 613.5 & 615.0 & 617.3 & jt + @xmath41&in - phase o2 " scissors-like & 482.5 & 495.7 & 502.6 & 513.1 & 518.4 & 525.0 & 528.4 & 533.6 & 537.0 & 539.2&[010 ] + @xmath42&out - of - phase mno@xmath9 bending & 430.0 & 444.8 & 449.1 & 460.9 & 464.6 & 468.9 & 473.7 & 477.5 & 481.1 & 481.5&[101 ] + @xmath43&o1(@xmath44 ) & 309.7 & 311.5 & 313.2 & & 323.8 & 329.2 & 330.9 & 335.6 & & 342.0 & @xmath0-shift ionic radius @xmath2.,width=264 ] the correlation between structural and physical parameters of @xmath0mno@xmath1 perovskites may be described as a function of different parameters , such as the ionic radius @xmath2 , tolerance factor @xmath45 or the averaged tilt angle @xmath46 .
the relation between these parameters , however , is close to linear@xcite and any one of them can be used for characterization of phonon mode frequency variations .
figure 2 shows the dependence of raman phonon frequencies on @xmath2 .
most raman frequencies @xmath47 exhibit nearly linear increase with @xmath2 , but the slope @xmath48 depends mainly on the type of the mode ( stretching , bending or rotational ) as well on the type of distortion that has activated the corresponding mode in the raman spectrum ( mno@xmath9 rotations around @xmath49_c$ ] or @xmath50_c$ ] cubic axes , janh - teller(jt ) distortion , and @xmath0-shift).@xcite in the ideal cubic abo@xmath1 perovskite ( space group @xmath51 , @xmath52 ) all atoms are at centrosymmetrical sites with fixed coordinates and there are no raman allowed modes .
the structure of stoichiometric @xmath0mno@xmath1 , described at room temperature by the @xmath13 space group ( @xmath53 ) , can be considered as orthorhombically distorted superstructure of the ideal perovskite with @xmath54_o$ ] , @xmath49_o$ ] , and @xmath55_o$ ] directions coinciding with the @xmath50_c$ ] , @xmath49_c$ ] and @xmath56_c$ ] directions of the parent cubic structure . in the @xmath13 structure
the atoms occupy four non - equivalent atomic sites ( @xmath0,mn , o1,o2 ) , of them only the mn site is a center of symmetry . from symmetry considerations only 5 of the 12 atomic coordinates
are fixed [ @xmath0 - 1,mn-3,o1 - 1 ] , while the remaining 7 variable coordinates can be considered as the lattice degrees of freedom producing the above mentioned lattice distortion .
these distortions give rise to 24 ( @xmath20 ) raman - allowed phonon modes , which in turn can be assigned to their distortions of origin ( [ 010]- , [ 101]-rotation , jt , or @xmath0-shift ) . with decreasing @xmath2 the distortions increase and most of the bond lengths shorten , which reasonably results in twofold effect : enhanced raman intensity and hardening of raman frequencies . at first sight
the [ 010 ] and [ 101 ] rotations can directly be obtained from the o2-mn - o2 and o1-mn - o1 angles , which practically have equal values.@xcite as shown by abrashev et al.@xcite , however , even in the case of equal o - mn - o angles , the [ 010 ] and [ 101 ] rotation angles will be different .
they are related to the deviations ( @xmath57 , @xmath58-atomic index ) of @xmath13 atomic coordinates from those in a hypothetical undistorted perovskite structure as @xmath59 } = \arctan(2|x_{o2}-z_{o2}|)\ ] ] @xmath60 } = \arctan(2^{5/2}|y_{o2}| ) \approx \arctan(2^{3/2}|z_{o1}|)\ ] ] the so defined atomic deviations can also be used to describe the other two basic distortions , namely , the jt distortion , characterized by the relative difference @xmath61 of the two pairs of mn - o2 distances , and the shift @xmath62 of the @xmath0-atoms in @xmath6-direction from their positions in an undistorted perovskite . in the case of small distortions : @xmath63 @xmath64 the variations of the basic distortions , calculated using the available data for atomic positions in @xmath0mno@xmath1@xcite are given in table ii . [ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] the frequency of a mode involving mainly stretching vibrations of o2 atoms in the @xmath65 planes is determined by the mn - o2 distances .
the short and long mn - o2 bond lengths vary with @xmath2 within 0.5% and 1.1% , respectively@xcite and therefore one expects that @xmath66 will change by no more than 1.5% . as illustrated in figure 3 , this is exactly the case for the @xmath33 mode at 606 - 617 @xmath22 , which has been assigned to the in - phase o2 stretching.@xcite .
the minimum of @xmath33 frequency near r = nd , sm reflects the fact that for these @xmath0 s the averaged mn - o distance has a maximum .
based on the same bond - length considerations , one could expect that the raman line of @xmath21 symmetry at 490 - 495 @xmath22 , corresponding to the in - phase anti - stretching vibrations of o2 in the @xmath65 plane,@xcite would behave in a similar way and its frequency would remain nearly constant through the whole @xmath0mno@xmath1 series . as seen from fig.4 , this is obviously not the case .
the data of figs . 1 and 4 clearly show a classic example for mixing of modes of same symmetries and close frequencies . for large @xmath2 ( @xmath0=la , pr , nd ) the lines at 490 - 492 and 445 - 465 @xmath22 can be considered as pure stretching(@xmath3 ) and bending(@xmath4 ) modes , respectively . with decreasing @xmath2 the @xmath4 frequency increases and approaches that of @xmath3 , resulting in strong mode mixing evidenced by mode repulsion and transfer of intensity .
the mixing is most strongly pronounced for @xmath0=sm to tb where the two modes are of comparable intensities and involve both stretching and bending atomic motions . with further decrease of @xmath2 ( @xmath0=dy ,
ho , y ) the modes become less mixed but now the higher mode is dominated by mno@xmath9 bending [ @xmath4-type ] motions and the lower one is @xmath3-type .
two phonon modes of same symmetries and close frequencies can be considered as coupled quantum oscillators . in a good approximation their frequencies
are given by @xmath67 where @xmath68 and @xmath69 are the mode frequencies without coupling and @xmath70 is the coupling constant .
it is reasonable to assume that without coupling the @xmath3 and @xmath4 mode frequencies would depend on @xmath2 in the same way as the @xmath33 and @xmath71 modes , respectively , as both modes of the @xmath3-@xmath33 ( stretching ) and @xmath4-@xmath42 ( bending ) pairs are activated by the same basic distortion and have similar shape .
therefore , as shown with dash lines in fig.4 , the @xmath72 dependence for the uncoupled @xmath3 mode can be approximated by scaling the experimental @xmath26 values , whereas that of uncoupled @xmath4 mode [ @xmath73 , like in the case of @xmath42 , can be approximated by a linear dependence . at the crossing point
@xmath74 ) , @xmath75 , which gives @xmath76 @xmath22 . of the frequencies of @xmath33 , @xmath4 and @xmath3 modes .
the dashed curves show the expected @xmath77 and @xmath78 dependences for pure stretching and bending modes.,width=264 ] the frequency vs @xmath79}$ ] dependence for the @xmath8 mode involving mainly mno@xmath9 @xmath6-rotations , shown in fig.4 with full circles , is close to proportionality .
expectedly , this mode exhibits `` soft mode '' behavior with @xmath80 for @xmath81 , the slope of @xmath82})$ ] being 23.5 @xmath22/deg . calculations of lattice dynamics for the two end compounds lamno@xmath1 and ymno@xmath1 have shown that the two weak @xmath21 modes at lower frequencies are strongly mixed and involve both in - phase mno@xmath9 @xmath38-rotations [ @xmath5 ] and o1(x)-motions [ @xmath7 .
the higher of these modes is @xmath84-like in lamno@xmath1 , but @xmath5-like in ymno@xmath1 . assuming that the higher mode remains @xmath5-like in other compounds with small @xmath2 ( homno@xmath1 , dymno@xmath1 ,
tbmno@xmath1 , gdmno@xmath1 ) , one finds that the frequency vs @xmath85}$ ] dependence for the partly mixed @xmath5 mode ( shown with empty squares ) is practically the same as that for the pure @xmath8 mode .
interestingly , the frequencies of [ 111]- rotational modes of structurally different rhombohedral lamno@xmath1 and laalo@xmath1 fit to these dependences , which is an indication for a quite general relationship between the angle of rotation and frequency of the rotational `` soft '' modes in perovskitelike compounds , independent on the direction of the rotational axis . }
$ ] or @xmath85}$ ] of the frequencies of @xmath8 and @xmath5 modes , involving mainly mno@xmath9 rotations.,width=264 ] the weak @xmath84 and @xmath43 involve mainly o1 motions in the @xmath65 plane , activated in the raman spectrum due to the shift of @xmath86 ions from their positions in an ideal perovskite .
it follows from fig.2 , table i and table ii that in the case of pure @xmath0-shift mode [ @xmath43 ] , despite of strongly increasing @xmath0-shift distortion between @xmath0=la and y , the corresponding increase of mode frequency is relatively modest ( see table ii ) .
it is plausible to expect similar dependence on @xmath2 for the @xmath84 mode , which explains the above discussed observation that the frequency vs @xmath2 dependence of the mixed @xmath88 modes is governed mainly by their rotational [ @xmath5 ] component .
the comparative study of polarized raman spectra of orthorhombic @xmath0mno@xmath1 series ( @xmath0=la , pr , nd , sm , eu , gd , tb , dy , ho , y ) shows that the variations of lattice distortions with @xmath2 affect significantly both the phonon frequencies and the shape of some raman phonon modes .
the strong mixing of phonon modes involving in - plane and out - of - plane oxygen motions may become a clue for understanding the change with @xmath2 of magnetic ordering at low temperatures .
the established proportionality between the frequency of a rotational mode and the angle of corresponding rotational distortion may be used for material characterization .
this work is supported in part by the state of texas through the texas center for superconductivity and the national science and engineering research council of canada .
the work of m.m.g . is supported by the bulgarian national research fund ( project f-1207 ) .
kusters , j. singleton , d.a .
keen , r. mcgreevy , and w. hayes , physica b * 155 * , 362 ( 1989 ) .
r. von helmolt , j. wecker , b. holzapfel , l. schultz , and k. samwer , phys .
lett . * 71 * , 2331 ( 1993 ) .
s. jin , t. h. tiefel , m. mccormack , r. a. fastnacht , r. ramesh , and l. h. chen , science * 64*,413 ( 1994 ) .
y. tokura , a. urushibara , y. moritomo , t. arima , a. asamitsu , g. kido , and n. furukawa , science * 63 * , 3931 ( 1994 ) .
a. munoz , m. t. casis , j. a. alonso , m. j. martinez - lope , j. l. martinez , and m. t. fernndez - daz , inorg .
chem . * 40 * , 1020 ( 2001 ) .
a. munoz , j. a. alonso , m. t. casais , m. j. martinez - lope , j. l. martnez , and m. t. fernndez - daz , j. phys .
: condens .
matter * 14 * , 3285 ( 2002 ) .
t. kimura , t. goto , h. shintani , k. ishizaka , t. arima , and y. tokura , nature * 426 * , 55 ( 2003 ) .
t. kimura , s. ishihara , h. shintani , t. arima , k. t. takahashi , k. ishizaka , and y. tokura , phys .
b * 68 * , 060403(r ) ( 2003 ) .
t. goto , t. kimura , g. lawes , a. p. ramirez , and y. tokura , phys .
lett . * 92 * , 257201 ( 2004 ) .
j. a. alonso , m. j. martnez - lope , m. t. casais , and m. t. fernndez - daz , inorg .
chem * 39 * , 917 ( 2000 ) .
m. v. abrashev , j. bckstrm , l. brjesson , v. n. popov , r. a. chakalov , n. kolev , r .-
meng , and m. n. iliev , phys .
b * 65 * , 184301 ( 2002 ) . | the polarized raman spectra of orthorhombic @xmath0mno@xmath1 series ( @xmath0=la , pr , nd , sm , eu , gd ,
tb , dy , ho , y ) were studied at room temperature .
the variation of phonon frequencies with @xmath0 ionic radius @xmath2 as a whole confirms the commonly accepted raman line assignments with two noticeable exceptions : ( 1 ) with decreasing @xmath2 the stretching @xmath3 and bending @xmath4 modes strongly mix for @xmath0=sm to tb , while for further decrease or @xmath2 ( @xmath0=dy , ho , y ) the @xmath4 mode is observed at higher frequency than @xmath3 mode ; ( 2 ) similar distortion - dependent mode mixing takes place for the rotational @xmath5 and o1(@xmath6 ) [ @xmath7 modes .
the mode mixing is particularly strong for the @xmath0mno@xmath1 compounds with @xmath2 values close to the transition from a - type to incommensurate sinusoidal antiferromagnetic ordering at low temperatures .
the frequency of rotational @xmath5 and @xmath8 modes scales to the angles of mno@xmath9 [ 101 ] and [ 010 ] rotations , respectively , and could be used as a measure of their value . |
the hubbard model is the simplest system which presumably incorporates the key features of the strong correlation limit .
understanding this model will be crucial for making any progress with cuprate superconductors , colossal magneto resistance systems or heavy fermions .
the special problem in this model is that near half - filling it represents a dense system of strongly interacting fermions , a situation in which a perturbation expansion in @xmath0 may not be expected to give any meaningful results . in dealing with this model , one
can then pursue two opposing strategies : one might expect that despite the strong interaction a perturbation expansion in @xmath0 remains a meaningful approximation , and apply conventional many - body theory .
the latter means that one is treating the kinetic energy exactly , and results in the validity of the luttinger theorem .
on the other hand , by adiabatic continuity this approach will never produce an insulator at half - filling so that we can be sure of its breakdown in the limit of large @xmath4 .
+ the opposite point of view was taken by hubbard@xcite : in his approximations , the interaction part of @xmath5 is treated exactly and approximations are made to the kinetic energy .
this results in the breakdown of the luttinger theorem , because the physical electron is effectively split into two particles , one of them corresponding to an electron moving between empty sites , the other to an electron moving between sites occupied by an electron of opposite spin .
the energies of formation of these effective particles differ by @xmath0 , whence for large @xmath0 the single free - electron band splits up into two bands formed predominantly by the two types of effective particles .
hubbard s approximations , and related schemes like the so - called @xmath6-pole approximations@xcite , have been dismissed by some authors as unphysical , because they do in fact violate the luttinger theorem away from half - filling . however , in a recent qmc study for the paramagnetic phase of the hubbard model@xcite we have shown that such criticism is entirely unwarranted : the fermi surface , if measured in an ` operational way ' from the fermi energy crossings of the quasiparticle band , indeed does violate the luttinger theorem .
the doping dependence of the fermi surface volume is qualitatively consistent with hubbard s results and the main discrepancy being the fact that one can rather clearly distinguish @xmath7 ` bands ' in the spectral functions rather than the @xmath6 bands predicted by the hubbard - i approximation .
more generally , exact diagonalization studies@xcite tend to produce for example a photoemission spectrum consisting of a relatively narrow ` quasiparticle band ' ( which forms the first ionization states ) and an ` incoherent continuum '
. a simple two - band form of the spectrum , as produced by the hubbard - i approximation , therefore can not give a quantitative description of the spectrum .
motivated by these numerical results we have re - examined the hubbard i approximation and attempted to find the most important corrections to this scheme for the limit of large @xmath0 and weak spin correlations .
we will see that the @xmath7-band structure observed in the qmc simulations can be reproduced quite well by adding two new effective particles which correspond to a composite object of a ` hubbard quasiparticle ' and a spin- , charge- , or @xmath1-excitation .
these composite objects actually are the quite obvious leading correction over the hubbard - i approximation , and we will see that including them into the equations of motion leads to an almost quantitative agreement with the numerical results for a variety of different systems .
we consider the hubbard hamiltonian in particle - hole symmetric form , @xmath8 with : @xmath9 here @xmath10 denotes summation over all pairs of nearest neighbors and @xmath11 . + for
bipartite lattices one can make use of two symmetry transformations .
the first one is the particle - hole transformation : @xmath12 , where @xmath13 .
if the kinetic energy contains only nearest neighbor hopping , this transformation leaves the hamiltonian invariant and exchanges the electron addition spectrum for momentum @xmath14 and removal spectrum for momentum @xmath15 at half - filling .
similarly the transformation @xmath16 , @xmath17 inverts the sign of @xmath18@xcite . at half - filling this allows to transform solutions of the positive - u model into those of the negative - u model .
this transformation implies that the single - particle spectral function at half - filling is identical for positive and negative @xmath0 . in the appendix some of the operators introduced in this work and
their transformation properties under these transformations are listed .
+ we proceed with the calculation of the single - particle green s function . as a first step ,
following hubbard@xcite , we split the electron annihilation operator into the two eigenoperators of the interaction part : @xmath19 these obey @xmath20 = \frac{u}{2 } \hat{d}_{i,\sigma}$ ] and @xmath21 = - \frac{u}{2 } \hat{c}_{i,\sigma}$ ] .
next , we consider the commutators of these ` effective particles ' with the kinetic energy . after some algebra , thereby using the identity @xmath22@xmath23@xmath24 , we find : @xmath25 & = & -t \sum_{j\in n(i)}\ ; [ ( 1-\frac{\langle n\rangle}{2 } ) c_{j,\uparrow } + ( c_{j,\uparrow } s_i^z + c_{j,\downarrow } s_i^- ) \nonumber \\ & & \;\;\;\;\;\ ; - \frac{1}{2 } c_{j,\uparrow } ( n_i - \langle n\rangle ) + c_{j,\downarrow}^\dagger c_{i,\downarrow } c_{i,\uparrow } ] , \nonumber \\ \ ; [ \hat{d}_{i,\uparrow } , h_t ] & = & -t \sum_{j\in n(i)}\ ; [ \frac{\langle n\rangle}{2 } c_{j,\uparrow } - ( c_{j,\uparrow } s_i^z + c_{j,\downarrow } s_i^- ) \nonumber \\ & & \;\;\;\;\;\ ; + \frac{1}{2 } c_{j,\uparrow } ( n_i - \langle n\rangle ) - c_{j,\downarrow}^\dagger c_{i,\downarrow } c_{i,\uparrow } ]
. \label{commu}\end{aligned}\ ] ] here @xmath26 denotes the @xmath27 nearest neighbors of site @xmath28 . keeping only the first term in each of the square brackets on the r.h.s .
( as we will do for the remainder of this section ) reproduces the hubbard i approximation .
we specialize to half - filling ( @xmath29 ) and introduce the green s functions @xmath30 where @xmath31 . then , using the anticommutator relations @xmath32 , @xmath33 , @xmath34 we obtain the following equations of motion : @xmath35 taking into account that the ordinary electron green s function @xmath36 and the green s function called @xmath37 by hubbard@xcite can be written as @xmath38 the resulting equations of motions are precisely those derived in the hubbard - i approximation : @xmath39 the present formulation , on the other hand , allows for an appealing physical interpretation of the hubbard - i approximation : we introduce free fermion operators @xmath40 and @xmath41 , which correspond to ` holes ' and ` double occupancies ' .
the hubbard operators are identified with these as follows : @xmath42 then , we can formally obtain the set of equations of motion ( [ simp ] ) from the following hamiltonian for the holes and double occupancies : @xmath43 the hamiltonian ( [ hubre ] ) is a quadratic form and readily solved by bogoliubov transformation : @xmath44 to yield the familiar dispersion relation @xmath45 . \label{hubdisp}\ ] ] to compute the spectral weight of the two bands we use @xmath46 , whence @xmath47 again , this is the correct hubbard - i result .
the above discussion shows the physical content of the hubbard - i approximation : the hamiltonian ( [ hubre ] ) describes fermionic particle - like and hole - like charge fluctuations , created by @xmath41 and @xmath48 , respectively .
these ` live ' in a background of singly occupied sites .
particle - like and hole - like charge fluctuations are created in pairs on nearest neighbors , and individually can hop between nearest neighbors .
the hopping integral for the hole - like particle has opposite sign as that for the electron - like fluctuation as it has to be , and the hopping integrals for both effective particles are @xmath49 times that for the ordinary electrons : this reflects the fact that due to the pauli principle ( say ) a spin up electron added to a ` background ' of singly occupied sites can propagate to a neighboring site only with a probability of @xmath49 ( we are neglecting any spin correlations of the background of singly occupied sites ) .
the factor of @xmath50 in ( [ egon ] ) is due to the fact that @xmath51 .
finally , the particle which stands for the double occupancy has an energy of formation of @xmath52 , the hole - like particle has an energy of @xmath53 .
as already mentioned , the hubbard i approximation therefore describes the splitting of the physical electron into two new effective particles which carry with them information on the ` environment ' in which the electron has been created .
one of them ( @xmath54 ) moves between sites occupied by an electron of opposite spin , the other one ( @xmath55 ) between empty sites .
this is a quite appealing physical idea , but the above formulation also very clearly highlights the weak points of the hubbard - i approximation .
in addition to the mere truncation of the commutators ( [ commu ] ) , which is an uncontrolled approximation , these are the following : adopting this picture we would have to assume that the states @xmath56 and @xmath57 are distinguishable ( and in fact orthogonal to one another ) .
this , however , is in general not the case : both states have one double occupancy on site @xmath58 and a hole on site @xmath28 , and the only difference is that they have been created in different ways .
in fact , using ( [ egon ] ) we find for their overlap @xmath59 where we have assumed a rotationally invariant ground state in the second line .
the hubbard - like approximation scheme thus should work only for a state with vanishing spin correlations - we will therefore henceforth assume the spin correlation function @xmath60 to be zero .
+ a second problem is , that the effective fermions have to obey a kind of hard - core constraint - a site can not be simultaneously occupied by a hole and a double - occupancy .
this constraint is not accounted for in the derivation of the hubbard - i approximation : the equations of motion are obtained from the hamiltonian ( [ hubre ] ) by treating the @xmath61 and @xmath62 as ordinary free fermion operators .
this problem is the source of the violation of certain sum - rules when the hubbard - i approximation is applied in the doped case , see the discussion given by avella _
et al._@xcite .
+ one last remark is that the commutators ( [ commu ] ) are invariant under the particle - hole transformation , i.e. they are transformed into each others hermitian conjugate .
this remains true for the truncated commutators , which give the hubbard - i approximation , so that the spectral function obtained from this is particle - hole symmetric .
this can also be verified directly using ( [ hubdisp ] ) and ( [ hubweight ] ) .
finally , the spectral function is manifestly invariant under sign change of @xmath0 , as it has to be .
we now want to try and derive an improved version of the hubbard - i approximation .
thereby we will address neither the problem of nonorthogonality of different hole / double - occupancy configurations nor the hard - core constraint - instead , in this work we will restrict ourselves entirely to an approximate way of treating the omitted terms in the commutator relations ( [ commu ] ) .
we expect that the present approximation is reasonable for large @xmath0 ( where the density of holes / double occupancies is small whence the hard - core constraint is of little importance ) and weak spin correlations ( where the nonorthogonality problem is small ) . throughout
we stick to the case of half - filling and no spin polarization , @xmath63 .
we return to the basic commutator relations ( [ commu ] ) and consider the terms on the r.h.s . which are omitted in the hubbard - i scheme .
the second term in the square bracket is the clebsch - gordan contraction of the spin-1/2 operator @xmath64 and the spin-1 operator @xmath65 into yet another spin-1/2 operator - it describes the coupling of the created hole / annihilated double occupancy to spin excitations .
this term may be expected to be the most important one in the limit of large positive @xmath0 .
the third term describes in an analogous way the coupling to density fluctuations , whereas the last term describes the coupling to the so - called @xmath1-pair excitation@xcite .
one may expect that in the case of _ negative _ @xmath0 the last two terms are the important ones . + in keeping with the basic idea of the hubbard approximations , namely to treat the dominant interaction terms exactly , we split also the composite operator into eigenoperators of the @xmath0-term and define : @xmath66 where @xmath67 . under the positive / negative @xmath0 transformation we have for example @xmath68 @xmath69 @xmath70 , i.e. keeping the at first sight unimportant ( for positive @xmath0 ) terms involving density and pairing fluctuations
is crucial for maintaining the exact symmetry under sign change of @xmath0 .
we then have @xmath71 & = & \frac{u}{2 } \hat{d}_{i , j,\sigma } , \nonumber \\ \ ; [ \hat{c}_{i , j,\sigma } , h_u ] & = & - \frac{u}{2 } \hat{c}_{i , j,\sigma } , \nonumber \\ \ ; [ \hat{c}_{i,\uparrow } , h_t ] & = & -t \sum_{j\in n(i)}\ ; [ \ ; \frac{1}{2 } c_{j,\uparrow } + \hat{c}_{i , j,\sigma } + \hat{d}_{i , j,\sigma } \ ; ] , \nonumber \\ \ ; [ \hat{d}_{i,\uparrow } , h_t ] & = & -t \sum_{j\in n(i)}\ ; [ \frac{1}{2 } c_{j,\uparrow } - \hat{c}_{i , j,\sigma } - \hat{d}_{i , j,\sigma } \ ; ] .
\label{bigcom}\end{aligned}\ ] ] the operators @xmath72 and @xmath73 may be thought of describing ` composite objects ' consisting of a charge fluctuation and a spin- , density- or @xmath1-excitation on a nearest neighbor .
ultimately these composite operators carry the quantum numbers of a single electron , i.e. spin 1/2 and charge @xmath74 .
we also note that under particle - hole transformation @xmath75 i.e. the composite particles transform in the same way as the @xmath76 and @xmath77 .
moreover , the commutators ( [ bigcom ] ) are invariant under particle - hole transformation , i.e. this transformation transforms them into each others hermitian conjugates .
+ we now enlarge the set of green s functions by allowing @xmath78 in ( [ green ] ) ; in the language of the ` effective fermions ' this means that we are introducing additional fermions corresponding to the composite objects . to obtain a closed system of equations of motion for these green s functions ,
we need equations for the @xmath79 and @xmath80 . as a first step we turn to the commutators
@xmath81 $ ] and @xmath82 $ ] . here
we have to distinguish three cases ( see figure [ fig1 ] ) : + a ) the hopping term may act along the bond @xmath83 and transport the hole back from @xmath58 to @xmath28 .
+ b ) it may transport the hole even further away from @xmath28 + c ) it may transport the spin- density- or @xmath1-excitation away from site @xmath28 .
+ if we want to restrict ourselves to the @xmath7 types of operators @xmath72 , @xmath84 , @xmath76 and @xmath77 , we have to neglect the contributions from the processes b and c. these processes would produce ` strings ' of excitations along a path of length 2 lattice spacings , and we would have to introduce even more complicated operator products to describe these . restricting ourselves to processes of the the type a , i.e. replacing @xmath85 , straightforward computation@xcite gives the surprisingly simple result @xmath86 & = & \frac{3t}{2 } \hat{d}_{j , i,\uparrow } + \frac{t}{2 } \hat{c}_{j , i,\uparrow } -\frac{3t}{4 } ( \hat{c}_{i,\uparrow } - \hat{d}_{i,\uparrow } ) , \nonumber \\ \ ; [ \hat{d}_{i , j\uparrow } , h_t ] & = & \frac{3t}{2 } \hat{c}_{j , i,\uparrow } + \frac{t}{2 } \hat{d}_{j , i,\uparrow } -\frac{3t}{4 } ( \hat{c}_{i,\uparrow } - \hat{d}_{i,\uparrow } ) .
\label{commu1}\end{aligned}\ ] ] again , these relations are particle - hole invariant , i.e. they are turned into each others hermitian conjugates by particle - hole transformation . in passing we
note that had we reduced the operators @xmath87 and @xmath88 to comprise only the terms involving spin excitations ( as might seem appropriate in the case of large positive @xmath0 ) , the commutators would have been much more complicated and in fact the ` hamilton matrix ' @xmath89 to be defined below would have been non - hermitian . + next ,
we need the anticommutators @xmath90 \nonumber \\ & + & \delta_{j , l } \ ; [ \frac{1}{2 } s_i^z - \frac{1}{4}\tilde{n}_i - \frac{1}{2 } ( \tilde{n}_i s_j^z + \tilde{n}_j s_i^z ) \ ; ] \nonumber \\ & + & \delta_{i , l } \ ; [ \hat{c}_{j,\downarrow } \hat{c}_{i,\downarrow}^\dagger - \hat{d}_{i,\downarrow } \hat{d}_{j,\downarrow}^\dagger \ ; ] \nonumber \\ \ { \hat{d}_{i , j,\uparrow } , \hat{c}_{l,\uparrow}^\dagger \ } & = & \delta_{i , l } \;[\ ; \hat{c}_{j,\downarrow } \hat{d}_{i,\downarrow}^\dagger + \hat{d}_{j,\downarrow}^\dagger \hat{c}_{i,\downarrow}\;]\end{aligned}\ ] ] taking the expectation value in the ground state most of the terms vanish on the basis of symmetries : @xmath91 vanishes due to inversion symmetry of the ground state , @xmath92 and @xmath93 vanish due to particle - hole symmetry at half - filling .
all terms containing unpaired spin operators vanish if we assume that the ground state is invariant under spin rotations ( which excludes ferro- or antiferromagnetic solutions ) .
finally we obtain : @xmath94 it is is easy to see that the expressions whose expectation values are taken are invariant under particle - hole transformation , and under the positive / negative u transformation we have : @xmath95 the expectation value of the anticommutator would be invariant under this positive / negative - u symmetry .
+ for large positive @xmath0 the terms @xmath96 and @xmath97 have a negligible expectation value and the only important term comes from the spin correlation . in keeping with our above remarks concerning the role of spin correlations in the hubbard i approximation
we will henceforth take the r.h.s .
of ( [ anticomm ] ) to be zero .
as was discussed above , the hubbard - i approximation implicitly assumes @xmath98 , and we will therefore keep this value also in ( [ anticomm ] ) .
we will discuss the consequences of not making this approximation later on .
+ using the above commutators and ( expectation values of ) anticommutators we are now in a position to set up a closed system of equations of motion . in the following we give explicit formulas only for a 1d chain with nearest - neighbor hopping , but the generalization to higher dimensions and/or longer range
hopping integrals will be self - evident .
we introduce the fourier transforms @xmath99 ( and analogously for the @xmath100 s ) and define the vector @xmath101 . here
@xmath102 is shorthand for @xmath103 . combining ( [ commu ] ) , ( [ compdef ] ) and ( [ commu1 ] ) , and performing a spatial fourier transformation the equations of motion
are readily found to be @xmath104 where the hermitian matrix @xmath89 is given by @xmath105 2 and @xmath106 .
the equation system ( [ eqnsys ] ) can be solved for each momentum and frequency by using the spectral resolution of the hamilton matrix @xmath89 .
this yields the green s functions @xmath107 and @xmath108 for each momentum and frequency . in an analogous way we can also derive an equation system for @xmath109 and @xmath110 .
thereby the matrix @xmath89 stays unchanged , whereas the r.h.s .
is changed into @xmath111 .
finally , the full electron green s function is obtained by adding up the four green s functions according to ( [ hubbi ] ) . upon forming
the laplace transform @xmath112 we finally obtain the spectral density @xmath113 in passing we note that this way of computing @xmath114 guarantees the validity of the sum rule @xmath115 since the particle - hole symmetry of the relations ( [ bigcom ] ) , ( [ commu1 ] ) and ( [ anticomm ] ) in turn guarantees particle - hole symmetry of the entire spectral function , we find that the sum - rule for the particle number is fulfilled automatically : @xmath116 as a last remark we note that going over to higher dimensions or adding longer ranged hopping integrals poses no problem whatsoever - for each spatial dimension @xmath117 we have to add four additional rows and columns containing the @xmath118 and @xmath119 where @xmath28 and @xmath58 are nearest neighbors in @xmath120-direction .
similarly , if we add an additional hopping integral @xmath2 between @xmath121 or @xmath122 nearest neighbors ( the number of whom we denote by @xmath123 ) to the hamiltonian , we have to add @xmath124 rows and columns , containing the @xmath118 and @xmath119 with @xmath121 or @xmath122 nearest neighbors @xmath28 and @xmath58 . in each case
these additional rows and columns contain only mixing terms amongst themselves or with @xmath125 and @xmath108 , so that the extension is completely trivial .
following the discussion in the preceding section we can calculate the full electron green s function , including the ( presumably ) most important corrections over the hubbard i approximation in the limit of weak spin correlations and large @xmath0 .
we now proceed to a comparison of the obtained results results for the spectral density @xmath114 , with the spectral density obtained from quantum monte - carlo ( qmc ) simulations .
thereby the temperature for the qmc simulation , @xmath126 , was chosen such that the spin correlation length is only approximately @xmath127 lattice spacings - the results thus are probably quite representative for the paramagnetic regime which our approximation aims to describe .
moreover , the value of @xmath128 is already rather large , so that we may also hope to have a small density of holes / double occupancies and the neglect of the hard - core constraint be justified . as a general remark concerning the qmc spectra we note that the maxent procedure used for the analytic continuation to the real axis es most reliable for ` features ' with large weight - this means that the position of tiny peaks is less accurate than that of large ones .
+ figure [ fig2 ] then compares the spectral density obtained from the hubbard i approximation , our extended hubbard approximation ( eha ) and qmc simulation .
the hubbard i approximation gives only a relatively crude fit to the actual spectral density obtained by qmc .
the extended hubbard approximation , on the other hand , gives an all in all quite correct description of the spectral density . out of the @xmath129 bands produced by diagonalizing @xmath89 , only four bands do have an appreciable spectral weight .
these four intense bands correspond rather well to @xmath7 broad ` bands ' of intense spectral weight which can be roughly identified in the qmc result .
it is interesting to note that a recent strong - coupling expansion for the hubbard model by pairault _ et al._@xcite also produced a @xmath7-band structure , although only for the 1d model .
the dispersion of the spectral weight along the bands is also reproduced quite satisfactorily be the extended hubbard approximation .
the main difference is the apparently strongly @xmath14-dependent width of the spectra produced by qmc , which however is outside the scope of the present approximation , which ( in the limit @xmath130 ) produces sharp @xmath131-peaks without any broadening .
a more severe deficiency of the eha is , that it tends to predict too high excitation energies , resulting in a somewhat too large value of the hubbard gap . in any way
, however , the magnitude of the hubbard gap comes out better than in the hubbard - i approximation .
+ we proceed to the hubbard model with an additional hopping integral @xmath2 between @xmath121 nearest ( i.e. ( 1,1)-like ) neighbors .
figure [ fig3 ] again compares the hubbard i approximation , the eha , and the results of a qmc simulation on an @xmath132 lattice .
the agreement between eha and the qmc result is again quite good , the main discrepancy being again an overall overestimation of the binding energies . on the other hand , the apparent @xmath7-band structure ,
the dispersion of the peak energies and the spectral weight agrees well with the numerical result . in particular , the @xmath6 bands in inverse photoemission ( i.e. @xmath133 ) predicted by the eha can be seen very clearly in the qmc spectra ) .
all in all the agreement is even better than for the case @xmath134 , which most probably is due to the fact that the spin correlations are weaker with @xmath2 , so that the assumption of an uncorrelated spin state is better justified in this case .
+ we proceed to the case of a hopping integral integral @xmath3 between @xmath122 nearest ( i.e. ( 2,0)-like ) neighbors . here
we have chosen @xmath135 , because for the larger value @xmath136 the qmc simulation still predicted a metallic state at @xmath128 .
figure [ fig4 ] shows the spectral functions for @xmath135 .
again , we can roughly identify @xmath7 bands and there is reasonable agreement for the dispersion .
the weight of the quasiparticle band in photoemission near @xmath137 ( at @xmath138 ) is not reproduced very well by the eha , but again the ubiquitous @xmath7-band structure is rather clearly visible .
+ next , we turn to a somewhat indirect check of the approximation . the ordinary electron creation operator is the symmetric combination of the hubbard operators .
however , we might also define the antisymmetric combination @xmath139 this operator has the same quantum numbers as the original electron operator , and therefore obeys the same selection rules .
it follows that when acting on the ground state , this operator probes the same final states as the electron operator , the only difference being the matrix element viz . the spectral weight of the respective peak in the spectral density .
in fact the green s function @xmath140 can be expressed as @xmath141 it therefore is easy to calculate within our approximation , and comparison with the qmc result provides an additional check for our description of the electronic structure .
note that the operator @xmath142 enhances the practically dispersionless bands at @xmath143 - since these bands now have a larger weight , their position and dispersion are more reliable than in the ` ordinary ' photoemission spectrum .
this is shown in figure [ fig5 ] , where indeed quite good agreement is found between the eha and the numerical result .
+ finally , we turn to the discussion of choosing a nonvanishing expectation value of the anticommutator in ( [ anticomm ] ) , i.e. we assume that @xmath144 .
this changes the rhs of the equation system ( [ eqnsys ] ) to @xmath145 but leaves the matrix @xmath89 unchanged .
in other words , the _ dispersion _ of the bands , which is determined by the eigenvalues of @xmath89 stays unchanged and only the spectral weight of the peaks changes .
moreover the sum rules ( [ sumrule1 ] ) and ( [ sumrule2 ] ) retain their validity also in this case . for large positive @xmath0 and @xmath146 charge fluctuations will be strongly suppressed and double occupancies will have a small probability , so that the dominant contribution to @xmath147 comes from the spin correlation function @xmath148 , whence we should choose @xmath149 .
assuming for example a negative @xmath147 of moderate value , @xmath150 , then leads to little change in the calculated spectral density ( see figure [ fig6 ] ) : the same two bands which had a large spectral weight for @xmath151 retain a large spectral weight also in this case .
there is , however , a rather undesirable feature associated with the bands with small spectral weight : numerical evaluation shows , that for some regions of @xmath14-space these bands acquire a small but _ negative _ weight .
+ the physical origin of this problem is probably related to nonorthogonalities of basis states : in principle we could interpret the matrix @xmath89 as a hamiltonian describing ( in 2d with only nearest neighbor hopping ) @xmath152 types of fermionic ` effective particles ' .
quite generally , the anticommutator - relation @xmath153 implies that the wave functions corresponding to the fermi operators @xmath154 and @xmath155 are non - orthogonal .
while an exact overlap matrix can never have negative eigenvalues but at most develop zero eigenvalues ( indicating that the set of basis states is overcomplete ) , any approximation to the matrix elements may lead , as an artefact of the approximation , to negative eigenvalues . since we are using only approximate values for the overlaps
, it may happen that we obtain states with a nominally negative norm , whence we can get poles of negative weight . setting @xmath151 throughout amounts to assuming that all our effective particles are orthogonal to one another and obviously removes the problem with nonorthogonalities .
this seems reasonable , because we are neglecting overlap terms proportional to the spin correlation function already in the hubbard - i approximation .
the lesson then is basically the same as discussed already before : the hubbard - i approximation is well defined only when applied to an ` ideally paramagnetic ' state with no correlations of finite range , and applying it to a state with finite spin correlations represents an approximation .
in summary , we have investigated the most important corrections over the hubbard - i approximation in the limit @xmath156 and electron density @xmath146 .
we have seen that the hubbard - i approximation describes charge fluctuations on a ` background ' of singly occupied sites , which is moreover assumed to have zero spin correlations .
the charge fluctuations are point - like , and correspond to an electron moving between empty sites and an electron moving between singly occupied sites .
we note that a very similar construction can also be applied to the kondo lattice@xcite and in fact reproduces the single particle spectra very well .
this is probably due to the fact that the kondo lattice has a unique ground state in the limit of zero kinetic energy , whereas the ground state of the hubbard model is highly degenerate in the case @xmath157 .
+ in our extended scheme for the hubbard model we have augmented these point - like charge fluctuations by additional ` particles ' which are composite in character and consist of a point - like charge fluctuation together with a spin- , density- , or @xmath1-excitation .
comparison of the obtained single - particle spectral density with qmc results for a variety of systems showed a quite reasonable agreement .
in particular the apparent @xmath7-band structure seen in the numerical spectra finds its natural explanation in the extended hubbard approximation .
we also note that qmc simulations where the spectra of the composite excitations have actually been computed@xcite , further support the present interpretation .
we thus have a quite successful method of computing the full quasiparticle band structure of the hubbard model , at least in the paramagnetic case and at half - filling .
+ the present scenario for the nature of the composite excitations also allows to make a connection with various theories for the hole motion in an antiferromagnet@xcite .
there , one is describing holes dressed by antiferromagnetic spin fluctuations .
when acting on the nel state the operators @xmath72 obviously describe precisely a hole together with a ` spin wave ' on a nearest neighbor or , put another way , a ` string ' of length one .
the terms which were omitted from the equation of motion for the @xmath72 then would correspond to strings of length two and so on . while such longer - ranged strings are apparently of minor importance in the paramagnetic phase , one may expect that they become more and more important for the description of the dispersion the stronger the antiferromagnetic correlations .
the relative importance of such longer ranged strings therefore may be the mechanism for the crossover from the hubbard - i like dispersion in the paramagnetic phase to the spin - density - wave like dispersion in the antiferromagnetic phase .
similarly , one might think of formulating the entire hubbard - i approximation also in the antiferromagnetic phase , by constructing the hamiltonian for charge fluctuations explicitly for a nel ordered spin background .
+ we thank h .-
evertz and w. hanke for discussions .
this work was supported by bmbf ( 05sb8wwa1 ) , computations were performed at hlrs stuttgart , lrz mchen and hlrz jlich .
transformation properties of various operators under particle - hole and positive / negative @xmath0 transformation .
+ j. hubbard , proc .
soc . a * 276 * , 238 ( 1963 ) .
j. hubbard , proc .
a * 277 * , 237 ( 1964 ) ; j. hubbard , proc .
a * 281 * , 401 ( 1964 ) .
l. m. roth , phys . rev . * 184 * , 451 ( 1969 ) .
g. geipel and w. nolting , phys .
b * 38 * , 2608 ( 1988 ) ; w. nolting and w. borgiel , phys . rev .
b * 39 * , 6962 ( 1989 ) .
b. mehlig , h. eskes , r. hayn , and m. b. j. meinders , phys .
b * 52 * , 2463 ( 1995 ) .
j. beenen and d. m. edwards , phys .
b * 52 * , 13636 ( 1995 ) . c. grber , m. g. zacher , and r. eder , cond - mat/9902015 .
e. dagotto , rev .
phys . * 67 * , 63 ( 1994 ) .
r. micnas , j. ranninger , and s. robaszkiewcz , rev .
* 62 * , 113 ( 1990 ) .
a. avella , f. mancini , d. villani , l. siurakshina , v.yu .
yushankhai , int .
b * 12 * , 81 ( 1998 ) ; see also f. mancini , phys .
a * 249 * , 231 ( 1998 ) . c. n. yang , phys .
* 63 * , 2144 ( 1989 ) ; s. c. zhang , phys .
. lett . * 65 * , 120 ( 1990 ) .
the more complicated ( anti)commutators in this work were obtained by a mathematica implementation of the fermion algebra .
s. pairault , d. senechal , and a .-
s. tremblay , phys .
lett . * 80 * , 5389 ( 1998 ) .
r. eder , o. rogojanu , and g. a. sawatzky , phys .
b * 58 * , 7599 ( 1998 ) . c. grber , r. eder , w. hanke ( in preparation ) .
l. n. bulaevskii , e. l. nagaev , and d. i. khomskii , sov .
jetp * 27 * , 638 ( 1967 ) .
s. a. trugman , phys .
b * 37 * , 1597 ( 1988 ) ; ibid . * 41 * , 892 ( 1990 ) .
r. eder and k. w. becker , z. phys .
b * 78 * , 219 ( 1990 ) .
o. p. sushkov , solid state commun . * 83 * , 303 ( 1992 ) .
g. reiter , phys .
b * 49 * , 1536 ( 1994 ) .
r. hayn , a. f. barabanov , j. schulenburg , and j. richter , phys .
b * 53 * , 11714 ( 1996 ) . | we reanalyze the hubbard - i approximation by showing that it is equivalent to an effective hamiltonian describing fermionic charge fluctuations , which can be solved by bogoliubov transformation . as the most important correction in the limit of large @xmath0 and
weak spin correlations we augment this hamiltonian by further effective particles , which describe composite objects of a fermionic charge fluctuation and a spin- , density- or @xmath1 excitation .
the scheme is valid for positive and negative @xmath0 .
we present results for the single particle green s function for the two - dimensional hubbard model with and without @xmath2 and @xmath3 terms , and compare to quantum monte - carlo ( qmc ) results for the paramagnetic phase .
the overall agreement is significantly improved over the conventional hubbard - i or two - pole approximation . 2 |
blood flows through a complex network of the circulatory system from large arteries to very tiny capillaries in order to ensure oxygen delivery and to remove metabolic waste .
this task is mainly carried out by red blood cells ( rbcs ) that are remarkably deformable , in healthy conditions , and therefore able to squeeze into tiny capillaries . a change in rheological and flow properties of the blood
is often associated with hematological diseases or disorders @xcite ( e.g. sickle - cell anemia , malaria , polycythemia vera ) .
understanding blood flow and its dependence on the mechanical properties of its constituents may improve and lead to new applications in biomedical technology , for example in blood substitutes development and transfusion techniques .
a major open problem in blood circulation is to understand the perfusion in the vasculature networks , especially in the microvasculature where rbcs accomplish their vital functions .
for example , an improper hematocrit distribution is observed in heart microcirculation with consequences such as occlusion zones ( within many patients with apparently healthy coronary arteries ) .
these abnormal traffic zones cause a lack of oxygen supply to tissues that leads to cardiac ischemic disease @xcite .
the precise origin of this dysfunction is still a matter of debate .
the principal mechanism that dictates blood heterogeneity in the microvascular networks is the hematocrit partition at the level of bifurcations .
rbcs do not behave as passive tracers . their shape flexibility and dynamics have a decisive role because their size is comparable to that of blood capillaries .
a well known phenomenon in microcirculation is the zweifach - fung effect @xcite : if we consider a bifurcation ( as in figure [ figillus ] ) , the child branch with the lower flow rate is depleted in rbcs as compared to the parent vessel , while the other , higher flow rate child branch is enriched .
that is , if in the parent vessel the total volumetric flow rate is @xmath9 and the rbc volumetric flux is @xmath10 , and in the child branch with the lower flow rate this flow rate is @xmath11 and the rbc flux @xmath12 , then @xmath13 . when the flow rate is sufficiently small , the hematocrit in the child branch can even drop down to zero , while it reaches high values in the other branch . obviously , this phenomenon has physiological consequences as it alters the transport of oxygen and other essential metabolites , and may even trigger pathological disorders ( e.g. occlusions in high hematocrit regions where the viscosity is higher and cell adhesion is favored ) .
previous studies have investigated the impact of the confinement @xcite , the bifurcation geometry @xcite , the hematocrit @xcite , and the rbcs deformability @xcite and aggregation @xcite .
most of these parameters influence rbcs distribution in the feeding flow , which is believed to dictates the partition at the bifurcation @xcite .
the zweifach - fung effect results from the existence of a cell free layer ( cfl ) close to the walls , which is only occupied by plasma .
the feeding flow is divided by a separating streamline into two parts , one feeding the low flow rate branch and the other feeding the high flow rate branch . due to the cfl
, the rbc fraction entering the low flow rate branch is smaller compared to the original rbc fraction in the total feeding flow .
the depletion in the low flow rate branch is accompanied by enrichment in the high flow rate branch .
in addition to the cfl as the main cause of the zweifach - fung effect , it has been argued that there is a relatively small counteracting effect where cells entering the bifurcation tend to be displaced towards the low flow rate branch compared to fluid streamlines ( a fact that slightly reduces the zweifach - fung effect ) @xcite , but this question is still debated @xcite .
the length ratio between the two child branches is set to 3 .
a , b : low viscosity contrast ( experiments with @xmath14=0.85 and simulation with @xmath14=1 ) .
c , d : high viscosity contrast ( experiments with @xmath14=10.3 and simulation with @xmath14=10).[figillus],width=648 ] the existence of a cfl is a consequence of the lateral migration of rbcs towards the vessel center .
this migration is a result of the wall - induced lift force due to hydrodynamic interactions @xcite , that depends on the nature of rbc dynamics ( like tank - treading or tumbling @xcite ) .
the final distribution of rbcs in a confined straight vessel is not only due to the lift force , but it is also influenced by other factors : ( i ) the curved velocity profile of the poiseuille flow @xcite , and ( ii ) the cell - cell hydrodynamic interactions @xcite . in the present work ,
we study the hematocrit partition at bifurcations using two - dimensional lattice boltzmann simulations , whose outcomes are validated and supported by microfluidic experiments . we show that rbcs deformability strongly impacts partition as long as the hematocrit is below @xmath15 ( within the normal range in microcirculation ) .
rbc deformability is governed by several parameters such as membrane stiffness ( shear , dilatation and bending elastic moduli ) , swelling degree , membrane viscosity and the viscosity contrast between the hemoglobin and the suspending fluid . here
we choose to tune the deformability through the latter parameter , the viscosity contrast , that controls the rbc dynamics ( tank - treading , tumbling or swinging ) then all the migration mechanisms at the origin of the cfl . on the other hand , and
more importantly , this study reveals that hematocrit partition can be completely reversed , that is the low flow rate child branch can be enriched in rbcs compared to the parent vessel .
this newly reported effect is an outcome of a subtle rbcs structuration in the microcirculatory system .
this highlights the importance of the notion of rbcs spatio - temporal organization as the main non - negligible ingredient to further understand blood perfusion in the microvasculature .
in both simulations and experiments , we use t - shaped bifurcations such as shown in fig.[figillus ] : a parent channel divides into two child branches with the same width , but with different lengths @xmath16 and @xmath17 ( @xmath18 ) .
the ratio of the flow rates in branches 1 and 2 is then given by @xmath19 , where @xmath20 and @xmath21 are the apparent viscosities of the suspension in branches 1 and 2 , respectively .
for dilute suspensions , where the viscosity is close to that of the suspending fluid , we simply have @xmath22 . in simulations
, we set the width of the channels to @xmath23 m and we vary @xmath24 from 1.43 to 3 . in experiments , we have @xmath25 , @xmath26 m and the height of the channel @xmath27 is 8 @xmath28 m .
the length of the parent vessel was chosen as long as possible to allow for the development of a stationary distribution of rbcs across the channel in the feeding flow ( 5 mm in experiments and 1.5 mm in simulations ) .
microfluidic channels were produced by standard soft lithography techniques , with molded pdms bonded to glass .
the rbc suspensions were perfused by a syringe pump ( kds legato 180 ) and imaging was performed by a video camera ( imaging source dmk 31af03 ) mounted on an inverted microscope with motorized stage ( olympus ix71 ) and a blue filter ( @xmath29 nm ) corresponding to an absorption peak of hemoglobin .
blood samples were provided by the etablissement franais du sang ( efs rhne - alpes ) from healthy donors .
rbcs were isolated by centrifugation after being washed twice in phosphate buffer saline ( pbs ) supplemented by 0.1 % bovine serum albumin ( bsa ) . to prevent sedimentation of rbcs in channels ,
the rbcs were re - suspended in density matching pbs and bsa solutions in a mixture of water and iodixanol ( optiprep from axis - shield ) .
this iso - dense solution was used either alone ( 1.94 mpa.s at 20 @xmath30c ) or after adding 5% dextran of molecular weight @xmath31 ( viscosity 23.4 mpa.s at 20 @xmath30c ) .
the viscosity of the internal hemoglobin solution of healthy rbcs is around 20 mpa.s at 20 @xmath30c @xcite .
this provides two values of the viscosity contrast @xmath14 , namely 10.3 and 0.85 .
the first value corresponds to the blood at 20 @xmath30c , while the physiological value at body temperature is around 5 - 6 @xcite .
note that we chose to vary the viscosity contrast @xmath14 as one way to tune deformability , and therefore the dynamics of lift and hydrodynamic interactions of cells .
stiffening cells using diamide or glutaraldehyde was another possibility .
however , from the experimental viewpoint , working with hardened cells at high volume fractions in such a confined environment is quite difficult due to jamming .
it would have been nearly impossible to inject a suspension of very stiff cells at hematocrits larger than 10% . also , the dynamics of glutaraldehyde hardened cells is pure tumbling , which corresponds to very high values of the viscosity ratio @xmath14 .
we do not expect the dynamics ( and therefore phase separation ) to change much at values of @xmath14 greater than 10 and we found more interesting to increase deformability by decreasing @xmath14 rather than trying to investigate less deformable cells ( with the experimental difficulties mentioned above ) .
local hematocrit measurements were made by comparing suspension flow images to a reference image without rbc , under identical illumination , and using the beer - lambert law of absorption .
the absorption coefficient was determined by a calibration with images at low hematocrit , where a direct measurement can be made by counting individual cells .
hematocrit in branch @xmath32 will be denoted @xmath33 . in experiments ,
@xmath27 is small enough ( @xmath34 m ) so that the flow is quasi two - dimensional . to allow a qualitative comparison with 2d numerical simulations ,
an area hematocrit @xmath35 was also derived by multiplying the number of cells per unit area by the average cross - sectional area of rbcs ( @xmath3619.8 @xmath28m@xmath37 ) . @xmath33 and @xmath35 are therefore linked by the relationship @xmath38 , where @xmath39 @xmath28m@xmath40 is the average volume of one cell . in simulations ,
we use lattice boltzmann method ( lbm ) to compute the fluid flow @xcite . each rbc is modeled by 60 nodes interconnected by a potential that allows bending , as well as a stretching modulus that penalizes distance variations between two adjacent nodes .
this achieves the rbc membrane incompressibility @xcite . in other words
, we set the spring constant to values as large as possible in order to keep the ratio between the membrane perimeter and area constant .
we define the reduced area as @xmath41 ( with p is the perimeter and a the enclosed area ) , which we set to 0.7 to produce a rbcs with a biconcave shape .
we use the immersed boundary method ( ibm ) to couple the fluid flow and rbc deformation @xcite . for comparison with experiments
, we set the viscosity contrast to @xmath42 and @xmath43 .
is set to 3 . a : simulations with @xmath14=1 and @xmath14=10 .
b : comparison between experiments ( @xmath14=0.85 and @xmath14=10.3 ) and the empirical law of pries _
@xcite ( solid line ) , for the same cross - sectional area .
the non - monotonous evolution of the relative hematocrit at low @xmath44 and high @xmath14 is related to the inversion of the zweifach - fung effect , on which we comment later on ( see fig .
[ fig - antizfexp]).[figphi1phi0],width=312 ]
as a guideline , we shall refer to the empirical law of pries _
et al . _ taken from refs @xcite that gives the hematocrit partition at a bifurcation : @xmath45 where @xmath46 , @xmath47 ( with @xmath48 the tube diameter in microns ) and @xmath49 $ ] .
@xmath44 is the volumetric hematocrit in the parent feeding branch , while @xmath50 is the hematocrit in a child branch .
@xmath51 is the feeding hematocrit in a reservoir that would be located right before the narrow feeding vessel . due to the fhrus effect
, @xmath51 is larger than @xmath44 and a relationship between both quantities is also given in @xcite : @xmath52 note that the partition law ( eq . [ eqpart ] ) has been validated through _ in - vivo _ experiments with rats ( thus at body temperature ) , with narrow capillaries ( of diameters @xmath48 lower than 30 microns ) , but with feeding hematocrits higher than 20% . as we shall compare predictions for 3d hematocrits in a cylindrical tube with either 2d simulations or experiments in a rectangular channel
, we should avoid any direct quantitative comparisons , but rather use pries _ et al . _ predictions as a guideline to identify where new behavior is exhibited .
for comparison with simulations , we set @xmath53 , where @xmath54 is the channel width , and we shall consider only the hematocrit ratios .
for the experiments in rectangular channels , we set @xmath48 to adjust the cross - sectional areas : @xmath55 .
we analyze in details how rbc deformability affects the hematocrit partition at the bifurcations .
figure [ figillus ] illustrates the zweifach - fung effect , observed in both experiments and simulations , at a feeding area hematocrit of 10% . in both cases ,
less rbcs enter the low flow rate branch ( the long branch ) simply due to the flow rates differences between the two child branches .
however , the asymmetry is significantly pronounced at low viscosity contrast @xmath14 ( when the suspending fluid is more viscous than the hemoglobin ) . to quantify the partition asymmetry , we measure the relative hematocrit @xmath56 ( or , equivalently , @xmath57 ) , in the low flow rate branch , while we vary the hematocrit in the parent branch ( fig . [ figphi1phi0]a ) . either in the simulations ( fig .
[ figphi1phi0]a ) or in the experiments ( fig .
[ figphi1phi0]b ) , we see less rbcs in the low flow rate branch than in the parent one ( @xmath58 ) , when the inlet hematocrits ( @xmath0 or @xmath44 ) lies between @xmath59 and @xmath60 , which is precisely a manifestation of the zweifach - fung effect . when the viscosity contrast is low , we observe a significantly strong reduction of hematocrit in the low flow rate branch , both in experiments and simulations , at moderate inlet hematocrit .
this interesting observation suggests that the rbcs mechanical properties can strongly impact the hematocrit partition _ in - vivo _ since the normal hematocrit is usually less than @xmath15 ( typically between 10 and 20 @xmath61 @xcite ) in microcirculation .
however , when the hematocrit is high enough , the viscosity contrast plays a minor role .
this is clear in simulations ( for @xmath0 larger than 25% , fig . [
figphi1phi0]1 ) .
similarly , in the experiments , above @xmath62 , data for both @xmath14 converge to the pries _
_ prediction .
the insensitivity to the viscosity contrast beyond a critical hematocrit ( @xmath63 ) is a robust phenomenon that is independent of the length ratio between the branches ( i.e. roughly the bulk flow rate ratio ) , as illustrated on figure [ figdepl1l2 ] ) . as a function of the hematocrit in the parent vessel @xmath0 , for several branches length ratios @xmath24 and viscosity contrasts @xmath14 . at low enough @xmath0 ,
the asymmetry between the two daughter branches is strongly enhanced as the viscosity contrast @xmath14 is decreased , while the partitioning becomes independent on @xmath14 for hematocrits above 20@xmath61 .
[ figdepl1l2],width=600 ] = 1 and @xmath14=10 , in simulations .
insets : snapshots showing the suspension at the bifurcation .
we define the cfl as a layer where the integrated concentration profile is below @xmath59 @xcite .
b : snapshots from experiments , for @xmath14=0.85 and 10.3 , and hematocrits @xmath6410 , 20 and 30% . every snapshot is a superimposition of 10 successive images in order to highlight the cfl in the parent vessel .
c : the stationary volume fraction density functions in the parent vessel obtained from simulations.[figprofiles],width=600 ] it is appealing to suggest that the dependency of the hematocrit partition upon the feeding hematocrit is the result of the up - stream organization of rbcs in the parent vessel due to hydrodynamic interactions . at low hematocrit flows ,
the cell - cell interaction is weak and the organization of rbcs , within the vessel , depends mainly on the dynamics of each rbcs , thus on @xmath14 .
the rbcs aggregate at the center of the vessel due to the wall - induced lift force , that increases with decreasing @xmath14 and increasing rbcs deformability @xcite .
this means that suspensions of rbcs with high viscosity contrasts have wider distributions ( smaller cfl ) in the channel as compared to suspensions having lower viscosity contrast . as a consequence
, the asymmetry in the partition is expected to increase when the viscosity contrast decreases , as shown in figure [ figdepl1l2 ] . to support this argument ,
the cfl thickness in the parent vessel and the configuration of rbcs before the bifurcation are reported in figure [ figprofiles]a , b .
we can clearly see that rbcs distribution at high viscosity contrast ( @xmath14=10 in simulations and @xmath14=10.3 in experiments ) is wider than that at low viscosity contrast ( @xmath14=1 in simulations and @xmath14=0.85 in experiments ) when the feeding hematocrit is low ( @xmath6520% , see also figure [ figprofiles]c - i , ii )
. however , when the hematocrit increases , the contribution of the hydrodynamic interactions among rbcs becomes stronger and stronger .
this causes a broadening of the distribution that acts against the lift force .
consequently , the partition between the two branches becomes more symmetric ( that is @xmath56 becomes close to 1 ) .
interestingly , those broad distributions are quasi independent of the viscosity contrast ( see figure [ figprofiles]a and c - iii , iv ) .
consequently , @xmath56 does not depend on @xmath14 either ( see figure [ figdepl1l2 ] ) .
thus the distribution is independent of the strength of the interactions between cells and between cells and walls , but it is mainly caused by geometrical constraints .
in other words , interaction between cells and the lift forces both depend on @xmath14 , and this result indicates that they depend more or less on @xmath14 in the same way .
the @xmath14 contributions cancel out once a critical feeding hematocrit is reached .
noteworthy , beyond this critical hematocrit the separating ratio @xmath56 quasi plateaus which enforces the idea that in this regime , the feeding flow can be considered as a three - layer fluid ( fluid - cell - fluid ) .
the width of each layer will depend neither on the strength of interaction ( which is related to deformability ) nor on the volume fraction . .
[ fig5 - 40],width=312 ] now we focus on the low hematocrit case , for which the partition depends strongly on the detail of the interactions and on the volume fraction .
moreover a peculiar effect arises due to the prevalence of the discrete nature of blood at that scale . for all hematocrits ,
the distribution of rbcs is not homogeneous , but rather exhibits two lateral peaks ( figure [ figprofiles ] ) .
this become more pronounced at low hematocrit ( @xmath66 ) , where a two - file distribution of rbcs is observed , as shown in figure [ fig5 - 40 ] . for @xmath14=10
, there is almost no cell flowing in the central part of the vessel , even though the wall - lift force tends to center them .
the structure adopted by the suspension can be viewed as a juxtaposition of layers with high and low hematocrits .
for example , as shown in figure [ fig5 - 40 ] ( top panel with @xmath43 ) the central part is depleted in rbcs , but it is escorted by two enriched layers , which themselves are surrounded by two depleted layers at the periphery ( close to the channel walls ) . this 5-layer configuration ( fluid - cell - fluid - cell - fluid )
has an extremely interesting impact on the partition .
this can be highlighted by measuring @xmath56 as a function of the bulk flow rate ratio @xmath67 between a child branch and the parent vessel , for fixed @xmath0 ( figure [ fig - antizf ] ) .
if we focus first on the results for high @xmath0 ( @xmath68 ) , we find again the insensitivity to @xmath14 .
as @xmath67 is increased from 0 to 0.5 , the low flow rate branch 1 recruits first the cfl and then the cells .
this implies that @xmath69 starts at 0 and increases until reaching @xmath0 when the situation is symmetric ( @xmath70 ) .
our results agree with a previous 2d simulation obtained for @xmath71 @xcite as well as with the empirical law of pries and coworkers @xcite . .
for simulations , the relative hematocrit is @xmath56 . for pries law ,
it is given by @xmath57 .
a : high hematocrit ( @xmath72 ) .
b : low hematocrit ( @xmath73).[fig - antizf],width=312 ] as a function of the hematocrit in the parent vessel @xmath44 . a : low @xmath14 ; b : high @xmath14 .
the length ratio between the branches is set to 3 .
the width @xmath54 of the inlet channel is set to 10 , 20 and 30 @xmath28 m .
the corresponding empirical laws of pries _ _ et al.__@xcite are also shown . the dotted line ( the one with highest slope ) corresponds to equal partition ( @xmath74 ) .
for @xmath75 @xmath28 m , the data are the continuation of the data already reported in figure [ figphi1phi0].c : snapshots for each width @xmath54 and two concentrations @xmath76 and 4.5% .
every snapshot is the superimposition of 10 successive images in order to highlight the structure of the suspension in the parent vessel[fig - antizfexp],width=600 ] an unexpected phenomenon is observed at low enough hematocrit , for @xmath77 and at high viscosity contrast @xmath43 , in contrast to the high hematocrit regime ( see figure [ fig - antizf ] b , @xmath43 ) . at low @xmath67 , the peripheral cfl in the parent vessel is recruited by the branch 1 so @xmath69 starts at 0 and increases when @xmath11 increases .
around @xmath78 , @xmath69 becomes larger than @xmath0 .
this means the hematocrit is increased in the low flow rate branch , which is the reverse behavior of the zweifach - fung effect .
the five - layer structure mentioned above is the key ingredient for understanding this unexpected behavior : in the intermediate range @xmath79 , the low flow rate branch recruits the lateral cfl layer plus the adjacent rbc - rich layer among the five layers .
by contrast , the high flow rate branch recruits the cfl layer close to the opposite wall plus its adjacent rbc - rich layer ( exactly as the low flow rate branch ) as well as the central ( and depleted ) layer .
thus , while both branches recruit approximately the same amount of cells per unit time , those are more dilute in the high flow rate branch , which receives more fluid , while in the classical zweifach - fung effect , the high flow rate branch is the one that receives more cells . for @xmath42 ( see figure [ fig - antizf]b with @xmath42 ) , the two - peak structure is not as marked as in the case of @xmath43 , so the reverse zweifach - fung effect is not as strong . the subtle role played by the suspension structuring at low hematocrit
is also supported by our experiments , where the interplay between the diffusion and the wall - lift force is controlled by varying the width @xmath54 of the inlet channel ( see figure [ fig - antizfexp ] ) .
when @xmath54 is low ( @xmath54 = 10 @xmath28 m ) , the hematocrit in the low flow rate branch is much lower than expected from pries _
_ predictions ( which were not validated on this confinement range ) .
this is caused by the cfl effect that becomes very strong .
as in the simulations , for @xmath80 , this effect is more pronounced at low @xmath14 , that corresponds to a more important wall lift force .
the 5-layer structure is clearly observed also in the experiments for @xmath81 , @xmath75 or 30 @xmath28 m and at low @xmath14 ( figure [ fig - antizfexp]c ) , but not at high @xmath14 , while it was more strongly marked at high @xmath14 in the simulations .
this indicates that this peculiar structure is very dependent on the mechanical properties of the cells and also on the degree of confinement .
nevertheless , a robust feature is valid in both simulations and experiments : when the two - file structure of rbcs takes place , a clear inversion of the blood partition at the bifurcation is observed . in the experiments
, this corresponds to figure [ fig - antizfexp]a , where some points lie above the equal partition line when @xmath82 .
there are also situations in which one of the two branches can be even completely devoid of rbcs ( figure [ fig - antizfexp]c ) .
a corresponding prolonged lack of rbcs perfusion to real blood vessels causes dysfunction and possibly ischemia disease . because rbc mechanical properties are affected by aging and pathologies
, these can induce abnormal partitions of the hematocrit in the vascular network .
as a result of the interplay between the zweifach - fung effect and the fhrus effect the hematocrit in microcirculation can reach values as low as 10 - 20% compared to the average hematocrit in human body ( 45% ) . at such a low hematocrit , our simulations and _ in vitro _
microfluidic experiments have revealed that rbcs partition at the level of bifurcations depends strongly on the viscosity contrast between the viscosities of the rbc hemoglobin and the suspending fluid . in the extreme hemodilution ,
our results exhibit a newly reported phenomenon : the low flow rate branch may receive higher hematocrit than the high flow rate branch in opposition to the known zweifach - fung effect .
this phenomenon is observed under moderate confinement and is the result of a peculiar structuring of the cell suspension . under stronger confinement , other strong discrepancy with pries _
empirical law was highlighted , with a strong asymmetry in the partitioning ( enhanced zweifach - fung effect ) .
our findings suggest that the viscosity contrast must be taken into consideration and carefully analyzed in order to have a firm understanding of rbc distribution in microcirculation .
this physiological parameter increases with aging as well as with some pathologies . the results of our present work
provide a valuable background needed to pinpoint the various rbcs properties that govern hematocrit partition , and thus oxygen delivery in the microcirculation in general .
z. s. thanks vassanti audemar for experimental advice and assistance . z. s. , g. c. , c. m. and t. p. acknowledge financial support from cnes ( centre national detudes spatiales ) and esa ( european space agency ) . the dyfcom team of liphy ( z. s. , g. c. , c. m. and t. p. ) is member of labex tec21 ( investissements davenir - grant agreement anr-11-labx-0030 ) , structure fdrative de recherche idysco ( cnrs ) , fdration galileo galilei de grenoble ( fr3345 cnrs - ujf - grenoble inp - irstea ) and groupement de recherche mecabio ( gdr3570 cnrs ) . j. o. barber , j. p. alberding , j. m. restrepo , and t. w. secomb .
simulated two - dimensional red blood cell motion , deformation , and partitioning in microvessel bifurcations .
_ , 36:0 16901698 , 2008 .
s. chien , c. d. tvetenstrand , m. a. epstein , and g. w. schmid - schonbein .
model studies on distributions of blood cells at microvascular bifurcations . _ am .
j. physiol .
heart circ .
_ , 248:0 h568h576 , 1985 .
d. fedosov , b. caswell , s. suresh , and g. karniadakis . quantifying the biophysical characteristics of plasmodium - falciparum - parasitized red blood cells in microcirculation .
_ proceedings of the national academy of sciences _ , 1080 ( 1):0 3539 , 2011 .
a. kumar , r. g. henrquez rivera , and m. d. graham .
flow - induced segregation in confined multicomponent suspensions : effects of particle size and rigidity .
_ j. fluid mech .
_ , 738:0 423462 , 1 2014 .
issn 1469 - 7645 .
j. l. mcwhirter , h. noguchi , and g. gompper .
flow - induced clustering and alignment of vesicles and red blood cells in microcapillaries .
_ proceedings of the national academy of sciences _ , 1060 ( 15):0 60396043 , 2009 .
j. m. sherwood , e. kaliviotis , j. dusting , and s. balabani .
hematocrit , viscosity and velocity distributions of aggregating and non - aggregating blood in bifurcating microchannel .
_ biomicrofluidics _ , 6:0 024119 , 2012 .
w. xiong and j. zhang .
two - dimensional lattice boltzmann study of red blood cell motion through microvascular bifurcation : cell deformability and suspending viscosity effects .
_ biomech . model .
_ , 11:0 575583 , 2012 .
j. zhang , p. c. johnson , and a. s. popel .
an immersed boundary lattice boltzmann approach to simulate deformable liquid capsules and its application to microscopic blood flows .
_ physical biology _ , 40 ( 4):0 285 , 2007 | partitioning of red blood cells ( rbcs ) at the level of bifurcations in the microcirculatory system affects many physiological functions yet it remains poorly understood .
we address this problem by using t - shaped microfluidic bifurcations as a model .
our computer simulations and _ in vitro _
experiments reveal that the hematocrit ( @xmath0 ) partition depends strongly on rbc deformability , as long as @xmath1% ( within the normal range in microcirculation ) , and can even lead to complete deprivation of rbcs in a child branch .
furthermore , we discover a deviation from the zweifach - fung effect which states that the child branch with lower flow rate recruits less rbcs than the higher flow rate child branch . at small enough @xmath0
, we get the inverse scenario , and the hematocrit in the lower flow rate child branch is even higher than in the parent vessel .
we explain this result by an intricate up - stream rbc organization and we highlight the extreme dependence of rbc transport on geometrical and cell mechanical properties .
these parameters can lead to unexpected behaviors with consequences on the microcirculatory function and oxygen delivery in healthy and pathological conditions .
affiliations : : : @xmath2 laboratoire interdisciplinaire de physique ( liphy ) umr5588 cnrs - universit grenoble alpes , grenoble f-38041 , france ; @xmath3 cnrs - sorbonne university , universit de technologie de compigne , umr7338 biomechanics and bioengineering , 60203 compigne , france ; @xmath4 tp1 , university of bayreuth , d-95447 bayreuth , germany ; @xmath5 laboratoire dhmatologie , chu , grenoble , france ; @xmath6timc - imag / therex , cnrs umr5525 , universit grenoble alpes , grenoble , france ; @xmath7 department of applied physics , eindhoven university of technology , p.o .
box 513 , 5600 mb eindhoven , the netherlands ; @xmath8 faculty of science and technology , mesa+ institute , university of twente , 7500 ae enschede , the netherlands abbreviated title : : : inversion of hematocrit partition corresponding author : : : thomas podgorski , e - mail : [email protected] * keywords : * microcirculation ; blood ; red blood cell ; microfluidics ; lattice boltzmann method |
the famous v.ya.ivriis conjecture @xcite says that _ in every billiard with infinitely - smooth boundary in a euclidean space of any dimension the set of periodic orbits has measure zero .
_ as it was shown by v.ya.ivrii @xcite , his conjecture implies the famous h.weyls conjecture on the two - term asymptotics of the spectrum of laplacian @xcite .
a brief historical survey of both conjectures with references is presented in @xcite . for the proof of ivrii s
conjecture it suffices to show that for every @xmath0 the set ot @xmath1-periodic orbits has measure zero . for @xmath2 this was proved in @xcite for dimension two and in @xcite for any dimension . for @xmath3 in dimension two this was proved in @xcite .
ivrii s conjecture is open already for piecewise - analytic billiards , and we believe that this is its principal case . in the latter case ivrii s conjecture
is equivalent to the statement saying that for every @xmath0 the set of @xmath1-periodic orbits has empty interior .
in the present paper we study a complexified version of ivrii s conjecture in complex dimension two .
more precisely , we consider the complex plane @xmath4 with the complexified euclidean metric , which is the standard complex - bilinear quadratic form @xmath5 .
this defines notion of symmetry with respect to a complex line , reflections with respect to complex lines and more generally , reflections of complex lines with respect to complex analytic ( algebraic ) curves .
the symmetry is defined by the same formula , as in the real case .
more details concerning the complex reflection law are given in subsection 2.1 .
one could have replaced the initial real euclidean metric by a pseudo - euclidean one : the geometry of the latter is somewhat similar to that of our complex euclidean metric .
billiards in pseudo - euclidean spaces were studied , e.g. , in @xcite .
proofs of the classical poncelet theorem and its generalizations by using complex methods can be found in @xcite . to formulate the complexified ivrii s conjecture ,
let us introduce the following definitions .
a complex projective line @xmath6 is _ isotropic _ , if either it coincides with the infinity line , or the complexified euclidean quadratic form vanishes on @xmath7 . or equivalently , a line is isotropic , if it passes through some of two points with homogeneous coordinates @xmath8 : the so - called _ isotropic points at infinity _ ( also known as _ cyclic _ ( or _ circular _ ) points ) .
[ deforb ] a _ complex analytic ( algebraic ) planar billiard _ is a finite collection of complex irreducible analytic ( algebraic ) curves @xmath9 that are not isotropic lines ; we set @xmath10 , @xmath11 .
@xmath1-periodic billiard orbit _ is a collection of points @xmath12 , @xmath13 , @xmath14 , such that for every @xmath15 one has @xmath16 , the tangent line @xmath17 is not isotropic and the complex lines @xmath18 and @xmath19 are symmetric with respect to the line @xmath17 and are distinct from it .
( properly saying , we have to take vertices @xmath20 together with prescribed branches of curves @xmath21 at @xmath20 : this specifies the line @xmath17 in unique way , if @xmath20 is a self - intersection point of the curve @xmath21 . ) [ defref ] a complex analytic ( algebraic ) billiard @xmath9 is _
@xmath1-reflective , _ if it has an open set of @xmath1-periodic orbits . in more detail
, this means that there exists an open set of pairs @xmath22 extendable to @xmath1-periodic orbits @xmath23 .
( then the latter property automatically holds for every other pair of neighbor mirrors @xmath21 , @xmath24 . )
* problem ( complexified version of ivrii s conjecture ) . * _ classify all the @xmath1-reflective complex analytic ( algebraic ) billiards .
_ contrarily to the real case , where there are no piecewise @xmath25-smooth 4-reflective planar billiards @xcite , there exist 4-reflective complex algebraic planar billiards . in the present paper we classify them ( theorem [ tclass ] stated at the end of the subsection ) .
basic families of 4-reflective algebraic planar billiards are given below .
theorem [ tclass ] shows that their straightforward analytic extensions cover all the 4-reflective algebraic planar billiards .
an @xmath7-reflective analytic ( algebraic ) billiard generates @xmath26-reflective analytic ( algebraic ) billiards for all @xmath27 .
therefore , @xmath1-reflective billiards exist for all @xmath28 .
now let us pass to the construction of 4-reflective complex billiards .
the construction comes from the real domain , and we use the following relation between real and complex reflection laws in the real domain .
[ rskew ] in a real billiard the reflection of a ray from the boundary is uniquely defined : the reflection is made at the first point where the ray meets the boundary . in the complex case , the reflection of lines with respect to a complex analytic curve is a multivalued mapping _ ( correspondence ) _ of the space of lines in @xmath29 : we do not have a canonical choice of intersection point of a line with the curve .
moreover , the notion of interior domain does not exist in the complex case , since the mirrors have real codimension two .
furthermore , the real reflection law also specifies the _ side _ of reflection .
namely , a triple of points @xmath30 , @xmath31 , @xmath32 , and a line @xmath33 through @xmath34 satisfy the _ real reflection law , _ if the lines @xmath35 and @xmath36 are symmetric with respect to @xmath37 , and also the points _ @xmath38 and @xmath39 lie in the same half - plane _ with respect to the line @xmath37 . the _ complex reflection law _ says only that the _ complex lines _ @xmath35 and @xmath36 are symmetric with respect to @xmath37 and does not specify the positions of the points @xmath38 and @xmath39 on these lines : they may be arbitrary . a triple of real points @xmath30 , @xmath31 , @xmath32 and a line @xmath33 through @xmath34 satisfy the complex reflection law , if and only if - either they satisfy the usual real reflection law ( and then @xmath38 and @xmath39 lie on the same side from the line @xmath37 ) , - or the line @xmath37 is the bissectrix of the angle @xmath40 ( and then @xmath38 and @xmath39 lie on different sides from the line @xmath37 ) .
in the latter case we say that the triple @xmath38 , @xmath34 , @xmath39 and the line @xmath37 satisfy the _ skew reflection law_. [ ex - lines ] consider the following complex billiard with four mirrors @xmath41 , @xmath42 , @xmath43 , @xmath44 : @xmath45 is a non - isotropic complex line ; @xmath42 is an arbitrary analytic ( algebraic ) curve distinct from @xmath41 ; @xmath44 is symmetric to @xmath42 with respect to the line @xmath41 .
this complex billiard obviously has an open set of 4-periodic orbits @xmath46 , these orbits are symmetric with respect to the line @xmath41 , see fig.1 .
[ ex - lines2 ] consider a complex billiard formed by four distinct lines @xmath41 , @xmath42 , @xmath43 , @xmath44 passing through the same point @xmath47 such that _ the line pairs _
@xmath48 and @xmath49 ( or equivalently , @xmath50 and @xmath51 ) are unimodularly isometric .
that is , there exists a complex euclidean isometry with unit jacobian that transforms one pair into the other : a _ complex rotation around the intersection point @xmath52 , _
see fig.2 .
this billiard is 4-reflective , or equivalently , the composition of symmetries @xmath53 , @xmath54 , with respect to the lines @xmath41 , @xmath42 , @xmath43 and @xmath44 that act on the dual projective plane is identity : @xmath55 on @xmath56 . indeed , the latter identity is equivalent to the same identity on the projective plane , i.e. , @xmath57 on @xmath29 .
the latter holds if there exists a complex rotation around @xmath52 sending the pair @xmath48 to @xmath49 .
this follows from the fact that the composition of symmetries with respect to two lines through a point @xmath52 is a complex rotation around @xmath52 and the commutativity of the group of complex rotations around @xmath52 .
let us prove the converse : assuming the same identity we show that @xmath49 is obtained from @xmath48 by complex rotation around @xmath52 .
let @xmath58 denote the image of the line @xmath42 under the complex rotation sending @xmath41 to @xmath44 .
one has @xmath59 , by the previous statement and assumption . therefore ,
@xmath60 , hence @xmath61 .
[ ex - circles ] consider the complex billiard with four mirrors @xmath41 , @xmath42 , @xmath43 , @xmath44 : @xmath45 and @xmath62 are complexifications of two distinct concentric circles on the real plane .
say , @xmath45 is the smaller circle .
we say that we reflect from the bigger circle @xmath62 in the usual way , `` from interior to interior '' , and we reflect from the smaller circle in the skew way : `` from interior to exterior '' and vice versa .
take arbitrary points @xmath63 and @xmath64 such that the segment @xmath35 lies outside the disk bounded by @xmath41 .
let @xmath65 denote the ray symmetric to the ray @xmath66 with respect to the diameter through @xmath34 .
let @xmath39 denote its second intersection point with the smaller circle mirror @xmath45 .
( we say that the ray @xmath65 passes through the first intersection point `` without noticing the mirror @xmath41 '' . )
consider the symmetry with respect to the diameter orthogonal to the line @xmath67 .
let @xmath68 denote the symmetric image of the point @xmath34 , see fig.3 .
thus , we have constructed a self - intersected quadrilateral @xmath46 depending analytically on two parameters : @xmath63 and @xmath64
. * claim . * _ the above quadrilateral @xmath46 is a 4-periodic orbit of the complex billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 , and the billiard is 4-reflective . _
the reflection law is obviously satisfied at the vertices @xmath69 .
it suffices to check the skew reflection law at the vertices @xmath70 . by symmetry
, it suffices to show that the tangent line @xmath71 is the bissectrix of the angle @xmath72 .
the union of lines @xmath73 and @xmath36 intersects the circle @xmath41 at four points with equal intersection angles , since these lines are symmetric with respect to a diameter . similarly , the lines @xmath35 and @xmath36 intersect the circle @xmath41 at equal angles : they are symmetric with respect to the diameter through @xmath34 .
the two latter statements together imply that the tangent line @xmath71 is the bissectrix of the angle @xmath72 .
thus , we have a two - parametric quadrilateral orbit family @xmath46 that extends analytically to complex domain .
hence , the billiard is 4-reflective .
this proves the claim .
consider a generalization of the above example : a complex billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 similar to the above one , but now @xmath45 and @xmath62 are complexifications of distinct _ confocal ellipses _ , say , @xmath45 is the smaller one . [ tell ] ( m.urquhart , see ( * ? ? ?
* , corollary 4.6 ) ) .
the above two confocal ellipse billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 , see fig.4 , is 4-reflective .
the main result of the paper is the following theorem .
[ tclass ] a complex planar algebraic billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 is 4-reflective , if and only if it has one of the following types : case 1 ) : some mirror , say @xmath41 is a line , @xmath45 , and the curves @xmath74 are symmetric with respect to the line @xmath41 , cf .
example [ ex - lines ] , fig.1 .
case 2 ) : @xmath41 , @xmath42 , @xmath43 , @xmath44 are distinct lines through the same point @xmath47 ; the line pair @xmath48 is sent to @xmath49 by complex rotation around @xmath52 , cf .
example [ ex - lines2 ] , fig.2 .
case 3 ) : @xmath45 , @xmath62 and they are distinct confocal conics , cf
. theorems [ tell ] and [ tquad ] , fig.3 , 4 . the notion of confocality of complex conics is the immediate analytic extension to complex domain of confocality of real conics . see more precise definition [ defconf ] in subsection 2.4 .
there is an analogue of ivrii s conjecture in the invisibility theory : plakhov s invisibility conjecture ( * ? ? ?
* conjecture 8.2 , p.274 ) .
its complexification coincides with the above complexified ivrii s conjecture @xcite . for more results on invisibility see @xcite .
another analogue of the 4-reflective planar ivrii s conjecture is tabachnikov s commuting billiard problem ( * ? ? ?
* , the last paragraph ) ; their complexifications coincide .
thus , results about the complexified ivrii s conjecture have applications not only to the original ivrii s conjecture , but also to plakhov s invisibility conjecture and tabachnikov s commuting billiard problem . theorem [ tclass ]
is proved in sections 24 . in section 6
we present its application ( theorem [ class ] ) , which gives the classification of so - called 4-reflective real algebraic pseudo - billiards : billiards that have open set of 4-periodic orbits with skew reflection law at some vertices and usual at the other ones .
the definition of confocal complex conics and the proof of 4-reflectivity of a billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 on distinct confocal conics @xmath41 and @xmath42 ( theorem [ tquad ] ) are given in subsection 2.4 .
the 4-reflectivity of billiards of types 1 ) and 2 ) in theorem [ tclass ] was already explained in examples [ ex - lines ] and [ ex - lines2 ] .
the main part of theorem [ tclass ] saying that each 4-reflective planar algebraic billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 is of one of the types 1)3 ) is proved in subsection 2.3 and sections 3 , 4 .
the main idea of the proof is similar to that from @xcite : to study the `` degenerate '' limits of open set of quadrilateral orbits , i.e. , quadrilaterals having either an edge tangent to a mirror at an adjacent vertex , or a pair of coinciding neighbor vertices , or a vertex that is an isotropic tangency point of the corresponding mirror .
we deal with the compact riemann surfaces @xmath75 , @xmath76 , @xmath77 , @xmath78 , the so - called normalizations of the curves @xmath41 , @xmath42 , @xmath43 , @xmath44 respectively that parametrize them bijectively except for self - intersections .
we study the closure in @xmath79 of the open set of 4-periodic orbits in the usual topology .
this is a purely two - dimensional algebraic set , which we will call the 4-reflective set and denote @xmath80 , that contains a zariski open and dense subset @xmath81 of 4-periodic billiard orbits ( proposition [ comp - set ] ) .
the above - mentioned degenerate quadrilaterals form the complement @xmath82 .
the proof of theorem [ tclass ] consists of the following steps .
step 1 . _ description of a large class of degenerate quadrilaterals in a billiard of type 3 ) will be given by theorem [ irrdeg ] in section 5 .
] @xmath83 _ ( subsections 2.1 and 2.2 ) .
1a ) case , when _
some vertex , say @xmath34 is an isotropic tangency point _ of the corresponding mirror @xmath42 .
we prove the _ isotropic reflection law : _ if @xmath84 , then at least one of the lines @xmath35 , @xmath36 coincides with the isotropic tangent line @xmath85 .
( propositions [ plimits ] and [ comp - set ] in subsection 2.1 ) 1b ) case , when _
some edge is tangent to the mirror through an adjacent vertex , _ say , @xmath86 .
we show that _ this can not be the only degeneracy _
( proposition [ tang ] ) .
we show that the case , when @xmath86 , @xmath87 and @xmath88 is also impossible without other degeneracies ( proposition [ neighbmir ] ) .
we then deduce ( corollary [ lnoncoinc ] ) that _ no 4-reflective algebraic planar billiard can have a pair of coinciding neighbor mirrors , _
say @xmath87 .
this is done by contradiction : assuming the contrary , we deform a quadrilateral orbit @xmath46 to a limit quadrilateral with @xmath88 forbidden by proposition [ neighbmir ] .
1c ) case , when @xmath86 _ and it is not an isotropic line , and there are no degeneracies at the neighbor vertices @xmath38 and @xmath39 .
_ we show that _ either @xmath89 , @xmath90 and @xmath45 , or @xmath91 is a cusp in appropriate coordinates .
the _ degree _ of a cusp is its intersection index with a generic line though the singularity ; the degree of a regular germ is one . ] with non - isotropic tangent line _ ( corollary [ connect ] in subsection 2.2 ) .
this follows from proposition [ tang ] and the fact ( proved in the same subsection ) that there exists no one - parameter family of quadrilaterals @xmath92 with @xmath93 .
corollary [ connect ] implies corollary [ a = c ] saying that _ if the mirror @xmath42 is not a line and @xmath44 has no cusps with non - isotropic tangent lines , then @xmath45 .
_ step 2 .
_ proof of theorem [ tclass ] in the case , when some of the mirrors , say @xmath41 is a line ( proposition [ straight ] in subsection 2.3 ) .
_ in the subcase , when some its neighbor mirror , say @xmath42 is not a line , this is done by considering a one - parametric family @xmath94 of degenerate quadrilaterals @xmath92 with @xmath86 .
the above - mentioned corollary [ connect ] together with the reflection law at @xmath38 imply that for every @xmath95 one has @xmath89 , the vertices @xmath34 , @xmath91 are symmetric with respect to the line @xmath41 ( an elementary projective duality argument ) and @xmath45 .
thus , the billiard is of type 1 ) in theorem [ tclass ] . in the subcase , when @xmath41 , @xmath42 and @xmath44 are lines , the above arguments applied to @xmath42 instead of @xmath41 imply that @xmath43 is also a line . the composition of symmetries with respect to the lines @xmath41 , @xmath42 , @xmath43 and @xmath44 acting on the dual projective plane is identity , by 4-reflectivity .
hence , the billiard is of type 2 ) , if the lines are distinct , and of type 1 ) otherwise .
the rest of the proof concerns the case , when no mirror is a line .
_ birationality of neighbor edge correspondence and rationality of mirrors _
( subsection 3.1 ) .
the image of the projection @xmath96 is a two - dimensional projective algebraic variety .
it defines an algebraic correspondence @xmath97
@xmath98 for every @xmath92 .
first we prove ( lemma [ lbirat ] ) that _ @xmath99 is birational , _ by contradiction .
the contrary assumption implies that some of the transformations @xmath100 , say @xmath99 has two holomorphic branches on some open subset @xmath101 : every @xmath102 completes to two distinct 4-periodic orbits @xmath103 .
it follows that the quadrilaterals @xmath104 form a two - parameter family of 4-periodic orbits of the billiard @xmath43 , @xmath44 , @xmath44 , @xmath43 with coinciding neighbor mirrors , and the latter billiard is 4-reflective , a contradiction to corollary [ lnoncoinc ] ( step 1b ) ) .
next we prove that _ the mirrors are rational curves ( corollary [ contrat ] ) . _
the proof is based on the observation that for every isotropic tangency point @xmath105 the transformation @xmath99 contracts the curve @xmath106 to a point @xmath107 with @xmath108 ( follows from the isotropic reflection law , see proposition [ comp - set ] , step 1a ) ) . applying the classical indeterminacy resolution theorem ( @xcite , p.545 of the russian edition ) to the inverse @xmath109 yields that the curve @xmath42 is rational . at the end of subsection 3.1
we deduce corollary [ bircusp ] , which deals with one - parametric family @xmath110 of quadrilaterals @xmath83 with @xmath91 being a _ fixed cusp with isotropic tangent line_. it shows that @xmath111 on @xmath110 and @xmath34 is a _ cusp of the same degree , as @xmath91 .
_ step 4 .
we show that _ all the mirrors have common isotropic tangent lines and each mirror has the so - called _ * property ( i ) : * _ all the isotropic tangencies are of maximal order , i.e. , the intersection of each mirror , say @xmath41 with its isotropic tangent line corresponds to a single point of its normalization @xmath75 _ ( lemma [ commi ] in subsection 3.2 ) . namely , given an isotropic tangency point @xmath105 of the curve @xmath41 , we have to show that the curve @xmath44 intersects the line @xmath112 at a single point .
this is equivalent to the statement saying that for every @xmath113 with @xmath114 the mapping @xmath99 contracts the curve @xmath106 to the point @xmath107 . or in other words , the projection to @xmath115 of the graph of the birational correspondence @xmath99 contains the curve @xmath116 .
the proof of this statement is split into the following substeps .
4a ) proof of the _ local version of the latter inclusion _ ( lemma [ lemgerm ] in subsection 2.6 ) , which deals with two distinct irreducible germs of analytic curves @xmath117 such that the line @xmath112 is isotropic and @xmath118 .
its statement concerns the germ of two - dimensional analytic subset @xmath119 in @xmath120 at the point @xmath121 defined as follows : we say that @xmath122 , if @xmath123 and either @xmath124 , or @xmath125 , or the lines @xmath126 and @xmath127 are symmetric with respect to the tangent line @xmath128 . by definition ,
the germ @xmath119 contains the germ of the curve @xmath129 .
lemma [ lemgerm ] states that under a mild additional condition _ the germ @xmath119 is irreducible_. the most technical part of the paper is the proof of lemma [ lemgerm ] in the case , when the germs @xmath41 and @xmath42 are tangent to each other .
the additional condition from the lemma is imposed only in the case of germs tangent to each other at a finite point and is formulated in terms of their _ puiseaux exponents : _ the germs are represented as graphs of multivalued functions in appropriate coordinates , and we take the lower powers in their puiseaux expansions .
the above additional condition is satisfied automatically , if either the germ @xmath41 is smooth , or its puiseaux exponent is no less than that of the germ @xmath42 .
the proof of lemma [ lemgerm ] for tangent germs is based on proposition [ tang - asympt ] from subsection 2.6 , which deals with the family of tangent lines to @xmath42 .
it relates the asymptotics of the tangency point with that of the intersection points of the tangent line with the curve @xmath41 .
4b ) we cyclically rename the mirrors so that the mirror @xmath41 has a germ tangent to @xmath37 that satisfies the condition of lemma [ lemgerm ] for every branch of the curve @xmath42 at its intersection point with @xmath37
. this will be true , e.g. , if we name @xmath41 the curve having a tangent germ to @xmath37 that is either smooth , or at an infinite point , or with the maximal possible puiseaux exponent .
we show that _ each one of the curves @xmath44 and @xmath42 intersects the line @xmath37 at a unique point .
_ this follows from the irreducibility of the germ @xmath119 ( lemma [ lemgerm ] ) , by elementary topological argument given by proposition [ clemgerm ] .
4c ) we show that _ each one of the curves @xmath75 and @xmath77 intersects the line @xmath37 at a unique point .
_ in the case , when the intersection point @xmath130 is either infinite , or smooth , this follows by applying step 4b ) to the curve @xmath44 instead of @xmath41 .
otherwise , if @xmath91 is a finite cusp , we show that the curves @xmath41 and @xmath43 are conics with focus @xmath91 , by using corollary [ bircusp ] ( step 3 ) and proposition [ char - conic ] ( subsection 2.4 ) .
finally , we show that @xmath45 , by using corollary [ a = c ] ( step 1c ) ) and proving that the curve @xmath44 has no cusps with non - isotropic tangent lines .
the latter statement is proved by applying riemann - hurwitz formula to the projection @xmath131 from appropriate point : the intersection point of two isotropic tangent lines to @xmath41 .
we show that _ the opposite mirrors coincide : @xmath45 , @xmath62 .
_ this follows immediately from corollary [ a = c ] and absence of cusps with non - isotropic tangent lines in rational curves with property ( i ) .
the latter follows from the description of rational curves with property ( i ) having cusps ( corollary [ cnform ] in subsection 2.5 ) .
we prove that _ the mirrors are conics _ ( theorem [ caust - conic ] in subsection 4.1 ) . to do this ,
first we show that they have no cusps ( lemma [ lem - cusp ] ) . assuming the contrary , we show that a mirror with cusps should have two distinct cusps of equal degrees ( basically follows from corollary [ bircusp ] , step 3 ) .
this would contradict the above - mentioned corollary [ cnform ] , which implies that there are at most two cusps and their degrees are distinct .
the rest of the proof of theorem [ caust - conic ] is based on the fact that _ for every @xmath63 with non - isotropic tangent line @xmath71 the collection of the tangent lines to @xmath42 through @xmath38 is symmetric with respect to @xmath71 _ ( corollary [ cortang ] , which follows immediately from corollary [ connect ] , step 2 ) .
we apply corollary [ cortang ] as @xmath38 tends to an isotropic tangency point , and deduce from symmetry that the isotropic tangency should be quadratic . in the case , when @xmath41 and @xmath42 are tangent to some isotropic line at distinct finite points , this is done by elementary local analysis . in the case ,
when @xmath41 and @xmath42 are isotropically tangent to each other , the proof is slightly more technical and uses proposition [ tang - asympt ] , see step 4a ) .
we prove that _ the conics @xmath45 and @xmath62 are confocal _ ( subsection 4.2 ) , by using confocality criterion given by lemma [ propconf ] in subsection 2.4 . if one of them is transverse to the infinity line , then their confocality immediately follows from the lemma and the coincidence of their isotropic tangent lines . in the case ,
when both @xmath41 and @xmath42 are tangent to the infinity line , the proof is slightly more technical and is done by using the above - mentioned corollary [ cortang ] and proposition [ tang - asympt ] .
we fix an euclidean metric on @xmath132 and consider its complexification : the complex - bilinear quadratic form @xmath5 on the complex affine plane @xmath133 .
we denote the infinity line in @xmath29 by @xmath134 . the _ symmetry _ @xmath135 with respect to a non - isotropic complex line
@xmath136 is the unique non - trivial complex - isometric involution fixing the points of the line @xmath37 . for every @xmath137 it acts on the space @xmath138 of lines through @xmath139 , and
this action is called _ symmetry at @xmath139_. if @xmath37 is an isotropic line through a finite point @xmath139 , then a pair of lines through @xmath139 is called symmetric with respect to @xmath37 , if it is a limit of pairs of lines @xmath140 through points @xmath141 such that @xmath142 and @xmath143 are symmetric with respect to non - isotropic lines @xmath144 through @xmath145 converging to @xmath37 .
if @xmath37 is a non - isotropic line , then its symmetry is a projective transformation .
its restriction to the infinity line is a conformal involution .
the latter is conjugated to the above action on @xmath146 via the projective isomorphism @xmath147 sending a line to its intersection point with @xmath148 .
[ lim - refl ] let @xmath37 be an isotropic line through a finite point @xmath139 .
two lines through @xmath139 are symmetric with respect to @xmath37 , if and only if some of them coincides with @xmath37 .
let us introduce an affine coordinate @xmath149 on @xmath148 in which the isotropic points @xmath150 , @xmath151 at infinity be respectively @xmath152 and @xmath153 . as it is shown below , lemma [ lim - refl ]
is implied by the following proposition [ tang - asympt1 ] the symmetry with respect to a finite non - isotropic line through a point @xmath154 acts on @xmath148 by the formula @xmath155 .
let @xmath156 and @xmath157 denote respectively the above line and symmetry .
then @xmath157 acts on @xmath148 by fixing @xmath158 and preserving the isotropic point set @xmath159 , by definition .
it can not fix @xmath152 and @xmath153 , since otherwise , it would fix three distinct points in @xmath148 and hence , would be identity there .
therefore , it would be identity on the whole projective plane , since it also fixes the points of a finite line @xmath156 , while it should be a nontrivial involution , a contradiction .
thus , @xmath160 is a conformal transformation fixing @xmath158 and permuting @xmath152 and @xmath153 .
hence , it sends @xmath149 to @xmath161 .
this proves the proposition .
* of lemma [ lim - refl ] .
* without loss of generality we consider that @xmath162 .
let @xmath163 be an arbitrary sequence of non - isotropic lines converging to @xmath37 , set @xmath164 : @xmath165 .
let @xmath166 be another line through @xmath139 , and let @xmath167 be a sequence of lines converging to @xmath168 .
set @xmath169 , @xmath170 : @xmath171 , @xmath141 .
then the image @xmath172 of the line @xmath167 under the symmetry with respect to @xmath163 is the line through the points @xmath145 and @xmath173 ( proposition [ tang - asympt1 ] ) .
one has @xmath174 .
hence , @xmath175 , as @xmath176 .
this proves the lemma .
the statement of lemma [ lim - refl ] is wrong in the case , when @xmath177 .
indeed , fix a non - isotropic line @xmath7 and consider a sequence of lines @xmath178 orthogonal to @xmath7 .
then for every @xmath179 the line symmetric to @xmath7 with respect to @xmath163 is the line @xmath7 itself , and it does not tend to @xmath37 . on the other hand , the next proposition provides an analogue of lemma [ lim - refl ] in the case , when the lines @xmath180 and @xmath178 are tangent to one and the same irreducible germ of analytic curve . in what follows we deal with irreducible germs of analytic curves that are not lines ; we will call them _ non - linear irreducible germs . _ for every curve @xmath181 and @xmath182 we identify the tangent line @xmath183 with the _ projective tangent line _ to @xmath110 at @xmath184 in @xmath29 .
an irreducible germ @xmath185 of analytic curve has _
isotropic tangency _ at @xmath52 , if the projective line tangent to @xmath110 at @xmath52 is isotropic .
[ plimits ] let @xmath186 be a non - linear irreducible germ of analytic curve in @xmath29 at its isotropic tangency point @xmath52 , set @xmath187 .
let @xmath188 be an arbitrary family of projective lines through @xmath182 that do not accumulate to @xmath37 , as @xmath189 .
let @xmath190 denote their symmetric images with respect to the tangent lines @xmath183 .
then @xmath191 , as @xmath189 .
proposition [ plimits ] together with its next more precise addendum will be proved below .
to state the addendum and in what follows , we will use the next convention and definition .
[ convaz ] fix an affine chart in @xmath29 ( not necessarily the finite plane ) with coordinates @xmath192 and the projective coordinate @xmath193 on its complementary `` new infinity '' line .
the _ complex azimuth _
@xmath194 of a line @xmath37 in the affine chart is the @xmath149-coordinate of its intersection point with the new infinity line .
we define the _ ordered complex angle _ @xmath195 between two complex lines as the difference @xmath196 of their azimuths .
we identify the projective lines @xmath197 by translations for all @xmath198 in the affine chart under consideration .
we fix a round sphere metric on @xmath199 .
[ defpuis ] consider a non - linear irreducible germ @xmath186 of analytic curve at a point @xmath47 and a local chart @xmath192 centered at @xmath52 with the tangent projective line @xmath200 being the @xmath139-axis .
then the curve @xmath110 is the graph of a ( multivalued ) analytic function with puiseaux asymptotics @xmath201 the exponent @xmath202 , which is independent on the choice of coordinates , will be called the _ puiseaux exponent _ of the germ @xmath186 .
* addendum to proposition [ plimits ] .
* _ in proposition [ plimits ] let us measure the angles between lines with respect to an affine chart @xmath192 centered at @xmath52 with @xmath37 being the @xmath139-axis .
let @xmath202 be the puiseaux exponent of the germ @xmath186 .
then the azimuth @xmath203 has one of the following asymptotics : _
case 1 ) : the affine chart is the finite plane , thus , @xmath52 is finite . then @xmath204 case 2 ) : the point @xmath52 is infinite . then @xmath205 @xmath206 @xmath207 in the proof of proposition [ plimits ] and its addendum and in what follows we will use the following elementary fact . [ pxt ]
let @xmath186 be a non - linear irreducible germ at the origin of analytic curve in @xmath4 that is tangent to the @xmath139-axis , and let @xmath202 be its puiseaux exponent .
for every @xmath208 let @xmath209 denote the intersection point of the tangent line @xmath183 with the @xmath139-axis
. then @xmath210 * of proposition [ plimits ] and its addendum . * the statement of proposition [ plimits ] obviously follows from its addendum , since @xmath211 by assumption .
thus , it suffices to prove the addendum .
case 1 ) : the affine chart is finite . without loss of generality we consider that the isotropic line @xmath37 passes through the point @xmath212 . then formula ( [ az - fin ] ) follows from proposition [ tang - asympt1 ] with @xmath213 . case 2 ) : the point @xmath52 is infinite .
let @xmath209 , @xmath214 , @xmath215 denote respectively the points of intersection of the ( true ) infinity line @xmath148 with the lines @xmath183 , @xmath190 , @xmath188 .
subcase 2a ) : @xmath216 .
then @xmath177 .
recall that @xmath37 is the @xmath139-axis , set @xmath217 .
one has @xmath218 the former equality follows from the condition of proposition [ plimits ] : the line @xmath188 through @xmath184 and @xmath219 has azimuth bounded away from below , since it does not accumulate to @xmath37 .
the latter equality follows from ( [ r-1r ] ) .
the @xmath139-coordinates of the points @xmath219 , @xmath209 , @xmath220 form asymptotically an arithmetic progression : @xmath221 .
indeed , the points @xmath209 and @xmath220 collide to @xmath216 , as @xmath189 , and the symmetry with respect to @xmath183 acts by a conformal involution @xmath222 fixing @xmath209 , permuting @xmath219 and @xmath220 and permuting distant isotropic points @xmath223 ( reflection law ) .
the involutions @xmath224 obviously converge to the limit conformal involution @xmath225 fixing @xmath52 and permuting @xmath223 .
this implies the above statement on asymptotic arithmetic progression .
the latter in its turn together with ( [ xptqu ] ) implies that @xmath226 , @xmath227 , as @xmath189 .
finally , the line @xmath190 passes through the points @xmath228 and @xmath229 , hence @xmath230 , @xmath231 .
this proves ( [ az - inf ] ) .
subcase 2b ) : @xmath52 is an isotropic point at infinity , say @xmath232 , and the line @xmath233 is finite .
we work in the above coordinate @xmath149 on the infinity line @xmath148 , in which @xmath52 is the origin .
we choose coordinates @xmath192 so that @xmath148 is the @xmath234-axis and @xmath235 there . here and in the next paragraph
we identify the points @xmath209 , @xmath220 , @xmath219 with their @xmath149-cordinates .
one has @xmath236 by proposition [ tang - asympt1 ] .
one has @xmath237 , by definition and transversality of the lines @xmath37 and @xmath148 ; @xmath238 , since the line @xmath188 does not accumulate to @xmath233 .
hence , @xmath239 , by ( [ pqt ] ) .
this implies that the line @xmath190 through the points @xmath228 and @xmath240 has azimuth of order @xmath241 and proves ( [ az - inf ] ) .
subcase 2c ) : @xmath232 and @xmath177 .
we choose the local coordinates @xmath192 centered at @xmath52 so that @xmath37 is the @xmath139-axis and @xmath242 there .
one has @xmath243 , @xmath244 as in subcase 2a ) . hence ,
@xmath245 , @xmath246 , by ( [ pqt ] ) .
therefore , the line @xmath190 through the points @xmath228 and @xmath247 has azimuth with asymptotics @xmath248 , @xmath249 .
this proves ( [ az - inf ] ) and finishes the proof of proposition [ plimits ] and its addendum
. we will apply proposition [ plimits ] to study limits of periodic orbits in complex billiards . to do this
, we will use the following convention .
[ conv2 ] an irreducible analytic ( algebraic ) curve @xmath250 may have singularities : self - intersections or cusps
. we will denote by @xmath251 its analytic parametrization by an abstract connected riemann surface @xmath252 that is bijective except for self - intersections .
it is usually called _
normalization_. in the case , when @xmath41 is algebraic , the riemann surface @xmath75 is compact .
sometimes we idendify a point ( subset ) in @xmath41 with its preimage in the normalization @xmath252 and denote both subsets by the same symbol . in particular , given a subset in @xmath29 , say a line @xmath7 , we set @xmath253 . if @xmath254 are two curves , and @xmath255 , @xmath256 , @xmath257 , then for simplicity we write @xmath31 and the line @xmath258 will be referred to , as @xmath35 .
[ defbranch ] let @xmath41 be an irreducible analytic curve in @xmath29 , and let @xmath75 be its normalization , see the above convention . for every @xmath105
the _ local branch @xmath259 of the curve @xmath41 at @xmath38 _ is the irreducible germ of analytic curve given by the germ of normalization projection @xmath260 . for every line @xmath261 through @xmath262 the _ intersection index at @xmath105 _ of the curve @xmath75 and the line
@xmath7 is the intersection index of the line @xmath7 with the local branch @xmath259 .
the tangent line @xmath263 will be referred to , as @xmath71 .
let @xmath264 be an analytic ( algebraic ) billiard , @xmath265 be the set of @xmath1-gons corresponding to its periodic orbits .
consider the closure @xmath266 in the usual topology .
we set @xmath267 the set @xmath80 will be called the * @xmath1-reflective set*. [ comp - set ] the sets @xmath266 and @xmath80 are analytic ( algebraic ) , and @xmath80 is the union of the two - dimensional irreducible components of the set @xmath266 .
the billiard is @xmath1-reflective , if and only @xmath80 is non - empty . in this case for every @xmath268 the projection @xmath269 is a submersion on an open dense subset in @xmath80 . in the @xmath1reflective algebraic case
the latter projection is epimorphic and the subset @xmath81 is zariski open and dense . for every @xmath270 and every @xmath268 such that @xmath271 the * complex reflection law * holds : - if the tangent line @xmath272 is not isotropic , then the lines @xmath18 and @xmath19 are symmetric with respect to @xmath273 ; - if @xmath273 is isotropic , then either @xmath18 , or @xmath19 coincides with @xmath273 .
the set @xmath274 is the difference @xmath275 of the two following analytic sets .
the set @xmath276 is locally defined by a system of @xmath1 analytic equations on vertices @xmath20 and @xmath277 , @xmath15 , saying that either the lines @xmath278 and @xmath19 are symmetric with respect to the tangent line @xmath17 , or the line @xmath17 is isotropic , or some of the vertices @xmath277 coincides with @xmath20 .
for example , if a @xmath1-gon @xmath279 has finite vertices @xmath280 , @xmath281 for some @xmath268 , then the corresponding @xmath268-th equation defining @xmath276 in its neighborhood can be written as @xmath282 the set @xmath283 consists of the @xmath1-gons having either some of the latter degeneracies ( isotropic tangency or neighbor vertex collision ) , or a vertex @xmath20 such that @xmath284 .
the set @xmath266 is the union of those irreducible components of the set @xmath276 that intersect @xmath274 .
hence , it is analytic ( algebraic ) .
similarly , the set @xmath80 is analytic ( algebraic ) , and it is the union of two - dimensional irreducible component of the set @xmath266 . the @xmath1-reflectivity criterion and submersivity follow from definition ; the epimorphicity in the algebraic case follows from compactness .
the ( zariski ) openness and density of the subset @xmath81 is obvious .
the reflection law follows from definition and proposition [ plimits ] . in more detail , let @xmath285 be a sequence of @xmath1-periodic orbits converging in @xmath266 .
let for a certain @xmath268 one have @xmath271 and the tangent line @xmath272 be isotropic .
if @xmath286 , then we are done .
otherwise @xmath287 , since the image @xmath288 of the line @xmath289 under the symmetry with respect to @xmath290 tends to @xmath273 , by proposition [ plimits ] .
this proves proposition [ comp - set ] .
here we deal with ( germs of ) analytic @xmath1-reflective planar billiards @xmath9 in @xmath29 : the mirrors are ( germs of ) analytic curves with normalizations @xmath291 , see convention [ conv2 ] ; @xmath292 are neighborhoods of zero in @xmath293 , @xmath15 .
let @xmath294 be the @xmath1-reflective set , see proposition [ comp - set ] , @xmath295 .
the main results of the subsection concern degenerate @xmath1-gons @xmath296 such that for a certain @xmath268 the mirror @xmath21 is not a line , @xmath297 and the latter line is not isotropic .
propositions [ tang ] and [ neighbmir ] show that they can not have types as at fig . 5 and 6a ) .
we deduce the following corollaries for @xmath3 : corollary [ lnoncoinc ] saying that every 4-reflective algebraic billiard has no pair of coinciding neighbor mirrors ; corollary [ connect ] describing the degeneracy at the vertex opposite to tangency ; corollary [ a = c ] giving a mild sufficient condition for the coincidence of opposite mirrors .
a point of a planar analytic curve is _ marked _
, if it is either a cusp , or an isotropic tangency point . given a parametrized curve @xmath251 as above , a point @xmath255 is marked , if it corresponds to a marked point of the local branch @xmath298 , see definition [ defbranch ] .
[ tang ] let @xmath9 and @xmath80 be as above .
then @xmath80 contains no @xmath1-gon @xmath23 with the following properties : - each pair of neighbor vertices correspond to distinct points , and no vertex is a marked point ; - there exists a unique @xmath299 such that the line @xmath300 is tangent to the curve @xmath301 at @xmath302 , and the latter curve is not a line , see fig.5 .
a real version of proposition [ tang ] is contained in @xcite ( lemma 56 , p.315 for @xmath3 , and its generalization ( lemma 67 , p.322 ) for higher @xmath1 ) .
suppose the contrary : there exists a @xmath1-gon @xmath303 as above .
without loss of generality we consider that @xmath304 .
moreover , without loss of generality we can and will assume that the above tangency is quadratic : the quadrilaterals with a tangency vertex @xmath305 form a holomorphic curve in @xmath80 with variable @xmath302 .
for every @xmath15 the reflection with respect to the local branch of the curve @xmath21 at @xmath20 induces a mapping in the space @xmath56 of projective lines .
more precisely , for every @xmath306 it induces a germ of biholomorphic mapping @xmath307 , since the line @xmath278 is transverse to @xmath17 for these @xmath268 . on the other hand ,
the germ @xmath308 is double - valued , with branching locus being the family of lines tangent to @xmath309 .
indeed , the image of a line close to @xmath310 under the reflection from the curve @xmath309 at their intersection point depends on choice of the intersection point .
the latter intersection point is a double - valued function with the above branching locus .
the product @xmath311 should be identity on an open set accumulating to the line @xmath312 , since @xmath23 is a limit of an open set of @xmath1-periodic orbits . but this is impossible , since the product of a biholomorphic germ @xmath313 and a double - valued germ @xmath308 can not be identity .
the proposition is proved .
[ neighbmir ] let @xmath9 and @xmath80 be as at the beginning of the subsection .
then @xmath80 contains no @xmath1-gon @xmath23 with the following properties : \1 ) each its vertex is not a marked point of the corresponding mirror ; \2 ) there exist @xmath314 , @xmath315 such that @xmath316 , @xmath317 , and @xmath41 is not a line ; 3 ) for every @xmath318 one has @xmath319 and the line @xmath18 is not tangent to @xmath21 at @xmath20 , see fig.6a .
the proof of proposition [ neighbmir ] repeats the above proof with some modifications . for simplicity
we give the proof only in the case , when the complement @xmath320 is non - empty . in the opposite case
the proof is analogous and is a straightforward complexification of the arguments from @xcite . without loss of generality
we consider that @xmath321 , then @xmath322 , and we will assume that the mirror @xmath323 has quadratic tangency with @xmath324 , as in the above proof .
consider the germs @xmath325 from the above proof
. set @xmath326 a holomorphic branch of the product @xmath327 following an open set of @xmath1-periodic orbits accumulating to @xmath328 should be identity , as in the above proof .
the germ @xmath329 is biholomorphic .
we claim that the germ @xmath330 is double - valued on a small neighborhood of the line @xmath331 . indeed , a line @xmath7 close to @xmath332 and distinct from the latter intersects the local branch @xmath333 at two distinct points close to @xmath334 .
let us choose one of them and denote it @xmath335 , and denote @xmath336 the other intersection point
. then the ordered pair @xmath337 extends to a unique orbit @xmath338 of length @xmath339 of the billiard on the local branch @xmath333 , see fig.6b .
the line @xmath340 is the image of the line @xmath7 under a branch of the mapping @xmath330 . for different choices of the intersection point
@xmath341 we get different lines @xmath342 .
indeed let us fix the above @xmath341 and complete the above orbit to the orbit @xmath343 on @xmath333 of length @xmath344 .
then the output line @xmath342 corresponding to the other intersection point @xmath336 is the line @xmath345 , by definition .
it is distinct from the line @xmath346 for an open set of lines @xmath7 , analogously to arguments from @xcite .
this together with the double - valuedness of the intersection point @xmath347 proves the double - valuedness of the above germ @xmath330 .
each @xmath1-periodic orbit @xmath348 corresponding to a point in @xmath349 close to @xmath23 contains a sub - orbit @xmath350 on @xmath333 as above .
this implies that the product of the above biholomorphic germ @xmath329 and double - valued germ @xmath330 should be identity , a contradiction .
this proves the proposition .
[ lnoncoinc ] there are no 4-reflective planar algebraic billiards with a pair of coinciding neighbor mirrors .
suppose the contrary : there exists a 4-reflective planar algebraic billiard @xmath41 , @xmath42 , @xmath42 , @xmath44
. then @xmath42 can not be a line , since otherwise , there would exist no 4-periodic orbit @xmath46 : by definition [ deforb ] , the lines @xmath351 and @xmath352 should be symmetric with respect to the non - isotropic line @xmath42 , which is impossible .
let @xmath353 denote the 4-reflective set .
it is two - dimensional and contains at least one irreducible algebraic curve @xmath110 consisting of quadrilaterals @xmath354 with coinciding variable vertices @xmath88 , by epimorphicity of the projection @xmath355 ( proposition [ comp - set ] ) .
let us fix the above @xmath110 .
there are three possible cases : case 1 ) : @xmath356 , @xmath357 on @xmath110 .
then @xmath358 , since the set of lines tangent to two algebraic curves at distinct points is finite .
this implies that @xmath359 and a generic quadrilateral @xmath360 represents a counterexample to proposition [ neighbmir ] with @xmath361 , @xmath362 .
case 2 ) : @xmath363 , @xmath357 on @xmath110 .
then we get a contradiction to proposition [ neighbmir ] with @xmath364 , @xmath362 .
the case @xmath356 , @xmath365 is symmetric . case 3 ) : @xmath366 on @xmath110 .
then we get a contradiction to proposition [ neighbmir ] with @xmath364 , @xmath367 .
corollary [ lnoncoinc ] is proved .
[ connect ] let @xmath41 , @xmath42 , @xmath43 , @xmath44 be a 4-reflective analytic planar billiard , and let @xmath42 be not a line .
let @xmath368 be the 4-reflective set .
let @xmath92 be such that @xmath31 , @xmath32 , the line @xmath369 is tangent to the curve @xmath42 at @xmath34 and is not isotropic .
then - either @xmath370 is tangent to the curve @xmath44 at @xmath91 , @xmath371 , @xmath45 and the corresponding local branches coincide , i.e. , @xmath372 ( see convention [ conv2 ] ) : `` opposite tangency connection '' , see fig.7a ; - or @xmath373 is a cusp of the branch @xmath374 and the tangent line @xmath375 is not isotropic : `` tangency cusp connection '' , see fig.7b . let @xmath376 denote an irreducible germ at @xmath46 of analytic curve consisting of quadrilaterals @xmath377 such that @xmath378 . then @xmath379 on @xmath94 , hence @xmath380 , @xmath381 , @xmath382 vary along the curve @xmath94 . without loss of generality
we consider that \(i ) the line @xmath369 has quadratic tangency with the curve @xmath42 at @xmath34 and is transverse to @xmath41 and @xmath43 at @xmath38 and @xmath39 respectively ; \(ii ) the points @xmath38 , @xmath34 , @xmath39 are finite and not marked ; \(iii ) @xmath383 and @xmath384 ; \(iv ) the germ at @xmath46 of the projection @xmath385 is open .
one can achieve this by deforming the quadrilateral @xmath46 along the curve @xmath94 .
indeed , it is easy to achieve conditions ( i ) and ( ii ) .
condition ( iv ) can be achieved , since the projection @xmath386 is open at a generic point of the curve @xmath94 .
indeed , it is a submersion on an open and dense subset in @xmath80 ( proposition [ comp - set ] ) .
hence , it is open outside at most countable union of curves contracted to points by @xmath386 ( if any ) , while @xmath110 is not contracted .
condition ( iii ) can be achieved , since @xmath387 and @xmath388 on @xmath94 .
indeed , if , e.g. , @xmath389 , then the one - parameter family of lines @xmath126 would be tangent to both curves @xmath41 and @xmath42 at distinct points @xmath390 , @xmath391 , which is impossible . then either @xmath392 is a marked point ( cusp or isotropic tangency ) that is constant on the curve @xmath94 , or the line @xmath393 is tangent to @xmath44 at @xmath394 for every @xmath395 .
this follows from conditions ( i)(iv ) , proposition [ tang ] and the discreteness of the set of marked points in @xmath396 .
case 1 ) : @xmath392 is a marked point that is constant on @xmath94 .
let @xmath91 be an isotropic tangency point .
then one of the lines @xmath397 or @xmath398 identically coincides with the isotropic tangent line @xmath375 ( proposition [ comp - set ] ) .
hence , either @xmath380 , or @xmath382 is constant on the curve @xmath94 , which is impossible , see the beginning of the proof .
thus , @xmath373 is a cusp and the tangent line @xmath375 is not isotropic .
case 2 ) : @xmath91 is not a marked point , the line @xmath370 is tangent to @xmath44 at @xmath91 , and this holds in a neighborhood of the quadrilateral @xmath46 in the curve @xmath94 . in the case , when @xmath399 on @xmath94 , one has @xmath400 , and we are done .
let us show that the opposite case is impossible .
suppose the contrary .
then deforming the quadrilateral @xmath46 along the curve @xmath94 , one can achieve that in addition to conditions ( i)(iv ) , one has @xmath401 .
thus , the line @xmath402 is identically tangent to both curves @xmath42 and @xmath44 at @xmath381 and @xmath394 respectively along the curve @xmath94 .
therefore , @xmath403 , @xmath62 and the tangent line @xmath126 to @xmath42 is orthogonal to both @xmath128 and @xmath404 identically on @xmath94 ( reflection law ) .
let us fix a local parameter @xmath184 on the curve @xmath42 in a neighborhood of the point @xmath34 .
consider a quadrilateral @xmath405 close to @xmath46 . then @xmath406 and @xmath407 .
let @xmath168 ( @xmath408 ) denote the line through @xmath409 ( @xmath410 orthogonal to @xmath411 ( respectively , @xmath412 ) .
these lines are tangent to @xmath42 at some points @xmath413 and @xmath414 respectively .
set @xmath415 . without loss of generality
we consider that the point @xmath373 is finite , since @xmath416 varies along the curve @xmath94 .
we measure angles between lines in the finite affine chart @xmath4 .
we show that the angle between the tangent lines @xmath417 and @xmath418 is of order @xmath158 on one hand , and of order @xmath419 on the other hand , as @xmath420 .
the contradiction thus obtained will prove the corollary .
we identify a point of the curve @xmath42 ( or its normalization @xmath76 ) with its @xmath184-coordinate .
the angle between the lines @xmath168 and @xmath421 is of order @xmath422 ( quadraticity of tangency ) .
this and analogous statement for @xmath423 together with the reflection law at @xmath409 , @xmath423 and @xmath424 imply the following asymptotics , as @xmath420 : \a ) @xmath425 , @xmath426 ; \b ) @xmath427 , @xmath428 .
let us prove statement b ) in more detail .
the lines @xmath421 and @xmath429 are symmetric with respect to the line @xmath168 , hence @xmath430 .
they intersect the local branch @xmath431 at the points @xmath432 and @xmath424 respectively .
this together with the quadraticity implies that @xmath433 . the latter implies statement b ) with appropriate constant @xmath428 .
the angle @xmath434 is an order @xmath158 quantity , by statement b ) and quadraticity . on the other hand , it is the angle between the symmetry lines for the pairs of lines @xmath435 and @xmath436 respectively .
this together with statement a ) , which says that the latter pairs are `` @xmath419-close '' , implies that the above angle should be also of order @xmath419 .
the contradiction thus obtained proves the corollary .
[ a = c ] let in a 4-reflective complex algebraic planar billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 the mirror @xmath42 be not a line and @xmath44 have no cusps with non - isotropic tangent lines
. then @xmath45 .
there exists an irreducible algebraic curve @xmath376 consisting of those quadrilaterals @xmath46 with variable @xmath34 for which @xmath86 : every 4-periodic billiard orbit can be deformed without changing the vertex @xmath34 to such a quadrilateral .
let us fix an @xmath95 with @xmath84 and non - isotropic tangent line @xmath85 .
its vertex @xmath91 is not a cusp with non - isotropic tangent line , by assumption .
therefore , @xmath45 , by corollary [ connect ] . here
we apply the results of the previous subsection to classify the 4-reflective algebraic billiards with at least one mirror being a line .
[ straight ] let @xmath41 , @xmath42 , @xmath43 , @xmath44 be a 4-reflective algebraic billiard in @xmath29 such that @xmath41 is a line . if some of the mirrors @xmath42 or @xmath44 is not a line , then @xmath45 and the curves @xmath42 , @xmath44 are symmetric with respect to the line @xmath41 , see fig.1 . if @xmath41 , @xmath42 , @xmath44 are lines , then @xmath43 is also a line , the lines @xmath41 , @xmath42 , @xmath43 , @xmath44 pass through the same point , and the line pairs @xmath48 , @xmath49 are sent one into the other by complex isometry with unit jacobian , see fig.2 .
we already know that @xmath437 , @xmath438 , by corollary [ lnoncoinc ] .
let us first consider the case , when one of the mirrors @xmath42 , @xmath44 , say @xmath42 is not a line .
let @xmath80 be the 4-reflective set , and let @xmath376 be an irreducible algebraic curve as in the proof of corollary [ connect ] : it consists of those quadrilaterals @xmath46 with variable @xmath34 , for which the line @xmath35 is tangent to @xmath42 at @xmath34 . for every @xmath439 such that @xmath34 is not a marked point and @xmath440 either the point @xmath373 is a cusp of the branch @xmath374 ( the same for all @xmath439 ) , or the line @xmath73 is tangent to @xmath44 at @xmath91 .
this follows from corollary [ connect ] .
the first , cusp case is impossible , by a projective duality argument . indeed ,
if @xmath91 were constant along the curve @xmath94 , then the lines @xmath35 with variable @xmath34 would intersect at one and the same point @xmath441 symmetric to @xmath373 with respect to the line @xmath41 ( the reflection law ) . on the other hand , for every @xmath439 and any other @xmath442 close to it the intersection point @xmath443 tends to @xmath444 , as @xmath445 .
therefore , @xmath446 , hence @xmath447 , as @xmath46 ranges in @xmath94 , a contradiction .
thus , for every @xmath95 the line @xmath73 is tangent to @xmath44 at the point @xmath91 .
finally , the family of tangent lines @xmath35 to @xmath42 is symmetric to the family of tangent lines @xmath73 to @xmath44 with respect to the line @xmath41 .
this implies that the curves @xmath42 and @xmath44 are also symmetric : the above argument shows that the intersection points @xmath448 and @xmath449 should be symmetric and tend to @xmath444 and @xmath373 respectively .
one has @xmath450 on @xmath94 , hence @xmath45 ( corollary [ connect ] ) .
the first statement of proposition [ straight ] is proved .
now let us consider the case , when @xmath41 , @xmath42 and @xmath44 are lines .
let us prove the second statement of the proposition .
if @xmath43 were not a line , then @xmath41 would also havent been a line , being symmetric to @xmath43 with respect to the line @xmath42 ( the first statement of the proposition ) , a contradiction to our assumption .
therefore , @xmath43 is a line .
the composition of reflections from the lines @xmath41 , @xmath42 , @xmath43 , @xmath44 is identity as a transformation of the space @xmath56 of projective lines ( 4-reflectivity ) .
this together with the last statement of example [ ex - lines2 ] proves the second statement of the proposition . let us prove that every 4-reflective billiard with at least one straight mirror is of one of the types 1 ) or 2 ) .
if each mirror is a line and some of them coincide , then the billiard is of type 1 ) . indeed ,
in this case the coinciding mirrors are opposite ( corollary [ lnoncoinc ] ) , say @xmath45 , and @xmath42 , @xmath44 are symmetric with respect to the line @xmath41 , by the isometry of the pairs @xmath48 and @xmath49 .
otherwise , a billiard with a straight mirror is of type either 1 ) , or 2 ) , by proposition [ straight ] .
this proves theorem [ tclass ] in the case of straight mirror .
here we recall the classical notions of confocality and foci for complex conics .
we extend urquhart s theorem [ tell ] and the characterization of ellipse as a curve with two given foci to complex conics ( theorem [ tquad ] and proposition [ char - conic ] respectively ) .
afterwards we state and prove lemma [ propconf ] characterizing pairs of confocal complex conics in terms of their isotropic tangent lines .
lemma [ propconf ] and its proof are based on the classical relations between foci and isotropic tangent lines ( propositions [ prklein ] and corollary [ same - lines ] ) .
let @xmath451 ( @xmath452 ) denote the space of all the conics in @xmath453 ( @xmath29 ) including degenerate ones : couples of lines .
this is a 5-dimensional real ( complex ) projective space .
one has complexification inclusion @xmath454 .
let @xmath455 denote the set of smooth ( non - degenerate ) complex conics .
consider the subset @xmath456 of pairs of confocal ellipses .
let @xmath457 denote the minimal complex projective algebraic set containing @xmath458
. the projective algebraic set @xmath459 is irreducible and has complex dimension 6 .
it is symmetric , as is @xmath458 .
the subset @xmath460 is zariski open and dense .
these statements follow from definition .
[ defconf ] two smooth planar complex conics are _ confocal , _ if their pair is contained in @xmath459 .
[ tquad ] for every pair of distinct complex confocal conics @xmath41 and @xmath42 the complex billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 is 4-reflective .
consider the fibration @xmath461 , @xmath462 : the @xmath463-fiber over a pair @xmath464 is the product @xmath465 .
let @xmath466 denote the set of pairs @xmath467 such that @xmath46 is an interior point of the 4-periodic orbit set of the billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 .
set @xmath468 for the proof of theorem [ tquad ] it suffices to show that @xmath469 .
the set @xmath470 is a difference of two analytic subsets in @xmath463 , as in the proof of proposition [ comp - set ] .
the set @xmath471 is constructible , by the latter statement and remmert s proper mapping theorem ( see , ( * ? ? ?
* of the russian edition ) ) .
the set @xmath471 contains a zariski dense subset @xmath472 , by urquhart s theorem [ tell ] .
hence , @xmath471 contains a zariski open and dense subset in @xmath473 .
now it suffices to show that the subset @xmath474 is closed in the usual topology .
that is , fix an arbitrary sequence @xmath475 converging to a pair of smooth distinct confocal conics @xmath48 , and let us show that the billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 is 4-reflective . to do this , fix an arbitrary pair @xmath476 that satisfies the following genericity conditions : @xmath31 ; @xmath38 and @xmath34 are finite and not isotropic tangency points of the corresponding conics ; the pair of lines symmetric to @xmath35 with respect to the lines @xmath71 and @xmath85 intersect the union @xmath477 at eight distinct points that are not isotropic tangency points .
pairs @xmath478 satisfying the latter conditions exist and form a zariski open subset in @xmath479 , since @xmath480 .
the pair @xmath478 is a limit of pairs @xmath481 extendable to periodic orbits @xmath482 of the billiard @xmath483 , @xmath484 , @xmath483 , @xmath484 , since the latter pairs form a zariski open dense subset in @xmath485 , by proposition [ comp - set ] . after passing to a subsequence
, the above orbits converge to a 4-periodic orbit @xmath46 of the billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 , by construction and genericity assumption .
thus , each pair @xmath478 from a zariski open subset in @xmath479 extends to a quadrilateral orbit , and hence , the billiard is 4-reflective .
theorem [ tquad ] is proved . in the above argument the assumption that @xmath480 is important .
otherwise , a priori it may happen that while passing to the limit , some neighbor vertices @xmath486 and @xmath487 of a 4-periodic orbit collide , and in the limit we get a degenerate 4-periodic orbit with coinciding neighbor vertices . [ prklein ] ( @xcite , ( * ? ? ?
* subsection 17.4.3.1 , p.334 ) , goes back to laguerre ) for every smooth real conic ( ellipse , hyperbola , parabola ) each its focus lies in two transverse isotropic tangent lines to the complexified conic .
[ same - lines ] @xcite every two complexified confocal real planar conics have the same isotropic tangent lines : a pair of transverse isotropic lines through each focus ( with multiplicities , see the next remark ) . for every conic @xmath488 and @xmath489
there are two distinct tangent lines to @xmath41 through @xmath39 .
but if @xmath490 , then @xmath491 is the unique tangent line through @xmath39 .
we count it twice , since it is the limit of two confluenting tangent lines through @xmath492 , as @xmath493 .
if @xmath490 is an isotropic point at infinity , then @xmath491 is a _
double isotropic tangent line to @xmath41 . _ every smooth complex conic has four isotropic tangent lines with multiplicities .
the _ complex focus _ of a smooth complex conic is an intersection point of some its two distinct isotropic tangent lines .
[ char - conic ] let @xmath494 be an unordered pair of points that does not coincide with the pair of isotropic points at infinity .
let @xmath250 be a parametrized analytic curve distinct from an isotropic line such that for every @xmath63 the lines @xmath495 and @xmath496 are symmetric with respect to the line @xmath71 .
then the curve @xmath41 is either a conic with foci @xmath497 and @xmath498 , or a line with @xmath497 and @xmath498 being symmetric with respect to @xmath41 .
none of the points @xmath497 and @xmath498 is an isotropic point at infinity . indeed ,
if @xmath499 , then @xmath500 , by symmetry , a contradiction to the assumption that @xmath501 . the curve @xmath41 is a phase curve of the following double - valued singular algebraic line field @xmath502 on @xmath29 : for every @xmath503 the lines @xmath495 and @xmath496 are symmetric with respect to the line @xmath504 .
the singular set of the latter field is the union of the isotropic lines through @xmath497 and those through @xmath498 .
each its phase curve is either a conic with foci @xmath497 and @xmath498 , or the symmetry line of the pair @xmath505 .
this follows from the same statement for real @xmath497 and @xmath498 in the real plane , which is classical , and by analyticity in @xmath505 of the line field family @xmath502 .
the proposition is proved .
a _ transverse hyperbola _ is a smooth complex conic in @xmath29 transverse to the infinity line .
generic hyperbola _ is a smooth complex conic that has four distinct isotropic tangent lines .
the complexification of a real conic @xmath41 is a generic hyperbola , if and only if @xmath41 is either an ellipse with distinct foci , or a hyperbola .
the complexification of a circle is a non - generic transverse hyperbola through both isotropic points @xmath212 , @xmath506 , with two double isotropic tangent lines at them intersecting at its center .
each generic hyperbola is a transverse one .
conversely , a transverse hyperbola is a generic one , if and only if it contains no isotropic points at infinity .
a conic confocal to a transverse ( generic ) hyperbola is also a transverse ( generic ) hyperbola . _ a generic hyperbola ( a complexified ellipse or hyperbola with distinct foci )
_ has _ four distinct finite complex foci _ ( including the two real ones ) .
[ propconf ] two smooth conics @xmath41 and @xmath42 are confocal , if and only if one of the following cases takes place : \1 ) @xmath41 and @xmath42 are transverse hyperbolas with common isotropic tangent lines ; \2 ) `` non - isotropic parabolas '' : @xmath41 and @xmath42 are tangent to the infinity line at a common non - isotropic point , and their finite isotropic tangent lines coincide ; \3 ) `` isotropic parabolas '' : the conics @xmath41 and @xmath42 are tangent to the infinity line at a common isotropic point , have a common finite isotropic tangent line and are obtained one from the other by translation by a vector parallel to the latter finite isotropic tangent line .
the first step in the proof of lemma [ propconf ] is the following proposition .
[ isofoci ] let @xmath41 be a generic hyperbola .
a smooth conic @xmath42 is confocal to @xmath41 , if and only if @xmath41 and @xmath42 have common isotropic tangent lines .
let @xmath507 denote the subset of generic hyperbolas .
set @xmath508 let @xmath509 denote the subset of pairs of conics having common isotropic tangent lines .
these are quasiprojective algebraic varieties .
we have to show that @xmath510 .
indeed , one has @xmath511 , by corollary [ same - lines ] and minimality of the set @xmath459 . for every quadruple @xmath498 of distinct isotropic lines , two through each isotropic point at infinity , let @xmath512 denote the space of smooth conics tangent to the collection @xmath498 . in other terms , the conics of the space @xmath513 are dual to the conics passing through the given four points dual to the lines in @xmath498 in the dual projective plane .
no triple of the latter four points lies in the same line .
this implies that the space @xmath513 is conformally - equivalent to punctured projective line .
the space @xmath514 is holomorphically fibered over the four - dimensional space of the above quadruples @xmath498 with fibers @xmath515 .
this implies that @xmath514 is a 6-dimensional irreducible quasiprojective variety containing another 6-dimensional irreducible quasiprojective variety @xmath516 .
the latter is a closed subset in the usual topology of the ambient set @xmath514 , by definition .
hence , both varieties coincide .
the proposition is proved .
* of lemma [ propconf ] .
* we will call a pair of conics _ tangentially confocal , _ if they satisfy one of the above statements 1)3 ) .
first we show that every pair @xmath48 of confocal conics is tangentially confocal .
then we prove the converse . in the proof of the lemma we use the fact that _ every pair of confocal conics is a limit of pairs of confocal generic hyperbolas _ , since the latter pairs form a zariski open and dense subset in @xmath459 .
this together with proposition [ isofoci ] implies that _ every two confocal conics have common isotropic tangent lines ( with multiplicities)_. for the proof of lemma [ propconf ] we translate the tangential confocality into the dual language .
let @xmath517 be the dual lines to the isotropic points at infinity , set @xmath518 .
a pair of smooth conics @xmath41 and @xmath42 are confocal generic hyperbolas , if and only if the dual curves @xmath519 and @xmath520 pass through the same four distinct points @xmath521 , @xmath522 ( proposition [ isofoci ] ) .
now let @xmath48 be a limit of pairs @xmath523 of confocal generic hyperbolas , let @xmath524 denote the above points of intersection @xmath525 , and let @xmath41 be not a generic hyperbola .
let us show that one has some of cases 1)3 ) . passing to a subsequence , we consider that the points @xmath526 converge to some limits @xmath527 .
then one of the following holds : ( i ) one has @xmath528 for all @xmath529 .
then @xmath530 , hence @xmath41 and @xmath42 are transverse hyperbolas , and we have case 1 ) .
\(ii ) one has @xmath531 , @xmath532 , and the conics @xmath519 and @xmath520 are tangent to each other at the limit @xmath533 of colliding intersection points @xmath534 .
this implies that the curves @xmath41 and @xmath42 are as in case 2 ) .
\(iii ) one has @xmath535 ( up to permuting @xmath212 and @xmath506 ) , and the conics @xmath519 and @xmath520 are tangent to each other with triple contact at the limit @xmath533 of three colliding intersection points @xmath536 .
the corresponding tangent line coincides with @xmath537 , since the points @xmath538 collide .
therefore , the conics @xmath41 and @xmath42 are tangent to the infinity line at @xmath506 and have triple contact there between them , and they have a common finite isotropic tangent line @xmath539 through @xmath212 .
below we show that then we have case 3 ) .
( iv ) all the points @xmath540 coincide with @xmath533
. then a smooth conic @xmath519 should be tangent at @xmath533 to both transverse lines @xmath541 and @xmath537 , analogously to the above discussion , a contradiction .
hence , this case is impossible .
[ ptransl ] let two distinct smooth conics @xmath254 be tangent to the infinity line @xmath148 at a common point
. then they have at least triple contact there , if and only if they are obtained from each other by translation .
one direction is obvious : if conics @xmath41 and @xmath42 tangent to @xmath148 are translation images of each other , then they have common tangency point with @xmath148 and at least triple contact there between them .
the latter follows from the elementary fact that in this case they may have at most one finite intersection point , which is a solution of a linear equation .
let us prove the converse .
suppose they are tangent to the infinity line at a common point and have at least triple contact there .
let @xmath63 and @xmath64 be arbitrary two finite points with parallel tangent lines .
the translation by the vector @xmath35 sends the curve @xmath41 to a curve @xmath542 tangent to @xmath42 at @xmath34 that has at least triple contact with @xmath42 at infinity . if @xmath543 , then their intersection index is at least 5 , a contradiction .
hence , @xmath544 .
this proves the proposition .
thus , in case ( iii ) @xmath41 and @xmath42 are translation images of each other , by proposition [ ptransl ] , and have a common finite isotropic tangent line ( hence , parallel to the translation vector ) .
therefore , we have case 3 ) . for every smooth conic @xmath41 let @xmath545 ( @xmath546 ) denote respectively the space of smooth conics confocal ( tangentially confocal ) to @xmath41 .
these are quasiprojective varieties .
we have shown above that @xmath547 , and for the proof of the lemma it suffices to show that @xmath548 . in the case ,
when @xmath41 is a generic hyperbola , this follows from proposition [ isofoci ] .
note that @xmath549 , since this is true for a zariski open dense subset in @xmath550 of generic hyperbolas @xmath41 ( see proposition [ isofoci ] and its proof ) and remains valid while passing to limits .
moreover , the subset @xmath551 is closed by definition . fix a smooth conic @xmath41 that is not a generic hyperbola .
let us show that @xmath546 is a punctured riemann sphere .
this together with the inclusion @xmath547 and closeness will imply that @xmath548 and prove the lemma .
we will treat separately each one of cases 1)3 ) ( or an equivalent dual case ( i)(iii ) ) .
case 3 ) is obvious : the space @xmath546 of images of the conic @xmath41 by translations parallel to a given line is obviously conformally equivalent to @xmath293 .
let us treat case 2)=(ii ) .
in this case the dual curve @xmath519 intersects the union @xmath552 at exactly three distinct points : @xmath553 , @xmath554 and @xmath555 .
the tangent line @xmath556 is transverse to the lines @xmath557 , @xmath558 , since @xmath519 is a conic intersecting each line @xmath559 at two distinct points .
the conics tangentially confocal to @xmath41 are dual to exactly those conics @xmath520 that pass through the points @xmath560 , @xmath561 , @xmath533 and are tangent to the line @xmath562 at @xmath533 . the latter three points and line being in generic position , the space of conics @xmath520 respecting them as above is a punctured projective line . in case ( i ) the proof is analogous and is omitted to save the space .
lemma [ propconf ] is proved .
[ tan - con ] let two confocal conics @xmath41 and @xmath42 be tangent to each other .
then each their tangency point lies on the infinite line , the corresponding tangent line is isotropic , and one of the following cases holds : \(i ) single tangency point of quadratic contact ; either the tangency point is isotropic and the tangent line is finite ; or it is non - isotropic , and the tangent line is infinite ; \(ii ) two tangency points , which are the two isotropic points at infinity ; the tangent lines are finite ; \(iii ) single tangency point of triple contact : an isotropic point at infinity , the tangent line is infinite .
let @xmath41 and @xmath42 be tangent confocal conics .
all their common tangent lines are isotropic , since this is true for generic hyperbolas and remains valid after passing to limit .
case 1 ) of the lemma corresponds to cases ( i ) ( first subcase ) or ( ii ) of the corollary .
case 2 ) of the lemma corresponds to case ( i ) , second subcase .
case 3 ) of the lemma corresponds to case ( iii ) of the corollary .
these statements follow from the proof of the lemma ( the arguments on the points @xmath540 of intersection of the dual conics ) and the fact that a tangency of two curves corresponds to a tangency of the dual curves .
for example , in case 1 ) ( or equivalently , case ( i ) from the proof of the lemma ) a tangency point @xmath52 of the conics @xmath41 and @xmath42 corresponds to a common tangency point of the dual conics @xmath519 and @xmath520 with a line @xmath559 , @xmath558 .
this implies that @xmath563 .
the other cases are treated analogously . in this subsection
we describe the class of special rational curves having property ( i ) introduced below ( proposition [ pisotr ] and corollary [ cnform ] ) .
we show in subsection 3.2 that mirrors of every 4-reflective algebraic billiard without lines belong to this class .
these results will be used in section 4 .
[ defi ] we say that a planar projective algebraic curve @xmath41 that is not a line _ has property ( i ) _ , if every its isotropic tangent line intersects its normalization @xmath75 at a single point @xmath38 , see convention [ conv2 ] ; the intersection index of the curve @xmath75 with @xmath71 at @xmath38 ( see definition [ defbranch ] ) then equals the degree of the curve @xmath41 .
every conic has property ( i ) .
corollary [ cnform ] below shows that the converse is not true .
a curve @xmath41 that is not a line has property ( i ) , if and only if its dual @xmath519 satisfies the following statement : ( @xmath564 ) for every @xmath558 and @xmath565 the germ @xmath566 is irreducible , the line @xmath567 is the only line through @xmath184 tangent to @xmath519 , and @xmath184 is their unique tangency point . each planar projective curve with property ( i )
has at least two distinct isotropic tangent lines .
let @xmath41 be a property ( i ) curve .
its isotropic tangent lines are dual to the points of non - empty intersection @xmath568 .
hence , the contrary to the corollary would imply that this intersection reduces to @xmath569 .
the germ @xmath570 is irreducible ( the above statement ( @xmath564 ) ) , and hence is not tangent at @xmath571 , say , to the line @xmath541 .
this implies that @xmath519 should intersect @xmath541 at some other point .
the contradiction thus obtained proves the corollary .
[ pisotr ] let a rational curve @xmath41 have property ( i ) .
then \(i ) either @xmath41 has at least three distinct isotropic tangency points : then it has no cusps , and at least one its isotropic tangency point is finite ; \(ii ) or it has exactly two distinct isotropic tangency points ; then at least one of them is an isotropic point at infinity , and @xmath41 has no cusps except maybe for some of the two latter points .
the curve @xmath41 has at least two distinct isotropic tangent lines , by the above corollary .
the tangency points should be distinct , since the contrary would obviously contradict property ( i ) .
case 1 ) : @xmath41 has at least three distinct isotropic tangency points @xmath38 , @xmath34 , @xmath39 .
at least one of them is finite .
indeed , otherwise @xmath572 , and some of them , say @xmath38 is not isotropic .
hence , the curve @xmath41 is tangent to the infinity line at @xmath38 and intersects it at @xmath573 , a contradiction to property ( i ) .
fix arbitrary two isotropic tangency points , say @xmath38 and @xmath34 .
we show that the curve @xmath41 has no cusps distinct from them . applying this to the other pairs @xmath574 and
@xmath575 will imply that @xmath41 has no cusps at all and will prove ( i ) .
let @xmath576 and @xmath577 denote the projective lines tangent to @xmath41 at @xmath38 and @xmath34 respectively .
they intersect @xmath41 only at @xmath38 ( respectively , @xmath34 ) , by property ( i ) .
this implies that @xmath578 and @xmath579 .
consider the projection @xmath580 : the composition of the parametrization @xmath581 and the projection from the point @xmath52 .
its global degree and its local degrees at its critical points corresponding to @xmath38 and @xmath34 are equal to the degree of the curve @xmath41 ( property ( i ) ) .
these are the only critical points , since they have maximal order and @xmath41 is rational .
hence , @xmath41 has no cusps distinct from @xmath38 and @xmath34 .
this together with the above discussion proves ( i ) .
case 2 ) : @xmath41 has exactly two isotropic tangency points @xmath38 and @xmath34 .
let us prove ( ii ) .
as is shown above , @xmath41 has no cusps distinct from @xmath38 and @xmath34 .
the dual curve @xmath519 intersects the union @xmath552 exactly at two distinct points @xmath582 and @xmath583 .
thus , @xmath519 intersects one of the lines @xmath559 , say @xmath541 at a unique point @xmath184 .
then @xmath519 is tangent to @xmath541 at @xmath184 ( irreducibility of the germ @xmath566 , by ( @xmath564 ) ) .
this implies that @xmath584 and proves ( ii ) and the proposition .
a system of _ isotropic coordinates _ on @xmath4 is a system of affine coordinates with isotropic axes .
[ cnform ] in case ( ii ) of proposition [ pisotr ] one of the following holds ( here @xmath44 is the degree of the curve @xmath41 ) : - either the curve @xmath41 is tangent to @xmath148 at an isotropic point at infinity and has another finite isotropic tangency point ; then in appropriate isotropic coordinates the curve @xmath41 is given by the following parametrization : @xmath585 - or the curve @xmath41 passes through the two isotropic points at infinity and in appropriate isotropic coordinates it is given by the following parametrization : @xmath586 in both formulas @xmath533 and @xmath44 are relatively prime . in particular , the curve @xmath41 is without cusps , if and only if it is a conic and in the above formulas @xmath587 , @xmath588 and @xmath589 respectively . recall that the curve @xmath519 intersects the union @xmath552 exactly at two distinct points , and one of the intersections @xmath590 is a single point , see the end of the above proof . at each point of intersection
@xmath590 the germ of the curve @xmath519 is irreducible , see @xmath591 .
therefore , we have the following possibilities ( up to permuting @xmath541 and @xmath537 ) : case 1 ) : @xmath519 passes through @xmath592 , @xmath593 , then @xmath519 is tangent to @xmath541 at @xmath571 ( statement ( @xmath564 ) ) , and @xmath519 intersects @xmath537 at a unique point @xmath184 different from @xmath571
. then @xmath41 is tangent to the infinity line at @xmath212 and has a finite isotropic tangent line @xmath594 through @xmath506 .
the line @xmath567 does not contain @xmath571 , since otherwise , @xmath595 and @xmath541 would be two distinct tangent lines to @xmath519 through @xmath571 , a contradiction to property @xmath591 . hence , the dual @xmath596 to the line @xmath567 is a finite point , and it lies in @xmath597 by duality .
the composition of the parametrization @xmath581 with the projection @xmath598 from the point @xmath212 is a branched covering @xmath580 .
either it is bijective , or it has exactly two critical points @xmath52 and @xmath212 , since @xmath41 has neither cusps distinct from them , nor finite tangent lines through @xmath212 . therefore ,
taking @xmath594 as the @xmath139-axis , @xmath52 as the origin and @xmath599 as the @xmath234-axis , we get ( [ form - pd ] ) after appropriate coordinate rescalings .
case 2 ) : @xmath519 intersects each line @xmath559 at a unique point @xmath600 and is tangent to @xmath559 there ( by statement ( @xmath564 ) ) , and @xmath601 .
hence , @xmath41 passes through both isotropic points transversely to the infinity line .
taking isotropic coordinates centered at the intersection @xmath602 , we get ( [ form - pq ] ) after appropriate rescalings .
the parametrizations in ( [ form - pd ] ) and ( [ form - pq ] ) can be chosen bijective ; then @xmath533 and @xmath44 are relatively prime .
hence , if @xmath603 , the curve @xmath41 has at least one cusp : either at an isotropic point at infinity , or at the origin ( the latter may take place only in case ( [ form - pd ] ) ) . the corollary is proved . in what follows , for every irreducible non - linear germ @xmath604 of analytic curve in @xmath605 ( or briefly , _
irreducible non - linear germ _ ) its puiseaux exponent ( see definition [ defpuis ] ) will be denoted by @xmath606 .
the main results of this subsection are the following lemma , proposition and corollary
. they will be used in subsection 3.2 in the proof of property ( i ) of mirrors of a 4-reflective billiard and coincidence of their isotropic tangent lines .
proposition [ tang - asympt ] stated below will be used in their proofs and also in section 4 , where we show that the mirrors are confocal conics .
[ lemgerm ] consider a pair of distinct non - linear irreducible germs @xmath604 and @xmath607 , set @xmath112 .
let @xmath37 be isotropic and @xmath118 .
let @xmath608 denote the germ at @xmath121 of two - dimensional analytic subset defined as follows : @xmath609 , if and only if @xmath610 and either @xmath611 , or @xmath612 and @xmath613 , or the lines @xmath614 , @xmath7 are symmetric with respect to @xmath615 .
( thus , @xmath119 contains the curve @xmath129 . )
let one of the following conditions hold : \(i ) @xmath31 ; \(ii ) @xmath616 , but @xmath617 ; \(iii ) @xmath616 , @xmath618 and @xmath38 is an infinite point ; \(iv ) @xmath616 is a finite point , @xmath618 and @xmath619 then the germ @xmath119 is irreducible .
[ refbir ] we say that three irreducible algebraic curves @xmath620 form a _ reflection - birational triple , _ if they are not lines , @xmath437 , @xmath438 and there exists a birational isomorphism @xmath621 such that for a non - empty zariski open set of pairs @xmath622 one has @xmath623 and the lines @xmath35 and @xmath73 are symmetric with respect to the tangent line @xmath71 .
[ clemgerm ] let @xmath42 , @xmath41 , @xmath44 be a reflection - birational triple , @xmath37 be an isotropic tangent line to @xmath41 .
for every their tangency point @xmath105 and every @xmath624 consider the germ @xmath119 constructed above for the local branches @xmath259 and @xmath625 .
let there exist a tangency point @xmath105 , @xmath626 , such that for every @xmath624 the corresponding germ @xmath119 is irreducible .
then the line @xmath37 intersects the curve @xmath78 at a unique point @xmath91 , and the transformation @xmath99 contracts the curve @xmath106 to the point @xmath107 . [ plemgerm ]
let @xmath42 , @xmath41 , @xmath44 be a reflection - birational triple , and let @xmath37 be an isotropic tangent line to @xmath41 .
let there exist a tangency point @xmath105 , @xmath626 , such that for every @xmath624 the local branches @xmath259 and @xmath625 satisfy one of the conditions ( i)(iv ) from lemma [ lemgerm ] . then the line @xmath37 intersects the curve @xmath78 at a unique point @xmath91 .
the lemma and the proposition are proved below .
the corollary follows immediately from them .
[ remiv ] if in the above condition ( iv ) inequality ( [ puiseq ] ) does not hold , then the germ @xmath119 is not irreducible , and some its irreducible component does not contain the curve @xmath129 .
this statement will not be used in the paper .
its proof omitted to save the space follows arguments similar to the proof of lemma [ lemgerm ] given below .
the author does not know whether the statement of proposition [ clemgerm ] holds in full generality , without requiring the irreducibility of all the germs @xmath119 corresponding to some @xmath38 . for the proof of lemma [ lemgerm ]
we introduce affine coordinates @xmath192 centered at @xmath38 so that @xmath37 is the @xmath139-axis .
we fix an arbitrarily small @xmath627 , and for every @xmath628 we consider the cone @xmath629 saturated by the lines through @xmath184 with moduli of azimuths greater than @xmath43 .
we denote @xmath630 @xmath631 we already know that the cone @xmath632 shrinks to @xmath37 , as @xmath633 ( proposition [ plimits ] ) , thus each connected component of the intersection @xmath634 shrinks to @xmath34 .
we show ( case by case ) that for every @xmath184 close enough to @xmath38 each one of the latter components is simply connected .
thus , for those @xmath184 the complement @xmath635 is connected , and it is the whole curve @xmath42 with small holes deleted ; the latter holes shrink to @xmath34 , as @xmath633 .
note that for every @xmath636 the line @xmath637 reflects from @xmath638 to a line @xmath188 through @xmath184 with modulus of azimuth no greater than @xmath43 .
moreover , each line @xmath188 through @xmath184 with azimuth less than @xmath43 corresponds to some @xmath636 .
let us localize the analytic set @xmath119 by the inequality @xmath639 with small @xmath627 .
then for every @xmath628 close enough to @xmath38 the preimage of the point @xmath184 under the projection @xmath640 is a connected holomorphic curve conformally projected onto @xmath641 that accumulates to the curve @xmath129 , as @xmath633 .
this implies the irreducibility of the germ @xmath119 .
the most technical cases of lemma [ lemgerm ] are cases ( iii ) and ( iv ) , when the germs @xmath604 and @xmath607 are tangent to each other . in the proof of the lemma in those cases we use proposition [ tang - asympt ] stated and proved below that concerns the family of tangent lines to the curve @xmath42 .
it describes the asymptotic relation between the tangency point and the intersection points of the tangent line with the curve @xmath41 .
it will imply that in case ( iii ) with @xmath642 and in case ( iv ) the cone @xmath632 contains no tangent line to @xmath42 .
this in its turn implies the simple connectivity of the components of the intersection @xmath643 . in case ( iii ) with @xmath644 we study the projections of the components of the intersection @xmath634 to the @xmath139-axis .
we show that the projection of each component lies in a disk disjoint from @xmath645 .
this together with the maximum principle implies that the minimal topological disk @xmath646 containing the component is disjoint from the vertical line @xmath647 .
this together with the maximum principle , now applied to the projection @xmath648 from the point @xmath184 implies that the intersection component under question is simply connected .
now let us pass to the proofs .
we consider parametrized curves ( germs ) and identify them with their parameter spaces ( disks in @xmath293 ) .
let @xmath604 and @xmath607 be distinct tangent irreducible non - linear germs : @xmath649 , @xmath650 .
let @xmath192 be affine coordinates centered at @xmath52 such that @xmath37 is the @xmath139-axis .
set @xmath651 the subset @xmath652 represents a germ of one - dimensional analytic set at @xmath653 .
we consider its irreducible components and their projections to the product @xmath654 : both @xmath184 and @xmath655 are projected to @xmath37 along the @xmath234-axis .
each irreducible component defines two implicit multivalued functions : - the function @xmath656 , whose graph is the image of the component under the above projection ; - the function @xmath657 : the azimuth of the tangent line @xmath658 .
we normalize the coordinates so that the curves @xmath41 and @xmath42 are graphs of functions @xmath659 [ tang - asympt ] let @xmath41 , @xmath42 be two tangent irreducible non - linear germs of analytic curves at a point @xmath47 .
let the coordinates @xmath192 , the number @xmath660 , the germ @xmath661 and the functions @xmath662 and @xmath663 be as above .
then for every irreducible component of the germ @xmath661 the corresponding implicit functions @xmath664 and @xmath665 have asymptotic puiseaux expansions at @xmath152 of the following possible types ; for every given pair @xmath48 all the corresponding asymptotics are realized by appropriate irreducible components : case 1 ) : @xmath666 .
two possible asymptotics for every @xmath48 : @xmath667 @xmath668 case 2 ) : @xmath669 , @xmath670 are relatively prime . then @xmath671 case 3 ) : @xmath642 . set @xmath672 , as above , @xmath673
. then @xmath674 the germ @xmath661 is given by zero set of an analytic function germ on @xmath479 at @xmath653 that has the type @xmath675 in the variables @xmath663 and @xmath662 . or equivalently , by an equation @xmath676 with the left - hand side being an analytic function of the parameters of the curves @xmath41 and @xmath42 .
an implicit function @xmath664 corresponding to an irreducible component of the germ @xmath661 is a solution to ( [ tang - form ] ) that has puiseaux expansion without free term .
hence , the restrictions to its graph of the three monomials @xmath677 , @xmath678 , @xmath679 should satisfy the following statements as multivalued functions in @xmath662 after substitution @xmath656 : - at least two of the above monomials have lower puiseaux terms in @xmath662 with equal powers ; we call them _ principal monomials _ ; their sum is of smaller order , i.e. , it starts with higher terms ; - the remaining monomial ( if any ) should be of smaller order than the principal ones . in more detail , consider the newton diagram in @xmath132 of the above triple of monomials .
that is , take the union of the translation images of the positive quadrant by the vectors @xmath680 , @xmath681 , @xmath682 .
the newton diagram is its convex hull .
edges _ are segments in its boundary that are not contained in the coordinate axes . for
every irreducible component of the germ @xmath661 the corresponding principal monomials should lie in the same edge of the newton diagram .
vice versa , each edge is realized by an irreducible component .
this is a version of a classical observation due to newton .
case 1 ) : @xmath666 .
then the newton diagram has two edges : the segments @xmath683 $ ] and @xmath684 $ ] .
these edges correspond to asymptotics ( [ case31 ] ) and ( [ case32 ] ) respectively .
case 2 ) : @xmath685 .
then there is a unique edge @xmath686 $ ] , the three above monomials lie there and are principal .
this implies ( [ case2 ] ) .
case 3 ) : @xmath642 .
we have one edge @xmath686 $ ] , the point @xmath681 is in the interior of the newton diagram , the principal monomials are @xmath677 and @xmath679 .
this implies ( [ case1 ] ) .
proposition [ tang - asympt ] is proved .
* of lemma [ lemgerm ] . * as it was shown above , for the proof of the lemma it suffices to prove that for every @xmath628 close to @xmath38 each connected component of the intersection @xmath687 is simply connected .
let us prove this case by case . to do this
, we consider the projection @xmath688 of the curve @xmath42 from the point @xmath184 .
note that the intersection @xmath634 is the preimage of a disk @xmath689 ; the symmetry with respect to the line @xmath638 sends the disk @xmath690 to another disk that correspond exactly to the lines through @xmath184 with moduli of azimuths greater than @xmath43 .
let @xmath192 be affine coordinates centered at @xmath38 with @xmath37 being the @xmath139-axis .
we identify all the projective lines @xmath691 by translations and introduce the coordinate @xmath692 on @xmath693 : @xmath694 .
this induces coordinate @xmath149 on each @xmath695 .
case ( i ) : @xmath31 .
then there exist neighborhoods @xmath696 and @xmath697 such that for every @xmath698 the image @xmath699 lies in the unit disk @xmath424 : if @xmath700 is close to @xmath34 and @xmath628 is close @xmath38 , then the line @xmath614 is close to @xmath701 .
this together with the maximum principle applied to the projection @xmath702 implies the simple connectivity of components of the intersection @xmath634 .
case ( ii ) : @xmath616 but @xmath703 .
then for every @xmath628 close enough to @xmath38 the cone @xmath632 contains no tangent lines to @xmath42 through @xmath184 , since @xmath632 shrinks to the line @xmath37 transverse to @xmath85 , as @xmath633 .
therefore , the projection @xmath704 of each component of the intersection @xmath634 to the disk @xmath690 is a branched covering either without critical points , or with exactly one critical point of maximal multiplicity .
the latter happens exactly when the intersection component under question contains @xmath34 and the latter is a cusp of the curve @xmath42 : this is the critical point . in both cases
the component is obviously simply connected .
case ( iv ) : @xmath649 is a finite point , @xmath650 , and @xmath705 .
we choose @xmath192 to be a finite affine chart .
let us show that for every @xmath628 close to @xmath52 the cone @xmath632 contains no tangent line to the curve @xmath42 , as in case ( ii ) .
indeed , the azimuths of all the lines forming the cone @xmath632 have uniform asymptotics @xmath706 , by ( [ az - fin ] ) . on the other hand ,
the azimuths of the tangent lines to @xmath42 through @xmath184 are of order @xmath707 , @xmath708 . indeed , in the case , when @xmath644 , one has @xmath709 , see ( [ case31])([case2 ] ) , hence @xmath710 . in the case ,
when @xmath642 , one has @xmath711 , by ( [ case1 ] ) , and hence @xmath708 : this is equivalent to the inequality @xmath705 from the assumption .
thus , the azimuths of the tangent lines uniformly asymptotically dominate the azimuths of the lines forming the cone @xmath632 .
hence , the cone @xmath632 contains no tangent lines , whenever @xmath184 is close enough to @xmath52 , and all the components of the intersection @xmath634 are simply connected , as in case ( ii ) . case ( iii ) : @xmath649 is an infinite point and @xmath650 .
then the azimuths of the lines forming the cone @xmath632 have uniform asymptotics @xmath712 , where @xmath713 , @xmath714 , by ( [ az - inf ] ) ; here @xmath660 is the same , as in ( [ sigm ] ) .
subcase ( iii a ) : @xmath642 .
then the azimuths of the tangent lines to @xmath42 through @xmath184 are of order @xmath707 , @xmath715 , see ( [ case1 ] ) .
therefore , the cone @xmath632 contains no tangent lines to @xmath42 , whenever @xmath184 is close enough to @xmath52 , and we are done , as in case ( ii ) .
subcase ( iii b ) : @xmath644 .
the intersection points @xmath716 satisfy the asymptotic equation @xmath717 substituting the expressions @xmath718 and @xmath719 , @xmath720 , @xmath721 , we get @xmath722 * claim 1 . *
_ let @xmath644 .
then there exists a finite subset @xmath723 depending only on @xmath724 , @xmath725 and @xmath660 such that the projection to the @xmath139-axis of the intersection @xmath634 lies in @xmath726- neighborhood of the subset @xmath727 , as @xmath189 . if there exists a family @xmath728 of intersection points such that @xmath729 ( along a sequence @xmath730 ) , then @xmath685 and @xmath731 .
_ in the case , when @xmath666 , all the solutions @xmath663 to ( [ yttt ] ) are @xmath732 . indeed ,
otherwise , if there existed a family of solutions @xmath663 to ( [ yttt ] ) that is not @xmath732 along a sequence of points @xmath189 , then the term @xmath733 in ( [ yttt ] ) would have dominated the rest of ( [ yttt ] ) , which is obviously impossible .
let us now treat the case , when @xmath734 . consider the three monomials @xmath735 , @xmath736 and @xmath737 in the left - hand side of ( [ yttt ] ) . in the case , when @xmath738 , their newton diagram consists of one edge @xmath739 $ ] : the point @xmath740 lies on this edge .
these are the principal monomials in a similar sense , as in the proof of proposition [ tang - asympt ] .
therefore , if @xmath741 , then each solution to ( [ yttt ] ) has asymptotics @xmath742 , with @xmath571 being a root of polynomial depending on @xmath202 , @xmath533 and @xmath660 , as in the same proposition .
the number @xmath533 being a three - valued function in @xmath202 , we get a triple of polynomials depending on @xmath202 and @xmath660 .
this proves the claim for the union @xmath743 of their roots .
let now @xmath587 : then the third monomial vanishes . for every family of solutions @xmath656 to ( [ yttt ] )
one has @xmath744 : otherwise , the monomial @xmath745 would have dominated the rest of the left - hand side in ( [ yttt ] ) along a subsequence @xmath730 , which is impossible , as above . substituting @xmath587 and @xmath744 to ( [ yttt ] ) yields that the expression @xmath746 should be of order @xmath747 .
this implies that each family of solutions @xmath656 to ( [ yttt ] ) has asymptotics either @xmath748 , or @xmath749 with @xmath750 .
this implies the first statement of the claim with @xmath743 consisting of zero and all the numbers @xmath751 .
let now there exist a family of solutions @xmath752 to ( [ yttt ] ) , i.e. , there exist a family of solutions @xmath655 to ( [ ytt ] ) with @xmath753 .
then @xmath685 , by the above discussion , and @xmath754 , by ( [ ytt ] ) .
this proves the claim .
given a family @xmath646 of connected components of the intersection @xmath634 , let us prove their simple connectivity , whenever @xmath184 is close enough to @xmath52 .
let @xmath755 denote the minimal topological disk containing @xmath646 .
without loss of generality we consider that there exists an @xmath756 such that for every @xmath184 close enough to @xmath52 the projection of the domain @xmath646 , and hence @xmath757 to the @xmath139-axis lies in @xmath732-neighborhood of the point @xmath758 , passing to subsequence @xmath730 ( claim 1 ) .
let @xmath759 .
then the domain @xmath757 does not contain @xmath721 , whenever @xmath184 is close enough to @xmath52 .
therefore , the image of its projection @xmath760 from the point @xmath184 is contained in the affine chart @xmath761 .
hence , @xmath762 is simply connected , by the maximum principle , as in case ( i ) .
let now @xmath364 .
we claim that then @xmath646 can not be tangent to a line through @xmath184 .
indeed , the @xmath139- and @xmath234- coordinates of all the points @xmath763 are asymptotically equivalent to those of the point @xmath184 , as @xmath189 , by the claim .
therefore , if there existed a @xmath763 such that the line @xmath614 is tangent to @xmath42 at @xmath655 , then the azimuth of the tangent line would have been asymptotically equivalent to that of the line @xmath638 : both of them should be equivalent to @xmath764 . thus , the cone @xmath632 would have contained a line with azimuth @xmath765 , while we already know that the azimuths of all its lines should have uniform asymptotics @xmath248 with @xmath713 , see ( [ az - inf ] ) .
the latter three possible values of the number @xmath533 are distinct from @xmath202 , since @xmath766 .
the contradiction thus obtained proves that in the case under consideration the component @xmath646 is not tangent to a line through @xmath184 , and hence , is simply connected , as in case ( ii ) .
this finishes the proof of simple connectivity of the components of the intersection @xmath634 .
together with the discussion following remark [ remiv ] , this proves lemma [ lemgerm ] .
* of proposition [ clemgerm ] . * the mapping @xmath99 contracts the curve @xmath106 to a point @xmath107 with some @xmath767 , by proposition [ plimits ] and birationality . fix an arbitrary @xmath767 , and let us show that @xmath99 contracts the curve @xmath106 to @xmath107 .
this together with holomorphicity of the mapping @xmath99 on finitely punctured curve @xmath106 ( isolatedness of indeterminacies ) implies uniqueness of the intersection point @xmath91 .
fix a line @xmath768 through @xmath38 , a point @xmath769 and consider the family of lines @xmath770 with @xmath771 : @xmath772 , as @xmath633 . for
every @xmath773 that is not an isotropic tangency point let @xmath190 denote the line symmetric to @xmath188 with respect to the tangent line @xmath638 .
one has @xmath191 , by proposition [ plimits ] .
fix a family of intersection points @xmath774 that tend to @xmath91 , as @xmath633 .
let @xmath775 denote the family of points such that @xmath776 : it exists and is unique by birationality .
the family of pairs @xmath777 is obviously an algebraic curve , hence , there exists a limit @xmath778 .
if @xmath779 , then @xmath780 , @xmath99 is holomorphic at @xmath781 , @xmath782 in a neighborhood of the pair @xmath781 ( proposition [ plimits ] ) , and we are done .
let now @xmath783 .
the irreducible surface germ @xmath119 contains the germs at @xmath121 of analytic curves @xmath784 and @xmath785 , where @xmath786 is the space of projective lines through the point @xmath38 .
the germ @xmath119 is smooth in the complement to the curve @xmath110 .
consider the mapping @xmath787 : @xmath788 .
this is a well - defined holomorphic mapping on the complement @xmath789 , by birationality and isolatedness of indeterminacies of the mapping @xmath99 .
one has @xmath790 , as @xmath791 . indeed , as @xmath792 , set @xmath793 , the points @xmath394 tend to the intersection @xmath794 .
the germ @xmath119 being irreducible , there are arbitrarily small connected neighborhoods of the point @xmath121 in @xmath119 , and their complements to @xmath110 are also connected . for every latter neighborhood @xmath795 and each @xmath796
the line @xmath127 is close to @xmath37 .
therefore , the corresponding point @xmath394 should stay in one and the same local branch intersecting @xmath37 of the curve @xmath78 , as @xmath139 ranges in @xmath797 .
this implies that the limit of the point @xmath394 exists and lies in @xmath37 , as @xmath791 . if @xmath791 along the curve @xmath798 , then @xmath799 , by construction .
hence , @xmath790 , as @xmath791 in @xmath789 .
the germ at @xmath121 of the curve @xmath800 is contained in @xmath119 , and @xmath801 contracts it to the point @xmath107 , by proposition [ plimits ] and the previous statement . or equivalently , @xmath99 contracts @xmath106 to @xmath107 .
this proves the proposition .
in this section and in what follows we consider that @xmath41 , @xmath42 , @xmath43 , @xmath44 is a 4-reflective billiard , and no its mirror is a line . in subsection 3.1
we prove birationality of neighbor edge correspondence ( lemma [ lbirat ] ) and deduce rationality of mirrors ( corollary [ contrat ] ) and description of the degenerate quadrilateral orbits with one of the vertices being a cusp with isotropic tangency ( corollary [ bircusp ] ) . in subsection 3.2
we show that the mirrors have property ( i ) , common isotropic tangent lines , and opposite mirrors coincide : @xmath45 , @xmath62 .
[ lbirat ] let @xmath802 be the 4-reflective set .
there exists a unique birational isomorphism @xmath803 such that @xmath623 for every @xmath92 .
in particular , the algebraic set @xmath80 is irreducible , the projection @xmath804 is birational and the curves @xmath42 and @xmath44 have equal degrees .
the algebraic set @xmath80 is epimorphically projected onto @xmath805 ( proposition [ comp - set ] ) , and its projection to @xmath115 defines an algebraic correspondence @xmath621 : @xmath623 for every @xmath92 .
suppose the contrary to birationality : one of the mappings @xmath806 , say , @xmath99 is multivalued on an open set .
then there exists an open subset @xmath101 such that for every @xmath807 there exist at least two distinct quadrilateral orbits @xmath808 and @xmath809 of the billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 , where @xmath810 and @xmath811 depend analytically on @xmath781 ; @xmath812 on @xmath795 .
the latter immediately implies that @xmath813 and @xmath814 on @xmath795 ( after shrinking @xmath795 ) .
indeed , if @xmath815 but @xmath814 on @xmath795 , then @xmath816 , since the lines @xmath817 and @xmath818 are the images of the lines @xmath36 and @xmath35 under the symmetries with respect to the lines @xmath819 and @xmath71 respectively .
thus , @xmath38 , @xmath39 and @xmath91 are on the same line for an open set of quadrilateral orbits @xmath46 , which is impossible .
the quadrilaterals @xmath820 form an open set of 4-periodic orbits of the billiard @xmath43 , @xmath44 , @xmath44 , @xmath43 with two pairs of coinciding neighbor mirrors , - a contradiction to corollary [ lnoncoinc ] .
this proves lemma [ lbirat ] .
[ contrat ] let @xmath41 , @xmath42 , @xmath43 , @xmath44 be a 4-reflective algebraic billiard , and the curve @xmath41 be not a line .
then @xmath42 and @xmath44 are rational curves .
thus , the mirrors of a 4-reflective algebraic billiard are rational , if none of them is a line . as it is shown below , corollary [ contrat ]
is immediately implied by lemma [ lbirat ] and the following well - known theorem from algebraic geometry .
it is a part of the indeterminacy resolution theorem for birational mappings .
[ trat ] ( implicitly contained in @xcite , p.546 of russian edition ) .
let @xmath821 be a birational isomorphism of smooth complex projective surfaces .
then each curve in @xmath822 contracted by @xmath330 to a point in @xmath823 is rational .
* of corollary [ contrat ] . *
it suffices to show that @xmath42 is rational , by symmetry .
there exists an isotropic tangency point @xmath105 , since @xmath41 is not a line .
the birational isomorphism @xmath621 from lemma [ lbirat ] contracts the fiber @xmath106 to a pair @xmath824 with @xmath825 , by proposition [ plimits ] , as in the proof of proposition [ clemgerm ] .
therefore , the fiber @xmath826 is a rational curve ( theorem [ trat ] ) , and thus , so is @xmath42 .
corollary [ contrat ] is proved .
[ bircusp ] let @xmath41 , @xmath42 , @xmath43 , @xmath44 be a 4-reflective algebraic billiard , and @xmath80 be its 4-reflective set .
let @xmath113 be a cusp with isotropic tangent line .
let @xmath827 be a parametrized analytic curve consisting of quadrilaterals @xmath92 with fixed @xmath91 and variable @xmath38 or @xmath39 .
then @xmath447 along @xmath110 , and @xmath34 is a cusp of the same degree ( see footnote 4 in subsection 1.2 ) , as @xmath91 .
let , e.g. , @xmath38 vary along the curve @xmath110 .
fix a quadrilateral @xmath828 with @xmath829 and @xmath38 being not a marked point ( @xmath356 , by corollary [ lnoncoinc ] ) .
then the birational transformations @xmath806 are biholomorphic at @xmath781 and @xmath107 respectively .
the biholomorphicity together with the reflection law imply that the intersection index of the variable line @xmath35 with the local branch @xmath625 is equal to that of the variable line @xmath73 with @xmath374 , i.e. , the degree of the cusp @xmath91 .
thus , it is greater than one , and we have two possibilities : either @xmath86 and @xmath34 varies along the curve @xmath110 ; or @xmath447 along @xmath110 and it is a cusp of the same degree , as @xmath91 , and we are done .
suppose the contrary : the former , tangency case takes place .
then @xmath830 along @xmath110 ( corollary [ lnoncoinc ] ) , and we can and will consider that @xmath32 and the line @xmath85 is not isotropic in our quadrilateral @xmath46 .
hence , @xmath831 , by corollary [ connect ] and since the tangent line @xmath375 is isotropic .
the contradiction thus obtained proves corollary [ bircusp ] .
the main result of the present subsection is the following lemma . [ commi ] the curves @xmath41 ,
@xmath42 , @xmath43 , @xmath44 have property ( i ) and common isotropic tangent lines , and @xmath45 , @xmath62 .
let us first prove property ( i ) and coincidence of isotropic tangent lines . to do this , it suffices to show that for every isotropic tangent line @xmath37 to any mirror each one of the curves @xmath75 , @xmath76 , @xmath77 , @xmath78 intersects @xmath37 at a single point .
fix an isotropic tangent line @xmath37 to some mirror . among the tangent branches to @xmath37 of the mirrors let us fix the one that either has infinite tangency point , or is smooth , or has the maximal possible puiseaux exponent .
let this be , say , the local branch of the curve @xmath41 at a point @xmath832 .
the curves @xmath42 , @xmath41 and @xmath44 taken in this or inverse order form a reflection - birational triple ( definition [ refbir ] and lemma [ lbirat ] ) . for every @xmath624 the pair of germs @xmath259 and @xmath625 satisfy some of the conditions ( i)(iv ) of lemma [ lemgerm ] : if @xmath616 is a finite point and @xmath833 , then inequality ( [ puiseq ] ) follows from the assumption that either the germ @xmath259 is smooth ( hence @xmath834 ) , or @xmath644 . therefore , the curve @xmath78 intersects the line @xmath37 at a single point @xmath91 ( corollary [ plemgerm ] ) , hence @xmath835 .
analogously , the curve @xmath76 intersects the line @xmath37 at a unique point , by the above arguments with @xmath42 and @xmath44 interchanged .
let us prove that each one of the curves @xmath75 and @xmath77 intersects the line @xmath37 at a unique point . to do this ,
note that the curves @xmath41 , @xmath44 , @xmath43 taken in this or inverse order also form a reflection - birational triple , by lemma [ lbirat ] . case 1 ) : @xmath91 is either an infinite point , or a finite point that is not a cusp
. then some of the conditions ( i)(iv ) of lemma [ lemgerm ] holds for @xmath604 , @xmath607 replaced by the local branches @xmath374 and @xmath259 respectively , as in step 1 .
therefore , the curve @xmath77 intersects the line @xmath37 at a unique point , as in step 1 , and analogously so does @xmath75 , by symmetry .
case 2 ) : @xmath91 is a finite cusp . consider a one - parametric family @xmath110 of quadrilaterals @xmath836 with the above fixed @xmath91 and variable @xmath837 . then @xmath838 on @xmath110 ,
and it is a cusp , by corollary [ bircusp ] .
thus , for every @xmath839 the lines @xmath126 and @xmath397 are symmetric with respect to the line @xmath128 ( reflection law ) . therefore , @xmath41 is a conic with foci @xmath381 and @xmath91 , by proposition [ char - conic ] .
it intersects the line @xmath37 at their unique tangency point @xmath38 .
let us show that @xmath45 .
the conic @xmath41 has at least two distinct isotropic tangent lines @xmath840 and @xmath841 , set @xmath842 .
each of them intersects the curve @xmath78 at a unique point , by step 1 .
these intersection points are distinct , as are their tangent lines @xmath840 , @xmath841 , and @xmath843 , by uniqueness .
consider the composition of the projection @xmath844 with the projection @xmath845 from the point @xmath497 .
this is a rational mapping @xmath846 that has two distinct critical points with the maximal multiplicity : the intersection points @xmath847 , @xmath558 .
therefore , it has no other critical points , and in particular , the curve @xmath44 has no cusps with non - isotropic tangent lines .
hence , @xmath45 , by corollary [ a = c ] , thus @xmath43 is a conic tangent to @xmath37 .
we have proved that the mirrors have property ( i ) and common isotropic tangent lines .
recall that a curve with property ( i ) has no cusps with non - isotropic tangent lines ( propositon [ pisotr ] and corollary [ cnform ] ) .
this together with corollary [ a = c ] implies that @xmath45 and @xmath62 .
lemma [ commi ] is proved .
[ cortang ]
let @xmath41 , @xmath42 , @xmath41 , @xmath42 be a 4-reflective complex planar algebraic billiard , and no mirror be a line .
then for every @xmath848 that is not an isotropic tangency point the collection of lines through @xmath38 tangent to @xmath42 ( with multiplicities ) is symmetric with respect to the tangent line @xmath71 .
it suffices to prove the statement of the corollary for every @xmath105 lying outside the finite set formed by the isotropic tangency points in @xmath75 and the intersection of the curve @xmath75 with the union of the curve @xmath42 , isotropic tangent lines to @xmath42 and the lines tangent to @xmath42 at the points of intersection @xmath849 .
the statement of the corollary remains valid after passing to limits .
fix an @xmath38 as above . for every @xmath850 such that @xmath851 the pair @xmath478 lifts to a quadrilateral @xmath92 with @xmath852 ( proposition [ comp - set ] )
; one has @xmath31 and @xmath32 , by construction .
then @xmath370 is the tangent line to @xmath42 symmetric to @xmath35 with respect to the line @xmath71 , by corollary [ connect ] and since @xmath42 has no cusps with non - isotropic tangent lines , as at the end of the above proof .
this proves the corollary .
recall that we assume that in the 4-reflective billiard under consideration no mirror is a line .
we have already shown that @xmath45 , @xmath62 and they are two distinct rational curves with property ( i ) and common isotropic tangent lines . in the next subsection we prove that they are conics .
their confocality will be proved in the second subsection .
the main tools we use here are corollaries [ cnform ] , [ cortang ] and proposition [ tang - asympt ] .
[ caust - conic ] let a pair of rational planar curves @xmath41 and @xmath42 different from lines form a 4-reflective billiard @xmath41 , @xmath42 , @xmath41 , @xmath42
. then @xmath41 and @xmath42 are conics .
theorem [ caust - conic ] is proved below .
first we show that the mirrors have no cusp ( the next lemma ) .
then we split the proof of theorem [ caust - conic ] into three cases ( proposition [ prop - cases ] ) , two of them will be treated in separate propositions .
[ lem - cusp ] in the conditions of theorem [ caust - conic ] none of the curves @xmath41 and @xmath42 has cusps . in the proof of this lemma we use the following proposition .
[ pr - cusp ] let a rational curve in @xmath29 with property ( i ) have a cusp .
then it has either only one cusp , or two cusps . in the latter case
its local branches at the cusps have distinct degrees .
the curve under consideration has one of the normal forms ( [ form - pd ] ) or ( [ form - pq ] ) in appropriate isotropic coordinates on the finite plane @xmath4 .
consider , e.g. , case ( [ form - pd ] ) : then @xmath533 and @xmath44 are relatively prime .
suppose the contrary : the curve has cusps both at the origin and at the infinity , and the corresponding local branches have equal degrees .
the latter degrees are equal to @xmath853 and @xmath854 respectively .
hence , @xmath533 and @xmath855 are not relatively prime , a contradiction .
case ( [ form - pq ] ) is treated analogously .
* of lemma [ lem - cusp ] .
* suppose the contrary : say , the curve @xmath42 has a cusp @xmath34 , let us denote its lifting to @xmath76 also by @xmath34 .
consider an irreducible algebraic curve @xmath110 consisting of quadrilaterals @xmath92 with fixed @xmath34 and variable @xmath38 .
then @xmath856 on @xmath110 , and it is a cusp of the same degree , as @xmath34 , by corollary [ bircusp ] .
let us show that @xmath857 .
the contradiction thus obtained to proposition [ pr - cusp ] will prove the lemma .
the curve @xmath42 has one of the normal forms ( [ form - pd ] ) or ( [ form - pq ] ) , and each its cusp is either an isotropic point at infinity , or the origin . if @xmath34 is an isotropic point at infinity , then the lines @xmath73 , which are symmetric to @xmath35 with respect to the line @xmath71 , should pass through the other isotropic point at infinity , and the latter obviously coincides with @xmath91 ; thus @xmath857 .
otherwise , @xmath34 is the origin in the normal form coordinates : then one has case ( [ form - pd ] ) , and the curve @xmath42 is tangent to the infinity line . if @xmath858 , then for every @xmath38 the line @xmath35 is orthogonal to the tangent line @xmath71 .
hence @xmath41 is a complexified circle , and thus , is tangent to finite lines at both isotropic points at infinity .
therefore , the infinite tangent line to the curve @xmath42 is transverse to @xmath41 , - a contradiction to the fact that @xmath41 and @xmath42 have common isotropic tangent lines .
hence , @xmath34 and @xmath91 are distinct cusps of equal degrees , a contradiction to proposition [ pr - cusp ] .
this proves lemma [ lem - cusp ] .
[ prop - cases ] let @xmath41 , @xmath42 be distinct rational curves with property ( i ) , common isotropic tangent lines and no cusps .
then one of the following holds : case 1 ) : some isotropic line is tangent to @xmath41 , @xmath42 at distinct finite points ; case 2 ) : @xmath41 , @xmath42 are tangent at their common finite isotropic tangency point ; case 3 ) : @xmath41 and @xmath42 are tangent to each other at both isotropic points at infinity ; then they are conics .
let us first assume that one of the curves , say @xmath41 has a finite isotropic tangency point @xmath859 , set @xmath860 .
then @xmath42 is tangent to @xmath37 at some point @xmath861 .
if @xmath862 , then we have case 2 ) .
otherwise , @xmath863 , hence @xmath864 ( property ( i ) of the curve @xmath41 ) . in this case
@xmath861 can not be infinite . indeed ,
otherwise , @xmath861 would be an isotropic point at infinity and @xmath37 would be the unique tangent line to @xmath42 through @xmath861 ( property ( i ) of the curve @xmath42 ) .
but there is another tangent line @xmath865 to @xmath41 through @xmath861 , since @xmath864 , @xmath41 has no cusps and by riemann
hurwitz formula for the projection @xmath598 from the point @xmath861 .
the line @xmath866 is isotropic and not tangent to @xmath42 , by construction and property ( i ) , a contradiction to the coincidence of isotropic tangent lines of the curves @xmath41 and @xmath42 .
thus , we have case 1 ) . if each one of the curves @xmath41 and @xmath42 has no finite isotropic tangency point , then one has case 3 ) ( proposition [ pisotr ] and corollary [ cnform ] )
. then the curves are conics , by the same corollary and absence of cusps .
this proves the proposition .
[ rdeg ] for every @xmath673 and every isotropic tangency point @xmath867 ( which is not a cusp by lemma [ lem - cusp ] ) the puiseaux exponent @xmath868 equals the degree of the curve @xmath869 , by property ( i ) .
[ reg - dist ] let a pair of distinct curves @xmath41 and @xmath42 that are not lines form a 4-reflective billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 .
let there exist an isotropic line @xmath37 that is tangent to @xmath41 and @xmath42 at distinct finite points .
then @xmath41 and @xmath42 are conics . by symmetry
, it suffices to prove that @xmath41 is a conic , i.e. , @xmath870 ( remark [ rdeg ] ) .
let @xmath871 be respectively the tangency points of the line @xmath37 with @xmath41 and @xmath42
. then @xmath872 ( property ( i ) of the curve @xmath42 ) .
there exists another line @xmath768 through @xmath859 that is tangent to @xmath42 , as in the above proof .
fix isotropic coordinates @xmath192 on the finite plane @xmath4 with the origin at @xmath859 , the @xmath139-axis @xmath37 and @xmath873 . for every @xmath628 close to @xmath859 there
exists a line @xmath190 tangent to @xmath42 through @xmath184 that tends to @xmath7 , as @xmath874 .
its image @xmath188 under the symmetry with respect to @xmath638 tends to @xmath37 , and @xmath875 , @xmath876 ( proposition [ tang - asympt1 ] ) . on the other hand ,
the line @xmath188 should be tangent to @xmath42 ( corollary [ cortang ] ) , and so is @xmath37 .
therefore , the intersection point @xmath877 tends to a finite point @xmath863 .
thus , @xmath188 is the line through the points @xmath878 , @xmath879 , and @xmath880 , @xmath881 .
hence , the azimuth @xmath882 is of order @xmath883 , but on the other hand , it is @xmath884 .
thus , @xmath885 and @xmath870 .
proposition [ reg - dist ] is proved .
[ tang - con ] let a pair of distinct curves @xmath41 and @xmath42 that are not lines form a 4-reflective billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 .
let they be tangent to each other at their common finite isotropic tangency point @xmath52 .
then they are conics .
let @xmath192 be affine coordinates on the finite plane with the origin at @xmath52 , and the tangent line @xmath886 be the @xmath139-axis .
let @xmath724 and @xmath725 be the puiseaux exponents at @xmath52 of the curves @xmath41 and @xmath42 respectively ( degrees , see remark [ rdeg ] ) , @xmath887 .
suppose the contrary : @xmath888 .
then there exists another line @xmath768 through @xmath52 that is tangent to @xmath42 , since the projection @xmath889 from @xmath52 has degree @xmath890 , @xmath42 has no cusps and by riemann
hurwitz formula . for every @xmath628 close to @xmath52 let @xmath190 be the tangent line to @xmath42 through @xmath184 such that @xmath891 , as @xmath189 .
its image @xmath188 under the symmetry with respect to the line @xmath638 is also tangent to @xmath42 , tends to @xmath37 , and @xmath892 , @xmath876 , as in the above proof .
on the other hand , the latter azimuth should be a quantity of order @xmath893 , @xmath894 , by proposition [ tang - asympt ] .
this is impossible , since @xmath895 .
proposition [ tang - con ] is proved .
* of theorem [ caust - conic ] . *
lemma [ lem - cusp ] shows that the curves @xmath41 and @xmath42 have no cusps .
proposition [ prop - cases ] lists all the possible cases 1)-3 ) .
propositions [ reg - dist ] and [ tang - con ] respectively show that in cases 1 ) and 2 ) the curves are conics . in case 3 ) the curves are automatically conics .
this proves theorem [ caust - conic ] .
we have shown that @xmath41 and @xmath42 are distinct conics with common isotropic tangent lines . here
we prove that they are confocal , by using confocality criterion given by lemma [ propconf ] .
this will finish the proof of theorem [ tclass ] .
case 1 ) : one of the conics , say @xmath41 is transverse to the infinity line ( i.e. , a transverse hyperbola ) . then @xmath42 is also a transverse hyperbola confocal to @xmath41 , since it has the same isotropic tangent lines and by lemma [ propconf ] , case 1 ) . case 2 ) : both @xmath41 and @xmath42 are tangent to the infinity line , and @xmath41 is tangent to it at a non - isotropic point @xmath859 .
then @xmath42 is also tangent to it at a non - isotropic point @xmath861 , by the coincidence of isotropic tangent lines
. * claim 1 .
* @xmath862 .
suppose the contrary : @xmath871 .
let @xmath896 be the isotropic points at infinity .
then @xmath859 , @xmath861 , @xmath212 , @xmath506 are four distinct points on the infinity line .
let us choose an affine chart @xmath192 with the origin @xmath212 , the @xmath139-axis being the infinity line with @xmath242 there .
fix an arbitrary point @xmath63 close to @xmath859 .
let @xmath897 denote the tangent line to @xmath42 through @xmath38 at a point close to @xmath861 : @xmath897 tends to the infinity line ( @xmath139-axis ) , as @xmath898 .
let @xmath899 denote the line symmetric to @xmath897 with respect to the line @xmath71 .
* claim 2 . * _ the line @xmath900 tends to the infinity line , as @xmath898 .
_ let @xmath901 , @xmath902 , @xmath903 denote respectively the @xmath139-coordinates of the points of intersection of the @xmath139-axis with the lines @xmath897 , @xmath71 and @xmath900 .
set @xmath904 , @xmath905 .
one has @xmath906
( reflection law and proposition [ tang - asympt1 ] ) , and @xmath907 , @xmath908 , as @xmath898 . therefore , @xmath909 .
thus , @xmath900 is the line through the points @xmath38 and @xmath910 , one has @xmath911 , as @xmath898 .
hence , @xmath900 tends to the @xmath139-axis , i.e. , @xmath148 .
this proves the claim .
thus , the line @xmath900 should be a tangent line to @xmath42 that tends to the infinity line , by corollary [ cortang ] and the above claim .
there are two tangent lines to @xmath42 through the point @xmath859 : the infinity line and a finite line , since @xmath872 .
thus , there are two tangent lines to @xmath42 through @xmath38 : the line @xmath897 and a line with a finite limit , as @xmath898 .
therefore , @xmath912 , thus @xmath897 is orthogonal to @xmath71 for all @xmath38 close enough to @xmath859 . in other words ,
every tangent line to @xmath42 close to the infinity line is orthogonal to the conic @xmath41 at both their intersection points .
hence , this is true for all the tangent lines to @xmath42 , by algebraicity .
thus , for every point @xmath63 there is only one tangent line to the conic @xmath437 through @xmath38 , which is impossible .
this proves claim 1 .
thus , the conics @xmath41 and @xmath42 have common isotropic tangent lines and are tangent to the infinity line at the same non - isotropic point , by claim 1 .
this together with lemma [ propconf ] implies that @xmath41 and @xmath42 are confocal .
case 3 ) : @xmath41 is tangent to the infinity line at an isotropic point , say @xmath212 .
then @xmath42 is also tangent to the infinity line at @xmath212 , since the conics @xmath41 and @xmath42 have common isotropic tangent lines .
recall that @xmath480 .
* claim 3 . * _ the conics @xmath41 , @xmath42 are tangent at @xmath212 with at least triple contact .
_ let @xmath192 be affine coordinates with the origin @xmath212 , the @xmath139-axis @xmath148 with @xmath242 there such that @xmath913 , as @xmath914 .
then @xmath915 , @xmath881 .
we have to show that @xmath916 . for
every @xmath628 close to @xmath212 set @xmath876 .
there are two tangent lines @xmath917 and @xmath918 to @xmath42 through @xmath184 , and they tend to the @xmath139-axis , i.e. , the infinity line , as @xmath919 .
their azimuths have asymptotics @xmath920 , @xmath558 , where @xmath921 are the roots of the quadratic equation @xmath922 , and their tangency points with @xmath42 have @xmath139-coordinates @xmath923 ( proposition [ tang - asympt ] ) .
therefore , the latter tangent lines intersect the @xmath139-axis at points with coordinates @xmath924 , while the tangent line @xmath638 intersects it at a point with coordinate @xmath925 ( proposition [ pxt ] ) . the pair of lines @xmath926 is symmetric with respect to the line @xmath638 ( corollary [ cortang ] ) . the symmetry can not fix each @xmath926 for all @xmath184 , since otherwise either two distinct lines @xmath917 and
@xmath918 would be orthogonal to @xmath638 , or one of them would be tangent to both @xmath41 and @xmath42 for all @xmath184 ; none of the latter is possible .
thus , the symmetry permutes the above lines .
hence , @xmath927 ( proposition [ tang - asympt1 ] ) , and thus , @xmath928 . on the other hand , @xmath929 , thus , @xmath916 .
the claim is proved .
claim 3 together with proposition [ ptransl ] imply that in our case the conics @xmath41 and @xmath42 are obtained one from the other by translation of the finite plane .
this together with the coincidence of their finite isotropic tangent lines and lemma [ propconf ] implies that @xmath41 and @xmath42 are confocal .
theorem [ tclass ] is proved .
here we consider a 4-reflective billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 on pair of distinct confocal conics @xmath41 , @xmath42 and prove the following classification of degenerate quadrilateral orbits .
note that smooth mirrors coincide with their normalizations .
[ irrdeg ] let @xmath930 be the 4-reflective set , and let @xmath81 be the subset of 4-periodic orbits .
the complement @xmath82 is the union of the following algebraic curves : \(i ) the curve @xmath931 consisting of quadrilaterals @xmath932 with @xmath933 , @xmath934 , @xmath935 and all the single - point quadrilaterals @xmath936 , @xmath937 ; @xmath938 on each component of the curve @xmath939 ; ( ii ) the curve @xmath940 defined analogously with @xmath41 and @xmath42 interchanged ; \(iii ) for every isotropic tangent line @xmath37 to @xmath41 and @xmath42 at points @xmath859 and @xmath861 respectively the curve @xmath941 consisting of quadrilaterals @xmath942 with variable @xmath38 and @xmath34 ; the curve @xmath943 is rational , and the natural projections @xmath944 , @xmath945 to the positions of the variable vertices @xmath38 and @xmath34 are bijective ; \(iv ) three other rational curves @xmath946 , @xmath947 , @xmath948 defined analogously to @xmath941 and consisting of quadrilaterals @xmath949 , @xmath950 and @xmath951 respectively with variable @xmath38 and @xmath34 .
each one of the curves @xmath931 and @xmath940 is - elliptic , if @xmath41 and @xmath42 are not tangent to each other ; - rational with one transverse self - intersection , if @xmath41 and @xmath42 are quadratically tangent at a unique point ; - rational with one cusp of degree two , if @xmath41 and @xmath42 are tangent with triple contact ; - a union of two smooth rational curves , if @xmath41 and @xmath42 are tangent at two distinct points . fix a quadrilateral @xmath83 .
let us show that it lies in the union of the above curves .
if its vertices are not isotropic tangency points and every two neighbor vertices are distinct , then it lies in @xmath952 , by corollary [ connect ] .
if it has an isotropic tangency vertex , then some its neighbor is also an isotropic tangency vertex ( reflection law and coincidence of isotropic tangent lines of the conics @xmath41 and @xmath42 ) .
therefore , given an isotropic line @xmath37 tangent to @xmath41 and @xmath42 at points @xmath859 and @xmath861 respectively , there exists a rational curve @xmath953 of quadrilaterals @xmath954 with variable @xmath38 ( see the proof of corollary [ contrat ] ) .
let us show that @xmath34 also varies along @xmath955 . indeed , in the contrary case , when @xmath447 on @xmath955 , one would have either @xmath956 , or the variable line @xmath35 is reflected to the constant line @xmath957 ( which is thus isotropic and coincides with @xmath85 ) .
in both cases @xmath833 , by coincidence of isotropic tangent lines of the conics @xmath41 and @xmath42 . hence , @xmath958 on @xmath955 .
therefore , @xmath959 is orthogonal to @xmath71 for every @xmath63 , and @xmath41 is a complex circle centered at @xmath861 . by confocality , @xmath42 is also a complex circle centered at @xmath960 , which is obviously impossible .
the contradiction thus obtained proves that both @xmath38 and @xmath34 vary along the curve @xmath955 .
the projection of the curve @xmath955 to the position of either @xmath38 , or @xmath34 is bijective , by birationality ( lemma [ lbirat ] ) .
a priori , @xmath46 may have yet another degeneracy : coinciding pair of neighbor vertices that are not isotropic tangency points .
first we show that the quadrilaterals in @xmath80 with the latter degeneracy form a finite set of points @xmath961 , @xmath937
. then we prove the last statements of the theorem on the structure of the curves @xmath931 and @xmath940 .
[ aaaa ] let @xmath962 be an intersection point such that the line @xmath71 is not isotropic .
then the line @xmath85 is also not isotropic , and the 4-reflective set @xmath80 contains no algebraic curve of quadrilaterals @xmath963 with fixed @xmath616 and variable @xmath39 or @xmath91 .
the non - isotropicity of the tangent line @xmath85 follows immediately from the coincidence of isotropic tangent lines to @xmath41 and @xmath42 ( confocality ) .
the line @xmath85 is transverse to @xmath41 , since each common tangent line to @xmath41 and @xmath42 is isotropic ( see the proof of corollary [ tan - con ] ) .
suppose the contrary to the second statement : there exists an algebraic curve @xmath964 consisting of quadrilaterals @xmath963 , say , with variable @xmath39 .
let @xmath965 denote the point of intersection @xmath966 distinct from @xmath38 .
then @xmath967 contains a quadrilateral @xmath968 with @xmath969 and @xmath970 .
the quadrilateral @xmath139 may be deformed to a quadrilateral @xmath971 with @xmath972 and @xmath973 , a contradiction to corollary [ connect ] .
the proposition is proved .
let @xmath38 be as in proposition [ aaaa ] .
the proposition immediately implies that the birational isomorphisms @xmath99 , @xmath974 and their inverses are biholomorphic at @xmath975 . let us show that they send @xmath975 to @xmath975 . indeed , take a point @xmath839 , the two tangent lines to @xmath42 through @xmath380 and the tangency points @xmath381 and @xmath394 .
one has @xmath976 , and @xmath977 , as @xmath978 .
this together with biholomorphicity and lemma [ lbirat ] implies that @xmath979 , @xmath980 and @xmath936 is the only quadrilateral in @xmath80 with a pair of neighbor vertices coinciding with @xmath38 . similarly , @xmath981 .
let us prove the last statement of theorem [ irrdeg ] for the curve @xmath931 ; the case of the curve @xmath940 is symmetric . in the case ,
when the conics @xmath41 and @xmath42 are not tangent , the ellipticity of the curve @xmath931 is classical , see @xcite .
namely , the projection @xmath982 : @xmath983 is a double covering over @xmath984 , branching over the points of intersection @xmath985 . if the conics are not tangent , then we have four distinct branching points , thus the curve @xmath931 is elliptic by riemann
hurwitz formula .
the cases of tangent confocal conics are described by corollary [ tan - con ] . in the case ,
when @xmath41 and @xmath42 are quadratically tangent at some point @xmath52 , the curve @xmath931 has two transversely intersected local branches through the point @xmath653 .
this easily follows from proposition [ tang - asympt ] , see ( [ case2 ] ) .
therefore , the projection @xmath986 does not branch at the double intersection @xmath52 and has only two branching points ; thus the curve @xmath931 is rational by riemann
hurwitz formula .
the case of two tangency points is treated analogously : the covering does not branch at all and has two univalent sheets bijectively projected to @xmath984 .
in the case of triple tangency we have two distinct branching points : the point @xmath52 and another branching point . the implicit function @xmath987 defined by the curve @xmath939
is double - valued in the neighborhood of the point @xmath52 , and its both branches have unit derivative at @xmath52 , by ( [ case2 ] ) .
this easily implies that the quadrilateral @xmath988 corresponds to a degree two cusp of the rational curve @xmath931 .
theorem [ irrdeg ] is proved .
here by _ real analytic curve _ we mean a curve @xmath989 analytically parametrized by either @xmath990 , or @xmath991 that is not the infinity line .
if a curve @xmath41 has singularities ( cusps or self - intersections ) , we consider its maximal real analytic extension @xmath581 , where @xmath75 is either @xmath990 , or @xmath991 , see ( * ? ? ?
* lemma 37 , p.302 ) .
the parametrizing curve @xmath75 will be called here the _ real normalization . _
the affine plane @xmath992 is equipped with euclidean metric .
a _ real planar analytic ( algebraic ) pseudo - billiard _ is a collection of @xmath1 real irreducible analytic ( algebraic ) curves @xmath993 .
@xmath1-periodic orbit _ is a @xmath1-gon @xmath23 , @xmath994 , such that for every @xmath15 one has @xmath319 , @xmath995 and the lines @xmath278 , @xmath19 are symmetric with respect to the tangent line @xmath17 .
the latter means that for every @xmath268 the triple @xmath996 , @xmath20 , @xmath997 and the line @xmath17 satisfy either usual , or skew real reflection law . for brevity
, we then say that usual ( skew ) reflection law is satisfied at the vertex @xmath20 . here
we set @xmath10 , @xmath13 , @xmath11 , @xmath14 .
a real pseudo - billiard is called _ @xmath1-reflective , _ if it has an open set ( i.e. , a two - parameter family ) of @xmath1-periodic orbits .
the interior points of the set of @xmath1-periodic orbits will be called _ @xmath1-reflective orbits . _ [ rpseudo ]
the complexification of every real @xmath1-reflective planar analytic pseudo - billiard is a @xmath1-reflective analytic complex planar billiard .
[ class ] a real planar algebraic pseudo - billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 is 4-reflective , if and only if it has one of the following types : \1 ) @xmath45 is a line , the curves @xmath74 are symmetric with respect to it ; \2 ) @xmath41 , @xmath42 , @xmath43 , @xmath44 are distinct lines through the same point @xmath998 , the line pairs @xmath48 , @xmath49 are transformed one into the other by rotation around @xmath52 ( translation , if @xmath52 is an infinite point ) ; \3 ) @xmath45 , @xmath62 and they are distinct confocal conics : either ellipses , or hyperbolas , or ellipse and hyperbola , or parabolas . in every 4-reflective orbit the reflection law at each pair of opposite vertices is the same ; it is skew for at least one opposite vertex pair .
* addendum 1 . * _ in theorem [ class ] , case 1 ) each 4-reflective orbit @xmath46 has the same type , as at fig.1 : it is symmetric with respect to the line @xmath45 and the reflection law is skew at @xmath38 and @xmath39 . in the subcase , when @xmath42 , @xmath44 are lines parallel to @xmath41 ,
the reflection law is usual at @xmath42 and @xmath44 . in the subcase ,
when @xmath62 is a line orthogonal to @xmath41 , the orbits are rhombi symmetric with respect to @xmath41 and @xmath42 : the reflection law is skew at each vertex , see fig.11a ) . if none of the latter subcases holds , then the billiard has both types of 4-reflective orbits @xmath46 : with usual reflection law at @xmath34 , @xmath91 and with skew one .
_ * addendum 2 . *
_ let a 4-reflective pseudo - billiard have type 2 ) . if the lines @xmath41 , @xmath42 , @xmath43 , @xmath44 are parallel and @xmath42 , @xmath44 lie between the lines @xmath41 and @xmath43 , then for every 4-reflective orbit the reflection law is usual at @xmath41 , @xmath43 and skew at @xmath42 , @xmath44 .
otherwise , there are three types of 4-reflective orbits : all the reflection laws are skew ; usual reflection law at @xmath41 , @xmath43 and skew at @xmath42 , @xmath44 ; vice versa , see fig.2 . _ * addendum 3 . *
_ let a 4-reflective pseudo - billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 have type 3 ) . in the case ,
when @xmath41 and @xmath42 are ellipses , all the 4-reflective orbits have the same reflection laws , as at fig.4 : two usual ones and two skew ones . in the case , when @xmath41 and @xmath42 are either hyperbolas , or ellipse and hyperbola , or parabolas , all the possible reflection law combinations are given in the next figures , up to symmetries with respect to the common symmetry lines of the conics @xmath41 and @xmath42 , renaming opposite mirrors and cyclic renaming of mirrors .
the main result of paper @xcite ( theorem 2 ) concerns usual real planar billiards with piecewise - smooth boundary ; the reflection law is usual .
it says that the set of quadrilateral orbits has measure zero . in the particular case of billiard with piecewise - algebraic boundary
this statement follows from the last statement of theorem [ class ] .
* of theorem [ class ] and its addendums . *
the pseudo - billiard @xmath41 , @xmath42 , @xmath43 , @xmath44 under question being 4-reflective , its complexification is 4-reflective ( remark [ rpseudo ] ) .
this together with theorem [ tclass ] implies that it has one of the above types 1)3 ) ( up to cyclic renaming of the mirrors ) .
let us show that each type is realized and prove the addendums .
* case of type 1 ) . *
each 4-reflective orbit @xmath46 is symmetric with respect to the line @xmath999 : the lines @xmath35 and @xmath73 are symmetric , the lines @xmath1000 and @xmath1001 are symmetric ; hence , the intersection points @xmath1002 and @xmath1003 are symmetric .
the statements of addendum 1 on the reflection law follow immediately from symmetry .
* case of type 2 ) .
* the 4-reflectivity was proved in example [ ex - lines2 ] ; the proof applies to the real case .
let us prove addendum 2 .
subcase 2a ) : @xmath41 , @xmath42 , @xmath43 , @xmath44 are parallel lines , say horizontal .
then the line pair @xmath48 is sent to @xmath49 by translation .
hence , the lower and upper lines are opposite mirrors , say @xmath41 and @xmath43 , the other intermediate lines @xmath42 and @xmath44 are opposite , the reflection law at @xmath41 and @xmath43 is usual , and that at @xmath42 and @xmath44 is skew .
subcase 2b ) : the mirrors pass through the same finite point @xmath52 .
* claim 1 . *
_ the lines @xmath42 and @xmath44 punctured at @xmath52 are contained in a union of two opposite quadrants of the complement @xmath1004 .
_ suppose the contrary .
then two rays @xmath1005 and @xmath1006 of the lines @xmath42 and @xmath44 respectively lie in the same half - plane with respect to the line @xmath41 and are separated by a ray @xmath1007 of the line @xmath43 ; all the rays are issued from the point @xmath52 .
this implies that the rotation around the point @xmath52 sending the ray @xmath1007 to @xmath1005 can not send @xmath1006 to a ray of the line @xmath41 , a contradiction to the condition that the pair @xmath48 is transformed to @xmath49 by rotation .
* claim 2 * _ there exist no 4-reflective orbits with distinct reflection laws at some pair of opposite vertices .
_ suppose the contrary : there exists a 4-reflective orbit @xmath46 , say , with usual reflection law at @xmath38 and skew at @xmath39 .
then the rays @xmath1008 and @xmath1009 are separated by the line @xmath43 and are not separated by the line @xmath41 , a contradiction to claim 1 .
this proves claim 2 .
* claim 3 . * _ there exists no 4-reflective orbit with usual reflection law at each vertex . _
the reflection law at the vertex with maximal distance to @xmath52 is obviously skew . let @xmath1010 ( @xmath1011 ) denote the smallest angle between the lines @xmath41 and @xmath42 ( respectively , @xmath43 and @xmath42 ) .
then the smallest angle between the lines @xmath44 and @xmath43 ( @xmath41 and @xmath44 ) equals @xmath1010 ( respectively , @xmath1011 ) , by isometry of pairs @xmath48 and @xmath49 .
one has @xmath1012 , since the lines are distinct .
let us name the mirrors so that @xmath1013 : then the line @xmath42 is closer than @xmath44 to the line @xmath41 .
one can achieve this by interchanging opposite mirrors @xmath41 and @xmath43 .
* claim 4 .
* _ there exists a 4-reflective orbit @xmath46 with usual reflection law at @xmath38 ( and hence , @xmath39 ) . _
fix a @xmath1014 , an @xmath1015 and a @xmath1016 such that the angle between the rays @xmath1017 and @xmath35 equals @xmath798 . then the line @xmath73 symmetric to @xmath35 with respect to the line @xmath41 intersects @xmath44 at a point @xmath91 lying together with @xmath34 in the same half - plane with respect to the line @xmath41 , by the inequality @xmath1018 .
the triangle @xmath72 thus constructed obviously extends to a 4-reflective orbit @xmath46 , since the composition of symmetries with respect to the mirrors is identity . by construction
, the reflection law at @xmath38 is usual .
the claim is proved .
* claim 5 . * _ there exists a 4-reflective orbit with skew reflection law at each vertex .
let us take the above - constructed 4-reflective orbit @xmath46 with usual reflection law at @xmath38 .
let @xmath1019 denote the line through @xmath38 parallel to @xmath42 , let @xmath1020 be their intersection point with the infinity line .
let us degenerate @xmath46 so that @xmath38 remains fixed and @xmath34 tends to @xmath1020 : the angle inequality @xmath1021 remains valid .
then the vertex @xmath39 tends to the finite intersection point @xmath382 of the line @xmath43 and a finite line @xmath1022 parallel to @xmath42 : the line @xmath1022 contains the point symmetric to @xmath38 with respect to the line @xmath42 .
let @xmath1023 denote the line symmetric to @xmath1019 with respect to @xmath41 .
the vertex @xmath91 tends to the intersection point @xmath394 of the lines @xmath44 and @xmath1023 .
the limit @xmath394 is finite , since the latter lines are not parallel , by the inequality @xmath1013 . by construction , the vertices @xmath38 and
@xmath382 are separated by the line @xmath42 , since they lie on lines parallel to @xmath42 and symmetric with respect to it .
therefore , as @xmath34 crosses the infinity line , the reflection law at @xmath34 remains skew , while the reflection law at @xmath38 changes from usual to skew .
finally , after crossing the infinity line by the vertex @xmath34 we obtain a new quadrilateral @xmath46 with skew reflection law at @xmath38 and @xmath34 .
the reflection law at the other vertices @xmath39 and @xmath91 is also skew , by claim 2 .
this proves claim 5 .
there exists a 4-reflective orbit with usual reflection law at @xmath42 and @xmath44 , by claim 4 , which also applies to @xmath34 and @xmath91 , by symmetry .
this together with claims 25 proves addendum 2 .
* case of type 3 ) : * a pseudo - billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 with @xmath41 and @xmath42 being real confocal conics .
the 4-reflectivity of such a pseudo - billiard in the case of ellipses is given by urquhart theorem [ tell ] .
its proof given in ( * ? ? ?
* , corollary 4.6 ) applies to the other types of confocal conics as well , as was mentioned in loc .
thus , all the pseudo - billiards listed in type 3 ) are realized .
let us prove the reflection law statement of theorem [ class ] and addendum 3 . to do this , we consider the 4-reflective orbit set @xmath1024 and its closure @xmath1025 in the usual topology .
the set @xmath1026 is an algebraic surface with birational projection to @xmath479 ( to positions of any two neighbor vertices ) , by proposition [ comp - set ] and lemma [ lbirat ] . in the case ,
when @xmath41 and @xmath42 are not parabolas , the latter birational projection is a diffeomorphism , and the surface @xmath1026 is a torus : the only indeterminacies of the complexified birational projection correspond to isotropic tangencies ( by proposition [ aaaa ] ) , which do not lie in the real domain .
we use the next proposition describing the complement @xmath1027 .
then we analyze how the reflection laws change as a 4-reflective orbit crosses a component of the latter complement .
the first step of this analysis is given by proposition [ crossma ] below .
[ class - deg ] the complement @xmath1028 is the union of the following sets : case ( i ) : @xmath41 and @xmath42 are ellipses , @xmath41 is smaller .
the union @xmath1029 of two disjoint closed curves @xmath1030 consisting of quadrilaterals of type @xmath1031 such that @xmath35 and @xmath1000 are tangent to @xmath41 at @xmath38 and @xmath39 respectively . case ( ii ) : @xmath41 and @xmath42 are hyperbolas , and @xmath42 separates the branches of the hyperbola @xmath41 .
the union @xmath1029 of two disjoint closed curves @xmath1032 defined as above .
four closed curves @xmath1033 , @xmath54 ; each curve @xmath1033 consists of quadrilaterals @xmath46 with infinite vertex @xmath1034 . case ( iii ) : @xmath41 and @xmath42 are respectively ellipse and hyperbola .
the union @xmath1035 of two disjoint closed curves @xmath1032 and the union @xmath1036 of two disjoint closed curves @xmath1037 defined as in case ( i ) . here and in the next cases the curves @xmath1038 and @xmath1039 contain appropriate single - point quadrilaterals @xmath936 , @xmath937 .
two closed curves @xmath1033 , @xmath1040 , defined as in case ( ii ) . case ( iv ) : @xmath41 and @xmath42 are codirected parabolas , @xmath41 lies inside the domain bounded by @xmath42 . two curves @xmath1032 as in case ( i ) that intersect at the point @xmath988 , @xmath1041 is the infinite intersection point . four closed curves @xmath1042 , @xmath1043 , @xmath1044 , @xmath1045 formed by quadrilaterals @xmath1046 , @xmath1047 , @xmath1048 , @xmath1049 respectively with variable @xmath38 and @xmath34 . case ( v ) : @xmath41 and @xmath42 are oppositely directed parabolas . the union @xmath1035 of two closed curves
@xmath1032 and the union @xmath1036 of two closed curves @xmath1037 defined as in case ( i ) . four closed curves @xmath1050 defined as above .
the curves @xmath1032 intersect only at the point @xmath988 , as do the curves @xmath1037 .
the statements of the proposition in cases ( i ) and ( ii ) follow immediately from theorem [ irrdeg ] and convexity .
let us prove them in case ( iii ) .
the complement @xmath1028 consists of the analytic set @xmath1029 of quadrilaterals of type @xmath1031 with tangencies to @xmath41 at @xmath38 and @xmath39 , analogous analytic set @xmath1051 and the above curves @xmath1033 , by theorem [ irrdeg ] .
the set @xmath1029 consists of two disjoint closed components @xmath1032 .
this follows from the same statement on the set of pairs @xmath476 such that the line @xmath35 is tangent to @xmath41 at @xmath38 .
indeed , each intersection point @xmath34 of a tangent line to the ellipse @xmath41 with the hyperbola @xmath42 lies in one of the two arcs forming the complement of @xmath42 to the interior domain of the ellipse @xmath41 .
choice of arc defines a component of the set @xmath1029 .
each component is diffeomorphically parametrized by the tangency vertex @xmath63 and hence , is closed .
the proof of the same statements for the set @xmath1036 is analogous .
the proof in cases ( iv ) and ( v ) is analogous ; the curves @xmath1052 are the real parts of the curves @xmath1053 from theorem 5.1 corresponding to the infinity line .
the proposition is proved .
[ crossma ] every quadrilateral @xmath1054 with finite vertices has a neighborhood @xmath1055 such that for every quadrilateral @xmath1056 the reflection laws at @xmath380 and @xmath382 are skew , and those at @xmath381 and @xmath394 are either both usual , or both skew , dependently only on @xmath139 .
the proposition follows from definition .
let us prove statements of addendum 3 for each case in proposition [ class - deg ] * subcase ( i ) of confocal ellipses : * each 4-reflective orbit deforms in @xmath1057 to the set @xmath1029 .
this together with proposition [ crossma ] and convexity implies that the reflection law is usual at @xmath42 and skew at @xmath41 and proves addendum 3 .
* subcase of confocal hyperbolas : * say , the hyperbola @xmath62 is inside the concave domain between the two branches of the hyperbola @xmath45 .
the reflection laws for a quadrilateral @xmath1058 do not change , as @xmath139 crosses one of the curves @xmath1032 , @xmath1037 outside the curves @xmath1033 , by proposition [ crossma ] .
they may change , as @xmath139 crosses a curve @xmath1033 , i.e. , some vertex crosses the infinity line .
note that the union of intersection points of different curves @xmath1033 is discrete and does not separate domains .
below we classify reflection law combinations near arbitrary quadrilateral @xmath1059 with unique infinite vertex : either @xmath859 , or @xmath861 ( cases of @xmath1060 or @xmath1061 are analogous ) .
this together with the previous statement and connectivity of the surface @xmath1026 implies that this classification covers all the possible reflection law combinations .
a ) @xmath1062 : only the vertex @xmath859 is infinite . set @xmath1063 .
there exists a path @xmath1064 , @xmath1065 $ ] , @xmath1066 such that @xmath1067 : then @xmath1068 and @xmath1069 .
the deformed vertices @xmath1070 and @xmath1071 do not cross the infinity line , hence they lie on the same affine branch @xmath1072 of the hyperbola @xmath42 , as for @xmath1073 .
let @xmath1074 denote the affine branch of the hyperbola @xmath41 that is contained in the convex domain bounded by @xmath1072 .
for every @xmath184 one has @xmath1075 , since the lines @xmath1076 and @xmath1077 are parallel and by focusing property of the convex mirror @xmath1072 .
hence , the path @xmath798 is disjoint from the union @xmath1078 .
fix a small neighborhood @xmath1079 of the point @xmath1080 .
every quadrilateral in @xmath1081 close to @xmath139 can be deformed to a quadrilateral in @xmath1082 along a path in @xmath1081 close to @xmath798 ; the reflection laws in the deformed quadrilaterals remain constant .
thus , it suffices to describe the types of quadrilaterals in @xmath1083 . for every quadrilateral @xmath1084 the reflection laws at @xmath34 and @xmath91 coincide ( easily follows from proposition 6.7 ) . as a quadrilateral in @xmath1082 crosses @xmath1085 ( i.e. , as @xmath38 crosses the infinity line )
, the reflection law at @xmath34 and @xmath91 changes to opposite .
this implies that every quadrilateral in @xmath1083 , and hence , every quadrilateral in @xmath1081 close to @xmath139 has one of the two types depicted at fig . 8 a ) ,
\b ) @xmath1086 : only the vertex @xmath861 is infinite .
there exists a path @xmath1087 , @xmath1065 $ ] , @xmath1066 , @xmath1088 ( thus , @xmath1089 ) , @xmath1090 , as in the above discussion . as above
, it suffices to describe the types of quadrilaterals in @xmath1081 close to @xmath1080 .
the points @xmath409 and @xmath423 lie on different affine branches of the hyperbola @xmath41 : the lines @xmath1091 and @xmath1092 are parallel to the asymptotic line @xmath1093 and are symmetric with respect to it , and @xmath1093 separates the branches of the hyperbola @xmath41 . in every quadrilateral @xmath1094 close to @xmath1080 the vertices @xmath34 and @xmath91
are either on the same branch of the hyperbola @xmath42 , or on different branches
. then we get fig.8b ) or fig.8c ) respectively ( up to interchanging names of either @xmath34 and @xmath91 , or @xmath38 and @xmath39 , or both ) .
this proves addendum 3 .
* subcase of confocal ellipse and hyperbola .
* let @xmath45 be an ellipse , and @xmath62 be a confocal hyperbola .
it suffices to describe the types of quadrilaterals in @xmath1081 close to a quadrilateral @xmath1095 with exactly one infinite point , say @xmath861 , as in the previous case .
the above deformation argument reduces this problem to the description of types of quadrilaterals in @xmath1081 close to a quadrilateral @xmath1096 . in every quadrilateral @xmath1094 close to @xmath234 the points @xmath34 and @xmath91
lie either on the same branch of the hyperbola @xmath42 , or on different branches .
we get figures 9a ) and 9b ) respectively .
this proves addendum 3 .
* subcase of confocal parabolas . *
any two confocal parabolas are either codirected ( fig.10a ) ) , or oppositely directed ( fig.10b ) ) .
let us consider the first case of codirected parabolas , say , @xmath41 is contained inside the convex domain bounded by @xmath42 .
fix a quadrilateral @xmath1097 .
its reflection law at @xmath42 is usual , by convexity .
we claim that the reflection law at @xmath41 is skew , i.e. , @xmath139 is as at fig.10a ) . to do this
, it suffices to show that @xmath139 deforms in @xmath1057 to a quadrilateral @xmath1098 ( proposition [ crossma ] ) .
there exists a continuous deformation @xmath1099 , @xmath1065 $ ] with finite @xmath1070 and @xmath1071 for all @xmath184 to a quadrilateral @xmath1100 with @xmath1101 , since the exterior parabola @xmath42 has only one infinite point .
then @xmath1098 , hence @xmath1069 .
the vertices @xmath1102 also remain finite : each quadrilateral in @xmath1026 with an infinite vertex has at least two infinite vertices , being contained in one of the curves @xmath1050 from propisiton [ class - deg ] .
this together with the above discussion implies that the quadrilateral @xmath139 is as at fig.10a ) .
let us now consider the case of oppositely directed parabolas .
it suffices to describe reflection laws only in those quadrilaterals in @xmath1057 that are close to the union of curves @xmath1050 , as in the case of hyperbolas .
that is , fix a quadrilateral @xmath1097 , say , with @xmath859 and @xmath861 close to infinity : @xmath859 ( @xmath861 ) lies outside the convex domain bounded by the parabola @xmath42 ( respectively , @xmath41 ) .
the quadrilateral @xmath139 deforms to @xmath1029 in @xmath1057 , as in the above subcase .
this implies that the reflection laws at @xmath859 and @xmath1060 are skew , as in the same subcase .
similarly , @xmath139 deformes to @xmath1051 , hence the reflection laws at @xmath861 and @xmath1061 are also skew .
thus , the quadrilateral @xmath139 is as at fig.10b ) .
the proof of theorem [ tclass ] and its addendums is complete .
the pseudo - billiard @xmath41 , @xmath42 , @xmath41 , @xmath42 formed by two orthogonal lines @xmath41 and @xmath42 is 4-reflective of type 1 ) , and its 4-reflective orbits are symmetric rhombi , see fig.10a ) . it is a limit of a 4-reflective pseudo - billiard of type 3 ) on confocal ellipse and hyperbola , as the ellipse tends to a segment and the hyperbola tends to the orthogonal line through the center of the limit segment .
it can be also viewed as a limit of a 4-reflective pseudo - billiard on a pair of confocal hyperbolas ( parabolas ) degenerating to orthogonal lines .
i wish to thank yu.s.ilyashenko , yu.g.kudryashov and a.yu.plakhov , who had attracted my attention respectively to ivrii s conjecture and its analogue in the invisibility theory .
i am grateful to them and to s.v.bolotin , e.ghys , a.v.klimenko , v.s.kulikov , k.a.shramov , b.sevennec , j .- c.sikorav , s.l.tabachnikov , v.a.timorin , r.uribe-vargas for helpful discussions .
i wish to thank a.l.gorodentsev and v.zharnitsky for helpful discussions and very valuable remarks on the text .
klein , f. , _ vorlesungen ber hhere geometrie , _ springer , 1926 .
dritte aufrage , nachdruck 1968 .
plakhov , a. _ exterior billiards .
systems with impacts outside bounded domains , _ springer , new york , 2012 .
vasilev , d. _ two - term asymptotics of the spectrum of a boundary value problem in interior reflection of general form , _ ( russian ) funktsional .
i prilozhen . , * 18 * ( 1984 ) , 113 , 96 ; english translation : functional anal . appl . , * 18 * ( 1984 ) , 267277 . | the famous conjecture of v.ya.ivrii ( 1978 ) says that _ in every billiard with infinitely - smooth boundary in a euclidean space the set of periodic orbits has measure zero_. in the present paper we study the complex algebraic version of ivrii s conjecture for quadrilateral orbits in two dimensions , with reflections from complex algebraic curves .
we present the complete classification of 4-reflective algebraic counterexamples : billiards formed by four complex algebraic curves in the projective plane that have open set of quadrilateral orbits . as a corollary ,
we provide classification of the so - called real algebraic pseudo - billiards with open set of quadrilateral orbits : billiards formed by four real algebraic curves ; the reflections allow to change the side with respect to the reflecting tangent line .
# 1#1 |
intersecting branes at arbitrary angles have been studied both in view of their susy properties in supergravity @xcite and their short - range interactions in string and m(atrix ) theory @xcite . despite the variety of the information on susy properties of branes at angles ,
less has been known about the structure of the supergravity solutions that describe this type of configurations .
the well known examples of these solutions include the 2-angle ( marginal ) configuration of two d2-branes at @xmath0 angles @xcite , and the 4-angle ( non - marginal ) configuration of two ns5-branes at @xmath1 angles @xcite and their generalizations to @xmath2 such branes @xcite ( for some other works in this respect see also @xcite ) .
the solution for the ( marginal ) configuration at @xmath0 angles of two arbitrary similar branes has been recently found in @xcite . nevertheless , there has been no solutions that describe , for example , marginal intersections of similar branes at three or more angles , or those for the non - similar branes at other than the right angles .
a systematic approach to the solution of this problem can be found in the works of @xcite , where such solutions for a _ distributed _ system of branes at angles are considered .
the analysis of these solutions , in the light of a general formulation in @xcite , reveals that , as far as we deal with flat p - branes , such solutions can not be realized in terms of the harmonic functions . that is , in general
, one can not find configurations of branes at angles constructed from parallel distributions of flat p - branes .
the physics underlying this property can be better understood , if one takes into account the role of the brane(s ) interactions and dynamics in forming a configuration at several angles .
in fact , determining the stability conditions for any configuration of p - branes , based on the worldvolume dynamics of the branes , is logically prior to any effort for finding a supergravity background that describes the configuration .
the interactions of d - branes at angles have been studied by calculating the amplitudes in the scattering processes , both in string and m(atrix ) theory , which determine the static potentials between pairs of these objects @xcite . that such a potential _ identically _ vanishes determines the necessary condition for the stability of the configuration .
one of the basic ingredients in such calculations , which makes them at all possible , is the assumption of _ flatness _ for the worldvolume geometries of the individual branes as well as their spacetime background .
however , there may exist configurations for which the the two branes are not flat and so all features of their motion can not be described by a single potential .
for example , they may tend to rotate relative to each other due to the ` twisting forces ' ( or the relative ` torques ' ) between themselves .
so one has to add the conditions that guaranty the balancing of such forces as well .
in fact , the full set of the stability conditions for a configuration of two interacting branes are those that satisfy the equations of motion for both of the branes when they are in _ flat _ states .
doing this , in general , involves extra complexities because of the need for determining the supergravity background for an arbitrarily curved p - brane .
for flat bps branes , however , this problem is considerably simplified by the fact that a bps brane in this state does not suffer from ` self - interaction ' forces .
indeed , in such a case the forces on each individual brane , due to coupling to its own supergravity background , are balanced against each other and hence , it can be studied as a _ probe _ scattered from another brane , which is the _ source _ of a well known supergravity background @xcite .
the aim of this paper is to study and classify possible configurations of branes at angles , assuming the dbi+wz worldvolume action as the branes dynamics . to this end , we first determine the general necessary and sufficient conditions that specify a stable configuration of two arbitrary flat branes in section 2
. then , in sections 3,4,5 , we use these general conditions for categorizing several configurations of two branes at angles for the cases of similar , non - similar and electromagnetic dual branes .
we will consider both the marginal and non - marginal bound states in these sections .
multi - angle marginal intersections are discussed in section 5 , and at the end , we derive physical interpretations for our stability conditions in section 7 .
we end the paper by a summary and some remarks .
as the discussion in the introduction implies , in all cases with flat worldvolume geometries , one can describe a two brane system , equivalently , by the worldvolume action for each of its constituent branes . for simplicity ,
we assume that @xmath3 and we take the @xmath4-brane as a probe moving in the background fields produced by a @xmath5-brane source . to describe this configuration
, we shall use an orthogonal coordinate system whose axes are defined using the tangent and normal directions of the two branes as indicared in table ( 1 ) . in
this table @xmath6 stands for the dimension of the hyperplane spanned by the world directions of the two branes , and @xmath7 denotes the spacetime dimension .
the situation is schematically displayed in figure ( 1 ) .
@xmath8 table ( 1 ) : decomposition of the spacetime coordinates we choose @xmath9 as the parameterizing coordinates of the probe and @xmath10 as its embedding coordinates .
according to the count of coordinates in table ( 1 ) , there are cases for which some of the coordinates in our decompoition are not present .
for example , when @xmath11 , we have no @xmath12 coordinates .
similarly , when @xmath13 ( i.e. , the two branes are parallel ) , there are no @xmath14 and @xmath15 coordinates . however , this has no effect on our general equations as they are formulated in terms of the coordinates @xmath16 and @xmath17 regardless their numbers .
thus , the embedding coordinates for a probe rotated / boosted relative to a fixed source are represented as y^a(x)=^a_x^+y^a_0 [ 1]with @xmath18 s and @xmath19 s representing constant slopes / velocities and shifts .
it is clear from the definitions in table ( 1 ) that ^a_=^i_=^a_=0 [ 2]that is , @xmath20 are constants and none of @xmath21 s depend on @xmath22 .
evidently , the remaining @xmath23 s can not have arbitrary values .
in fact , they have to be chosen in a way that eq.([1 ] ) solves the @xmath4-brane equations of motion : _ ( ) -=0 [ 3]where @xmath24 denotes the relevant worldvolume lagrangian .
that is , @xmath18 s must be such the identities ( ^b _ -^b_a)0 [ 4]hold for all values of @xmath25 . here
, @xmath26 have to be considered as the function @xmath27 , whose @xmath21-dependences turn to appear in the form ( y^a;^a_)=(h ( ) ; ^a _ ) [ 5]where @xmath28 represents a harmonic function in the @xmath29-directions @xcite . taking this into account , and that @xmath30 s are independent coordinates as far as @xmath14 s are
so , one can see that eq.([4 ] ) leads to z^a=0 [ 6](^r _ -^r_s)=0 [ 7]we will refer to eqs.([6]),([7 ] ) as the ` no - force ' and ` no - torque ' conditions , respectively , since they are indeed the conditions for the vanishing of the oscillations along the transverse directions of the two branes , as well as their ( relative ) rotations along the relative transverse directions ( see section 7 ) . by definition , a ` marginal ' configuration of two branes is the one which is stable at any arbitrary separations of the two branes , i.e. , for any _ constant _ values of @xmath31 s .
this , according to eq.([6 ] ) , implies ( h;)_0( ) [ 8]showing that the total potential energy between the two branes has a constant value .
it is obvious that eq.([7 ] ) in this case is automatic and hence the above equation is the unique description of the ` marginal ' configurations .
+ there exist other configurations , however , that can be formed only at zero separations of the two participating branes , i.e. , when the centers of the two branes in their transverse space are coincident and they actually intersect each other .
we refer to these as the non - marginal configurations @xcite . for such solutions , with @xmath32 , eq.([6 ] )
is automatic and hence eq.([7 ] ) or its equivalent as ^r_-^r_s^r_s ( ) [ 9]provides a unique description for the non - marginal configurations . in the following three sections ,
we analyze the eqs.([8]),([9 ] ) for classifying several marginal and non - marginal configurations that may occur in cases with a pair of similar @xmath33 , non - similar @xmath34 and electromagnetic dual @xmath35 branes .
the notations will be mainly identical to those introduced in @xcite .
using the general expression of the lagrangian for a @xmath36-probe in a @xmath36-source background @xcite , the function @xmath37 is found to be = h^-1[det^1/2(+h)-1 ] [ 10]where the angular information is encoded in the matrix @xmath38 having the components ^_^r^r _ [ 11]and the indices of the type of @xmath39 are raised and lowered by @xmath40 .
the matrix @xmath38 can be diagonalized by choosing the coordinates @xmath41 so that the two branes to be related by a set of ` commuting ' rotations and a boost . in such coordinates , @xmath38 has the components ^_=diag(-v^2,tan^2_1,,tan^2_d-1)= : ^ _ _ [ 12]here , @xmath42 represents the magnitude of the velocity vector in the directions transverse to the @xmath36-source and @xmath43 s are the angles of the @xmath36 commuting rotations ( obviously , a number @xmath44 of these angles are vanishing by assumption ) .
inserting the expression ( [ 10 ] ) for @xmath26 into the ` marginality ' condition , eq.([8 ] ) , one obtains det(+h)(1+_0h)^2 ( h>0 ) [ 13]in the basis that @xmath38 has the form of eq.([12 ] ) , this identity reads ( 1-v^2h)(1+tan^2_1h)(1+tan^2_d-1h)(1+_0h)^2 [ 14]thus the only possibility for the rotation and boost parameters is & & _ 1=_2= : + & & v=_3==_d-1=0 [ 15]that is : + the only marginal configuration of two similar branes are the static configurations with two non - vanishing angles obtained by an abelian subgroup of the @xmath0 rotations .
+ this is the same as the 1/4 susy configuration whose supergravity solution had been found earlier in @xcite as a generalization to the solution of 2-branes at angles @xcite .
the constant value of @xmath26 in this case is @xmath45 . +
another possibility , which one might consider for two similar branes , was the brane - anti - brane system . in this case
the constant term @xmath46 in brackets in eq.([10 ] ) would be flipped to ( + 1 ) which is equivalent to flipping @xmath47 to @xmath48 in eq.([13 ] ) or ( [ 14 ] ) . however , because @xmath49 in this case , eq.([14 ] ) could not be satisfied for any real values of @xmath42 and @xmath43 s . in other words ,
a brane - anti - brane pair with any combination of a boost and several rotations is unstable and can not form a marginal configuration .
it is easy to see that the ` non - marginality ' condition , eq.([9 ] ) , after some algebra reduces to ( ^-1)^^r_^s_- ^rsh^-1(-1)=^rs [ 16]where @xmath50 , @xmath51 and @xmath52 are a set of angle - dependent constants . in the basis that makes @xmath38 diagonal ( eq.([12 ] ) ) , the above equation is also diagonal and after some manipulations yields _ ( 1+h_)(1+h_r)^2(1+h_r)^2 ( h>0 ) [ 17]where @xmath53 . for this equation to be satisfied identically , one needs & & _
1=_3 , _ 2=_4 + & & v=_5==_d-1=0 [ 18]that is : + the only ` non - marginal ' configurations of two ` similar ' branes are the static configurations with four angles obtained by two independent abelian subgroup of the @xmath0 rotations . + it is easy to check that , unless one of the @xmath54 vanishes , in this case @xmath26 is not a constant , but it is a linear function of @xmath28 .
orthogonal configurations of the 4-angle intersections have been found previously in the literature of the supergravity composite brane solutions @xcite .
the most famous examples in this category consist of the @xmath55 in @xmath56 ii a , b , @xmath57 in @xmath56 ii b and @xmath58 in d=11 @xcite . in a recent paper @xcite ,
a 4-angle configuration of this type for the ii a ns5-branes at sp(2 ) angles has been found , by directly solving the supergravity and killing spinor equations , showing that it preserves at least 3/32 of the susy ( see however @xcite ) .
the general worldvolume lagrangian for a @xmath4-probe in a @xmath5-source background , with @xmath59 @xcite , gives rise to an expression for @xmath37 as = h^-m/2det^1/2(+h ) [ 19]where @xmath38 is given by eq.([11 ] ) and @xmath60 is a function of dimensions defined as m(d_1,d_2)= 2(d_1)(d_2)+d_1_2 [ 20]for marginal configurations , @xmath60 specifies the number of ( non - vanishing ) angles ( see @xcite and below ) . here , @xmath61 represents the dilaton - d - form coupling constant satisfying @xcite ^2(d)=1-d [ 21 ] the marginality condition , eq.([8 ] ) , in this case gives _ ( 1+h_)_0 ^ 2h^m ( h>0 ) [ 22]obviously , such an identity can be true whenever @xmath62 , and a number @xmath60 of @xmath63 s to be @xmath64 , while the remaining are vanishing .
this implies that & & _ 1==_m=/2 + & & _ m+1==_d_1 - 1=v=0 [ 23]the number of common directions then is @xmath65 .
therefore : + the only ` marginal ' bound state of a pair of ` non - similar ' @xmath66-branes are the ` static orthogonal ' configurations in which the two branes share @xmath67 of their directions , where @xmath44 must satisfy = -2(d_1)(d_2)+d_1d_2 [ 24]this is , in fact , the so called ` intersection rule ' of the intersecting brane systems , which was originally found in the study of their supergravity solutions @xcite , and then re - appeared in the reduced lagrangian approach to the distributed brane systems in @xcite and was interpreted there as an algebraic constraint ) can be viewed as a relation between the gravitational and dilatonic ` charges ' @xmath68 , @xmath69 of the two branes , with @xmath70 being their masses , for the corresponding forces to cancel each other @xcite . ] required by the no - force conditions @xcite .
the ` non - marginality ' condition , eq.([9 ] ) , in this case gives the counterpart of eq.([16 ] ) as h^-m/2\ { h(^-1)^^r_^s_- ^rs } = ^rs [ 25]which in the @xmath38-diagonalizing basis ( eq.([12 ] ) ) takes the form of the identity _
( 1+h_a)_r^2h^m(1+h_r)^2 ( h>0 ) [ 26]which is the analogue of eq.([17 ] ) .
obviously , such an identity holds ( for each @xmath71 ) , if and only if @xmath72 , and further @xmath73 of @xmath63 s are infinity while the others are vanishing .
therefore , & & _
1==_m+2=/2 + & & _ m+3==_d_1 - 1=v=0 [ 27]these are just the conditions ( [ 23 ] ) in which @xmath60 has been replaced by @xmath73 .
so we have the result : + the only ` non - marginal ' bound state of a pair of ` non - similar ' @xmath66-branes are the ` static orthogonal ' configurations in which the two branes share @xmath67 of their directions , with @xmath44 satisfying + 2=-2(d_1)(d_2)+d_1d_2 [ 28]which shifts @xmath44 by a @xmath74 relative to the one given by the rule ( [ 24 ] ) .
this is just the same equation identified as the intersection rule of the ` localized intersections ' in @xcite . by the above derivation
, however , it has to be identified as a rule for the ` non - marginal ' intersections .
an ` electromagnetic ' dual pair of branes , a priori , can not be placed in either of the two categories studied in sections 3,4 .
this is due to the fact that an electromagnetic dual pair with @xmath75-branes is the source of a single @xmath76-form potential @xmath77 , which couples to the two branes through @xmath78 and its dual @xmath79 respectively @xcite . taking this point into account , and assuming @xmath80 to adapt with the conventions of section 2 , one obtains the modified version of eq.([4 ] ) as ( ^n _ -_mn - c_mnh ) 0 [ 29]where @xmath26 is defined in eq.([19 ] ) in which @xmath81 , and the constants @xmath82 are defined as _ mn= _
mnl_1l_d^_1_d ^l_1__1^l_d__d/(d!)^2 [ 30]here , the two epsilons stand for the levi - civita symbols in the subspaces of @xmath83 and @xmath84 , respectively . the modifying term @xmath85 in eq.([29 ] ) , however , does _ not _ modify the results of section 4 for the branes at angles with @xmath86 .
the reason , as can be checked using eq.([12 ] ) , is that in all @xmath87 cases one obtains @xmath88 .
thus the rule ( [ 24 ] ) ( with @xmath89 ) and eq.([21 ] ) imply @xmath90 . hence : + the only marginal configuration of an electromagnetic dual pair of @xmath75-branes is a ` static ' configuration in which the two branes ` orthogonally ' intersect ( overlap ) on a string .
on the other hand , a non - marginal configuration should obey the rule ( [ 28 ] ) , which in this case implies that @xmath91 .
this is in obvious contradiction with the static - ness property asserted above the rule ( [ 28 ] ) which means that such configurations of @xmath92 can not be realized .
the results of the previous sections regarding the non - marginal configurations , in particular their intersection rule ( eq.([28 ] ) ) , are based on the no - torque condition ( eq.([7 ] ) ) which severly relies on the existence of at least one pair of coordinates @xmath41 in table ( 1 ) . since for a pair of parallel branes ( i.e. a configuration of the form @xmath93 ) there are no such coordinates , hence the stability condition for such configurations should have a different form .
whatever the form of these conditons , they should guarantee that the equations of motion ( eq.([4 ] ) ) are satisfied . in the case at hand
all @xmath94 s are zero and this equation reduces to : @xmath95 , where @xmath96 is the same as @xmath97 . noting that @xmath28 in this case depends only on @xmath31 through @xmath98 , and further that @xmath32 in the non - marginal case , the above equation yields : |_r=0=0 [ * ] where @xmath99 . for a pair of non - similar branes we find from eq.([19 ] ) that v(r)=(h(r))^-m/2=(1+q_2)^-m/2 here @xmath100 is the transverse dimension and @xmath101 is ( proportional to ) the charge of @xmath5-brane and @xmath60 is defined as in eq.([20 ] ) .
( we have assumed that @xmath3 and @xmath102 . )
it is obvious that @xmath103 is proportional to the potential energy between the two branes at a separation @xmath71 and eq .
( [ * ] ) indicates that their mutual forces tend to balance each other as @xmath104 .
now , since in this limit @xmath105 , the above condition means that the function @xmath103 should be regular near @xmath106 at least to the first order in its derivatives , requiring that m(m-2)>2 or equivalently _ 1_2>2(d-2 ) -d_1d_2 ^ 2 this inequality together with @xmath3 and @xmath102 in turn specifies all the non - marginal bound states of the form of branes within branes
. special cases of such configurations are those with a self dual pair , i.e. with @xmath107 . since in this case @xmath108 , @xmath81 , the above condition ( assuming @xmath109 ) yields : @xmath110 showing that such configurations are possible only in @xmath111 dimensions .
famous examples of such configurations are @xmath112 in @xmath113 and the dyonic membrane ( bound state of an electric and a magnetic 2-brane ) in @xmath114 dimensions , which were known through their supergravity solutions @xcite .
the results of the previous sections indicate that , except for the 2-angle static configuration of similar branes at @xmath0 angles , no other configuration of two flat branes with several angles and boost can be marginally stable . while , both the scattering amplitude and superalgebra computations @xcite indicate that multi - angle intersections , under certain conditions among their angles , can form a marginal configuration , it is amazing that the worldvolume solutions appropriate to a pair of flat branes are not realizable .
this apparent contradiction is resolved by recalling that all the scattering amplitude and superalgebra computations rely heavily on the basic assumption for the existence of an ` asymptotic state ' in which the branes behave like flat hypersurfaces in a minkowski space . in the string theory language ,
the perturbative calculation of the scattering amplitudes are reliable only in the weak coupling region of the theory where one deals with weak ( linearized ) gravitational interactions .
this restricts such calculations to the ` far ' or ` asymptotic ' region of the two branes where they look like flat hypersurfaces in a flat spacetime . on the susy side , also
, one does not need to solve a killing spinor equation in all of the space to determine susy fraction preserved by a brane configuration . to do this
, it suffices to solve only the asymptotic ( algebraic ) killing spinor equation @xcite , which encodes _ only _ the asymptotic form of the spacetime metric and of the geometry of branes . as a result
, there may be bps states of _ curved _ p - branes which look like asymptotically as multi - angle configurations of flat p - branes . in the asymptotic region
, we will see that the no - force condition to first order will require a certain angular constraint which is just the same that characterizes these bps states @xcite .
however , the same no - force condition to higher orders , as well as the no - torque condition , are not satisfied except for the marginal configurations which were categorized in sections 3,4 .
this means , firstly , that the marginal multi - angle configurations , generally , can not consist of flat p - branes .
secondly , even in the case of an asymptotically force - free configuration , the relative angular position of the two branes is influenced by a non - vanishing torque which eventually brings them together by counterbalancing the forces that act between themselves .
in general , the force and torque conditions , eq.([8]),([9 ] ) , are equivalent to a set of algebraic constraints relating @xmath115 s together .
this can be seen easily by putting @xmath116 and expanding @xmath117 as ( 1+h;)=_n=0 ^ 1_n()h^n [ 32]upon this expansion , eqs.([8]),([9 ] ) give respectively _ n()=0 [ 33]^r__n-^r_s_n=0 [ 34]where @xmath118 . for a ` real ' marginal bound state of flat p - branes , these two sets of equations for all @xmath2 restrict possible configurations to those of sections 3,4 .
however , for a configuration of curved p - branes with asymptotic flat geometries , we may continue to _ define _ the marginality property by demanding that @xmath117 to be constant _ only _ to first order in @xmath119 in the asymptotic region @xmath120 .
such ` asymptotic marginal ' configurations , thus , are distinguished by a condition on the angles as @xmath121 .
this condition , though provides a mean of translational stability , does _ not _ insure rotational stability of the two brane system , which to be guaranteed by the eq.([34 ] ) for @xmath122 . these two conditions together ,
will be seen that , restrict the possible marginal configurations to those obtained in sections 3,4 .
we now examine the explicit expressions of these conditions in the previous cases .
it is easy to see , by expanding eq.([10 ] ) to @xmath123 , that in this case & & _
0()=det^1/2(+)-1 + & & _ 1()=det^1/2(+)\ { 12 tr ( ) -1}+1 [ 35]where @xmath18 dependences are encoded in @xmath38 as is defined in eq.([11 ] ) .
( the expression for @xmath47 is given only for later reference . ) diagonalizing @xmath38 as in eq.([12 ] ) , the @xmath124 equation becomes f()= _
= 0^d-1cos_+ 12_=0^d-1sin^2_-1=0 [ 36]where , for convenience , we have included the velocity @xmath42 in @xmath43 s by defining @xmath125 . solving eq.([36 ] ) for @xmath42 , however , shows that combinations of boost and rotations for less than three non - vanishing angles do _ not _ define allowable configurations .
indeed , one can see using eq.([36 ] ) , that the one- and two - angle configurations are limited only to the parallel and @xmath0-rotated static configurations respectively .
if , in addition to the above condition , one requires rotational stability of the configuration , eq.([34 ] ) for @xmath122 gives = sin_(_cos _ -cos_)=0 [ 37]the only simultaneous solutions of eqs.([36]),([37 ] ) , when more than two @xmath126 s exist , are the one with _ 1=_2 , _ 0=_3==_d-1=0 [ 38]and its permutations for @xmath126 s ( @xmath127 ) .
that is , the only rotationally stable marginal configurations are those with @xmath0 angles . in this case eq.([19 ] ) gives & & _
0()=det^1/2(+ ) + & & _ 1()=12det^1/2(+)\ { tr ( ) -m } [ 39]so , in the basis of eq.([12 ] ) , the @xmath124 condition becomes & & f()= _
= 0^d_1 - 1sin^2_-m=0 + & & -2(d_1)(d_2)+d_1d_2=_=0^d_1 - 1cos^2 _ [ 40]this gives , in fact , a modification of the usual intersection rule , eq.([24 ] ) , to the general case involving arbitrary boost and angles between the two branes .
obviously , an ` orthogonal static ' limit , with @xmath126 s equal to @xmath128 , exists only in cases with @xmath72 .
that is , for an orthogonal intersection , the rule ( [ 24 ] ) must hold and in such a case @xmath60 counts the number of @xmath129 angles . despite eq.([36 ] ) , the eq.([40 ] ) allows for the possibility of the combinations of a boost with any number of angles . specially , when @xmath130 , one can find ( asymptotically ) boosted configurations of two parallel branes having a relative velocity @xmath131 .
however , configurations defined by eq.([40 ] ) are not rotationally stable , unless we have = 2sin_cos_=0 [ 41]which means that @xmath126 s must be @xmath128
. therefore , the rotationally stable marginal configurations are the static ones with orthogonal branes obeying the rule ( [ 24 ] ) .
so far , we have stressed on the fact that every ` equilibrium ' configuration of two flat p - branes is specified by the two requirements , eqs.([6]),([7 ] ) , which we have interpreted as the no - force and no - torque conditions respectively . in this section ,
we try to make this correspondences explicit by perturbing the worldvolume of one of the branes ( the probe ) with respect to its flat state , while the other brane ( the source ) is kept flat and fixed .
this type of description is similar to the one which is used in @xcite . to be concrete ,
we first present a perturbative formulation of the classical solutions for a general field theory with one expansion parameter , and then consider its application to the worldvolume field theory of the probe .
assume that a field theory , for the variable(s ) @xmath132 , is defined by means of the perturbative lagrangian = _
0[]+_1[]+ ^22_2[]+ [ 42]with @xmath133 being the perturbation parameter . now , take @xmath134 to represent a classical solution of the unperturbed lagrangian @xmath135 $ ] , i.e. , it satisfies the equation of motion : @xmath136 .
the question , which we like to answer , is that how can we specify a solution of the perturbed lagrangian which tends to @xmath134 when @xmath137 .
obviously , such a solution must obey the expansion ( x;)=_0(x)+_1(x)+^22_2(x)+ [ 43]putting this expansion in eq.([42 ] ) , and using the functional taylor series expansion of @xmath138 s , the overall expansion of @xmath139 $ ] in powers of @xmath133 takes the form s are in the form of local operators , acting on the functions which are multiplied by themselves , summed over the field indices and are finally integrated over the spacetime coordinates .
] = _ 0[_0 ] + ( _ 0_1+_1)+ ^22(_0_2 + ^2_0_1 ^ 2 + 2_1_1 + _ 2)+ [ 44]where all @xmath138 s and their functional derivatives are evaluated at @xmath140 . treating @xmath141 as a set of independent variables , and varying @xmath26 with respect to these variables , one picks the set of equations & & _
0=0 + & & ^2_0_1+_1=0 + & & ^2_0_2 + ^3_0_1 ^ 2 + 2 ^ 2_1_1 + _
2=0 , etc .
[ 45]these equations , in principle , determine the solutions for @xmath142 in the successive order .
it is worth - pointing that , except for @xmath134 which is assumed to be a given solution of @xmath47 , all other @xmath143 s are determined by solving a linear inhomogeneous pde of the form _
n=_n [ 46]where the source term @xmath144 in the @xmath2-th stage is a known function of @xmath145 , constructed from @xmath146 .
for vanishing boundary conditions on @xmath143 s , one solves eq.([46 ] ) symbolically as @xmath147 , with @xmath148 representing the green s function of the linear operator @xmath149 at @xmath140 .
if @xmath47 , as in usual , is a first order lagrangian in terms of the @xmath150-derivatives , then @xmath149 will be a second order linear differential operator with ( in general ) spacetime dependent coefficients , whose green s function is constructed in the usual manner .
the above procedure , thus eventually , determines @xmath151 to any arbitrary order in the expansion parameter @xmath133 .
we consider a source - probe configuration of arbitrary branes , as in @xcite , and choose the embedding of the probe , as in section 2 , to be represented by @xmath152 .
it is clear that the worldvolume lagrangian of the probe has a generic form as = ( h(y^a),_y^a ) [ 47]where the explicit dependences on @xmath21 are encoded in the harmonic function @xmath153 of the transverse distance from the source , which is proportional to its charge or tension @xmath154 , and vanishes asymptotically .
obviously , taking the limit @xmath155 is equivalent to going to the asymptotic region of @xmath21 , where the source and probe have a large separation , and in this limit eq.([47 ] ) takes the form of an ordinary lagrangian of a minimal surface in the minkowski space . taking this as the unperturbed lagrangian @xmath47 , and treating @xmath154 as a perturbation parameter
, the perturbed lagrangian @xmath26 will be expanded as = _
n=0 ^ 1h^n(y)_n(y ) [ 48]which is nothing but the eq.([32 ] ) with @xmath156 s replaced by @xmath157 s . also , it is clear that @xmath47 has classical solutions which are in the form of flat hypersurfaces , similar to the one in eq.([1 ] ) , which plays here the role of the ` unperturbed solution ' @xmath158 .
the main advantage of this choice for @xmath158 is that it renders @xmath159 $ ] s as constant parameters @xmath160 .
this means that we are interested in a probe whose worldvolume geometry , in the region far from the source , is that of a flat hypersurfaces , though it may be curved in the near region .
thus , perturbing the flat solutions as @xmath161 , and using the general formulation of the previous subsection , one finds that the first order perturbation , @xmath162 , obeys the equation _ _
y^b_1 + ( ^a__1-_ab_1 ) _ bh(y_0)=0 [ 49]the eq.([49 ] ) represents a system of second order pde with the _ constant _ coefficients @xmath163 and source terms which are linear combinations of @xmath164 .
now , using the expression ( [ 35 ] ) or ( [ 39 ] ) for @xmath165 , one can show that = _ 0\ { ^ab^+^a_c^b _ ( ^c^-^c^-^^c ) } [ 50*]where @xmath166^{\a}_{\ \b}$ ] . inserting this into the eq.([49 ] ) , we obtain i_ab^__y^b_1 + 1 ( ^a__1-^ab_1 ) _ bh(y_0)=0 [ 51*]where @xmath167 .
it is easy to see , using eq.([2 ] ) , that the above equation for @xmath168 ( as defined in table ( 1 ) ) is decomposed into the three _ uncoupled _ equations & & ^a _ _ y^i_1=0 [ 50 ] + & & ^a _ _
z^a_1 -_1_ah(y_0)=0 [ 51 ] + & & i_rs^a__y^s_1 + 1 ( ^r__1-_rs_1 ) _ sh(y_0)=0 [ 52]where we have used the fact that @xmath169 .
the operator @xmath170 appearing above is indeed the d alembertian operator along the probe s worldvolume coordinates , as can be seen by the eq.([12 ] ) , ^a _ _
= -1_0^ 2+cos^2_1_1^ 2++cos^2_p_p^ 2 [ 52*]thus the eqs.([50])-([52 ] ) are written as the equations describing the propagation of waves along the probe with or without external sources .
perturbations in the @xmath12 directions propagate as free waves , as is expected by the homogeneity of the space along these coordinates , while those in @xmath171 and @xmath15 directions propagate as the forced oscillations .
the eq.([51 ] ) , describing the transverse oscillations , resembles a force equation while eq.([52 ] ) , describing the longitudinal oscillations , is the reminiscent of a torque equation . in the equilibrium conditions , with @xmath172 , eqs.([51]),([52 ] ) reduce to the eqs.([33]),([34 ] ) for @xmath122 component .
these same conditions , for higher order @xmath173 s , reproduce the higher @xmath2 components of the eqs.([33]),([34 ] ) , and finally one recovers the eqs.([8]),([9 ] ) . as a result ,
our perturbative approach provides physical interpretations for eqs.([8]),([9 ] ) as the balancing conditions of the force and torque , respectively .
this paper categorizes several configurations of two arbitrary branes at angles which are derivable from the dbi+wz action for p - branes . in using this dynamics for p - branes
, we have implicitly assumed that all types of internal gauge fields of the branes as well as the background @xmath174 field are vanishing . in this way all types of p - branes ( e.g. , ns- , d- and m - branes ) are treated in a similar manner by using the dynamics of dbi+wz action .
further , we have assumed that neither of the two branes is affected by the fields that originate from itself , at least when it is stretched as a flat hypersurface ( the bps or no - force condition for single branes ) .
thus , the wz term contribute to the dynamics , when we deal with ` similar ' branes carrying the same ( p - form ) charges , and it is vanishing when the the branes are ` non - similar ' carrying different charges . under these assumptions ,
the analysis for both of these cases reveals two types of configurations : the marginal and the non - marginal ones . in the marginal case
, we found that the only configuration of similar branes at angles is the one with two angles in a subgroup of @xmath0 , while the only one for non - similar branes is an orthogonal configuration obeying the ordinary intersection rule @xcite . in this case , no configuration with more than two angles can be found @xcite . in the non - marginal case
, on the other hand , we saw that the only configuration with similar branes is the one with four angles in two independent subgroups of @xmath0 , while the only one for non - similar branes is an orthogonal configuration obeying an unusual rule of intersection , previously identified as the localized branes intersection rule @xcite .
while , the whole analysis in this paper considers only two brane configurations , the @xmath175 brane configurations can also be undertaken by a similar analysis , provided one knows the background fields for each @xmath176 of these branes .
though , for marginal configurations , this does not seem to give additional information other than those for pair - wise intersections , it may be a useful device for investigating about the non - marginal configurations made up of several branes .
it should be emphasized here that _
marginal _ multi - angle configurations , other than those stated in the above which all had been found in the context of the classical supergravity solutions , no other marginal configuration can be constructed from a set of _ flat _ p - branes .
this is why the solutions of this kind had not been discovered in the supergravity solutions literature .
( this , however , does not prevent the possibility of having non - marginal solutions with several angles . )
it is tempting to ask that whether one can form marginal multi - angle configurations by putting suitable curvatures on their worldvolumes .
of course , in such a case one has to define rigorously the ` marginality ' property . in the case of asymptotic flat p - branes
, we defined it as the stability of the configuration at _ arbitrary _ separation of the two branes as measured in their _ asymptotic _ region .
we have answered the above question , only partially , by perturbing the worldvolume of one of the branes when it is posed to the background of the other ` unperturbed ' brane , and find that the marginality requirement , in general , breaks the conditions for the flatness of the worldvolume .
a general treatment requires propagating both of the branes and looking for the conditions that need to be asymptotically flat .
+ + i would like to thank h. arfaei for fruitful discussions and m.m .
sheikh - jabbari for careful reading of the paper and valuable suggestions . | in this paper we investigate about several configurations of two intersecting branes at arbitrary angles .
we choose the viewpoint of a brane source and a brane probe and use the low - energy dynamics of p - branes . for each p
- brane this dynamics is governed by a generic dbi action including a wz term , which couples to the sugra background of the other brane .
the analysis naturally reveals two types of configurations : the `` marginal '' and the `` non - marginal '' ones .
we specify possible configurations for a pair of similar or non - similar branes in either of these two categories . in particular , for two similar branes at angles
, this analysis reveals that all the marginal configurations are specified by @xmath0 angles while the non - marginal configurations are specified by @xmath1 angles . on the other hand
, we find that no other configuration of two intersecting branes at non - trivial angles can be constructed out of flat p - branes .
so in particular , two non - similar branes can only be found in an orthogonal configuration . in this case
the intersection rules for either of the marginal or non - marginal configurations are derived , which thereby provide interpretations for the known results from supergravity .
-15 mm 0 mm 0 mm 0 mm 0 mm 160 mm 230 mm .65 mm 0.5 cm * r. abbaspur * + .25 in _ institute for studies in theoretical physics and mathematics , + p.o .
box 19395 - 5531 , tehran , iran .
+ department of physics , sharif university of technology , + p. o. box 19365 - 9161 , tehran , iran . _ 0.5 cm : = c v |
ternary chalcopyrites are promising for optical frequency conversion applications in solid state laser systems , such as optic parametric oscillators ( opo ) and frequency doubling devices @xcite .
zinc germanium diphosphide ( zngep@xmath0 ) is an excellent nonlinear optical material which exhibits good optical transparency over the 0.712 @xmath5 m wavelength region .
this uniaxial crystal has positive birefringence , and its conventional conversion efficiency , or figure of merit ( fom=@xmath6 where @xmath7 is the index of refraction ) exceeds that of linbo@xmath8 , a presently commonly used material , by an order of magnitide .
it is therefore , a very good candidate to producing tunable laser output in the near infrared @xcite .
cdgeas@xmath0 has even larger @xmath3 and fom .
to the best of our knowledge this material has the highest nonlinear optical coefficient in the class of phase - matchable compounds @xcite .
recent crystal growth technologies have made great progress towards growth of large crystals of both zngep@xmath0 and cdgeas@xmath0 with improved optical quality .
these two materials , however , are part of a larger family of ii - iv - v@xmath0 compounds and although the crystal growth perfection of these has not been pursued to the extent of the above two , it is of interest to ask whether other materials in this class , or alloys between them might offer increased flexibility in terms of combining high @xmath3 with other ranges of transparency and/or non - critical phase - matching .
more generally , it is of interest to understand the origin of the high @xmath3 and the degree of birefringence in these materials as well as to study the trends within this family of compounds . in the present paper
we systematically study the electronic structure and optical properties of a class of ternary pnictides with formula abc@xmath0 ( a = zn , cd ; b = si , ge ; c = as , p ) which are crystallizing in the chalcopyrite structure .
the measured shg coefficients have been reported for some of these materials , and their values are generally rather high ( at least not much lower than the shg in gaas ) @xcite . our main purpose is to understand the trends in this family of materials and to check whether simple schemes for extrapolation ( e.g. , based on inverse power relations to the band gap ) to other compounds in this family are valid .
the choice of a , b , and c atoms allows us to consider the trends in the group - ii , group - iv , and group - v atoms separately in this class of materials which has the general formula ii - iv - v@xmath0 .
we also note that some of these materials have direct gaps while others have indirect or pseudodirect gaps .
thus , we can investigate whether there is any relation between the type of the gap and the @xmath3 values .
in fact , we find little evidence for such a correlation .
this result could be expected from the fact that @xmath3 depends on a brillouin zone average of interband vertical optical transitions and is , therefore , little affected by the bands at a few particular * k*- points determining the band edges .
typically , one indeed finds smaller band gap materials to have higher @xmath3 @xcite .
it is expected from the basic expressions for the nlo response functions @xcite and from empirical observations that nlo susceptibilities scale roughly inversely with some power of the interband energy differences , and , therefore , with the minimum gaps if these dominate the response .
however , we show that the situation is not that straightforward , and the real value of the zero frequency limit of shg is a result of a very delicate balance between different contributions to nonlinear optical response , e.g. , the intraband and interband contributions .
it turns out that cdgeas@xmath0 is a rather exceptional material in this context .
a second purpose is to investigate whether or not the chalcopyrite structure plays a special role in leading to high @xmath3 values .
we can test the effect of crystal structure at least qualitatively by considering an alternative crystal structure corresponding to the same overall chemical formula or stoichiometry . to this end
we consider a ( 001 ) oriented @xmath4 superlattice structure and compare its band structure , total energy , and @xmath3 with that of chalcopyrite for zngep@xmath0 .
we note that this particular structure still maintains the local tetrahedral environment of the chalcopyrite corresponding to each anion being surrounded by two cations of each type .
our results show that the alternative structure considered generally has a much lower band gap and hence higher @xmath3 .
however , we find that this is related to the occurence of alternatingly compressive and tensile biaxial strain in the layers which make the structure unstable .
thus , we do not think that this alternative structure gives a promise for further increases of @xmath3 .
the chalcopyrite structure is typically accompanied by a distortion from the ideal @xmath9 ratio one would obtain by simply substituting ii and iv atoms in the cubic zincblende structure in the appropriate ordering .
it is thus also of interest to investigate whether this has an important effect on @xmath3 .
we find that these effects are significantly larger for the cd than for the zn compounds which are closer to the ideal structure . while the gaps are generally increased and hence , both intra- and interband contributions are significantly decreased , the net value of @xmath3 is only slightly changed as a result of the compensation noted above .
the computational approach used in the present investigation was described in our previous paper@xcite and successfully applied to sic polytypes @xcite and gan / aln superlattices @xcite .
the rest of the paper is organized as follows . in section [ sec : comp ] we describe briefly the necessary details of our computational approach . in section [ sec : res ] we present our calculated results for the electronic band structures , static second harmonic generation coefficients and birefringence and address the various questions raised in this introduction .
we analyze the results in terms of the decomposition of the frequency dependent im@xmath10 function on intra- and interband terms as defined in refs . and .
a summary of the main conclusions of this work is given in sec .
[ sec : con ] .
our calculations are based on the linear muffin - tin orbitals method ( lmto ) @xcite within the atomic sphere approximation ( asa ) .
this method is quite efficient mainly because it employs a rather small basis set .
it allows to easily deal with systems containing a large number of atoms per unit cell while maintaining a sufficiently large number of * k*-points so as to ensure converged brillouin zone ( bz ) integrations .
extensive checks performed in our previous paper @xcite demonstrated that our asa - lmto based approach yields results for the second order response functions of comparable accuracy as the full - potential linear augmented plane - wave method ( flapw ) method@xcite .
the formulas used here to calculate @xmath3 were given in a previous publication.@xcite they are based on the so - called `` length - gauge '' formalism of sipe and ghahramani,@xcite and aversa and sipe@xcite .
this formalism uses the independent particle approximation and is presently restricted to undoped semiconductors .
also , it does not include the local field effects .
this method has several apparent advantages , e.g. , the manifest absence of unphysical singularities of nonlinear optical responses in the zero - frequency limit and the explicit satisfaction of the kleinman relations.@xcite the self - consistent lmto electronic structure calculations were carried out within the framework of the local density approximation ( lda ) in the density functional theory .
the exchange - correlation potential has been taken in the form of hedin and lundqvist.@xcite one of the central problems in calculation of optical response functions in semiconductors is the problem of the gap corrections in the lda @xcite which appears from the fact that response functions have to deal with the actual quasiparticle excitations rather than the kohn - sham eigenstates .
the presently most accurate approach , the gw method , is unfortunately rather computationally demanding because it needs calculations of matrix elements of the nonlocal self - energy operator in terms of the one particle green s function and the screened coulomb interaction which requires itself calculation of the inverse dielectric response function including local field effects.@xcite in practice , a simplified approach , the so - called `` scissors operator '' is often used .
it assumes the rigid shift in energy @xmath11 of the conduction band with subsequent renormalization of the velocity ( momentum ) operator matrix elements ( see , e.g. , refs . for details ) . in our previous paper
@xcite we noted that this approach is not entirely satisfactory because it breaks the consistency between the eigenvectors and eigenvalues of the one particle hamiltonian .
this manifests itself rather clearly when applying this approach to the calculation of effective masses . in a rigid shift ,
the masses should not change , but applying the `` scissors '' renormalization to the momentum matrix elements and using the well - known @xmath12 expression for the inverse masses we see that the effective mass decreases .
we proposed an alternative approach , which consists in adding semi - empirical corrections to the diagonal elements of the lmto . in some sense
, it is an attempt of describing the scissor s operator @xmath13 their shifts , explicitly at the level of the hamiltonian .
the approach is based on the observation that the states at the bottom of the conduction band in tetrahedrally bonded semiconductors have a characteristic predominance of certain lmto basis orbitals in their wave functions : e.g. , cation @xmath14- at @xmath15 and a mixture of cation @xmath14- and empty sphere @xmath14- orbitals at the @xmath16- point of the zincblende bz .
therefore , by shifting these @xmath14- orbitals , one shifts the corresponding states .
the states at @xmath17 typically behave intermediately between those at @xmath15 and @xmath16 .
thus , an advantage is that one can include * k*- dependent shifts if these are known .
this alternative approach is closely related to the one suggested by christensen @xcite in which it is further used that only @xmath14- states are sensitive to sharply localized @xmath18-function - like potentials . in our previous paper @xcite
we showed that better results for the frequency dependent shg ( at least for gaas and gap ) can be obtained in this way .
however , this procedure allows us only to shift the lowest conduction bands while for some materials the shift of the whole set of conduction bands is a better approximation ( e.g. , for sic , see ref . for details ) .
in other words , it is a reasonable approach only if one is interested in the highest valence band and the lowest conduction band exclusively .
its success for gaas and gap observed in our previous work is thus based on the fact that these bands tend to dominate the second order response functions in that material but this is not a general fact for all semiconductors .
in the present study , in order to study trends , it is important to follow a systematic and consistent treatment for all the compounds , including those for which detailed band gap information at different * k*- points is not available .
therefore , the bulk of our results was obtained using the simple `` scissors operator '' approach .
unfortunately , we can not use that approach directly for the case of cdgeas@xmath0 ( where lda does not give any gap at all ) . as a result ,
the bands in cdgeas@xmath0 have a completely changed topology near the `` overlap '' of the valence and conduction bands at @xmath15 . to deal with this case we first applied a shift of the hamiltonian @xmath14- states to get some `` bare minimum '' gap of at least 0.3 ev and subsequently applied the usual scissors operator to obtain the experimental gap values .
we emphasize that the primary purpose of this paper is to study the chemical and structural effects on the electronic structure and @xmath3 functions .
the above described problems with the gap corrections , although not handled to our entire satisfaction , are expected to only have a minor effect on the main conclusions of our work as long as a consistent treatment is followed . the detailed description of the procedure for calculating the linear response and frequency dependent shg has been described elsewhere ( see , e.g. , ref . ) .
the real part of the frequency dependent response functions has been obtained from the imaginary part by using the the kramers - kronig transformation .
for the zero - frequency limit of shg we use the simplified and computationally more convenient formulas .
it avoids the need for calculation of the response function in a wide frequency region required for an accurate application of the kramers - kronig transformation and is less singular as regards * k*- point summations , so it can be converged with fewer terms .
the * k*- space integration may be limited to the irreducible wedge of the brillouin zone by performing a prior symmetrization of the product of the three momentum matrix elements over all the transformation of the point group of the crystal . for the frequency dependent shg
, we use the usual tetrahedron scheme of the integration with linear interpolation of the band energies and the products of the matrix elements .
for the zero - frequency limit on the other hand , we employ a semi - analytical linear interpolation scheme which is more efficient and produces a smaller error.@xcite
the chalcopyrite abc@xmath0 materials ( with chemical composition ii - iv - v@xmath0 and i - iii - vi@xmath0 ) can be obtained from iii - v compound and ii - vi compounds respectively by replacing every two group iii ( ii ) atoms per cell by a ii and iv ( i and iii ) atom .
for example , from gap we derive zngep@xmath0 by this chemical substitution .
their crystal structure is a body - centered tetragonal lattice which has eight atoms per unit cell .
it can be thought of as a tetragonally distorted a@xmath0b@xmath0 ( or 2 + 2 ) face centered cubic ( fcc ) superlattice in the \{201 } direction of the cations a and b with an interpenetrating fcc lattice of common anions c displaced by ( 1/4,1/4,@xmath19 ) .
the structural parameters are the lattice constant @xmath20 which corresponds to the cubic lattice constant of the zincblende structure from which the chalcopyrite structure is derived , the ratio @xmath21 , and the internal displacement parameter @xmath19 . in the ideal structure @xmath22 = 2 and @xmath19 = 1/4 .
the nonideal value of @xmath19 is due to the distortion of the anion sublattice involving a shift by each anion away from one neighboring cation in the direction of the another cation ( of different sort ) .
it was shown that such a shift of anions is a consequence of atomic sizes @xcite .
usually the column - ii atom is larger than the column - iv one , and the anions are closer to b than to a atoms . in principle , one can obtain the @xmath20 , @xmath22 and @xmath19 parameters by total energy minimization .
it was shown in a previous paper by one of the authors that such a relaxation can significantly affect the band structure @xcite .
we did not do such a relaxation in the present work but preferred to use the available experimental values of these structural parameters .
some comparisons to the ideal structure were also carried out .
table [ tab : str ] shows experimental values of the lattice parameters @xmath20 , @xmath22 , and @xmath19 for several chalcopyrites abc@xmath0 from the considered group .
the most complete set of parameters can be found in refs . and .
the early measurements of ref . did not pay too much attention on detailed measurements of the parameter @xmath19 but the measured values of the parameters @xmath20 and @xmath22 are very close in both the references . below
we show that the distortions from the ideal chalcopyrite structure ( @xmath22=2 , @xmath19=1/4 ) can be neglected only when they are small .
otherwise , they can change significantly the value of the gaps and the shg .
it would be interesting to mention that the substitution of phosphorus by arsenicum increases the lattice constant @xmath20 by the factor of 3.54 % while the value of @xmath21 changes by less than 1 % .
the substitution of ge by si does not change either @xmath20 or @xmath22 by more than 1.5 % .
table [ tab : shg ] shows the experimental values of the minimum band gaps in these materials from ref . compared to our lda gaps and their difference in the first three columns .
the nature of the band gaps varies from direct to pseudodirect or indirect . by pseudodirect
is meant that the gap is nominally direct but corresponds to a weak , almost forbidden , optical transition . in the chalcopyrites ,
this situation can occur because the zone edge states of the parent iii - v zincblende compound at @xmath23 are folded onto the @xmath15-point in the chalcopyrite structure .
thus , when the zincblende @xmath24 state is lower than the direct gap at @xmath15 ( which happens in gap ) , a pseudodirect gap is expected in a chalcopyrite corresponding to a small perturbation from gap . the correspondence between states in chalcopyrite notation and
the states in zincblende is as follows : the conduction band minimum at @xmath16 , @xmath25 , the next higher state at @xmath16 , @xmath26 and @xmath27 .
the latter is predominantly s - like while the other two have important p - contributions .
the optical matrix elements between the crystal - field splitted valence band maxima @xmath28 or @xmath29 , which are both p - like , and the @xmath30 conduction band state are large while the ones with @xmath31 and @xmath32 are small . by inspection of the band dispersion , or the matrix elements , or the eigenvectors , one can easily identify the nature of the lowest conduction band at @xmath15 as being @xmath30 or @xmath31 like .
our calculated gaps indicted as being pseudodirect in fact correspond to the @xmath31 minimum while the direct ones correspond to a @xmath30 minimum .
the experimental assignments are mainly based on the optical absorption being weak or strong @xcite along with some general theoretical considerations based on zone folding arguments of the type explained above . experimentally , it is very difficult to distinguish between a pseudodirect and a truly indirect gap .
although all the lda gaps are lower than the experiment by about 1 ev , we note that the assignments of the direct or pseudodirect nature agree with experiment with two exceptions .
we find zngep@xmath0 to be actually indirect with a conduction band minimum location at a point near to the folded @xmath16-point .
this will be further discussed elsewhere @xcite and is still in agreement with the absorption measurements indicating a weak transition .
also , we find znsias@xmath0 to be direct while experiment indicates a pseudodirect gap .
we note however , that in this case , the @xmath31 pseudodirect gap as well as the lowest indirect gap in our calculation are only 0.1 ev above the direct @xmath30 gap . errors of the order of 0.1 ev in the ordering of the conduction band states are common , a well - known example being ge , for which it was found that subtle effects such as core - polarization @xcite need to be included to obtain the correct ordering .
we note that if we order these semiconductors according to increasing gap , the lower gap ones are direct , the higher gap ones are pseudodirect ( or indirect ) with the crossing taking place at about 1.7 ev , i.e. , cdgep@xmath0 and znsias@xmath0 being borderline cases . from the above discussion relating the bands to those of the parent iii - v compounds , the pseudodirect situation is clearly expected in znsip@xmath0 , because zn is in the same row of the periodic table as ga and the si s - state occur higher than the si - p states ( @xmath33 ) .
thus one expects the @xmath30 state to be mainly zn- like , and a a small perturbation from gap is expected . in zngeas@xmath0 , on the other hand , we clearly expect a truly direct ( @xmath30-like ) gap because gaas is direct .
next , we note that cd being a heavier element has lower s - like states than zn and thus replacement of zn by cd will tend to make the gap direct as opposed to pseudodirect .
ge also has a lower @xmath30 state than si . thus replacing si by ge
will also tend to make the gap direct .
this is consistent with the borderline case cdgep@xmath0 being direct . also , in zngep@xmath0 , we find the @xmath30 state slightly below the @xmath31 but nevertheless it is indirect with a minimum gap at another * k*- point . again , in this case the difference between direct and indirect or pseudodirect gaps is of the order of 0.1 ev only . in conclusion ,
the nature of the gaps is generally well described by our lda calculations .
next , we consider the gap correction from the lda .
we note that both the zn and cd compounds have a gap correction @xmath11 of about 1 ev ( @xmath11 varies from 0.83 ev for znsias@xmath0 to 1.13 ev for cdsias@xmath0 ) .
this is consistent with the gap corrections of gap ( 1.2 ev),@xcite gaas @xcite ( 0.9 ev ) , and zn - vi compounds @xcite ( 1.21.6 ev ) .
other aspects of our band structures not discussed here are in general agreement with those of previous studies,@xcite for some of these compounds although there are some differences in the location of the zn3d band .
thus , apart from the gap corrections , which for simplicity we take from experiment and apply as a `` scissors '' shift to all conduction states , our band structures seem to describe the electronic structure of these systems rather well . in any rate ,
the optical response functions considered in this paper are not very sensitive to the fine band structure details such as changes from direct to indirect at the level of 0.1 ev , because they are derived from integrals over the whole brillouin zone .
also , the contributions of the zn or cd d - bands in the optical transitions determining @xmath3 are quite negligible being far too deep although they do have an indirect effect on the lattice constants and the states near the valence band maximum via their hybridisation with the d - bands . for practical applications in shg and opo s
an important quantity is the birefringence because it enters in the phase - matching condition .
it can be calculated from the linear response functions from which the anisotropy of the index of refraction is obtained .
figure [ fig : eps_comp ] shows the imaginary part of the dielectric function ( df ) @xmath34 for zngep@xmath0 and cdgeas@xmath0 in a wide energy region and for different light polarizations . in this subsection we concentrate exclusively on these two materials because of recent experimental measurements of high accuracy of frequency and temperature dependence of birefringence .
@xcite in general , the @xmath34 s are rather similar in these materials , except for the difference in the gap value .
it consists mostly of one broad hump with a few additional fine structures which depend on the light polarization .
the birefringence is the difference between the extraordinary and ordinary refraction indices , @xmath35 , where @xmath36 is the index of refraction for an electric field oriented along the c - axis and @xmath37 is the index of refraction for an electric field perpendicular to the c - axis .
it is positive for both the materials zngep@xmath0 and cdgeas@xmath0 . in the low energy region (
@xmath38 0.1 ev ) phonon absorption is starting to play an important role , and this was clearly observed in the temperature dependence of the birefringence in refs .. we can not study the temperature dependence in the present publication . however , when the temperature is changing between 14 k and 450 k , the change of the birefringence at a given frequency is not larger than 1015 % of its value at room temperature , which is about the error of our present estimation which is limited in its accuracy among others by the uncertainties in the gap corrections and by the neglect of local field effects .
figure [ fig : bir_comp ] shows @xmath39 in a wide energy region . of course , in practice the birefringence is important only in the non - absorbing region , i.e. , below the gap .
the presense of an absorption makes the use of the nonlinear crystal in opo s or frequency doubling device quite difficult or impossible .
however , such a curve is illustrative to show the general aspects of the difference of the df for different polarization .
one may note that the general shape of the curves for zngep@xmath0 and cdgeas@xmath0 is rather similar , indicating the same frequency regions where the @xmath40 functions are enhanced or decreased in one polarization or the other .
this , of course , is due to the similarities in their underlying band structures at the ev scale .
nevertheless , we can recognize a stronger negative peak around 2.53.0 ev in cdgeas@xmath0 than in zngep@xmath0 and a higher positive value in the low energy region .
the birefringence of zngep@xmath0 and cdgeas@xmath0 in the non - absorbing energy region together with experimental data is shown in figures [ fig : bir_zgp ] and [ fig : bir_cga ] . as one could expect the calculated curve
is growing quadratically with energy ( in the experiment @xmath39 starts to grow again at low energies due to the phonon absorption ) .
the curve is continuing to grow with energy below the gap reaching a non - monotonic behavior when the interband absorption processes start . for both zngep@xmath0 and
cdgeas@xmath0 we predict fairly accurate values of the birefringence to within about 10 % in the middle of the gap region .
it correctly accounts for the fact that it is about twice larger in cdgeas@xmath0 than in zngep@xmath0 .
for zngep@xmath0 the numerical values are in quite reasonable agreement with experiment in all the non - absorbing region . for cdgeas@xmath0
the agreement with measurements is good in a low energy region ( 0.10.3 ev ) but at higher energies the experimental curve starts to grow faster than the theoretical one .
the reason for this disagreement can be different , e.g. , ( i ) the measurements could be prepared on a crystal which is not perfect and contains impurities and cracks which increase the scattering ; ( ii ) the excitons ( non included in our calculations ) could be important in this materials and give a significant contribution in refraction index near the band gap edge ; ( iii ) our calculations of the dielectric function for cdgeas@xmath0 might be not accurate enough . therefore , our calculations of birefringence should be considered as qualitative estimates only .
nevertheless , it is gratifying to see that even such an estimation exhibits the correct trends for the dependence of the birefringence on the chemical composition in these two chalcopyrites .
this shows that our calculations will be predictive for other chalcopyrites and may be used for preliminary investigations of the phase - matching conditions in these and related materials .
in this section we focus on the results for the shg coefficients in the zero frequency limit .
for the material with the point group 42@xmath41 there are only two independent component of the shg tensor , namely , the 123 and 312 components ( 1 , 2 , and 3 refer to the x , y , and z axes respectively , which are chosen along the cubic axes ) . in the static limit
, these two components are becoming equal according to kleinman `` permutation '' symmetry , @xcite which dictates additional relations between tensorial components beyond the purely crystallographic symmetry .
for a non - ideal chalcopyrite structure ( when @xmath19 @xmath42 1/4 ) the point group symmetry is distorted and some additional components of shg ( e.g. , 311 or 131 ) appear .
however , direct numerical calculations show that these additional components are not larger than 35 % of the main 123 component for the values of @xmath19 corresponding to real chalcopyrites .
therefore , we ignore them in the bulk of our results and make all the calculations for the 42@xmath41 point group .
for the considered ternary chalcopyrite semiconductors including the orbitals with angular moments @xmath43=3 in the basis set changes the results by the factor of 510 % only . in most of our calculations
we neglect them including orbitals with moments up to @xmath43=2 .
in addition to the gaps discussed earlier , table [ tab : shg ] shows the values of the calculated and measured ( where available ) shg in the zero frequency limit in abc@xmath0 compounds ( a = zn , cd ; b = ge , si ; c = as , p ) .
we recall that all values were calculated using the `` scissor '' approach with a rigid shift of the conduction bands chosen such that the minimum gap is adjusted to the experimental value .
first , we discuss the trends with chemical element .
one can see that substitution of p by as keeping the other elements fixed increases @xmath3 in all cases .
this is also consistent with the values of gaas ( 105 pm / v ) and gap ( 48 pm / v ) calculated using the same approach ( lda+scissors ) in ref .. similarly , substitution of si by ge and zn by cd increases @xmath3 .
second , one may notice that this is correlated with the band gaps .
substitution of as by p , or ge by si , or cd by zn , each increase the gap and decrease @xmath3 .
however , these qualitative trends can not easily be turned into quantitative scaling laws .
if one more carefully considers the ratios of the gaps and the ratios of @xmath3 for each of the above substitutions , no inverse power law between gaps and @xmath3 can be extracted in the remaining four element combinations .
for example , for substitution of as to p , the ratios of the gaps for znge , znsi , cdge and cdsi are 1.8 , 1.2 , 3.0 and 1.4 , respectively , but the inverse ratio of the @xmath3 s are 1.8 , 1.7 , 4.0 and 1.9 . on the basis of the theoretical expressions,@xcite one might have guessed a scaling with the third power of the gap because three energy denominators occur .
clearly , such a strong scaling with the gap is not present .
the reason for this is that there are several terms with positive and negative contributions which to a great extent cancel each other and secondly , the @xmath3 is not dominated by the terms involving the minimum gap but arises from a * k*- point summation over the whole brillouin zone and a sum over several band - to - band terms .
thus , no simple scaling with the gaps emerge .
the smallest value of the gap occurs for cdgeas@xmath0 which also has the largest @xmath3 . from the above discussion of the trends
, one can certainly say that cdgeas@xmath0 has the mostly favorable chemical combination of the elements among the class of chalcopyrites considered here .
however , these arguments do not suffice to explain why the second harmonic generation coefficient in cdgeas@xmath0 is so exceptionally large . before doing a more detailed analysis of the frequency dependent @xmath3 in the next section we first analyze the next two columns of the table where we put separately interband versus intraband contributions in shg .
the meaning of these terms needs perhaps some clarification . in general , the formulas for second order response include many terms , and its division in several groups called intra- or interband is rather conventional and depends somewhat on the formulation of the problem .
for instance , it appears at first to be different in the formulas using the length - gauge and the momentum gauge formulations .
the most convincing decomposition based on a careful analysis of the underlying time dependent perturbation theory is the following @xcite : the `` interband '' contribution corresponds to polarization loop where all the three green s function belong to different electronic bands , while the `` intraband '' processes include the modulation of the linear response by the intraband motion as well as the modification of the intraband motion by the interband polarization processes .
analysis of the intraband and interband @xmath3 contributions shows that , ( i ) for all the considered abc@xmath0 compounds the interband contribution is negative while intraband is positive , therefore , these two terms work in opposite directions ; ( ii ) for all the materials except cdgeas@xmath0 and cdsias@xmath0 both absolute values of the intraband and interband terms are larger than the resulting total shg ; ( iii ) the absolute value of the intraband term is always larger than those of interband , i.e. , the resulting total shg is positive .
we also notice that cdgeas@xmath0 and cdsias@xmath0 are exceptions which have an exceptionally small value of @xmath44 , i.e. , the intraband term is dominating the shg in the zero frequency limit .
this term for cdgeas@xmath0 is nearly the same as in zngeas@xmath0 .
however , there is no interband contribution to compensate it .
this explains why the second response function is so large in this material .
the value of @xmath45 in cdsias@xmath0 is a few times smaller than in cdgeas@xmath0 because of differences in the value of the gap .
another point of interest is a comparison with the parent iii - v compounds gaas and gap .
the values obtained at the same level of calculation@xcite have already been given above .
clearly , all phosphides considered here have a higher value of @xmath3 than gap .
they also have consistently lower gaps .
for the arsenides , we also find the @xmath3 values to be consistently higher or equal in the chalcopyrites than in gaas , and to have lower gaps , except for znsias@xmath0 which has a slightly higher gap and the same value of @xmath3 as gaas .
the chalcopyrite type chemical substitution iii - v to ii - iv - v@xmath0 thus clearly appears to be favorable for @xmath3 .
in addition , and even more important in practice , the anisotropy of the chalcopyrite crystal structure allows for angular phase - matching while this is not possible in the cubic zincblende crystals .
the comparison of calculations with experiment is somewhat complicated because the reported experimental results for these materials differ considerably among each other .
the last two columns of table [ tab : shg ] show experimental values of shg from the two different handbooks ( ref . and ref . ) .
the values from ref .
are systematically higher than those from ref . by a factor of 1.31.4 .
the reason of such a discrepancy is not completely clear .
one possible explanation is that most of the original values of shg were measured relatively to quartz , and the absolute value of second harmonic generation of pure quartz has been remeasured a few times ( see ref . for more recently recommended value of shg in quartz ) . for zngep@xmath0 , znsias@xmath0 , and
cdgep@xmath0 our calculations are closer to ref . while for cdgeas@xmath0 they agree well with ref .. in general , we know that the `` scissors '' corrected values are overcorrected , i.e. , the actual value of shg has to be higher @xcite .
such an overcorrection is an intrinsic problem of the `` scissors '' approach . with these cautions in mind ,
some of our values for the cases where no experimental values are available can be used as predictions . in order to better understand the origin of the relative magnitudes of the intra- and interband contributions , we now consider the frequency dependent @xmath3 functions , or ,
more precisely , their imaginary part from which the real part and in particular its static value can be obtained by a kramers - kronig transformation .
figures [ fig : shg_zgp ] and [ fig : shg_cga ] show the imaginary part of the frequency dependent @xmath46 for the 123 and 312 components in zngep@xmath0 and cdgeas@xmath0 .
first , it would be worth to mention that the shg curves for 123 and 312 components look very similar being different just in fine structure details .
this is not very surprising because in the initial zincblende material which was used to derive the chalcopyrite material by chemical substitutions these two components were equal at arbitrary frequency .
the shape of the curves is also qualitatively very similar in the two materials .
it consists of a broad hump with positive shg values between the half of the band gap and 1.52 ev , then it changes sign and exhibits another broad dip between 2 and 3 ev . at
energies higher than 4 ev the imaginary part of shg drops to zero very fast .
the amplitudes of both broad features are somewhat higher in cdgeas@xmath0 than in zngep@xmath0 .
this is probably due to the smaller value of the gap in the first material .
im@xmath46 for the other ternary pnictides abc@xmath0 considered above are looking very similar , and we do not show them .
as we see , the analysis of im@xmath46 does not answer the question concerning the origin of the extremely large value of the zero frequency shg in cdgeas@xmath0 . to this end , it is more useful to analyze different contributions to the total shg curves .
figures [ fig : ei_zgp ] and [ fig : ei_cga ] show the intraband and interband contribution separately . for
zngep@xmath0 both the interband and intraband contributions in the most interesting energy region below 4 ev ( for the kramers - kronig integral which gives the zero frequency value ) do not change the sign ( the first is negative , the other is positive ) .
this means that both the inter- and intraband contributions to the static shg are large and have opposite signs . in cdgeas@xmath0
the situation is different .
while the intraband term behaves in the similar way to those in zngep@xmath0 ( does not change the sign below 4 ev ) , the interband term is positive between 0.3 and 1.1 ev and negative between 1.1 and 3 ev .
thus , the low energy region ( below 4 ev ) of the interband term by itself already exhibits a good deal of compensation between positive and negative values when the integral appearing in the kramers - kronig transformation is calculated .
there is no simple qualitative explanation of the fact why the resulting kramers - kronig integral which gives the zero frequency value of the interband contribution to shg is as small as the numerical calculations show : less than 0.5 % of the intraband term in our calculations , see the previous section ) .
this appears to be a rather coincidental interplay between different interband contributions .
the important fact is that the contributions from different groups of interband transitions in this material have different sign , and they compensate each other in the kramers - kronig integral .
this example shows that a small value of the gap may be favorable for but is clearly not sufficient or the dominant contributing factor to obtain a large value of shg . the value of the shg results from a delicate balance between different contributions too complicated to capture in a simple scaling law with the gap .
unfortunately , we can not presently further reduce this question to a simple explanation in terms of particular band - to - band transitions .
nevertheless , it shows that the entire band structure matters , not simply the gaps .
therefore , there appears , at least in principle , to be an ample room for optimization of @xmath3 by band structure engineering .
of course , we can not control the signs of different interband contributions in a definite bulk chemical compound at will even with quite elaborate chemical substitutions .
however , further opportunities to modify the band structures could exist for semiconducting heterostructures , such as multiple quantum wells , superlattices , layered materials , quantum dots , etc . in these structures
the energies of the transitions between different levels and their matrix elements can to some extent be controlled by geometrical parameters of the structure .
for example , it has already been shown that the asymmetric quantum wells can display very large values of second order optical response ( see , e.g. , refs . ) .
however , the situation in these systems is complicated , and theoretical studies show that excitonic effects play an important role,@xcite in addition to the single electron part considered in our work .
nevertheless , the idea of parametrically controlled heterostructures with large @xmath3 is a challenging task which we hope to pursue in future work because of the practical importance of obtaining materials with higher @xmath3 values .
as we already mentioned above , the distortions from the ideal chalcopyrite structure can be neglected in calculations of optical spectra and shg only if they are small enough as in the case of zngep@xmath0 .
figure [ fig : bnd1_zgp ] compares the electronic bands for zngep@xmath0 along the @xmath47 and @xmath48 lines calculated in two different crystal structures , namely , the real experimental structure ( with the parameters from table [ tab : str ] ) , and the ideal chalcopyrite structure with the same lattice constant @xmath20 .
the band structures look very similar being different only in some fine details ( e.g. , the splitting of the upper valence band at the @xmath15- point is different in the two cases due to different crystal fields ) .
both crystals have indirect ( @xmath48 ) band gaps of nearly the same value .
the zero - frequency shg values calculated in lda are also very close ( 102 pm / v for the experimental structure and 106 pm / v for ideal chalcopyrite , i.e. , the difference between them is only about 4 % ) . in materials with larger distortions
the situation is different .
one has to study how the deviations of both the parameters @xmath22 and @xmath19 from those of the ideal chalcopyrite structure ( @xmath22 = 2 , @xmath19 = 1/4 ) affect both the band structure and optical constants .
figure [ fig : bnd1_csp ] shows how the value of the parameter @xmath19 affects the electronic bands in cdsip@xmath0 ( this compound is the most illustrative because the deviations of both @xmath22 and @xmath19 are the largest ones when compared with other materials of the considered group ) . except for some fine details ,
the effect of @xmath19 can be considered as a rigid shift upwards ( by about 0.4 ev ) of the low conduction bands .
the character of the gap ( pseudodirect ) does not change with @xmath19 .
the behavior of the band structure is completely different when we change the parameter @xmath22 ( fig .
[ fig : bnd2_csp ] ) .
namely , when the value of @xmath22 is approaching 2 , the two lowest conduction bands are changing their positions , and the band gap is becoming direct as opposed to indirect . this illustrates the differences of the behavior of the electronic spectra with changes in @xmath22 and @xmath19 . for the other materials the situation is quite similar .
the change of @xmath19 produces nearly rigid shift of the few lowest conduction bands while the changes of the parameter @xmath22 can mix these bands and change the character of the band gap .
moreover , the value of the rigid shift is nearly linear in @xmath49 and depends very weakly on the material .
thie means that the deformation potential associated with this structural change is almost constant in this class of compounds .
why this is so is presently not clear although it is not entirely unexpected since , in general , these materials are of course all closely related to gaas or gap . in order to study how these changes affect the optical constants we calculated the zero frequency shg coefficients at different values of @xmath22 and @xmath19 . for the band structures of cdsip@xmath0 shown in figures [ fig : bnd1_csp ] and [ fig : bnd2_csp ] this calculation gives @xmath3 = 73 pm / v for the real experimental structure , @xmath3 = 90 pm / v for @xmath22 = 1.836 and @xmath19 = 1/4 , and @xmath3 = 113 pm / v for the ideal chalcopyrite structure .
the value of the lattice constant @xmath20 and the value of the gap correction used was the same in all the three cases .
these calculations illustrate qualitatively how shg depends on distortions of the ideal chalcopyrite structure . in the considered case
both the deviations of @xmath22 and @xmath19 tend to decrease the value of shg .
another interesting question is to study whether or not the chalcopyrite type of ordering of the cations plays a special role in high @xmath3 values in these materials . for this purpose
we considered an alternative crystal structure , a ( 001 ) oriented @xmath4 superlattice ac+bc , which still maintains the local tetrahedral environment of the chalcopyrite where each anion is surrounded by two cations of each type but which nevertheless exhibits a different ordering vector .
this superlattice structure has the same overall chemical formula and stoichiometry as the chalcopyrite structure .
figure [ fig : bnd2_zgp ] shows the lda band structures of the @xmath4 zngep@xmath0 superlattice along the same directions of the brillouin zone as in fig .
[ fig : bnd1_zgp ] .
the real cell for the superlattice is a tetrahedral one which is twice smaller than the cell for the chalcopyrite structure .
we also calculated the electronic structure of @xmath4 superlattices for other abc@xmath0 materials considered before .
however , in some materials which exhibit a small lda gap in the chalcopyrite structure , the gap for the @xmath4 superlattice becomes negative .
here we show the results for zngep@xmath0 only in order to illustrate how the band structure changes with modifications in the crystal structure .
the gap is much smaller in the @xmath4 structure ( @xmath50 = 0.3 ev ) than in the chalcopyrite ( @xmath50 = 1.16 ev ) . as a result ,
the zero frequency shg is about 190 pm / v for the superlattice , i.e. , about twice higher than shg in the chalcopyrite zngep@xmath0 ( we used the same value of the gap correction ( @xmath11 = 0.89 ev ) for the superlattice structure which has been used before for the chalcopyrite zngep@xmath0 ) . while this is at first sight promising for obtaining higher @xmath3 we caution that this structure is unlikely to be stable .
the atomic sizes of cations are different , and one can expect an occurance of alternatingly compressive and tensile biaxial strain in the cation layers .
the calculations show that the difference of the total energy per chemical unit between the superlattice and the chalcopyrite structure is @xmath51 = 0.204 ev , i.e. , is rather large .
this confirms our hypothesis that the alternative structure is not stable with respect to different structural rearrangements minimizing the total energy .
the linear and nonlinear optical properties for important group of chalcopyrites abc@xmath0 ( a= zn , cd ; b= ge , si ; c= as , p ) have been calculated over a wide energy range .
we studied any possible combination of a , b , and c. this allowed us to study the trends in the second order optical response with chemical composition .
the results for the zero - frequency limit of shg are in good agreement with available experimental results .
the calculated birefringence for zngep@xmath0 and cdgeas@xmath0 also shows a fair agreement with recent experimental data in the energy region corresponding to the middle of the gap .
for all the considered compounds the second harmonic generation coefficient @xmath3 is of the order of 100200 pm / v , i.e. , of the order of and in fact larger than the shg in the initial zincblende material ( gaas or gap ) from which these compounds were created by chemical substitutions .
the only exception is cdgeas@xmath0 , which has much higher nonlinear response ( @xmath38 500 pm / v ) than all the other materials from this group .
this value of the shg can not be explained only by a small value of the band gap .
it appears as a result of a very delicate balance between different terms which contribute to second order response .
the results of the calculations are rather stable with respect to small structural modifications .
they do not change much with distortions of the ideal chalcopyrite structure in materials in which @xmath22 and @xmath19 are close to those in ideal chalcopyrite structure .
however , when these deviations are large ( like in cdsip@xmath0 ) the gaps as well as shg s can be strongly affected by the values of @xmath22 and @xmath19 chosen in calculations , and the ideal structure does not describe a real experimental situation correctly .
the lda calculations for the ( 001)- oriented @xmath4 superlattice which has the same overall chemical formula and stoichiometry as the chalcopyrite structure exhibits a much smaller lda band gap than the chalcopyrite structure and larger values of the shg coefficients . however , this structure has a very high total energy which is probably related to large strains in the cation layers .
these strains should make the structure unstable with respect to different structural rearrangements . in conclusion , we can certainly say that a small value of the gap is favorable for larger shg . however , this is by no means sufficient for quantitative predictions of the @xmath3 values from the known values of the gaps . within this family of compounds all calculated in essentially the same approximation ,
no clear scaling of calculated @xmath3 values with a power law of the gap was obtained .
because of the experimental uncertainties in determining @xmath3 values for compounds with different degree of crystalline ideality and different optimization of phase - matching conditions , we consider this test as a more reliable test of such scaling laws than those based on experimental values .
the origin of @xmath3 in terms of the underlying band structure is clearly too complex for such a simple minded extrapolation to be valid .
one has to take into account all the contributing terms and analyze them carefully .
even then , gaining understanding is far from easy and we were only partially successful at unraveling the origin of the high value of @xmath3 for cdgeas@xmath0 .
nevertheless , we hope that our present analysis in terms of the intra- and interband terms provides at least a first step towards such understanding .
both the intra- and interband terms are generally found to be decreasing with increases of the gap because both term contain some power of the interband transition energy in the denominator .
even , so , they do not simply scale with the gap and their sum which is the total value of @xmath3 can either increase or decrease depending on the degree of compensation between the two contributions which was found to depend sensitively on details of the calculations , for example , the `` non - ideality '' of the parameter @xmath19 .
thus , only a very complicated interplay between these different terms is forming the total value of the nonlinear optical response function .
a detailed analysis of these contributions appears to be necessary to understand and predict confidently the expected nlo response for new materials .
s. k. kurtz , j. jerphagnon , and m. m. choy , in _ landolt - brnstein numerical data and functional relationships in science and technology _
, new series , vol . iii/11 eds .
hellwege and a.m. hellwege , ( springer , berlin - heidelberg , 1979 ) , p. 671 .
s. n. rashkeev , w. r. l. lambrecht , and b. segall , in _ nitride semiconductors _ , eds . f. a. ponce , s. p. denbaars , b. k. meyer , s. nakamura , and s. strite , mater .
. proc . * 482 * , ( mrs pittsburgh 1998 ) , p. 857 . | a first - principles study of the birefringence and the frequency dependent second harmonic generation ( shg ) coefficients of the ternary pnictide semiconductors with formula abc@xmath0 ( a = zn , cd ; b = si , ge ; c = as , p ) with the chalcopyrite structures was carried out . it uses a recently developed computational approach based on the self - consistent linear muffin - tin orbital ( lmto ) band - structure method , which is applied using the local density approximation to density functional theory with a simple _ a - posteriori _ gap correction .
the susceptibilies are obtained in the independent particle approximation , i.e. , without local field and excitonic effects .
the zero frequency limits of @xmath1 were found to be in reasonable agreement with available experimental data for all the considered materials .
we found that substitution of p by as , si by ge , and zn by cd is favorable to get a higher value of @xmath2 .
however , the anomalously high value of the zero frequency shg in cdgeas@xmath0 ( the material with the most favorable combination of a , b , and c ) is rather exceptional than typical for this group of compounds .
an analysis of the different contributions in the frequency dependent shg spectra shows that this value appears as a result of a very small interband term in the zero frequency limit which can not compensate the large intraband contribution as it happens in most of the other materials of this class .
the smallness of the interband term in cdgeas@xmath0 is a result of a very delicate balance between different interband transitions , any of which can give positive or negative contribution in shg .
while we find the empirical observation that a smaller value of the gap is strongly correlated with a larger value of the shg to be generally true when comparing various element substitutions within this family in a qualitative sense , simple inverse power scaling laws between gaps and @xmath3 values are not supported by our results .
the case of cdgeas@xmath0 clearly shows that this is an oversimplification and that for reliable predictions of trends of shg coefficients one has to study the interplay between different terms which contribute in shg .
we have also studied the relation between @xmath3 and the chalcopyrite crystal structure by considering some of these materials in an alternative layered zincblende type latttice .
we find that the ( 001 ) oriented @xmath4 superlattice structure has significantly lower gaps than the chalcopyrite and correspondingly higher @xmath3 .
however , this smaller gap structure is characterized by a large alternatingly compressive and tensile lateral strain in the layers , which makes it unfavorable .
we also find that the distortions from the ideal chalcopyrite tend to increase the gap and decrease both the inter- and intraband contributions to @xmath3 , but the net value of @xmath3 is only slightly changed in most cases .
these effects are larger for the cd compounds than for the zn - compounds . as far as the birefringence is concerned
, we find our calculations for zngep@xmath0 and cdgeas@xmath0 to be in fair agreement with experiment ( discrepancies being rather constant and of order 10 % ) in the frequency range corresponding to the middle of the gap but to deviate from the data when the absorption edges of the band gap at high energy and the phonon absorption bands at low energy are approached .
it is presently not clear whether this reflects a deficiency in the theory or imperfections in the samples . |
[ resultsb ] | we have searched for signatures of polarization in hadronic jets from @xmath0 decays using the `` jet handedness '' method . the polar angle asymmetry induced by the high slc electron - beam polarization
was used to separate quark jets from antiquark jets , expected to be left- and right - polarized , respectively .
we find no evidence for jet handedness in our global sample or in a sample of light quark jets and we set upper limits at the 95% c.l . of 0.063 and 0.099
respectively on the magnitude of the analyzing power of the method proposed by efremov _
et al . _ * a search for jet handedness in hadronic @xmath1 decays * _ adelphi university , garden city , new york 11530 infn sezione di bologna , i-40126 bologna , italy boston university , boston , massachusetts 02215 brunel university , uxbridge , middlesex ub8 3ph , united kingdom california institute of technology , pasadena , california 91125 university of california at santa barbara , santa barbara , california 93106 university of california at santa cruz , santa cruz , california 95064 university of cincinnati , cincinnati , ohio 45221 colorado state university , fort collins , colorado 80523 university of colorado , boulder , colorado 80309 columbia university , new york , new york 10027 infn sezione di ferrara and universit di ferrara , i-44100 ferrara , italy infn lab .
nazionali di frascati , i-00044 frascati , italy university of illinois , urbana , illinois 61801 lawrence berkeley laboratory , university of california , berkeley , california 94720 massachusetts institute of technology , cambridge , massachusetts 02139 university of massachusetts , amherst , massachusetts 01003 university of mississippi , university , mississippi 38677 nagoya university , chikusa - ku , nagoya 464 japan university of oregon , eugene , oregon 97403 infn sezione
di padova and universit di padova , i-35100 padova , italy infn sezione di perugia and universit di perugia , i-06100 perugia , italy infn sezione di pisa and universit di pisa , i-56100 pisa , italy rutgers university , piscataway , new jersey 08855 rutherford appleton laboratory , chilton , didcot , oxon ox11 0qx united kingdom sogang university , seoul korea stanford linear accelerator center , stanford university , stanford , california 94309 university of tennessee , knoxville , tennessee 37996 tohoku university , sendai 980 japan vanderbilt university , nashville , tennessee 37235 university of washington , seattle , washington 98195 university of wisconsin , madison , wisconsin 53706 yale university , new haven , connecticut 06511 deceased also at the universit di genova also at the universit di perugia _ the transport of parton polarization in strong interactions is of fundamental interest .
it is at present an open question whether the polarization of quarks or antiquarks produced in hard collisions is observable via the final - state fragmentation products in the resulting jets .
the @xmath1 resonance is an ideal place to study this issue as the partons in @xmath2 decays are predicted by the standard model ( sm ) to be highly longitudinally polarized .
if a method of observing such polarization were developed , it could be applied to jets produced in a variety of hard processes in order to elucidate the spin dynamics of the underlying interactions . in the process
@xmath3 the relative cross sections for production of left- and right - handed quarks of flavor @xmath4 are given at tree level by @xcite : @xmath5 where @xmath6 , @xmath7 , @xmath8 is the longitudinal polarization of the electron beam , @xmath9 and @xmath10 are the vector and axial - vector couplings of fermion @xmath4 to the @xmath1 , and @xmath11 is the polar angle of the outgoing quark with respect to the electron beam direction .
the quark and antiquark in a @xmath1 decay have opposite helicities so that @xmath12 .
the sm predicts @xmath13 , @xmath14 and @xmath15 so that the quarks are produced predominantly left - handed and the antiquarks predominantly right - handed . in order to observe a net polarization in an ensemble of jets from @xmath1
decays it is therefore necessary to distinguish quark jets from antiquark jets .
this separation can be achieved at the slac linear collider ( slc ) where @xmath1 bosons are produced in collisions of highly longitudinally polarized electrons with unpolarized positrons . in 1993
the average electron beam polarization was [email protected] @xcite . in this case
the sm predicts a large difference in polar angle distributions between quarks and antiquarks , providing an unbiased separation of quark and antiquark jets .
we define the `` helicity - based '' polarization of jets at a given @xmath17 : @xmath18 this jet polarization is independent of flavor , and reaches 0.72 and 0.52 in magnitude at large @xmath19 for beam polarizations of @xmath200.63 and @xmath210.63 respectively .
an alternative variable is the `` chirality - based '' polarization of jets : @xmath22 this jet polarization is independent of @xmath23 and electron beam polarization but depends on quark flavor . it is accessible by charge ordering of the tracks used in the analysis as described below .
the experimental challenge is to find observables sensitive to one or both of these expected jet polarizations .
nachtmann @xcite and efremov _
et al_. @xcite have speculated that the polarization may be observable inclusively via a triple product of track momenta in jets . arguing that quark fragmentation may resemble a multi - body strong decay
, they note that the simplest observable with the same transformation properties under parity inversion as @xmath24 ( eqn .
2 ) has the form @xmath25 , where @xmath26 is a unit vector along the jet axis , corresponding to the spin direction of a longitudinally polarized parton that produced the jet , and @xmath27 and @xmath28 are the momenta of two particles in the jet chosen by some charge - independent prescription , e.g. , @xmath29 and referred to some suitable frame .
alternatively , if @xmath27 and @xmath28 are the momenta of a positively and a negatively charged particle , the cross product can be ordered by charge , e.g. , @xmath30 for a given flavor @xmath31 has the same transformation properties under parity inversion as @xmath32 ( eq . 3 ) and
so might be sensitive to @xmath32 .
jets from @xmath1 decays comprise a mixture of flavors that might yield different signals since quark charges and fragmentation properties depend on flavor .
taking into account only the sign @xmath33 of the charge of quarks of flavor @xmath4 , one expects a net polarization @xmath34 , where @xmath350.17 and @xmath360.22 are the sm fractions of hadronic @xmath1 decays into @xmath37 or @xmath38 and @xmath39 , @xmath40 or @xmath41 respectively .
although no quantitative estimate of the size of @xmath42 is given , it is argued in @xcite that such a term can arise from interference between two processes , for example fragmentation into @xmath43 and fragmentation into @xmath44 where @xmath45 .
thus @xmath42 might be largest for triplets of pions nearby in rapidity in which a zero - charge pair has invariant mass near the @xmath46 mass , and the unpaired track together with the oppositely charged track in the pair is used to calculate @xmath42 in the 3-pion rest frame .
it is also argued that evidence of polarization is more likely to be visible in the highest - momentum tracks in jets .
ryskin has proposed @xcite an alternative physical model of the transport of parton polarization in the context of a string fragmentation scheme , which gives a nonzero expectation value of @xmath42 in the laboratory frame if @xmath27 and @xmath28 are the momenta of two hadrons containing partons from the same string breakup .
a jet may be defined as left- or right - handed if @xmath42 is negative or positive respectively . for an ensemble of jets
the jet handedness @xmath47 is defined as the asymmetry in the number of left- and right - handed jets : @xmath48 it can then be asserted that @xmath49 where @xmath50 is the expected polarization of the underlying partons in the ensemble of jets , and @xmath51 is the analyzing power of the method . in this letter we present the results of the first search for jet handedness in @xmath0
decays using a sample of approximately 50,000 hadronic @xmath1 decays collected by the sld experiment @xcite in 1993 .
we have applied the methods suggested in @xcite and @xcite , and have extended them to be more inclusive . for each method we used both helicity- and chirality - based definitions of @xmath42 to calculate @xmath47 .
a handedness signal may be diluted in heavy flavor events , @xmath52 or @xmath41 , since a large fraction of the tracks in each jet is from the decay of a spinless heavy meson .
dalitz _
et al . _ have concluded @xcite that any effect resulting from @xmath53 or @xmath54 decays should be very small .
we therefore divided our data into samples enriched in light , @xmath55 , @xmath39 or @xmath40 , and heavy flavor events using hadron lifetime information , and sought evidence for jet handedness in each .
the trigger and initial selection of hadronic events is described in @xcite .
the analysis presented here is based on charged tracks measured in the central drift chamber and vertex detector .
a set of cuts was applied to select events well - contained within the detector acceptance .
tracks were required to have ( i ) a closest approach to the beam axis within 5 cm and within 10 cm along the beam axis of the measured interaction point , ( ii ) a polar angle @xmath11 with respect to the beam axis with @xmath56 @xmath57 0.80 , and ( iii ) a minimum momentum transverse to this axis of @xmath58 @xmath59 150 mev/@xmath60 .
events were required to contain a minimum of five such tracks , a thrust @xcite axis polar angle with respect to the beam axis @xmath61 within @xmath62 @xmath57 0.71 , and a minimum charged visible energy @xmath63 gev , where all tracks were assigned the charged pion mass .
events with hard gluon radiation were rejected by requiring events to contain only two jets , defined using the jade clustering algorithm @xcite at @xmath64 , which were back - to - back within @xmath65 .
a data sample comprising 17,853 events passed these cuts .
in addition to considering this global sample , events were classified as being of light or heavy flavor based on impact parameters of charged tracks measured in the vertex detector .
the 9,977 events containing no track with normalized transverse impact parameter with respect to the interaction point @xmath66 were assigned to the light - flavor sample and all other events were assigned to the heavy - flavor sample .
the light - flavor contents of these two samples were estimated from monte carlo simulations to be 84% and 30% respectively @xcite .
following @xcite we first considered jets in which the three highest - momentum tracks had total charge @xmath161 and the invariant mass of each zero - charge pair satisfied @xmath67 gev/@xmath68 .
all tracks were assigned the charged pion mass and their momenta were boosted into the three - track rest frame .
the tracks forming the higher - mass zero - charge pair were used to calculate @xmath69 and @xmath70 , where @xmath71 and @xmath72 is the thrust axis signed so as to point along the jet direction .
a signal would be visible as a nonzero mean @xmath42 , which in the case of the helicity - based analysis would be of opposite sign for events produced with positive and negative electron - beam polarization and for jets with positive and negative @xmath73 .
the distributions of sgn(@xmath74 and @xmath31 are shown for the light - flavor sample in fig . 1 .
both distributions appear to be symmetric about zero , implying that any jet handedness is small . also shown in fig .
1 are the predictions of the jetset @xcite monte carlo simulation program for @xmath2 decays , in which spin transport was not simulated , combined with a simulation of the sld .
these simulations give a good description of our measured inclusive track and event topology distributions @xcite .
the means of the simulated @xmath42 distributions are consistent with zero within statistical errors , limiting any analysis bias to 0.008 on @xmath47 .
the variances of the measured distributions are reproduced by the simulation to within 5% relative , although the details of the shapes are not .
the jet handedness for the helicity - based analysis was calculated in bins of jet @xmath23 according to eq .
( 6 ) separately for events produced with positive and negative electron - beam polarization .
results are shown in fig . 2 for the light - flavor sample ;
similar results ( not shown ) are obtained for the global and heavy - flavor samples . in all cases
the measured jet handeness is consistent with zero , and there is no evidence for an angular dependence .
equation ( 7 ) was fitted simultaneously to the @xmath75 measured in events produced with positive and negative electron - beam polarizations , by averaging @xmath76 ( eq .
2 ) over each @xmath23 bin and allowing the analyzing power @xmath51 to vary .
the result of the fit to the light - flavor sample is shown as the solid lines in fig .
2 , and the fitted analyzing powers for all three samples are listed in table i. the jet - handedness values for the chirality - based analysis were calculated from the unbinned @xmath31 distributions and the analyzing powers were calculated from eq .
( 7 ) , where @xmath77 ( eq . 3 ) was averaged over the flavor composition of each sample , estimated from the simulations and weighted by the sign of the quark charge .
the analyzing powers are shown in table i. since all @xmath51 are consistent with zero , we set upper limits at the 95% confidence level on their magnitudes , also shown in table i. since the helicities of the quark and antiquark in a given event are opposite , one might expect a correlation between @xmath42 values in the two jets in an event , which would be negative for the helicity - based analysis and positive for the chirality - based analysis .
we found correlation coefficients to be consistent with zero , within statistical errors of @xmath160.02 , for both analyses and for the global , light- , and heavy - flavor samples .
we extended this method to use the @xmath78 highest - momentum particles in each jet , with @xmath79 .
we considered all zero charge pairs @xmath80,@xmath81 among these @xmath78 particles , without imposing mass cuts , and calculated @xmath82 and @xmath83 for each pair in the @xmath78-particle rest frame .
jets with fewer than @xmath78 tracks were excluded .
we then considered both the average , @xmath84 , and the @xmath85 with largest magnitude @xmath86 . in both cases
the jet handedness calculated from the global , light- and heavy - flavor samples was consistent with zero for all @xmath78 and for both helicity- and chirality - based analyses . for @xmath87
upper limits on the magnitudes of the analyzing powers in the range 0.05 - 0.11 were derived . for
larger @xmath78 the reduced sample size limits our accuracy .
following @xcite we then attempted to select pairs of tracks likely to contain partons from the same string breakup . in studies using jetset
we found the relative rapidity with respect to the jet axis of the tracks in a pair to be useful for this . requiring zero charge
does not improve this selection , but was used in the chirality - based analysis . in each jet the tracks
were ordered in rapidity and assigned a number @xmath88 , where @xmath89 for the track with highest rapidity .
we then required pairs of tracks @xmath80,@xmath81 to have @xmath90 and max(@xmath91,@xmath92)@xmath93 .
since the signal is expected @xcite to increase with momentum @xmath94 transverse to the thrust axis , we also required @xmath95 .
we calculated @xmath83 and @xmath82 in the laboratory frame for each pair satisfying these criteria and then considered both the average , @xmath84 , and the @xmath85 with largest magnitude @xmath86 .
we varied @xmath96 , @xmath97 , and @xmath98 in an attempt to maximize the handedness signal . in all cases
the jet handedness calculated from the global , light- and heavy - flavor samples was consistent with zero .
we obtained upper limits in the range 0.05 - 0.11 for @xmath99 , @xmath100 and @xmath1012 gev / c .
statistics become poor in the potentially interesting high-@xmath98 region .
a number of systematic checks was performed .
the results of all analysis methods were found to be insensitive to the track and event selection cuts , to the jet - finding algorithm and @xmath102 values used to select 2-jet events , and to the values of the selection criteria for tracks used to define @xmath42 .
each analysis was performed on samples of monte carlo events in which spin transport was not simulated , yielding @xmath47 consistent with zero within @xmath103 . in conclusion
, we have searched for evidence of parton polarization in hadronic @xmath1 decays using the jet handedness methods proposed in @xcite and @xcite . in an attempt to optimize a signal we studied a wide range of parameters for each method . in each case
we applied both helicity- and chirality - based analyses , and sought signals separately in samples of light- and heavy - flavor jets as well as in the global sample .
we found no evidence for a non - zero jet handedness , implying that the transport of polarization through the quark fragmentation process is small .
the method proposed in @xcite , applied to a sample of light - flavor jets , yielded upper limits of 0.099 and 0.070 on the magnitudes of the analyzing powers of helicity- and chirality - based analyses respectively .
similar limits were derived for all other methods we applied .
we thank the personnel of the slac accelerator department and the technical staffs of our collaborating instutions for their outstanding efforts on our behalf .
we thank r. dalitz , l. dixon , a. efremov , r. jaffe , m. peskin and m. ryskin for helpful comments relating to this analysis .
see e.g. , b. mele and g. altarelli , phys .
lett . *
b299 * ( 1993 ) 345 .
sld collab .
, k. abe _
et al_. , phys .
rev .
lett .
* 73 * ( 1994 ) 25 .
o. nachtmann , nucl . phys .
* b127 * ( 1977 ) 314 .
a.v .
efremov _ et al_. , phys
. lett . *
b284 * ( 1992 ) 394 .
n.a .
trnqvist , hu
tft9250 ( 1992 ) .
m.g .
ryskin , phys . lett .
* b319 * ( 1993 ) 346 .
sld design report , slac report 273 ( 1984 ) .
r.h .
dalitz _
et al . _ , z. phys .
* c42 * ( 1989 ) 441 .
e. farhi , phys .
rev .
lett .
* 39 * ( 1977 ) 1587 .
jade collab .
, w. bartel _
et al .
_ , z. phys . *
c33 * ( 1986 ) 23 .
a full discussion of flavor tagging can be found in : sld collab .
, k. abe _
et al_. , slac
pub6569 , to be submitted to phys .
rev .
d. t. sjstrand and m. bengtsson , comput .
. commun .
* 43 * ( 1987 ) 367 .
sld collab .
, k. abe _
et al_. , slac
pub6641 , submitted to phys .
rev . |
the goal of this paper is to define and study a class of objects called _
semimatroids_. a semimatroid can be thought of as a matroid - theoretic abstraction of the dependence properties of an affine hyperplane arrangement .
many properties of hyperplane arrangements are really facts about their underlying matroidal structure . therefore , the study of such properties can be carried out much more naturally and elegantly in the setting of semimatroids .
the paper is organized as follows . in section [ sec : semimats ] we define semimatroids , and show how we can think of a hyperplane arrangement as a semimatroid .
the following sections provide different ways of thinking about semimatroids .
section [ sec : modid ] shows how a semimatroid extends " to a matroid , and determines a modular ideal inside it .
the semimatroid can be recovered from the matroid and its modular ideal .
section [ sec : preim ] describes the close relationship between semimatroids and strong maps .
semimatroids are described in terms of elementary preimages and single - element coextensions .
section [ sec : pointed ] gives a bijection between semimatroids and pointed matroids .
section [ sec : semilat ] gives a new characterization of geometric semilattices as posets of flats of semimatroids , extending the classical correspondence between geometric lattices and simple matroids .
the final sections are geared towards the study of the tutte polynomial of a semimatroid .
section [ sec : del - cont ] defines the concepts of duality , deletion and contraction .
section [ sec : tuttesemi ] defines the tutte polynomial , and shows that it is the unique tutte - grothendieck invariant for the class of semimatroids .
finally , section [ sec : intext ] gives a combinatorial interpretation for the non - negative coefficients of the tutte polynomial .
it is worth pointing out that kawahara discovered semimatroids independently , and described their orlik - solomon algebra in @xcite .
las vergnas s work on the tutte polynomial of a quotient map @xcite is also closely related to our work ; we will say more about this in section [ sec : tuttesemi ] .
[ def : semi ] a _ semimatroid _ is a triple @xmath0 consisting of a finite set @xmath1 , a non - empty simplicial complex @xmath2 on @xmath1 , and a function @xmath3 , satisfying the following five conditions . * ( r1 ) * if @xmath4 , then @xmath5 and @xmath6 , then @xmath7 . *
( r3 ) * if @xmath8 and @xmath9 , then @xmath10 . * ( cr1 ) * if @xmath8 and @xmath11 , then @xmath9 . * ( cr2 ) * if @xmath8 and @xmath12 , then @xmath13 for some @xmath14 .
we call @xmath1 , @xmath2 and @xmath15 the _ ground set _ , _ collection of central sets _ and _ rank function _ of the semimatroid @xmath0 , respectively
. sometimes we will slightly abuse notation and denote the semimatroid @xmath2 , when its ground set and rank function are clear .
we will denote subsets of @xmath1 by upper case letters , and elements of @xmath1 by lower case letters .
we will need the fact that semimatroids satisfy a local " version of ( r1 ) and ( r2 ) and a stronger version of ( cr1 ) and ( cr2 ) , as follows . *
( r2 ) * if @xmath16 then @xmath17 or @xmath18 . * ( cr1 ) * if @xmath8 and @xmath11 , then @xmath9 and @xmath19 . * ( cr2 ) * if @xmath8 and @xmath20 , then @xmath13 and @xmath21 for some @xmath14 . _ proof of _ ( r2 ) . from ( r2 ) we know that @xmath22 . from ( r3 )
we know that @xmath23 , and this is @xmath24 or @xmath18 by ( r1 ) .
@xmath25 _ proof of _ ( cr1 ) .
the hypotheses imply that @xmath26 .
then ( r2 ) says that @xmath27 , while ( r3 ) says that @xmath28 .
@xmath25 _ proof of _ ( cr2 ) . by applying ( cr2 ) repeatedly
, we see that we can keep on adding elements @xmath29 of @xmath30 to the set @xmath31 , until we reach a set @xmath32 such that @xmath33 .
now we claim that @xmath34 for some @xmath35 . if that was not the case then , since @xmath36 , ( cr1 ) applies to @xmath37 and @xmath38 .
therefore @xmath39 and @xmath40 .
then ( cr1 ) applies to @xmath41 and @xmath42 , so @xmath43 and @xmath44 .
continuing in this way , we conclude that @xmath45 and @xmath46 , a contradiction .
@xmath25 let us now explain the connection between semimatroids and hyperplane arrangements .
given a field @xmath47 and a positive integer @xmath48 , an _ affine hyperplane _ in @xmath49 is an @xmath50-dimensional affine subspace of @xmath49 .
a _ hyperplane arrangement _ @xmath51 in @xmath49 is a finite set of affine hyperplanes in @xmath49 .
a subset ( or _ subarrangement _ ) @xmath52 of hyperplanes is _ central _ if the hyperplanes in @xmath53 have a non - empty intersection .
the _ rank function _
@xmath54 is defined for each central subset @xmath53 by the equation @xmath55 .
[ prop : arrissemi ] let @xmath51 be an affine hyperplane arrangement in @xmath49 .
let @xmath56 be the collection of central subarrangements of @xmath51 , and let @xmath54 be the rank function of @xmath51 . then @xmath57 is a semimatroid . _
_ to each hyperplane @xmath58 we can associate a vector @xmath59 and a constant @xmath60 , so that @xmath61 is the set of points @xmath62 such that @xmath63 , with the usual inner product on @xmath49 .
it is easy to see that the rank of a central subset @xmath64 is equal to the rank of the set @xmath65 in @xmath49 . from this point of view , axioms ( r1 ) ,
( r2 ) , ( r3 ) are standard facts of linear algebra applied to the vector space @xmath49 .
we now check axioms ( cr1 ) and ( cr2 ) . to check axiom ( cr1 ) , assume that @xmath66 and @xmath67 .
let @xmath68 be the intersection of the hyperplanes in @xmath31 , and similarly let @xmath69 . since @xmath70 and @xmath71 , we must have @xmath72 .
also , @xmath73 implies @xmath74
. therefore @xmath75 , and every hyperplane in @xmath76 contains @xmath77 .
it follows that @xmath9 . to check axiom ( cr2 ) , assume that @xmath8 and @xmath78 .
let @xmath79 and define similarly @xmath80 .
since @xmath81 , there exists a vector @xmath82 , corresponding to a hyperplane @xmath83 , which is not in the span of @xmath84 .
thus @xmath85 has a non - empty intersection with @xmath86 .
@xmath25 semimatroids , like matroids , have several equivalent definitions . in their context , it is possible to talk about flats , independent sets , spanning sets , bases , circuits , and most other basic matroid concepts .
we will say more about this in section [ sec : intext ] . until then
, we will use the rank function approach of definition [ def : semi ] throughout most of our treatment .
we will also need some facts about the closure approach , which we now present . for a semimatroid @xmath87 and a set @xmath4 ,
the _ closure of @xmath31 in @xmath2 _ is @xmath88 .
we will sometimes drop the subscript and write @xmath89 instead of @xmath90 when it causes no confusion .
the closure operator of a semimatroid satisfies the following properties , for all @xmath66 and @xmath91 . *
( clr1 ) * @xmath92 and @xmath93 . * ( cl1 ) * @xmath94 . * ( cl2 ) * if @xmath6 then @xmath95 . *
( cl3 ) * @xmath96 * ( cl4 ) * if @xmath16 and @xmath97 , then @xmath13 and @xmath98 .
_ proof . _ to check ( clr1 ) , let @xmath99 .
we repeat the argument of the proof of ( cl2 ) . since @xmath100 , ( cr1 ) applies to @xmath101 and @xmath102
, so @xmath103 and @xmath104 . ( cr1 ) then applies to @xmath105 and @xmath106 , so @xmath107 and @xmath108 .
continuing in this way , we conclude that @xmath109 and @xmath110 .
( cl1 ) is trivial . to check ( cl2 ) ,
. then @xmath16 and @xmath112 . applying ( cr1 ) to @xmath113 and @xmath30 ,
we conclude that @xmath114 and @xmath115 .
therefore @xmath116 .
we know that @xmath117 ; so to prove ( cl3 ) it suffices to check the reverse inclusion .
let @xmath118 . then @xmath119 and @xmath120 therefore , since @xmath121 , we have @xmath16 and @xmath112 also ; _
i.e. _ , @xmath111 .
finally , we check ( cl4 ) . the assumption that @xmath122 implies that @xmath123 and @xmath124 . since @xmath13 , the assumption that @xmath125 implies that @xmath126 .
these two results together give @xmath127 ; _ i.e. _ , @xmath98 .
@xmath25 we will later need the following definitions . a _ flat _ of a semimatroid @xmath2 is a set @xmath128 such that @xmath129 .
the _ poset of flats _
@xmath130 of a semimatroid @xmath2 is the set of flats of @xmath2 , ordered by containment .
in the following sections , we will present bijections between the class of semimatroids and other important classes of objects .
figure [ fig : bijections ] at the end of section [ sec : preim ] should be useful in understanding these bijections , and is worth keeping in mind especially in sections [ sec : modid ] , [ sec : preim ] and [ sec : pointed ] . in this section ,
we show that a semimatroid is essentially equivalent to a pair @xmath131 of a matroid @xmath132 and one of its modular ideals @xmath133 .
we start by showing how we can naturally construct a matroid @xmath134 from a given semimatroid @xmath135 , by extending the rank function @xmath15 from @xmath2 to @xmath136 .
[ prop : semi - mat ] let @xmath87 be a semimatroid
. for each subset @xmath137 , let @xmath138 .
then @xmath139 is the rank function of a matroid @xmath140 .
_ it is clear , but worth remarking explicitly , that @xmath141 if @xmath4 . it will be most convenient to check the three _ local _ axioms ( r1)-(r3 ) for the rank function of a matroid @xcite .
let @xmath142 and @xmath143 be arbitrary . *
( r1 ) * @xmath144 .
this is trivial . *
( r2 ) * @xmath145 or @xmath18 .
this is easy .
it is immediate from the definition that @xmath146 .
now let @xmath147 for @xmath148 , @xmath149 .
then @xmath150 is also in @xmath2 , so @xmath151 . *
( r3 ) * if @xmath152 , then @xmath153 .
this takes more work .
assume that @xmath154 but @xmath155 .
let @xmath156 , @xmath157 be such that @xmath158 .
notice that @xmath159 must contain @xmath160 ; otherwise we would have @xmath161 and @xmath162 .
similarly , @xmath159 contains @xmath163 .
so let @xmath164 ; clearly @xmath165 .
we have @xmath166 .
therefore @xmath167 . similarly , @xmath168 .
then , by the submodularity of @xmath15 , @xmath169 .
now , since @xmath170 , we can find @xmath171 , @xmath172 such that @xmath173 .
so we have @xmath174 with @xmath175 . by ( cr2 )
, we can add an element of @xmath176 to @xmath177 and obtain a set @xmath149 with @xmath178 such that @xmath179 .
notice that @xmath180 and @xmath181 .
thus @xmath182 similarly , @xmath183 and @xmath184 now we have @xmath185 with @xmath186 . once again , ( cr2 ) guarantees that we can add an element of @xmath187 to @xmath30 to obtain an element of rank @xmath188 in @xmath2 .
but @xmath189 , so the only elements of @xmath187 which may not be in @xmath30 are @xmath160 and @xmath163 .
also , we saw that @xmath190 .
this is a contradiction .
@xmath25 the following definitions will be important to us .
a pair @xmath191 of subsets of @xmath1 is a _ modular pair _ of the matroid @xmath192 if @xmath193 .
@xcite a _ modular ideal _ @xmath133 of a matroid @xmath194 is a non - empty collection of subsets of @xmath1 satisfying the following three conditions . *
( mi1 ) * @xmath133 is a simplicial complex . * ( mi2 ) * @xmath195 for every non - loop @xmath160 of @xmath132 . * ( mi3 ) * if @xmath191 is a modular pair in @xmath132 and @xmath196 , then @xmath197 .
[ prop : modideal ] for any semimatroid @xmath135 , the collection @xmath2 is a modular ideal of @xmath134 .
_ we denote the rank function of @xmath134 by @xmath139 and the rank function of @xmath2 by @xmath15 .
of course , @xmath15 is just the restriction of @xmath139 to @xmath2 .
axioms ( mi1 ) and ( mi2 ) of a modular ideal are satisfied trivially .
we reformulate ( mi3 ) as follows : * ( mi3 ) * if @xmath198 are pairwise disjoint , @xmath199 and @xmath200 , then @xmath201 .
we can assume that @xmath77 and @xmath202 are non - empty ; if one of them is empty , the claim is trivial .
we prove ( mi3 ) by induction on @xmath203 ; let @xmath204 and @xmath205 .
first assume that @xmath206 and @xmath207 are different ; say , @xmath208 .
by ( cr2 ) , we can add some element of @xmath209 to @xmath210 and obtain a set in @xmath2 .
this element can only be @xmath211 , so @xmath212 .
assume then that @xmath213 .
if @xmath214 , ( cr1 ) implies that @xmath215 . assume then that @xmath216 , and therefore @xmath217 by hypothesis .
there is a subset of @xmath218 in @xmath2 of rank @xmath188 ; since it can not be contained in @xmath219 or @xmath209 , it must be of the form @xmath220 for some @xmath221 .
but then we have @xmath222 . by ( cr2 )
, we can add some element of @xmath223 to @xmath210 and obtain a set in @xmath2 .
this element can only be @xmath211 , so @xmath215 .
having established the base case of the induction , we proceed with the inductive step .
assume that @xmath224 and , without loss of generality , @xmath225 .
let @xmath226 . applying the submodularity of @xmath139 twice
, we get that @xmath227 it follows that @xmath228 also .
we can apply the induction hypothesis to the sets @xmath229 , since @xmath230 and @xmath231 .
we conclude that @xmath232 .
we can then apply the induction hypothesis to the sets @xmath233 , since @xmath234 and @xmath235 .
we conclude that @xmath201 , as desired .
@xmath25 propositions [ prop : semi - mat ] and [ prop : modideal ] show us how to obtain a pair @xmath131 of a matroid @xmath132 and one of its modular ideals @xmath133 , given a semimatroid @xmath2 .
now we show that it is possible to recover @xmath2 from the pair @xmath236 .
[ prop : mat , modid - semimat ] let @xmath194 be a matroid , and let @xmath133 be a modular ideal of @xmath132 .
let @xmath237 be the restriction of the rank function of @xmath132 to @xmath133 .
then @xmath238 is a semimatroid .
_ the rank function @xmath237 inherits axioms ( r1)-(r3 ) from @xmath239 .
( cr1 ) is easy .
if @xmath196 and @xmath240 , then @xmath241 by submodularity . thus @xmath191 is a modular pair in @xmath132 , and @xmath242 .
now we check ( cr2 ) .
we start by showing that @xmath133 must contain every independent set of @xmath132 .
in fact , assume that @xmath243 is a minimal independent set which is not in @xmath133 .
since @xmath133 contains all non - loop elements , @xmath243 has at least two elements @xmath160 and @xmath163 .
then @xmath133 contains the modular pair @xmath244 , so it contains their union @xmath243 , a contradiction .
now take @xmath196 with @xmath245 , and pick @xmath246 such that @xmath247 .
let @xmath248 be an independent subset of @xmath31 of rank @xmath249 ; then @xmath250 is an independent set of rank @xmath251 .
therefore @xmath133 contains the modular pair @xmath252 , so it contains their union @xmath253 . @xmath25
[ th : semimatmatid ] let @xmath1 be a finite set .
let @xmath254 be the set of semimatroids on @xmath1 .
let @xmath255 be the set of pairs @xmath131 of a matroid @xmath132 on @xmath1 and a modular ideal @xmath133 of @xmath132 . 1 .
the assignment @xmath256 is a map @xmath257 .
the assignment @xmath258 is a map @xmath259 .
3 . the two maps above are inverses , and give a one - to - one correspondence between @xmath254 and @xmath255 . _ proof . _
the first and second parts are restatements of propositions [ prop : semi - mat ] and [ prop : modideal ] and proposition [ prop : mat , modid - semimat ] , respectively .
denote the maps @xmath257 and @xmath260 above by @xmath261 and @xmath262 respectively .
it is immediate that @xmath263 is the identity map in @xmath254 . to check that @xmath264 is the identity map in @xmath255 , we need to show the following . given a matroid @xmath192 and a modular ideal @xmath133 of @xmath132 , @xmath265 for all @xmath142 .
but this is easy : it is clear that @xmath266 .
equality is attained because @xmath31 has an independent subset @xmath248 of rank @xmath249 ; since @xmath248 is independent , it is in @xmath133 .
@xmath25 before we continue our analysis , we state explicitly a simple property of semimatroids and modular ideals which is implicit in the proofs above . in a semimatroid @xmath0 ,
all the maximal sets in @xmath2 have the same rank , which we denote @xmath15 . in a modular ideal @xmath133 of a matroid @xmath194 ,
all the maximal sets have maximum rank @xmath267 .
now we show that a semimatroid is also equivalent to a pair @xmath268 of a matroid @xmath132 and one of its rank - increasing single - element coextensions @xmath269 .
to do it , we outline the correspondence between the modular ideals , the elementary preimages and the rank - increasing single - element coextensions of a matroid .
this correspondence is just the dual of the well understood correspondence between the modular filters , the elementary quotients , and the rank - preserving single - element extensions of a matroid @xcite , @xcite , @xcite .
therefore we omit all the proofs of these results , and refer the reader to the relevant literature . a _ quotient map _
@xmath270 is a pair of matroids @xmath271 on the same ground set such that every flat of @xmath132 is a flat of @xmath272 .
there are several other equivalent definitions of quotient maps , including the following .
* proposition 8.1.6 ) let @xmath132 and @xmath272 be two matroids on the set @xmath1 . the following are equivalent : +
( i ) @xmath270 is a quotient map .
+ ( ii ) for any @xmath273 , @xmath274 . + ( iii ) for any @xmath275 , @xmath276 .
an _ elementary quotient map _ is a quotient map @xmath277 such that @xmath278 or @xmath18 .
we will focus our attention on elementary quotient maps .
their importance is the following .
perhaps the most useful notion of a morphism in the category of matroids is that of a _
strong map_. every strong map between matroids can be regarded essentially as a quotient map , followed by an embedding of a submatroid into a matroid .
also , every quotient map can be factored into a sequence of elementary quotient maps . therefore , elementary quotient maps are essentially the building blocks of strong maps . for more information on this topic , we refer the reader to @xcite .
an _ elementary preimage _ of a matroid @xmath132 is a matroid @xmath272 on the same ground set such that @xmath270 is an elementary quotient map .
the following proposition explains the relevance of elementary preimages and quotient maps in our investigation .
* proposition 6.5 ) [ th : idealpreim ] let @xmath279 be a matroid . let @xmath280 be the set of modular ideals of @xmath132 and let @xmath281 be the set of elementary preimages of @xmath132 . 1
. given @xmath282 , define the rank function @xmath283 by : @xmath284 then @xmath285 is a matroid , and @xmath286 .
2 . given @xmath286 , let @xmath287 then @xmath282 .
the two maps @xmath288 and @xmath289 defined above are inverses .
they establish a one - to - one correspondence between @xmath280 and @xmath281 .
[ th : semimatmatpreim ] given a finite set @xmath1 , let @xmath290 be the set of pairs @xmath291 of a matroid @xmath132 on @xmath1 and one of its elementary preimages @xmath272 .
then there are one - to - one correspondences between @xmath254 , @xmath255 and @xmath290 . _ proof . _
combine theorems [ th : semimatmatid ] and [ th : idealpreim ] .
@xmath25 .
let @xmath132 be a matroid on the ground set @xmath1 and let @xmath292 be an element not in @xmath1 .
a _ single - element coextension _ of @xmath132 by @xmath292 is a matroid @xmath293 on the set @xmath294 such that @xmath295 .
@xmath293 is _ rank - increasing _ if @xmath296 .
it is worth remarking that most single - element coextensions of @xmath132 by @xmath292 are rank - increasing .
the only one which is not rank - increasing is the matroid @xmath293 on @xmath294 such that @xmath297 for all @xmath298 ; _ i.e. _ , the one where @xmath292 is a loop .
( * ? ? ? * dual of theorem 8.3.2 ) [ th : preimcoext ] let @xmath132 be a matroid and @xmath292 be an element not in its ground set .
let @xmath299 be the set of rank - increasing single - element coextensions of @xmath132 by @xmath292 . 1
. given @xmath286 , define @xmath300 by @xmath301 for @xmath273 .
then @xmath302 is a matroid , and @xmath303 .
if @xmath303 , then the matroid @xmath304 is in @xmath281 .
3 . the two maps @xmath305 and @xmath306 defined above are inverses .
they establish a one - to - one correspondence between @xmath281 and @xmath299 .
[ th : semimatmatcoext ] given a finite set @xmath1 and an element @xmath307 , let @xmath308 be the set of pairs @xmath309 , where @xmath132 is a matroid on @xmath1 and @xmath293 is one of its rank - increasing single - element coextensions by @xmath292 .
then there are one - to - one correspondences between @xmath254 , @xmath255 , @xmath290 and @xmath308 . _ proof . _
combine theorems [ th : semimatmatid ] , [ th : idealpreim ] and [ th : preimcoext ] .
@xmath25 we briefly mention that given a matroid @xmath132 , there are other objects in correspondence with the modular ideals of @xmath132 .
two such examples are the _ modular cocuts _ of @xmath132 and the _ colinear subclasses _ of @xmath132 .
they are the duals of modular cuts and linear subclasses , respectively .
a modular cocut @xmath310 of @xmath132 is a collection of circuit unions of @xmath132 satisfying two conditions .
first , if @xmath311 are circuit unions and @xmath312 , then @xmath313 .
second , if @xmath314 and @xmath315 is a modular pair in @xmath132 , then @xmath316 .
a colinear subclass @xmath2 of @xmath132 is a set of circuits of @xmath132 such that if @xmath317 and @xmath318 , and @xmath319 is a circuit , then @xmath320 .
the details and proofs of the ( dual ) correspondences appear in ( * ? ? ?
* theorem 7.2.2 ) and @xcite , respectively .
we end this section by reviewing the correspondences and objects of sections [ sec : modid ] and [ sec : preim ] with an example .
let @xmath321 be the semimatroid consisting of the set @xmath322 $ ] , the collection of central sets @xmath323 } - \{12 , 123\}$ ] , and the rank function @xmath324 for @xmath4 .
it is easy to check that this is , indeed , a semimatroid .
the first diagram of figure [ fig : bijections ] depicts the poset on @xmath2 ordered by inclusion ; below and to the right of each node we have written the set in @xmath2 corresponding to it , and above and to its left we have written its rank . and its corresponding matroids.,height=480 ] to the semimatroid @xmath2
, we have assigned a pair @xmath325 , a pair @xmath326 and a pair @xmath327 . to obtain the matroid @xmath132
, we add the subsets of @xmath328 $ ] not in @xmath2 to the diagram above , to get the boolean algebra @xmath329}$ ] .
we have placed big nodes on the sets of this poset which are in the diagram of @xmath2 , and small nodes on the new sets . to obtain the rank function of @xmath132
, we copy the rank function of @xmath2 on the big nodes . on each small node
, we put the largest number that we can find on a big node below it .
the big nodes form the modular ideal @xmath133 of @xmath132 . to obtain the matroid @xmath272
, we simply leave the rank function of @xmath132 fixed on the big nodes , and increase it by @xmath18 on the little nodes . finally , to obtain the matroid @xmath293
, we glue two boolean algebras @xmath329}$ ] , to obtain a boolean algebra @xmath330}$ ] on @xmath331 elements .
( we have omitted most of the `` diagonal '' edges of this poset for clarity . ) on the lower copy of the boolean algebra , we put the rank function of @xmath272 . on the upper copy
, we put the rank function of @xmath132 , increased by @xmath18 .
we now establish a correspondence between semimatroids and pointed matroids .
@xcite a _ pointed matroid _ is a pair @xmath332 of a matroid @xmath132 and a distinguished element @xmath292 of its ground set . pointed matroids
are a combinatorial tool often used in the study of affine hyperplane arrangements .
the connection between them is the following .
consider an affine arrangement @xmath333 in @xmath334 , where @xmath61 is defined by the equation @xmath335 .
[ def : cone ] the _ cone over @xmath51 _ is the arrangement @xmath336 in @xmath337 , where @xmath338 is defined by the equation @xmath339 for @xmath340 , and @xmath341 is the _ additional hyperplane _ @xmath342 .
being a central arrangement , @xmath343 has a matroid @xmath344 on the ground set @xmath343 associated to it .
to the arrangement @xmath51 , we associate the pointed matroid @xmath345 . [
th : matcoextpointedmat ] let @xmath1 be a set and let @xmath307 .
let @xmath346 be the set of pointed matroids @xmath332 on @xmath294 such that @xmath292 is not a loop of @xmath132 .
there are one - to - one correspondences between @xmath254 , @xmath255 , @xmath290 , @xmath308 and @xmath346 .
_ it suffices to show a correspondence between @xmath308 and @xmath346 .
the elements of @xmath308 are the pairs @xmath347 for all matroids @xmath293 on @xmath294 such that @xmath348 ; _ i.e. _ , such that @xmath292 is not a loop .
the map @xmath349 establishes the desired bijection . @xmath25 at this point , given a set @xmath1 and an element @xmath307 , we have bijections between @xmath254 , @xmath255 , @xmath290 , @xmath308 and @xmath346 , provided by theorems [ th : semimatmatid ] , [ th : semimatmatpreim ] , [ th : semimatmatcoext ] , and [ th : matcoextpointedmat ] .
the bijection @xmath350 is an important one .
we have obtained it as the composition of four bijections , and now we wish to describe it explicitly .
[ th : pointedmatsemimat ] let @xmath1 be a set and let @xmath307
. 1 . for @xmath351 , let @xmath352 and let @xmath15 be the restriction of @xmath353 to @xmath2 .
then @xmath0 is a semimatroid .
2 . for @xmath354 , define @xmath355 by @xmath356 for @xmath273 .
then @xmath353 is a rank function on @xmath357 , and @xmath351 .
3 . the two maps @xmath350 and @xmath358 defined above are inverses .
they establish a one - to - one correspondence between @xmath346 and @xmath254 . _ proof .
_ we will show that , if we start with @xmath351 and trace the bijections of theorems [ th : matcoextpointedmat ] , [ th : preimcoext ] , [ th : idealpreim ] and [ th : semimatmatid ] , we obtain the semimatroid @xmath359 . under the bijection of theorem [ th : matcoextpointedmat ]
, @xmath351 corresponds to @xmath360 . under the bijection of theorem [ th : preimcoext ]
, @xmath361 corresponds to @xmath362 .
@xmath362 , under the bijection of theorem [ th : idealpreim ] , corresponds to the modular ideal @xmath363 . since @xmath292 is not a loop of @xmath293 , @xmath364 and @xmath365 .
therefore @xmath366 . finally , under the bijection of theorem [ th : semimatmatid ] , @xmath367 corresponds to @xmath368 .
similarly , if we start with a semimatroid @xmath0 and keep track of its successive images under the bijections of theorems [ th : semimatmatid ] , [ th : idealpreim ] , [ th : preimcoext ] and [ th : matcoextpointedmat ] , we get the pointed matroid @xmath369 described .
this theorem then becomes a consequence of the previous ones .
@xmath25 it is not difficult to see that , under the coning construction , the central subsets of a hyperplane arrangement @xmath51 correspond to the subsets of @xmath343 whose closure in @xmath344 does not contain the additional hyperplane @xmath341 . theorem [ th : pointedmatsemimat ] shows that , for semimatroids , the natural analog of the cone of a semimatroid @xmath2 is the matroid @xmath293 of the pointed matroid @xmath369 corresponding to it .
the triple of matroids @xmath370 is sometimes called the _ triple of the pointed matroid _ @xmath369 .
we will also call it the _ triple of the semimatroid _ @xmath2 .
we now discuss geometric semilattices and their relationship to semimatroids .
we start by recalling some poset terminology . for more background information , see for example ( * ? ? ?
* chapter 3 ) .
a _ meet semilattice _ is a poset @xmath371 such that any subset @xmath372 has a _ greatest lower bound _ or _ meet _
@xmath373 : an element such that @xmath374 for all @xmath375 , and @xmath376 for any @xmath377 such that @xmath378 for all @xmath379 .
such posets have a minimum element @xmath380 .
notice that if a set @xmath1 of elements of a meet semilattice has an upper bound , then it has a least upper bound , or _ join _
it is the meet of the upper bounds of @xmath1 .
a _ lattice _ is a poset @xmath382 such that any subset @xmath383 has a greatest lower bound and a least upper bound . clearly , if a meet semilattice has a maximum element , then it is a lattice .
a meet semilattice @xmath371 is _ ranked _ with _ rank function _
@xmath384 if , for all @xmath385 , every maximal chain from @xmath380 to @xmath386 has the same length @xmath387 .
atom _ is an element of rank @xmath18 .
a set of atoms @xmath388 is _ independent _ if it has an upper bound and @xmath389 .
a _ geometric semilattice _ is a ranked meet semilattice satisfying the following two conditions . *
( g1 ) * every element is a join of atoms . * ( g2 ) * the collection of independent set of atoms is the collection of independent sets of a matroid . a _ geometric lattice _ is a ranked lattice satisfying ( g1 ) and ( g2 ) .
geometric lattices arise very naturally in matroid theory from the following result . recall that a matroid @xmath192 is _ simple _ if @xmath390 for all @xmath391 and @xmath392 for all @xmath393 .
@xcite , @xcite [ th : latmat ] a poset is a geometric lattice if and only if it is isomorphic to the poset of flats of a matroid .
furthermore , each geometric lattice is the poset of flats of a unique simple matroid , up to isomorphism . from this point of view
, semimatroids are the right " generalization of matroids , as the following theorem shows .
a semimatroid @xmath394 is _ simple _ if @xmath395 and @xmath396 for all @xmath391 , and @xmath397 for all @xmath398 with @xmath399 .
[ th : semilatsemimat ] a poset is a geometric semilattice if and only if it is isomorphic to the poset of flats of a semimatroid .
furthermore , each geometric semilattice is the poset of flats of a unique simple semimatroid , up to isomorphism . to prove theorem [ th : semilatsemimat ] we use the following two propositions .
[ prop : semilatlat]@xcite a poset @xmath371 is a geometric semilattice if and only if there is a geometric lattice @xmath382 with an atom @xmath292 such that @xmath400$].$ ] denotes the interval of elements greater than or equal to @xmath292 in the poset @xmath382 .
] furthermore , @xmath382 and @xmath292 are uniquely determined by @xmath371 .
[ prop : flatsofsemimat ] let @xmath401 be a semimatroid , and let @xmath369 be the pointed matroid on @xmath294 corresponding to it under the bijection of theorem [ th : pointedmatsemimat ] .
let @xmath130 and @xmath402 be the posets of flats of @xmath2 and @xmath293 . then @xmath403 $ ] . _ proof .
_ since both posets are ordered by containment , we only need to show the equality of the sets @xmath130 and @xmath404 $ ] .
first we show that @xmath405 $ ] .
let @xmath406 .
then for all @xmath407 such that @xmath408 , @xmath409 , and therefore @xmath410 . to check that @xmath31 is a flat in @xmath293
, we need to show that this equality still holds if @xmath411 .
this is not difficult : if that is the case and @xmath412 , then @xmath413 .
clearly then equality must hold .
the case @xmath414 is easier , but needs to be checked separately .
hence @xmath415 , and since no element of @xmath2 contains @xmath292 , @xmath405 $ ] .
the inverse inclusion is easier . if @xmath31 is a flat in @xmath293 not containing @xmath292 , then @xmath416 for all @xmath407 .
when @xmath16 , this equality says that @xmath417 .
therefore @xmath31 is a flat in @xmath2 also .
@xmath25 _ proof of theorem [ th : semilatsemimat ] .
_ it is not difficult to check that the bijection of theorem [ th : pointedmatsemimat ] restricts to a bijection between _ simple pointed matroids _
( pointed matroids @xmath418 such that @xmath293 is simple ) and simple semimatroids . the result
then follows combining this fact with theorem [ th : latmat ] and propositions [ prop : semilatlat ] and [ prop : flatsofsemimat ] .
like matroids , semimatroids have natural notions of duality , deletion and contraction , which we now define .
let @xmath87 be a semimatroid . extend the function @xmath15 to a matroid rank function @xmath419 as in proposition [
prop : semi - mat ] . define the simplicial complex @xmath420 , and the rank function @xmath421 by @xmath422 .
the _ dual of _
@xmath2 is the triple @xmath423 .
@xmath424 is a semimatroid .
_ it is possible to simply check that @xmath424 satisfies the axioms of a semimatroid .
it is shorter to proceed as follows .
consider the pair @xmath326 associated to @xmath2 under corollary [ th : semimatmatpreim ] .
it is known ( * ? ? ?
* proposition 8.1.6(f ) ) that if @xmath272 is an elementary preimage of @xmath132 , then @xmath425 is an elementary preimage of @xmath426 . from the pair @xmath427 , we then get a semimatroid using corollary [ th : semimatmatpreim ] again .
it is straightforward to check that this semimatroid is precisely @xmath424 .
@xmath25 for any semimatroid @xmath2 , we have that @xmath428 .
_ this is easy to check directly from the definition . @xmath25
let @xmath87 be a semimatroid and let @xmath429 be such that @xmath430 .
let @xmath431 and , for @xmath432 , let @xmath433 .
the _ contraction of @xmath434 from @xmath2 _ is the triple @xmath435 .
@xmath436 is a semimatroid .
_ checking the axioms of a semimatroid is straightforward .
@xmath25 let @xmath87 be a semimatroid and let @xmath429 be such that @xmath430 . let @xmath437 and , for @xmath438 , let @xmath439 .
the _ deletion of @xmath434 from @xmath2 _ is the triple @xmath440 .
@xmath441 is a semimatroid .
_ checking the axioms of a semimatroid is straightforward .
@xmath25 again , as with matroids , there are two special kinds of elements that we need to pay special attention to when we perform deletion and contraction .
a _ loop _ of a semimatroid @xmath87 is an element @xmath429 such that @xmath430 and @xmath442 .
an _ isthmus _ of a semimatroid @xmath87 is an element @xmath429 such that , for all @xmath128 , @xmath443 and @xmath444 .
[ lemma : loop ] if @xmath429 is a loop of the semimatroid @xmath87 , then @xmath445 .
otherwise , @xmath446 .
_ clearly @xmath447 .
if @xmath434 is a loop , consider any @xmath128 .
( cr1 ) applies to @xmath448 and @xmath388 , so @xmath449 and @xmath450 .
therefore the maximum rank @xmath15 in @xmath2 is achieved for some @xmath432 .
but then we have @xmath451 , so @xmath445 . if @xmath434 is not a loop , then for all @xmath432 we have @xmath452 , so @xmath453 .
equality holds : if we start with @xmath430 and repeatedly apply ( cr2 ) with an element of @xmath2 of rank @xmath15 , we can obtain a set @xmath454 of rank @xmath15
. then @xmath455
. @xmath25 [ lemma : isthmus ] if @xmath429 is an isthmus of the semimatroid @xmath456 , then @xmath457 .
otherwise , @xmath458
. _ proof . _ clearly @xmath459 .
if @xmath434 is an isthmus then it is clear from the definition that @xmath460
. if @xmath434 is not an isthmus , there are two cases . if there is an @xmath461 such that @xmath462 , take a maximal one .
it is also a maximal set in @xmath2 , so it has maximum rank @xmath15 ; and @xmath463 , so @xmath464 .
the other possibility is that for all @xmath128 , we have @xmath449 and @xmath465 . in this case
it is also clear that @xmath464 .
@xmath25 [ lemma : loopisthmus ] if @xmath429 is a loop or an isthmus of the semimatroid @xmath466 , then @xmath467 .
_ this is clear from lemmas [ lemma : loop ] and [ lemma : isthmus ] and their proofs .
with the background results that we have established , we are now able to define and study the tutte polynomial of a semimatroid .
the _ tutte polynomial _ of a semimatroid @xmath87 is defined by @xmath468 if @xmath51 be a hyperplane arrangement and @xmath56 is the semimatroid determined by it , then the tutte polynomial of the semimatroid @xmath56 is precisely the tutte polynomial of the arrangement @xmath51 , as defined and studied in @xcite . that paper focuses on enumerative aspects arising from the computation of these polynomials
; here we will concentrate our attention on matroid - theoretical considerations .
_ example .
_ figure [ fig : a ] shows a hyperplane arrangement @xmath51 in @xmath469 , consisting of the five planes @xmath470 and @xmath471 in that order .
table [ table : tutte ] shows all the central subsets of @xmath51 , and their contributions to the tutte polynomial of @xmath51 .
.,width=336 ] .computing the tutte polynomial @xmath472 [ cols=">,<",options="header " , ] [ l : phi1 ] the map @xmath473 maps @xmath474 to @xmath2 . _ proof .
_ let @xmath475 . for all @xmath476
, @xmath477 is central and @xmath478 , so @xmath479 .
therefore @xmath480 and @xmath481 .
since @xmath482 , this implies that @xmath483 , and @xmath484 as well .
@xmath25 [ l : phi2 ] the map @xmath485 maps @xmath2 to @xmath474 . _ proof .
_ let @xmath4 .
let @xmath486 and @xmath487 , so that @xmath488 .
we need to show three things .
first , we need @xmath489 to be a basis for @xmath31 .
this is immediate .
next , we need the elements of @xmath490 to be internally active in @xmath489 .
let @xmath491 . since @xmath489 is the smallest basis for @xmath2 containing @xmath492 , for any element @xmath493 not in @xmath489 we have @xmath494 .
by submodularity , we can conclude that @xmath495 , which is exactly what we wanted . finally , we need to show that the elements of @xmath496 are externally active in @xmath489 .
let @xmath497 .
first notice that @xmath498 , because @xmath499 is dependent : @xmath500 . also notice that @xmath501 is central , applying ( cr1 ) to @xmath502 and @xmath489 .
now observe the following .
we know that @xmath492 is the largest basis for @xmath31
. therefore @xmath503 for all @xmath504 . by submodularity
, it follows that @xmath505 .
we can rewrite this as @xmath506 since @xmath492 is independent . since @xmath507
, submodularity implies that @xmath508 .
this shows that @xmath386 is an externally active element in @xmath489 .
@xmath25 [ pr : key ] the map @xmath473 is a bijection from @xmath474 to @xmath2 , and the map @xmath485 is its inverse .
proposition [ pr : key ] is the main ingredient of our proof of theorem [ th : intext ] . before proving it , we need some lemmas . [
l : rank ] for all @xmath475 , we have @xmath509 .
_ we start by showing that @xmath510 for all @xmath511 . if @xmath512 , do the following . since @xmath35 is internally active , @xmath513 , and therefore @xmath514 . otherwise , if @xmath515 , then @xmath516 . since @xmath434 is externally active , @xmath517 .
submodularity then implies that @xmath518 .
now that we know this , submodularity implies that @xmath519 for all @xmath520 .
applying submodularity again , we get @xmath509 for all @xmath521 . @xmath25 [ l : d ] for all @xmath475 , we have @xmath522 . _ proof . _
lemma [ l : rank ] tells us that @xmath523 is a basis for @xmath524 ; we need to show that it is the largest one .
consider an arbitrary @xmath525 subset @xmath31 of @xmath526 with @xmath527 .
we will show that @xmath31 is not a basis for @xmath528 .
let @xmath529 , where the @xmath530 s are in @xmath523 and the @xmath531 s are in @xmath532 .
since @xmath533 we can assume , without loss of generality , that @xmath534 . from lemma [ l : rank ] we know that @xmath535 , having @xmath536 , @xmath537 and @xmath538 implies that @xmath539 .
combining these two inequalities and using submodularity , we get that @xmath540 for all @xmath340 . invoking submodularity once again
, we get that @xmath541 .
therefore @xmath542 it follows that @xmath31 is not a basis for @xmath524 .
@xmath25 [ l : ud ] for all @xmath475 , we have @xmath543 .
_ in view of lemma
[ l : d ] , we need to show that @xmath544 . clearly @xmath77 is a basis of @xmath2 containing @xmath523 ; now we show that it is the smallest one .
let @xmath545 be an @xmath139-tuple smaller than @xmath77 , where the @xmath530 s are in @xmath243 ( since @xmath31 must contain @xmath523 ) and the @xmath546 s are in @xmath1 .
we will show that @xmath31 is not a basis for @xmath2 .
once again we can assume , without loss of generality , that @xmath547 .
since each @xmath530 is internally active , @xmath548 , and hence @xmath549 .
submodularity gives @xmath550 , which in turn gives @xmath551 .
@xmath25 so far we have only defined @xmath552 for independent sets @xmath31 of @xmath2 .
we can extend the definition to arbitrary subsets @xmath142 as follows .
if @xmath31 is dependent , then there is no basis of @xmath2 containing it .
instead , we consider all the minimal sets of rank @xmath139 which contain @xmath31
. let @xmath552 be the lexicographically smallest of those sets .
then we can say even more .
[ l : du ] for all @xmath475 , we have @xmath553 and @xmath554
. we will not need lemma [ l : du ] to prove proposition [ pr : key ] and theorem [ th : intext ] .
we state it for completeness , but we omit its proof , which is very similar to the proofs of lemmas [ l : d ] and [ l : ud ] . _
proof of proposition [ pr : key ] .
_ checking that @xmath555 is the identity map in @xmath2 is immediate , and lemmas [ l : d ] and [ l : ud ] imply that @xmath556 is the identity map in @xmath474 .
@xmath25 _ proof of theorem [ th : intext ] .
_ using the bijection of proposition [ pr : key ] , the sets in @xmath2 are precisely the sets of the form @xmath528 , where @xmath77 is a basis , @xmath557 and @xmath558 . also , from lemma [
l : rank ] , @xmath559 .
therefore we have @xmath560 as desired .
@xmath25 into intervals.,width=432 ] regard the simplicial complex @xmath2 as a poset , ordering its faces by inclusion .
there is a nice way to understand theorem [ th : intext ] in terms of this poset .
proposition [ pr : key ] gives us a way of classifying the faces of @xmath2 according to the basis of @xmath2 that they correspond to under the map @xmath561 ( or @xmath562 ) .
this classification decomposes the poset into disjoint intervals , where each interval is a boolean algebra of the form @xmath563 $ ] for a basis @xmath77 .
this is illustrated in figure [ fig : boolean ] for the arrangement @xmath51 considered at the beginning of section [ sec : tuttesemi ] ; recall table [ table : intext ] .
if we look at the interval corresponding to basis @xmath77 , and add the contributions of its elements to the right - hand side of ( [ eq : tutte ] ) , we simply get the monomial @xmath564 .
the present work is chapter 3 of the author s ph.d .
thesis @xcite .
i am extremely grateful to seth chaiken , vic reiner , gian - carlo rota , richard stanley , and tom zaslavsky ; the content and exposition of this paper benefitted greatly from instructive conversations with them .
f. ardila .
enumerative and algebraic aspects of matroids and hyperplane arrangements , ph.d .
thesis , massachusetts institute of technology , 2003 .
f. ardila . computing the tutte polynomial of a hyperplane arrangement , _ preprint _ , 2002 .
g. birkhoff .
abstract linear dependence in lattices , _ amer .
j. math . _ * 57 * , 1935 , 800 - 804 .
a. bjrner .
homology and shellability of matroids and geometric lattices , in n. white ( ed . ) , _ matroid applications _ , encyclopedia of mathematics and its applications , 40 , cambridge univ . press , cambridge , 1992 , 123 - 225 .
t. brylawski . a combinatorial model for series - parallel networks , _ trans .
* 154 * ( 1971 ) , 1 - 22 .
t. brylawski and j. oxley .
the tutte polynomial and its applications , in n. white ( ed . ) , _ matroid applications _ , encyclopedia of mathematics and its applications , 40 , cambridge univ . press , cambridge , 1992 , 123 - 225 .
t. a. dowling and d. g. kelly .
elementary strong maps between combinatorial geometries , in _ colloquia internazionale sulle teorie combinatorie , tomo ii _ ,
lincei , rome , 1976 , 121 - 152 .
y. kawahara .
on matroids and orlik - solomon algebras , preprint , 2003 .
j. p. s. kung .
strong maps , in n. white ( ed . ) , _ theory of matroids _ , encyclopedia of mathematics and its applications , 26 , cambridge univ . press , cambridge , 1986 , 298 - 310 .
m. las vergnas .
the tutte polynomial of a morphism of matroids i. set - pointed matroids and matroid perspectives , _ ann .
fourier , grenoble _ * 49 * 3 ( 1999 ) , 973 - 1015 .
p. orlik and h. terao .
_ arrangements of hyperplanes _ , springer - verlag , berlin / heidelberg / new york , 1992 .
j. g. oxley .
_ matroid theory _ , oxford university press , new york , 1992
. v. reiner .
an interpretation for the tutte polynomial , _
european j. combin . _ * 20 * ( 1999 ) , 149 - 161 .
r. p. stanley .
enumerative combinatorics , vol .
1 _ , wadsworth and brooks - cole , belmont , ca , 1986 ; reprinted by cambridge university press , cambridge , 1997 . w. t. trotter .
_ combinatorics and partially ordered sets _ , the johns hopkins university press , baltimore and london , 1992 .
w. t. tutte .
lectures on matroids , _
_ * 69b * ( 1965 ) , 1 - 48 . m. l. wachs and j. w. walker . on geometric semilattices , _ order . _
* 2 * ( 1986 ) , 367 - 385 .
d. j. a. welsh .
_ matroid theory _ , academic press , new york , 1976 .
n. white , ed . _
theory of matroids _ , cambridge university press , cambridge , 1986 . | we define and study _ semimatroids _ , a class of objects which abstracts the dependence properties of an affine hyperplane arrangement . we show that geometric semilattices are precisely the posets of flats of semimatroids .
we define and investigate the tutte polynomial of a semimatroid .
we prove that it is the universal tutte - grothendieck invariant for semimatroids , and we give a combinatorial interpretation for its non - negative coefficients . [ chapter : semimatroids ] |
generating function ( gf ) is a terminology frequently used in combinatorial mathematics as a bridge between discrete mathematics and continuous analysis .
it functions like a clothesline on which we hang up a sequence of numbers for display @xcite .
a gf may have different aliases in different contexts , e.g. partition function in statistical physics @xcite and @xmath2-transform in signal analysis @xcite .
it can be used to find certain characteristics of a sequence like expectation value or standard deviation @xcite .
sometimes that will be much more efficient than a direct evaluation .
there are many ways to construct a gf for a sequence , the simplest of which is using power series . given @xmath3 ( @xmath4 )
, we may construct its gf as @xmath5 note that @xmath6 can take a continous value and @xmath7 is analytic with respect to @xmath6 , which may give us some computational power using techniques of derivation or integration .
the paper is organized as follows .
section ii shows how a gf can find the number of distinct ways of assigning @xmath0 identical bosons or fermions to @xmath1 different states .
section iii discusses the gf approach to find the degeneracies of energy levels of a harmonic oscillator and a confined free particle both in higher dimensions . in our discussion
, we follow the notation @xmath8\,f(t)$ ] used in ref .
@xcite to mean the coefficient of @xmath9 term in the power series expansion of @xmath7 .
for example , we have @xmath8\,\mathrm{e}^t=1/{n!}$ ] @xmath10 .
the problem of how many ways there are to distribute @xmath0 identical particles among @xmath1 different states lies at the heart of quantum statistics .
it determines the specific distribution function that identical particles obey , i.e. bose - einstein distribution for bosons and fermi - dirac distribution for fermions .
textbooks generally tackle the problem by a combinatorial method @xcite . here
we show how to solve it by a gf .
pauli s exclusion principle has that a quantum state can accommodate one fermion at most .
this makes the counting problem to be much easier than that of bosons .
the number is simply the binomial coefficient @xmath1 choose @xmath0 : @xmath11 we can do it in another way .
the occupation number for fermions is @xmath12 and @xmath13 only .
we thus associate each state with a gf @xmath14 .
it is obvious that the coefficient of term @xmath15 in the expansion of @xmath16 is exactly the total ways that we can pick out @xmath0 candidates from @xmath1 states : ( @xmath17 ) @xmath18 } f^m ( t ) & = & { [ t^n]}\,(1+t)^m\nonumber\\ & = & \frac{1}{n!}\,{[t^0]}\,\frac{\mathrm{d}^n}{\mathrm{d}x^n}\,(1+t)^m\nonumber\\ & = & \frac{1}{n!}\,{[t^0]}\,m(m-1)\cdots(m - n+1)(1+t)^{m - n}\nonumber\\ & = & \frac{m!}{n!\,(m - n)!}\nonumber\\ & = & { m \choose n}.\end{aligned}\ ] ] the gf approach recovers the binomial coefficient @xmath1 choose @xmath0 .
furthermore , the approach can be easily adapted to the boson case
. there can be an arbitrary number of bosons in one quantum state in principle , i.e. the occupation number can be @xmath12,@xmath13,@xmath19,@xmath20 .
it is not easy to figure out a combinatorial method to find the total ways in this case . with the help of gf
, we now associate each quantum state with @xmath21 .
notice that @xmath22 has an explicit form when @xmath6 is within its region of convergence : @xmath23 just as what we have done for fermions , we multiply @xmath1 copies of @xmath22 together . as a result ,
the number of total ways to distribute @xmath0 bosons among @xmath1 different states is just the coefficient of term @xmath15 in the expansion of @xmath24 : @xmath18 } b^m ( t ) & = & { [ t^n]}\,\frac{1}{(1-t)^m}\nonumber\\ & = & \frac{1}{n!}\,{[t^0]}\,\frac{\mathrm{d}^n}{\mathrm{d}t^n}\,\frac{1}{(1-t)^m}\nonumber\\ & = & \frac{1}{n!}\,{[t^0]}\,(-m)(-m-1)\cdots(-m - n+1)\frac{(-1)^n}{(1-t)^{m+n}}\nonumber\\ & = & \frac{(n+m-1)!}{n!\,(m-1)!}\nonumber\\ & = & { n+m-1 \choose n}.\end{aligned}\ ] ] this is exactly the same as what is derived from a combinatorial method @xcite , but we are using techniques from calculus for continuous variables .
the gf approach also has an advantage when the occupation numbers are distributed in some random way , e.g. , @xmath13,@xmath19,@xmath25,@xmath20 .
energy levels are usually quantized in a quantum system . furthermore
, degeneracies often occur in a system with independent degrees of freedom . in this section
we employ the gf to count the degeneracies .
problem of this kind can be formulated as follows .
suppose the system hamiltonian @xmath26 can be written as @xmath27 where the subscripts @xmath28,@xmath29,@xmath30,@xmath20 represent independent degrees of freedom .
suppose further that the energy spectrum of @xmath31 is @xmath32 with degeneracies @xmath33 , @xmath34 .
since the degrees of freedom are independent , the schrdinger equation of the system is separable .
therefore , the energy spectrum of the system hamiltonian @xmath26 consists of all possible sums of @xmath35 .
the degeneracy of a certain energy level @xmath36 is then given by the total number of @xmath1-tuples @xcite @xmath37 that satisfy @xmath38 with @xmath1 being the total number of degrees of freedom .
now we associate each degree of freedom @xmath39 with a gf @xmath40 and multiply them together over all @xmath39 . as a result
, we obtain a polynomial @xcite of @xmath6 , whose coefficients give exactly the degeneracies we want . to be specific , the degeneracy of energy level @xmath36 is given by @xmath41}\,\prod_{\delta}g_{\delta}(t).\end{aligned}\ ] ] a harmonic oscillator ( ho ) is not only exactly soluble in both classical and quantum physics , but also of great physical relevance .
for example , the problem of finding the landau levels of an electron in a magnetic field reduces to finding energy levels of an equivalent ho .
a 1d ho has a simple structure of energy spectrum , whose levels are non - degenerate and equally spaced : @xmath42 consider now a 2d homogeneous ho with potential @xmath43 its energy spectrum can be solved in a planar polar coordinate system , noting that @xmath44 .
a much simpler way is to treat the 2d system as two uncoupled 1d hos .
consequently , the energy levels are determined by a couple of integers @xmath45 as @xmath46 where @xmath47 . for a given integer @xmath48
, there correspond @xmath49 different couples of quantum numbers @xmath45 , namely , @xmath50 , @xmath51 , @xmath20 , @xmath52 .
the degeneracy of energy level @xmath53 is therefore @xmath54
now we apply the gf approach to find the degeneracy .
subtracting the zero point energy @xmath55 from the spectrum of each 1d ho and setting @xmath56 , we obtain @xmath57 , @xmath58 .
then we associate each 1d ho with a gf @xmath59 which reduces to @xmath60 for @xmath61 . the degeneracy @xmath62 of energy level @xmath63 is therefore given by @xmath64}\,h_{\mathrm{1d}}^2(t)\nonumber\\ & = & { [ t^n]}\,\frac{1}{(1-t)^2}\nonumber\\ & = & { n+2 - 1 \choose n}\nonumber\\ & = & n+1,\end{aligned}\ ] ] using result from section ii - b .
we can go further to find the degeneracies of energy levels for a 3d homogeneous ho whose potential is @xmath65 it is straightforward to write down that @xmath66 and the corresponding degeneracy @xmath67}\,h_{\mathrm{1d}}^3(t)\nonumber\\ & = & { [ t^n]}\,\frac{1}{(1-t)^3}\nonumber\\ & = & { n+3 - 1 \choose n}\nonumber\\ & = & \frac12 ( n+1)(n+2).\end{aligned}\ ] ] the sequence of @xmath68 goes like @xmath13,@xmath69,@xmath70,@xmath71,@xmath72,@xmath73,@xmath20 , which is consistent with the results derived by solving in spherical coordinates .
it is not difficult to generalize above to the case of inhomogeneous hos , e.g. @xmath74 .
as long as the ratio @xmath75 is some rational number , we can always find appropriate gfs to calculate the degeneracies of the system energy levels .
consider first a 1d free particle confined to some region of length @xmath78 .
it has non - degenerate energy levels @xcite as @xmath79 in the following , we choose @xmath78 such that @xmath80 for simplicity .
accordingly , the levels @xmath81 have a gf as @xmath82 @xmath83 can be used to find the degeneracies of energy levels of a free particle confined in a square of @xmath84 or even a cube of @xmath85 .
finding the degeneracies in each case ( 1d , 2d and 3d ) is a basic problem in quantum mechanics . in solid state physics @xciteit
is also concerned with the density of states per unit energy @xmath86 . in two dimensions ,
the density of states @xmath87 is independent of the energy @xmath36 , whereas @xmath88 in 1d and @xmath89 in 3d .
textbooks @xcite usually derive the density of states by using a quasi - continuous approximation of the distribution of energy levels in @xmath90-space .
below we use @xmath83 to find the level degeneracies in 2d and 3d .
following eq .
( 9 ) , the degeneracies are @xmath94\,f_{\mathrm{1d}}^2(t),\\ d_{\mathrm{3d}}(e ) & = & [ t^e]\,f_{\mathrm{1d}}^3(t).\end{aligned}\ ] ] unfortunately , series like @xmath83 does not have a known explicit sum .
thanks to the advanced math softwares like mathematica , we can use a computer to perform the evaluation .
coefficientlist " function can extract the coefficients in the expansion of eqs . ( 20 - 22 ) .
then we are able to count the total number of states whose energies fall into the interval @xmath95 $ ] .
table [ degeneracies ] gives a detailed list of number of states within certain energy intervals for a 1d , 2d and 3d free particle , respectively .
the list plots of the number of states are further presented in fig .
[ density plot ] .
there are totally @xmath96 energy intervals and each interval has a length of @xmath97 . since the interval length is far more than the minimum separation between energy levels , the quasi - continuous approximation is valid and the plots will reflect approximately the dependence of density of states on the energy . a comparison between the discrete number of states and the continuous density of states
is made in fig .
[ density plot ] ( depicted as dots and solid lines , respectively ) .
it can be seen that the exact numbers of states derived by a gf approach comply well with @xmath98 , @xmath92 and @xmath89 .
herbert s. wilf , _ generatingfunctionology _ ( a k peters , ltd . , 2006 ) , 3rd ed .
see , for example , kerson huang , _ introduction to statistical physics _
( crc press , 2001 ) see , for example , alan v. oppenheim et al .
, _ signals and systems _
( prentice hall , 1996 ) , 2nd ed .
it is commonly used in probability theory .
see , for example , athanasios papoulis and s. unnikrishna pillai , _ probability , random variables and stochastic processes _ ( mcgraw - hill , 2002 ) , 4th ed .
david j. griffiths , _ introduction to quantum mechanics _
( benjamin cummings , 2004 ) , 2nd ed .
if an element @xmath99 in the tuple corresponds to a level of degeneracy @xmath100 , that tuple should be counted for @xmath100 times .
assume the energy levels are quantized so that their separations are multiples of some unit .
that is usually the case in a simple quantum system .
the spin degeneracies are excluded from the discussion .
see , for example , charles kittel , _ introduction to solid state physics _
( wiley , 2004 ) , 8th ed . | in this paper we present a generating function approach to two counting problems in elementary quantum mechanics . the first is to find the total ways of distributing @xmath0 identical particles among @xmath1 different states .
the second is to find the degeneracies of energy levels in a quantum system with multiple degrees of freedom .
our approach provides an alternative to the methods in textbooks . |
quantum zakharov equations ( qzes ) @xcite describe the nonlinear interaction of high - frequency ( hf ) quantum langmuir waves ( qlws ) and the low - frequency ( lf ) quantum ion - acoustic waves ( qiaws ) in quantum plasmas .
this set of equations basically extends the classical zakharov equations ( czes ) @xcite with higher - order dispersive terms associated with the bohm potential . unlike czes , qzes have been deduced using a quantum fluid model under the similar quasineutral assumption and the multiple time scale technique .
recently , much attention has been paid to investigate the dynamics of qzes in the context of e.g. , the formation of langmuir solitons through modulational instability ( mi ) as well as langmuir turbulence through the process of chaos ( see , e.g. , refs .
the statistical properties of the qzes have been analyzed using kinetic treatment and to show that the quantum coupling parameter ( @xmath0 ) can be responsible for reducing the mi growth rate @xcite .
the latter is , however , shown to be maximized in the limit of @xmath1 .
a variational approach was conducted to study the quantum effects on localized solitary structures @xcite .
moreover , some general periodic solution using lie point symmetries @xcite , some exact solutions @xcite as well as both temporal @xcite and spatiotemporal dynamics @xcite in the context of chaos and langmuir turbulence of the qzes have been investigated in the recent past .
furthermore , a comprehensive work on the dynamics of langmuir wave packets in three spatial dimensions @xcite as well as some investigations on arrest of langmuir wave collapse @xcite by the quantum effects can be found in the recent works .
notice that when the wave electric field is strong such that it approaches the decay instability threshold , the interaction of qlws and the qiaws is said to be in ` weak turbulence ' , and then qlws are essentially scattered off qiaws . on the other hand , when the electric field intensity is so strong that the mi threshold is exceeded , the interaction results in ` strong turbulence ' regime in which qlws are typically trapped by the density cavities associated with the qiaws .
such phenomena can frequently occur in plasmas .
moreover , the transfer of energy to few stronger modes with small spatial scales can take place due to a chaotic process , and this energy transfer can be faster when the chaotic process in a subsystem of the qzes is well developed @xcite . in the present work , we will investigate the full qzes numerically especially when the wave number of modulation is small enough from its critical value .
the latter , in turn , excites many unstable harmonic modes which in a longer time collide and fuse into fewer new incoherent patterns due to qiaw emission .
we will show that since the critical wave number of modulation ( below which the mi sets in ) depends on the quantum coupling parameter @xmath2 the length scale of excitation for the transition from temporal chaos ( tc ) to spatiotemporal chaos ( stc ) is to be larger than the classical case .
moreover , lower the values of @xmath3 , the higher is the number of unstable harmonic modes .
furthermore , we will show that the solitary waves thus formed due to mi will lose their strength after a long time through random collision and fusion among the patterns under strong qiaw emission .
this process becomes quicker whenever the density correlation due to quantum fluctuation becomes strong with higher values of the quantum parameter @xmath0 .
the stc state is then said to emerge , and the energy of the system is thus redistributed to new incoherent patterns as well as to few stronger modes with small length scales .
the nonlinear interaction of qlws and qiaws is described by the following one - dimensional qzes @xcite .
@xmath4@xmath5 where @xmath6 is the slowly varying wave envelope of the hf electric field and @xmath7 is the lf plasma density perturbation due to qiaw fluctuation .
also , @xmath8which is associated with the bohm potential , represents the ratio of the ion plasmon energy to the electron thermal energy . here
@xmath9 is the scaled planck s constant , @xmath10 is the boltzmann constant , @xmath11 is the electron temperature and @xmath12 is the ion ( electron ) plasma frequency with @xmath13 denoting the constant background density and @xmath14 the ion ( electron ) mass .
the electric field @xmath15 is normalized by @xmath16 , the density @xmath17 by @xmath18 .
moreover , the space and time variables are rescaled by @xmath19 and @xmath20 , where @xmath21 is the electron debye length .
note that by disregarding the term @xmath22 in eqs .
( [ 1 ] ) and ( [ 2 ] ) , one recovers the well - known czes @xcite .
the latter have been widely studied in the context of solitons , chaos and langmuir turbulence in many areas of plasma physics ( see , e.g. , refs .
it is thus of natural interest to investigate the qzes in the quantum realm , which may be useful for understanding the onset of plasma wave turbulence at nanoscales in both laboratory and astrophysical plasmas .
the growth rate of mi can be obtained from eqs .
( [ 1 ] ) and ( [ 2 ] ) by assuming the perturbations of the form exp@xmath23 from a spatially homogeneous pump electric field @xmath24 as @xcite @xmath25 ^{1/2 } , \label{3}%\ ] ] where @xmath26 and @xmath27 it can be shown that the growth rate is maximum at @xmath1 and decreases for increasing values of @xmath0 with cut - offs at lower wave numbers of modulation @xcite . in order to solve numerically the eqs . ( [ 1 ] ) and ( [ 2 ] ) we choose the following initial condition @xcite @xmath28 , \text { } n(x,0)=-\sqrt{2}% e_{0}k\beta\cos(kx ) , \label{4}%\ ] ] where @xmath24 is the amplitude of the pump langmuir wave field and @xmath29 is a constant of the order of @xmath30 to emphasize that the perturbation is very small .
we use runge - kutta scheme with time step size @xmath31 and consider the grid size @xmath32 so that @xmath33 corresponds to the grid position @xmath34 .
the spatial derivatives are approximated with centered second - order difference approximations .
the results are presented in figs . 1 - 6 after the end of the simulation with @xmath35 and @xmath36 ( upper panel ) and the associated density fluctuation @xmath37 ( lower panel ) with respect to the space @xmath38 after time @xmath35 in the numerical simulation of the qzes for @xmath39 , @xmath1 and @xmath40.,width=288,height=288 ] constant corresponding to the evolution as in fig
this shows that pattern selection leads to many solitary patterns which due to qiaw emission collide and get fused into few incoherent patterns after a long time .
the stc state emerges .
, width=288,height=288 ] note that as above the mi sets in for wave numbers satisfying @xmath41 where @xmath42 is a real root of the cubic @xmath43 and @xmath44 defines the curve along which pitchfork bifurcation takes place .
moreover , the dynamics is subsonic in the regime @xmath45 where the mi growth rate is small .
as @xmath46 is lowered from @xmath47 many unstable modes with higher harmonic modes will be excited . again
, the master mode can , in principle , result in the excitation of @xmath48 unstable harmonic modes where @xmath49 $ ] with @xmath50 there may also exist many solitary patterns with spatially modulated length @xmath51 where @xmath52 is for master mode ( @xmath53 ) and @xmath54 are for the unstable harmonic modes . as a result
, the envelope @xmath15 can be expressed as : @xmath55 @xmath56 in which the first term on the right - side of eq .
( [ 5 ] ) comes from the master mode and unstable harmonic modes with @xmath57 being due to pattern selection , whereas the second term is due to the nonlinear interaction among the patterns formed .
, @xmath58 and other parameters remain the same as in fig .
1.,width=288,height=288 ] constant with respect to space and time corresponding to the evolution as in fig .
this shows that pattern selection leads to many more solitary patterns than the classical case ( fig .
2 ) which due to qiaw emission collide and get fused into few incoherent patterns after a long time .
the collision is random and not confined between two patterns .
few stable modes are also excited .
the system energy is then redistributed to new incoherent patterns and few stable modes with small length scales .
the system is in the state of stc.,width=288,height=288 ] the profiles of the electric field and the density fluctuation at the end of simulation are shown in fig.1 for @xmath39 and @xmath1 .
we observe an excited electric field of the order @xmath59 highly correlated with density depletion @xmath60 or @xmath61 from fig .
2 , we find that many solitary patterns are formed from the master mode and unstable harmonic modes by means of pattern selection .
two solitary patterns initially peaked in @xmath62 collide to form a new mode which again collides with the master pattern peaked at @xmath63 and fuses into a new strong mode .
after few more other collisions and fusions , there remain only four distorted patterns . the system is then said to emerge the stc state . when @xmath46 is further lowered many solitary trains
are excited and saturated .
as soon as the collisions and fusions among the solitary patterns take place and the new incoherent trains are formed under strong qiaw emission , these incoherent patterns then collide with some others repeatedly and finally there remain few incoherent patterns in stochastic motion with the greater part of the system energy .
meanwhile many stronger modes are also excited to share the energy .
these are shown in figs . 3 , 4 and figs . 5 , 6 for different values of @xmath0 : @xmath64 and @xmath65 respectively .
we observe that though the number of remaining modes are the same for both the cases , however , many more ( compared to fig .
6 ) solitary patterns are seen to form initially ( fig .
4 ) which later participate into the collisions and fusions under qiaw emission . moreover , the new incoherent patterns formed with the lapse of long time are more stronger as in fig .
6 than those in fig .
4 . furthermore , for higher values of @xmath0 ( fig .
6 ) the collisions and fusions take place at an earlier time than the case of lower @xmath0 ( fig.4 ) .
thus , a certain amount of energy which was initially distributed among many solitary waves , will now be transferred to a few incoherent patterns as well as to some stable higher harmonic modes with short wave lengths .
so , if initially there form many solitary pattern trains due to mi with different modulational lengths , collision and fusion among most them can lead to the stc state . there must exist a critical wavelength or wave number @xmath66 at which the transition from temporal chaos to stc takes place .
this value , in the quantum case ( @xmath67 ) is quite different from the classical one ( @xmath1 ) . for detail investigations readers
are referred to works , e.g. , in refs .
@xcite . .
other parameters remain the same as in fig .
3 . , width=288,height=288 ] constant with respect to space and time corresponding to the evolution as in fig . 5 . here
the pattern selection leads to many solitary patterns , but less than that in fig .
4 . though , the collision is random and not confined between two patterns , but they fuse into new incoherent patterns in a shorter time than the case of @xmath64 ( fig .
4 ) due to strong qiaw emission .
the system is also in the stc state.,width=288,height=288 ]
we have performed a simulation study of the qzes to show that many coherent solitary patterns can be excited and saturated through the mi of unstable harmonic modes by a modulation wave number of qlws .
it is observed that there exist critical values of @xmath46 for which the motion of the coherent solitary patterns is in the state of tc or in stc .
the transition from tc to stc occurs when @xmath68 for @xmath1 or , for @xmath69when @xmath70 it is shown that the dispersion due to quantum tunneling induces strong qiaw emission leading to collision and fusion among the patterns to occur at an earlier time than the classical case .
the collision and fusion among some trains take place and the new incoherent pattern trains are formed accompanying strong qiaw emission due to quantum effects . the stc state is then said to emerge . as a result ,
the system energy in the stc state is spatially redistributed in the process of pattern collision , fusion and distortion , which may switch on the onset of langmuir turbulence in quantum plasmas .
this work was prepared in honor of professor p. k. shukla s 60th birthday .
m gratefully acknowledges support from the kempe foundations , sweden . | wave - wave interaction in plasmas is a topic of important research since the 16th century .
the formation of langmuir solitons through the coupling of high - frequency ( hf ) langmuir and low - frequency ( lf ) ion - acoustic waves , is one of the most interesting features in the context of turbulence in modern plasma physics .
moreover , quantum plasmas , which are ubiquitous in ultrasmall electronic devices , micromechanical systems as well as in dense astrophysical environments are a topic of current research . in the light of notable interests in such quantum plasmas , we present here a theoretical investigation on the nonlinear interaction of quantum langmuir waves ( qlws ) and quantum ion - acoustic waves ( qiaws ) , which are governed by the one - dimensional quantum zakharov equations ( qzes ) .
it is shown that a transition to spatiotemporal chaos ( stc ) occurs when the length scale of excitation of linear modes is larger than that of the most unstable ones .
such length scale is , however , to be larger ( compared to the classical one ) in presence of the quantum tunneling effect .
the latter induces strong qiaw emission leading to the occurrence of collision and fusion among the patterns at an earlier time than the classical case .
moreover , numerical simulation of the qzes reveals that many solitary patterns can be excited and saturated through the modulational instability ( mi ) of unstable harmonic modes . in a longer time , these solitons are seen to appear in the state of stc due to strong qiaw emission as well as by the collision and fusion in stochastic motion .
the energy in the system is thus strongly redistributed , which may switch on the onset of langmuir turbulence in quantum plasmas . |
the first reduction games which have appeared in the literature are perhaps the lipschitz game @xmath2 and the wadge game @xmath3 ( both were defined by wadge in his ph.d .
thesis , see @xcite ) .
they are a special kind of infinite two - player zero sum games on @xmath0 with perfect information , and are designed in such a way that if player @xmath4 has a _ legal _ strategy @xmath5 in @xmath2 ( resp .
@xmath3 ) then from @xmath5 it can be recovered in a canonical and fixed way a function @xmath6 from the baire space @xmath7 into itself which is lipschitz with constant @xmath8 ( resp .
continuous ) .
conversely , given a lipschitz with constant @xmath8 ( resp . continuous ) function @xmath9 one can construct a legal strategy @xmath5 for player @xmath4 in the corresponding game such that @xmath10 .
these games were introduced to study the relations ( also called _ reducibilities _ ) @xmath11 and @xmath12 , where for every @xmath13 @xmath14 the link between these preorders and the corresponding games is the following : given @xmath15 , a payoff set for @xmath2 ( resp .
@xmath3 ) is canonically constructed ( see section [ sectionreductiongames ] ) in such a way that player @xmath4 has a _ winning _ strategy in @xmath16 ( resp .
@xmath17 ) if and only if @xmath18 ( resp .
@xmath19 ) .
assuming the _ axiom of determinacy _
@xmath20 , or even just the determinacy of the corresponding games @xmath2 and @xmath3 , wadge proved that both @xmath11 and @xmath12 induce well - behaved stratifications of the subsets of @xmath7 which have turned out to be very useful in various parts of set theory ( see e.g. @xcite ) .
some years later , van wesep defined , building on work of wadge , another reduction game , the _ backtrack game _ @xmath21 , but at that time it was not clear which should be the `` topological '' class of functions @xmath22 corresponding to legal strategies for player @xmath4 in @xmath21 .
it was andretta who solved this problem in @xcite , by showing that such @xmath22 is exactly the collection of those @xmath23 for which there is a partition @xmath24 of @xmath7 into closed sets such that @xmath25 is continuous for every @xmath26 ( see ( * ? ? ? * theorem 21 ) ) , which in turn coincide with the collection of the @xmath27-functions by a theorem of jayne and rogers ( see e.g. ( * ? ? ? * theorem 1.1 ) for a proof of this last result ) .
another reduction game , namely the _ eraser game _
@xmath28 , was defined ( essentially ) by duparc in such a way that @xmath29 is a baire class @xmath8 function if and only if there is some legal strategy @xmath5 for @xmath4 in @xmath30 such that @xmath10 .
finally , some work related to this topic was developed in @xcite ( although in this case there are no reduction games directly involved ) .
having all these useful reduction games , it is quite natural to ask if one could also define reduction games for other `` natural '' collections of functions ( this question was explicitly posed by andretta in his @xcite : `` is there a wadge - style game for higher levels of reducibility , like @xmath31 and such ? '' ) .
more precisely : say that a set of functions @xmath22 is _ playable _ if there is some reduction game @xmath32 such that for every @xmath33 , @xmath34 if and only if there is a legal strategy @xmath5 for @xmath4 in @xmath32 for which @xmath35 ( this notion will be completely formalized in section [ sectionreductiongames ] ) .
clearly not every set of functions is playable : for example , the collection of _ all _ functions from @xmath7 into itself is not playable , as a simple cardinality argument shows ( the strategies for _ any _ game on @xmath0 are always at most @xmath36 ) .
nevertheless we can ask the following : [ question ] which ( topologically defined ) classes of functions are playable ?
the motivation for this problem mainly relies on the fact that the presence of a reduction game provides combinatorial tools for the study of the reducibility induced by the corresponding set of functions see e.g. @xcite .
the first partial answer to this general problem was given by semmes in @xcite and in his ph.d .
thesis @xcite : there he proposed a game ( called _ tree game _ ) which corresponds to the borel functions , and some other games ( the _ multitape game _ @xmath37 , the _ multitape eraser _
game @xmath38 , and the game @xmath39 ) which correspond , respectively , to the functions strictly continuous on a @xmath40-partition , to the functions which are of baire class @xmath8 on a @xmath40-partition , and to the baire class @xmath41 functions see next section for the definitions of these classes of functions . in this paper
we somewhat extend these results providing a positive answer to question [ question ] for a wide class of subsets of the borel functions , and for @xmath1-measurable functions ( where @xmath1 is any boldface pointclass closed under countable unions and finite intersections ) : therefore the paper is in some respect unusual for a research publication in mathematics , as it mainly consists of definitions and of proofs that these definitions are correct . nevertheless , in the last section we will also provide some applications of these games which motivate our interest in this subject .
the material of this paper ( except for section [ sectiongamma ] ) mainly comes from sections 2.2 .
and 4.8 of the author s ph.d .
thesis @xcite or is obtained via minor variations of the constructions contained therein , but for the reader s convenience ( and to avoid confusions ) in the present paper we have adapted most of the terminology and notation used in @xcite to the one already used in @xcite , with the following exception : because of the applications of reduction games to reducibilities for sets of reals given in section [ sectionaxioms ] , in this paper the payoff set of a reduction game will be defined starting from two sets of reals ( see section [ sectionreductiongames ] ) , whereas in @xcite it was defined starting from a ( partial ) function from the reals into the reals ( nevertheless , it is quite easy to see how to modify one kind of presentation into the other ) .
the constructions we are going to present rely on a very general way of defining games for sets of functions which are piecewise defined , for sets of functions which are ( pointwise ) limits of certain sequences of functions , and for @xmath1-measurable functions : most of the proofs involve some sort of operation for games which allows to transform a sequence of already known reduction games ( representing some classes of functions ) into a new reduction game which represent the larger class of those functions piecewise in the old classes on a definable partition , or the class of the pointwise limits of the old functions .
the paper is reasonably self - contained and is organized as follows : in section [ sectionpreliminaries ] we will fix some notation , while in section [ sectionreductiongames ] we will give a precise definition of what should be meant by _ reduction game _ and _ playable _ set of functions , and give some basic examples ( both old and new ) . in sections [ sectionplayable ] and [ sectiongamma ] we will prove our main results , showing how to construct new reduction games ( building on other known games ) : this will give a `` uniform '' solution to our problem for almost all sets of functions involved in ( generalizations of ) wadge s theory , for @xmath1-measurable functions , and for some other related sets of functions .
in section [ sectionaxioms ] we will give some examples of how to apply the techniques arising from these games to the study of various reducibilities , and we will prove some relationships between the corresponding determinacy axioms .
finally , in section [ sectionsmallness ] we will give the optimal condition under which constructions like those presented in section [ sectionplayable ] can be carried out : even if this technical improvement allows to deal with a strictly larger class of sets of functions , for the sake of simplicity we have postponed it to the last section because it complicates very much the presentation without adding relevant ideas for the construction of the new reduction games .
in most of the applications involving games , one usually assumes @xmath20 ( or some other axiom of this kind ) and then uses the combinatorics arising from the winning strategies in the games under consideration to prove the desired results .
however , @xmath20 ( and , in general , all known determinacy principles which are not restricted to the context of a small definable pointclass , like the pointclass of borel sets ) contradicts the full axiom of choice @xmath42 , and therefore in presenting new games and their applications one has to be careful and just use choice principles which do not contradict @xmath20 . in this paper
, we will always work in @xmath43 except for section [ sectionaxioms ] , in which we will sometimes need the _ axiom of dependent choice ( over the reals ) _ @xmath44 and the axiom @xmath45 , that is the statement `` every set of reals has the baire property '' .
our notation is quite standard and we refer the reader to the monograph @xcite for all the undefined symbols and notions . given two sets @xmath46 and @xmath47 , we will denote by @xmath48 the collection of all functions from @xmath47 to @xmath46 .
thus we will denote by @xmath49 the set of all @xmath0-sequences of elements of @xmath46 , while the collection of the _ finite _ sequences of elements of @xmath46 will be denoted by @xmath50 ( we will refer to the length of a finite sequence @xmath51 with the symbol @xmath52 ) .
as usual in descriptive set theory , the elements of the baire space @xmath7 will be called _
reals_. if @xmath53 we will write @xmath54 for the sequence @xmath55 and @xmath56 for the @xmath0-sequence @xmath57 . for simplicity of notation , we will also put @xmath58 , @xmath59 and @xmath60 .
let @xmath61 be the bijection @xmath62 .
then we can define the homeomorphism ( where @xmath49 is endowed with the product topology of the discrete topology on @xmath46 ) @xmath63 and , conversely , the `` projections '' @xmath64 defined by @xmath65 : clearly , every @xmath66 is surjective , continuous and open .
unless otherwise specified , in what follows all functions should be intended as _ partial _ functions , i.e. defined just on some @xmath67 ( endowed with the relative topology inherited from @xmath7 ) , and not necessarily on the whole space @xmath7 .
we denote by @xmath68 the collection of the lipschitz functions with constant less or equal than @xmath69 , and put @xmath70 .
since they played a special role as reducibilities , we will denote by @xmath71 the set @xmath72 , and by @xmath73 the set of all continuous functions .
moreover , the collection of the baire class @xmath74 functions ( equivalently , @xmath75-measurable functions see the next paragraph for the definition ) will be denoted by @xmath76 . finally , given any nonzero countable ordinal @xmath74
, we will denote by @xmath77 the collection of all _ @xmath78-functions _ ,
i.e. of those @xmath79 such that @xmath80 for every @xmath81 .
a pointclass @xmath82 is said to be _ boldface _ if it is closed under continuous preimages , and is called _ @xmath83-pointclass _ if it is boldface and closed under countable unions and finite intersections . a set @xmath84 is said to be _
universal for @xmath1 _ if @xmath85 , where @xmath86 .
a function @xmath79 is _
@xmath1-measurable _ if @xmath87 for every open set @xmath88 .
the collection of such functions is denoted by @xmath89 .
note that if @xmath1 is a @xmath83-pointclass then @xmath90 if and only if @xmath91 for any clopen subbasis @xmath92 of the topology of @xmath7 .
@xmath1-partition _ of @xmath67 is a family @xmath93 of pairwise disjoint sets of @xmath94 such that @xmath95 .
if @xmath1 is a @xmath83-pointclass then every @xmath1-partition of @xmath96 is automatically a @xmath97-partition . given a sequence for every @xmath98 we will systematically use the symbol @xmath22 instead of @xmath99 in all the notation . ] @xmath100 of sets of functions and a @xmath83-pointclass @xmath1 , we will denote by @xmath101 the collection of those @xmath102 for which there is a @xmath1-partition ( equivalently , a @xmath97-partition ) @xmath103 of @xmath96 and a sequence @xmath104 of functions _ with domain @xmath96 _ such that @xmath105 and @xmath106 for every @xmath26 . if @xmath107 for some @xmath74 , we will simply write @xmath108 instead of @xmath109 ( in this case we can replace `` @xmath78-partition '' with `` @xmath110-partition '' in the definition above ) . in particular , a function in @xmath111 will be said _
continuous on a @xmath78-partition _
( equiv .
_ on a @xmath110-partition _ ) .
the following minor variation of the previous definition in general gives a different set of functions see ( * ? ? ?
* remark 6.2 ) : @xmath112 ( resp.@xmath113 ) denotes the collection of those @xmath79 for which there is a @xmath97-partition ( resp . a @xmath78-partition or , equivalently , a @xmath110-partition ) @xmath103 of @xmath96 such that @xmath114 for every @xmath26 .
a function in @xmath115 will be said _ strictly continuous on a @xmath78-partition _
( equiv .
_ on a @xmath110-partition _ ) .
both the previous definitions are useful : for instance @xmath111 has been used in ( * ? ? ?
* section 6 ) as a natural example of borel - amenable reducibility , while @xmath116 has been used in ( * ? ? ?
* section 5.2 ) to find a generalization for the level @xmath117 of the theorem of jayne and rogers mentioned in the introduction , i.e. to show that @xmath118 . finally , given @xmath99 as above , we will denote by @xmath119 the collection of those @xmath79 for which there is a sequence of functions @xmath120 with domain @xmath96 such that @xmath105 and @xmath121 is the _ pointwise _ limit of the sequence @xmath122 . in section [ sectionaxioms ]
, we will deal with various reducibilities for sets of reals and with the corresponding hierarchies of complexity of @xmath123 .
therefore for all the terminology and the results about these concepts we refer the reader to @xcite in fact we suggest to keep a copy of that paper while reading that section in order to compare the various results with the combinatorial arguments proposed here .
the unique modification is that here we will sometimes consider reductions from some @xmath67 to @xmath7 : given such an @xmath96 , a set of functions @xmath22 with domain @xmath96 , and sets @xmath15 , we say that @xmath46 is _
@xmath22-reducible _ to @xmath47 ( in symbols @xmath124 ) just in case there is some @xmath34 such that _ for every @xmath125 _
@xmath126 that is @xmath127 .
finally , given a set @xmath22 of _ totally defined _
functions , recall that the _ semi - linear ordering principle _ for @xmath22 ( denoted by @xmath128 ) is the statement @xmath129 where @xmath130 denotes @xmath131 .
as recalled in the introduction , the first examples of reduction games are @xmath2 , @xmath3 , @xmath21 and @xmath30 ( defined in ( * ? ? ? * and 64 ) , @xcite , and @xcite , respectively ) , which represent , respectively , the classes of functions @xmath71 , @xmath73 , @xmath132 and @xmath133 ( see e.g. ( * ? ? ?
* theorem b8 ) and ( * ? ? ?
* theorems 3.1 , 4.1 , and 5.1 ) ) .
here we just give a brief and informal description of the rules of these games .
in @xmath2 , both @xmath134 and @xmath4 have to play a natural number at each of their turns .
the game @xmath3 is a variation of @xmath2 in which @xmath4 has the further option of `` passing '' ( i.e. skipping her move at some turn ) , but with the condition that at the end of the run she has played infinitely many natural numbers , i.e. she has enumerated a real .
the backtrack game @xmath21 is a further variation of @xmath3 in which @xmath4 can still pass ( with the same condition above ) , but even _ backtrack _ , i.e. she can delete all her previous moves at once and start to play natural numbers ( or pass ) anew , with the restriction that in each run she can use this option only finitely many times ( this guarantees that at the end of the run she has indeed played some real ) . finally , the eraser game @xmath30 can be described in the following way : @xmath134 must play a natural number on each of his turn , while @xmath4 can either play a natural number or erase the last natural number which appears on her board , but to guarantee that at the end of each run @xmath4 has indeed played a real , we require that for each @xmath125 and each @xmath26 there must be some @xmath135 such that for every @xmath136 we have @xmath137 , where @xmath138 is the sequence of natural numbers that @xmath4 has played ( after possible erasings ) when @xmath134 has enumerated @xmath139 . in other words , @xmath4 has to enumerate a real @xmath140 and she has the option of changing the @xmath98-th digit of @xmath141 at any time , but for each @xmath98 she can take this option only finitely many times .
many other games ( both old and new ) can be obtained by modifying one of these games , for example : * a simple variation of @xmath3 leads to the game @xmath142 , in which @xmath4 can still pass but _ at most _ @xmath143 times in a run ( this is also equivalent to requiring that @xmath4 pass for the first @xmath143 turns and then plays the rest of the game without passing any more[klip ] ) : as we will see in proposition [ theorlipstrategies ] , the strategies for @xmath4 in @xmath144 induce exactly the functions in @xmath68 ; * a variation of @xmath21 leads instead to the game @xmath145 in which @xmath4 can still backtrack finitely many times but is no more allowed to pass : legal strategies for @xmath4 in @xmath145 induce exactly the functions in @xmath146 ( which is a proper subset of @xmath132 see @xcite ) ; * another variation of @xmath3 leads to the multitape game @xmath37 defined by semmes in @xcite , which can be described as follows : @xmath134 has to play his numbers on a single rows , while @xmath4 has to play her numbers on a table containing @xmath0-many rows . at each turn
, @xmath134 plays a natural number on his unique row , while @xmath4 has first to choose one of her rows and then either pass or play a new natural number on it , with the condition that in each run of the game she has to play an infinite amount of numbers on exactly one of her rows : by andretta - semmes ( * ? ? ?
* theorem 6.1 ) , legal strategies for @xmath4 in this game induce the functions in @xmath147 ; * the game @xmath30 can be `` iterated '' by using @xmath98 eraser operators ranked with a priority in order to obtain games @xmath148 whose legal strategies for @xmath4 induce exactly the baire class @xmath98 functions .
various other reduction games , like the multitape eraser game @xmath38 defined in ( * ? ? ?
* section 7 ) which represents @xmath149 , can be obtained in a similar way . however , each of these games seems to be strictly related to the particular presentation of the game itself , and therefore it seems difficult to guess which should be the definition of a game representing a more complex class of functions , like e.g. @xmath150 , @xmath151 , or @xmath152 .
in order to have a uniform approach to the problem of representing classes of functions by means of games , it is useful to first abstractly define the notion of _ reduction game _ ( see subsection [ sectionredg ] ) , to isolate some basic examples of such games ( like the already defined @xmath3 , or games representing class of the form @xmath89 see subsection [ sectionexamples ] and section [ sectiongamma ] ) , and then find some operations which correspond to the analogous topological operations used in the definition of the new classes of functions , like the operation of taking pointwise limits , or of giving piecewise definitions on a definable partition of the space ( this is done in sections [ sectionplayable ] and [ sectionsmallness ] ) .
a reduction game is a tuple @xmath153 ( where @xmath154 is some symbol which identify the game ) such that @xmath67 , @xmath155 is a countable set disjoint from @xmath0 ( called _ set of moves _ ) , @xmath156 ( called _ set of rules _ ) and @xmath157 is a function from @xmath158 into @xmath7 ( called _ interpretation function _ )
. the rules of the game are as follows : @xmath134 plays elements of @xmath0 , while @xmath4 plays elements of @xmath159 , so that after @xmath0-many turns @xmath134 will have produced a real @xmath160 while @xmath4 will have produced @xmath161 ( @xmath141 is called _ complete play _ of @xmath4 ) .
then we give the following two conditions : 1 .
if @xmath162 then @xmath134 loses ; 2 . if @xmath125 but @xmath163 then @xmath4 loses .
if @xmath164 is a strategy for @xmath134 and @xmath141 is the complete play of @xmath4 in a run of @xmath32 , then @xmath165 denotes the real enumerated by @xmath134 while following @xmath164 against @xmath141 .
similarly , if @xmath5 is a strategy for @xmath4 and @xmath160 , we denote by @xmath166 the complete play produced by @xmath4 in the run of @xmath32 in which @xmath134 enumerates @xmath167 and @xmath4 plays according to @xmath5 ( @xmath168 and @xmath169 are defined in a similar way for every @xmath170 and @xmath171 ) . a strategy @xmath164 for @xmath134
is said to be _ legal _ if @xmath172 for every @xmath161 , while a strategy @xmath5 for @xmath4 is said to be _ legal _ if @xmath173 for every @xmath125
. the collection of legal strategies for @xmath4 in @xmath32 will be denoted by @xmath174 .
given @xmath13 , @xmath175 is defined by the following winning condition : if neither ( 1 ) or ( 2 ) have occurred , then @xmath4 wins if and only if @xmath176 ( in this case @xmath177 are called _ payoff sets _ of @xmath175 and @xmath178 is called _ play _ or _
output real _ of @xmath4 ) .
a strategy ( for either @xmath134 or @xmath4 ) is said to be _ winning _ in the game @xmath175 if it is legal and always guarantees the victory of the corresponding player , whatever his or her opponent plays .
notice that every @xmath179 canonically induces the unique function @xmath180 ( in this case we will say that @xmath5 _ represents _ @xmath121 ) , whereas for some function @xmath79 there can be distinct @xmath181 such that @xmath182 .
we will put @xmath183 . with this notation
, @xmath4 has a _ winning _ strategy in the game @xmath175 if and only if @xmath46 is _
@xmath184-reducible _ to @xmath47 : this is why the games described above are called _ reduction _ games . usually , the topological definition of a certain class of functions is virtually independent from the particular domain of the functions under consideration ( apart from its topology , of course ) , meaning that the definition of such class uses @xmath96 just as a sort of parameter .
for example , a function is said to be continuous if the preimage of an open set is still open , and this definition does not involve any other information on the domain of the function except for its topology
. therefore , to have a decent notion of _ representation _ of a ( topologically defined ) class of functions @xmath22 ( with domain arbitrary subsets of @xmath7 ) by means of a certain _ set _ of reduction games @xmath185 , it seems natural to stipulate by convention that all the games in @xmath186 share the same set of moves , set of rules and interpretation function , or at least , when considering games related to boldface pointclasses @xmath1 larger than the collection of borel sets . ] that all sets of rules ( and consequently all the interpretation functions ) of the games in @xmath186 are defined by a single formula which uses @xmath96 as a parameter .
such classes of games are said to be _
parametrized_. with a little abuse of notation , if @xmath186 is a parametrized class of games we will denote by @xmath184 again the collection of all functions induced by the legal strategies _ in one of the games of @xmath186_. let @xmath22 be any set of functions from subsets of @xmath7 into @xmath7 .
we say that @xmath22 is _ playable _ if there is a parametrized class @xmath186 of reduction games such that @xmath187 .
in this case , we also say that the class @xmath186 _ represents _ @xmath22 .
finally , for @xmath188 we will denote by @xmath189 ( or simply @xmath190 ) the statement : `` for all @xmath13 , the game @xmath175 is determined ( i.e. either @xmath134 or @xmath4 has a winning strategy ) '' .
@xmath190 is obviously a consequence of @xmath20 ( to see this it is enough to `` code '' in the natural way the reduction game @xmath32 into a classical game on @xmath0 ) .
moreover , if @xmath158 and @xmath157 are not too complicated , the above implications have also `` local versions '' .
for example , if @xmath158 is a borel subset of @xmath191 ( endowed with the product topology ) and @xmath157 is borel , then borel determinacy is sufficient to have that for every _ borel _ @xmath13 the game @xmath175 is determined ( and , more generally , local versions of @xmath20 imply local versions of @xmath190 ) . as we will see , if @xmath22 is a playable subset of the borel functions it is in practice always the case that the parametrized class @xmath186 of reduction games which represents @xmath22 has a borel set of rules and a borel interpretation function ( defined independently from @xmath96 ) .
now we want to give some examples on how to formalize the games presented at the beginning of this section into reduction games .
as the names suggest , the set @xmath155 will be used to code the alternative moves ( like `` pass '' , `` backtrack '' , `` erase '' , and so on ) of @xmath4 , @xmath158 will be used to code the rules of the game ( that is the rules that @xmath4 must respect in order to have a chance of victory ) , and @xmath157 will be used to recover from the ( play of @xmath134 and the ) complete play of @xmath4 the real that must be used in checking the winning condition .
we will present just three cases , namely continuous functions , baire class @xmath8 functions , and lipschitz functions with constant @xmath69 : however , we will prove that the corresponding reduction game really represents the desired class of functions just for the last case , as the other two well - known proofs can be obtained using classical arguments ( see e.g. ( * ? ? ?
* theorems 3.1 and 5.1 ) ) .
[ exw ] define the reduction game @xmath192 by : - @xmath193 ( the symbol @xmath194 will be interpreted as `` pass '' ) ; - @xmath195 ; - @xmath196 .
a function @xmath79 is continuous if and only if @xmath197 for some @xmath198 .
[ exe ] define the reduction game @xmath199 by : - @xmath200 ( the symbol @xmath201 will be interpreted as `` erase '' , and will correspond to the backspace key of a usual computer keyboard ) ; - for @xmath202 , inductively define @xmath203 by letting @xmath204 , @xmath205 ( for @xmath26 ) , and @xmath206 ; - @xmath207 ; - @xmath208 , where @xmath209 is the smallest @xmath135 such that @xmath210 .
a function @xmath79 is of baire class @xmath8 if and only if @xmath10 for some @xmath211 .
define the reduction game @xmath212 by as the game in which @xmath4 pass exactly for the first @xmath143 turns see page . ] : - @xmath213 ( the symbol @xmath194 will be interpreted as `` pass '' again ) ; - @xmath214 ; - @xmath215 .
[ theorlipstrategies ] for every @xmath216 , a function @xmath79 is in @xmath217 if and only if there is some @xmath218 such that @xmath10 .
let @xmath218 .
then @xmath219 has length @xmath220 , which implies that @xmath221 for every @xmath222 . for the other direction ,
let @xmath79 be in @xmath68 .
given @xmath223 , let @xmath224 be the unique sequence such that @xmath225 ( such a sequence must exists since @xmath226 ) , and define the strategy @xmath5 for @xmath4 in @xmath142 by @xmath227 if @xmath228 and @xmath229 otherwise .
it is clear that @xmath5 is legal and such that @xmath10 .
let us start with a technical definition .
[ defadequate ] let @xmath153 be a reduction game and @xmath184 be the set of functions induced by the strategies in @xmath174 .
we say that @xmath32 ( or @xmath184 ) is _ @xmath194-closed _ if @xmath184 coincides with the set of functions induced by the legal strategies for @xmath4 in the new reduction game @xmath230 defined by : - @xmath231 , where @xmath194 is a new symbol not in @xmath159 ; - @xmath232 ; - @xmath233 . a set of functions @xmath22 is _ adequate _ if it contains the identity function and there is a parametrized class @xmath186 of @xmath194-closed reduction games which represents @xmath22 . roughly speaking
, the condition for a set @xmath184 of being @xmath194-closed is the natural counterpart in terms of strategies of the property of being closed under right - composition with continuous functions from @xmath96 into itself , while a set of functions is adequate if it is not too small .
the previous definition could seem a little bit obscure , but it covers almost all the important cases and is designed in such a way that the arguments presented in the next subsections can be carried out in a very general way , adequateness is not the optimal condition for our purpose : nevertheless , it allows to give an easier presentation on the subsequent constructions avoiding some technical and notational complications , and thus seems to be a good compromise between generality of the arguments and clearness of exposition . ] .
for instance , it is obvious that @xmath3 is @xmath194-closed , but let us check as a nontrivial example that @xmath30 ( and hence @xmath133 ) is @xmath194-closed as well .
consider the game @xmath234 : clearly @xmath235 and , conversely , every @xmath236 can be converted in a legal strategy for @xmath4 in @xmath30 by substituting every use of the symbol @xmath194 with the pair of moves `` play @xmath237 and then play @xmath201 '' .
however , one has also to notice that not all the sets of functions considered in this paper are @xmath194-closed for a counterexample just take @xmath68 . in the next subsections
we will show how to construct games for the classes of functions @xmath108 , @xmath238 and @xmath239 , provided that @xmath74 is some fixed countable nonzero ordinal and @xmath99 is a countable sequence of not too small ( i.e. adequate ) playable sets of functions ( albeit for simplicity of presentation in subsections [ sectionpiecewise ] and [ sectiontilde ] we will just deal with the simpler case in which @xmath99 is constantly equal to some fixed adequate @xmath22 : this covers all the most important cases that one encounters in practice , and the general case can easily be recovered from these particular examples ) .
taking @xmath240 in theorems [ theorgames ] and [ theorgames2 ] we get , in particular , a generalization of the games @xmath21 and @xmath37 for ( simultaneously ) _ all _ higher level ( the existence of such games was still an open problem ) . moreover , since it is easy to see that the constructions below always produce reduction games which are @xmath194-closed , by iterating those constructions one can get a wide class of reduction games , namely games representing @xmath241 and @xmath242 for every @xmath243 .
fix any increasing sequence of ordinals @xmath245 cofinal in @xmath74 and , for each @xmath246 , a set @xmath247 which is @xmath248-complete ( we will see in claim [ claimchangecontrolsets ] that the choice of the @xmath249 s and of the @xmath247 s is not essential ) .
let @xmath22 be an adequate set of functions , and let @xmath250 be a @xmath194-closed reduction game representing the subset of @xmath22 consisting of those function which have domain @xmath96 .
let @xmath251 be the game defined in example [ exw ] representing the set of ( totally defined ) continuous functions @xmath73 .
now define @xmath252 as follows : - @xmath253 ; - @xmath254 ; - @xmath255 , where @xmath256 is smallest such that @xmath257 .
the game @xmath258 can be visualized as a two - player game in which @xmath134 has to fill in a table with a single row , while @xmath4 has to fill in a _ table with @xmath0-many rows_. [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] at the @xmath143-th turn , @xmath134 plays a natural number on his ( unique ) row , while player @xmath4 has two options ( in what follows @xmath98 is the unique natural number such that @xmath259 for some / any @xmath260 ) : pass or play a natural number on the @xmath98-th row if @xmath261 is even ( but at the end of the run she must have played infinitely many natural numbers on such row , i.e. she must have produced a real @xmath262 on it ) , or else play an element of @xmath159 on her @xmath98-th rows if @xmath98 is odd .
the even rows are _ control _ rows which can _ activate _ the rows immediately below them ( the odd ones ) , and this happens exactly when the real @xmath262 played on the @xmath263-th row belongs to the _ control set _ @xmath264 . in every run of the game , @xmath4 has to activate _ at least _ one of the odd rows , and she has to make sure that the sequence @xmath265 she has played on the @xmath266-st row belongs to the set of rules @xmath158 for every @xmath267 ( i.e. she must `` respect the rules '' of @xmath32 on the odd rows ) .
finally , the output real played by @xmath4 is exactly @xmath268 , where @xmath256 is least such that the @xmath266-st row is activated . +
every strategy @xmath5 for @xmath4 in @xmath258 can be seen as a sequence @xmath269 of legal strategies for @xmath4 in the game @xmath270 or in @xmath271 ( depending on whether @xmath98 is even or odd ) , each of which is used on the corresponding row .
in fact , if @xmath272 is such a sequence we can define the strategy @xmath273 for @xmath4 in @xmath258 as follows : for each @xmath274 , let @xmath275 be such that @xmath276 , and define @xmath277 .
it is not hard to check that @xmath273 is a ( non necessarily legal ) strategy for @xmath4 in @xmath258 such that @xmath278 for every @xmath125 .
thus to define a legal strategy for @xmath4 in @xmath258 it is enough to give a sequence @xmath279 of legal strategies for @xmath4 in @xmath3 ( resp .
@xmath32 ) if @xmath98 is even ( resp.odd ) , and check that for every @xmath125 there is some @xmath267 such that @xmath280 .
conversely , given a legal strategy @xmath5 for @xmath4 in @xmath258 and a natural number @xmath26 , we can define the legal strategy @xmath281 for @xmath4 in , respectively , @xmath282 if @xmath98 is even or in @xmath230 if @xmath98 is odd ( where @xmath282 and @xmath230 are defined as in definition [ defadequate ] ) as follows : for each @xmath283 , define @xmath284 if there is no @xmath285 such that @xmath276 , and @xmath286 otherwise . since both @xmath3 and @xmath32 are @xmath194-closed , with a little abuse of notation we will confuse each @xmath287 ( respectively , @xmath288 ) with any legal strategy in @xmath3 ( resp .
@xmath32 ) which induces the same function @xmath289 ( resp.@xmath290 ) on @xmath96 .
it is not hard to check that the operations @xmath66 on strategies `` commute '' with @xmath291 : given a strategy @xmath5 for @xmath4 in @xmath258 , for every @xmath125 and @xmath267 we have that @xmath292 and @xmath293 . in particular
, @xmath5 is a _ legal _ strategy for @xmath4 in @xmath258 if and only if @xmath294 is , and in the positive case the two strategies induce the same function on @xmath96 ( i.e.@xmath295 ) . + we will now prove that , as already announced , the choice of the ordinals @xmath249 s and of the sets @xmath247 s is not essential .
let @xmath296 be a ( non necessarily increasing ) sequence of ordinals cofinal in @xmath74 , and for every @xmath26 let @xmath297 be @xmath298-complete .
let @xmath299 be the game defined as at the beginning of this section but using the @xmath297 s instead of the @xmath247 s .
[ claimchangecontrolsets ] for every legal strategy @xmath300 for @xmath4 in @xmath299 there is a legal strategy @xmath5 for @xmath4 in @xmath258 such that @xmath301 , and , conversely , for every legal strategy @xmath302 for @xmath4 in @xmath258 there is a legal strategy @xmath303 for @xmath4 in @xmath299 such that @xmath304 .
therefore , @xmath258 and @xmath299 represent the same set of functions .
since the @xmath249 s are cofinal in @xmath74 and the @xmath247 s are @xmath248-complete , for every @xmath216 there is some @xmath305 such that @xmath306 , thus there is a winning strategy @xmath307 for @xmath4 in @xmath308 .
moreover , we can fix some @xmath309 for every @xmath98 , and for every @xmath140 let @xmath310 and @xmath311 be such that @xmath312 is constantly equal to @xmath141 and @xmath313 is the identity function . finally , given @xmath314
let @xmath315 be any strategy such that @xmath316 .
now define @xmath317 and @xmath318 if @xmath319 for some @xmath216 , and @xmath320 and @xmath321 otherwise .
finally put @xmath322 .
given @xmath160 , it is not hard to check that @xmath323 and @xmath324 , while @xmath325 for every @xmath98 which is not of the form @xmath305 for some @xmath216 .
hence @xmath5 is a legal strategy for @xmath4 in @xmath258 if and only if @xmath300 were a legal strategy for @xmath4 in @xmath299 , and moreover for every @xmath160 we have that @xmath326 , where @xmath143 is least such that @xmath327 ( which implies that @xmath305 is the least @xmath328 such that @xmath329 ) .
using the same argument , one can convert any @xmath330 into a @xmath331 such that @xmath332 , hence we are done .
thus from this point onward we can take the option of changing the sets @xmath247 s in the definition of @xmath258 at our pleasure , provided that the new ones are @xmath298-complete for some ( non necessarily increasing ) sequence of ordinals @xmath333 cofinal in @xmath74 .
[ theorgames ] for every @xmath334 and every @xmath335 we have that : a. @xmath336 if and only if there is some @xmath337 such that @xmath338 ; b. if @xmath134 has a winning strategy in @xmath339 , then @xmath134 has also a winning strategy in @xmath16 . in particular , in this case ,
@xmath340 , i.e. there is a _ contraction _
( that is a lipschitz function with constant @xmath341 ) @xmath342 such that @xmath343 for every @xmath160 , and @xmath344 .
first suppose that @xmath335 is in @xmath244 : let @xmath345 be a sequence of @xmath110-sets such that @xmath346 is a partition of @xmath96 , and @xmath347 be a sequence of continuous functions from @xmath96 into @xmath7 such that @xmath348 for every @xmath349 . by definition of the sets @xmath247 s
, we can find an increasing sequence of natural numbers @xmath305 such that @xmath4 has a winning strategy @xmath307 in @xmath350 for each @xmath216 .
moreover , we can choose the reals @xmath309 and define the strategies @xmath351 and @xmath352 as in the proof of claim [ claimchangecontrolsets ] . finally , since @xmath353 for each @xmath216
, we can find strategies @xmath354 such that @xmath355 , i.e. such that @xmath356 for every @xmath125 . now put @xmath357 and @xmath358 if @xmath319 for some @xmath216 , and @xmath359 and @xmath321 otherwise .
clearly @xmath273 is a legal strategy for @xmath4 in @xmath258 . moreover , for every @xmath125 there is a unique @xmath143 such that @xmath360 , so that @xmath361 just for @xmath362 : thus for every @xmath216 and @xmath360 @xmath363 conversely , given a legal strategy for @xmath4 in @xmath258 , define @xmath364 clearly the @xmath365 s form a @xmath78-partition of @xmath96 and @xmath366 for every @xmath98
. thus each @xmath367 induces a function @xmath368 in @xmath22 , and it is easy to check that @xmath369 for every @xmath26 , that is @xmath370 . finally , let @xmath302 be a winning strategy for @xmath134 in @xmath339 .
we define a strategy @xmath164 for @xmath134 in @xmath16 in the following way : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath140 be the real enumerated by @xmath4 in @xmath16 . consider the run of the auxiliary game @xmath371 in which @xmath134 enumerates @xmath141 and @xmath4 follows @xmath352
. now fix @xmath372 , and consider the run of a second auxiliary game @xmath373 in which @xmath134 follows @xmath302 and @xmath4 uses @xmath374 on the even rows , and `` copy '' the moves of @xmath4 in the previously described run of @xmath375 on the odd ones ( the strategy for @xmath4 defined in this way is clearly legal since @xmath372 ) . then at each turn
`` copy '' the corresponding move made by @xmath134 in the run of @xmath376 described above .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is not hard to check that since @xmath302 is winning then @xmath172 and @xmath377 : thus @xmath164 is a winning strategy for @xmath134 in @xmath16 .
the rest of part _ ii ) _ follows by standard arguments .
[ remsemmes ] although the game @xmath258 and the multitape game @xmath37 defined in @xcite were developed independently , one should notice that the easiest direction of the proof of theorem [ theorgames ] ( the one which goes from strategies to functions ) presents some affinity ( at least in spirit ) with the corresponding direction of the proof of ( * ? ? ?
* theorem 6.1 ) .
however , the definition of @xmath37 ( and , consequently , the whole theorem 6.1 of @xcite ) does not seem to admit a simple and straightforward generalization for higher levels : this should be contrasted with the definition of the games @xmath378 ( and of the games @xmath379 defined in the next subsection ) , which simultaneously gives a sort of `` uniform '' definition for all levels @xmath74 of games representing @xmath111 and @xmath380 in fact the games @xmath378 and @xmath379 can be seen as a direct `` translation '' of the definitions of @xmath111 and @xmath380 into the game - theoretic formalism .
we now want to prove that also the collection @xmath381 is playable by showing how to modify the game @xmath258 to obtain the game @xmath382 , which will represent this new set of functions .
the idea is to allow @xmath4 to not follow the rules on some of her rows .
here is the formal definition of @xmath383 : - @xmath384 ; - @xmath385 ; - @xmath386 , where @xmath256 is smallest such that @xmath257 .
thus the game @xmath382 can be visualized as the variant of the game @xmath258 in which @xmath4 must `` respect the rules '' _ just on all the activated rows _ ( rather than on all her odd rows ) .
every strategy @xmath5 for @xmath4 in @xmath382 can again be seen as a product @xmath387 of strategies @xmath388 for @xmath4 in @xmath270 and @xmath389 ( where now _
@xmath390 is a subset of @xmath96 _ depending on the index @xmath98 ) .
one can also define the projections @xmath66 on strategies as in the previous subsection , and check that they `` commute '' with the operation @xmath291 .
note that given a sequence of strategies @xmath388 as above , @xmath387 is a _ legal _ strategy for @xmath4 in @xmath382 if and only if @xmath391 and @xmath392 cover @xmath96 . using this fact one can prove the following theorem in a similar way to theorem [ theorgames ] .
[ theorgames2 ] for every @xmath334 and every @xmath335 we have that : a. @xmath393 if and only if there is some @xmath394 such that @xmath338 ; b. if @xmath134 has a winning strategy in @xmath395 , then @xmath134 has also a winning strategy in @xmath16 . the idea to require @xmath4 to fill a table with @xmath0-many rows
allows us also to define a ( quite trivial ) game for @xmath119 ( hence , in particular , for all @xmath76 s ) .
since in this case considering an arbitrary sequence @xmath397 of adequate playable sets of functions does not significatively increase the complexity of the presentation , we will not restrict ourselves to a constant @xmath99 .
suppose that the functions in @xmath398 have all domain @xmath96 , and let @xmath399 be a sequence of @xmath194-closed reduction games , each representing the corresponding @xmath398 .
the reduction game @xmath400 is defined as follows : - @xmath401 ; - @xmath402 ; - @xmath403 .
the game @xmath404 can be visualized as the game in which at each turn @xmath134 must play a natural number on his ( unique ) row , while @xmath4 has to play either a natural number or a symbol from @xmath405 on the @xmath98-th row of her table with @xmath0-many rows , with the condition that she must `` respect the rules '' of the corresponding game on each of these rows and that @xmath406 must exists , where @xmath407 is the value of @xmath408 on ( i.e. the `` interpretation '' of ) what @xmath4 has played on the @xmath98-th row : in this case , the output real of @xmath4 is exactly @xmath406 . as for the games @xmath258 , it is easy to check that every strategy @xmath5 for @xmath4 in @xmath404 can be decomposed into @xmath0-many strategies @xmath281 for @xmath4 in @xmath409 ( one for each row ) , and conversely @xmath0-many strategies @xmath388 for @xmath4 in @xmath409 can be coded up into a unique strategy @xmath410 for @xmath4 in @xmath404 .
moreover it is easy to check that @xmath411 is in @xmath239 if and only if there is some @xmath412 such that @xmath10 ( this is because the use of the table with @xmath0-many rows allows to directly code the definition of `` being limit of a sequence of functions '' into a single game ) .
let @xmath1 be any @xmath83-pointclass .
the main goal of this section is to construct games representing the collection @xmath89 of @xmath1-measurable functions @xmath79 .
when @xmath413 ( for some @xmath414 ) this just give an alternative way of defining games for baire class @xmath74 functions ( see section [ sectionlim ] ) , but note that since e.g. @xmath415 is trivially a @xmath83-pointclass ( for every @xmath26 ) , the main result of this section gives also a new way ( albeit less informative than the construction given in @xcite ) of defining a game for the class of all borel functions ( taking @xmath416 ) , and simultaneously solves the problem of finding games for the projective functions posed by semmes in his ph.d .
thesis @xcite .
fix a universal set @xmath417 for @xmath1 and let @xmath418 be the wadge game representing ( totally defined ) continuous functions .
here is the definition of the game @xmath419 : - @xmath420 ; - @xmath421 , where @xmath422 ; - @xmath423 , where @xmath424 is smallest such that @xmath425 .
the game @xmath426 can be visualized as follows : player @xmath134 must fill , as usual , a single row by playing a natural number at each of his turn ( thus he produces a real @xmath125 ) .
player @xmath4 is in charge of filling again a table with @xmath0-many rows : she can pass , but at the end of the round she must have enumerated a real @xmath427 on her @xmath98-th row ( for each @xmath246 ) .
the rules for @xmath4 are that each @xmath428 must code a set @xmath429 whose intersection with @xmath96 is in @xmath430 ( i.e. such that there is @xmath431 for which @xmath432 ) , and for every @xmath98 there must be an @xmath135 such that @xmath433 . the output real @xmath434
is then defined by @xmath435 if and only if @xmath135 is the smallest @xmath143 such that @xmath436 . as usual , any strategy @xmath5 for @xmath1
can be seen as a product @xmath387 of legal strategies for @xmath4 in @xmath3 , and one can define the projections @xmath66 of strategies in @xmath437 in such a way that they `` commute '' with the operation @xmath291 .
[ theorgamma ] for every @xmath334 and every @xmath79 we have that : a. @xmath90 if and only if there is some @xmath438 such that @xmath10 ; b. if @xmath134 has a winning strategy in @xmath439 , then @xmath134 has also a winning strategy in @xmath16 .
first assume that @xmath90 .
since the sets @xmath440 form a clopen subbasis for the usual topology of @xmath7 , we have that @xmath441 .
let @xmath442 be such that @xmath443 , @xmath444 be a code for @xmath445 , and @xmath125 be the real enumerated by @xmath134 : if we put @xmath273 , where @xmath446 is any strategy representing the constant function with value @xmath444 , then @xmath5 is clearly a legal strategy for @xmath4 in @xmath426 such that @xmath10 .
assume now @xmath447 .
then @xmath448 because @xmath449 , where @xmath450 is such that @xmath451 .
hence @xmath452 .
finally , let @xmath302 be a winning strategy for @xmath134 in @xmath439 , and @xmath453 be codes for , respectively , @xmath7 and @xmath454 ( as elements of @xmath1 ) .
then the strategy @xmath164 for @xmath134 in @xmath455 defined in the following way is clearly winning ( the proof being the same as in theorem [ theorgames ] ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath140 be the real enumerated by @xmath4 in @xmath16 , and consider the run of the auxiliary game @xmath439 in which @xmath134 plays according to @xmath302 and @xmath4 enumerates @xmath456 or @xmath457 on her @xmath458-th row depending on whether @xmath459 or @xmath460 ( since @xmath461 for any @xmath135 , this strategy for @xmath4 is clearly legal )
. then copy at each turn the corresponding move made by @xmath134 in the run of @xmath439 described above .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ notice that , contrarily to the games defined in all the previous sections , it is no more true that e.g. if @xmath462 then borel determinacy implies that @xmath439 , where @xmath463 , is determined .
this is because of the use of codes for sets in @xmath97 , which generally makes the set of rules more complicated than @xmath1 itself : in fact , in most cases , to say that `` @xmath167 codes a @xmath97-set '' require roughly speaking at least one real quantifier over a predicate of the same complexity as @xmath97 ( it is well - known e.g. that the set of codes for the borel sets forms a @xmath464-complete set ) . nevertheless ,
if @xmath465 ( that is if @xmath107 for some countable @xmath74 , being @xmath1 a @xmath83-pointclass ) one can redefine the games @xmath466 in such a way that the new sets of rules and the interpretation functions remain borel ( so borel determinacy will imply that these new games are determined whenever @xmath467 ) .
this can be obtained by fixing in advance a sequence of @xmath248-complete sets @xmath247 ( where @xmath468 is an increasing sequence of countable ordinals cofinal in @xmath74 ) as in the definition of the games @xmath258 , and then using the fact that for every @xmath469 set @xmath470 ( hence also for each @xmath471 , where @xmath472 and the @xmath473 s are defined as above ) there is a @xmath474-partition @xmath475 of @xmath96 such that @xmath476 for some @xmath477 : roughly speaking , in the new games player @xmath4 will have again to completely fill a board with @xmath0-many rows , but the function @xmath6 will be determined by checking which of the reals that appears on the rows of @xmath4 s table ( instead of the real @xmath167 enumerated by @xmath134 ) belongs to the corresponding set @xmath247 .
we leave to the reader the exact definition of these games , as well as the proof that they represent @xmath478 .
+ the same kind of construction introduced for the games @xmath426 allows also to define games for @xmath479 or @xmath480 ( where @xmath99 is a sequence of adequate playable sets of functions ) for an arbitrary @xmath83-pointclass @xmath1 . for simplicity of presentation
, we will deal again only with the case @xmath481 .
the idea is simply to take the game @xmath258 and , instead of fixing in advance the control sets @xmath247 , require @xmath4 to produce on each control row the _ code _ for some control set in @xmath430 : a ( non control ) row will be activated just in case the real @xmath125 enumerated by @xmath134 belongs to the set @xmath482 coded on the corresponding control row .
more precisely , given a @xmath194-closed reduction game @xmath483 representing the functions of @xmath22 with domain @xmath96 , define @xmath484 as follows ( where @xmath485 is the wadge game representing continuous functions ) : - @xmath486 ; - @xmath487 } \wedge { \exists i ( x \in d_{y , i } ) } \}$ ] , where @xmath488 and @xmath489 is a fixed universal set for @xmath1 ; - @xmath490 , where @xmath256 is smallest such that @xmath491 . combining the proofs of theorem [ theorgames ] and theorem [ theorgamma ] it is not hard to check that : for every @xmath334 and every @xmath79 we have that : a. @xmath492 if and only if there is some @xmath493 such that @xmath10 ; b. if @xmath134 has a winning strategy in @xmath494 , then @xmath134 has a winning strategy in @xmath16 as well .
in this section we will analyze the relationships among some determinacy axioms , and show how to apply the techniques arising from reduction games to the study of the reducibilities between sets of reals induced by the corresponding sets of functions . here
we will just present two cases , namely the cases corresponding to @xmath495 and @xmath111 ( for any fixed @xmath74 ) . notice that all reduction games used in this section are always intended to be of the form @xmath496 _ with
@xmath497_. we first consider the following axioms which are related to the games @xmath142 ( @xmath498 ) : this will lead in theorem [ theoraxioms ] to a slight extension of the results concerning the equivalence of some determinacy axioms obtained by andretta in his @xcite and @xcite , although we must note that the most difficult implication involved in such extension was already proved in those papers .
@xmath499 : : for every @xmath500 the game @xmath501 is determined .
@xmath502 : : for every @xmath13 there is some @xmath216 such that @xmath501 is determined .
@xmath503 : : for every @xmath13 and for every @xmath216 the game @xmath501 is determined .
[ lemmaaxioms ] a. @xmath504 ; b. @xmath505 for every @xmath216 ; c. @xmath506 .
the first part is obvious since @xmath20 easily implies that every reduction game is determined , and the equivalence and the first two implications of part _ ii ) _ are obvious as well
. the third implication of part _ ii ) _ can be proved using the trivial observation that winning strategies for @xmath134 in any of the games @xmath142 induce contractions , while the last implication follows from @xmath507 .
it remains only to prove part _
iii)_. fix @xmath508 and @xmath216 .
if @xmath509 then @xmath501 is trivially determined ( @xmath4 has a winning strategy if @xmath510 , and @xmath134 has a winning strategy if @xmath511 ) : hence we can assume @xmath512 and fix some @xmath513 .
consider the auxiliary game @xmath514 : if @xmath134 has a winning strategy @xmath164 for @xmath515 , @xmath134 can also win @xmath501 simply playing @xmath516 for the first @xmath143 turns ( i.e. for @xmath517 ) , and then playing @xmath518 if @xmath4 has enumerated a sequence of the form @xmath519 ( for some @xmath520 ) in the game @xmath501 , and @xmath237 otherwise .
conversely , if @xmath4 has a winning strategy @xmath5 in the game @xmath515 , then she can also win @xmath501 by playing @xmath194 for the first @xmath143 rounds , and then playing @xmath521 , where @xmath51 is the sequence enumerated by @xmath134 in @xmath501 , if @xmath522 , or enumerating @xmath141 otherwise .
[ theoraxioms ] let @xmath523 be one of the axioms @xmath499 , @xmath502 and @xmath503 . then @xmath524 . by lemma [ lemmaaxioms ]
we have @xmath525 , and since under @xmath526 we have from ( * ? ? ?
* proposition 15 and theorem 18 ) that @xmath527 , we get the desired equivalence .
in particular , this theorem implies that all the results about the @xmath495-hierarchy obtained in @xcite ( such as the fact that the structure induced by @xmath528 can be completely determined and is a well - founded semi - linear order whose antichains have size at most two , or the relationship between this hierarchy and the ones induced by @xmath11 and @xmath12 ) hold under any of the axioms listed above ( together with @xmath526 ) .
we now turn our attention to the @xmath111-hierarchies ( for some fixed nonzero @xmath414 ) .
recall from @xcite that there are two operations @xmath529 and @xmath530 such that @xmath531 are the successors of @xmath46 in the @xmath111-hierarchy whenever @xmath532 . here
are the definitions : @xmath533 and @xmath534 where the @xmath247 s are @xmath248-complete for an increasing sequence of ordinals @xmath468 cofinal in @xmath74 and @xmath535 .
there is a strict relationship between the games @xmath258 ( in particular when @xmath240 ) and these successor operations in fact our definition of @xmath258 was originally motivated by the definitions of @xmath529 and @xmath530 .
[ propgamessuccessor ] for every @xmath13 , the following are equivalent : a. @xmath536 ; b. @xmath537 _ and _ @xmath538 ; c. @xmath537 via some function @xmath121 such that @xmath539 ; d. @xmath538 via some function @xmath121 such that @xmath539 . obviously , _
iii)__@xmath540_iv ) _ since @xmath541 for every set @xmath542 . moreover , _
iii ) _ and _ iv ) _ together trivially imply _ ii ) _ , and _
i ) _ implies _ iii ) _ and _ iv ) _ since , by definition of @xmath378 , every winning strategy for @xmath4 in @xmath543 can be obviously converted into a winning strategy for @xmath4 in both @xmath544 and @xmath545 . to see that _ ii ) _ implies _ i ) _ , let @xmath546 and @xmath547 be , respectively , winning strategies for @xmath4 in @xmath544 and @xmath545 . as already observed in claim [ claimchangecontrolsets ] , we can change the sets @xmath247 in the definition of @xmath378 with some suitable @xmath297 s , and it will suffice to show that @xmath4 has a winning strategy in @xmath548 , where @xmath549 is the game defined using the @xmath297 s instead of the @xmath247 s .
choose for every @xmath26 and @xmath550 a strategy @xmath551 representing @xmath552 .
then put @xmath553 for every @xmath26 and set @xmath273 , where @xmath554 , @xmath555 , @xmath556 and @xmath557 ( @xmath216 ) .
notice that each @xmath297 is @xmath298-complete , where @xmath558 for @xmath559 ( so that @xmath333 is a sequence of ordinals cofinal in @xmath74 ) .
we claim that @xmath560 .
let @xmath160 .
since the @xmath561 s are _ winning _ strategies in the corresponding games , we have that for every real @xmath167 @xmath562 thus there must be some @xmath98 such that either @xmath563 or else @xmath564 .
but this implies that either @xmath565 or @xmath566 , hence @xmath5 is legal .
to finish the proof , let @xmath98 be the smallest natural number such that @xmath567 : then if @xmath568 ( @xmath569 ) we clearly have @xmath570 and @xmath571 where @xmath572 and @xmath573 .
therefore @xmath5 is a winning strategy for @xmath4 in @xmath548 .
as for the lipschitz games @xmath574 , the games @xmath378 allow us to introduce new determinacy axioms ( one for each @xmath74 ) : @xmath575 : : for every @xmath13 the game @xmath543 is determined . clearly @xmath576 , but proposition [ propgamessuccessor ] allows us to prove the following stronger corollary .
[ cordeterminacy ] @xmath577 implies @xmath575 .
let @xmath46 and @xmath47 be two subsets of @xmath7 . by @xmath577 , the games @xmath544 and @xmath545 are determined . if @xmath134 has a winning strategy in one of these two games , then he can convert this strategy into a winning strategy for @xmath134 in @xmath543 in the obvious way , hence we can assume that @xmath4 wins both the games .
but in this case @xmath4 has a winning strategy in @xmath543 by proposition [ propgamessuccessor ] , hence we are done .
there is a natural question arising from the previous corollary , namely : [ questionconverse ] assume @xmath578 . given a countable ordinal @xmath579 ,
does the converse to corollary [ cordeterminacy ] hold ?
this question was answered positively for @xmath580 by andretta in his @xcite , where it is shown that in fact @xmath581 ( which is a direct consequence of @xmath582 ) implies @xmath583 if we assume @xmath526 .
the proof is carried out with an induction on the @xmath132-hierarchy of degrees ( which can be determined under @xmath584 ) , but even if we will show that for any @xmath74 the axioms @xmath585 are indeed strong enough to determine the @xmath77-hierarchy of degrees as well ( see theorem [ theorstruc ] below ) , it seems that the argument used by andretta does not generalize in a straightforward way to higher levels .
therefore question [ questionconverse ] is still completely open for @xmath586 . we will now prove that , as announced in the previous paragraph , the axiom @xmath575 is strong enough either @xmath536 or @xmath587 , which by theorem [ theorgames ] is a ( seemingly weaker ) consequence of @xmath575 . ] to determine ( together with @xmath45 and @xmath44 ) the degree - structure induced by @xmath111 , or even by any borel - amenable set of reductions @xmath588 ( see @xcite for a general introduction to such degree - structures )
this shows that to study a borel - amenable reducibility @xmath22 we just need to assume an axiom which is `` of the same level '' of @xmath22 , rather than the seemingly stronger @xmath589 . toward our goal
, we will simply modify the arguments presented in @xcite whenever an axiom stronger than @xmath575 was required .
we start by proving a lemma ( analogous to ( * ? ? ?
* lemma 2.1 ) ) under the new axiomatization .
[ lemmabasic2 ] assume @xmath575 .
for every set of reductions @xmath590 and every @xmath13 we have @xmath591 . since @xmath592 , from @xmath593
we have @xmath594 .
but then @xmath4 can not win @xmath595 ( if this would happen , then @xmath596 and hence also @xmath597 ) .
therefore by @xmath575 we have that @xmath134 has a winning strategy in the same game , and hence @xmath18 by theorem [ theorgames ] .
moreover @xmath598 since otherwise @xmath599 ( here we use the fact that @xmath600 ) , and thus @xmath601 .
[ lemmawellfounded ] assume @xmath602 .
for every set of reductions @xmath588 the relation @xmath603 is well - founded
. it clearly suffices to prove that there is no @xmath603 descending chain the equivalence between this statement and well - foundness can be obtained in the usual way using the existence of a surjection @xmath604 ( see ( * ? ? ?
* corollary 2.2 ) ) .
so assume towards a contradiction that @xmath605 is such a chain .
we claim that @xmath606 and @xmath607 for every @xmath26 , i.e. that @xmath134 wins both @xmath608 and @xmath609 : applying then the classic martin - monk argument to these winning strategies , we can construct the flip - set which contradicts @xmath45 , finishing our proof . first note that for every @xmath26
, we have @xmath610 by @xmath128 ( which follows from @xmath575 ) .
from this fact one can conclude , arguing as in lemma [ lemmabasic2 ] , that @xmath4 can not win neither @xmath611 nor @xmath612 : but then @xmath134 wins both games by @xmath575 , and hence @xmath613 by theorem [ theorgames ] . [ theorstruc ] assume @xmath602 , and let @xmath588 be any borel - amenable set of reductions .
then the degree - structure induced by @xmath603 is completely determined and looks like the wadge one .
since @xmath614 and @xmath603 is well - founded by lemma [ lemmawellfounded ] , we have that theorem 3.1 and theorem 4.6 of @xcite are provable under our new axiomatization ( for part _ vii ) _ of theorem 3.1 , assume @xmath615 : since @xmath4 can not win neither @xmath616 nor @xmath617 , we have @xmath618 and @xmath619 by theorem [ theorgames ] ) .
moreover , ( * ? ? ?
* theorem 5.3 ) follows from @xmath45 alone , hence we are done . finally , notice
that we can also reprove theorem 4.7 of @xcite in this new context using lemma [ lemmabasic2 ] instead of ( * ? ? ?
* lemma 2.1 ) : therefore under @xmath602 we have that for every pair @xmath620 of borel - amenable sets of reductions , @xmath22 is equivalent to ( i.e. induces the same hierarchy of degrees as ) @xmath621 just in case they have the same characteristic set , that is just in case @xmath622
this final section is devoted to a technical refinement of the notion of being adequate for a certain set of functions in relationship to the possibility of representing such set by means of reduction games ( using the ideas coming from sections [ sectionplayable ] and [ sectiongamma ] ) .
we start with a significative example .
as observed after definition [ defadequate ] , the class of all lipschitz functions @xmath495 is not adequate ( being not @xmath194-closed ) , but still it is possible ( and useful ) to consider e.g. classes of the form @xmath623 , which are proper subsets of @xmath111 ( see @xcite ) . roughly speaking , in order to define the game @xmath624 ( whose legal strategies for @xmath4 will induce the functions in @xmath623 )
, it is enough to modify the algorithm to pass ( that is making @xmath4 always play a natural number on some of her rows ) does not give the desired result , because legal strategies for @xmath4 in such game would induce functions uniformly continuous ( rather than lipschitz ) on a definable partition and these two sets of functions are distinct by @xcite . ] that @xmath4 must follow to fill in the @xmath0-many rows of her table in the game @xmath258 : in fact , in the game @xmath624 , @xmath4 will have to _ simultaneously _ play a new natural number on a certain _ finite set _ of rows at each of her turns .
more precisely : for @xmath625 let @xmath626 be an enumeration without repetitions of @xmath627 , and let @xmath468 and @xmath24 be chosen as in subsection [ sectionpiecewise ] . then define @xmath628 by : as for the games @xmath258 , a strategy @xmath635 for @xmath4 in @xmath624 can be constructed from a sequence @xmath269 of strategies for @xmath4 in @xmath636 , where @xmath209 is the unique natural number such that either @xmath637 or @xmath638 : in fact it is enough to define @xmath639 , where @xmath135 is such that @xmath640 , and it is easy to check that @xmath641 just in case for every @xmath160 there is some @xmath26 such that @xmath642 .
conversely , given a strategy @xmath641 one can construct the strategies @xmath643 ( for @xmath550 ) by letting @xmath644 is @xmath228 and @xmath645 otherwise . assume first that @xmath646 , and let @xmath648 and @xmath649 be a sequence of @xmath110-sets such that @xmath346 is a partition of @xmath96 such that @xmath348 . by borel determinacy
is such that @xmath650 then @xmath651 : otherwise @xmath652 by @xmath653 for borel sets , contradicting the @xmath654-properness of @xmath655 . ]
, we can find an increasing sequence @xmath656 of natural numbers such that @xmath651 .
moreover , let @xmath657 be smallest such that @xmath658 , and inductively define @xmath659 and @xmath660 , so that @xmath661 ( hence also @xmath662 ) and @xmath663 for every @xmath216 .
let @xmath307 be a winning strategy for @xmath4 in @xmath664 and @xmath665 be such that @xmath666 . finally , fix @xmath309 and for every @xmath140 let @xmath667 be such that @xmath668 is constantly equal to @xmath141 .
now define @xmath357 and @xmath669 if @xmath670 , and @xmath671 otherwise . if we construct the strategy @xmath635 for @xmath4 in @xmath624 as explained above , it is not hard to check that @xmath360 if and only if @xmath672 , and that in this case @xmath673 for @xmath674 and @xmath675 conversely , let @xmath365 be the set of those @xmath167 for which @xmath98 is least such that @xmath676 .
clearly these @xmath365 s form a @xmath78-partition of @xmath96 and , as already observed , @xmath677 induces a function @xmath678 .
thus @xmath679 and we are done .
theorem [ theorglipxi ] clearly allows us to reprove all the results of section [ sectionaxioms ] ( except for proposition [ propgamessuccessor ] and its corollary ) , but using @xmath624 and @xmath623 instead of @xmath378 and @xmath111 . the obstacle in reproving also proposition [ propgamessuccessor
] simply relies in the definitions of the operations @xmath529 and @xmath530 : nevertheless , using the `` coding '' introduced in the definition of @xmath624 it is possible to define a set @xmath680 and two new operations @xmath681 and @xmath682 such that @xmath683 and @xmath684 for every @xmath685 , and with the further property that proposition [ propgamessuccessor ] holds whenever we replace all occurrences of @xmath111 , @xmath529 , @xmath530 and @xmath686 in its statement with @xmath623 , @xmath681 , @xmath682 and @xmath680 . from the construction of
@xmath624 we can now infer which are the minimal conditions on the @xmath398 s under which one can carry out the constructions above and define the games @xmath687 which represent @xmath108 .
[ defdelayable ] let @xmath153 be a reduction game and @xmath184 be the set of functions induced by legal strategies for @xmath4 in @xmath32 .
we say that @xmath32 ( or @xmath184 ) is _ delayable _ if for every @xmath26 the set @xmath184 is still represented by each of the new reduction games @xmath688 defined by : as for @xmath194-closure , one could note that the property of being delayable corresponds to the property of being closed under right - composition with lipschitz functions from @xmath96 into itself . from this definition and from the construction above , it turns out that we can still define reduction games representing @xmath108 , @xmath238 and @xmath119 whenever each element @xmath398 of the sequence @xmath99 contains the identity function and is represented by a parametrized class @xmath692 of _ delayable _ ( rather than @xmath194-closed ) reduction games .
this technical condition is optimal if we want to define reduction games like those presented in section [ sectionplayable ] , in which @xmath4 has to fill in a table with @xmath0-many rows : in fact , any reduction game is by definition formalizable as a game on @xmath0 , and this essentially means that in each turn @xmath4 can make at most a finite numbers of moves on a finite number of rows of her table , condition which easily leads to our definition of delayability .
however , there are still examples of natural playable sets of functions ( and even of reducibilities for sets of reals , like the set @xmath71 ) which are clearly non - delayable .
this leaves open the following question : given @xmath579 and a non - delayable playable set of functions @xmath22 , are there reduction games representing the classes of functions @xmath244 and @xmath381 ?
in particular , are there reduction games representing @xmath693 and @xmath694 ? | we present a general way of defining various _ reduction games _ on @xmath0 which `` represent '' corresponding topologically defined classes of functions . in particular , we will show how to construct games for piecewise defined functions , for functions which are pointwise limit of certain sequences of functions and for @xmath1-measurable functions .
these games turn out to be useful as a combinatorial tool for the study of general reducibilities for subsets of the baire space @xcite . |
in all geometries , the time averaged fields should be homogeneous in each horizontal ( @xmath62 ) plane . the simulated domain is divided into 100 horizontal layers at differing heights @xmath44 , at intervals of @xmath71 . throughout , to calculate layer - wise variables , we first take the spatial average of some variable of interest , say @xmath114 , in the whole layer ( @xmath62 planes ) and then average the instantaneous values , @xmath115 , arithmetically in time , @xmath116 . for physically consistent spatial averaging ,
kinematic variables are weighted by the areas of the cross - sections of the particles in the corresponding layers .
for example , the instantaneous spatially averaged velocity in layer @xmath117 is @xmath118 where @xmath119 is the particle - wise velocity at position @xmath120 and time @xmath121 and @xmath122 is the area of cross - section of @xmath123th particle in the layer @xmath117 at time @xmath121 .
the layer - wise mean velocity fluctuation @xmath25 is calculated differently as the square root of _ granular temperature _ : @xmath124 as defined above , before calculating the velocity fluctuation , we need the instantaneous mean velocities of each layer and then interpolate the particle - wise mean instantaneous velocity using the position @xmath120 of each particle between two layers . then the squared velocity deviations are averaged in both space ( shown above ) and time .
this definition is quite similar to the kinetic part of the stress tensor defined in @xcite except that we subtract the instantaneous particle - wise mean velocity instead of the time averaged one , which can eliminate the temporal velocity fluctuation of the whole layer .
for example , if a box of particles were very tightly packed and its boundaries given a rigid - body oscillation in time , grains inside would still appear locked in their cage of neighbors .
the temperature definition we use ensures this rigid motion is excluded from the temperature calculation , which is important as it should not influence the apparent rheology . to verify the averaging method described in the main text
, we compare the average stress field with the solution of the equilibrium equations in steady state : @xmath125 for @xmath126 where @xmath127 is the cauchy stress and @xmath128 is the body force ( gravity ) . choosing the @xmath129 basis shown in the main text ,
the side walls are periodic boundary conditions and we have @xmath130 . due to the symmetry , certain stress components become determined solely from equilibrium i.e. are statically determinate which gives a means to test the validity of the stress averaging method
. in homogeneous planar shear flow , the stress components @xmath131 and @xmath132 throughout should be the same as the confining wall stress components . as shown in fig .
[ fig1 : 1a ] , @xmath133 to within @xmath134 , not only showing that the averaging method is consistent with the equilibrium equations , but also meaning the feedback control method of applying top - wall pressure is capable of applying the target pressure . as shown in fig .
[ fig1 : 1a ] , the shear component @xmath132 is also constant in @xmath44 , to within @xmath134 . 0.37 and @xmath132 in the three kinds of configurations tested are plotted against the height . (
a ) the planar shear flow ( @xmath55 , @xmath135 ) ; ( b ) the planar shear flow with gravity ( @xmath136 , @xmath137 ) ; ( c ) the inclined chute flow ( @xmath138 ) .
the black solid lines show @xmath139 , the blue solid lines show @xmath132 , the orange dash dot lines show the predicted @xmath139 in ( a ) and ( b ) , and the predicted @xmath139 in ( c ) and @xmath132 in ( a ) , ( b ) , ( c ) up to a constant .
, title="fig : " ] 0.37 and @xmath132 in the three kinds of configurations tested are plotted against the height .
( a ) the planar shear flow ( @xmath55 , @xmath135 ) ; ( b ) the planar shear flow with gravity ( @xmath136 , @xmath137 ) ; ( c ) the inclined chute flow ( @xmath138 ) .
the black solid lines show @xmath139 , the blue solid lines show @xmath132 , the orange dash dot lines show the predicted @xmath139 in ( a ) and ( b ) , and the predicted @xmath139 in ( c ) and @xmath132 in ( a ) , ( b ) , ( c ) up to a constant . , title="fig : " ] 0.37 and @xmath132 in the three kinds of configurations tested are plotted against the height .
( a ) the planar shear flow ( @xmath55 , @xmath135 ) ; ( b ) the planar shear flow with gravity ( @xmath136 , @xmath137 ) ; ( c ) the inclined chute flow ( @xmath138 ) .
the black solid lines show @xmath139 , the blue solid lines show @xmath132 , the orange dash dot lines show the predicted @xmath139 in ( a ) and ( b ) , and the predicted @xmath139 in ( c ) and @xmath132 in ( a ) , ( b ) , ( c ) up to a constant . , title="fig : " ] in the cases with nonzero gravity , equilibrium gives @xmath140 and @xmath141 given @xmath142 , which can be examined in the configurations of the planar shear flow with gravity ( fig .
[ fig1 : 1b ] ) and inclined chute flow ( fig .
[ fig1 : 1c ] ) . in the chute flow case ,
since the confining pressure and shear stresses at the top and bottom are not specified beforehand , solutions can be determined analytically only up to a constant .
since @xmath29 is in fact varying with height over the domain , we integrate @xmath143 and @xmath144 with respect to @xmath44 using the averaged layer - wise @xmath145 to calculate the analytical stresses .
the slow flowing regions whose strain over the sampling time is small , show expectedly stronger fluctuations about the average ( @xmath146 in fig .
[ fig1 : 1b ] and @xmath147 in fig .
[ fig1 : 1c ] ) .
the agreement of the averaged stress components and the values given through equilibrium in fig .
[ fig1:stress test ] support the definitions of average packing fraction and stress used in this work . here
we make a comparison with granular kinetic theory , which can also lead to the form @xmath148 as mentioned in the main text , where @xmath149 is some function of @xmath29 .
regarding kinetic theory , in lun et al.s work @xcite , @xmath2 and the viscosity @xmath81 in a granular gas are functions of @xmath29 and granular temperature @xmath82 : @xmath83 where @xmath84 is the density @xmath85 and @xmath86 are functions of packing fraction @xmath29 . in haff s work
@xcite , the dependency of @xmath2 and @xmath81 on the velocity fluctuation @xmath25 and average separation of neighbouring grains @xmath150 , which is a function of @xmath29 , is : @xmath151 where @xmath152 and @xmath153 are dimensionless constants .
either of the theories above would give a form @xmath148 , though neither of them predict @xmath154 the same as the fitted @xmath72 in the main text .
related to kinetic theory , losert et al.s locally newtonian continuum model @xcite proposes @xmath155 where @xmath156 is the density at random close packing , and @xmath157 is a dimensionless number proposed to be a function of packing fraction to account for the divergence of viscosity near random close packing . from the latter formula , by using @xmath26 , we can directly get @xmath158 it is supposed in @xcite that @xmath159 , where @xmath160 is the packing fraction of random close packing .
this implies @xmath161 , which is tested in fig .
[ fig2:test other function ] . only some of the data can be fitted using such a power law . substituting the pressure with its proposed density and temperature dependent formula ,
the locally newtonian continuum model predicts : @xmath162 this would mean , algebraically , that @xmath163 should be a single function of packing fraction @xmath29
. such a deduction is tested in fig .
[ fig3 : comparison(a ) ] using the same dem results in multiple configurations .
we notice that when the packing fraction is less than 0.57 , the data points lay on a single curve , however , as the packing fraction increases , different cases have different curves .
interestingly , the normalized fluidity @xmath28 proposed in the main text and shown in fig .
[ fig3 : comparison(b ) ] collapses the same data more strongly , which suggests the kinetic pressure formula used in the former test may be less reliable in dense , non - uniform flow fields .
this point is discussed in more depth next .
an equation of state relating the pressure , volume and temperature ( pvt relationship ) is a common assumption in fluids . in the hypothesized formula for @xmath17
, @xmath29 corresponds to `` v '' and @xmath25 corresponds to `` t '' . in the following , we discuss possible relations between @xmath2 , @xmath29 and @xmath25 to see if one of the independent variables in the hypothesis @xmath166 can be replaced with @xmath2 , thereby reducing the number of physical kinematic variables introduced to one .
alternatively , it is also natural to ask if one of the independent variables tends to be ` relatively constant ' compared to the other , such that the profile of @xmath17 can be discerned largely from _ either _ @xmath25 or @xmath29 .
we consider these possibilities next . 1 .
_ is it possible @xmath167 or @xmath168 ? _ + if we are only allowed to use @xmath25 and @xmath2 as the independent variables , then there are two dimensionless numbers : @xmath28 and @xmath169 , where @xmath6 is the solid density of the grains . fig .
[ fig5 : gn - dvn ] shows that fluidity @xmath17 is not generally well described solely by @xmath25 and @xmath2 .
+ considering @xmath29 and @xmath2 as the independent variables , we can get two dimensionless numbers : @xmath170 and @xmath29 .
[ fig5:gnp - phi ] shows that fluidity @xmath17 is not generally well described solely by @xmath29 and @xmath2 .
+ these suggest a non - unique relationship between @xmath2 , @xmath29 and @xmath25 in dense inhomogeneous flows . in the absence of any dimensional constant , using @xmath2 , @xmath29 and @xmath25 ,
we can get two dimensionless numbers : @xmath171 and @xmath29 .
existence of a pvt equation of state would thus imply a one - to - one relation between @xmath171 and @xmath29 . as plotted in fig .
[ fig5 : pn - phi ] such an equation of state does not collapse out of the data from our many tests in inhomogeneous flows .
in particular , the spread is most evident for high packing fractions where a larger role for cooperativity exists , conceivably influencing the interrelationship between these variables .
it is certainly the case that for static , dense packings , the density and pressure need not related uniquely for hard particles .
is @xmath17 typically sensitive both to @xmath25 and to @xmath29 , or is one variable dominant ? _
+ as shown in the main text , the collapse of @xmath28 onto @xmath72 spans an order of magnitude .
hence , at hypothetically fixed velocity fluctuations , the fluidity would still be nontrivially influenced by @xmath29 in typical flows .
the fluidity is proportional to @xmath25 at fixed @xmath29 , so nontrivial variations in @xmath25 would cause nontrivial variations in @xmath17 . unlike equilibrium materials where internal temperature rapidly equilibrates with an external bath , all fluctuations in granular flow are generated from plastic flow itself , absent wall vibrations .
this can produce largely inhomogeneous spatial fields for @xmath25 , varying one or more orders of magnitude in typical flows @xcite .
hence , @xmath25 can not be treated as a relatively constant field for the purposes of computing granular fluidity in an arbitrary flow . + as a direct example , the depth profiles of velocity in inclined plane flow geometries having the same inclination angle but different heights of the granular layers differ from each other @xcite , whereas the packing fraction and @xmath7 profiles are quite homogeneous and almost at the same value through all cases . because @xmath172 , the observed variations in strain - rate imply @xmath29 is not adequate on its own to describe the granular fluidity in inclined flows .
the inclined flow cases show that the effect of @xmath25 is not negligible and @xmath166 is generally not able to be simplified further .
0.37 replaced with @xmath2 : ( a ) test @xmath173 ( b ) test @xmath174 .
data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).(c ) normalized pressure plotted against packing fraction .
data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).,title="fig : " ] 0.37 replaced with @xmath2 : ( a ) test @xmath173 ( b ) test @xmath174 . data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).(c ) normalized pressure plotted against packing fraction .
data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).,title="fig : " ] 0.37 replaced with @xmath2 : ( a ) test @xmath173 ( b ) test @xmath174 .
data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).(c ) normalized pressure plotted against packing fraction .
data points from all chute flow tests ( @xmath175 ) , planar shear with gravity tests ( @xmath164 ) and homogeneous planar shear flow tests ( @xmath165).,title="fig : " ] 45ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.104.078302 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | a recent granular rheology based on an implicit ` granular fluidity ' field has been shown to quantitatively predict many nonlocal phenomena . however , the physical nature of the field has not been identified . here
, the granular fluidity is found to be a kinematic variable given by the velocity fluctuation and packing fraction .
this is verified with many discrete element simulations , which show the operational fluidity definition , solutions of the fluidity model , and the proposed microscopic formula all agree .
kinetic theoretical and eyring - like explanations shed insight into the obtained form .
the rheology of dry granular materials is commonly studied in homogeneous simple ( planar ) shear tests . in these tests ,
the _ inertial granular rheology _ can be observed , in which a one - to - one relationship exists between two dimensionless numbers : @xmath0 , the ratio of shear stress @xmath1 to normal stress @xmath2 , and the inertial number @xmath3 , which nondimensionalizes the shear rate @xmath4 by the ( mean ) particle size @xmath5 , @xmath2 , and the solid density @xmath6 @xcite .
empirically , the bijection between @xmath7 and @xmath8 is often fitted to the form @xcite @xmath9 where @xmath10 is a static yield value , below which the system does not flow , @xmath11 is an upper limit for @xmath7 at high rates , @xmath12 , and @xmath13 is a dimensionless constant . despite its effectiveness in steady simple shearing , granular behavior in more general circumstances can be observed to deviate from the inertial rheology . in inclined
plane flows , where the @xmath7 field is spatially homogeneous and given by the tilt angle , the angle at which a flowing layer stops depends explicitly on the size ( thickness ) of the pile @xcite . in steady but non - uniform flow geometries , flow is observed in zones where @xmath14 and the @xmath15 relation is not one - to - one in these regions @xcite .
moreover , a `` secondary rheology '' has been observed in which the dynamics of a loaded probe submerged in quiescent material is influenced by the motion of far - away boundaries of the granular system @xcite .
such phenomena deviating from the inertial law are describable only by considering nonlocal effects .
various microscopic notions have been considered to understand the origins of this nonlocality @xcite .
recently , a size - dependent granular rheological framework has been proposed based on a state field called the `` granular fluidity . '' with minimal fitting parameters , the model has shown the capability of quantitatively predicting a range of nonlocal effects in multiple geometries , including all the deviations from @xmath16 behavior described above @xcite .
the granular fluidity field , denoted @xmath17 , is presumed to be governed by a dynamical partial differential equation ( pde ) @xcite : @xmath18 where @xmath19 is a dimensionless constant called the _ nonlocal amplitude _
, @xmath20 is a time - scale , and @xmath21 .
the @xmath17 field influences the flow through its role in the constitutive relation between stress and strain - rate : @xmath22 .
together , the result is a flow model with an intrinsic length - scale given by @xmath5 .
the inertial law , eq .
( [ local_law ] ) , can be obtained when the flow field is homogeneous ( @xmath23 ) and in steady state . while these equations define the model from a mathematical perspective , the physical nature of the granular fluidity field is not clear . to be valid in its role within the constitutive model
, granular fluidity should be a kinematically observable state variable , as was stressed in @xcite and was assumed in @xcite where eq ( [ g_pde ] ) was reconciled form a variational argument .
what _ is _ the granular fluidity ? in this letter , we identify a microphysical definition for the granular fluidity field , which defines the fluidity in terms of the velocity fluctuation and packing fraction , two kinematic variables . using discrete element method ( dem ) simulations in multiple configurations at steady - state
, we compare the predictions of this microscopic formula with the constitutive definition @xmath24 , as well as steady solutions of @xmath17 from the pde for granular fluidity .
the strong agreement found gives evidence that the pde is in fact a model for the behavior of this kinematic field .
lastly , we attempt to explain the microphysical description of @xmath17 using kinetic theory and also using an eyring - like model , which illustrates a possible fluctuation activated process of granular flows .
we start with the hypothesis that for hard particles , @xmath17 should relate to velocity fluctuations @xcite , @xmath25 , among other possible state variables . from its operational usage in the flow - rule
, @xmath26 has dimensions of inverse time .
we propose the relevant time scale is @xmath27 such that the fluidity @xmath17 can be nondimensionalized as @xmath28 .
if we suppose the only other state variable affecting @xmath17 is the packing fraction @xmath29 , the normalized fluidity should be expressible as @xmath30 where @xmath31 is an unknown function . to evaluate the hypothesis , dem simulations of three kinds of configurations
were implemented in the open - source software lammps @xcite .
the particles simulated are spheres with solid density @xmath32 , mean diameter @xmath33 , and polydispersity of 20% to prevent crystallization .
the particle interaction model @xcite contains elastic forces , damping effects , and coulomb friction using a spring - dashpot law defined by stiffness in the normal and tangential directions @xmath34 , @xmath35 , damping coefficient in the normal and tangential directions @xmath36 , @xmath37 and surface friction coefficient @xmath38 .
we adopt @xmath37 to be 0 and calculate @xmath36 with the restitution coefficient @xmath39 : @xmath40 . throughout
, @xmath41 is kept , which ensures that the deformation of the particles is small enough to be in the hard particle regime .
the simulated system is a cuboid domain ( @xmath42 , 9096 particles in total ) of particles sheared between two planar rough walls made of particles of the same properties at the top and bottom .
the bottom wall is fixed in all the cases .
if a pressure boundary condition at the top wall is needed , the top wall s height @xmath43 is controlled to set the pressure using a feedback process @xcite . to apply a fixed volume boundary condition ,
the top wall is held stationary in @xmath44 .
periodic boundary conditions are applied to the other four boundaries .
the time step is chosen as @xmath45 .
when the steady state is reached , we output data every 20,000 steps , collecting a total of @xmath46=3000 snapshots for each simulation .
we used three families of configurations : homogeneous planar shear , planar shear with gravity , and chute flows , as shown in fig .
[ fig1:geometries ] .
the confining pressure at the top boundary is @xmath47 , where @xmath48 , the gravity is @xmath49 , where @xmath50 , and the horizontal velocity of the top wall is @xmath51 , where @xmath52 , @xmath53 and @xmath54 are dimensionless factors to control the boundary conditions and the gravity . in gravity - free planar shear cases , the confining pressure is chosen as @xmath55 and ten different @xmath54 s are examined . in the cases of planar shear with gravity ,
@xmath53 is fixed at @xmath56 and five cases are simulated with four different @xmath54 s and two @xmath52 s . in chute flow cases , inclined angles of @xmath57 are simulated .
gravity is increased to @xmath58 in the slanted configurations and the top wall is fixed in the @xmath59 direction , @xmath60 .
we test fixed volume top - wall constraints for all chute cases .
we also perform fixed top - wall pressure constraints in the @xmath61 chute case , where , to make a direct comparison , the confining pressure is chosen to be the same as the mean pressure at the top wall in the corresponding fixed volume case . in total , twenty different flows are simulated .
0.135 0.135 0.135 in all geometries , the time averaged fields should be homogeneous in each horizontal ( @xmath62 ) plane . dividing the simulated domain into layers at differing heights @xmath44 , we first take the spatial average of the variable of interest in the whole layer ( @xmath62 planes ) , and then average the instantaneous layer - wise values arithmetically in time .
when the layer - wise instantaneous velocity has been calculated , we subtract the instantaneous mean velocity from the particle velocity to obtain the velocity deviation .
the velocity fluctuation @xmath25 is defined as the root of _ granular temperature _ which is the mean square of velocity deviation .
the particle - wise stress tensor is calculated considering both the contact contribution and the kinetic contribution , as defined in @xcite .
after the average cauchy stress tensor has been obtained in a layer , the pressure is defined as @xmath63 and the shear stress is defined as the equivalent shear stress @xmath64 , where @xmath65 is the stress deviator .
see supplemental material can be reduced[note1 ] ] for the detailed averaging method , its verification , and particular details of the calculation of @xmath25 .
to be free of wall effects , we have excluded layers at distances smaller than @xmath66 from the walls . to ensure that the variables are at steady - state , we accept data from layers that have experienced a strain no smaller than @xmath67 in sampling .
the span of inertial numbers in the data presented is 0.0007 to 0.6 , which covers the range of flow regimes , from quasi - static to collisional . to determine
if @xmath68 , we calculate the layer - wise fluidity field @xmath26 , @xmath25 , and @xmath29 in all cases . for added precision , here
we evaluate @xmath5 as the layer - wise mean particle size ; though quasi - monodisperse , the packing shows slight ( @xmath69 ) spatial variation in @xmath5 .
we plot normalized fluidity @xmath28 against packing fraction @xmath29 in fig .
[ fig3:verification ] for all 20 flows simulated .
each data point corresponds to a different @xmath44 value , except gravity - free planar shear cases , which are nearly uniform and show a single data point per test .
@xmath29 is smoothed in a band as wide as @xmath70 at @xmath44 intervals of @xmath71 .
the data from different configurations collapse well , suggesting that packing fraction and velocity fluctuation are sufficient to define the granular fluidity .
@xmath72 has a nearly constant behavior for low @xmath29 values , which transitions into a roughly linear decrease at high values .
this behavior can be fit to a hyperbola @xmath73 which vanishes at @xmath74 , approximately random close packing .
an interpretation for the functional form is given later .
for further examination , in fig .
[ fig4 : predictions ] the predicted fluidity fields using the microscopic formula are compared with @xmath75 and solutions of eq .
( [ g_pde ] ) in all configurations tested . in the pde , @xmath76 , @xmath77 and @xmath78
are obtained by fitting the @xmath16 relation in the homogeneous planar shear cases as shown in fig . [
fig4 : predictions(a ) ] , and we choose @xmath79 .
all these values are close to the values found for glass spherical beads @xcite , with the exception of @xmath11 which we find to be larger in our discrete simulations . is irrelevant here , as we study only steady - state . ]
the boundary condition for @xmath17 is taken as the wall value of @xmath75 from dem and the pde is run to steady - state using a finite - difference method to obtain steady @xmath17 profiles .
note , the fluidity profiles of @xmath80 chute flow cases with fixed volume boundary conditions and fixed pressure boundary conditions , compared in fig .
[ fig4 : predictions(c ) ] , match well , showing that the type of boundary condition does not have a significant influence on the constitutive behavior in the interior .
this is also the case if the incline angle is varied , which we verified in additional tests .
overall , the collapse of the three different definitions of @xmath17 in the various cases is strong , evidencing the generality of the @xmath68 formula .
0.245 0.245 0.245 0.245 + 0.245 0.245 0.245 0.245 there are at least two ways to reconcile these results physically , one with kinetic theory and another in terms of an activated process .
regarding kinetic theory , in lun et al.s work @xcite , @xmath2 and the viscosity @xmath81 in a granular gas are functions of @xmath29 and granular temperature @xmath82 : @xmath83 where @xmath84 is the density @xmath85 and @xmath86 are functions of packing fraction @xmath29 . under these relations , the operational definition of fluidity
would imply @xmath87 whose form is the same as eq .
( [ g_form ] ) .
though the fitted @xmath72 is different from @xmath88 given in @xcite , the similarity of the form is suggestive .
also , since the nonlocal effects captured by the fluidity model are most evident in quasi - static regions with enduring contacts , it seems the extended kinetic theory @xcite would be needed to further this connection , in which dense behavior beyond binary collisions is modeled .
kinetic theory based justifications have previously been used to explain the apparent density- and temperature- dependent viscosity of dense granular flows in annular shear experiments @xcite , which foreshadows the functional form shown above . see supplemental materials for more analysis of the connection with these theories .
eyring s equation @xcite for activated processes has an analogy in the flow behavior of amorphous solids @xcite and has been considered previously in the context of granular flows @xcite . here , as in @xcite
, we can view a material volume element as a collection of microscopic ` sites ' .
each site able to undergo a shear event or ` hop ' has a barrier that must be overcome in order to do so .
we express the element s total shear rate through the product : @xmath89 we assume @xcite that the fraction of potential sites for a shear event is a function of packing fraction , @xmath90 .
it should approach @xmath91 at a jammed @xmath29 value , @xmath92 .
the second term can be expressed as the product of an attempt frequency for perturbations , @xmath93 , and the net probability , @xmath94 , that a perturbation causes a positive shear event vs a backward ( negative ) event on a site .
the frequency of attempts should be related to the fluctuational motion of the particles , so we assume that @xmath93 is given by @xmath95 , i.e. the ratio of fluctuation velocity to a characteristic length @xmath96 .
the exponent of the probability of reaction is typically given by an arrhenius form , which depends on an energy barrier and a distribution of energies in an attempt , which depends on temperature .
the limit of rigid particle interactions in granular systems complicate energetic considerations in this context , but one can instead express @xmath97 in terms of a critical shear stress barrier @xmath98 for forward or reverse shear events on a site , and a probability distribution for the stress fluctuation delivered by a perturbation , as in @xcite .
we similarly assume that @xmath99 , where the mean applied shear stress on the element , @xmath1 , is seen to bias the barrier , and that the cumulative distribution function for stress perturbations is exponentially decaying , as observed in @xcite , with mean on the order of @xmath2 .
therefore , @xmath100 where @xmath101 is a constant factor .
lastly , we make the common assumption @xcite that the strain per flow event is a constant , @xmath102 , related to a typical ` jump distance ' .
we can now multiply the three factors together and approximate @xmath103 to get @xmath104 because @xmath105 , the error of the linear approximation is always less than @xmath106 , and much smaller in quasi - static media . upon dividing by @xmath7 one obtains @xmath107 since @xmath11 , @xmath101 , and @xmath102 are constant , eq .
( [ eyring_explain ] ) has the same form as eq .
( [ g_form ] ) .
this analysis suggests shear flow in granular media may be a fluctuation activated process . supposing @xmath108 is the dominant contribution to @xmath31 , fig .
[ fig3:verification ] suggests a natural interpretation . when @xmath109 , the packing is open enough that all sites are able to flow , and @xmath108 holds at its maximum value in this range . as packing fraction increases above @xmath110 , the number of sites that have enough free volume for flow shows an expected monotonic decrease .
the cohen - turnbull theory of free - volume distribution in glassy materials gives a similar behavior for the fraction of sites above a critical free - volume threshold @xcite .
since eq .
( [ g_pde ] ) evolves @xmath17 as a single state variable , it is natural to ask if the observed relation @xmath111 can be reduced to depend on only one state variable rather than two .
for example , through the assumption of a local equation of state , one might suppose the pressure could be used to eliminate @xmath29 or @xmath25 from the system .
in view of the data from the many geometries tested , we find this is not the case ( see supplemental material ) , suggesting the equation of state in inhomogeneous dense flow may need to include dependence on gradients and/or nonlocal influences . because @xmath29 and @xmath25 both play an irreducible and influential role in defining @xmath17 , a microphysical derivation for the granular fluidity pde , eq .
( [ g_pde ] ) , may involve combining dominant terms from a heat equation and a density evolution rule , though first attempts using kinetic theory forms produce a system more complex than eq .
( [ g_pde ] ) . deriving eq .
( [ g_pde ] ) from the microscopic description of @xmath17 remains crucial future work .
herein , we have proposed a relation connecting granular fluidity to granular velocity fluctuation and packing fraction .
this relation is demonstrated in dem simulations of multiple configurations .
all three descriptions of the granular fluidity field ( i ) its operational definition ( @xmath26 ) extracted from dem simulations , ( ii ) its definition from the fluidity governing pde , and ( iii ) the new microphysical definition match well with each other in multiple geometries under multiple conditions .
granular fluidity could be related to a fluctuation activated process as a measure of the rate of the number of flowable microsites perturbed by attempts .
it is also interesting to compare the description herein to previous models for non - granular materials ( e.g. emulsions , suspensions ) , which evolve the ` standard ' fluidity @xmath112 ( inverse viscosity ) rather than @xmath17 .
theories have previously suggested @xmath113 relates microscopically to the rate of plastic events @xcite , which has been correlated to shear - rate fluctuations in experiments @xcite .
our result that a fluctuation variable is key as well in describing granular fluidity suggests a bridge between the different amorphous material classes . |
magnetic fields in diffuse media are extremely important for many astrophysical phenomena ( see @xcite ) .
magnetic fields are important for star formation , propagation of cosmic rays , transport of heat etc .
the observational study of magnetic fields , however , is limited by the ability of the available techniques to trace magnetic fields . for instance , in spite of the both progress of observational techniques and instrumentation for measuring emission ( see @xcite ) and absorption polarization @xcite and the recent progress in understanding grain alignment ( see for recent reviews ) the reliable high resolution tracing of magnetic fields is not readily available and requires a lot of resources ( e.g. planck example ) .
in this situation is is very much advantageous to have alternative and synergetic ways of probing magnetic fields .
astrophysical magnetic fields are turbulent and observations testify that turbulence is really ubiquitous in astrophysics @xcite .
magnetic field makes turbulence anisotropic with turbulent eddies elongated along magnetic field ( see @xcite , for a review ) . as a result ,
observed velocity correlations are expected to be elongated along magnetic field and this property was demonstrated with synthetic observations ( @xcite , henceforth el05 ) .
later the study of anisotropies was performed using the principal component analysis ( pca ) and applied to observations @xcite .
there is another way to employ properties of mhd turbulence in order to study magnetic fields .
the aforementioned turbulent eddies aligned with magnetic field entail that the velocity gradients should have larger value when calculated perpendicular to the magnetic field .
this property of magnetic turbulence was employed by @xcite ( @xcite , henceforth gl16 ) who proposed tracing magnetic fields using velocity centroids gradients ( vcgs ) .
the technique was elaborated by @xcite ( @xcite , henceforth yl17 ) and the new way of magnetic field tracing was successfully compared with observations of planck polarization .
these papers introduced a new way of magnetic field studies with spectroscopic data .
velocity centroids present one way of representing velocities .
however , centroids have the contribution not only of velocity , but also of density ( see el05 ) and the density fluctuations are not as good aligned with magnetic field as velocity ( see @xcite ) . at the same time , the analytical study in @xcite revealed that channels maps are sensitive only to velocity fluctuations if the corresponding turbulent density fluctuations are dominated by large scale contributions .
this motivates our present study to explore the ability of velocity channel gradients ( vchgs ) of magnetic field tracing . in what follows
, we discuss in [ sec : theory ] the theoretical motivation of this work .
[ sec : numerics ] discusses the numerical methods on simulations and ways of analysis .
we explore the gradients in the channel maps from turbulent velocities in [ sec : gradients - in - fsa ] . in [
sec : fluctuations ] we compare the performance of density fluctuations and velocities from channel maps .
we examine the performance of channel map gradients and correlation anisotropies in [ sec : anisotropy ] .
we explore the reduced centroid gradient in [ sec : reducedcentroid ] .
we testify our method in observation in [ sec : obs ] .
we discuss our result in [ sec : discussion ] , and conclude in [ sec : conclusion ]
mhd turbulence theory is the an old subject that has been boosted recently by the ability of performing high resolution 3d numerical simulations . before that there was no way of testing theoretical constructions and many competing theories describing mhd turbulence could co - exist .
for instance , the original studies of alfvenic turbulence by @xcite and @xcite were based a hypothetical model of isotropic mhd turbulence .
some of the later studies ( see @xcite ) pointed out to the anisotropic nature of the mhd cascade .
however , the modern theory of mhd turbulence originates from the prophetic work by @xcite , henceforth gs95 ) .
given originally rather lukewarm acceptance by the mhd turbulence pandits , this theory nevertheless was supported by further theoretical and numerical studies ( @xcite , henceforth lv99 , @xcite , see @xcite for a a review ) that extended the theory and provided its rigorous testing .
our present study is based on the modern understanding of the mhd turbulence cascade and the statistical properties of turbulence confirmed numerically .
the original gs95 theory is the incompressible mhd theory , i.e. the theory of alfvenic turbulence and the turbulence of pseudo - alfven waves .
the latter represent fundamental slow waves in the incompressible limit .
the third fundamental component of mhd perturbations , namely fast waves , do not exist in the incompressible limit where the sound velocity is infinitely high .
the incompressible mhd turbulence was successfully tested in @xcite and @xcite .
the numerical studies of the applicability of gs95 ideas and its generalizations for the compressible case ( see @xcite ) were first performed in @xcite .
they testified that the gs95 incompressible scaling is applicable to the alfven and slow modes , while revealed that fast modes provide isotropic spectrum .
in fact , these studies revealed that the fundamental modes can be considered separately , as for non - relativistic mhd turbulence the coupling between different types of fundamental modes is a subdominant effect therefore , in what follows we can consider three distinct cascades , namely , the cascade of alfven , slow and fast modes .
numerical simulations show that the alfven and slow modes are dominant component of the cascade and those correspond to anisotropic perturbations aligned with the magnetic field .
this provides the theoretical justification for relating gradients and magnetic fields in the observational studies . in the subalfvenic regime ,
i.e. for the injection velocity @xmath0 being less than the alfven velocity @xmath1 , the alfven modes initially evolve by increasing the perpendicular wavenumber while keeping the parallel wavenumber stays the same ( see lv99 , @xcite ) .
the increase of the perpendicular wave number makes the the alfvenic wavevectors more and more perpendicular to magnetic field .
therefore the gradients of magnetic field and velocity get aligned perpendicular to the magnetic - field direction .
the increase of perpendicular wavenumbers while the parallel wavenumber stays the same , increases the energy cascading .
when the fraction of cascaded energy becomes @xmath2 the gs95 cascade takes over .
this happens at the scale @xmath3 , where @xmath4 is the turbulence injection scale and @xmath5 is the alfven mach number ( see lv99 , @xcite ) . in the gs95 cascade
both parallel and perpendicular wavenumbers of alfvenic perturbations increase . to quantify
this one should adopt the system of reference aligned with the local magnetic field . in this system of reference
the so - called critical balance condition should be satisfied , which states that the time of the interaction of the oppositely moving wavepackets / eddies @xmath6 , where @xmath7 is the parallel to magnetic field scale of the wavepacket / eddy is equal to the perpendicular shearing time of the wavepacket / eddy @xmath8 , where @xmath9 is the perpendicular scale of the wavepacket / eddy and @xmath10 is the turbulent velocity associated with this scale .
this is how the cascade proceeds in the strong regime with the wavepackets getting more and more elongated according to ( lv99 ) : @xmath11 which testifies that for @xmath12 much smaller than the injection scale @xmath4 the perturbations are strongly elongated and aligned with the local magnetic field .
therefore the gradients of the velocities are expected to be aligned with the _ local _ magnetic field and by measuring them one can trace the local variations of the magnetic field within a turbulent volume . within the gs95 picture and its generalization to the compressible media
( gs95 , @xcite ) the slow modes are slaved by alfven modes . indeed ,
alfven modes shear and cascade slow mode perturbations , which is the process that is confirmed by numerical simulations .
thus , in agreement with numerical studies , the the anisotropic scaling given by eq .
( [ lpar ] ) is valid the slow mode eddies . as a result
both the gradients of alfven modes and of slow modes are expected to be perpendicular to the local direction of magnetic field .
this is the key idea behind tracing magnetic - field directions with velocity gradients .
while velocity gradients are directly not available from astrophysical observations of diffuse media , other measures can be constructed using observational data .
velocity channel maps can be constructed using spectroscopic observations of doppler shifted lines .
the statistics of these maps has been described in @xcite ( @xcite , henceforth lp00 ) for the optically thin data and in @xcite for the observations in the presence of absorption . in what follows we concentrate on the optically thin case and only mention some the possible effects of optically thick data.we note that when we discuss thin and thick slices of data , we mean not the effects of absorption but the thickness of channel maps .
the minimal thickness of the latter is determined by the spectral resolution @xmath13 of the instrument , but it can be increased by integrating the spectroscopic data over larger @xmath13 . an important prediction in lp00 is that velocity caustics create fluctuations of intensity of the channel maps and the relative importance of the velocity and density fluctuations changes with the thickness of the channel maps .
in particular , lp00 identified a regime of `` thin velocity slices '' and found that in this regime the intensity fluctuations in the slice are dominated by the velocity fluctuations provided that the density fluctuations have the 3d spectrum that is steep , i.e. most of the fluctuating energy is concentrated at the large scales .
the aforementioned statement about the steep density spectrum can be expressed in terms of the 3d power spectrum @xmath14 . in terms of this spectrum
the kolmogorov cascade corresponds to @xmath15 and it is steep .
the borderline spectrum is @xmath16 , with turbulence having spectrum @xmath17 , @xmath18 containing more energy at the small scale and therefore being shallow . for subsonic flows
the density spectra in mhd turbulence are steep ( see @xcite ) .
thus for such flows the intensity fluctuations in thin velocity slices of position - position - velocity ( ppv ) data are influenced only by the turbulent velocity statistics .
naturally , the thermal broadening interferes with the minimal slice of the ppv for which fluctuations can be studied .
this means that for studying thin slices of subsonic flows one should use heavier species that those of the main hydrogen astrophysical flow .
the heavier species which are referred usually in the interstellar literature as `` metals '' can be used .
if the density spectrum is shallow , i.e. @xmath18 , the contribution of density and velocities to the statistics of intensity fluctuations of the velocity slices was evaluated in lp00 . for thin slices ,
both velocity and density are important , while the contribution of velocity decreases as the slice thickness increases .
when the integration is performed over the entire line , only density fluctuations determine the fluctuations of the resulting intensity distribution .
the anisotropy of mhd turbulence results in the anisotropy of the intensity distribution within the ppv slices .
this was first demonstrated with the observational data in @xcite where a new technique of studying magnetic field direction was suggested . later a systematic numerical study of the change of the anisotropy of intensity fluctuations in ppv slices was performed in @xcite .
an analytical study of the effect of the alfven , slow and fast modes on the anisotropy of intensity fluctuations in the ppv slices was performed in @xcite ( @xcite , henceforth klp16 ) .
this study opens an avenue for separating the contributions of the different modes using the anisotropies of the measured correlation functions of the slice intensities . in view of the above ,
it is important to perform the study of the gradients of intensities within ppv slices . in terms of the separating the compressible and incompressible components in klp16 ,
such a study provides an additional test that the anisotropies are actually caused by turbulence . by varying the thickness of the slice one can study the variation of gradients as the relative contribution of density and velocity changes
this may be important as the density and velocity , in general , have different statistics and the fluctuations of these fields are aligned with magnetic field to a different degree .
for instance , the strongly supersonic flows demonstrate isotropic spectra of density . in addition
, for studies of galactic flows the regular shear opens a way to study different turbulent regions separately and therefore different channel maps can be associated with different locations within the galactic volume . in what follows the advancements of the understanding of the theory of ppv anisotropies in klp16
guide us in studying gradients in the ppv velocity slices .
note that the criterion for the velocity slice of being thin or thick as it given in lp00 is as follows : the velocity slice is thin if the square root of turbulent velocity dispersion on the scales that the slice is being studied @xmath19 is _ greater _ than the thickness of the slice @xmath13 , i.e. @xmath20 where @xmath21 is the separation of the correlating points over plane of the sky , i.e. the pp separation . in the following , we would characterize the channel thickness by whether the ratio @xmath22 passes through unity .
the numerical data was obtained by numerical simulations using a single fluid , operator - split , staggered grid mhd eulerian code zeus - mp/3d @xcite to set up a three - dimensional , uniform , isothermal turbulent medium .
periodic boundary conditions are applied to emulate a part of interstellar cloud .
solenoidal turbulence injection was employed .
our simulations consists of various alfvenic mach number @xmath5 and sonic mach number @xmath23 , where @xmath0 is the injection velocity , while @xmath24 and @xmath25 are the alfven and sonic velocities respectively , which they are listed in[tab : simulationparameters ] .
the domain @xmath26 corresponds to the simulations of plasmas with magnetic pressure larger than the thermal pressure , i.e. plasmas with low @xmath27 , while the domain @xmath28 corresponds to the pressure dominated plasmas with @xmath29 . to investigate the detail structure of gradients from different @xmath30 and @xmath31
, we employ the wave mode decomposition method in @xcite to extract alfven , slow and fast modes from velocity data .
the corresponding equations determining the basis for the decomposition into modes are : [ eq : fsa - decompositions ] @xmath32 where @xmath33 , @xmath34 , and @xmath35 .
we would only use the los component of the decomposed velocities for velocity channels calculations.that is to say , the three velocity modes can then be acquired by @xmath36(\hat{\zeta}_{f , s , a } \cdot \hat{\zeta}_{los})\ ] ] where @xmath37 is the fourier transform operator .
the upper panel of figure [ fig : fsa ] illustrates the decomposition procedure that takes place in the fourier space .
the resulting 3 data cubes are dominated by the alfven , slow and fast modes , respectively . in the middle and lower panels of figure [ fig : fsa ]
we show the results of decomposed velocity cube which being projected along the x - axis .
we illustrate the decomposition method using two cubes with low and high @xmath38 .
the properties of the fast and slow modes differ in low and high @xmath38 plasmas , therefore we study these two cases separately in the following sections . [
tab : simulationparameters ] c c c c c model & @xmath30 & @xmath31 & @xmath39 & resolution + a1 & 0.2 & 0.2 & 0.02 & @xmath40 + a2 & 0.2 & 0.2 & 0.2178 & @xmath40 + a3 & 0.2 & 0.2 & 2 & @xmath40 + a4 & 0.2 & 0.2 & 21.78 & @xmath40 + a5 & 0.2 & 0.2 & 200 & @xmath40 + b11 & 0.4 & 0.04 & 0.02 & @xmath40 + b12 & 0.8 & 0.08 & 0.02 & @xmath40 + b13 & 1.6 & 0.16 & 0.02 & @xmath40 + b14 & 3.2 & 0.32 & 0.02 & @xmath40 + b15 & 6.4 & 0.64 & 0.02 & @xmath40 + b21 & 0.4 & 0.132 & 0.2178 & @xmath40 + b22 & 0.8 & 0.264 & 0.2178 & @xmath40 + b23 & 1.6 & 0.528 & 0.2178 & @xmath40 + b31 & 0.4 & 0.4 & 2 & @xmath40 + b32 & 0.8 & 0.8 & 2 & @xmath40 + b41 & 0.132 & 0.4 & 18.3654 & @xmath40 + b42 & 0.264 & 0.8 & 18.3654 & @xmath40 + b51 & 0.04 & 0.4 & 200 & @xmath40 + b52 & 0.08 & 0.8 & 200 & @xmath40 + gradients were calculated following the procedures described in yl17 . for determining whether the gradients are probing thin or thick channel map we use the following criterion : the channel is thin if the gradient is calculated over the patch of size @xmath21 for which the criterion given by eq . ( [ criterion ] )
is satisfied ; otherwise , the channel is thick . according to the thickness conditions
, we construct the _ column density channel map _ about some velocity slice @xmath41 by : @xmath42 the respective _ velocity channel map _ is @xmath43 to extract the statistics from gradients map , we use the sub - block averaging method introduced in yl17
. we also use a way to produce channel maps that contain only spatial frequencies for which the slice is thin . for this purpose we provide filtering of the high spatial frequencies for which the criterion given by eq .
( [ criterion ] ) is not satisfied . as the velocity dispersion described in eq .
( [ criterion ] ) depends on the block size .
we use the lowest velocity dispersion value in all blocks for the production of channel maps .
as the velocities , unlike densities , are directly related to mhd turbulence , we first consider only intensity fluctuations that arise from only turbulent velocity . for this purpose
we create data cubes using the velocity field obtained from our 3d numerical simulations but substitute the actual densities by a constant density .
figure [ fig : illus-1 ] illustrates the relative orientation of ppv velocity slice intensity gradients or velocity channel gradients ( vchgs ) and the projected magnetic field on a _ thin _ slice setting .
the gradients are all rotated for 90 degrees , which will be annotated as _ rotated _ gradients .
the rotated gradients according to the theoretical considerations above correspond to the magnetic field directions . in the following sections ,
gradients are assumed to be rotated , unless emphasized specifically .
illustrations on the the intensity ( left , green ) and velocity ( right , red ) channel _ rotated _ gradients from a thin slice map with respect to the projected magnetic field ( blue ) .
notice that the velocity field have not been decomposed by the cl03 method yet.,title="fig:",scaledwidth=22.0% ] illustrations on the the intensity ( left , green ) and velocity ( right , red ) channel _ rotated _ gradients from a thin slice map with respect to the projected magnetic field ( blue ) .
notice that the velocity field have not been decomposed by the cl03 method yet.,title="fig:",scaledwidth=22.0% ] the fact that the vchgs for thin maps are strongly influenced by the velocity field shows the promise of the approach for tracing magnetic fields using thin channel maps .
below we consider the gradients arising from alfven , slow and fast modes separately .
the theoretical considerations giving the guidance for this study are provided in klp16 , where the anisotropies of the correlation of the intensities arising from velocities were studied .
the dynamics of alfvenic modes in the strong mhd turbulence are very different from the dynamics of freely propagating alfven waves .
the modes cascade on the scale of the order of one period and the wavevector of the alfvenic perturbations in strong turbulence is nearly perpendicular to the local direction of the magnetic field .
the anisotropy of the intensity correlation of the iso - contours of correlation is elongated _ parallel _ to magnetic field .
the latter is essential for tracing of magnetic fields , as the gradients that these modes are inducing are also perpendicular to the local direction of magnetic field , which is employed in gl16 , yl17 and consequent studies from them . for alfven modes
the theoretical expectation is very natural : the anisotropy of the intensity correlations in a thin slice increases with the decrease of the alfven mach number , i.e. the smaller the velocity perturbations , the more anisotropic is the velocity distribution .
figure [ fig : alf-1 ] illustrates the velocity channels in low and high @xmath38 and their gradients within a thin slice that are induced by alfvenic modes .
the right panel of figure [ fig : alf-2 ] shows the change of the @xmath44 with respect to @xmath31 .
according to @xcite the change of alfven modes with @xmath30 is marginal .
this is shown in the left panel of [ fig : alf-2 ] .
the ( constant density ) velocity channel map for low ( left ) and high ( right ) @xmath38from alfven mode in a thin slice map .
their respective gradients ( red for vchgs , green for igs ) are overplotted by projected magnetic field ( blue).,title="fig:",scaledwidth=22.0% ] the ( constant density ) velocity channel map for low ( left ) and high ( right ) @xmath38from alfven mode in a thin slice map .
their respective gradients ( red for vchgs , green for igs ) are overplotted by projected magnetic field ( blue).,title="fig:",scaledwidth=22.0% ] as we increase the thickness of channel maps , the @xmath44 of alfven modes is decreasing , which is shown in the blue curves in [ fig : alf-3 ] .
at the same time the contrast of the gradients given by @xmath45 is decreasing . due to this decrease
the role of velocity fluctuations in the channel map fluctuations is decreasing and the thick channels get dominated by density effects .
this is illustrated in figure [ fig : alf-4 ] . a more detailed discussion on the contribution of density effect will be in 5 .
one should remember , however , that whether the channels are thin or thick depends on the size of the eddies that are being studied ( lp00 ) .
therefore , if we calculate the gradients for the large blocks , we are sampling large eddies and for such large eddies the slices can be still thin .
variation of @xmath46 with respect to channel map thickness.,scaledwidth=48.0% ] slow waves present perturbations of magnetic field and density that propagate along magnetic field lines . in the limit of incompressible media , slow waves are pure magnetic compressions that propagate along magnetic field lines .
formally the incompressible case corresponds to @xmath47 and in this limit the slow modes are frequently called preudo - alfven modes . for @xmath48 the slow waves are density perturbations propagating along magnetic field lines . in the presence of alfvenic turbulence , slow modes do not evolve on their own , but are sheared by alfven modes . as a result , the features of alfvenic turbulence , e.g. spectrum and anisotropies , are imprinted on the slow modes ( see gs95 , @xcite ) .
that also means the perpendicular velocity gradients in alfven modes will also be inherited by slow modes .
figure [ fig : slow-1 ] illustrates the channel maps for a thin slice for the case of low @xmath38 ( left panel ) and high @xmath38 ( right panel ) .
it is obvious that low @xmath38 vchgs performs better , which is also shown in figure [ fig : alf-3 ] .
in fact , when @xmath38 is small , its respective channel maps behave more likely to that of alfven mode channel maps . on the other hand ,
channel maps from high @xmath38 systems are not as highly structured as the maps with low @xmath38 as @xmath49 and @xmath50 are complementary measures of the physical conditions in mhd turbulence , we can study the change of @xmath38 by varying @xmath30 and @xmath31 .
figure [ fig : slow-2 ] shows the change of the @xmath44 of the vchgs in the thin channels induced by slow modes for different @xmath38 by changing @xmath31 .
indeed , while according to klp16 the anisotropies of correlation functions of the intensities in thin slice for slow modes are different for the case of low-@xmath38 and high-@xmath38 plasmas , the properties of turbulence are also changing with @xmath30 .
indeed , for @xmath51 we expect the formation of shocks in the media .
the _ thin _ slice ( constant density ) velocity channel maps for low @xmath38 ( left ) and high @xmath38 ( right ) for slow modes , title="fig:",scaledwidth=23.0% ] the _ thin _ slice ( constant density ) velocity channel maps for low @xmath38 ( left ) and high @xmath38 ( right ) for slow modes , title="fig:",scaledwidth=23.0% ] the change of @xmath44 of slow modes vchgs with respect to @xmath38 by varying @xmath31,scaledwidth=48.0% ] on the change of channel widths to @xmath44 , the green curves from figure [ fig : alf-3 ] shows the superiority of alfven modes when compared to slow modes .
moreover , the change of @xmath44 from slow modes due to a change of channel width is much more significant than that of alfven modes . while these modes are not direct observables , as both modes contributes to velocity gradient alignments .
the result of figure [ fig : alf-3 ] suggests velocity centroid maps without proper selection of channel widths may only have fair performance due to pollution of @xmath44 from slow modes .
this in turn suggesting a reduction of channel widths can enhance the performance on tracing magnetic fields using the gradient technique .
similar to the properties of the slow modes , the properties of fast modes are different in low and high @xmath38 plasmas . for high @xmath38 plasmas
the fast wave are similar to sound waves that propagate with sound speed irrespectively of the magnetic field direction .
similar to acoustic turbulence , the corresponding turbulence is expected to be isotropic . in low @xmath38
the fast modes correspond to magnetic field compressions that propagate with the alfven velocity . in terms of correlation function anisotropy of fast modes ,
the elongation of the iso - contours of correlation is elongated _ perpendicular _ to magnetic field .
the alignment of undecomposed velocity gradients will contain contributions from the three modes .
therefore for fast mode dominated environments , one should expect the gradients to be parallel instead of perpendicular to local magnetic field .
the _ thin _ slice ( constant - density ) velocity channel maps with _ unrotated _ gradients for low @xmath38 ( left ) and right @xmath38 ( right ) for fast modes .
the red vectors are the _ unrotated _ vchgs , and the blue vectors are projected magnetic field direction.,title="fig:",scaledwidth=23.0% ] the _ thin _ slice ( constant - density ) velocity channel maps with _ unrotated _ gradients for low @xmath38 ( left ) and right
@xmath38 ( right ) for fast modes .
the red vectors are the _ unrotated _ vchgs , and the blue vectors are projected magnetic field direction.,title="fig:",scaledwidth=23.0% ] figure [ fig : fast-1 ] shows the channel maps with low and high @xmath38 respectively .
the low @xmath38 map carries more structures than that of high @xmath38 map , but they are obviously carrying different anisotropy direction when examining the contours in the maps . in the language of fast mode gradients ,
the maximal gradient is _ parallel _ to local magnetic field direction .
figure [ fig : fast-2 ] illustrates the @xmath44 of fast modes vchgs which the gradient vectors are not rotated .
in contrast to the decreasing @xmath44 for rotated alfven and slow mode vchgs , the @xmath44 of unrotated fast mode vchgs increases significantly when the channel width passes through unity for both low and high @xmath38 case .
this suggests a decrease of channel width can suppress the contribution of fast modes in velocity gradient calculations .
numerical simulations ( e.g. @xcite ) indicate that fast modes are subdominant at least for the cases of incompressible driving of turbulence .
our study indicates that in terms of vchgs one can expect a further suppression of the fast mode contribution in thin channels .
this important as the directions of gradients induced by fast modes are orthogonal to those by fast and slow modes and , therefore , potentially their interference could make magnetic field tracing less reliable . at the same time , this peculiar properties of gradients arising from fast modes can be used to separate their contribution . note , that fast modes are identified in @xcite as the dominant source of cosmic ray scattering within galactic environments .
to study the relative importance of density and velocity fluctuations , we create synthetic maps by transposing the velocity data cube on the plane - of - sky plane , if the original velocity cube is @xmath52 , and the line - of - sight along the x - direction , then the transposed cube is @xmath53 .
computationally , we would calculate the correlation function anisotropy of a quantity called _ transposed centroid _
@xmath54 : @xmath55 the reason behind this is that both densities and velocities are anisotropic and trace the local magnetic field direction .
after the transpose procedure , the velocity anisotropies are orthogonal to that from density data and the relative influence of the two contributions to the centroid anisotropy can be evaluated .
when examining the anisotropy using the correlation function @xmath56 in the transposed centroid map , the ratio between the axis perpendicular ( the elongation direction of @xmath54 ) to magnetic field direction over that of parallel one ( the elongation direction of @xmath57 ) will reflect the relative importance of velocity to density fluctuations .
a scatter plot showing the change of axis ratio relative to velocity slice thickness.,scaledwidth=48.0% ] figure [ fig : fluc-1 ] shows the decreasing trend of the aforementioned axis ratio as velocity slice thickness increases .
it is evident that even though density information can contaminate the contribution of alignment to magnetic field from velocity information , the overall anisotropy comes from the velocity centroid , which essentially a weighted product of density and velocity data , is still carrying the anisotropic nature along local magnetic field directions .
nevertheless , using a thin slice can reduce the contribution from density data , which in turn provide a more reliable detection of local magnetic field .
it was demonstrated in @xcite that the anisotropies in channel maps can trace magnetic fields . while these anisotropies can be very informative in terms of determining the relative contribution of compressible versus incompressible modes ( see kandel et al .
2016 , 2017 ) , the tracing of the detailed magnetic field structure is problematic with them . in figure
[ fig : cfa-1 ] we show the vchgs alignment measure @xmath44 versus the alignment measure of the directions of the maximal anisotropy ( elongation ) of the correlation functions of the channel map intensities .
we clearly see that the channel map intensities are not good for detailed magnetic field tracing , which is consistent to our previous studies on the comparisons of alignments from correlation function anisotropies and different types of gradients ( see yl17 , @xcite , yl17b ) . a comparison of @xmath44 on a thin channel map when applying the gradient technique and correlation function anisotropy techniques to alfven mode vchgs .
, scaledwidth=48.0% ] the reason of this is easy to understand .
the velocity gradients correspond to the angular velocities of turbulent eddies .
these individual eddies are aligned with the local magnetic field . on the contrary ,
the correlation functions are the measures that are well defined only after the ensemble averaging .
thus the averaging over small patches does not produce good statistics necessary for the correlation functions to be well defined .
our study above suggests that the decrease of channel width improves the alignment the gradient and magnetic field alignment .
however , it is important to understand whether one can use only a part of the spectral line for the analysis . given the channel map data one can form _ reduced centroids _ that contain only part of the line : @xmath58 where the index @xmath59 determines the order of the centroid .
increasing @xmath59 may enhance the effects of velocity , but it also can increase the noise .
these measures are useful both in the case of studying gradients from extended galactic disk data and also for the data from the wings of the absorption lines . with the dispersion of velocities
@xmath60 from equation [ criterion ] and the average width of all the doppler shifted lines @xmath61 , we can classify the line into three regions , namely the central part , the middle part and the wing .
we would like to illustrate the alignment of gradients for the reduced centroid from each of these three portion of the lines using the alfven mode .
figure [ fig : rc-1 ] illustrates the @xmath62 reduced centroid from a selected alfven mode velocity data cube .
the central portion is very much the same as the thin slice result , giving an excellent alignment in respect to the magnetic field .
the middle portion of the reduced centroid also shows a fair alignment , but the number of data points gets limited due to the limitations of the numerical resolution and the shot noise increase . nevertheless , the gradient map after sub - block average still provides a nice fit to magnetic field direction .
the wing portion for this illustration is very much limited by the discrete data : it does not contain enough data points even for showing the distribution of centroids .
in fact , in mhd turbulence we expect that the overall profile to be determined by the largest eddies that produce most of the dispersion .
the gradients are not expected to depend on this .
therefore all three parts of the line should be influenced by the small scale eddies that are aligned in respect to the magnetic fields local to them . by looking at the contour structure of the three maps ,
we actually see the correspondence of the elongation to magnetic field directions obtained with different parts of the line .
for observations of galactic hi the use of different portions of the line is not any more just a test of our theoretical concepts . due to the galactic rotation ,
different parts of the galactic hi have different velocities in respect to the observer .
therefore one can trace the variations of magnetic field using the reduced centroids . to demonstrate the effect we use the galfa data and compare our results with the averaged along the line of sight magnetic field directions as they are traced by planck polarization . unlike the study with full centroids in yl17
now we do not expect good tracing if the perpendicular direction of magnetic field is changing along the line of sight .
following the same procedure as in yl17 and selecting the same region with the same resolution , we calculate the reduced centroid by varying velocity thickness @xmath13 .
figure [ fig : obs-1 ] shows the resulting map from the central , middle and wing parts of the line that are obtained using the reduced centroids .
we see notable variations of the alignment which may be indicative of the magnetic field direction variation along the line of sight
. this conclusion can be tested by using starlight polarization from the stars with known distances .
turbulence can be studied using both velocity centroids and channel maps ( see in terms of the properties of correlation functions both techniques have full analytical treatment based on the statistics of the ppv developed in lp00 and @xcite .
the anisotropies of the statistics of the channel maps is described in klp17 and of the velocity centroids in @xcite .
the two techniques have proven to be complementary .
for instance , studies of supersonic turbulence spectra is heavily distorted for velocity centroids @xcite . at the same time , supersonic turbulence is the natural domain for velocity channel analysis ( vca ) suggested in lp00 ( see e.g. @xcite ) . in view of
that we feel it is advantageous to use gradients for both velocity centroids and channel maps .
the most interesting case for the gradient technique that we explore is the case of gradients in thin channel maps . in this regime
the gradients of channel map intensities are affected mostly by velocity fluctuations , the latter tracing magnetic field better than densities . on the contrary , the gradients in the velocity channels in the limiting case when the velocity channels are wide to encompass the entire line correspond the intensity gradients .
the latter case was studied in @xcite .
the statistics of channel of intermediate thickness reflects both density and velocity statistics
. for the studies of magnetic fields in the disk of the galaxy gradients within channel maps allow to investigate the line of sight distribution of magnetic fields using the galactic rotation curve . however , the interpretation of the results for the diffuse distribution should include the dispersion induced by turbulence .
it is not possible to increase the 3d spacial resolution of the magnetic field directions by going to the thin velocity channels .
the change of the channel thickness , nevertheless , changes the relative contributions of the velocity and density fluctuations into the channel map . in terms of resolving the spacial structure of magnetic field , for the atomic hydrogen studies the spatial resolution depends both on the direction of study , the velocity range as well as on the turbulent velocity dispersion .
the latter provides the grid size @xmath63 that should be multiplied to the visual shear along the line of sight @xmath64 in the direction of the observations .
individual clouds with high density contrast can be studied separately .
this can be relevant to e.g. co@xmath65 .
if we imagine a situation that direction of magnetic field is changing along the line of sight the techniques that are discussed here can provide a coarse graded picture of the changes of the magnetic field direction perpendicular to the line of sight . in this way we can distinguish the variations of regular magnetic field over the cells that are larger than @xmath66 . choosing velocity slice thickness much less than @xmath0 we are not getting higher spacial resolutions , but changing the relative contributions of the velocity and density fluctuations in the channels . if we choose the velocity slice thickness of the order of the turbulent velocity dispersion it produces a rough structure of the column densities over the size of the large scale eddies .
therefore the corresponding channel map gradients can be associated with the igs that trace magnetic field with the limitations relevant to the ig tracing .
in the situation when the effects of velocities should be stressed , the reduced centroids can also be used .
in fact , or the hi line broadened through galactic rotation if the averaging over the entire turbulent velocity dispersion and taking gradients brings us back to the well known igs , the calculation of the reduced centroids over the velocity dispersion provides the vcg measure .
in the paper above we have shown that the gradients calculated within velocity channel maps , i.e. vchgs , can trace magnetic field in diffuse media .
these measures are complementary to the velocity centroids gradients , i.e. vcgs .
our decomposition of the turbulent velocities into alfven , slow and fast modes shows that these modes provide different signatures in terms of the vchgs . in particular , 1 .
we tested that the alignments of alfven and slow modes increases as the slice gets thinner .
we showed that fast modes , whose gradients are parallel to magnetic field , can be suppressed by using thin slices .
we showed that density contributions can be suppressed also by using thin slices .
we constructed observable measures termed `` reduced centroids '' and explored their ability to trace magnetic fields .
* acknowledgments .
* elucidating discussions with chris mckee and susan clarke are acknowledged .
al acknowledges the support the nsf grant ast 1212096 , nasa grant nnx14aj53 g as well as a distinguished visitor pve / capes appointment at the physics graduate program of the federal university of rio grande do norte , the inct inespao and physics graduate program / ufrn .
the stay of khy at uw - madison is supported by the fulbright - lee fellowship .
clemens , d. p. , cashman , l. , hoq , s. , montgomery , j. , & pavel , m. d. 2014 , american astronomical society meeting abstracts # 224 , 224 , 220.06 correia , c. , lazarian , a. , burkhart , b. , pogosyan , d. , & de medeiros , j. r. 2016 , , 818 , 118 , a. , & lazarian , a. 2005 , http://dx.doi.org/10.1086/432458 [ , 631 , 320 ] esquivel , a. , & lazarian , a. 2005 , , 631 , 320 esquivel , a. , lazarian , a. , horibe , s. , et al . 2007 , , 381 , 1733 fernandez , e. r. , zaroubi , s. , iliev , i. t. , mellema , g. , & jeli , v. 2014 , , 440 , 298 esquivel , a. , & lazarian , a. 2011 , , 740 , 117 gaensler , b. m. , haverkorn , m. , burkhart , b. , et al .
2011 , http://dx.doi.org/10.1038/nature10446 [ , 478 , 214 ] galtier , s. , pouquet , a. , & mangeney , a. 2005 , physics of plasmas , 12 , 092310 , c. a. , burkhart , b. , lazarian , a. , gaensler , b. m. , & mcclure - griffiths , n. m. 2016 , http://dx.doi.org/10.3847/0004-637x/822/1/13 [ , 822 , 13 ] hayes , j. c. , norman , m. l. , fiedler , r. a. , et al .
2006 , , 165 , 188 heyer , m. , gong , h. , ostriker , e. , & brunt , c. 2008 , , 680 , 420 - 427 , j. c. 1984 , http://dx.doi.org/10.1086/162481 [ , 285 , 109 ] lazarian , a. , pogosyan , d. , & esquivel , a. 2002 , seeing through the dust : the detection of hi and the exploration of the ism in galaxies , 276 , 182 lazarian , a. http://dx.doi.org/10.1086/505796[2006 , 4 ] . 2016 , http://dx.doi.org/10.3847/1538-4357/833/2/131 [ , 833 , 131 ] lazarian , a. , & pogosyan , d. 2000 , , 537 , 720 lazarian , a. , & pogosyan , d. 2004 , , 616 , 943 lazarian , a. , & pogosyan , d. 2006 , , 652 , 1348 , a. , & pogosyan , d. 2012 , http://dx.doi.org/10.1088/0004-637x/747/1/5 [ , 747 , 5 ] , k. c. 1959 , http://dx.doi.org/10.1086/146713 [ , 130 , 241 ] yan , h. , & lazarian , a. 2002 , physical review letters , 89 , 281102 yan , h. , & lazarian , a. 2012 , numerical modeling of space plasma slows ( astronum 2011 ) , 459 , 40 yuen , k. h. , & lazarian , a. 2017 , arxiv:1701.07944 | we explore the ability of gradients of velocity channel map intensities to trace magnetic fields in turbulent diffuse media .
this work capitalizes both on the modern theory of mhd turbulence that predicts the magnetic eddies tend to be aligned with the local direction of magnetic field , and the theory of position - position - velocity ( ppv ) statistics that describes how the velocity and density fluctuations in real space are being mapped into the ppv space .
we show that for steep , e.g. kolmogorov - type density spectrum , the velocity channel gradients ( vchgs ) in thin velocity channels are dominated by velocity contributions . while for the velocity channel thickness comparable to turbulent injection velocities , the vchgs are dominated by the properties of turbulent densities . as turbulent velocity structures
are better aligned with magnetic fields , the tracing with thin channels has the ability of representing the magnetic field better .
we decompose the results of 3d mhd simulations into alfven , slow and fast modes and analyze synthetic maps produced with these modes .
we show that alfven and slow modes act in unison to trace magnetic field , while the velocity gradients produced by the fast mode are orthogonal to those produced by the first two modes .
however , for thin channel maps the contributions from the alfven and slow modes are shown to dominate which allows a reliable magnetic field tracing .
we also introduce centroids that use only part of the spectral line rather the entire spectral line and apply them to galfa 21 cm data .
we compare the directions obtained with the gradients of these `` reduced centroids '' and the magnetic field directions as they are traced by the planck polarization .
we believe that the observed deviations can potentially reveal the variations of the magnetic field along the line of sight . |
microlens planet detections routinely yield the planet - star mass @xmath7 , but since the host star is generally very difficult to observe , the host mass @xmath8 ( and so the planet mass @xmath9 ) and the system distance @xmath10 usually remain undetermined . in principle , one can determine these quantities from the microlensing event itself , provided the angular einstein radius @xmath2 and the microlens parallax @xmath0 are both measured @xcite @xmath11 where @xmath12 is the ratio of the earth s orbit to the einstein radius projected on the observer plane , and @xmath13 $ ] is the lens - source relative parallax , with @xmath14 generally being known from direct observation of the source . to date , @xmath2 has been measured in the majority of planetary events , even though it is rarely ( @xmath15 ) measured in ordinary events .
this is because planets are detected when the source passes near or over a caustic induced by the planet , which yields a measurement of the source size relative to the einstein radius .
hence , if a method could be found to measure @xmath0 for a large fraction of events , the information returned from microlensing planet searches would be radically improved .
the microlens parallax is actually a vector @xmath16 , with the direction being that of the lens - source relative motion .
it can in principle be measured either from the ground @xcite , by combining space - based and ground - based observations @xcite , or from space alone @xcite . when i proposed the pure ground - based method @xcite , i illustrated it with an event that lasted 4 years , so that the lightcurve oscillated annually around the standard microlensing form .
in practice , such 4-year events are never seen , but @xmath16 can be measured from less regular features in shorter events . however , typical events have timescales @xmath17days , so that the great majority are too short to permit a measurement .
@xcite proposed a purely space - based method using _ hubble space telescope ( hst ) _ observations .
since the _ hst _ orbital period is only 1.5 hours , the oscillations are obviously much shorter than @xmath18 .
however , the _ hst _ orbital radius is only @xmath19 , whereas typically @xmath20 .
hence , the oscillations would typically be minuscule . to overcome this obstacle , @xcite proposed that the observations should be made during caustic crossings , which he showed dramatically increase the strength of the signal .
however , alerting _
hst _ to imminent caustic crossings is extremely difficult , more so for planetary events , and in fact such measurements have never been carried out , nor even attempted .
the ground - space combination originally proposed by @xcite remains feasible in principle , but each event must be observed very frequently @xcite , implying a huge space - craft investment .
recently , @xcite showed that a much simpler and cheaper ground - space approach was feasible for high - magnification events , but these constitute a small subset . here
i show that a single survey satellite in geosynchronous orbit is a near - optimal solution to this problem . as in my original proposal ,
oscillations are induced on a standard microlensing lightcurve , but since these have periods of one day rather than one year , they are shorter than the event .
the amplitude of the oscillations is 6.6 times larger than those of the @xcite proposed _ hst _ observations . moreover , no special warning system is required .
it is true that this approach only works for moderately high - magnification ( @xmath21 ) events , but these are more sensitive to planets than more typical events @xcite .
hence , this approach can yield parallaxes for a large fraction of planetary events .
moreover , such a geosynchronous survey satellite is a very real possibility .
the @xcite report recommended a _ wide field infrared space telescope ( wfirst ) _ as its highest priority . a large fraction of the observing time would be devoted to microlensing planet searches . although originally recommended for l2 orbit , a geosynchronous orbit is now being actively considered .
hence , it is most timely to investigate the implications of such an orbit for microlensing parallaxes .
consider an observatory in circular orbit with period @xmath22 and semi - major axis @xmath23 . if this is an earth orbit , these two quantities
are obviously related , but here i begin with the general case .
let the microlensing target be at a latitude @xmath24 with respect to the plane of this orbit , so that the projected semi - minor axis is @xmath25 .
i then normalize these axes to an au @xmath26 now consider that continuous observations of a microlensing event are made at a rate @xmath27 per einstein timescale @xmath18 , with signal - limited flux errors @xmath28 this condition is not as restrictive as may first appear .
we are only concerned with scaling of the errors when the source is relatively highly magnified , so that this form need not hold all the way down to baseline , where sky and blending may be important .
let the lens pass the source on its right at an angle @xmath29 relative to the major projected axis of the satellite orbit , with impact parameter @xmath30 ( in units of @xmath2 ) .
then the lens - source separation vector @xmath31 ( in units of @xmath2 ) is given by @xmath32 where @xmath33 , @xmath34 is the time of closest approach , @xmath35 , and @xmath36 is the orbit phase relative to peak . by definition ,
hence , @xmath38 the magnification is given by @xcite @xmath39 so that the derivatives of the flux with respect to the parallax parameters are @xmath40 where @xmath41 is the unmagnified source flux . in order to make a fisher - matrix analysis
, i first assume that the magnification does not change much over an orbit .
this will ultimately restrict the result to events for which @xmath42 .
i will discuss the more general case further below .
i then evaluate the mean contribution to the fisher matrix , averaged over one orbit @xmath43
@xmath44 @xmath45 and so obtain the inverse covariance matrix by integrating over all observations @xmath46 where @xmath47 in the limit @xmath48 , it is straightforward to show that @xmath49 figure [ fig : deviate ] shows that this approximation holds quite well for @xmath50 ( i.e. , @xmath51 ) . using this approximation
, we obtain , @xmath52 since the ratio of the quantities in the right - hand vector is never greater than @xmath53 ( and is @xmath54 for @xmath55 , which is typical of galactic bulge fields and an equatorial orbit ) , the parallax error is reasonably well characterized by the harmonic rms of these two values , @xmath56
the fiducial values in equation ( [ eqn : sigmapie ] ) are plausible for a _ wfirst _ style mission . for example
, a continuous sequence of 3-minute exposures yields @xmath57 for a @xmath58day event .
a large fraction of sources would have 1% errors in such an exposure , whether continuous or co - added .
note that the formula does not really specify @xmath59 and @xmath27 separately , but just @xmath60 .
moreover , for typical @xmath58day events and a @xmath61day orbit , the assumptions of the derivation basically apply for @xmath62 , i.e. , @xmath63 .
of course , the integral in equation ( [ eqn : gdef ] ) goes to infinity , while the observations do not , but the overwhelming contribution to this integral is from a few @xmath64 near peak .
the fiducial parallax error in equation ( [ eqn : sigmapie ] ) is quite adequate to measure the mass and distance of disk lenses but somewhat marginal for bulge lenses .
for example , for @xmath65 , a typical disk lens at @xmath66kpc has @xmath67 , but a typical bulge lens at @xmath68kpc has @xmath69 . at the very least ,
however , such a measurement would distinguish between bulge and disk lenses .
naively , equation ( [ eqn : sigmapie ] ) appears to require `` effective errors '' of @xmath70 , which must be achieved in the face of both astrophysical and instrumental effects . in particular
, it would seem to challenge the systematics limit for real crowded - field photometry , even from space .
however , what is actually required is not control of systematics to this level , but only control of systematic effects that correlate with orbital phase .
this requirement is still not trivial for geosynchronous orbit because of phase - dependent temperature variations , but it is not as intractable as vetting all systematic errors at this level .
nothing is known about stellar variability at this level in @xmath71 ( likely wfirst passband ) . even _
kepler _ has only a handful of stars for which it can make measurements of this precision in its optical passband ( fig . 4 of @xcite ) .
however , again what is relevant is not variability per se , but variability on 1-day timescales , and @xcite report that fgk stars typically vary on @xmath72-day timescales .
moreover , most forms of intrinsic variability are very subdued in @xmath71 band relative to the optical . see @xcite for a spectacular example of this in a microlensing event . finally , @xcite find that variability declines with increasing proper motion ( so presumably age ) , and the bulge is mostly much older than the disk , though perhaps not entirely @xcite
one would nevertheless like a rigorous test that undetectable variability and systematic effects are not corrupting the results .
fortunately the satellite data themselves provide such a test .
@xcite showed that even very short microlensing events yield `` one - dimensional parallaxes '' from the asymmetry in the lightcurve induced by the earth s ( approximately constant ) acceleration toward the sun .
that is , @xmath73 is well constrained while @xmath74 is very poorly constrained .
normally , such 1-d parallaxes are not considered useful because the amplitude of @xmath0 ( and so @xmath8 ) is also poorly constrained . however , in the present case these 1-d parallaxes serve two very important functions . figure [ fig : lc ] shows the lightcurve of a simulated event with 72 days of observations centered on the vernal equinox in each of 5 years , similar to the anticipated schedule of wfirst .
the event is assumed to have the fiducial parameters from equation ( [ eqn : sigmapie ] ) : @xmath58days , @xmath62 , @xmath75 , with @xmath34 right at the vernal equinox .
the lower panel shows the difference between this event as observed from a geosynchronous orbit , and the same event observed from an inertial platform .
note that since the orbit is equatorial , @xmath76 , i.e. , the components of @xmath16 in the east and north directions projected on the sky .
the event is at @xmath77(18:00:00,-30:00:00 ) , and is assumed to have a parallax @xmath78 .
the oscillations near peak are due to the satellite s orbital motion , while the asymmetries in the wings are from the earth s orbit .
the latter are much larger , but yield only 1-d information .
this is illustrated in figure [ fig : ell ] , which shows the error ellipses due to 1 ) earth - only , 2 ) satellite - only , 3 ) earth+satellite . also shown is the result of the analytic calculation ( which assumed observations extending to infinity ) .
note that there are two sets of solutions in each case , corresponding to the ( @xmath79 ) degeneracy @xcite .
the first point is that if @xmath80 as derived from geosynchronous parallax agrees with the much more precise value derived from the earth s orbit , then one can have good confidence in @xmath81 , for which there is no direct test .
this is true on an event by event basis , but more true for the ensemble of parallax measurements .
the second point is that these two parallax measurements will be automatically combined in the fit to any lightcurve , which means that @xmath80 will be much better determined than @xmath81 .
depending on the relative values of these two components , this may add important information in some cases .
finally , the ensemble of geosynchronous measurements of @xmath80 provides a test of the accuracy of the 1-d earth - orbit measurement of this quantity from the wings of the event .
this is important because the source stars may show variability on 10-day timescales , even if they do not vary on 1-day timescales . using techniques similar to those used to derive equation ( [ eqn : stdev ] ) , it is straightforward to show that the 1-d parallax error due to earth acceleration for short , relatively high - magnification events ( and infinite observations ) is @xmath82 where 58 days is one radian of the earth s orbit and @xmath83 is the projected earth - sun separation in au at time @xmath34 . since @xmath84 for bulge observations made during the equinoxes ,
the ratio of geosynchronous - to - earth parallax errors is @xmath85 since the effective @xmath86day , this implies that the earth - orbit parallax will essentially always yield a precise check on the geosynchronous parallax in one direction .
the derivation underlying equation ( [ eqn : sigmapie ] ) breaks down for @xmath87 : the errors continue to decline with falling @xmath30 , but no longer linearly .
they also become dependent on the orientation and phase of the orbit in a much more complicated way . from the present perspective , the main point is that the formula with @xmath88 provides an upper limit on the errors for events with yet higher peak magnification .
next , the errors derived here assume a point - lens event .
however , since the observations would be near - continuous , it is likely that caustic crossings or near approaches would be captured . as pointed out by @xcite such caustic effects
can significantly enhance the signal .
another feature of these ( and most ) parallax measurements is that they work better at low mass , simply because @xmath89 is bigger .
space - based microlensing measurements have the potential to directly detect the lens when it is more massive ( so , typically , brighter ) .
for example , as the lens and source separate after ( or before ) the event , their joint light becomes extended and the centroids of the blue and red light separate ( if the source and lens are different colors ) .
these effects allowed @xcite and @xcite to measure the host masses in two different planetary events using followup _ hst _ data . because geosynchronous parallax works better at low mass ,
while photometric / astrometric methods work better at high mass , they are complementary .
finally , i note that such parallaxes would be of great interest in non - planetary events as well . without @xmath2 ,
such measurements do not yield masses and distances , but they do serve as important inputs into bayesian estimates of these quantities
. moreover , since the direction of @xmath16 is the same as that of the lens - source relative proper motion @xmath90 , a parallax measurement provides an important constraint when trying to detect / measure the source - lens displacement away from the event .
if the magnitude @xmath91 can be measured from these data , then so can @xmath92 , which in turn yields the mass and distance . | i show that for a substantial fraction of planets detected in a space - based survey , it would be possible to measure the planet and host masses and distances , if the survey satellite were placed in geosynchronous orbit .
such an orbit would enable measurement of the microlens parallax @xmath0 for events with moderately low impact parameters , @xmath1 , which encompass a disproportionate share of planetary detections .
most planetary events yield a measurement of the angular einstein radius @xmath2 . since the host mass is given by @xmath3 where @xmath4 is a constant , parallax measurements are the crucial missing link .
i present simple analytic formulae that enable quick error estimates for observatories in circular orbits of arbitrary period and semi - major axis , and arbitrary orientation relative to the line of sight .
the method requires photometric stability of @xmath5 on @xmath6 orbit timescales .
i show that the satellite data themselves can provide a rigorous test of the accuracy of the parallax measurements in the face of unknown systematics and stellar variability , even at this extreme level . |
the discovery @xcite of a new boson with a mass close to 125 gev has been well anticipated as the standard model higgs boson @xcite and provided the first experimental evidence of the higgs mechanism @xcite .
it is a great triumph , but not an end , of the giant campaign for higgs hunting in the development of particle physics .
although the subsequent more precise measurements @xcite at the lhc have shown the properties of the higgs boson are well consistent with the predictions of the standard model ( sm ) , the precision of the current experimental data still leave open the possibility of an extended higgs sector @xcite . among many new physics scenarios beyond the sm , the two higgs doublet models ( 2hdm ) @xcite are the simplest extensions of the sm . in the 2hdms
, an additional higgs doublet is introduced to the sm higgs sector , which could result in rich phenomena , in collider physics @xcite , flavor physics @xcite , neutrino physics @xcite , dark matter @xcite and cosmology @xcite . however , unlike the sm , unwanted tree - level flavor - changing neutral current ( fcnc ) interactions in the 2hdm are not forbidden by the glashow - illiopoulos - maiani ( gim ) mechanism . besides some other solutions @xcite , this issue is usually addressed by the natural flavor conservation ( nfc ) hypothesis through imposing a discrete @xmath1 symmetry @xcite . according to different @xmath1 charge assignments
, there are four types of the nfc 2hdm , referred to as the type - i , type - ii , type - x and type - y 2hdm , respectively .
of course , there are new parameters in the 2hdms to be determined or excluded by the measurements of electro - weak processes . to this end , b - meson decays are usually employed to constrain their parameter spaces . among the rare b - meson decays , the leptonic processes @xmath11 ( @xmath12 , or @xmath13 ) are of special interest @xcite .
they suffer from very few hadronic uncertainties and are induced by fcnc transitions , which make them sensitive probes to the effects of physics beyond the sm , especially models with a non - standard higgs sector @xcite .
recently , the next - to - leading order ( nlo ) electroweak corrections and the next - to - next - to - leading order ( nnlo ) qcd corrections @xcite in the sm have been calculated . on the bsm side ,
a full one - loop calculation in the aligned 2hdm ( a2hdm ) has been performed in ref .
@xcite . motivated by this progress ,
in this paper we perform a detailed study of the @xmath0 decay within the 2hdms with @xmath1 symmetry . at present , this process is calculated in the type - ii 2hdm in large @xmath2 limit only @xcite . using the higgs base correspondence between the a2hdm and the 2hdms , we will derive the relevant full one - loop wilson coefficients of the four variant 2hdms contributing to the @xmath0 decay from the recent a2hdm results @xcite without the large @xmath2 approximation .
we also investigate the possibility to discriminate the four different types of 2hdm in the light of the recent collider and flavor physics data , as an update of our previous work @xcite .
we combine the constraints from @xmath14 , @xmath15 mixing , @xmath16 and @xmath17 @xcite , with the experimental data from the direct search for higgs bosons at lep @xcite , tevatron @xcite and lhc @xcite , and the constraints from perturbativity , tree - level vacuum stability and perturbative unitary . for the @xmath0 decay ,
the correlations between its branching ratio and the mass - eigenstate rate asymmetry @xmath6 are also reevaluated with the constrained parameter space of the 2hdms obtained in this paper .
we have found that @xmath6 can slightly deviate from the sm prediction in the type - ii 2hdm only , and that the ratio of time - integrated @xmath5 gets similar contributions from the four 2hdms ; this makes it very hard to discriminate the four types of 2hdms with the correlation between @xmath6 and @xmath5 as suggested in our previous work @xcite .
the paper is organized as follows . in section [ sec : bs_lm_lm ] , we give a brief overview of the @xmath0 decay . in section [ sec : wc ] , full one - loop contributions from the 2hdms with @xmath1 symmetry are derived explicitly . in section [ sec : analysis ] , we give our detailed numerical results and discussions .
we conclude in section [ sec : conclusion ] .
the relevant theoretical formulas are recapitulated in the appendix .
in the sm , the leptonic decays @xmath11 ( @xmath12 or @xmath13 ) arise from the @xmath18 box and @xmath19 penguin diagrams .
generally , these decays can be described by the low - energy effective hamiltonian @xmath20 where @xmath21 denotes the qed fine - structure constant and @xmath22 the ckm matrix elements .
the semi - leptonic operators are defined as @xmath23 in the sm , the contributions from the scalar operators @xmath24 and @xmath25 are highly suppressed ( the corresponding wilson coefficients are given in eq . . ) , but @xmath26 will play the dominant role .
its explicit expressions up to the nlo qcd corrections can be found in ref .
recently , calculations of the nlo ew @xcite and nnlo qcd @xcite corrections have also been completed @xcite .
this progress will be incorporated into our calculations . with the effective hamiltonian eq .
, the branching ratio of @xmath11 reads @xmath27 where @xmath28 , @xmath29 and @xmath30 denote the mass , mean lifetime and decay constant of @xmath31 meson respectively .
the short - distance contributions @xmath32 and @xmath33 are defined as @xmath34 as discussed in the following section , there is no bsm phase in the 2hdms with @xmath1 symmetry .
therefore , we only consider the case that both @xmath32 and @xmath33 are real in this paper . as pointed out in ref .
@xcite , the measured branching ratio of @xmath11 should be the time - integrated one , denoted by @xmath35 . in order to compare with the experimental measurements , the sizable effect of @xmath36 oscillations
should be taken into account @xcite , and one has @xmath37 where the mass - eigenstate rate asymmetry @xmath6 can be expressed as @xmath38 the observable @xmath6 is independent of the branching ratio of @xmath0 and provides complementary information on the short - distance structure of this decay . in the sm , @xmath39 .
following ref .
@xcite , it is convenient to introduce the ratio @xmath40 where both hadronic uncertainties and ckm matrix elements are canceled out .
in the 2hdms with @xmath1 symmetry , @xmath41 processes receive contributions from box diagrams with charged higgs and penguin diagrams with @xmath19 boson and neutral higgs bosons .
the wilson coefficient @xmath26 has been calculated in the type - ii 2hdm @xcite . for @xmath42 and @xmath43 ,
only the leading contributions in the large @xmath2 limit have been computed in the type - ii model @xcite .
however , the remaining contributions could be important for some specific @xmath2 values in the other types of 2hdms . in this section
, we first of all give a brief introduction to the 2hdms with @xmath1 symmetry , and then show that the wilson coefficients could be derived explicitly from the recent full one - loop results of the a2hdm @xcite .
the 2hdm extends the sm higgs sector with an additional scalar doublet . with the two higgs
doublets @xmath44 and @xmath45 , the cp - conversing 2hdm potential with a softly broken @xmath1 symmetry reads @xcite @xmath46,\end{aligned}\ ] ] where @xmath47 is a soft @xmath1 symmetry breaking term and the parameters @xmath48 and @xmath49 are real .
the two higgs doublets @xmath44 and @xmath45 can be generally parameterized as @xmath50 where the two vacuum expectation values ( @xmath51 ) @xmath52 and @xmath53 are real and positive . from the vacuum condition
@xcite @xmath54 they can be expressed as other parameters in the higgs potential , where @xmath55 is defined . by introducing the @xmath51 @xmath56 ( @xmath57 ) , the mixing angle @xmath58 and the soft @xmath1 symmetry breaking parameter @xmath3 as @xmath59 , @xmath60 and @xmath61
, we can use @xmath62 as independent 2hdm potential parameters .
physical higgs states are obtained by the following rotations : @xmath63 where the rotation matrix is given by @xmath64 the mixing angle @xmath65 is determined by the higgs potential of eq .
@xcite , @xmath66 in the 2hdm with @xmath1 symmetry , the physical higgs spectrum consists of five degrees of freedom : two charged scalars @xmath67 , two cp - even neutral scalars @xmath68 and @xmath69 , and one cp - odd neutral scalar @xmath70 . the quartic couplings @xmath71 in the higgs potential can be expressed in terms of their masses as @xcite @xmath72 therefore , the eight parameters in the higgs potential @xmath48 and @xmath49 can be rewritten equivalently by the four physical higgs masses @xmath73 , @xmath74 , @xmath75 , @xmath76 , the two mixing angles @xmath65 and @xmath58 , the @xmath51 @xmath77 , and the @xmath1 symmetry breaking parameter @xmath3 . in the case of @xmath78 , which is considered in ref @xcite ,
@xmath3 can be eliminated and the 2hdm potential parameters can be expressed by seven parameters @xmath79 as @xmath80 in the interaction basis , the general yukawa lagrangian of the 2hdm can be written as @xmath81 where @xmath82 , @xmath83 and @xmath84 denote the sm quark and lepton doublets , and @xmath85 , @xmath86 , and @xmath87 are the right - handed up - type quark , down - type quark and lepton singlet , respectively .
the yukawa coupling matrices @xmath88 are @xmath89 complex matrices in flavor space .
l c c c c c c & @xmath44 & @xmath45 & @xmath85 & @xmath86 & @xmath90 & @xmath83 , @xmath84 + ' '' '' height 1pt @a xhline type - i & @xmath91 & @xmath92 & @xmath92 & @xmath92 & @xmath92 & @xmath91 + type - ii & @xmath91 & @xmath92 & @xmath92 & @xmath91 & @xmath91 & @xmath91 + type - x & @xmath91 & @xmath92 & @xmath92 & @xmath92 & @xmath91 & @xmath91 + type - y & @xmath91 & @xmath92 & @xmath92 & @xmath91 & @xmath92 & @xmath91 + in order to avoid tree - level fcnc , a discrete @xmath1 symmetry is introduced @xcite . all the possible nontrivial @xmath1 charge assignments are listed in table [ tab : type ] , which define the four well - known types of 2hdm , i.e. type - i , type - ii , type - x and type - y . in the mass - eigenstate basis
, the yukawa interactions can be written in the form @xmath93\nonumber\\ & + \frac{\sqrt 2}{v}\bar u \left ( m_u v \xi_a^u p_l+ v m_d\xi_a^d p_r \right ) d h^+ + \frac{\sqrt2m_\ell\xi_a^\ell}{v}\bar\nu_l \ell_r h^+ + \text{h.c.},\label{eq : yukawa}\end{aligned}\ ] ] where @xmath94 .
the yukawa couplings @xmath95 in the four types of 2hdm are listed in table [ tab : yuk ] .
in addition , the couplings of the light cp - even higgs boson @xmath68 to gauge bosons @xmath96 or @xmath97 can be written as @xmath98 , which is normalized to the corresponding couplings of the sm higgs boson @xmath99 @xcite .
recently , the lhc run i data confirm the sm higgs - like nature of the @xmath100 boson discovered at the lhc @xcite .
if the light cp - even higgs @xmath68 in the 2hdm is identified with the observed @xmath100 boson , global fits to the lhc higgs data suggest that all four types of 2hdm should lie close to the so - called alignment limit @xcite @xmath101 where both the yukawa and the gauge couplings of @xmath68 are identical to the values of the sm higgs boson . from eqs . and , the alignment limit can be achieved when the quartic couplings in the higgs potential satisfy @xcite @xmath102 for recent studies on the alignment limit in the 2hdm , we refer to ref .
@xcite .
lccccccccc & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 + ' '' '' height 1pt @a xhline type - i & @xmath112 & @xmath113 & @xmath113 & @xmath114 & @xmath114 & @xmath114 & @xmath115 & @xmath116 & @xmath116 + type - ii & @xmath112 & @xmath117 & @xmath117 & @xmath114 & @xmath118 & @xmath118 & @xmath115 & @xmath119 & @xmath119 + type - x & @xmath112 & @xmath113 & @xmath117 & @xmath114 & @xmath114 & @xmath118 & @xmath115 & @xmath116 & @xmath119 + type - y & @xmath112 & @xmath117 & @xmath113 & @xmath114 & @xmath118 & @xmath114 & @xmath115 & @xmath119 & @xmath116 + since the 2hdms with @xmath1 symmetry are particular cases of the a2hdm @xcite , there exists a one - to - one correspondence for yukawa couplings between these two models .
however , the correspondence is not so straightforward for higgs cubic couplings . unlike the 2hdms with @xmath1 symmetry ,
the a2hdm potential is usually defined in the so - called `` higgs basis '' @xcite , in which only one higgs doublet gets a nonzero @xmath51 .
therefore , the parameter @xmath2 defined in the nfc 2hdms is not a physical parameter in the a2hdm @xcite . in both the a2hdm and the nfc 2hdms , @xmath0 decay is induced by gauge boson @xmath19 , goldstone boson @xmath120 , and higgs bosons @xmath121 penguin diagrams , as well as box diagrams mediated with @xmath122 , @xmath67 , and @xmath123 . to one - loop level ,
their contributions to the wilson coefficients are divided into the following different parts : @xmath124 where each part in the parentheses is gauge invariant .
this gauge invariance is validated by the actual calculation in both the feynman and the unitary gauges in the a2hdm @xcite .
the wilson coefficients labeled with `` sm '' denote the contributions from the diagrams involved with only the sm fields ( with the goldstone bosons but not the higgs boson ) , whose expressions are given in appendix [ sec : app ] . those with `` 2hdm '' contain the higgs contributions . for simplicity ,
their explicit expressions are given in the unitarity gauge in the following , where the goldstone boson contributions are absent .
the higgs bosons affect the box and @xmath19 penguin diagrams with yukawa interactions .
their contributions to wilson coefficients in the nfc 2hdms can easily be obtained from the a2hdm results with replacement of the yukawa couplings , @xmath125 for self - contained , we present the wilson coefficients after the correspondences made in appendix [ sec : app ] .
the higgs penguin diagrams involve yukawa couplings as well as higgs - gauge couplings and higgs cubic couplings .
therefore , their wilson coefficients can not be derived from the a2hdm results so straightforwardly as in the box and @xmath19 penguin diagrams , as discussed in previous section . since the a2hdm wilson coefficients are given for individual higgs penguin diagrams in ref .
@xcite , we use the following approach . for every higgs penguin diagram in the nfc 2hdms ,
its contribution is derived from the a2hdm results with the replacement of the higgs - gauge vertex and the triple higgs vertex .
then the total contributions to the wilson coefficients are obtained , @xmath126 where @xmath127 , @xmath128 , the functions @xmath129 defined in eq .
, and the higgs cubic couplings are defined as @xmath130 where the soft @xmath1 symmetry breaking parameter @xmath3 has been defined in sec [ sec:2hdm ] . in the literature @xcite , it is found that the wilson coefficients can receive large @xmath2 enhancement only in the type - ii 2hdm and the branching ratio with large @xmath2 depends only on the higgs masses
@xmath76 , @xmath74 , @xmath73 and the mixing angle @xmath65 . however , as shown by eqs . and , a term proportional to @xmath131 in our full one - loop wilson coefficient @xmath42 is also enhanced by @xmath4 , which comes from the heavy higgs @xmath69 penguin diagrams mediated by charged higgs bosons . using the parameter @xmath132 in the higgs potential of eq .
directly , this term is proportional to @xmath133 and enhanced by @xmath134 .
this @xmath3 dependent term has not been considered yet in the previous studies in the literature . therefore , its effects are worthy of a detailed investigation .
the soft @xmath1 symmetry breaking parameter @xmath3 is associated with the spontaneous cp breaking @xcite and characterizes the masses of all the higgs bosons @xcite .
this parameter enters the @xmath0 decays through the higgs penguin diagrams .
however , it is found that the @xmath3 term can not make more significant contributions than other terms of the wilson coefficient @xmath42 . here , we would choose @xmath68 as the higgs boson discovered by atlas @xcite and cms @xcite and take the alignment limit @xmath135 , which is favored by the current 2hdm fits @xcite .
then the cubic couplings in eq .
read @xmath136 focusing on the coupling @xmath137 , it can be seen from eqs . and that large contributions from this coupling would require @xmath138 . however , we know @xmath139 from the 2hdm vacuum condition @xcite and perturbativity @xcite .
it is also noted that , the higgs penguin diagrams can be enhanced by very large @xmath2 or @xmath140 . in all the four types of 2hdms
, the @xmath141 contributions could be enhanced by large @xmath142 . in practice , @xmath143 has been excluded by the perturbativity @xcite . similarly , the coupling @xmath144 can make a large contribution if @xmath145 . among the four models ,
this contribution is enhanced by @xmath4 only in type - ii 2hdm .
however , the ratio @xmath131 still suffers from the theoretical constraints , which will be discussed with numerical results in the following section .
although the effects from the operators @xmath24 and @xmath25 are suppressed by @xmath146 , these two scalar operators can make significant contributions in the two parameter regions : ( i ) in the type - ii 2hdm , since both @xmath42 and @xmath43 contain @xmath2 enhanced terms , the effects of the scalar operators are enhanced in the parameter space with large @xmath2 .
( ii ) the contributions from the cp - odd higgs penguin diagrams are inversely proportional to the mass of the cp - odd higgs boson @xmath70 .
thus , the wilson coefficient @xmath43 becomes much more significant in the region with small values of @xmath75 @xcite .
] in all the four 2hdms . in the particular case of the type - ii 2hdm
, our result of @xmath26 agrees with the one calculated in ref .
@xcite . for
the wilson coefficients @xmath42 and @xmath43 in the 2hdm , the calculations have been performed by various groups @xcite .
the latest results are presented in these three papers @xcite , where the 2hdm contributions are computed in the type - ii model in some specific cases . in ref .
@xcite , the wilson coefficients are calculated in large @xmath2 limit , i.e. , only @xmath4 enhanced terms are kept .
however , the higgs penguin diagrams with trilinear @xmath147 and @xmath148 couplings are not considered . in refs .
@xcite and @xcite should be included .
] , after including these penguin diagrams , the calculations are performed again in the large @xmath2 limit but with the assumption @xmath149 for the couplings in the higgs potential ) is different from the one defined in eq . .
this condition is also expressed as @xmath78 by the couplings used in our paper . ] . considering only terms proportional to @xmath4
, our result agrees with the one of ref .
@xcite in the case of @xmath150 , and those of ref .
@xcite in the case of @xmath78 .
generally , the 2hdm contains eight free parameters , i.e. , @xmath48 and @xmath49 in the higgs potential of eq . .
they can be rewritten equivalently in terms of the higgs masses @xmath73 ,
@xmath74 , @xmath75 , @xmath76 , the mixing angles @xmath65 and @xmath58 , the parameter @xmath3 , and the @xmath51 @xmath77 . if the condition @xmath78 is assumed , @xmath3 can be expressed by the other parameters , as shown in eq . .
it is the reason why terms depending on the @xmath1 symmetry breaking parameter @xmath3 were absent in the previous calculations @xcite , but are present in this paper .
searches for @xmath151 decays have been performed at the babar , belle , and tevatron ( for a review , see ref .
@xcite ) . at the lhc ,
measurements by cms @xcite and lhcb @xcite collaborations with the full data of lhc run i have resulted in the averaged values for the time - integrated branching ratios @xcite @xmath152 where the errors are dominated by the statistical uncertainties and expected to be significantly reduced in the near future . both of them are in good agreement with the latest updated sm predictions @xcite , @xmath153 and @xmath154 , in which the nlo ew @xcite and the nnlo qcd @xcite corrections have been included .
thus , strong constraints on the 2hdm parameters are expected . in the nfc 2hdms , the relevant parameters are the two mixing angles @xmath65 and @xmath58 , four higgs mass parameters @xmath76 , @xmath73 , @xmath74 , and @xmath75 . in
the @xmath151 decays , the @xmath1 symmetry breaking parameter @xmath3 also enters into the decay amplitude and is independent from these parameters . as discussed in ref .
@xcite , we choose the light neutral higgs @xmath68 in the 2hdm as the sm higgs observed at the lhc and adopt the alignment limit @xmath155 .
then the model parameters are reduced to @xmath156 . as discussed in ref .
@xcite , we shall restrict these parameters in the following ranges : @xmath157\gev , \qquad m_{h^\pm},m_a , m\in [ 1,1000]\gev,\qquad \tan\beta \in [ 0.1 , 100].\end{aligned}\ ] ] starting from these parameter spaces , we will start our numerical scan . in the numerical analysis , we impose experimental constraints in the same way as in ref . @xcite . to constrain the 2hdm parameters ,
we have taken into account ( i ) flavor processes : @xmath15 mixing , @xmath17 , @xmath158 and @xmath14 decays , ( ii ) direct searches for higgs bosons at lep @xcite , tevatron @xcite and lhc @xcite , both of which have been discussed in detail in our previous work @xcite .
additionally , we also consider the oblique parameter @xmath159 in the ew precision measurement @xcite and require the couplings @xmath49 to satisfy ( iii ) theoretical constraints : perturbativity @xcite , tree - level vacuum stability @xcite and perturbative unitarity @xcite ( see ref .
@xcite for the expressions ) . for @xmath0 decay , both the nnlo qcd and the nlo ew corrections in the sm and the full one - loop contributions in the 2hdm are included . as discussed in sec .
[ sec : wc ] , the effects of the soft @xmath1 symmetry breaking parameter @xmath3 can be enhanced by large @xmath2 in the type - ii 2hdm .
the @xmath3 dependence of the branching ratio @xmath160 is shown in figure [ fig : m ] in the type - ii 2hdm for various @xmath2 and @xmath74 values .
as expected , the effects of @xmath3 become significant when the two ratios @xmath131 and @xmath2 are large .
however , it is found that the theoretical constraints from perturbativity , vacuum stability , and perturbative unitarity have put the bound @xmath161 ( and @xmath162 ) in the parameter space of eq . .
therefore , the soft @xmath1 symmetry breaking parameter @xmath3 can not make more significant effects than the other @xmath2 enhanced terms in @xmath42 and @xmath43 . after considering the current experimental data ,
the allowed parameter spaces of all the four 2hdms are obtained .
since the constraints from @xmath163 appear to be more or less weaker than those from @xmath0 , we only show the results from the latter one , which are plotted in the @xmath164 plane in figure [ fig : ps ] . compared to our previous results @xcite ,
the parameter space with small @xmath2 is excluded for all the four types of 2hdms .
this change is caused by the contributions with small @xmath2 neglected in the previous calculations @xcite but included in the present full one - loop computation as discussed in section .
[ sec : wc ] . for the large @xmath2 region ,
only the type - ii model is bounded , which is in agreement with our previous result but still weaker than the one from @xmath165 . combining all the constraints aforementioned ,
we obtain the survived parameter space of all the four types 2hdms , as an update of our previous results @xcite , which is shown in the @xmath166 plane in figure [ fig : ps_all ] . it is found that the small @xmath2 region is restricted for all the four models by @xmath36 mixing and @xmath167 , while the large @xmath2 region is constrained only in the type - ii 2hdm by @xmath16 and @xmath0
decays . compared to our previous results ,
the current constraints on the large @xmath2 region in the type - ii 2hdm are more stringent .
this is mainly because the theoretical constraints are included in the current analysis . in these constrained parameter spaces of the four 2hdms ,
the correlations between the observables @xmath6 and @xmath168 defined in eqs . and are reevaluated , which are presented in figure [ fig : correlation ] . unlike our previous results @xcite ,
the correlations in the four different types of 2hdms are almost indistinguishable .
the allowed ranges of @xmath168 are the same for all the four models , while @xmath6 can deviate slightly from the sm prediction only in the type - ii 2hdm .
it is found that the difference from our previous results is mainly caused by the theoretical constraints and the new full one - loop wilson coefficients considered in the current analysis . in the type - ii 2hdm , the bounds on @xmath2 are more stringent compared to our previous results as discussed above .
thus , the allowed range of @xmath42 is restricted more stringently in the current analysis . in this case
, @xmath6 can deviate from the sm prediction very tiny , which can be seen from eq . .
as discussed in sec .
[ sec : wc ] , our results of the wilson coefficients can also be applied to the small @xmath2 region in all the four models , while some terms are not included in @xmath43 used in our previous analysis . in the case of small @xmath75 , @xmath43 is enhanced and these terms make the allowed regions of @xmath168 in the type - i and type - y 2hdms as large as the one in the type - x model .
meanwhile , the value of @xmath168 is almost independent of @xmath6 in the type - ii 2hdm .
in this paper , we have performed an updated analysis of the rare leptonic decay @xmath0 in the 2hdm with a softly broken @xmath1 symmetry .
we have derived the full one - loop wilson coefficients @xmath26 , @xmath42 and @xmath43 from the recent a2hdm results @xcite , which can be applied to the contributions of all the four types of 2hdms for both large and small @xmath2 value .
our main conclusions are summarized as follows : * compared to @xmath26 , the wilson coefficients @xmath42 and @xmath43 are negligible in the entire 2hdm parameter space , except for large @xmath2 in the type - ii 2hdm or small cp - odd higgs mass @xmath75 in the four models .
in addition , only the wilson coefficients @xmath42 and @xmath43 in the type - ii 2hdm can be enhanced by large @xmath2 . *
the soft @xmath1 symmetry breaking parameter @xmath3 enters into the higgs penguin diagrams and affects the wilson coefficient @xmath42 .
the dominant contributions are proportional to @xmath131 and enhanced by @xmath4 in the type - ii 2hdm , which have not been considered in the literature @xcite . however , after combing the theoretical constraints from perturbativity , vacuum stability and perturbative unitarity , we have found that the parameter @xmath3 can not make more significant contributions than other terms in the wilson coefficients . * after imposing the experimental constraints , regions with small @xmath2 are excluded for all the four types of 2hdm , which are quite different from our previous results @xcite .
as expected , large @xmath2 region is only excluded in the type - ii 2hdm . as an update of our previous analysis @xcite
, we have also investigated the possibility to distinguish the four types of 2hdm in light of the recent updated flavor physics data , the collider data from the direct searches for higgs bosons and the theoretical progresses .
the combined bounds on the 2hdm parameters have been derived for the four models . in the survived parameter regions , the correlations between @xmath6 and @xmath168 in all the four of the 2hdms are almost indistinguishable from each other . in the 2hdms with @xmath1 symmetry , @xmath6 can only have a very tiny deviation from the sm prediction , while @xmath168 could deviate from the sm one sizably .
this could be tested by the much more precise measurement of @xmath0 at the lhc in the coming years .
the work was supported by the national natural science foundation of china ( nsfc ) under contract nos . 11225523 , 11221504 and 11435003 .
xy is also supported in part by the nrf grant funded by the korean government of the mest ( no .
2011 - 0017430 ) and ( no . 2011 - 0020333 ) , and by the open project program of skltp , itp , cas ( no . y4kf081cj1 ) .
xy acknowledges the hospitality of the itp , cas , where this work was finalized .
we thank xin - qiang li for useful discussions .
in this appendix , we recapitulate the relevant expressions of the wilson coefficients in the sm and the four types of the 2hdms for completeness , which are obtained from ref .
@xcite . in the sm
, the one - loop wilson coefficients of the scalar operators can be written as @xmath169 in the unitary gauge , their expressions read @xmath170\nonumber \\ & - \frac{s_w^2}{36}\left[\frac{x_t(18 x_t^3 - 139 x_t^2 + 274 x_t-129)}{2 ( x_t-1)^3 } + \frac{24 x_t^4 - 33 x_t^3 - 45
x_t^2 + 50 x_t-8}{(x_t-1)^4 } \ln x_t\right]\ , , \nonumber\end{aligned}\ ] ] where @xmath171 denotes the contributions from the sm higgs penguin diagrams .
the other wilson coefficients @xmath172 and @xmath173 are same in the sm and the 2hdms . in the four types of the 2hdms , the various contributions in the wilson coefficients of eq .
are obtained by the replacement of the yukawa couplings in eq . , which are given in the unitary gauge , @xmath174 -\xi_a^d\xi_a^\ell \frac{\xt\ln(\xhc/\xt)}{4(\xhc-\xt)}\ , , \\[0.8em ] c_{p,\,\unitary}^{\rm box,\,\thdm}= & + \frac{\xi_a^u\xi_a^\ell\xt}{8(\xhc-\xt ) } \left[1 + \frac{2\xt^2-\xhc\xt-\xhc}{(\xt-1)(\xhc-\xt)}\ln\xt + \frac{\xhc ( 1 - 2\xt + \xhc)}{(\xhc-1)(\xhc-\xt)}\ln\xhc \right ] \nonumber\\ & + \xi_a^d\xi_a^\ell \frac{\xt\ln(\xhc/\xt)}{4(\xhc-\xt)}\ , , \nonumber\\[0.8em ] c^{z,\ , \thdm}_{10,\,\unitary}= & + ( \xi_a^u)^2\,\frac{x_t^2}{8}\left[\frac{1}{\xhc - x_t } - \frac{\xhc\ln(\xhc/\xt)}{(\xhc - x_t)^2}\right]\ , , & \nonumber\\[0.8em ] c_{p,\,\unitary}^{z,\,\thdm}= & + \frac{x_t}{4(\xhc - x_t)^2}\ , \biggl\lbrace -\xi_a^d\xi_a^u\biggl[- \frac{x_t+\xhc}{2 } + \frac{x_t \xhc}{\xhc - x_t}\ln\frac{\xhc}{\xt}\biggr ] \nonumber\\ & \hspace{7.38em } + ( \xi_a^u)^2\biggl[\frac{\xhc^2 - 8\xhc x_t-17x_t^2}{36(\xhc - x_t ) } -\xt ( \xhc-\xt ) \nonumber\\ & \hspace{11.3em}+\left(\frac{x_t^2(3\xhc+x_t)}{6(\xhc - x_t)^2}+\xt\xhc\right)\ln\frac{\xhc}{\xt}\biggr]\biggr\rbrace \nonumber\\[0.5em ] & + \frac{s_w^2 x_t}{6(\xhc - x_t)^2}\,\biggl\lbrace-\xi_a^d\xi_a^u\,\biggl[\frac{5 x_t-3 \xhc}{2 } + \frac{\xhc(2\xhc-3x_t)}{\xhc - x_t}\ln\frac{\xhc}{\xt}\biggr ] \nonumber\\ & \hspace{7.52em } - ( \xi_a^u)^2\biggl[\left(\frac{4 \xhc^3 - 12 \xhc^2 x_t+9 \xhc x_t^2 + 3 x_t^3}{6(\xhc - x_t)^2}+\frac{3}{2}\xt\xhc\right)\ln\frac{\xhc}{\xt } \nonumber\\ & \hspace{11.3em}- \frac{17 \xhc^2 - 64 \xhc x_t+71 x_t^2}{36(\xhc - x_t ) } -\frac{3}{2}\xt(\xhc-\xt)\biggr]\biggr\rbrace \,.\nonumber\end{aligned}\ ] ] the higgs penguin contributions
@xmath175 have been given in eq .
, where the functions @xmath176 are defined as @xmath177\ , , \\
g_1^{(a)}(x_t,\xhc,-\xi_a^u,\xi_a^d ) & = -\frac{\,3\,}{4 } -\xi_a^u\xi_a^d ( f_1+f_2+f_3 ) + ( \xi_a^u)^2 ( f_4-f_5)\ , , \nonumber\\ g_2^{(a)}(x_t,\xhc,-\xi_a^u,\xi_a^d ) & = -(\xi_a^d)^2\xi_a^u f_1 + \xi_a^d(\xi_a^u)^2 ( f_3 + f_2 ) + ( \xi_a^u)^3 ( f_5 - f_4 ) + \xi_a^u ( f_7 - f_6 ) + \xi_a^d f_1\ , , \nonumber\\ g_3^{(a)}(x_t,\xhc,-\xi_a^u,\xi_a^d ) & = -(\xi_a^d)^2 \xi_a^u f_1 + \xi_a^d(\xi_a^u)^2 ( f_3 - f_2 ) -(\xi_a^u)^3 ( f_5 + f_4 ) -\xi_a^u ( f_7 + f_6 ) + \xi_a^d f_1\,.\nonumber\end{aligned}\ ] ] here the one - loop functions @xmath178 are abbreviated as @xmath179 with the definitions @xmath180\ , , \\ f_2(x , y ) = & \frac{1}{2(y - x)^{\phantom{2 } } } \left[x - \frac{y x}{y - x}(\ln y-\ln x)\right]\ , , \nonumber\\ f_3(x , y ) = & \frac{1}{2(y - x)^{\phantom{2 } } } \left[y-\frac{y^2 \ln y}{y - x}+\frac{x ( 2y- x ) \ln x}{y - x}\right]\ , , \nonumber\\ f_4(x , y ) = & \frac{1}{4(y - x)^2 } \left[\frac{x \left(3 y - x\right)}{2}-\frac{y^2 x}{y- x}(\ln y-\ln x)\right]\ , , \nonumber\\ f_5(x , y ) = & \frac{1}{4(y - x)^2 } \left[\frac{x ( y-3 x)}{2}-\frac{y x ( y-2 x)}{y- x}(\ln y-\ln x)\right]\ , , \nonumber\\ f_6(x , y ) = & \frac{x\left(x^2 - 3yx+9y-5 x-2\right)}{8(x-1)^2(y - x ) } + \frac{y \left(y x-3y+2 x\right ) } { 4 ( y-1)(y- x)^2}\ln y\nonumber \\[0.2 cm ] & + \frac{y^2 \left(-2 x^3 + 6 x^2 - 9 x+2\right)+3 y x^2 ( x^2 - 2 x+3)-x^2 \left(2 x^3 - 3 x^2 + 3 x+1\right)}{4 ( x-1)^3
( y- x)^2}\ln x\ , , \nonumber\\ f_7(x , y ) = & \frac{\left(x^2+x-8\right)}{8 ( x-1)^2}-\frac{y ( y+2 ) } { 4 ( y-1)(y - x)}\ln y+\frac{y \left(x^3 - 3 x^2 + 3 x+2\right)+3 \left(x-2\right ) x^2}{4 ( x-1)^3(y - x)}\ln x\,.\nonumber\end{aligned}\ ] ] it is noted that the divergence in the higgs penguin diagrams at one - loop level is canceled by a fcnc local operator in the a2hdm @xcite . in the 2hdms with @xmath1 symmetry
, we find that the divergence automatically vanishes after adding all the higgs penguin contributions . for the four types of 2hdm , the values of the relevant yukawa couplings are listed in table [ tab : yuk ] . when deriving the expressions of the higgs penguin diagrams in eq .
, the following identities have been used : @xmath181 and @xmath182 which can be obtained from table [ tab : yuk ] .
it should be noted that there is a freedom in the definitions of the functions @xmath178 , since adding the lhs of eq . to eq . does not change @xmath176 .
j. albrecht , mod .
a * 27 * , 1230028 ( 2012 ) [ arxiv:1207.4287 [ hep - ex ] ] .
s. schael _ et al . _
[ aleph and delphi and l3 and opal and lep working group for higgs boson searches collaborations ] , eur . phys .
j. c * 47 * ( 2006 ) 547 [ hep - ex/0602042 ] . | we revisit the rare leptonic decay @xmath0 in the two - higgs doublet models with a softly broken @xmath1 symmetry , namely type - i , type - ii , type - x and type - y 2hdms .
we have derived the relevant full one - loop wilson coefficients of the four 2hdms from the recent calculation in the aligned two - higgs doublet model by li , lu and pich , which could be mapped to all the four 2hdms for both large and small @xmath2 .
it is found that a new term associated with the soft @xmath1 symmetry breaking parameter @xmath3 can be enhanced by @xmath4 in the type - ii 2hdm , which has not been considered in the literature . imposing both theoretical and experimental constraints , we have renewed the bounds on the parameter spaces of the four 2hdms .
different from our previous paper , however , we find that all the four 2hdms give sizable and similar contributions to @xmath5 within the stringently restricted parameter spaces , but very tiny as regards the mass - eigenstate rate asymmetry @xmath6 ; this makes it unfeasible to discriminate the four types of 2hdm with the correlations between the observables in @xmath0 decay .
_ @xmath7institute of particle physics and key laboratory of quark and lepton physics ( moe ) , + central china normal university , wuhan , hubei 430079 , p. r. china _
+ _ @xmath8college of physics and electronic engineering , xinyang normal university , + xinyang , henan 464000 , p. r. china _ + _ @xmath9department of physics and ipap , yonsei university , seoul 120 - 749 , korea _ + _ @xmath10state key laboratory of theoretical physics , institute of theoretical physics , + chinese academy of sciences , beijing 100190 , p. r. china _ |
since the discovery of @xmath6 violation in 1964 @xcite , great effort has been placed in understanding the origin and mechanism of this asymmetry . in 1973 ,
kobayashi and maskawa proposed a model where @xmath6 violation is accommodated within the weak interaction as an irreducible complex phase in the quark mixing matrix @xcite .
today we refer to this as the ckm matrix and with three known fermion families it can be written as @xmath12 the unitary nature of the matrix implies there are six orthogonality conditions obtained by row - column multiplication with its inverse .
one of these conditions which is relevant to @xmath7 decays is : @xmath13 each orthogonality condition requires the sum of three complex numbers to be zero , and therefore can be represented geometrically in the complex plane as a triangle .
these are known as ` unitary triangles ' , and for the condition expressed in equation [ buni ] , we refer to the angles in the triangle as @xmath10 = @xmath14 , @xmath15 = @xmath16 and @xmath17 = @xmath14 .
two @xmath7-factories have been built , babar at slac usa and belle at kek japan , with the aim of measuring @xmath6 violation in the @xmath7 system and to constrain further the ckm matrix using information from @xmath7 meson decays .
we can write the time dependent amplitude of the @xmath18 state at @xmath19 = 0 decaying into a final state @xmath20 as @xmath21,\ ] ] where @xmath22 = @xmath23 is the amplitude of the @xmath18 decay to the final state @xmath20 , @xmath24 is the amplitude for @xmath25 and @xmath26 give the weak eigenstates @xmath27 in the ( @xmath18 , @xmath25 ) basis @xmath28 . the average between the h and l masses
is @xmath29 and the difference is @xmath30 , whilst we make the approximation @xmath31 .
when both @xmath18 and @xmath25 can decay to the same final state @xmath20 , @xmath6 violation can occur through the interference between mixing ( @xmath32 ) and the decay ( @xmath33 ) , even if @xmath6 is conserved in both @xmath34 .
pairs of @xmath7 mesons are generated at the @xmath35 resonance where the two mesons oscillate coherently between the @xmath18 and @xmath25 until one decays .
we consider the time interval @xmath36 between the decay of one @xmath7 flavor eigenstate ( or `` tag '' ) and the decay of the other @xmath7 to a @xmath6 eigenstate @xmath20 where the decay rate in terms of @xmath36 can be written as @xmath37 where @xmath38 and it is assumed @xmath39 and the @xmath40 sign represents the @xmath18(@xmath25 ) tag . in the @xmath35 center of mass frame the @xmath7 meson is nearly at rest , leading to a strong correlation between the reconstructed mass and the missing mass of the partner @xmath7 .
therefore the typical choice for an independent pair of kinematic variables is @xmath41 where the asterisk refers to the @xmath35 frame and the subscript @xmath7 denotes the reconstructed @xmath7 .
true @xmath7 decays tend to be peak at @xmath42 = 0 and @xmath43 = @xmath44 .
the tagged @xmath7 does not need to be fully reconstructed as the important information required is simply the decay vertex and whether it s a @xmath18 or @xmath25 .
the flavor of the tagged @xmath7 is determined by the sign of the charge in the ( @xmath45 ) , leading charged track .
the efficiency @xmath46 and the mistag fraction @xmath47 for each tagging algorithm is measured with fully reconstructed eigenstate decays .
the effective efficiency is given by @xmath48 , where the babar and belle experiments report average measured values of ( @xmath49)% and ( @xmath50)% respectively .
the decay rate as a function of @xmath36 is modified as a consequence of the tagging efficiency and also the convolution with the resolution function @xmath51 .
the resulting decay rate becomes @xmath52 where @xmath53 is the @xmath7 lifetime .
we look for asymmetries between @xmath18 and @xmath25 tagged events having a @xmath54 ) dependence with a known angular frequency @xmath30 and an amplitude that is given by @xmath55 multiplied by the dilution factor @xmath56 .
the interesting physics lies within @xmath55 as it contains the factor @xmath32 which is common to all decay modes . depending on the decay mode under study , it is @xmath55 which is measured to help provide information on @xmath10 , @xmath15 or @xmath57 .
considering decays @xmath60 ( @xmath58 or @xmath59 ) , the time dependent decay rate can be further written as @xmath61,\ ] ] where the @xmath6 violating asymmetry parameters @xmath62 and @xmath63 and @xmath64 will be used throughout the document with subscripts referring to the decay in question . ]
(= @xmath65 ) are expressed as @xmath63 = ( 1 @xmath66)/(1 + @xmath67 ) and @xmath62 = 2im@xmath68/(1 + @xmath67 ) , where @xmath68 is a parameter that depends on @xmath69 mixing and the amplitudes for @xmath18 and @xmath70 decaying to @xmath71 . if these decays proceed only via @xmath72 tree amplitudes , @xmath73 and @xmath74
. if there are penguin contributions to the amplitude then @xmath62 = @xmath75 and @xmath76 .
a limit on the difference between @xmath10 and @xmath77 can be set using the grossman - quinn bound @xcite which is @xmath78 where @xmath79 is the branching fraction of @xmath80 and @xmath81 is the branching fraction of @xmath82 .
based on a dataset of 81 fb@xmath83 and 140 fb@xmath83 used by the babar and belle experiments respectively , the results in table [ tab : pipi ] where obtained for the decay @xmath9 .
._babar and belle measurements of the @xmath6 violating parameters @xmath84 and @xmath85_. [ cols="^,^,^ " , ]
given the semileptonic decay @xmath86 one can reconstruct the neutrino where then it is possible to measure the hadronic mass @xmath87 @xcite . in the case where @xmath88 the hadronic mass is greater than or equal to the mass of the @xmath89 meson .
however when @xmath90 the hadronic mass distribution extends below the @xmath89 meson mass such that a cut @xmath91 significantly enhances the relative contribution @xmath72 compared to @xmath92 . a measurement of the rate can then be used directly to measure @xmath3 or @xmath4 .
this approach was first used by cleo @xcite and is now being used by babar @xcite and belle @xcite .
we can study the moments of the invariant mass distributions , @xmath93 and @xmath94 where the moments are extracted directly from the measured @xmath87 and @xmath95 distributions directly , taking into account corrections for mass scales , the detector efficiency and small residual backgrounds .
theoretical calculations of the second moment @xmath94 have been performed using an operator product expansion ( ope ) in powers of the strong coupling constant @xmath96 and @xmath97 up to the order of @xmath98 . a review on this subject
can be found here @xcite .
combining measurements of @xmath94 for values of the momentum threshold with other measurements of semileptonic branching fractions and @xmath7 lifetimes we can further constrain the @xmath99 quark mass , @xmath100 in the kinematic mass scheme @xcite and the ckm matrix element @xmath101 @xcite .
based on a dataset of 82 fb@xmath83 corresponding to 89 million @xmath7 meson pairs the following results where obtained .
@xmath101 = ( 41.4 @xmath102 0.4@xmath103 @xmath102 0.4@xmath104 @xmath102 0.2@xmath105 @xmath102 0.6@xmath106 @xmath100 = 4.61 @xmath102 0.05@xmath103 @xmath102 0.04@xmath104 @xmath102 0.02@xmath105 gev / c@xmath107 where in both cases the first error is due to the experimental uncertainty of the measurement , the second is a consequence of heavy quark expansion ( hqe ) and the third is due to the uncertainty contribution from @xmath108 . for @xmath101
the last error is due to the uncertainty from the hqe of the semileptonic width @xmath109 .
babar use a data sample of 82 fb@xmath83 containing 89 million @xmath110 pairs . one @xmath7 is fully reconstructed and the other @xmath7 is analyzed for semileptonic decays .
for semileptonic decay candidates , a high momentum lepton is required such that @xmath111 1.0 gev / c .
all the reconstructed particles not associated with the fully reconstructed @xmath7 are used to calculate the missing four - momentum of the semileptonic decay . this missing four - momentum
is taken as being the neutrino four - momentum .
a kinematic fit is performed on the event to determine @xmath87 .
additional cuts are then applied to improve the quality of events , such as requiring the invariant mass of the neutrino to be small in order to improve the @xmath87 resolution .
a fit is performed on the @xmath87 distribution whereby the region @xmath112 1.55 gev / c@xmath107 is dominated by @xmath113 and is used to fix the scale for the @xmath92 distribution in @xmath87 .
the number of @xmath114 events below @xmath115 1.55 gev / c@xmath107 is then extracted from the fit from which they calculate : @xmath116 = ( 4.62 @xmath102 0.28 @xmath102 0.27 @xmath102 0.40 @xmath102 0.26 ) @xmath117 where the first error is statistical , the second is systematic , the third is due to the error on @xmath114 modeling and the fourth is due to the error on the known relationship between the rate and @xmath116 .
belle follow a similar procedure to that descried above .
they require one fully reconstructed @xmath7 in the events and look for the other @xmath7 to decay semileptonically . with the knowledge of which particles belong to the fully reconstructed @xmath7 and which belong to the semileptonic decay , the values for @xmath87 and @xmath118 ( mass of the virtual @xmath119 squared ) are calculated .
candidates are required to have @xmath120 1.5 gev / c@xmath107 and @xmath121 7 gev / c@xmath107 .
the @xmath92 events are subtracted and from the @xmath114 events in the @xmath87 distribution they find @xmath116 = ( 4.66 @xmath102 0.28 @xmath102 0.35 @xmath102 0.58 @xmath102 0.17 @xmath102 0.08 ) @xmath117 where the first error is statistical , the second is systematic , the third is theoretical and the fourth and fifth errors are due to @xmath92 and @xmath72 modeling respectively .
babar and belle have established @xmath6 violation in @xmath18 decays through the measurement of @xmath122 , where the constraint on the ckm unitarity triangle is shown in figure [ fig : tri ] .
this @xmath6 asymmetry is well accommodated within the standard model , however we are starting to see evidence for inconsistencies from charmless decays .
also , @xmath6 violation in the decay @xmath9 has been reported by belle , though not yet confirmed by babar .
we look forward to additional data which will provide definitive results regarding these measurement , helping to further constrain the angles @xmath10 and @xmath17 and with the help of theorists provide more precise measurements of the ckm elements @xmath3 and @xmath4 .
i am extremely grateful to the organizers of the conference and would like to thank them for providing an interesting and enlightening physics program .
it is also a pleasure to thank the babar and belle experiments for their achievements in obtaining high luminosity with which they have provided a wide range of physics results .
christenson _ et al _ , phys .
* 13 * , 138 ( 1964 ) .
m. kobayashi and t. maskawa , prog .
. phys . * 49 * , 652 ( 1973 ) .
y. grossman and h.r .
quinn , phys .
d * 58 * , 017504 ( 1998 ) .
m. benecke and m. neubert , hep - ph/0308039 ( 2003 ) .
i.dunietz , phys . lett .
b * 427 * , 179 ( 1998 ) .
l. wolfenstein , phys .
lett . * 51 * , 1945 ( 1983 ) .
d. cronin - hennessy _ et al _ ( cleo collaboration ) , phys .
rev . lett . *
87 * , 251808 ( 2001 ) .
scott e. roberts , ph.d thesis , univ . of rochester ( 1997 ) .
b. aubert _ et al _ ( babar collaboration ) , hep - ph/0307062 . h. kakuno ( belle collaboration ) , cipanp2003 .
n. uraltsev , hep - ph/0409125 ( 2004 ) .
i. i. bigi , m. shiffman , n. uraltsev and a. vainshtein , phys .
d * 56 * , 4017 ( 1997 ) .
p. gambino and n. uraltsev , eur .
j. c * 34 * , 181 ( 2004 ) . | results by the babar and belle experiments of the measurements @xmath0 , @xmath1 and @xmath2 along with the ckm matrix elements @xmath3 and @xmath4 are presented .
the babar and belle experiments @xmath5 results for @xmath1 are in good agreement with each other , comfortably establishing @xmath6 violation within @xmath7 decays .
however , there is a 3.5 standard deviation between this result and @xmath1 measured by the belle experiment using @xmath8 decays .
belle also find evidence for @xmath6 violation through time dependent measurements of the decay @xmath9 , whilst a tighter constraint has been placed on the unitarity angle @xmath10 by babar , using time dependent studies of the decay @xmath11 .
= 14.5pt |
the jet in m87 has been observed in many wavelength bands , yet some puzzles remain about the nature of its emissivity . polarization observations in both the optical and radio have shown that the emission from the jet at wavelengths longward of @xmath6 is most likely of synchrotron origin ( @xcite , @xcite ) .
collimated structure can be traced back to @xmath7 from the core using vlbi ( @xcite ) , and terminates @xmath8 away from the core ( 2 kpc at the assumed distance of 17 mpc ) in the western radio lobe , with optical emission still detectable at this distance .
the jet looks very similar at optical and radio wavelengths , although sparks , biretta , & macchetto ( 1996 , hereafter @xcite ) showed that differences ( e.g. , in the transverse brightness profiles ) do exist .
far from having a smooth appearance , the jet exhibits a series of bright knots at intervals roughly @xmath9 apart .
the nature of the knots is uncertain , but they are usually attributed to internal shocks from either flow instabilities ( bicknell & begelman 1996 , hereafter @xcite ) or variable outflow at the source ( @xcite ) .
both the knots and the interknot regions exhibit power law spectra of index @xmath10 in the radio , connecting to the optical data with a power law of index @xmath11 , and steepening to @xmath12 between @xmath13 and @xmath14 in the optical .
this steepening trend is also found in observations at infrared ( e.g. , @xcite ) and ultraviolet ( @xcite ) wavelengths , as confirmed recently by _
hst _ observations ( @xcite , @xcite ) .
x - ray observations made with _ einstein observatory _ ( @xcite ) and _ rosat _ ( @xcite and @xcite ) reveal x - ray emission from several spots along the jet ( mainly the core and knot a , possibly also from knots d and b ) .
however , the origin of the x - ray emission is unknown and it is not clear whether the spectrum breaks between optical and x - ray wavelengths to a spectral index of @xmath15 , with the x - ray emission still being of synchrotron origin , or whether the x - ray emission is produced by a different mechanism , e.g. , inverse compton scattering or bremsstrahlung . the onesidedness of the m87 jet has been interpreted as the result of relativistic beaming of the emission from an intrinsically bipolar jet , which implies bulk velocities corresponding to lorentz factors of @xmath16 with line of sight inclinations of @xmath17 .
this would explain the existence of the radio lobes on both sides of the core .
new infrared detections of counterjet emission ( stiavelli , peletier , & carollo 1997 ) seem to confirm this hypothesis .
the doppler beaming interpretation is bolstered by proper motion measurements of the knots ( @xcite ) , which show characteristic velocities @xmath18 in knots a and b , with some features exhibiting much larger proper velocities ( a subfeature of knot d appears to show superluminal motion ) .
these motions broadly support the interpretation of the knots as relatively weak , _ oblique _ shocks moving down the jet with pattern speeds significantly smaller than the bulk speed of the flow ( @xcite ) .
one might hope to detect the effects of synchrotron cooling and relativistic particle acceleration by studying the spectrum as a function of position along the jet .
such measurements ( @xcite ) show that the radio to optical spectral index , @xmath19 , is very nearly uniform , while the optical spectral index , @xmath20 , is anti correlated with the brightness , i.e. , the optical spectrum is flatter in regions of higher intensity . modeling the optical steepening as a high - energy cutoff imposed on a power law spectrum , neumann , meisenheimer , & rser ( 1997 , hereafter @xcite ) and @xcite find a corresponding correlation between brightness and cutoff frequency , i.e. , a higher frequency at higher intensities .
both these results are striking in the lack of a strong secular decline in the cutoff frequency with distance from the core , as would be expected naively if the steepening were due to synchrotron cooling . indeed
, these observational results do not compare well with simple quantitative models of synchrotron cooling in the m87 jet .
the usual assumption of equipartition leads to estimates of the magnetic field of order @xmath21 , with values up to @xmath22 in knot a. the synchrotron lifetime for electrons with lorentz factors of @xmath23 , needed to produce the optical emission in a @xmath24 field , is only @xmath25 , which , for mildly relativistic bulk velocities ( e.g. , @xmath26 ) implies a travel distance of less than @xmath27 . but
the projected length of the jet is about @xmath28 , and even the distances between the most prominent knots are longer than the estimated cooling length . yet
the spectrum between radio and optical bands remains remarkably constant along the jet , with only minor variations in the optical spectral index .
this presents a paradox : after a travel distance of @xmath29 one would expect the cutoff to have moved to below @xmath30 , a factor of @xmath31 smaller than what is actually observed . the discrepancy becomes worse if the magnetic field is stronger than the equipartition value , as suggested by owen , hardee , & cornwell ( 1989 ) in order to explain the confinement of the overpressured jet via the magnetic tension force .
( an alternate explanation for the confinement that the radio cocoon surrounding the jet is overpressured with respect to its surroundings has been proposed by @xcite . )
three explanations for the discrepancy have appeared in the literature : 1 .
first - order fermi re acceleration in the knots , interpreted as shocks , could produce high energy electrons ( and possibly positrons ) from a synchrotron - cooled distribution , with a power law of roughly the index observed .
it could also explain the observed x - ray emission .
however , since the power - law index produced by fermi acceleration is a strong function of the compression ratio one would need some fine tuning to explain the observed constancy of the radio to optical spectral index , which does not equal the limiting value for a strong adiabatic shock .
furthermore , there does not seem to be a significant amount of cooling between the knots even at the highest optical frequencies observed , which would be expected for the assumed @xmath32-fields and interknot distances .
particles could be transported in a loss - free channel in the interior of the jet , with the bulk of the emission produced in a thin outer layer of high magnetic field strength ( @xcite ) . in this picture , the knots and filaments would be interpreted as instabilities with greatly increased magnetic field strength wrapped around the jet . the emission would then be fed by particles from this channel .
the radio brightness profiles across the jet seem to suggest a limb - brightened emission , but a reinvestigation of the hst observations ( @xcite ) shows that the optical emission is more concentrated to the inner regions of the jet . also , @xcite s deconvolution places the brightest spots in the jet interior .
this argues against a field - free zone in the jet interior .
3 . on the spot reacceleration by a yet unknown process
could maintain the cut - off particle momentum at the observed level , as has been proposed by meisenheimer et al .
( 1996 ) in a model similar to ours ( see [ sec : meismod ] ) .
this has the advantage of explaining all the observed features , but invokes unknown physics to explain the apparent lack of cooling .
inspired by the observed correlation between the emissivity variations and the cut - off frequency , by the newly deconvolved volume emissivity ( @xcite ) , and by the new evidence for relativistic bulk velocities ( @xcite ) , we propose a simple way of explaining the observations .
the only standard assumption we give up is the assumption of equipartition , which does not seem to have a very firm physical foundation anyway .
magnetic fields _ smaller _ than equipartition by a factor of @xmath33 , coupled with bulk lorentz factors in excess of @xmath34 , can readily explain the lack of evidence for synchrotron cooling . in our model the fluctuations of the cutoff frequency
are produced by weak shocks , so that the influence of the compressions on the plasma distribution function can be considered to be adiabatic . as a result , we are able to explain the general behavior of the cutoff reasonably well ; the magnetic fields we derive from our own fits to the data are below equipartition with a convincing level of confidence , but the inferred total pressures do not necessarily need to exceed the equipartition values .
additionally , for relativistic jets the equipartition values for @xmath32-field and pressure are less than in the nonrelativistic limit , so the pressures we derive can fall below the equipartition value in the nonrelativistic case .
we organize the paper as follows . in
[ sec - adiabatic ] we present a nonrelativistic treatment of the synchrotron emissivity , taking into account cooling but assuming that the particles and fields respond adiabatically to changes in the flow density .
once relativistic effects are incorporated into the treatment in
[ sec - relativistic ] , we use the data of @xcite and @xcite to constrain the magnetic field strength of the jet .
section [ sec - discussion ] discusses confinement and stability of the jet in the light of the pressures derived from the results of [ sec - relativistic ] , the production of x - ray emission in knot a , polarization , and limits on the particle acceleration site ; and [ sec - conclusion ] gives a brief summary of the results and future prospects .
our model rests on the hypothesis that fermi acceleration is unnecessary to explain the fluctuations of radio to optical emissivity and cutoff frequency along the m87 jet .
given certain assumptions about the orientation and degree of disorder in the magnetic field , and the degree of anisotropy permitted in the relativistic electron distribution , we can relate changes in both the emissivity and the cutoff frequency uniquely to changes in the density of the jet fluid .
these adiabatic effects are readily combined with the effects of synchrotron cooling ( @xcite ) . in effect , given the emissivity map of @xcite , we can _ predict _ the run of the cutoff frequency along the jet , and vice - versa . since we also have nmr s observations of the cutoff frequency as a function of position ,
our adiabatic model is subject to a powerful self - consistency check .
observationally , the main changes in emissivity and cutoff frequency are rather localized , and associated with the positions of the knots .
these small - scale fluctuations therefore provide the strongest check on our assumption of adiabaticity .
the _ large - scale _ trends then determine the best fit to the magnetic field strength .
we have already seen that the apparent lack of a large - scale synchrotron cooling trend is incompatible with a magnetic field strength as large as the mean equipartition value ( see fig .
[ fig : freqfit ] ) . neglecting relativistic and projection effects
, it would require a magnetic field as low as @xmath35 to obtain a cooling length of @xmath29 at optical frequencies .
this is an order of magnitude smaller than the mean equipartition field and would require a total ( particle + magnetic ) pressure of @xmath36 to produce the observed average amount of synchrotron emission , compared to an equipartition value of @xmath37 .
as noted earlier , the physical basis of equipartition is weak .
estimates of @xmath32-fields in radio hot spots based on the synchrotron cooling time indicate that in some cases equipartition might be correct up to a factor of @xmath38 ( @xcite ) , but the conditions in the jet might very well be different from those in the lobes .
we are therefore free to consider the magnetic field strength to be a free parameter . applying our adiabatic model to the observational data
, we can derive an estimate for the magnetic field strength . in the case of the m87 jet ,
this estimate lies _ below _ equipartition , even when relativistic effects are taken into account ( [ sec - relativistic ] ) .
our model for the evolution of the particle distribution function follows that of coleman & bicknell ( 1988 ) .
we assume that pitch angle scattering due to plasma micro instabilities keeps the particles close to an isotropic distribution in the fluid rest frame . in the absence of cooling
, this would imply that the relativistic electrons respond to compressions like a @xmath39 ( i.e. , ultrarelativistic ) fluid , but this behavior will be modified by synchrotron cooling . because the magnetic field is frozen into the plasma , its strength should change as the plasma density fluctuates along the jet . depending on the orientation and the degree of disorder of the field ,
its variation will depend roughly on the density change to some power @xmath40 : @xmath41 , where @xmath42 is the proper particle density and @xmath43 is the distance from the core . for a completely disordered magnetic field , @xmath44 , whereas for a homogeneous field the power depends on the orientation of the field with respect to the compression normal , with @xmath45 for an orthogonal orientation , @xmath46 for parallel orientation .
since the polarization of the jet is of order 10%-20% in the interknot regions , compared to the maximum polarization of 70% for a homogeneous field , it is likely that the magnetic field has a disordered component , so it is reasonable to assume @xmath47 to be of order @xmath48 ( but see [ sec : polar ] ) .
the ordered @xmath32-component is aligned with the jet axis almost everywhere except in the brightest knots , as can be seen from polarization measurements ( @xcite ) .
we do not know its orientation with respect to the compression normal , because the orientations of the ( presumably oblique : see @xcite ) shocks are unknown . however , our results are not very sensitive to what the actual value of @xmath47 is , as we will show later .
we therefore make the simplifying assumption of a single exponent describing the field variations . using the scaling relation for the synchrotron emissivity of a power law momentum distribution ( @xmath49 ,
corresponding to a spectral index of @xmath50 ) under adiabatic compression ( e.g. , @xcite ) : @xmath51 where @xmath52 is the angle between magnetic field and line of sight , we can express the field relative to its value at @xmath53 , the ( arbitrary ) injection point at which we start the calculation , as a function of the emissivity ratio @xmath54 : @xmath55 where @xmath56^{-1}$ ] .
a necessary condition for this approach to be valid ( in addition to the assumed isotropy and the absence of fermi acceleration ) is the assumed steady - state injection of relativistic particles and fields by the central engine , which allows us to relate densities and fields at each @xmath43 to the corresponding values at the injection point @xmath57 at a given instant of time .
is it plausible to neglect fermi acceleration in the shocks that comprise the knots ? except for knot a , the brightness changes along the jet are moderate . in knot
a , the brightest feature , the emissivity changes by a factor of order @xmath58 ( @xcite ) , which , if entirely due to a sudden compression of the plasma , can be produced by a proper compression ratio of @xmath59 as measured in the respective rest frames of the plasma if we take @xmath60 . for the other knots we infer smaller density contrasts , @xmath61 .
consistent with this observation , we will henceforth take the knots to be weak shocks , in accordance with the suggestion by @xcite that the knots are highly oblique ( and therefore weak ) shocks .
thus , because fermi acceleration leaves the spectral index unchanged if the shock is weak enough , we will henceforth neglect its effect on the cutoff frequency .
section [ sec - discussion ] discusses fermi acceleration in more detail , with particular focus on the possibility of fermi acceleration occurring in knot a. furthermore , because the shocks are believed to be oblique , we take the fluid velocity to be constant to first order , both in magnitude and direction . for the shock jump conditions in the non relativistic limit ( which we consider in this section ) the velocity component perpendicular to the shock plane @xmath62 is inversely proportional to the density , thus for a proper compression ratio of @xmath63 , as seen in knot a , the perpendicular velocity component should change by a factor of @xmath64 .
for highly oblique shocks , @xmath62 is small compared to @xmath65 and the velocity will not change significantly .
moderate changes in velocity would be easy to incorporate in principle ; yet with our current ignorance of the velocity field and shock parameters , such a level of detail is unwarranted .
we will comment on the validity of this assumption in [ sec - relativistic ] .
we also postpone a treatment of the motion of the knots until [ sec - relativistic ] , and assume them to be stationary for the rest of this section . with these assumptions we are ready to calculate the downstream cutoff frequency for a given initial cutoff momentum in the injected particle distribution and a given @xmath32-field at @xmath66 .
we use the transport equation as presented by coleman & bicknell ( 1988 ) : @xmath67 written in the rest frame of the fluid . here
, @xmath68 is the electron distribution function and @xmath69 is the synchrotron loss term .
the equation is valid for the assumed case of isotropy and negligible compton losses ( for a brief discussion of compton losses see [ sec - discussion - confinement ] ) .
the solution of this equation is @xmath70 where @xmath71 is the injected momentum distribution and @xmath72 ( see @xcite ) where the subscript 0 denotes the values at injection point @xmath66
. equation ( [ eq : solution ] ) describes how the momentum of a given particle changes along a streamline .
thus , if the distribution initially cuts off at @xmath73 , we can calculate the cutoff momentum @xmath74 downstream .
because in our model the density @xmath75 is proportional to @xmath76 and because we know the scaling of @xmath32 with @xmath43 from equation ( [ eq : magnetic ] ) , we can eliminate @xmath42 and @xmath77 from equation ( [ eq : solution ] ) .
the remaining parameters are @xmath78 , @xmath73 , and @xmath79 , the latter being provided by the sparks et al
. data .
the cutoff momentum @xmath80 is related to the observed cutoff frequency @xmath81 by the expression @xmath82 equation ( [ eq : critical ] ) contains another parameter , @xmath83 , the angle between the line of sight and the magnetic field .
for now we shall set the factor @xmath84 , which is valid for disordered fields , since 1 ) the regions in which the field is perpendicular to the line of sight have the highest emissivity , and 2 ) assuming randomly oriented fields , half of the field orientations lie in the range from @xmath85 to @xmath86 to the line of sight , i.e. , @xmath87 .
thus the cutoff frequency is mainly determined by field orientations close to @xmath86 or @xmath88 .
we can now determine the free parameters @xmath89 and @xmath73 by applying a least chi - squared method to fit the observed cutoff frequency @xmath90 with the value determined from equations ( [ eq : solution ] ) and ( [ eq : critical ] ) , using the emissivity map @xmath91 provided by @xcite .
we prefer to average the emissivity across the jet , which minimizes small scale variations probably due to the deprojection procedure .
because @xcite also averaged across the jet , this seems to be the most appropriate way of calculating the cutoff frequency .
figure [ fig : freqfit ] ( calculated for a bulk lorentz factor of @xmath92 , a radio spectral index of @xmath93 , @xmath94 , and @xmath44 ) shows the observed cutoff frequency @xmath90 ( vertical bars ) with error bars and the best fit curve ( solid line ) , which seems to reproduce the scaling of the cutoff frequency reasonably well .
( the radio spectral index seems to break to @xmath95 at @xmath96 ghz , so we have used both @xmath97 and @xmath98 in our fits with insignificant differences in the average parameters but smaller chi - squared for @xmath99 ; see [ sec - relativistic ] . ) for comparison the plot also shows the best fit cutoff frequency for equipartition @xmath32-fields ( dashed line ) .
the mean @xmath32-field is of order @xmath100 , even smaller than the zeroth order estimate made at the beginning of this section .
assuming ( arbitrarily ) a lower cutoff at @xmath101 and the observed high - frequency cutoff at @xmath102 yields an average total pressure of @xmath103 for the given parameters , compared to an equipartition value of @xmath104 . in calculating absolute values for both pressure and @xmath32-field , projection ( i.e. , foreshortening and length scale ) effects
must be taken into account , since the emissivity was derived for a side - on view of the jet .
this introduces a factor of @xmath105 in intrinsic emissivity and pressure for a given magnetic field and a factor of @xmath106 in equipartition pressure .
it seems that the proposed modest compressions can account for the fluctuations seen in the spectral cutoff , and the large scale decrease in @xmath81 is well reproduced
. it would be helpful to determine both the emissivity and the spectral index maps from the same method and data , thus eliminating errors due to different reduction procedures .
we comment on the deviations and uncertainties in this fit in
[ sec - relativistic ] .
it is important to note that this technique should be independent of what the actual shape of the particle distribution is , because it simply tracks the behavior of a single feature in the spectrum , which could be identified with either a break or a cutoff .
as seen in this section , one runs into problems with the jet pressure for nonrelativistic bulk velocities .
also , the observed mildly relativistic proper motion of the knots and the jet s onesidedness favor a relativistic interpretation , as do the knot spacing and morphology ( @xcite ) .
the results indicate that lorentz factors of order @xmath107 fit the observations best . in the next section
we will investigate the effects of these suggested relativistic bulk velocities .
relativistic motions not only explain the onesidedness of the m87 jet , but also help to solve the synchrotron cooling problem mentioned in the introduction .
the travel time in the electron rest frame is reduced by a factor @xmath1 due to time dilation , the intrinsic emissivity is reduced by a factor @xmath108 ( where @xmath109 is the doppler factor @xmath110^{-1}$ ] ) , and the intrinsic cutoff frequency is doppler shifted downward by a factor @xmath109 . as a result , the apparent synchrotron lifetime can be a significant underestimate of the intrinsic value .
biretta ( 1993 ) estimates the lower limit on the jet to counterjet radio brightness ratio to be @xmath111 ( the higher value corresponds to the assumption that jet and counterjet have identical appearance ) .
based on this limit we adopt line of sight angles @xmath112 and lorentz factors @xmath113 . note that even doppler factors smaller than unity can lead to a large jet to counterjet brightness ratio , as the counterjet brightness is severely reduced for large @xmath1 .
we repeat the analysis of
[ sec - adiabatic ] using the emissivity profile of @xcite , this time corrected for doppler boosting and projection effects .
again , for fitting the cutoff frequency only emissivity ratios are important , so these corrections do not change the fitting procedure as long as changes in the bulk velocity can be neglected .
we are confident that at least the direction of the flow is not changed significantly before knot c. the only knot for which such effects could be important is knot a , because it displays a jump in emissivity of @xmath114 , whereas in the other knots the emissivity is increased by a factor of order @xmath115 only , which implies very moderate compressions . using the relativistic continuity equation for an oblique shock we have estimated the post knot a lorentz factor to be @xmath116 for a pre knot a @xmath117 , a compression ratio of 3 and intrinsic obliquities @xmath118 .
although the change in @xmath119 might seem large at first glance , the impact such a velocity change has on our fits is not large , as we will explain below .
we modify equation ( [ eq : transport ] ) to follow the electron distribution in the _ fluid rest frame _ by replacing @xmath120 with @xmath121 and @xmath122 with @xmath123 , where @xmath124 is the proper time .
the same changes apply to equation ( [ eq : solution ] ) . strictly speaking , the fluid frame is not an inertial frame and we would have to include accelerational terms into the equation , introducing an anisotropy .
but our assumption should be adequate , provided that isotropization takes place over short enough scales .
treating the response of the distribution function to compressions as adiabatic and assuming isotropy ( i.e. , an adiabatic index of @xmath125 ) we calculate the changes in @xmath32 and cutoff momentum ( measured in the fluid frame ) from the emissivity changes in the fluid frame .
the substitution @xmath126 takes care of the time dilation effects . in order to incorporate the observed motion of the knots , we must correct @xmath127 and @xmath128 in equation ( [ eq : solution ] ) for light travel time effects between the source and the observer
this is because the knots move during the time it takes a particle to travel from @xmath129 to @xmath43 .
we need to know the ratios @xmath130 and @xmath131 experienced by a particle as a function of _ proper time _
@xmath132 in order to be able to do the integration in equation ( [ eq : solution ] ) . because we infer the relative values of @xmath32 and @xmath42 at a given position from the emissivity ratio @xmath133 , it is important to know the velocity of the emissivity pattern .
we assume that the pattern of density and field fluctuations retains its shape and moves along the jet at a fixed speed @xmath134 , taken to be smaller than @xmath135 and set equal to @xmath136 everywhere in our calculations for simplicity .
if the present field distribution ( i.e. , at a given time @xmath137 in our frame , corrected for light travel time effects ) is expressed by @xmath138 , then the field at time @xmath139 and position @xmath140 is given by @xmath141 .
now , for a particle currently at @xmath43 , the equation of motion is @xmath142 .
therefore , the field distribution experienced by the particle as a function of time is @xmath143 $ ] .
appropriate modifications to equation ( [ eq : solution ] ) are straightforward .
the effect of the pattern speed on the result is not very dramatic , reducing @xmath144 by about @xmath145 . with this set of assumptions
we can once again proceed to integrate the modified equation ( [ eq : solution ] ) for various @xmath146 and @xmath147 . using a minimum chi - square routine
we can determine the best fit values for @xmath148 , and @xmath149 .
relation ( [ eq : magnetic ] ) then yields @xmath138 .
figure [ fig : contourplot ] shows a chi - square plot for @xmath150 , @xmath151 , and @xmath44 .
the equipartition value for the average @xmath32-field is shown as a shaded area at @xmath152 .
the upper limit on @xmath153 , set by @xmath154 contours , lies at @xmath155 , 75% above the best fit value , @xmath156 .
the lower limit set by @xmath157 is @xmath158 , 80% below the best fit value .
the average equipartition field of @xmath159 lies above even the @xmath160 contour .
the lower limit on @xmath32 is not nearly as strict , due to the fact that cooling is not dominant , i.e. , we can produce a similar spectral behavior by reducing the magnetic field and increasing the particle energy , which produces the tear - shaped appeareance of the contours .
thus , strictly speaking , the best fit values for the @xmath32-field should be regarded as upper limits . the reduced @xmath144 values ( i.e. , @xmath144 divided by the number of degrees of freedom ) fall above 44 , which is uncomfortably high . however , because we do not have formal errors for the emissivity deprojection by @xcite , which will introduce a significant uncertainty , a high value for @xmath144 is not all that discouraging .
estimating the average uncertainty in the emissivity by comparing the averaged emissivity to that derived from taking only a slice along the jet yields an uncertainty of order 50% , which leads to uncertainties in the predicted cutoff frequency of roughly 20% .
this is significantly higher than the formal error in @xcite s data and will reduce the @xmath144 by a factor of approximately 10 .
the @xmath144 values are dominated by the region beyond knot a. the post knot a residuals in our fit are not larger than the residuals in the pre knot a region , but because the post knot a region is brighter , the error bars on the measured cutoff frequency are smaller , which increases the @xmath161 . the deprojection procedure , which assumed an axially symmetric flow , breaks down beyond knot a , which will introduce significant uncertainty .
also , non uniformities in the emissivity could lead to large errors if the optical emission peaks at different locations than does the radio emission . field orientation effects and changes in @xmath1 and @xmath162 might also contribute to the error .
we performed the same procedure just out to knot a and found that , with the _ same _ parameters , the reduced @xmath161 shrinks to 13 . leaving @xmath32 and @xmath163 as _ free _ parameters reduces @xmath161 to 10 , but also reduces the @xmath32-field significantly .
because in this case the algorithm mainly fits the region around knot a ( where the error bars are smallest ) , we can not expect the global run of @xmath164 to have significant impact on the fit , which would be necessary to extract information about the average magnetic field .
we conclude that the reproduction of fine detail is not satisfactory in the region beyond knot a. however , the gross run of @xmath164 , which is principally responsible for constraining our parameters , is reasonably well reproduced .
the best fit average magnetic field @xmath153 , as a function of @xmath1 and @xmath162 , is plotted in figure [ fig : bmean ] .
figure [ fig : brels1p5 ] shows the average ratio @xmath165 as a function of @xmath147 for @xmath166 to @xmath167 in increments of @xmath168 , and the area corresponding to the limiting jet to counterjet brightness ratio of 150 - 380 .
note that the equipartition magnetic field has to be corrected by a factor of @xmath169 for projection and doppler boosting of the emissivity ; this has already been taken into account in the figure . clearly , for @xmath147 in the range @xmath170 and @xmath171 the departure from equipartition is not very large ( roughly a factor of @xmath172 ) .
in order to test the dependence of the best fit @xmath148 on the parameter @xmath173 we have calculated the same curves for @xmath45 and @xmath174 .
figure [ fig : berr ] shows the fractional deviation @xmath175 from the @xmath176 curve for models with @xmath177 .
the deviation is small compared to the expected errors introduced by the simplifications we made and to the range in @xmath32 allowed by our minimum chi - square procedure , at most @xmath178 for @xmath179 and small @xmath1 .
this is not a very reasonable value for @xmath173 in any case , because the field has a random component , thus @xmath173 should be higher , and the probability of the field being in the shock plane ( thus having @xmath45 ) is twice as high as for the field being normal to the shock .
we conclude that our ignorance of the precise behavior of @xmath32 under compression is not a serious obstacle to the application of our model .
we also tested the impact a change in @xmath180 at knot a might have on our results . as we mentioned earlier ,
the best fit @xmath32-field values we derive are upper limits .
this is the reason why a change in @xmath119 at knot a does not change our results significantly : generally , lower lorentz factors require lower fields to explain the observed lack of cooling .
if the jet is slowed down beyond knot a , we will need lower average fields to fit this region . however , lowering the field does not change the quality of the fit much ( the @xmath181 is essentially unchanged ) , so the global field strength is simply set by the region with the lower lorentz factor , @xmath182 ( the relative scaling of @xmath32 is still determined from equation [ eq : magnetic ] , taking relativistic beaming into account ) .
we have introduced by hand a change of @xmath119 at knot a into our model ( we solved the continuity equation at the shock , assuming an obliquity of @xmath85 , for the velocity change that would reproduce the observed emissivity jump of 11 , including relativistic beaming and adiabatic compression ) , and calculated the fractional deviation @xmath183 of the derived averaged @xmath32-field . here , @xmath184 is the best fit average @xmath32-field derived for uniform @xmath119 and @xmath185 is the best fit field for a jet slowing down from @xmath186 to @xmath182 at knot a. figure [ fig : berr ] shows @xmath187 for uniform jet models with @xmath119 set to either @xmath186 or @xmath182 ( filled light and dark grey regions , respectively ) .
the latter is always less than 18% for the parameter range we used .
note that for post knot a @xmath182 s above 5 , the pre knot a @xmath186 exceeds 8.5 , thus @xmath188 can be ruled out on the basis of gross energy balance arguments ( see next section ) . for completeness
we have shown the deviation of the best fit average @xmath32-field @xmath189 for a 2 cm radio spectral index of @xmath190 instead of @xmath191 as the black region in figure [ fig : berr ] .
one can see that the difference is negligible compared to other uncertainties .
in the preceding sections we demonstrated that a ) magnetic fields slightly below equipartition and b ) moderately relativistic effects are able to explain the general behavior of the spectrum in m87 . in this section we will examine the confinement properties of the jet and compare our model with a previous model by meisenheimer et al .
. we also comment on the production of x - ray emission in knot a , and on the consistency of our model with polarization measurements , and we estimate the minimum distance from the core at which particle acceleration has to occur . naturally the question arises whether the jet can be confined under the conditions we proposed above .
the usual assumption for a jet to be confined is that it is in pressure equilibrium with its surroundings .
alternatively , one could imagine the jet to be freely expanding into an underpressured surrounding medium .
@xcite argue that in order to produce shocks via kelvin - helmholtz instability , some interaction between jet and surrounding medium has to take place , as opposed to a free expansion scenario .
they also show that the minimum lorentz factor @xmath1 for a freely expanding jet with no cold matter content is at least @xmath192 , much higher than the values we have used above . because we would most certainly fall out of the beaming cone for such a high @xmath1 jet
, the intrinsic emissivity would be much higher than the observed value .
as @xcite point out , the energy flux of the jet would far exceed the estimates made on the basis of the expanding bubble the m87 jet blows into the ism .
a @xmath1 that high would also raise questions about the location at which the jet is decelerated to nonrelativistic velocities , and seems inconsistent with the claimed detection of ir counterjet emission by stiavelli et al .
we can therefore rule out the picture of a freely expanding jet . as a consequence
we need a mechanism to provide confinement , i.e. , we need to set the jet pressure in relation to the ambient pressure . the ambient gas pressure in the center of m87 has been derived by @xcite from fitting cooling flow models to the _ einstein _
x - ray observations .
the values they find fall into the range @xmath193 to @xmath194 .
it is important to note that the pressure of the interstellar medium in m87 might not be representative of the pressure of the immediate environment of the jet .
in fact , @xcite s analysis of the helical kelvin - helmholtz instability leads to the conclusion that the ambient medium of the jet is significantly overpressured with respect to the interstellar medium in m87 .
note also that the pressure in the knots might well exceed the ambient pressure without losing confinement , as long as the average pressure does not .
we have calculated the average total pressure @xmath195 in the jet from the averaged emissivity and the best
fit @xmath32-field for various angles and lorentz factors , as shown in figure [ fig : logpressure ] .
the assumptions we have made in constructing this plot are analogous to those of @xcite , who ( arbitrarily ) assumed a lower cutoff at @xmath196 , a high energy cutoff at @xmath197 , a spectral index of @xmath198 , and equipartition between heavy particle and electron energy .
( note : because the spectrum is steeper than @xmath199 above 10 ghz , our estimate of the pressure is likely to be an overestimate . )
we have also assumed isotropic emission in the plasma rest frame by using an average value of @xmath200 for the term @xmath201 in the emissivity equation ( [ eq : emissivity ] ) . for comparison
we have also calculated the equipartition pressure and plotted the ratio of pressure to equipartition pressure in figure [ fig : relpressure ] .
it is obvious that we are far above equipartition for small values of @xmath1 and large @xmath162 , but as we approach the favored range of @xmath202 and @xmath203 , @xmath204 approaches the equipartition value .
the exact value of the pressure depends critically on the details we put into the model spectrum . for a jet composed entirely of electrons and positrons , the pressure would go down by a factor of @xmath205 , whereas the equipartition pressure would only decrease by a factor of @xmath206 .
the lack of information about the low frequency spectrum inhibits any statements about the low energy particle distribution .
however , it is safe to assume that the power law does not continue down to non - relativistic energies .
it is obvious that a magnetic field far below equipartition alone can not explain the behavior of the jet it might account for the spectral changes but it requires the pressure to be much higher than that of the surrounding medium .
field orientations close to the line of sight will lead to an underestimate in emissivity , pressure , and intrinsic cutoff frequency and will require even lower magnetic fields and even higher pressures in order to prevent significant cooling .
this changes as we increase @xmath1 : the inferred pressures are close to the value for the interstellar medium in m87 as derived by @xcite , for @xmath207 and @xmath208 .
this , combined with the possible overpressure of the jet s immediate environment relative to the ism , leads to the conclusion that there is no confinement problem . for small values of the magnetic field
, one might ask if inverse compton losses become dominant .
a simple order of magnitude estimate shows that this is not the case .
the ratio of synchrotron to inverse compton loss timescales is equal to the ratio of photon energy density to magnetic field energy density ( @xcite ) .
the magnetic field strength has been estimated above . to derive an estimate of the photon energy density produced by the synchrotron emission , we normalize to the radio luminosity corrected for beaming and projection effects and integrate over an @xmath209 radio to optical power law that cuts off at @xmath210 hz .
this shows that the inverse compton lifetime due to just the synchrotron radiation field of the jet is roughly an order of magnitude longer than the synchrotron lifetime for the parameter range we suggested above .
the starlight background at the center of m87 also contributes to the photon energy density . using an isothermal sphere profile , normalized to the total luminosity of m87
, we arrive at a central photon energy density roughly an order of magnitude smaller than that of the magnetic field , small enough to justify the assumption of negligible compton losses .
it is instructive to compare our model to an earlier _ ad - hoc _ model by meisenheimer et al .
( 1996 , see also @xcite ) , which bears a lot of similarity to our model .
they start from the same assumption that the spectral changes along the jet can be explained by simple compressions and assume that the cutoff momentum @xmath211 is almost constant along the jet , parameterizing it as a function only of the transverse jet radius ( measured from the 2 cm radio map ) : @xmath212 note that for an adiabatic compression of the plasma _ transverse _ to the flow @xmath211 goes as @xmath213 .
they take the @xmath32-field to consist predominantly of a toroidal component , @xmath214 , and hold the poloidal component @xmath215 fixed .
they determine the longitudinal compression ratio of @xmath214 and of the particle density @xmath216 from their fit to the cutoff frequency @xmath164 with equation [ eq : critical ] .
however , in an adiabatic compression , the cutoff momentum varies as @xmath217 and will therefore be affected by longitudinal compressions as well ( here is where our assumption of a disordered field allows us to determine a relation between density and magnetic field , so we can solve equation [ eq : critical ] uniquely for @xmath32 ) .
they neglect the fact that the synchrotron emissivity is enhanced in adiabatic compressions by @xmath218 ( equation [ eq : emissivity ] ) rather than @xmath219 . since the shocks might well be oblique , their assumptions that @xmath214 scales as the longitudinal compression ratio and that @xmath215 is constant might also not be valid .
meisenheimer et al .
( 1996 ) favor an intrinsically onesided , subrelativistic jet , viewed close to perpendicular ( @xmath220 ) .
knot a would be a head on shock in this scenario .
as we have mentioned above , for this set of parameters additional acceleration has to be provided to maintain the optical emission out to large distances from the core .
meisenheimer et al .
( 1996 ) favor an unknown global acceleration process to explain the constancy of the cutoff momentum . with these assumptions , their model yields similar results to ours in that it reproduces the small scale brightness variations on the basis of the changes in cutoff frequency .
our model could thus be regarded as an extension of their approach , putting it on the theoretical basis of adiabatic expansions , with a different mechanism for providing the large scale constancy of the spectrum .
an important result from the analysis above is that cooling longward of uv energies is not important over the length of the jet the proper time is reduced by a significant factor and the magnetic fields are small enough to leave the spectral shape unchanged .
this conclusion begs the question of the origin of the x - ray emission detected by the _
einstein observatory _ and _ rosat_. both observations show emission from the core / knot d region and knot a. ultra - high - energy particles , capable of radiating in the x - ray regime , could be carried out from knot d , where x - ray emission is observed , to knot a , and reaccelerated in the shock by the adiabatic compression mechanism discussed above .
for this to happen the spectrum would have to break rather than cut off in the optical .
the presence of a break instead of a cutoff would not change our fits , as long as the break is located above the frequency we fitted .
we have calculated the behavior of particles with x - ray emitting energies along the jet and found that for our best fit @xmath32-fields cooling out to knot a will have produced a spectral cutoff at @xmath221 hz which is where most of the _ einstein _ hri s sensitivity lies .
since the @xmath32-fields we derived are upper limits , a lower field could leave the distribution function unchanged even at such high energies .
therefore this mechanism of producing x
rays is marginally consistent with our model .
it might also account for at least part of the x rays .
note that , as suggested by various authors ( e.g. , @xcite ) , the x - ray emission could also be of non synchrotron origin altogether .
the compression of a factor @xmath222 inferred from the analysis above indicates that fermi acceleration might be present in knot a , although the moderate change in emissivity and the constancy in radio and optical spectral index suggests that it might not be very efficient .
it is possible that fermi acceleration occurs at parts of the shock only , resulting in a particle distribution superposed of a compressed and a fermi - accelerated preshock distribution . in order for fermi acceleration to take place at all
the shock has to be subluminal , i.e. , the intersection point of a given magnetic field line and the shock front has to move with a speed smaller than the speed of light , in which case we can find a frame in which the magnetic field is perpendicular to the shock front .
this is the case for fields not too closely aligned with the shock plane . in the nonrelativistic case the field orientations leading to a superluminal shock are rare , and subluminal shocks are the rule rather than the exception ,
so one would expect fermi acceleration to take place . because in relativistic shocks the percentage of superluminal field orientations rises sharply with @xmath223
, fermi acceleration should become less important . in this limit
, most of the particle acceleration would occur through the mechanism of `` shock drift acceleration '' .
@xcite presented a theory of this process valid in the relativistic case .
they show that the adiabatic approximation is still accurate in the limit of @xmath224 , where @xmath225 is the upstream lorentz factor in the perpendicular shock frame and @xmath226 the corresponding velocity .
we have calculated this quantity for various obliquities and field orientations appropriate for knot a and it seems to fall into the desired range . for a disordered magnetic field
we would have to average the resulting spectrum over all possible field orientations .
the more superluminal the shock the less important would effects of fermi acceleration be . relativistic corrections to shock drift acceleration are only important for high @xmath227 , which , again , depends on the field orientation .
averaging over all possible field orientations would probably render these corrections unimportant .
fermi acceleration both changes the shape of an incoming power law spectrum and amplifies it . for an incoming electron spectrum of the form @xmath228 in the test particle limit ( i.e. , the pressure provided by the accelerated particles is negligible ) , and a nonrelativistic shock ,
the change of the spectrum depends on @xmath229 , where @xmath43 is the compression ratio . if @xmath230 the spectral index is changed to @xmath231 .
if @xmath232 the slope remains unchanged but the spectrum is still amplified by a factor @xmath233 ( @xcite ) .
the radio spectral index is @xmath234 , implying @xmath235 . to provide a boost in emissivity by a factor of @xmath236 , @xmath231 has to be @xmath237 , implying a compression ratio of @xmath238 ( assuming @xmath60 ) .
note that this is very close to the compression ratio one derives for an adiabatic compression .
the spectral index produced by such a shock is @xmath239 , consistent with the observed @xmath240 .
@xcite showed that the produced power law softens as one departs from the test particle limit . also , the simple treatment stated in this paragraph breaks down in the case of relativistic shocks , where the spectral index is no longer a simple function of the compression ratio .
recent investigations by @xcite have shown that oblique relativistic shocks can produce rather steep spectra , but other results indicate that they might be more efficient in accelerating particles ( i.e. , producing flatter spectra ) than their nonrelativistic counterparts ( @xcite ) .
@xcite find an optical to x - ray spectral index of @xmath15 for the knot a region , which seems to be consistent with low efficiency acceleration .
if fermi acceleration were present at the shock and effective enough to change the spectral shape in the optical , it would no longer be feasible to use the data from the whole jet to determine the magnetic field .
rather , the same analysis could simply be carried out seperately for the pre and post knot a regions of the jet .
we would then have to make an estimate of the shock strength based on the known parameters in order to determine the ratio of pre to post shock lorentz factor @xmath241 . even in this case
we would need lorentz factors of order @xmath242 to solve the cooling problem .
however , based on the observed mild spectral changes , and the assumption that pitch angle scattering is strong , we conclude that fermi acceleration , if present , will not be efficient enough to affect the spectrum below the cutoff .
inefficient fermi acceleration might very well be present in knot a , producing the x - ray emitting particles observed .
prediction of the produced high energy spectral index has to wait for more conclusive results on fermi acceleration at relativistic oblique shocks .
an important complication to the treatment above is the fact that the magnetic field will not be completely disorganized polarization measurements show that in some regions a homogeneous component is present .
in fact , it is possible that cancellations between regions with homogeneous fields but different orientations occur along the line of sight ( @xcite ) . in such a case ,
the assumption of disorganized fields , leading to @xmath243 , is no longer justified . however , since the impact that the parameter @xmath173 has on the fit is minor ( fig .
[ fig : berr ] ) , we feel this caveat is not very severe and merely mention this complication here .
in addition to the unknown orientation of the jet itself , the field orientation is also unknown .
since synchrotron emission depends on the magnetic field orientation to the line of sight @xmath52 as @xmath201 , it can be strongly peaked away from the field direction .
the cutoff frequency also depends on @xmath244 .
because the cutoff frequency is determined by sampling all regions along the line of sight , it is not obvious which value to choose for @xmath52 .
fortunately , in a domain of disordered field the regions with the field oriented close to perpendicular will contribute most of the flux , thus the error we make by setting @xmath244 to 1 will not be too large .
note that we used a value of @xmath245 in calculating the pressure , the appropriate average of @xmath246 over @xmath247 sterad .
knot a shows a polarization of order @xmath248 and a field orientation close to the shock plane . neglecting relativistic effects , the small amount of upstream polarization , and the fact that shear might reduce the compression of the field
, we can use the approximate formula ( @xcite ) @xmath249 where @xmath250 is the fractional polarization , @xmath43 is the proper compression ratio , and @xmath251 is the line of sight angle from the shock plane , to obtain a lower limit on the compression ratio in knot a of @xmath252 , valid for the inferred range of viewing angles with respect to the shock plane ( bb ) .
this is consistent with the compression ratios inferred above .
equation ( [ eq : polar ] ) also provides an upper limit of @xmath253 on @xmath251 ; however , the shock is assumed to be oblique , so we can not use this result to constrain @xmath254 .
having an estimate of both magnetic field and cutoff momentum at the injection radius @xmath66 , we can now try to determine where the actual particle acceleration has to occur .
we assume some radial dependence for the magnetic field in the _ inner _ portion of the jet ( i.e. , smaller than @xmath255 ) , for example a power law : @xmath256 .
furthermore we make the simplifying assumption of a constant @xmath1 . by also taking @xmath32 to be proportional to @xmath257 , which determines the radial dependence of @xmath42
, we can invert equation ( [ eq : solution ] ) and solve for the radius at which the cutoff momentum approaches infinity , in other words , the minimum radius inside of which acceleration has to occur : @xmath258 where @xmath259 , and @xmath260 .
the acceleration radius @xmath261 approaches zero if @xmath262 .
it is obvious that the estimate of @xmath261 depends critically on the value of @xmath263 . in figure
[ fig : racc ] we have plotted @xmath261 as a function of @xmath263 for @xmath264 and @xmath265 . in the same figure we have also plotted @xmath261 for the case in which b no longer scales like @xmath257 in this case we have taken @xmath266 and used the same values for @xmath162 and @xmath1 ( dashed line ) . if the jet expands at constant opening angle and with uniform @xmath267 , the decline of b with radius should correspond to @xmath268 , since the density scales like @xmath269 and dissipative effects will probably limit the rate of decline . adopting an upper limit on the magnetic field at @xmath270 of @xmath271 ( @xcite ) limits @xmath263 to values smaller than @xmath272 . for @xmath265 , @xmath273 , and @xmath60 ,
this constrains the acceleration radius to be @xmath274 , or @xmath275 . on the other hand
, the radius of acceleration can not lie inside the schwarzschild radius of the central black hole , which is of order @xmath276 for a @xmath277 black hole .
the dependence of @xmath261 on @xmath263 , @xmath173 , and the core magnetic field is too strong to make any detailed predictions about where the acceleration actually has to occur .
however , the @xmath261curve is rather flat throughout most of the possible range for @xmath263 , which suggests that the most plausible value for @xmath261 falls between @xmath272 and @xmath58 pc .
this is intriguingly far away from the central engine .
we have proposed that the apparent lack of synchrotron cooling in the m87 jet most likely indicates the presence of a sub - equipartition magnetic field . while the total ( particle + magnetic ) pressure needed to explain the observed synchrotron emissivity is uncomfortably high in the nonrelativistic limit , doppler beaming effects consistent with bulk lorentz factors in the range @xmath107 lower the pressure requirements considerably . fluctuations in the synchrotron emissivity and spectral cutoff frequency are consistent with adiabatic changes in the magnetic field strength and particle energies that accompany compressions and rarefactions along the flow fermi acceleration along the flow is not necessary to explain the observations .
the knots are identified with relatively weak shocks , as inferred from other data by @xcite . the first - order fermi acceleration expected to occur at such shocks ,
if any , would generate a particle energy distribution steeper than the @xmath278 needed to produce the radio to optical synchrotron spectrum .
thus , effects of particle acceleration along the jet might be apparent only shortward of the cutoff frequency , e.g. , in the x - ray band .
however , a disordered magnetic field or a `` superluminal '' field orientation with respect to the shock front ( begelman & kirk 1990 ) could further reduce the efficiency of fermi acceleration , hence we should continue to regard the origin of the x - ray emission as unknown . the pressure estimates we derive are consistent with the assumption that the m87 jet is embedded in a moderately overpressured bubble , as suggested by @xcite . as a result , it seems that the set of parameters we have suggested above can solve the cooling problem of the m87 jet , as sub - equipartition fields are able to explain both the behavior of the cutoff frequency and the confinement of the jet . using derived values for the magnetic field and the cutoff momentum at @xmath66 we can put an upper limit of @xmath279 on the radius
at which most of the particle acceleration occurs . explaining why the acceleration is confined to a particular scale ( which may be quite large compared to the size of the central black hole ) poses an interesting problem for future work .
the method we have developed should be applicable to other radio sources as well , as long as propagation effects on the particle momentum distribution can be treated as adiabatic .
thus , as soon as comparable data becomes available for other jets , it would be easy to apply the same analytic techniques to them and obtain limits on the magnetic field and the pressure in these sources .
the analysis of the m87 jet we carried out in this paper could be improved by better deprojection models of the jet .
higher quality spectral data both in the near infrared and the x - ray would help to verify the assumption of constancy of the underlying spectral shape and decrease the uncertainty in the location of spectral features ( the knowledge of which is essential for this technique to work ) .
_ hst / nicmos _ will be ideally suited to mapping out the jet properties in the near infrared , a spectral region in which the run of the spectral index is still fairly poorly determined .
_ axaf _ , providing imaging spectroscopy at sub - arcsecond resolution , will be the ideal tool to obtain more detailed information about the x - ray spectrum and emissivity along the jet , hopefully allowing us to determine the radiation mechanism at high energies .
the model we have described above fits in nicely with the new picture of m87 that has emerged in several recent papers .
a moderately relativistic jet and magnetic fields somewhat below equipartition can explain the lack of cooling observed at optical wavelengths .
oblique shocks , weak enough to render fermi acceleration unimportant , can explain both the small scale variability of the cutoff frequency and the brightness variations via adiabatic compressions .
we believe this model can tie together most of the observations available today and will take us closer to the true nature of the jet .
we are grateful to klaus meisenheimer for an exceptionally thorough and detailed referee report , which helped us to improve the paper .
klaus meisenheimer and bill sparks also provided us with the observational data files that we used to fit our models .
this work has been partially supported by nsf grants ast-95 29175 and ast-91 20599 and a fulbright up95/96/97 grant .
baade , w. 1956 , , 119 , 221 ballard , k.r . , &
heavens , a.f .
1992 , , 259 , 89 bicknell , g.v . , & begelman , m.c .
1996 , , 467 , 597 ( bb ) begelman , m.c . , & kirk , j.g .
1990 , , 353 , 66 biretta , j.a .
1993 , in astrophysical jets .
space telescope science institute symposium series vol.6 , d. burgarella , m. livio , & c. odea ( eds . ) ( cambridge : cambridge university press ) , 263 biretta , j.a . , & meisenheimer , k. 1993 , in jets in extragalactic radio sources .
lecture notes in physics vol .
421 , h.j .
rser & k. meisenheimer ( eds . ) ( berlin : springer ) , 159 biretta , j.a . ,
stern , c.p .
, & harris , d.e .
1991 , , 101 , 1632 biretta , j.a .
, zhou , f. , & owen , f.n .
1995 , , 447 , 582 boksenberg et al .
1992 , , 261 , 393 coleman , c.s . , &
bicknell , g.v .
1988 , , 230 , 497 drury , l.o .
, axford , w.i . , & summers , d. 1982 , , 198 , 833 hughes , p.a . ,
& miller , l. 1995 , in beams and jets in astrophysics .
cambridge astrophysics series vol .
hughes ( ed . )
( cambridge : cambridge university press ) , 1 junor , w. , & biretta 1995 , j.a . , , 109 , 500 kirk , j.g .
1994 , in plasma astrophysics .
saas - fee advanced course vol .
benz & t.j .- l .
courvoisier ( eds . ) ( berlin : springer ) , 225 kirk , j.g .
, & heavens , a.f .
1989 , , 239 , 995 meisenheimer , k. , rser , h.j . & schltelburg , m. 1996 , , 307 , 61 meisenheimer , k. , rser , h.j . ,
hiltner , p.r . ,
yates , m.g .
, longair , m.s . ,
chini , r. , & perley , r.a .
1989 , , 219 , 63 meisenheimer , k. 1992 , in physics of active galactic nuclei , w.j .
duschl & s.j .
wagner ( eds . ) ( berlin : springer),525 neumann , m. , meisenheimer , k. , rser , h.j . , &
fink , h.h .
1997a , , 318 , 383 , in preparation ( nmr ) owen , f.n .
, hardee , p.e . , & cornwell , t.j . 1989 , , 340 , 698 perola , g.c . , & tarenghi , m. 1980 , , 240 , 2 rees , m.j . 1978 , , 184 , 61p , 283 , 873 , 283 , l111 rybicki , g.b . , &
lightman , a.p .
1979 , radiative processes in astrophysics , ( new york : wiley ) sparks , w. , biretta , j.a .
, & macchetto , f. 1996 , , 473 , 254 ( sbm ) stiavelli , m. , peletier , r. , & carollo , c.m .
1997 , , 285 , 181 stocke , j.t . ,
rieke , g.h . , & lebofsky , m.j .
1981 , , 294 , 5839 , 319 white , r.e . , & sarazin , c.l .
1988 , , 335 , 688 | we propose that the intensity changes and spectral evolution along the m87 jet can be explained by adiabatic changes to the particle momentum distribution function and the magnetic field .
this is supported by the lack of any significant variation in the radio to optical spectral index along the jet and the moderate changes in radio brightness . assuming a simple scaling law between magnetic field and density
, we use the deprojection of a 2 cm vla intensity map by sparks , biretta , & macchetto ( 1996 ) to predict the spectral evolution along the jet .
we derive limits for the magnetic field and the total pressure by comparing our results with the spatially resolved fit to spectral data by neumann , meisenheimer , & rser ( 1997 ) of a model spectrum that cuts off at @xmath0 . to explain the weakness of synchrotron cooling along the jet
, the magnetic field strength must lie below the equipartition value .
although the inferred pressure in the limit of nonrelativistic bulk flow lies far above the estimated pressure of the interstellar matter in the center of m87 , bulk lorentz factors @xmath1 in the range of @xmath2 and inclination angles @xmath3 lead to pressure estimates close to the ism pressure .
the average best fit magnetic fields we derive fall in the range of @xmath4 , departing from equipartition by a factor @xmath5 .
this model is consistent with the proposal by bicknell & begelman ( 1996 ) that the knots in the m87 jet are weak , oblique shocks . first
order fermi acceleration will then have a minimal effect on the slope of the radio to optical spectrum while possibly accounting for the x - ray spectrum . |
the production of dust in the ejecta of core - collapse supernovae ( cc - sne ) is supported by observations @xcite , while theoretical studies of dust condensation in primordial cc - sne have predicted that 0.10.3
@xmath6of dust could be produced @xcite .
evidence for dust formation by ( at least ) some sne comes from precise isotopic abundance ratio studies of grain inclusions found in meteorites , with many ( including graphite and silicon carbide inclusions ) exhibiting isotopic distributions that differ significantly from those found in the sun and earth ( e.g. * ? ? ?
* ; * ? ? ?
. however , direct evidence that sne play a major role in the dust budget of galaxies is still sparse , since of the few supernovae that have been probed for dust production , most appear to have formed much less dust than predicted by models .
for example , analyses of the photometric and spectroscopic evolution of sne 1987a and 1999em from 400800 days after outburst yielded estimates of a few@xmath7 @xmath6of dust formed per sn @xcite , which translates into condensation efficiencies , defined here as ( mass of refractory elements condensed into dust)/(mass of refractory elements in ejecta ) , of @xmath8 @xcite .
in contrast , for sne to have produced the majority of the @xmath9@xmath10 @xmath6of dust found in high - redshift ( @xmath11 ) quasars @xcite , their condensation efficiency must have been around @xmath120.2 @xcite . in the cases of both sne 1987a and 1999em , the above investigators all separately noted that the dust masses could be significantly larger if the dust was clumped rather than distributed homogeneously throughout the ejecta .
the presence of clumped dust is supported by the wavelength - independent circumstellar extinction seen in some sne @xcite , and by the theoretical expectation that rayleigh - taylor instabilities should form in post - shock ejecta @xcite .
@xcite confirmed this hypothesis for dust formation within sn 2003gd , by demonstrating with three - dimensional monte - carlo radiative transfer ( rt ) models that clumpy , inhomogeneous dust models can require up to an order of magnitude more mass , than implied by either analytic estimates or by homogeneous dust models , to produce a given optical extinction and mid - infrared ( ir ) excess .
sugerman et al .
found that clumped models implied that up to @xmath13 @xmath6of dust had formed in the ejecta of sn 2003gd by day 678 , corresponding to a dust condensation efficiency of up 0.12 , which is interestingly close to the efficiency of @xmath120.2 required to match high redshift dusty sources . since the previous , rather low , dust mass estimates for sn 1987a were based on analytical techniques , we were therefore motivated to re - analyse the available observational data for this object using modelling techniques similar to those used by sugerman et al . for sn 2003gd .
in addition to allowing for clumped dust distributions , we also study the effects that the geometry of the medium and the spatial distribution of the luminosity source have on the emerging spectral energy distributions ( sed ) and , as a consequence , on dust masses determined by fitting rt models to observations .
radiative transfer in an inhomogeneous medium has been the subject of a number of studies , including the simplest approach of a two - phase medium consisting of dense clumps embedded in a less dense interclump medium ( icm ) .
these studies employed a combination of analytical techniques and monte carlo modelling with dust models of varying complexity , in order to study dust in different phases of the interstellar medium ( ism ) and in protogalactic and galactic environments ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
all the models presented in this work were performed using the 3d monte carlo rt code mocassin @xcite , which can account for both ( primary and secondary ) components of the radiation field in a fully self - consistent manner through absorption , re - emission and scattering of photons .
this is necessary if one wants to deal with the transfer through inhomogeneous clumpy distributions , both for the case of a diffuse illuminating source distributed through the medium or for the case of a central illuminating source .
section 2 describes the basic assumptions used for the modelling , while the results of our parameter investigation are presented and discussed in section 3 . in section 4 ,
we readdress the issue of the mass of dust formed by sn 1987a , constraining our models primarily with the observations between @xmath120.3@xmath14 m and 30@xmath14 m published by ( * ? ? ?
* hereafter w93 ) .
using both smoothly varying and clumped grain density distributions , the 3d monte carlo radiative transfer ( rt ) code mocassin @xcite was used by @xcite to derive the mass of dust condensed by the type ii sn 2003gd , by fitting optical ( b , v , r , i ) and _ spitzer space telescope _
3.6@xmath14 m , 4.5@xmath14 m , 5.8@xmath14 m , 8.0@xmath14 m , and 24@xmath14 m fluxes observed at two epochs ( days 499 and 678 after outburst ) .
their results clearly showed that analytical analyses of optical and ir data , as well as rt models with smoothly varying dust density distributions , can underestimate the dust mass by an order of magnitude or more compared to models that allow for clumping .
@xcite provided evidence that optically thick clumps may exist in the envelope of sn 1987a , implying that the dust mass may be underestimated by sed - based smooth density distribution models .
these authors argued that amorphous part dust was required by the wavelength - dependent blueward shift of various emission lines , while clumps are required to explain the wavelength independent reddening effect of the dust .
successively , w93 summarised the evidence for significant clumping in the ejecta of sn 1987a , which was provided by observations ranging from @xmath0-ray and x - ray to infrared wavelengths .
we have therefore revisited the observational data for sn 1987a , employing a similar modelling strategy to that used for sn 2003gd , i.e. we assume that dust condensed in clumps within the ejecta , with the heating source distributed within the remnant shell .
our models are appropriate for type ii sne older than a few hundred days , when the energy input is dominated by @xmath0-rays from @xmath1co decay ( * ? ? ?
we consider two clumping models , designed to represent the extreme cases of likely clumping behaviour : ( i ) the diffuse heating source is mixed with the dust grains , which are present in both the clumps and in the interclump medium ( icm ) ; and ( ii ) the diffuse illumination source is confined to the virtually dust - free icm , with all dust grains confined to the clumps .
the latter model appears the most physically plausible , given that the outer c , o , and si - rich shells , where carbonaceous and silicate grains are expected to form , should never completely mix with the inner @xmath1ni/@xmath1co zone @xcite .
only macroscopic mixing has been found in the clumpy ejecta of the type ii supernova remnant cas a @xcite , suggesting that the element layers were not homogeneously mixed . for completeness
, we also consider a `` smooth '' , clump - free model .
@xcite reported early ir observations ( day 265 and 267 ) and found no evidence for any emission from dust that might have formed in the sn ejecta or from pre - existing dust in the surrounding medium .
a variety of observations indicated that the onset of dust formation in the ejecta of sn 1987a occurred at about day 530 ( * ? ? ?
* w93 ) . given that the featureless 10-@xmath14 m spectrum of sn 1987a after day 530 ( w93 ; * ? ? ?
* ) precludes a significant silicate contribution , the dust was assumed to be either amorphous carbon ( optical constants from * ? ? ?
* ) or graphite ( optical constants from * ? ? ?
in view of theoretical predictions that supernova grain size distributions should be relatively lacking in large particles @xcite , we adopted a standard ( * ? ? ?
* hereafter mrn ) ism grain size distribution , n(@xmath15 ) @xmath16 , but with the grain radii @xmath15 truncated at the upper end , such that @xmath17 = 0.005 @xmath14 m and @xmath18 = 0.05 @xmath14 m .
we found however that for the models presented in section 3 , the seds predicted by considering a full mrn distribution versus a truncated size distribution are quite similar and that the choice of either did not affect the general results of this paper .
following our parameter investigation , we determine best fits to the w93 day 615 and day 775 spectra for all three dust configurations ( smooth , clumpy i and clumpy ii ) and in section 4 present revised dust mass estimates for sn 1987a on days 615 and 775 . model and the green - dashed line is for a model with @xmath19 . , width=321 ] the 3d monte carlo photoionisation and dust rt code mocassin was described and benchmarked in @xcite .
we used version 2.02.21 of the code in its dust - only rt mode , which allows a multi - grid resolution approach in order to increase the spatial resolution in the clumps .
the radiation field is described by discrete monochromatic packets of energy @xcite , whose trajectories through the shell ( mapped on multiple cartesian grids ) are characterised by absorption , re - emission and scattering events , according to the local medium opacities and emissivities . to ensure that energy conservation is strictly enforced at each location ,
all energy packets are followed until they escape the shell , contributing to the emergent sed .
the initial frequencies of the energy packets , emitted in this case by an extended diffuse source , are determined by sampling the probability density functions ( pdf s ) obtained from the blackbody distribution for the temperature of the source .
the total bolometric luminosity , @xmath20 , integrated over the diffuse source equals that of an equivalent centrally - located point source and the energy carried by each individual energy packet in unit time is simply @xmath21 , where @xmath22 is the total number of energy packets employed in the simulation .
our models consist of a spherical shell , with inner radius @xmath23 , outer radius @xmath24 and a @xmath25 density profile .
two basic dust density distributions are considered : one where the density varies smoothly as @xmath26 and the other with dense homogeneous clumps embedded in a less dense icm , with a density profile varying as @xmath26 .
the clumps have radius @xmath27 and a volume filling factor , @xmath28 .
the density contrast between the clumps and the icm is defined by @xmath29 , where @xmath30 and @xmath31 are the clump and icm dust densities at the inner edge of the shell .
the absolute dust density in the clumps is therefore constant for all clumps , regardless of their radial distance , meaning that the density contrast with the icm is larger at larger shell radii .
our choice to employ this model is based upon the assumption that as early as a few hours after outburst , post - shock ejecta become rayleigh - taylor unstable @xcite , forming a clumpy distribution .
the clumps would then simply move outward with the original imprint of density enhancement remaining intact .
the clump positions are assigned stochastically , with the probability of a given grid cell being occupied by a clump being proportional to @xmath26 .
the total number of clumps in a given model is controlled by the volume covering factor , @xmath28 , and the clump size parameter , which is kept constant at a value of @xmath32 = @xmath33/30 .
the smooth models use 61@xmath34 cells to describe the sn shell , although we make use of the assumed symmetrical nature of the object and restrict our simulations to only one eighth of the total volume , reflecting the packets at the cartesian planes @xcite .
figure [ fig : clumpy ] shows a 3d representation of the density distribution in the clumpy models , consisting of cross - sections taken along the positive domains of the orthogonal cartesian planes
. the clumpy models also use a _ mother _ grid of 61@xmath34 cells to describe the icm , whilst each individual clump is described by a subgrid of 5@xmath34 cells which occupies the volume of a _ mother _ grid cell .
we experimented with higher resolution grids and subgrids and found that they had negligible consequences on our final results .
for clumpy model configurations i and ii we first vary the dominant clump parameters and show the effects of these variations on the emergent seds and equivalent optical depths , defined in section [ sec : difil ] below . the illuminating source is adopted to have a total bolometric luminosity @xmath35 at day 615 ( from figure 3 of w93 ) radiating as a blackbody of @xmath36 = 7000 k. we experimented with lower blackbody effective temperatures and came to the conclusion that the observations could be best fitted by a value of t@xmath37 = 7000 k , which may be interpreted as the superposition of the 5000 - 5500 k optically thick ` hot ' component and the h i bound - free continuum component identified in the w93 fits to the sn 1987a observations ( their figure 2 ) .
we also note that for the same epoch ( day 616 ) @xcite and @xcite found @xmath38 and @xmath19 , respectively .
however in this work , we adopt the lower luminosity value which is consistent with and returns the better fit to w93 s data , as shown in figure [ fig : ltest ] , where we compare the seds obtained by identical smooth density models of day 615 , but with different bolometric luminosity .
the red - dotted line is for a model with @xmath35 model and the green - dashed line is for a model with @xmath19 .
the input parameters for all models will be discussed in more details in the following sections .
the diffuse source is distributed either throughout the medium , including in the clumps ( the clumpy i models , in which the diffuse heating source intensity follows the dust density ) ; or located only in the icm ( the clumpy ii models ) , such that each radiating cell @xmath39 emits a luminosity @xmath40 proportional to the local interclump medium density . in the case of a sn shell illuminated by a diffuse source of radiation distributed within the shell , optical depths calculated along a line of sight from the centre of the sphere
have little observational relevance . to account for the fact that the photons emitted from different locations within the shell will encounter very different opacities
, we calculate the _ effective optical depth _
, @xmath41 , from the ratio of the emerging unattenuated flux at @xmath42 for the case where there is no dust present , @xmath43 , to the emergent flux after attenuation by dust in the shell , @xmath44 : @xmath45 this method is completely independent of geometry and therefore does not require averaging along a discrete number of lines of sight ; furthermore it automatically takes into account the effects of photons scattering off the grains , which increases their path lengths through the shell , thus increasing the probability of their being absorbed by dust before escaping .
in this section we present results obtained from clumpy models where both clumps and icm radiate with luminosity @xmath40 proportional to the local medium density .
the effects of varying the volume filling factor of the clumps , @xmath28 , the clump - to - icm density ratio , @xmath46 , and the dust mass , @xmath47 on the emerging seds and effective optical depths of the models are illustrated in figures [ fig : sedcef ] to [ fig : sedcemass ] .
the seds are plotted as dimensionless , distance- and luminosity - independent spectral shapes , or relative flux , @xmath48/@xmath49 , versus @xmath42 in microns ..
figure [ fig : sedcef ] shows that increasing the volume filling factor @xmath28 of the clumps has dramatic effects on the effective optical depth and therefore on the emergent seds , for models with @xmath46 = 100 and the same total dust mass @xmath50 = 5.6@xmath210@xmath51 , with the individual dust clumps each having a mass of 10@xmath52 .
when @xmath28 = 0.05 , each of these clumps has a centre - to - edge radial optical depth in the r - band of 1.22 for the amorphous carbon models .
the same clumps containing graphite grains @xcite , rather than amorphous carbon grains , would have a centre - to - edge radial optical depth in the r - band of 0.46 .
it is clear that an increase of the clump volume filling factor causes a decrease in the effective optical depth of the models , with less optical radiation being reprocessed at ir wavelengths .
this somewhat counter - intuitive effect is in fact a direct result of the luminosity source being embedded in the clumps , so for constant @xmath47 and @xmath46 , an increase in @xmath28 implies a decrease in the clump dust density and thus a reduction of the optical depth through each knot , facilitating the escape of photons emitted by a given knot .
this is opposite to the behaviour of models which have a central source of radiation , or ones in which only the icm radiates , as described in the next section . in figure
[ fig : sedcealpha ] we show the seds ( left panels ) and the equivalent optical depths ( right panels ) for models with a constant @xmath28 = 0.05 and a total dust mass in all clumps of @xmath50 = 5.6@xmath210@xmath51 , for different values of @xmath46 .
variations in @xmath46 do not appear to have large effects on the emerging seds , except perhaps for the @xmath46 = 1 model ( red - dotted line ) , where there is no dust density enhancement in the clumps with respect to the icm .
we note , however , that the icm dust density follows an @xmath26 dependency , whereas the density of the clumps stays constant with distance from the centre , implying that the clump to icm density ratio increases as the square of the distance from the centre .
the main point is that for a model where the intensity of the diffuse heating source follows the dust density , dust grains can not be hidden in optically thick clumps .
an increase in dust mass , e.g. a factor of two , will change the optical and ir sed dramatically , as shown in figure [ fig : sedcemass ] , where in the left- and right - hand panels we plot the emergent seds and equivalent optical depths for models where @xmath28 and @xmath46 stay constant at values of 0.1 and 100 , respectively , and the total dust mass @xmath47 varies from @xmath125@xmath210@xmath51 to @xmath121@xmath210@xmath53 .
finally , it is clear that for these models a degeneracy exists between @xmath28 and total dust mass @xmath50 , which can not be resolved solely by sed observations .
we now present the results obtained from clumpy models ii having exactly the same clump distribution as for i , but this time assuming that dust is only present in the clumps and that the luminosity source is only embedded in the icm .
the parameter @xmath46 tends to infinity for these models , so we only investigate the effects of varying @xmath28 and @xmath47 on the emerging seds and equivalent optical depths .
the left and right panels of figure [ fig : sedcenef ] show the seds and equivalent optical depths corresponding to models with constant total dust mass @xmath47 = 3.5@xmath210@xmath51 and @xmath28 = 0.01 ( black , dotted ) , 0.02 ( red , dashed ) , 0.05 ( green , dash - dot ) and 0.1 ( blue , dash - dot - dot ) . the left and right panels of figure [ fig : sedcenemd ] show the seds and equivalent optical depths corresponding to models with constant @xmath28 = 0.1 and @xmath47 between @xmath54 . for the reference model , with @xmath28 = 0.1 and and @xmath47 = 3.5@xmath210@xmath51 , each clump has a mass of 3.5@xmath210@xmath55 and a centre - to - edge radial optical depth in the v - band of 1.33 and 0.41 , for the amorphous carbon and graphite models , respectively .
it is clear from figure [ fig : sedcenef ] that @xmath28 directly controls the effective optical extinction , i.e. the amount of radiation intercepted by the clumps and reprocessed at ir wavelengths .
the behaviour of the seds and equivalent optical depths here is the same as for a centrally located source and opposite to that shown by the clumpy i models where the heating radiation is emitted within both clumps and icm . in the current case of a dust - free icm with the dust confined to clumps , for optically thick clumps the equivalent optical depth in the visual region
is governed only by the covering factor of the clumps , which determines the amount of radiation intercepted by the clumps .
this would imply that the actual dust mass in the clumps would be unconstrained by the measured extinction in the optical and that virtually unlimited amounts of dust could be hidden in optically thick clumps .
this is apparent in figure [ fig : sedcenemd ] , where it can be seen that the models with 10 , 50 and 100@xmath56 3.5@xmath210@xmath51 show the same sed between @xmath120.3@xmath14 m and @xmath122@xmath14 m .
however the quantity of dust in the clumps still has a significant effect on the grain temperature distribution and therefore on the appearance of the sed at ir wavelengths .
the larger the dust mass for a given @xmath28 , the cooler the average grain temperature and therefore the greater the shift of the reprocessed radiation peak to longer wavelengths .
so , for the clumpy ii case the total dust mass is not unconstrained , even when the clumps are optically thick .
if both optical and ir observational data are available , it appears that the best modelling strategy for optically thick clumps , when segregation between the dust and the heating source is assumed , is ( i ) to start with a large dust mass ( enough for the clumps to be optically thick in the visual region ) and to vary @xmath28 until a fit is achieved for the optical part of the sed ; ( ii ) keeping @xmath28 fixed , to vary the dust mass until a fit to the ir part of the sed is also achieved .
if the dust mass is reduced to the point where the optical sed starts changing again , this implies that the clumps are becoming optically thin
. this may be acceptable , or one may try a lower covering factor and restart the fitting process .
[ cols="<,^,^,^,^,^,^,^",options="header " , ] taking into account the results presented above , we now derive revised dust mass estimates for sn 1987a .
our analysis is based on model fits to the 0.330@xmath14 m spectrophotometry from days 615 and 775 published in figures 1 and 2 of w93 , which were digitised using the _
tool provided by the nasa ads , then dereddened using the @xcite extinction law , scaled to fit the extinction measured toward sn 1987a in @xcite .
our models are also required to match the optical extinction @xmath57 of the sn , which was first shown to increase with time by @xcite . however , as discussed in @xcite , the method adopted by @xcite to measure the dust extinction from the broad - band light curves is likely to have significantly underestimated the @xmath58 and @xmath59-band values .
m. the red dotted line corresponds to a smooth density distribution model , the dashed green line to our clumpy i model and the blue dot - dash line to our clumpy ii models .
model parameters are summarised in table [ tab : bestfit].,width=321 ] but for graphite model fits.,width=321 ] m spectrophotometry published by w93 ( black line ) and the day 793 50- and 100-@xmath14 m photometry of @xcite ( triangles ) .
all models used an mrn size distribution truncated to a maximum grain radius of 0.05 @xmath14 m .
the red dotted line corresponds to a smooth density distribution model , the dashed green line to our clumpy i model and the blue dot - dash line to our clumpy ii models .
model parameters are summarised in table [ tab : bestfit ] . ,
width=321 ] but for graphite models.,width=321 ] our smooth density distribution and our clumpy ii models for day 615 ( red dotted and blue dash dot lines in figure [ fig : sn87a615 ] for amorphous carbon and figure [ fig : sn87a615gr ] for graphite ) both fit the ir and optical sed of sn 1987a quite well the smooth density distribution graphite model ( figure 8) provides a somewhat better fit to the 20-@xmath14 m spectrophotometry than the clumped models , but at the expense of under - predicting the flux level in the short wavelength optical region . the effective extinction predicted by our clumpy ii models for day 615 is a(u ) = 1.69 , a(b ) = 1.66 , a(v ) = 1.61 , a(r ) = 1.54 and a(i ) = 1.48 , for amorphous carbon grains and a(u ) = 1.64 , a(b ) = 1.54 , a(v ) = 1.44 , a(r ) = 1.38 and a(i ) = 1.31 , for graphite grains .
although our absolute values are larger than those reported by @xcite see discussion in @xcite about the extinction being underestimated by the method adopted by @xcite , the slope of the computed extinctions is in agreement with the results of @xcite , who noted that the empirical extinctions were rather flat in the optical .
our day 615 clumpy i models , whilst providing satisfactory fits to the ir sed , provides a poorer match in the optical region . for the day 775 observations ( figure [ fig : sn87a775 ] for amorphous carbon and figure [ fig : sn87a775gr ] for graphite ) , the smooth and clumpy i models produce too much extinction at the shorter optical wavelengths , particularly for the graphite case .
all inputs for our models are listed in table 1 , where we also summarise our dust mass estimates , r - band model extinctions and the r - band radial centre - to - edge extinction within an individual clump . the day 615 smooth density distribution amorphous carbon model requires a total dust mass of 2.0@xmath210@xmath3m@xmath4 , while the amorphous carbon dust masses derived from our clumpy i and clumpy ii models are the same within the uncertainties , 2.0@xmath210@xmath3m@xmath4 and 2.2@xmath210@xmath3m@xmath4 , respectively .
our graphite grain models require approximately 3 times more mass to fit the sed of sn 1987a than do our amorphous carbon models , due to the significantly lower optical to mid - ir absorption coefficients of @xcite graphite compared to hanner ( 1988 ) amorphous carbon ( see figure [ fig : amcvgraphvsilmrn ] ) .
our smooth density distribution graphite model requires a dust mass of 4.2@xmath210@xmath3m@xmath4 , a factor of 10 higher than estimated by w93 for day 615 using their simple large - grain analytic model .
our clumpy
i and ii graphite models both require @xmath126.5@xmath210@xmath3m@xmath4 , factor of two larger than derived by w93 for day 615 using an analytical clumpy model with graphite absorption coefficients . for day 775
, our smooth density distribution amorphous carbon model requires a dust mass of 3.0@xmath210@xmath3m@xmath4 , while our clumpy i and clumpy ii models require 2.0@xmath210@xmath3m@xmath4 and 4.2@xmath210@xmath3m@xmath4 , respectively .
our smooth , clumpy i and clumpy ii graphite models ( fig .
[ fig : sn87a775gr ] ) require masses of 4.5@xmath210@xmath3m@xmath4 , 6.5@xmath210@xmath3m@xmath4 and 7.5@xmath210@xmath3m@xmath4 respectively to fit the day 775 sn 1987a data .
the day 775 smooth density distribution amorphous carbon and graphite models both provide better fits to the observed seds landwards of 20 @xmath14 m than do the clumpy models , but both of the smooth density distribution models fail to match the observed flux level in the short wavelength optical region .
the ir spectrum seems to be too sharply peaked at short wavelengths to be fitted with simple models .
the residuals are also broader than typical atomic line widths , which could be interpreted as evidence for solid - state features in the dust possibly stochastically heated , as discussed bouchet et al . ,
our smooth density distribution graphite mass estimate is 7 times larger than w93 s analytic large - grain estimate for this epoch , while our clumpy i and clumpy ii graphite grain mass estimates are factors of 1.31.5 larger than w93 s analytic clumpy model estimate of 5.0@xmath210@xmath3m@xmath4 for graphite grains .
dwek et al .
( 1992 ) also estimated a dust mass of 5@xmath210@xmath3m@xmath4 for the ejecta of sn 1987a , based on fitting a 150 k modified blackbody to day 1144 ir photometric data .
although their estimate is somewhat uncertain , due to the small number of photometric points in the mid - ir region , it does indicate that the mass of dust in the ejecta of sn 1987a did not increase significantly after day 775 .
furthermore , @xcite presented mid - ir imaging of sn 1987a and reported , in particular , the detection of the sn ejecta in the n - band .
they concluded that their measurements imply a dust temperature in the 90100 k range , and a mass range of @xmath60 , again ruling out a significant variation in dust mass between day 775 and day 6067 . our own dust mass estimates in table [ tab : bestfit ]
imply that for the graphite models the total dust mass hardly increased between days 615 and 775 .
finally , bouchet et al .
( 2004 ) detected mid - ir silicate emission from the circumstellar ring that is believed to have been produced by mass loss by the progenitor star of sn 1987a .
this ring is responsible for most of the present - day mid - ir emission , but bouchet et al .
( 2004 ) showed that this contribution started after day 4200 only , and therefore does not affect the present findings .
we also made estimates for the circumstellar extinction in the envelope of sn 1987a at late epochs in order to constrain our models .
for this we used the methods described in greater detail by @xcite for the case of sn 2003gd .
using the photometric data of @xcite for days 1 - 813 and @xcite thereafter , we estimated the circumstellar r - band extinction of sn 1987a at days 615 and 775 by comparing the observed level of the light curve at those epochs to that expected for a light curve powered purely by the decay of @xmath1co that had been normalised to the photometric data obtained earlier than day 400 .
this yielded @xmath57 s of 1.7 and 2.4 magnitudes at days 615 and 775 , respectively .
a somewhat more conservative estimate of the circumstellar extinction can be obtained if we assume that the intrinsic light curve has contributions from other sources of decay , such as @xmath61co , and also allow for the decreasing opacity of the envelope to the gamma rays that power the light curve .
this yielded @xmath57 s of 1.2 and 1.6 magnitudes at days 615 and 775 , respectively .
corresponding estimates for the r - band circumstellar extinction produced by the models were obtained by comparing the r - band flux levels obtained with and without dust .
inspection of table 1 ( final row for each model ) shows that all of our day 615 models satisfy the @xmath57 = 1.21.7 observational constraint . for the case of our day
775 models the predicted @xmath57 range is 1.52.4 , which can be compared to our estimated observational constraint of @xmath57 = 1.62.4 . ) of @xcite graphite compared to @xcite amorphous carbon and to @xcite astronomical silicate .
the coefficients were computed using mie scattering routines using an mrn size distribution with minimum and maximum grain radii of 0.005 and 0.05 @xmath14 m , identical to that used in all models presented in this paper.,width=321 ] for clumpy i amorphous carbon or graphite models , the derived day 615 and day 775 dust masses are the same , while the inner radius for the day 775 models is 5.7/5.0 times larger than for day 615 . yet the day 775 models yield higher @xmath57 s than the day 615 models .
the equivalent optical depth increase from day 615 to 775 is due to the fact that for the volume filling factor to decrease from 5 to 3% whilst keeping the same dust mass , the dust number density in each individual clump must increase , hence making it harder for radiation produced in the internal regions of the clumps to escape .
the decrease of the clump covering factor from day 615 to 775 is physically justified under the assumption that no clumps and no extra dust condensed between day 615 to day 775 .
if this were true then the expansion of the inner radius from 5@xmath5610@xmath62 cm to 5.7@xmath5610@xmath62 cm would imply a decrease of the volume covering factor from @xmath28=0.05 to @xmath28=0.034 . however , best fit of the sed implies @xmath28 = 0.03 , causing therefore the increase of @xmath57 from 1.4 to 2.0 reported in table 1 .
the behaviour of clumpy ii models is somewhat different ; the volume filling factor here has to increase from 0.1 to 0.3 in order to fit the decrease of the sed in the optical from day 615 to day 775 .
as discussed in section 3.2 , once the clumps become optically thick , the only factor affecting the optical sed is @xmath28 , with the main effect of varying the dust mass being on the temperature of the grains and therefore on the ir emission . in this model
the total dust mass also increases between the two epochs , implying that as new grains form new clumps also form .
an alternative interpretation is that clumps form early on by rayleigh - taylor instabilities in the explosion , and remain imprinted in the ejecta .
the increase in dust mass , might therefore be due to additional accretion .
m. the red dotted line corresponds to a graphite @xcite model .
the dashed green line corresponds to a model with a composition of 50%:50% astronomical silicates : graphite by mass @xcite , the blue dash - dot line corresponds to 25%:75% astronomical silicates : graphite , and the cyan dotted line corresponds to 15%:85% astronomical silicates : graphite @xcite .
the mixed composition models shown all correspond to a dust mass of 1.3@xmath210@xmath5 m@xmath4 , twice as large as for the clumpy ii day 615 graphite model that is plotted . , width=321 ] ) superimposed to the 712 @xmath63 m region of the observed day 615 spectrum from figure 1 of w93 .
the red dotted line corresponds to a graphite @xcite model .
the dashed green line corresponds to a model with a composition of 50%:50% astronomical silicates : graphite by mass @xcite , the blue dash - dot line corresponds to 25%:75% astronomical silicates : graphite , and the cyan dotted line corresponds to 15%:85% astronomical silicates : graphite @xcite .
, width=321 ] our dust mass estimates are dependent on the assumed dust chemistry .
as shown in figure [ fig : amcvgraphvsilmrn ] , amorphous carbon grains with hanner ( 1998 ) optical constants have higher absorption coefficients per unit mass in the optical and near - ir than either astronomical graphite or silicate grains with @xcite optical constants .
the amorphous carbon models therefore require the least mass to fit the observed seds of sn 1987a .
graphite models require between 2 and 3 times as much mass to provide a fit .
silicate - only models would require even larger dust masses , due to their even lower optical and near - ir absorption coefficients ; however pure silicate models are ruled out for sn 1987a by the lack of detectable 10- and 18-@xmath14 m silicate emission or absorption features in any of the mid - ir spectra of sn 1987a ( w93 ; roche et al .
1993 ; bouchet & danziger 1993 ) .
after day 530 , the emission - line strengths of [ si i ] 1.65@xmath14 m , [ mg i ] 4571 , and [ o i ] 6300 decreased significantly @xcite , consistent with the formation of _ some _ fraction of silicate dust .
we therefore tried a number of mixtures of astronomical silicates and graphite .
figure [ fig : sp ] shows fits to the day 615 w93 data obtained with our clumpy ii model using 50% : 50% ( green dash - dot line ) , 25% : 75% ( blue dash - dot line ) and 15% : 85% ( cyan dash - dot - dot line ) astronomical silicate : graphite mixtures , all with masses of 1.3@xmath210@xmath5m@xmath4 ( red dotted line ) , i.e. twice as large as for the clumpy ii graphite model .
the 50% : 50% silicate : graphite and 25% : 75% mixtures produce too strong 10- and 18-@xmath14 m silicate absorption features compared to the observational limits .
however , the 15% : 85% silicate : graphite model produces a smooth enough mid - ir spectrum to be consistent with the observations .
due to the high optical absorption coefficients of amorphous carbon grains , the clumpy ii day 615 amorphous carbon mass estimate of 2.2@xmath210@xmath3 m@xmath4 ( table [ tab : bestfit ] ) is likely to represent a lower limit to the total mass of dust that condensed in the ejecta of sn 1987a , while the mixed silicate : graphite mass of 1.3@xmath210@xmath5 m@xmath4 , discussed in the previous paragraph , probably represents close to an upper limit .
iron grains have optical absorption coefficients in the optical that are intermediate between those of astronomical graphite and silicate , while lacking the strong infrared features that eliminate silicate - only models . however , direct observational evidence for iron grains in circumstellar or interstellar environments is still lacking .
m photometric points at day 793 ( triangles ) are from @xcite.,width=321 ] figure [ fig : comp ] shows a comparison between the day 678 optical - ir spectrum of sn 2003gd @xcite and the day 615 and day 775 spectra of sn 1987a ( w93 ) , after scaling to a common distance of 51 kpc .
it shows similarities between their seds in the optical and 1630-@xmath14 m regions , but differences in the 38-@xmath14 m region . the day 678 clumpy model for sn 2003gd @xcite , required a dust mass that is 13 times higher than the maximum mass allowed for mixed graphite : silicate models of sn 1987a and 27 times higher than our day 775 clumpy ii pure graphite mass estimate for sn 1987a .
the main reason for this difference is the very different opacities of the dust mixtures required to fit the seds of the two sne ; in the case of sn 2003gd , a dust mixture composed of 15% amorphous carbon and 85% astronomical silicates provided the best fit @xcite , while the maximum silicate mass fraction allowed for sn 1987a is only 15% , due to its featureless mid - ir spectrum at late epochs .
astronomical silicates are much more transparent to optical and near - ir radiation than are amorphous carbon or graphite grains , and therefore require larger dust masses to match a given sed .
furthermore , the much higher optical albedos of silicate grains allows much larger masses to be inside clumps , for similar levels of ir emission , though the silicate 10-@xmath14 m feature should be present in emission or absorption ( mainly the latter ) .
we note that models containing only carbon grains did not provide a good fit to the day 499 optical - ir spectrum of sn 2003gd , producing too much emission in the 210-@xmath14 m region .
the model fits imply that the proportions of silicates and carbon grains were reversed in sn 1987a and sn 2003gd .
the mass of the immediate progenitor of sn 2003gd is estimated to have been 8@xmath64 m @xmath4 @xcite , while the mass of sk 69 202 , the progenitor of sn 1987a , is estimated to have been @xmath1220 m@xmath4 @xcite .
nucleosynthetic calculations show that although large masses of oxygen should be produced by high mass supernovae , below 20 m @xmath4 the yield of oxygen from the o - rich zone can drop below the yield of carbon from the c - rich zone @xcite , although the results are highly sensitive to details of some of the adopted reaction rates @xcite .
irrespective of this , @xcite and @xcite have shown that due to the destruction by ionizing particles of co , and the consequent disruption of the normal co sink for carbon atoms in oxygen - rich regions , large quantities of carbon - rich dust particles can form easily in sn ejecta zones that have c / o@xmath651 .
what remains to be elucidated is what causes either silicate or carbon grains to dominate the dust produced by particular supernovae .
we have studied the effects of a number of clumping parameters on the seds emerging from two geometrical configurations , consisting of a spherical shell of clumpy medium illuminated by a diffuse radiation source .
clumpy i models assume full mixing between the dust and the radiating medium , while clumpy ii models assume that dust is only present in the clumps and the illuminating radiation comes only from the icm . 1 . varying the volume filling factor , @xmath28 has opposite effects in clumpy i and clumpy ii models . for a constant dust mass , @xmath47 and clump to icm ratio , @xmath46 ,
an increase in @xmath28 in clumpy i models causes a decrease in the effective optical depth , with less optical radiation being reprocessed at ir wavelengths . in clumpy
ii models @xmath28 directly controls the amount of optical radiation being intercepted by the dust , since the illumination source is always external of the clumps .
this is the same behaviour shown by centrally illuminated shell models . 2 .
in the range explored , variations of @xmath46 in clumpy i models ( constant @xmath47 and constant @xmath28 ) have only a minor impact on the emerging seds .
3 . for models
where the emitting source is intermixed with the dusty medium , dust grains can not be hidden in optically thick clumps .
for these models a degeneracy exists between @xmath47 and @xmath28 which may not be resolved with ir sed observations alone .
4 . for clumpy
ii models with optically thick clumps , an increase in @xmath47 results in cooler grains which will therefore radiate at longer ir wavelengths .
we used both smoothly varying and clumped dust density distributions to obtain new estimates for the mass of dust condensed by the type ii sn 1987a by fitting the 0.3 - 30@xmath14 m spectrophotometric data of @xcite at two epochs ( day 615 and day 775 ) . from our modelling we conclude that between 2@xmath7 and 1.3@xmath66 m@xmath4 of mainly carbon - based grains had formed in the ejecta by day 615 ( 47@xmath7 m@xmath4 ) for pure graphite models ) with the derived dust masses being found to be independent of whether the dust was clumped or not .
this result differs from the conclusions of w93 , who derived unclumped graphite dust masses of 46@xmath67 m@xmath4 on days 615 and 775 , versus graphite masses from their analytic clumped dust modelling of 35@xmath7 m@xmath4 .
the latter values are close to our own clumped graphite dust mass estimates from monte carlo modelling . for a 20 m@xmath4 sn progenitor , between 0.30.4 m@xmath4 of refractory elements
are expected to form ( woosley & weaver 1995 ) , so the estimate of 2@xmath7 to 1.3@xmath66 m@xmath4 of new dust formed in the ejecta of sn 1987a corresponds to a condensation efficiency of only 5@xmath7 to 4@xmath66 , significantly lower than the dust condensation efficiency of @xmath120.12 estimated for sn 2003gd by sugerman et al .
the cause of such very different dust yields amongst type ii core - collapse supernovae remains to be explained . | we present a study of the effects of clumping on the emergent spectral energy distribution ( sed ) from dusty supernova ( sn ) shells illuminated by a diffuse radiation source distributed throughout the medium .
these models are appropriate for type ii sne older than a few hundred days , when the energy input is dominated by @xmath0-rays from @xmath1co decay .
the fully 3d radiation transport problem is solved using a monte carlo code , mocassin , and we present a set of models aimed at investigating the sensitivity of the seds to various clumping parameters . we find that , contrary to the predictions of analytical prescriptions , the combination of an optical and ir observational data set is sufficient to constrain dust masses even in the case where optically thick clumps are present .
using both smoothly varying and clumped grain density distributions , we obtain new estimates for the mass of dust condensed by the type ii sn 1987a by fitting the optical and infrared spectrophotometric data of @xcite at two epochs ( day 615 and day 775 ) .
when using amorphous carbon grains , our best fits to the observational data imply that about 2.0@xmath210@xmath3 m@xmath4 of dust had condensed in the envelope of sn1987a by day 615 and between 2.0@xmath210@xmath3 and 4.2@xmath210@xmath5 m@xmath4 by day 775 .
we find that the absence of a silicate emission or absorption feature in the observed mid - ir spectra implies that no more than 15% of the dust formed around sn 1987a can have been in the form of silicate particles .
our models require larger dust masses for the case of graphite grains , namely between 4.2@xmath210@xmath3 and 6.6@xmath210@xmath3 m@xmath4 at day 615 and between 4.5@xmath210@xmath3 and 7.5@xmath210@xmath3 m@xmath4 at day 775 . from our numerical models we derive dust masses for sn 1987a that are comparable to previous analytic clumped graphite grain mass estimates , and at least two orders of magnitude below the 0.10.3 m@xmath4 that have been predicted to condense as dust grains in primordial core collapse supernova ejecta .
this low condensation efficiency for sn 1987a is in contrast to the case of sn 2003gd , for which a dust condensation efficiency as large as 0.12 has recently been estimated .
supernovae : individual : sn 1987a |
the _ local network effect _ , which is a relatively new idea in economics@xcite , means that the decision of one entity can influence those by whom that entity is connected to .
particularly , the network effect manifests through inactivation as the value of exiting the enterprise is enhanced through the catalytic action of the connections between participants .
homophily@xcite spells that a contact between similar people occurs at a higher rate than among dissimilar people"@xcite , and strongly influences contagions that diffuse through social links@xcite .
mlm participants are thus more likely to connect with others of the same ses , consequently elevating homogeneity ( or depressing diversity ) in the firm .
diversity here is measured by the simpson index @xmath92 serving as a dimensionless potential function minimised when @xmath93 ( at highest diversity ) .
the network effect is constituted as @xmath24 , for @xmath94 ; hence , the network effect is stronger at less diversity . the generalized hurst method@xcite has been coded by one of its authors , t. aste .
the code was downloaded from matlab file exchange website , http://www.mathworks.com/matlabcentral/fileexchange/30076 , and was used with default settings in the calculation of @xmath2 and @xmath3 .
characteristic timescale is chosen at @xmath35 days ( i.e. , @xmath101 month @xmath102 @xmath103 days ) .
assuming that the system - size parameter is of the order , @xmath104 individuals , and the unit @xmath105 individuals , then setting @xmath106 implies an average per - capita encounter rate @xmath21 between @xmath101 and @xmath107 per month , which is a reasonable estimate .
100 beamish , t. d. & biggart n. w. mesoeconomics : business cycles , entrepreneurship , and economic crisis in commercial building markets . in lounsbury , m & hirsch , p. m. ( ed . ) , markets on trial : the economic sociology of the u.s .
financial crisis : part b _ res .
org . _ * 30 * , 245 - 280 ( 2010 ) .
koellinger , p. d. & thurik , a. r. entrepreneurship and the business cycle .
_ , doi:10.1162/rest_a_00224 ( 2011 ) .
eagle , n. , macy , m. & claxton , r. network diversity and economic development .
_ science _ * 328 * , 1029 - 1031 ( 2010 ) .
granovetter , m. the impact of social structure on economic outcomes . _ j. econ .
perspect . _
* 19 * , 33 - 50 ( 2005 ) .
minsky , h. p. monetary systems and accelerator models .
_ * 47 * , 860 - 883 ( 1957 ) .
sims , c. a. macroeconomics and reality .
_ econometrica _ * 48 * , 1 - 48 ( 1980 ) .
colander , d. , howitt , p. , kirman , a. , leijonhufvud , a. & mehrling , p. beyond dsge models : toward an empirically based macroeconomics .
papers & proc . _ * 98 * , 236 - 240 ( 2008 ) .
elsner , w. why meso ?
on aggregation " and emergence " , and why and how the meso level is essential in social economics . _ for .
econ . _ * 36 * , 1 - 16 ( 2007 ) .
albaum , g. & peterson , r. a. multilevel ( network ) marketing : an objective view . _ the market .
* 11 * , 347 - 361 ( 2011 ) .
morales , r. , di matteo , t. , gramatica , r. & aste , t. dynamical generalized hurst exponent as a tool to monitor unstable periods in financial time series . _
physica a _ * 391 * , 3180 - 3189 ( 2012 ) .
di matteo , t. multi - scaling in finance .
_ quant . fin .
_ * 7 * , 21 - 36 ( 2007 ) . burns , a. f. _ the frontiers of economic knowledge _ ( princeton u. p. , 1954 ) .
dopfer , k. the origins of meso economics : schumperter s legacy and beyond .
_ j. evol .
* 22 * , 133 - 160 ( 2012 ) .
coughlan , a. t. & grayson , k. network marketing organizations : compensation plans , retail network growth , and profitability . _ int .
_ * 15 * , 401 - 426 ( 1998 ) .
borland , l. option pricing formulas based on a non - gaussian stock price model .
lett . _ * 89 * , 098701 ( 2002 ) .
gimeno , j. competition within and between networks : the contingent effect of competitive embeddedness on alliance formation .
j. _ * 47 * , 820 - 842 ( 2004 ) .
simpson , e. h. measurement of diversity .
_ nature _ * 163 * , 688 ( 1949 ) .
sundararajan , a. local network effects and complex network structure .
_ * 7 * , doi:10.2202/1935 - 1704.1319 ( 2008 ) .
galeotti , a. , goyal , s. , jackson , m. o. , vega - redondo , f. & yariv , l. network games .
stud . _ * 77 * , 218 - 244 ( 2010 ) .
jackson , m. o. & yariv , l. diffusion of behavior and equilibrium properties in network games .
econ . rev .
_ * 97 * , 92 - 98 ( 2007 ) .
goeree , j. k. , mcconnell , m. a. , mitchell , t. , tromp , t. & yariv , l. the @xmath108 law of giving .
j. microecon . _
* 2 * , 183 - 203 ( 2010 ) .
apicella , c. l. , marlowe , f. w. , fowler , j. h. & christakis , n. a. social networks and cooperation in hunter - gatherers .
_ nature _ * 481 * , 497 - 501 ( 2012 ) .
mcpherson , m. , smith - lovin , l. & cook , j. m. birds of a feather : homophily in social networks .
_ * 27 * , 415 - 444 ( 2001 ) .
aral , s. , muchnik , l. & sundararajan , a. distinguishing influence - based contagion from homophily - driven diffusion in dynamic networks .
u s a _ * 106 * , 21544 - 21549 ( 2009 ) .
campbell , k. e. , marsden , p. v. & hulbert , j. s. social resources and socioeconomic status . _ social networks _ * 8 * , 97 - 117 ( 1986 ) .
van kampen , n. g. _ stochastic processes in physics and chemistry _
( north - holland , 2007 ) .
gillespie , d. t. a general method for numerically simulating the stochastic time evolution of coupled chemical reactions .
_ j. comput .
* 22 * , 403 - 434 ( 1976 ) .
blume , l. e. & easley , d. optimality and natural selection in markets .
_ j. econ .
theor . _ * 107 * , 95 - 135 ( 2002 ) .
webb , j. n. _ game theory : decisions , interaction and evolution _ ( springer - verlag , 2007 ) .
koehn , d. ethical issues connected with multi - level marketing schemes .
* 29 * , 153 - 160 ( 2001 ) . | business - cycle phenomenon has long been regarded as an empirical curiosity in macroeconomics . regarding its causes
, recent evidence suggests that economic oscillations are engendered by fluctuations in the level of entrepreneurial activity@xcite .
opportunities promoting such activity are known to be embedded in social network structures@xcite . however , predominant understanding of the dynamics of economic oscillations originates from stylised pendulum models on aggregate macroeconomic variables@xcite , which overlook the role of social networks to economic activity echoing the so - called _ aggregation problem _ of reconciling macroeconomics with microeconomics@xcite . here
i demonstrate how oscillations can arise in a networked economy epitomised by an industry known as multi - level marketing or mlm@xcite , the lifeblood of which is the profit - driven interactions among entrepreneurs . quarterly data ( over a decade ) which i gathered from public mlms reveal oscillatory time - series of entrepreneurial activity that display nontrivial scaling and _ _ persistence__@xcite .
i found through a stochastic population - dynamic model , which agrees with the notion of profit maximisation as the organising principle of capitalist enterprise , that oscillations exhibiting those characteristics arise at the brink of a critical balance between entrepreneurial activation and inactivation brought about by a homophily - driven _
network effect_. oscillations develop because of stochastic tunnelling permitted through the destabilisation by noise of an evolutionarily stable state .
the results fit altogether as evidence to the burns - mitchell conjecture that economic oscillations must be induced by the workings of an underlying network of free enterprises searching for profit"@xcite .
i anticipate that the findings , interpreted under a mesoeconomic framework@xcite , could open a viable window for scrutinising the nature of business oscillations through the lens of the emerging field of network science .
enquiry along these lines could shed further light into the network origins of the business - cycle phenomenon .
known widely in literature as _ network marketing _ , mlm executes through embedded social networks its essential business functions such as goods distribution , consumption , marketing , and direct selling@xcite .
that makes mlm a stark microcosm of a networked economy .
one salient yet so far overlooked feature of mlm dynamics is the aperiodic oscillations in firm size @xmath0 , quantified by the number of participating entrepreneurs ( fig . 1 ) .
empirical quarterly firm - size data have been collected from four public mlms ( supplementary data ; supplementary methods , s1 ) : nuskin enterprises ( nus ) , nature sunshine ( natr ) , usana health sciences ( usna ) , and mannatech inc .
( mtex ) .
publicly - listed firms are chosen because they are required to disclose accurate business data on a regular basis .
the average revenue ( i.e. , total revenue divided by @xmath1 for any given quarter ) does not proportionately rise with @xmath1 ( fig .
1c , d ) , implying that firm - size expansion does not inevitably translate into revenue growth .
firm size is thus a more reliable quantifier for entrepreneurial activity than is total revenue .
the scaling property of the @xmath0 time - series is examined via hurst analysis@xcite ( methods ; supplementary methods , s1 ) .
the hurst exponents , @xmath2 and @xmath3 , quantify the scaling of the absolute increments and of the power spectrum , respectively .
if time series were generated by a wiener process , such as in the black - scholes model@xcite , then @xmath4 . but
@xmath5 indicates _ persistence _
, i.e. , changes in one direction usually occur in consecutive periods ; whereas @xmath6 suggests _ anti - persistence _ ,
i.e. , changes in opposite directions usually appear in sequence@xcite .
ideally , a single scaling regime means @xmath7 , which applies to time - series generated by unifractal processes such as the wiener process and the fractal brownian motion .
table 1 presents hurst exponents for different mlms .
generally , @xmath5 except for nus north asia and natr with @xmath8 ( within standard deviation ) ; and @xmath9 within standard deviation .
overall , these features of the time - series suggest that mlm firm dynamics is a non - wiener unifractal process@xcite .
unifractality implies self - similarity such that conclusions drawn at one timescale remains statistically valid at another timescale .
.hurst exponents @xmath2 and @xmath3 for different mlms estimated using the generalised hurst method@xcite .
@xmath5 indicates persistence , whereas @xmath6 indicates anti - persistence .
@xmath3 values , which are closely related to the scaling of the power spectrum , are also shown .
generally , @xmath10 within standard deviation .
the standard deviation values are determined from a pre - testing procedure ( supplementary methods , s1 ) . [ cols="^,^,^",options="header " , ] [ tab : hursts ] [ fig : empirical ] an mlm firm is considered as a population of profit - seeking entrepreneurs
. this population exhibits disorder through the presence of two entrepreneur types distinguished by socio - economic status ( ses ) .
let @xmath11 and @xmath12 denote these types , wherein @xmath11 has higher ses than @xmath12 , and @xmath13 and @xmath14 denote their subpopulation size , respectively .
the total population at any given time @xmath15 is thus @xmath16 .
three major processes run the population dynamics : entrepreneurial activation by recruitment ; competitive inactivation ; and catalytic inactivation due to a network effect .
recruitment is expressed in the following reaction equations : @xmath17 wherein @xmath18 and @xmath19 are per - capita rates of recruitment of types @xmath11 and @xmath12 , respectively , and @xmath20 because @xmath11 s higher ses implies a faster rate of entrepreneurial activation through the support of bigger capital and vaster social resources .
competitive inactivation occurs due to market overlap , or _
_ niche overlap__@xcite , as participants can go head - to - head over the same clientele or market .
an encounter rate @xmath21 , which can be related to the density of the embedding social network , quantifies the probability of market overlaps .
thus , competitive inactivation is : @xmath22
lastly , catalytic inactivation , which denotes the _ network effect _ ( methods ) , is expressed as : @xmath23 the network structure of the mlm catalyses inactivation of existing participants at the rate @xmath24 , where @xmath25 is a measure of the probability that two members drawn randomly from the mlm belong to the same type .
it has been widely used in literature as a diversity index@xcite . due to interconnectedness and homophily@xcite
, the inactivation of one could ( like a contagion ) infect another to follow suit .
combining equations ( [ eq : recruitment] [ eq : catalytic ] ) results to a master equation ( supplementary equation 1 ) for the state probability density .
perturbation analysis accounts for the fluctuations arising from demographic stochasticity@xcite . in terms of the system size @xmath26 ( roughly the size of that part of the overall population considered _ fit _ for entrepreneurial activities ) the following ansatz
is made : @xmath27 and @xmath28 , where @xmath29 and @xmath30 are the average concentrations , and @xmath31 and @xmath32 are the magnitude of the fluctuations of the stochastic variables @xmath13 and @xmath14 , respectively ( supplementary methods , s2 ) . the highest order in the expansion expresses the macroscopic rate equations for @xmath29 and @xmath30 : @xmath33 meanwhile , the next highest order term gives the fokker - planck equation , or fpe ( supplementary equation 2 ) , governing the dynamics of the probability density for the magnitude of the fluctuations . from the fpe , the expectation values of the stationary fluctuations are @xmath34 , which supports the interpretation that the deterministic solutions to ( [ eq : macroscopic ] ) are the correct average values .
the model is nondimensionalised by setting the characteristic timescale at @xmath35 days ( methods ) . consequently
, the rates can be squarely related to empirical data by rescaling to appropriate units .
dimensionless rates take on simplified yet meaningful values : @xmath36 ; @xmath37 ; and @xmath38 .
bifurcation analysis ( supplementary methods , s4 ) of the nondimensionalised equation ( [ eq : macroscopic ] ) unveils a bifurcation manifold @xmath39 , where @xmath40 equation ( [ eq : bifurcation ] ) coincides with an evolutionarily stable state ( ess ) of a population game between the types ( supplementary methods , s3 s4 ) .
the fraction of @xmath11 entrepreneurs is @xmath41 , where @xmath42 , and @xmath43 is the fluctuation component .
analysis of the second moments from the fpe shows that the variance diverges as @xmath44 as @xmath45 ( supplementary methods , s5 ) .
that is a signature of criticality through which the ess , where @xmath46 and @xmath47 , is ( quite counterintuitively ) destabilised as the bifurcation manifold is approached .
this mechanism is hereby referred as _
stochastic tunnelling _ wherein noise enables the state trajectory to cross a phase barrier that could not have been otherwise traversed without actively tuning the bifurcation parameter ( supplementary methods , s4 ) .
stochastic tunnelling drives the business oscillations ( fig . 2 ) .
time series is generated by solving the model using a numerical technique , known as gillespie s stochastic simulation algorithm@xcite , which directly integrates the master equation ( supplementary methods , s6 ) .
diverging variance indeed allows the solution to wander far enough from the ess and closer to an unstable point ( uep ) which pushes that solution toward the boundary state , where @xmath48 and @xmath49 ( fig .
2a ) .
that noise also enables the solution to sling back to the ess consequently forming loops in the phase portrait , hence , oscillations in the time series ( fig .
2b ) .
the time series consist of upswings associated with increasing diversity and downswings with decreasing diversity , i.e. , @xmath50 as @xmath51 .
the remarkable observation is that recovery from low points of the series coincide with periods when @xmath11 is dominant a case of the fitter entrepreneurs surviving through recessions"@xcite .
profit maximisation is an axiom of capitalist enterprise@xcite .
mlm may enhance profitability by maximising the proportion of @xmath11 entrepreneurs ( supplementary methods , s7 ) .
thus , the time - average value @xmath52 is examined for various pairs of @xmath18 and @xmath53 which consequently depicts the phase diagram of the model ( fig .
3a ) .
business oscillations come about as a result of stochastic tunnelling through the critical boundary @xmath54 .
phase ii , where @xmath11 stably dominates ( supplementary fig .
s2b ) , can be considered pareto - optimal as the mlm maximises profitability as a whole .
but high levels of targeted recruitment , i.e. , @xmath55 , are required .
entrepreneurial activation , however , might in reality be less discriminatory and thus @xmath56 , which denotes higher entropy ( supplementary discussion ) .
the critical boundary delineates , for any magnitude @xmath53 of the network effect , the minimum @xmath18 that promotes long - run dominance of @xmath11 entrepreneurs .
nevertheless , a stronger network effect tends to frustrate that dominance as catalytic inactivation increasingly outpaces activation , leading to degradation of entrepreneurial activity ( supplementary fig .
s2d ) .
the ess at the iii - iv boundary ( fig .
3a ) is therefore pareto - dominated @xcite .
the hurst maps ( fig .
3b , c ) locate where the real mlms are on the phase diagram .
the hurst exponents are determined from the same exact method .
clusters appear in the vicinity of the iii - iv boundary . on these clusters @xmath2 and @xmath3 are approximately between @xmath57 and @xmath58 , about the same range of values found in real mlms ( see table 1 ) . a correlative plot ( fig .
3d ) between @xmath2 and @xmath3 further confirms not only agreement between model and empirical data , but also their unifractality .
overall , these findings suggest that real - world mlms are pareto - dominated economic systems@xcite , which are operating in an environment characterised by high entropy , i.e. , @xmath56 , and by a strong network effect ( i.e. , @xmath59 ) .
the study paints an illuminating insight about the nature of mlm operations .
mlms have been accused in several instances by discontent participants for ethical violations concerning its business practices@xcite .
the model justifies such disgruntlement for two reasons .
first , that profit is closely associated with recruitment implies less selective entrepreneurial activation .
second , that recruitment proceeds through embedded networks connotes strong network effects .
less - fit entrepreneurs can join the market in droves but are weeded out too soon@xcite because of the pareto - dominated nature of the venture .
the feeling of being victimised is thus not at all surprising .
the mesoeconomic framework ( i.e. , linking microeconomic foundations with macroeconomic phenomena@xcite ) puts the present study in a broader economic context . a more network - dynamic approach to viewing business cycles is hereby encouraged .
lastly , the mathematical model could be extended or refined , such as by generalising the network effect using the hlder mean such that @xmath60{x_a^q+x_b^q},\ ; \forall q>1 $ ] ( supplementary discussion , supplementary fig .
s1 ) ; whereas empirical data of higher temporal resolution may become available in the future , to further test the implications that came forth .
[ fig : phaseportrait ] , @xmath61 , @xmath62 for a total time of @xmath63 @xmath64day periods @xmath65 year . * a , * phase portrait showing one stochastic realization for @xmath0 ( in units @xmath66 ) versus @xmath67 ; @xmath68 marks the initial condition : @xmath69 and @xmath70 .
the dashed curves and lines are the nullclines of the replicator equations from the evolutionary game ( supplementary methods , s3 ) , which intersect at the evolutionarily stable state ( ess , @xmath71 ) and at an unstable equilibrium point ( uep , @xmath72 ) . the trajectory of the solution forms a loop indicating the oscillations .
* b , * time series for @xmath0 and @xmath67 .
hurst analysis gives @xmath73 and @xmath74 for the @xmath0 series.,title="fig : " ] [ fig : phasediagram ] for different pairs of @xmath18 and @xmath53 ( resolution : @xmath75 ) .
phase i represents the regime @xmath76 where the critical manifold @xmath39 is inaccessible due to the constraint @xmath77 .
phase ii denotes the pareto - optimal region @xmath78 and @xmath79 where @xmath80 .
phase iii , where @xmath81 is similar to phase i except that here the critical manifold is accessible .
phase iv is where @xmath82 but @xmath83 resulting to degradation of @xmath1 due to @xmath84 ( supplementary fig .
2d ) .
_ phase boundaries _ :
i ii , iii , @xmath85 ; iii
ii , iv , @xmath39 ; and ii
iv , @xmath86 . *
b , c , * map of the hurst exponents @xmath2 and @xmath3 , respectively . the phase boundaries are superimposed .
the data are generated by simulating the model for @xmath87 @xmath88day periods @xmath89 years with initial population @xmath90 for different pairs of @xmath18 and @xmath53 ( resolution : @xmath91 ) .
each data pixel is an average of four stochastic realisations .
* d , * correlative plot between @xmath2 and @xmath3 .
empirical data are those listed in table 1 . the dashed line @xmath7 denotes unifractality of the time series.,title="fig : " ] |
although @xmath7 galaxies are being found in increasing numbers ( _ e.g. _ , c. steidel ; j. lowenthal , this volume ) , their study has been almost exclusively through their ultraviolet ( uv ) flux redshifted into the optical . as such , we merely know about the near - instantaneous star formation rate the uv continuum is dominated by massive short - lived ob stars , and hence is sensitive to star formation on the time - scale of only @xmath8 years . from studies of local galaxies , morphologies in the uv can be very different from the appearance in the optical ( _ e.g. _ , colley _
_ 1996 , apjlett 473 , 63 ) . to comprehensively study the underlying stellar populations
, one should also utilize longer rest wavelengths , where the light from older and less massive stars may dominate .
a useful spectral feature in the study of stellar populations is the 4000 break , which arises from metal - line blanketing in late - type stars .
even if the stellar population is only moderately evolved , there will still be the balmer break . at high-@xmath9
, these rest - frame optical breaks are redshifted into the near - ir ( figure 1 ) . here
we present preliminary results from an infrared study of stellar populations in a newly - discovered lensed galaxy at @xmath0 . at this redshift , in most cosmologies
the age of the universe is less than @xmath10gyr .
evidence for any semi - evolved stellar population would have significant implications for galaxy evolution in the early universe .
= 0.5 = 0.5 =
deep ground - based imaging and more recent hst studies of the cluster abell 2390 ( @xmath12 ) have revealed many giant arcs , arising from gravitational lensing of background galaxies . in particular
, a @xmath13 elliptical close to the critical curve of the cluster bisects an extended pair of thin , near - straight arcs ( figure 2 ) .
the brighter northern arc ( @xmath14 ) is @xmath15 long , with the southern component ( @xmath16 ) @xmath17 in length .
the similarity in colours ( @xmath18 , objects 4&5 in pell _
_ 1991 , apj 366 , 405 ) and their positions relative to the caustic suggest that these arcs are lensed images of the same source .
imaging with hst / wfpc2 reveals that the arcs are marginally resolved in the radial direction ( @xmath19 fwhm ) . taking a nave model of a tangentially - stretched spherical source implies a magnification @xmath20 , and more sophisticated lens geometries lead to a magnification of @xmath21 ( frye & broadhurst 1998 , apjlett _ submitted _ , astro - ph/9712111 ) .
the hst images also reveal the presence of two smaller counter - arcs . an arclet redshift survey by bzecourt & soucail ( 1997 , a&a 317 , 661 ) detected a single emission line at 6137 from the northern component ( their arclet#20 ) , and subsequent keck / lris spectroscopy ( frye & broadhurst 1998 ) confirmed this detection and identified the line as ly-@xmath6 at @xmath0 , on the basis of several rest - frame uv interstellar absorption features .
the southern arc and one of the counter - arcs were also confirmed to be at @xmath0 .
these arcs are amongst the highest - redshift galaxies known , and are not pre - selected for some agn characteristic or extreme luminosity ( they are magnified ) , nor are they preferentially colour - selected for a young stellar population .
thus , we may presuppose them to be lensed images of a ` normal ' galaxy , and this affords us the opportunity to study stellar populations in the early universe , uncontaminated by any contribution from a central engine .
at @xmath22 , the rest - frame optical is redshifted to the near - infrared , and the balmer+4000 break is placed between the @xmath3-band ( at @xmath23 m ) and the @xmath24-band ( at @xmath25 m ) . in figure 1 ( left panel ) we plot the evolution of the @xmath26 colour with age for an instantaneous starburst at @xmath0 .
for this system , the @xmath26 colour is an excellent discriminator of the stellar ages along the arcs .
we recently obtained deep @xmath3 & @xmath4 ( @xmath27 m ) imaging of the arcs in good seeing with the near infrared camera ( nirc ) on the w. m. keck - i telescope ( figure 2 ) .
the @xmath4-band approximates to the rest - frame @xmath28-band , and the integrated flux along the brighter northern arc ( @xmath29 ) suggests an intrinsic luminosity of @xmath30 for a @xmath31 cosmology , where @xmath32 is the magnification , and @xmath33 if @xmath34 .
comparison with the wfpc2 images reveals colour gradients along both arcs .
optical spectroscopy shows that only one of the four bright knots of the northern arc has prominent ly-@xmath6 emission ( labeled n4 in figure 2 ) .
the optical / near - ir colours of this are consistent with an ob star , implying a region of active star formation .
however , the broad - band colours integrated along the entire northern arc ( figure 1 , right panel ) show that the overall spectrum is best fit by a more evolved ( @xmath35myr ) population with extinction intrinsic to the source of @xmath36 .
this implies that there are older stars outside the actively star - forming h ii regions . in some areas , @xmath37 , indicating that @xmath38myr has elapsed since the major episode of star formation ( figure 1 , left ) .
our keck @xmath3- and @xmath4-data allow us to ` age - date ' the stellar populations in this spatially - resolved @xmath11 galaxy by the inferred amplitude of the balmer+4000 break along the transverse extent of the arcs . a more detailed treatment of the near - ir / optical colours will be given in bunker _
et al . _
( 1998 , _ in preparation _ ) .
we would like to thank the organizing committee .
we are grateful to daniel stern for many useful discussions .
this work is based on observations obtained at the w.m .
keck observatory , and with the nasa / esa hubble space telescope , obtained from the data archive at stsci which is operated by the association of universities for research in astronomy , inc .
, under the nasa contract nas 5 - 26555 . | we have recently obtained deep near - ir keck imaging of a newly - discovered @xmath0 galaxy ( frye & broadhurst 1998 ) .
this is lensed by the rich foreground cluster abell 2390 ( @xmath1 ) into highly - magnified arcs
@xmath2 in length .
our @xmath3- and @xmath4-band keck / nirc imaging allows us to map the balmer@xmath5 break amplitude . in combination with high - quality archival hst / wfpc2 data
, we can spatially resolve stellar populations along the arcs .
the wfpc2 images clearly reveal several bright knots , which may correspond to sites of active star formation . indeed ,
in some spatial regions the keck / lris discovery spectra are consistent with ob - star spectral energy distributions in the rest - ultraviolet
. however , there are considerable portions of the arcs which appear redder with no ly-@xmath6 emission , consistent with being post - starburst regions . |
massive stars in close binaries are destined to interact in some way over their lifetimes .
the initially more massive star is the first to evolve , and after shedding its outer , h - rich envelope by stellar winds , it may begin mass transfer of cno nuclear processed gas to its companion @xcite .
after completion of the mass transfer stage , the abundance ratio of n / c in the photosphere of the gainer may have increased from the solar value of 0.3 to an enriched value of 3 @xcite , or even higher if the mass accretion had induced large scale mixing in the mass gainer @xcite . however , there is conflicting evidence about whether or not mass transfer actually occurs in real binaries composed of massive , o - type stars @xcite .
the massive x - ray binaries ( mxrbs ) , for example , generally do not exhibit evidence of n - enrichment , and this fact suggests that the supernova ( sn ) progenitor may eject mass from the entire system , perhaps during a luminous blue variable stage @xcite . we were surprised , therefore , to discover that the @xmath14 lines in the spectrum of the mxrb and microquasar ls 5039 ( type o6.5 v((f ) ) ; @xcite ) displayed a weakness indicative of cno processed gas @xcite .
ls 5039 is also unusual in several other respects .
it is one of only a few confirmed mxrbs with associated radio emission @xcite .
it has radio - emitting relativistic jets characteristic of galactic microquasars , and it is probably a high energy gamma ray source as well @xcite .
we found that the system is a short period binary with the highest known eccentricity ( @xmath15 ) among o star binaries with comparable periods @xcite .
this high eccentricity probably resulted from the huge mass loss that occurred with the sn explosion that gave birth to the compact star in the system @xcite .
binaries that suffer large mass loss in a sn are expected to become runaway stars , and recently both we @xcite and @xcite found that ls 5039 has a significant proper motion indicating a large peculiar space velocity . here we present results from new optical and ultraviolet spectroscopy that confirm the presence of cno processed gas in the photosphere of ls 5039 .
we give revised orbital elements for the binary ( 2 ) , and we use our combined optical spectra to assign ls 5039 to the category of the n - rich on stars ( 3 ) .
we present the first uv spectra of ls 5039 ( made with hst / stis ) that we analyze to determine the stellar wind properties ( 4 ) and interstellar reddening ( 5 ) .
we use the new results to revisit the issue of the masses of the o - star and compact companion ( 6 ) .
we also present simple models for the change in orbital elements caused by the supernova ( 7 ) , which require a modest kick velocity in order to reproduce the observed eccentricity and runaway velocity .
finally we discuss the possible explanations for the n - enrichment given our current understanding of the evolutionary history of the system ( 8) .
we obtained new spectra of ls 5039 on four different observing runs , and the details about the spectrograph setup in each case are summarized in table 1 .
the primary set of blue region spectra come from observations with the university of texas mcdonald observatory 2.7 m telescope and large cassegrain spectrograph ( lcs ) on two observing runs ( see @xcite for details ) .
all these were made with a ti ccd ( @xmath16 , 15.2 @xmath17 m pixels ) except on one night when we used a loral / fairchild ccd ( @xmath18 , 12 @xmath17 m pixels ) .
most of the spectra were made in a single 1200 s exposure that resulted in a signal - to - noise ratio of s / n = 120 in the continuum .
a secondary set of blue spectra were made with the ctio 1.5 m telescope and cassegrain spectrograph ( cs ) using a loral ccd ( @xmath18 , 15 @xmath17 m pixels ) .
these were usually 1800 s exposures that produced a s / n = 80 in the continuum .
we also observed the red spectral region during a kpno coude feed run in 2002 june .
these spectra are comparable to those we presented in @xcite except that these new spectra were obtained with a ti ccd that had smaller wavelength coverage .
finally , we obtained one additional red spectrum of lower resolving power with the ctio 1.5 m cs on 2003 march 21 .
all these spectra were extracted and calibrated using standard techniques in iraf .
we removed atmospheric telluric lines from the red spectra using spectra of broad - lined a - type stars ( see @xcite ) .
our first task was to re - evaluate the orbital elements presented in @xcite using an enlarged set of radial velocity measurements .
we measured radial velocities for the blue spectra by cross - correlating each spectrum with the average of the blue spectra made at mcdonald observatory ( the largest homogeneous set ) .
we omitted spectral regions containing interstellar features .
these relative velocities were converted to an absolute scale by adding the radial velocity of the average spectrum as determined by parabolic fitting of the cores of the strong lines , @xmath19 km s@xmath3 ( using rest wavelengths from @xcite ) .
radial velocities from the red spectra were also determined by cross - correlation techniques , but in this case the reference spectrum was formed by a shift - and - add average of the red spectra obtained earlier @xcite .
this cross - correlation function is generally based on the positions of the h@xmath10 and @xmath20 features , but for the ctio red spectrum we also included the spectral regions containing @xmath21 and @xmath22 .
we also determined the cross - correlation positions of the sharp interstellar line at 6614 that we used to make small velocity corrections between observatories .
the 14 new radial velocity measurements are collected in table 2 . our first attempt to revise the orbital period and other elements starting with the period we found earlier ( @xmath23 d ; @xcite ) failed , with 8 of the 14 new measurements falling far from the calculated radial velocity curve .
thus , we made a new period search using the discrete fourier transform method of @xcite .
there are a number of candidate peaks in the periodogram that are clustered around a frequency of 0.25 cycles d@xmath3 , and most of these are aliases introduced by our sparse data sampling @xcite .
we tested each of these candidate periods using the nonlinear , least - squares fitting program of @xcite to determine the orbital elements , and only one of the test periods , @xmath4 d , led to consistent placement of all the new measurements .
this value was also the primary signal found using the phase dispersion minimization technique of @xcite that is useful for finding periodicities in non - sinusoidal variations .
significantly , this period also fits new radial velocity data obtained by j. casares , m. rib , j. m. paredes , and j. mart ( in preparation ) .
the revised orbital elements are given in table 3 , and the observations and radial velocity curve are shown in figure 1 .
this fit assumed equal weights for each radial velocity measurement ; fits weighted using the inverse square of the standard deviation of each measurement led to identical results within errors .
aside from the orbital period , the new elements do not differ greatly from those we presented earlier @xcite .
most of our observations are clustered into runs of a week or so ( comparable to the period ) so that while the placement of these groups of measurements in phase is somewhat altered , the overall shape of the curve is much the same .
@xcite discuss the appearance of the optical and near - ir spectrum of ls 5039 , from which they derive a spectral classification of o6.5 v((f ) ) for the mass donor star .
here we examine the line patterns in our optical spectra to confirm this classification .
we formed an average blue spectrum by calculating a shift - and - add mean spectrum from our mcdonald and best ctio spectra ( based upon the line shifts from the orbital solution in the last section ) .
the average interstellar spectrum was divided out of each observed spectrum prior to forming the average .
the resulting mean spectrum is shown in the upper plot in figure 2 together with identifications for the more prominent lines .
we also show in this diagram the blue spectra of the standard stars hd 93146 ( o6.5 v((f ) ) ; from the atlas of @xcite ) and 15 mon ( o7 v((f ) ) ; obtained during our ctio run ) . @xcite
have recently made available a library of synthetic stellar spectra for a grid of non - lte atmospheres appropriate for o - type stars , and we show in figure 2 a flux rectified version of one of their models ( with stellar parameters discussed below ) .
this spectrum was convolved with broadening functions to simulate stellar rotational and spectrographic instrumental broadening ( discussed below ) .
the spectral subtype among o - type stars is defined by the relative strengths of the and lines @xcite .
the relative strengths of the @xmath24 and @xmath25 lines in the spectrum of ls 5039 make a good match with the o6.5 v((f ) ) standard star ( but not with the cooler o7
v((f ) ) spectrum in which @xmath25 is stronger ) .
the primary luminosity criterion at this spectral type is the strength of the @xmath26 line that ranges from strong absorption to strong emission between main sequence and luminous supergiant stars ( see fig . 9 in @xcite ) .
the deep @xmath26 absorption line in the spectrum of ls 5039 clearly indicates a luminosity class v object .
we also observe weak emission in the @xmath27 blend , which is indicated by the classification suffix ( ( f ) ) .
our results therefore confirm the spectral classification obtained by @xcite of o6.5 v((f ) ) .
we note , however , the unusual strength of the @xmath28 and @xmath29 features in the spectrum of ls 5039 compared to the other reference spectra .
these lines tend to peak near type o8.5 , and at any given subtype they increase in strength with luminosity @xcite .
the equivalent width of the @xmath29 blend in the spectrum of ls 5039 is @xmath30 , which is larger than that in all 13 n - strong stars with spectral types o7 or earlier listed by @xcite ( with the exception of the n enhanced , o7
v((f ) ) star , hd 90273 ) .
furthermore , we observe that the @xmath31 blend is weak or absent in the spectrum of ls 5039 , and other c features are also unusually weak ( @xmath32 , @xmath33 emission , @xmath34 ) . we show below that this trend of strong n and weak c is also found in the uv stellar wind lines ( 5 ) .
these characteristics are the hallmark of the on type stars that appear to be contaminated with cno nuclear processed gas @xcite .
one of the hottest stars that is currently assigned to the on group is hd 110360 ( on7 v ; @xcite ) , and the strong similarity in appearance between ls 5039 and hd 110360 ( see fig . 5 in @xcite ) suggests that ls 5039 also belongs to this group .
thus , we advocate a classification of on6.5 v((f ) ) for the spectrum of ls 5039 . the appearance of the optical line spectrum can also be used to help estimate the physical properties of the star using the new non - lte model spectra from @xcite .
a direct comparison of their grid of model spectra with our observed spectrum can provide reliable estimates of the stellar projected rotational velocity @xmath35 , the effective temperature @xmath36 , and the gravitational acceleration @xmath37 .
we adopted the @xcite solar metallicity models , which assume an atmospheric microturbulent velocity of 10 km s@xmath3 .
we used our relatively higher dispersion red spectra ( @xmath38 ; @xcite ) to find a value of @xmath35 that would produce a match of the model and observed profiles of the strong @xmath21 line .
we formed a rotational broadening function for a linear limb darkening law @xcite using a limb darkening coefficient from the tables of @xcite , and then the @xcite model spectrum was convolved with this and a gaussian instrumental broadening function for direct comparison with the observed profile .
the best fit was made with @xmath7 km s@xmath3 , which is in good agreement with our earlier result of @xmath39 km s@xmath3 based upon the half - width of the @xmath40 line @xcite .
we estimated the effective temperature by comparing the observed to line ratios with those predicted from the grid of model spectra from @xcite .
we measured equivalent widths of the major he lines by gaussian fitting of the observed spectral profiles , and we then formed logarithms of the important ratios used in spectral classification ( table 4 ) .
we made numerical integrations across the same lines in the grid of model spectra by @xcite and then formed plots of @xmath41@xmath42@xmath43 $ ] versus @xmath36 .
we restricted the grid to models with @xmath44 , which is the appropriate gravity for main sequence stars ( see below ) .
table 4 lists the resulting values of @xmath36 made by interpolation with the observed ratios in these curves .
we formed a final mean temperature by double weighting the result from the primary ratio that defines the classification sequence , @xmath24 to @xmath25 , to arrive at @xmath45 kk .
this temperature is somewhat low for spectral type
temperature calibrations based upon pure h and he model atmospheres @xcite but it agrees well with more recent calibrations that rely on blanketed , non - lte models @xcite . the stellar gravity , @xmath37 , is presumably the key parameter defining the appearance of the @xmath26 line for stars of this temperature @xcite .
however , the emergence of this feature as an emission line in more luminous stars is the result of their vigorous stellar winds , and unfortunately the @xcite models are based on static , plane parallel atmospheres that will not predict the @xmath26 line strength correctly in most cases .
instead , we can use the stark broadening of the h balmer lines to help estimate the atmospheric pressure and hence gravity .
h@xmath46 is the most suitable feature in our spectra for this purpose since h@xmath47 is severely blended with @xmath48 at this spectral resolution .
we show in figure 3 the observed and model h@xmath46 profiles for @xmath49 , 4.0 , and 4.3 ( cgs units ) .
these model profiles were convolved with both the rotational and instrumental broadening functions .
the comparison indicates that @xmath50 .
however , we caution that this value may be biased upwards by the presence of lines in the wings of h@xmath46 ( wavelengths from the nist database ) that are probably underestimated by the @xcite models in this particular case .
a reliable abundance analysis is clearly desirable , but will require higher dispersion spectra to account for the effects of line blending in the vicinity of the key cno lines .
however , we made preliminary estimates by calculating line profiles for a range of assumed abundances using the radiative transfer code _
synspec _ and the non - lte stellar atmosphere associated with our derived values of @xmath36 and @xmath37 @xcite .
we found that the equivalent width of the @xmath51 blend ( @xmath30 ) corresponds to a n abundance of @xmath52 dex relative to the solar value .
the equivalent width of @xmath32 ( @xmath53 ) yields a c abundance of @xmath54 dex ( i.e. , sub - solar ) .
these trends are consistent with our expectations for cno processed gas @xcite .
the ultraviolet spectra of o - type stars are a key source of information about their photospheres and stellar winds @xcite , and we were fortunate to obtain the first ever uv spectra of ls 5039 with the _ hubble space telescope
_ space telescope imaging spectrograph ( stis ) in 2002 august . during the first visit to the target ( 2002 aug 13 ) we obtained two exposures ( exposure times of 2087 and 1814 s ) with the g140l grating and one exposure ( 600 s ) with the g230l grating .
the g140l grating covers the wavelength range 1150 1730 with a resolving power of @xmath55 while the g230l grating records the range 1570 3180 with a resolving power of @xmath56 .
a second visit was made 7 days later ( 2002 aug 20 ) to obtain a single exposure ( 2087 s ) with the g140l grating .
these visits correspond to orbital phases 0.145 and 0.708 , respectively , when the o - star primary was oriented at an angle from the line of the nodes in the orbital plane of @xmath57 and @xmath58 , respectively ( 2 ) .
thus , the first visit occurred just past the quadrature phase when the orbital axis was orthogonal to our line of sight and the second visit was made shortly after primary superior conjunction ( compact companion in the foreground ) .
there were no obvious flux or profile variations between these visits , so we formed an exposure weighted average fuv spectrum for examination .
we obtained a single uv spectrum from the archives of the _ international ultraviolet explorer satellite _ of the same o6.5 v((f ) ) reference standard shown above in figure 2 , hd 93146 , to compare to that of ls 5039 .
we rebinned the high dispersion _ iue _ spectrum ( swp 11136 ) of hd 93146 onto the wavelength grid of the stis fuv spectrum , and both the reference and target spectra were rectified to a pseudo - continuum by division of a polynomial fit made through relatively line - free regions .
two sections of these spectra in the immediate vicinity of the strong stellar wind lines @xmath0 and @xmath1 are illustrated in figures 4 and 5 , respectively .
there is generally good agreement between the spectra of ls 5039 and hd 93146 in the uv photospheric lines .
there are differences , however , in the interstellar line strengths and especially in the wind lines .
we find that the @xmath0 p cygni profile has both a deeper blue absorption component ( due to stellar wind material projected against the disk of the star ) and a stronger red emission component ( from scattered light surrounding the star ) than is seen in the spectrum of hd 93146 .
the differences occur in the opposite sense in the @xmath1 wind profile .
the enhanced n and weak c features mirror the results found in the optical spectra , and these same wind line differences are found in other on type stars @xcite .
both wind profiles have shapes that are similar to their counterparts in the spectrum of hd 93146 , and in particular , there are two minima in the absorption part of the @xmath0 feature in the spectra of both stars .
these minima appear much narrower in the high dispersion version of the hd 93146 spectrum , and their spacing corresponds to the highly blueshifted positions of the doublet .
they are examples of narrow ( or discrete ) absorption components ( nac ) that often appear in the uv wind lines @xcite .
detailed time sequences demonstrate that nac usually proceed blueward through the trough and become narrower .
@xcite show that the sum @xmath59 is approximately constant and equal to the extreme velocity found in saturated line cores , and they argue that this quantity is a good estimate of the wind terminal velocity , @xmath60 .
we measured the doppler shifts of both minima in the @xmath0 profile of ls 5039 , and we find @xmath61 km s@xmath3 .
the resolving power of the stis spectra is too low to determine the nac width , so we assumed it is the same as @xcite found for hd 93146 , @xmath62 km s@xmath3 .
thus , we estimate that the wind terminal velocity for ls 5039 is @xmath2 km s@xmath3 , which agrees well with the mean for the o6.5 v subtype of @xmath63 km s@xmath3 @xcite . in some mxrbs
these wind lines display orbital phase variations due to excess ionization of the wind in a zone surrounding the x - ray source ( the _ hatchett - mccray effect _ ; @xcite ) .
the wind profiles of ls 5039 that we observed with hst / stis showed no evidence of changes between the quadrature and superior conjunction phases .
the x - ray flux was moderately low at the time of the observations ( see fig . 7 below ) .
if the x - ray source ionized a large volume that included regions projected against the stellar disk in the superior conjunction phase , then portions of the blue absorption trough would weaken as the scattering ions are ionized to higher levels @xcite .
however , we suspect that the ionized region is small compared to the system dimensions in the case of ls 5039 .
@xcite use the same wind accretion model we discuss in 6 to show that the @xmath64 parameter defining the extent of the ionization zone scales as @xmath65 among mxrbs , where @xmath66 is the separation of the components and @xmath67 is the wind velocity relative to the neutron star . for the separation at the time of the _ hst _ near superior conjunction observation
, we estimate that this parameter falls in the range @xmath68 , based upon a comparison with results for vela x-1 from @xcite .
this corresponds to a strmgren sphere radius of approximately @xmath69 of the system separation ( see eq .
[ 4 ] in @xcite ) .
the minimum projected separation between the outer edge of the ionization sphere and stellar limb at the observed phase is approximately @xmath70 ( for an inclination @xmath71 ) , so that no occultation is expected .
we suspect that any wind profile variations will be restricted to phases very close to superior conjunction and will only be observed if the system has a relatively large orbital inclination ( 6 ) .
the reddening and extinction of ls 5039 are directly related to the issues of the star s distance and space velocity ( 6 ) . @xcite and
@xcite both argue that the reddening is @xmath72 , and @xcite suggest that the distance is @xmath73 kpc based upon an assumed absolute magnitude .
the flux in the uv spectral range is very sensitive to the extinction curve @xcite , so we decided to re - investigate the reddening using the new stis spectra .
the stis spectra are shown in figure 6 in a version rebinned to a resolving power of @xmath74 .
we have included in figure 6 fluxes based upon johnson - cousins and near - ir photometry from @xcite and the 2mass survey @xcite , plus new strmgren @xmath75 , and @xmath76 measurements we made on 2002 june 24 using the ctio 0.9 m telescope and site @xmath77 ccd .
we also observed six standard stars ( hd 105498 , hd 128726 , hd 156623 , hd 157795 , hd 167321 , and hd 216743 ) , taken from lists by @xcite and @xcite , in each band at a minimum of three different airmasses each to calibrate the photometry .
all of the instrumental magnitudes were determined using a large aperture of 8 arcsec , and the photometric transformation equations from @xcite were applied to determine the apparent magnitudes of ls 5039 : @xmath78 , @xmath79 , @xmath80 , and @xmath81 .
the errors in the magnitudes are from the combined estimated instrumental errors and the errors in the transformation constants .
the ir measurements were transformed to fluxes using the nicmos units conversion tool .
we determined fluxes from the optical band photometry using the iraf routine @xmath82 in the _ hst _ _ stsdas / synphot _ package .
this was done by comparing the difference in calculated magnitude between a reddened model spectrum of ls 5039 ( a preliminary version of that discussed below ) and a spectrum of vega with the observed difference in magnitude ( using vega magnitudes from @xcite and @xcite ) .
the model fluxes at the central filter wavelengths were then prorated to match the calculated and observed magnitude differences .
we assumed that the model fluxes from @xcite for the spectroscopically determined parameters @xmath83 kk and @xmath44 provided a reliable estimate of the spectral flux distribution at the stellar photosphere .
we then calculated how the observed spectrum at earth appears based upon the interstellar reddening curve given by @xcite .
there are three parameters that determine the appearance of the transformed spectrum : the reddening , @xmath84 , the ratio of total to selective extinction at @xmath85 , @xmath86 , and the ratio of stellar radius to distance , @xmath87 ( equivalent to half the angular diameter in radians ) .
the first two parameters determine the shape of the extinction curve , while the final parameter is a normalization factor .
we formed @xmath88 residuals for a grid of @xmath84 and @xmath89 values to find the best fit values that minimized @xmath88 .
the best fit , shown by the solid line in figure 6 , was made with @xmath8 , @xmath90 , and @xmath91 ( angular diameter = 0.03 mas ) .
this transformed spectrum fits well the entire fuv to ir range .
we also show @xmath92 deviations from this fit in @xmath84 ( _ dotted line _ ) and in @xmath89 ( _ dashed line _ ) .
there are two nearby stars along the line of sight to ls 5039 that have comparable extinction : bd@xmath93 ( b1 iv ) with @xmath94 and @xmath95 kpc @xcite and bd@xmath96 ( o8 v ) with @xmath97 and @xmath98 kpc @xcite .
our derived extinction of @xmath99 indicates that the distance to ls 5039 is greater than 1.9 kpc , which is consistent with prior estimates ( @xmath100 kpc ; @xcite ) .
in our previous paper @xcite , we estimated the probable secondary mass for a range in primary mass by comparing the observed and predicted x - ray fluxes for wind accretion .
here we re - consider those arguments in light of the new results given above .
the independent parameter is the mass of the o star , @xmath101 , that sets many of the other stellar parameters .
the gravity determination from 3 , @xmath50 , establishes the relationship between mass and radius , @xmath102 , where the ratio of radiative acceleration by electron scattering to gravitational acceleration is @xmath103 .
the specific relationship for ls 5039 based upon the derived effective temperature ( 3 ) is @xmath104 .
the fit of the observed flux distribution yields the angular diameter ( 5 ) , which leads to a distance estimate of @xmath105 kpc ( somewhat closer than our prior estimate of @xmath106 kpc ; @xcite ) .
we estimate the wind mass loss rate using the equivalent width of the h@xmath10 emission line and the method of @xcite .
the h@xmath10 profile appears as a deep absorption line in stars with minimal wind , and the feature becomes progressively filled - in as the mass loss rate increases ( appearing as a pure emission line in stars with strong winds ) .
the mass loss rate depends upon the net emission equivalent width ( i.e. , the difference between the observed and purely photospheric absorption equivalent widths ) , the choice of the wind velocity law exponent @xmath107 , the wind terminal velocity @xmath60 , and the stellar effective temperature @xcite .
we revised our assumptions about several of these parameters .
we adopted @xmath108 kk ( 3 ) and used the detailed optical flux model from @xcite to find the appropriate photospheric equivalent width for h@xmath10 , @xmath109 .
we adopted a velocity law index of @xmath110 that @xcite advocate as the appropriate value for the weak wind case we find here .
finally , we set the terminal velocity to @xmath111 km s@xmath3 , the value we estimated from the wind profiles in the stis spectra ( 4 ) .
most of these changes result in a moderate decrease in the estimated mass loss rate ( table 5 , row 3 ) compared to our earlier results @xcite .
we noted in the previous paper that the h@xmath10 strength appears to be variable on time - scales longer than our observing runs @xcite , which implies that there are long term variations in the mass loss rate .
@xcite have recently shown that there are also long term variations in the x - ray flux that appear to be related to those observed in h@xmath10 .
we show in figure 7 the time evolution of the x - ray flux in the 0.3 10 kev range reported by @xcite .
figure 7 also illustrates the mass loss rate variations determined from the h@xmath10 equivalent widths from @xcite ( 3 measurements ) , @xcite ( 2 measurements ) , and our two recent observations ( @xmath112 and 2.95 for 2002 june 25 and 2003 march 21 , respectively ) .
the @xmath76-axes in figure 7 were scaled to overlap the x - ray flux and mass loss estimates as closely as possible , and we confirm that these two quantities appear to vary in concert in the limited data available .
this result supports the wind accretion model and the validity of the mass loss estimates
. it also points out the need to compare the predicted and observed accretion luminosities at an epoch when nearly simultaneous optical and x - ray data are available . here
we will make the comparison for the closely spaced kpno coude feed @xcite and bepposax @xcite observations made in 2000 october , which correspond to a mass loss rate and x - ray flux minimum in figure 7 .
we calculated the predicted wind accretion x - ray luminosity for a given @xmath101 and a grid of companion masses @xmath113 using the method of @xcite , and then we determined which @xmath113 value produced an x - ray luminosity that matched the observed value .
this procedure was essentially the same as described earlier @xcite , but we improved our accretion rate estimate by finding the mean of rates for equal time samplings around the elliptical orbit and by adopting the observed projected rotational velocity ( 3 ) in the calculation of the wind
companion relative velocity ( see eq . [ 7b ] in @xcite ) .
the resulting best - fit masses are listed in table 5 and are illustrated in a mass plane diagram in figure 8 .
note that the stellar radius is well within its critical roche radius at periastron in all these models , so that mass accretion by roche lobe overflow is probably not a significant process in ls 5039 .
our revised @xmath113 estimates are smaller than but comparable to what we found previously ( see fig . 2 in @xcite ) .
the main reason for this difference is that we relied in the earlier paper on an x - ray luminosity estimate that turned out to be a local maximum ( the first point shown in fig .
7 ) , and this consequently required a larger @xmath113 to account for the accretion luminosity .
the lightly shaded region surrounding the best - fit mass estimates in figure 8 shows how the solutions vary with changes in assumed parameters .
the lower envelope corresponds to the calculations made using a wind velocity law exponent @xmath114 instead of the 1.0 value recommended by @xcite .
a lower exponent corresponds to a higher mass loss estimate ( see fig . 15 in @xcite ) , and this must be balanced by a lower mass accretor in order to match the observed x - ray luminosity ( see eq . [ 21 ] in @xcite ) .
the upper envelope shows the results obtained by assuming that the h@xmath10 equivalent width was @xmath92 larger ( @xmath115 ) than actually observed in 2000 october ( @xmath116 ) .
the increased equivalent width corresponds to a weaker stellar wind ( less emission ) , and in fact the assumed value is close to that for no wind ( @xmath109 ) . here
a reduction in wind strength is compensated by an increase in accretor mass in order to maintain the observed x - ray luminosity .
these tests indicate that our estimates for @xmath113 based upon the accretion model have formal errors of approximately @xmath117 . however , we caution that the actual errors may be larger because the simple wind accretion model does not account for the effects of the ionization of the wind by the x - ray source and because the efficiency parameter for conversion of accreted matter into x - ray radiation , @xmath118 , is only known approximately
. nevertheless , our results for @xmath113 are consistent with a neutron star companion .
there are several considerations that can help limit the range of acceptable mass for the o - star .
the observed reddening of ls 5039 compared to nearby , line - of - sight stars indicates a minimum distance of 1.9 kpc ( 5 ) , and this distance corresponds to a lower limit on the stellar mass of @xmath119 .
the wind accretion solutions at the high mass end require increased mass loss rates and larger system dimensions ; such combinations occur at high inclinations according to the orbital mass function .
the dashed line in figure 8 indicates the inclinations at which eclipses would begin to occur , and we see that for large @xmath101 the best - fit accretion models predict that eclipses should be observed . however , there is no evidence of either x - ray @xcite or optical @xcite eclipses despite significant efforts to find them .
thus , the probable absence of eclipses and the results of the wind accretion model suggest an upper mass limit of approximately @xmath12 ( although masses as large as @xmath120 still fall within the error zone of the wind accretion model results ) .
the wind properties offer additional constraints on the o - star mass .
hot massive stars have a wind momentum product , @xmath121 , that is directly related to the stellar luminosity @xcite , and rows 3 and 4 of table 5 compare our h@xmath10 derived mass loss rate with that predicted by the empirical wind momentum relationship for galactic o - stars .
the agreement appears best at the low mass end , but there is @xmath122 dex scatter in the wind momentum product among the stars that define the empirical relationship , so the full 20 @xmath123 range remains viable .
table 5 also lists the ratio @xmath124 , which typically has a value of 2.6 in o - type stars @xcite .
this criterion would again favor the lower mass range for ls 5039 , @xmath125 , but we caution that there is a large spread in this ratio among hot stars .
indeed , @xcite find a ratio @xmath126 for the comparable on8 v star , hd 14633 , and @xcite quote ratio estimates in the range 0.8 2.8 in other mxrbs .
the final row of table 5 gives estimates of the equatorial velocity @xmath127 that are derived from the projected rotational velocity assuming that the spin axis is aligned with the orbital axis ( see table 6 below ) and that the orbital inclination can be set from the wind accretion models .
the eccentricity of the system probably results from the supernova event that formed the neutron star , and the current value of the eccentricity can be used to infer the pre - sn orbital parameters , assuming that no asymmetrical kick velocity was imparted to the neutron star during the explosion and that the eccentricity has not decreased since the sn @xcite .
we show in the top portion of table 6 estimates of the pre - sn orbital period , @xmath128 , and mass of the sn - progenitor just prior to the sn , @xmath129 , based upon the current orbital elements , mass solutions from the wind accretion model , and the assumption of @xmath130 .
there are several problems with these models that result from our revisions to the distance and eccentricity compared to earlier work @xcite .
the paramount difficulty is the apparent discrepancy between the runaway velocity expected from the eccentricity and masses , @xmath131 ( predicted ) , and the peculiar space velocity , @xmath132 ( derived ) ( based upon the observed proper motion and radial velocity , the distance associated with a given stellar mass @xmath101 , and the method described in @xcite ) . the former is approximately twice as large as the latter , a much greater difference than expected from their formal errors ( approximately 14% and 19% , respectively ) .
the second problem is that the predicted pre - sn orbital period is so short that the roche radius of the o - star is smaller than or equal to the estimated stellar radius .
a third , potential , difficulty is the relatively large mass of the sn - progenitor compared to theoretical expectations @xcite . all of these problems can be alleviated by removing the assumption that the sn kick velocity was zero .
there is now a significant body of evidence that non - zero kick velocities must occur in sn @xcite .
the introduction of a sn kick velocity will dramatically influence the post - sn orbital parameters and system runaway velocity . following the development of @xcite
, we can express the final outcome parameters in terms of two ratios , @xmath133 and @xmath134 where @xmath135 and @xmath136 are the values of the kick velocity and pre - sn relative orbital velocity , respectively .
we also need to specify the direction of the kick in terms of @xmath137 , the angle between the sn - progenitor s orbital motion and the kick - velocity vector projected onto the pre - sn orbital plane , and @xmath138 , the angle between the kick direction and the pre - sn orbital plane ( see fig . 1 in @xcite ) .
with these four parameters selected we can determine the post - sn eccentricity , system runaway velocity , and angle @xmath139 of misalignment between the o - star rotation axis and the current orbital angular momentum vector ( see eq .
[ 2.8 , 2.10 , 2.16 ] in @xcite ) .
we explored the possible solutions applicable to ls 5039 by calculating the predicted final eccentricity and runaway velocity over a grid of @xmath140 , @xmath141 , @xmath137 , and @xmath138 .
our strategy was to compute @xmath142 and @xmath131 for each set of parameters , and if the results in both cases were within @xmath143 of the observed values , then we saved the increment of solid angle corresponding to the test values of @xmath137 and @xmath138 in our grid .
we omitted any models that led to a pre - sn orbit with an o - star roche radius smaller than the observed value and any kick directions that would have the sn - remnant strike the o - star .
our final results are shown in figure 9 for the masses associated with our @xmath144 wind accretion model .
the grayscale intensity in this figure is directly proportional to the integrated solid angle of acceptable solutions for given values of @xmath140 and @xmath141 , so that the figure may be interpreted as a geometrical probability estimate ( white corresponding to no solutions possible , black corresponding to the maximum integrated solid angle of acceptable solutions in the @xmath145 grid ) .
the high frequency `` ripples '' in the image result from the coarseness of our angular grid ( increments of @xmath146 in @xmath137 and @xmath138 ) and our hard boundaries of @xmath147 for acceptable solutions .
there is a broad range of solutions in the @xmath145 plane , but the geometrically favored solutions are concentrated in two regions .
the first occurs along a ridge in the lower portion of the diagram near @xmath148 and @xmath149 .
the solutions correspond to kicks in the direction @xmath150 , i.e. , in approximately the same direction as the orbital motion of the progenitor at the time of the sn .
the pre - sn orbital parameters for such prograde ( @xmath151 ) solutions are given in the middle portion of table 6 , which lists the solid angle weighted average values of initial period , kick velocity , and misalignment .
we find initial periods similar to the no kick case but the progenitor masses are lower .
the associated kick velocities are small and consistent with the low values found by @xcite and by @xcite for other mxrbs .
@xcite suggest that small kick velocities may occur in close binaries in which the sn - progenitor was a rapid rotator at the time of the sn ( i.e. , in those which avoided spin down in a red - supergiant phase ) .
the other geometrically favored region occurs near @xmath152 and @xmath153 , and these solutions correspond to retrograde kicks ( @xmath154 ) with @xmath155 .
the lower portion of table 6 gives the pre - sn orbital parameters for such solutions that have a longer initial period ( and larger roche radii ) than the other solutions .
the kick velocities are much larger than before , but still within the range considered reasonable in other investigations @xcite .
however , the largest kick velocities permitted in figure 9 ( for @xmath156 km s@xmath3 and @xmath157 ) are probably too extreme for most of the commonly accepted kick velocity distributions .
we expect that the binary probably suffered a significant reduction in orbital period during a spiral - in , common envelope phase prior to the sn explosion @xcite . at the conclusion of this stage
the stars were in a short period orbit , and the strong tidal forces would ensure quick spin synchronization and circularization of the orbit @xcite .
we can estimate the average synchronous rotation speed prior to the sn , @xmath158 , using @xmath159 ( table 5 ) and @xmath160 ( table 6 ) , and we list these estimates for the @xmath151 and @xmath154 cases in table 6 .
the synchronous rotation speeds are smaller than the observed value of projected rotational velocity ( @xmath7 km s@xmath3 ) for the @xmath154 models , so we prefer the solutions for the @xmath151 models . the sn progenitor may also have been a rapid rotator as a result of tidal synchronization , and the lower kick velocities associated with the @xmath151 models are consistent with the suggestion that smaller kicks occur in rapid rotators @xcite
. however , regardless of the choice of the model for the sn event in ls 5039 , it is clear that a modest , non - zero kick velocity is required to explain consistently the observed eccentricity and runaway velocity .
it is somewhat surprising that we find clear evidence of cno processed gas in the photosphere of ls 5039 while the spectral signature of cno processed gas is generally absent in other mxrbs ( with the interesting exceptions of vela x-1 = hd 77581 @xcite and 4u170037 = hd 153919 @xcite ) .
the n - enriched gas has two possible sources , nuclear processed gas from the interior of the o - star and/or processed gas from the envelope of the neutron star progenitor .
here we briefly discuss the issues associated with both sources of enriched gas .
the origin of the he and n enrichment in the obn and other evolved massive stars is probably related to rotationally induced mixing of the stellar interiors . @xcite and @xcite have constructed stellar models that include rotation , and they find that significant mixing of cno processed gas into the photosphere can occur over a main sequence lifetime in rapidly rotating , massive stars .
@xcite found that the obn stars have a projected rotational velocity distribution consistent with the assumption that they represent a class of rapid rotators , and they showed that he is enhanced in several very rapidly rotating o - type stars .
the projected rotational velocity of ls 5039 ( @xmath7 km s@xmath3 ) is not unusually large for o - type stars @xcite , and the star s spin axis inclination , @xmath161 , would need to be relatively small if the star is currently a rapid rotator ( see @xmath127 estimates in table 5 ) .
we showed above that the star could have been a rapid rotator due to tidal interactions prior to the sn ( see the @xmath151 solutions in table 6 ) .
nevertheless , we doubt that significant rotationally induced mixing has occurred since then because the elapsed time since the sn implied by the kinematical age ( @xmath162 my ; @xcite ) is much smaller than the characteristic time for large scale mixing ( comparable to the main sequence lifetime , @xmath163 my ; @xcite ) . therefore , we doubt that the n - enrichment of ls 5039 is due to internal mixing unless the o - star was born as a rapid rotator and mixing has progressed for a long time ( i.e. , the lifetime of the neutron star progenitor plus the kinematical age ) .
the alternative explanation is that the nuclear processed gas was accreted from the sn progenitor .
the system had a short orbital period prior to the sn explosion ( table 6 ) that probably resulted from a spiral - in during a common - envelope phase ( cep ) @xcite .
the surviving o - star probably accreted relatively little material during the cep because of its short duration , and accretion following the cep must have been very limited because mass transfer at that point would have caused an expansion of the orbit @xcite . instead we suspect that the cno - enriched gas was transferred from the progenitor during a period immediately prior to the start of the cep .
massive donor stars with radiative envelopes can in some circumstances sustain a stable mass transfer rate through roche lobe overflow until the envelope is lost @xcite , and it is possible that several solar masses of enriched gas were accreted by the o - star gainer during this evolutionary stage .
other explanations such as direct wind fed accretion @xcite and accretion of the impacting sn shell @xcite are less compelling because the amount of enriched gas accreted in both cases is relatively minor .
many of the black hole mxrbs are also thought to have evolved from widely separated binaries into compact systems through a common envelope stage @xcite , and it is curious that these do not generally show the n enrichment we find in ls 5039 .
the explanation may be related to the youth of ls 5039 @xcite .
we suspect that thermohaline mixing @xcite , rotational mixing , and stellar wind mass loss will act to remove the enriched gas from the atmosphere so that the n enrichment may decline with the time elapsed since the sn .
ls 5039 may represent an unusually young mxrb in which these processes have yet to make a significant change in the atmospheric abundances , so that the products of the intense binary interaction remain vividly intact in the stellar atmosphere .
we thank the staffs of ctio , kpno , and mcdonald observatory for their assistance in making these observations possible .
we are especially grateful to dr .
todd henry and the smarts consortium for the ctio 1.5 m time awarded to this project .
we also thank dr .
thierry lanz and dr .
ivan hubeny for the use of and advice about their codes _ tlusty _ and _ synspec _ , dr . nolan walborn for comments about the on classification , dr . marc rib for communicating his radial velocity results in advance of publication , and the referee dr . philipp podsiadlowski for his insights on binary evolution .
pjw is grateful for continuing hospitality at the department of astrophysical sciences , princeton university .
support for this work ( hst proposal number 9449 ) was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
additional financial support was provided by the national science foundation through grant ast@xmath1640205297 ( drg ) .
institutional support has been provided from the gsu college of arts and sciences and from the research program enhancement fund of the board of regents of the university system of georgia , administered through the gsu office of the vice president for research .
this research made use of the multimission archive at the space telescope science institute ( mast ) , nasa s astrophysics data system bibliographic service , and the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and nsf .
berger , d. h. , & gies , d. r. 2001 , , 555 , 364 bessell , m. s. 1983 , , 95 , 480 bhattacharya , d. , & van den heuvel , e. p. j. 1991 , physics rep . , 203 , 1 bolton , c. t. , & rogers , g. l. 1978 , , 222 , 234 brandt , n. , & podsiadlowski , ph .
1995 , , 274 , 461 claret , a. , & cunha , n. c. s. 1997 , , 318 , 187 clark , j. s. , et al .
2001 , , 376 , 476 clark , j. s. , goodwin , s. p. , crowther , p. a. , kaper , l. , fairbairn , m. , langer , n. , & brocksopp , c. 2002 , , 392 , 909 clausen , j. v. , larsen , s. s. , garcia , j. m. , gimnez , a. , & storm , j. 1997 , , 122 , 559 cousins , a. w. j. 1987 , south african astron .
circ . , 11 , 93 cutri , r. m. , et al .
2003 , the 2mass all - sky catalog of point sources ( pasadena : univ .
ipac ) dearborn , d. s. p. 1977 ,
aplett , 19 , 15 de greve , j. p. , & de loore , c. 1992 , , 96 , 653 de loore , c. , & vanbeveren , d. 1994 , , 292 , 463 fitzpatrick , e. l. 1999 , , 111 , 63 fryxell , b. a. , & arnett , w. d. 1981 , , 243 , 994 gies , d. r. , huang , w. , & mcswain , m. v. 2002 , , 578 , l67 gray , d. f. 1992 , the observation and analysis of stellar photospheres , 2nd ed .
( cambridge : cambridge univ .
press ) gray , r. o. 1998 , , 116 , 482 hatchett , s. , & mccray , r. 1977 , , 211 , 552 he , l. , whittet , d. c. b. , kilkenny , d. , & spencer jones , j. h. 1995 , , 101 , 335 heger , a. , & langer , n. 2000 , , 544 , 1016 henden , a. a. , & kaitchuck , r. h. 1982 , astronomical photometry ( new york : van nostrand reinhold ) hilditch , r. w. 2001 , an introduction to close binary stars ( cambridge : cambridge univ .
press ) hjellming , m. s. , & webbink , r. f. 1987 , , 318 , 794 howarth , i. d. , & prinja , r. k. 1989 , , 69 , 527 howarth , i. d. , siebert , k. w. , hussain , g. a. j. , & prinja , r. k. 1997 , , 284 , 265 howarth , i. d. , & smith , k. c. 2001 , , 327 , 353 hughes , a. , & bailes , m. 1999 , , 522 , 504 kaper , l. , hammerschlag - hensberge , g. , & van loon , j. th .
1993 , , 279 , 485 kaper , l. , henrichs , h. f. , nichols , j. s. , & telting , j. h. 1999 , , 344 , 231 kilkenny , d. 1993 , south afr . astron . obs .
circ . , 15 , 53 kudritzki , r .-
p . , & puls , j. 2000 , , 38 , 613 lamers , h. j. g. l. m. , & cassinelli , j. p. 1999
, introduction to stellar winds ( cambridge : cambridge univ . press ) lamers , h. j. g. l. m. , van den heuvel , e. p. j. , & petterson , j. a. 1976 , , 49 , 327 langer , n. , wellstein , s. , & petrovic , j. 2003 , in a massive star odyssey , from main sequence to supernova , proc .
k. a. van der hucht , a. herrero , & c. esteban ( san francisco : a.s.p . ) , 275 lanz , t. , & hubeny , i. 2003 , , 146 , 417 martins , f. , schaerer , d. , & hillier , d. j. 2002 , , 382 , 999 mathys , g. 1989 , , 81 , 237 mcswain , m. v. , & gies , d. r. 2002 , , 568 , l27 mcswain , m. v. , gies , d. r. , riddle , r. l. , wang , z. , & wingert , d. w. 2001 , , 558 , l43 meynet , g. , & maeder , a. 2000 , , 361 , 101 morbey , c. l. , & brosterhus , e. b. 1974 , , 86 , 455 nelemans , g. , tauris , t. m. , & van den heuvel , e. p. j. 1999 , , 352 , l87 paredes , j. m. , mart , j. , rib , m. , & massi , m. 2000 , science , 288 , 2340 paredes , j. m. , rib , m. , ros , e. , mart , j. , & massi , m. 2002 , , 393 , l99 penny , l. r. 1996 , , 463 , 737 pfahl , e. , rappaport , s. , podsiadlowski , ph .
, & spruit , h. 2002 , , 574 , 364 podsiadlowski , ph . , rappaport , s. , & han , z. 2003 , , 341 , 385 podsiadlowski , ph . ,
rappaport , s. , & pfahl , e. d. 2002 , , 565 , 1107 prinja , r. k. , barlow , m. j. , & howarth , i. d. 1990 , , 361 , 607 puls , j. , et al .
1996 , , 305 , 171 reig , p. , rib , m. , paredes , j. m. , & mart , j. 2003 , , 405 , 285 rib , m. , paredes , j. m. , romero , g. e. , benaglia , p. , mart , j. , fors , o. , & garca - snchez , j. 2002 , , 384 , 954 rib , m. , reig , p. , mart , j. , & paredes , j. m. 1999 , , 347 , 518 roberts , d. h. , lehr , j. , & dreher , j. w. 1987 , , 93 , 968 schnberner , d. , herrero , a. , becker , s. , eber , f. , butler , k. , kudritzki , r. p. , & simon , k. p. 1988
, , 197 , 209 stellingwerf , r. f. 1978 , , 224 , 953 taam , r. e. , & sandquist , e. l. 2000 , , 38 , 113 vacca , w. d. , garmany , c. d. , & shull , j. m. 1996 , , 460 , 914 vanbeveren , d. , & de loore , c. 1994 , , 290 , 129 van den heuvel , e. p. j. , & van paradijs , j. 1997 , , 483 , 399 van loon , j. th . , kaper , l. , & hammerschlag - hensberge , g. 2001 , , 375 , 498 wade , r. a. , & rucinski , s. m. 1985 , , 60 , 471 walborn , n. r. 1976 , , 205 , 416 walborn , n. r. , & fitzpatrick , e. l. 1990 , , 102 , 379 walborn , n. r. , lennon , d. j. , heap , s. r. , lindler , d. j. , smith , l. j. , evans , c. j. , & parker , j. w. 2000 , , 112 , 1243 walborn , n. r. , nichols - bohlin , j. , & panek , r. j. 1985 , international ultraviolet explorer atlas of o - type spectra from 1200 to 1900 ( nasa rp-1155 ; washington , dc : nasa ) wellstein , s. , & langer , n. 1999 , , 350 , 148 wellstein , s. , langer , n. , & braun , h. 2001 , , 369 , 939 zahn , j .-
1977 , , 57 , 383 llrcccc 2002 jun 07 & mcd 2.7 m / lcs & 1200/
4000/1 & & ti & 2500 & 40604770 + 2002 jun 2328 & kpno 0.9 m / cf & 316/12000/2 & og550 & ti & 11630 & 65326708 + 2003 mar 17 & ctio 1.5 m / cs & 831/ 8000/2 & cuso@xmath165 & loral & 2320 & 40714741 + 2003 mar 1820 & ctio 1.5
m / cs & 831/ 8000/2 & bg39 & loral & 2320 & 40714741 + 2003 mar 21 & ctio 1.5 m / cs & 831/ 8000/1 & gg495 & loral & 1800 & 54886799 + 2003 apr 26 & mcd 2.7 m / lcs & 1200/ 4000/1 & & loral & 2500 & 40714790 + 2003 apr 2728 & mcd
2.7 m / lcs & 1200/ 4000/1 & & ti & 2500 & 40564764 + lcccl 2432.902 & 0.901 & @xmath166 & @xmath167 & mcd blue + 2448.858 & 0.505 & @xmath168 & @xmath169 & kpno red + 2449.839 & 0.727 & @xmath170 & @xmath171 & kpno red + 2450.811 & 0.946 & @xmath172 & @xmath173 & kpno red + 2453.801 & 0.622 & @xmath174 & @xmath175 & kpno red + 2715.894 & 0.829 & @xmath176 & @xmath177 & ctio blue + 2716.902 & 0.057 & @xmath178 & @xmath179 & ctio blue + 2717.901 & 0.282 & @xmath180 & @xmath181 & ctio blue + 2718.839 & 0.495 & @xmath182 & @xmath183 & ctio blue + 2719.881 & 0.730 & @xmath184 & @xmath185 & ctio red + 2755.901 & 0.867 & @xmath186 & @xmath187 & mcd blue + 2756.849 & 0.081 & @xmath188 & @xmath189 & mcd blue + 2756.957 & 0.106 & @xmath190 & @xmath191 & mcd blue + 2757.956 & 0.331 & @xmath192 & @xmath193 & mcd blue + lc @xmath194 ( d ) & 4.4267 @xmath195 0.0005 + @xmath196 ( hjd - 2,450,000 ) & 2756.49 @xmath195 0.07 + @xmath197 ( km s@xmath3 ) & 17.6 @xmath195 1.3 + @xmath198 ( km s@xmath3 ) & 4.1 @xmath195 0.8 + @xmath142 & 0.48 @xmath195 0.06 + @xmath199 ( deg ) & 268 @xmath195 10 + rms ( km s@xmath3 ) & 5.5 + @xmath200 ( @xmath201 ) & 0.0017 @xmath195 0.0005 + @xmath202 ( @xmath203 ) & 1.36 @xmath195 0.12 + lccc @xmath159 ( @xmath203 ) & 7.0 & 8.5 & 9.9 + @xmath124 & 2.5 & 2.2 & 2.1 + @xmath216 ( @xmath201 y@xmath3 ) & @xmath217&@xmath218&@xmath219 + @xmath216 ( @xmath201 y@xmath3 ) & @xmath220&@xmath221&@xmath222 + @xmath113 ( @xmath201 ) & 1.3 & 1.4 & 1.5 + @xmath223 ( @xmath203 ) & 10.0 & 11.9 & 13.3 + @xmath224 ( kpc ) & 1.9 & 2.4 & 2.7 + @xmath127 ( km s@xmath3 ) & 198 & 168 & 149 + lcccc @xmath128 ( d ) & 0 & 1.39 & 1.39 & 1.39 + @xmath129 ( @xmath201 ) & 0 & 11.6 & 16.6 & 21.5 + @xmath131 ( predicted ) ( km s@xmath3 ) & 0 & 181 & 210 & 233 + @xmath132 ( derived ) ( km s@xmath3 ) & & 95 & 118 & 137 + & & & & + @xmath160 ( d ) & @xmath225 & 1.52 & 1.58 & 1.63 + @xmath129 ( @xmath201 ) & @xmath225 & 6.9 & 10.4 & 14.5 + @xmath226 ( km s@xmath3 ) & @xmath225 & 58 & 51 & 41 + @xmath227 ( deg ) & @xmath225 & 3 & 2 & 1 + @xmath158 ( km s@xmath3 ) & @xmath225 & 232 & 273 & 306 + & & & & + @xmath160 ( d ) & @xmath164 & 4.61 & 4.60 & 4.81 + @xmath129 ( @xmath201 ) & @xmath164 & 7.4 & 11.5 & 16.0 + @xmath226 ( km s@xmath3 ) & @xmath164 & 182 & 211 & 226 + @xmath227 ( deg ) & @xmath164 & 13 & 12 & 11 + @xmath158 ( km s@xmath3 ) & @xmath164 & 77 & 94 & 104 + | we present an investigation of new optical and ultraviolet spectra of the mass donor star in the massive x - ray binary ls 5039 .
the optical band spectral line strengths indicate that the atmosphere is n - rich and c - poor , and we classify the stellar spectrum as type on6.5 v((f ) ) .
the n - strong and c - weak pattern is also found in the stellar wind p cygni lines of @xmath0 and @xmath1 ( narrow absorption components in the former indicate that the wind terminal velocity is @xmath2 km s@xmath3 ) .
we suggest that the n - enrichment may result from internal mixing if the o - star was born as a rapid rotator , or the o - star may have accreted n - rich gas prior to a common - envelope interaction with the progenitor of the supernova .
we re - evaluated the orbital elements to find an orbital period of @xmath4 d. we compared the spectral line profiles with new non - lte , line - blanketed model spectra from @xcite , from which we derive an effective temperature @xmath5 kk , gravity @xmath6 , and projected rotational velocity @xmath7 km s@xmath3 .
we fit the uv , optical , and ir flux distribution using a model spectrum and extinction law with parameters @xmath8 and @xmath9 .
we confirm the co - variability of the observed x - ray flux and stellar wind mass loss rate derived from the h@xmath10 profile @xcite , which supports the wind accretion scenario for the x - ray production in ls 5039 .
wind accretion models indicate that the compact companion has a mass @xmath11 , consistent with its identification as a neutron star .
we argue that the o - star has a mass in the range 20 @xmath12 ( based upon a lower limit for the distance and the lack of eclipses ) .
the observed eccentricity and runaway velocity of the binary can only be reconciled if the neutron star received a modest kick velocity due to a slight asymmetry in the supernova explosion ( during which @xmath13 was ejected ) . |
when the phase coherence of electrons passing through a mesoscopic system is retained , a number of quantum interference phenomena will occur .
recent advances in nano - technologies have attracted much attention to the quantum coherence phenomena in the resonant tunneling processes of the quantum dot(qd ) systems,@xcite in which the typical length scale can be shorter than or comparable to the mean free path of electrons and the wave nature of electrons plays a decisive role . in the past decade , the widely adopted method of studying the phase coherence of a traversing electron through a qd is to measure the magnetic - flux - dependent current through an aharonov - bohm(ab ) interferometer by inserting the qd into one of its arms.@xcite the observed magnetic - oscillation of the current will be the indication of the coherent transport through the qd , provided that at least partial coherence of electrons is kept .
@xcite fano resonances is another good probe for the phase coherence in the qd system.@xcite it is known that the fano resonance stems from quantum interference between resonant and nonresonant processes,@xcite and manifests itself in spectra as asymmetric lineshape in a large variety of experiments . unlike the conventional fano resonance,@xcite the fano effect in qd system
has its advantage in that its key parameters can be readily tuned .
suppose that a discrete level inside the qd acts as a breit - wigner - type scatter and is broadened by a factor of @xmath0 due to couplings with the continua in leads .
the key to realize the fano effect in the conductance spectra is that , within @xmath0 the phase of the electron should smoothly change by @xmath1 on the resonance.@xcite the first observation of the fano lineshape in the qd system was reported by gres _
et al_.@xcite in the single - electron - transistor experiments .
recently , k. kobayashi _ et al . _ carried out research on magnetically and electrostatically tunable fano effect in a qd embedded in an ab ring,@xcite and a.c .
et al_. investigated a tunable fano interferometer consisting of a qd coupled to a one - dimensional channel via tunneling and observed the coulomb - modified fano resonances.@xcite the double quantum dot(dqd ) system , including the series - coupled dqd@xcite and parallel - coupled dqd,@xcite makes the quantum transport phenomena rich and varied .
the parallel - coupled dqd system is of particular interest , in which two qd s are respectively embedded into opposite arms of the ab ring , coupled each other via barrier tunneling , and coupled to two leads roughly equally . as a controllable two - level system , it is appealing for the parallel - coupled dqd system to become one of promising candidates for the quantum bit in quantum computation based on solid state devices.@xcite the entangled quantum states required for performing the quantum computation demand a high degree of phase coherence in the system.@xcite being a probe of phase coherence,@xcite if the fano effect in the parallel - coupled dqd system is tunable and exhibits the swap effect , it is certainly of practical importance .
inspired by recent experimental advances in the parallel dqd,@xcite several groups have attempted to address this multi - path system theoretically , and predicted the fano resonance in the parallel - coupled dqd system.@xcite however , it seems that a physically transparent picture for the fano effect in the dqd system is still lacking .
for example , it is unclear what the resonant and nonresonant channels are in this fano system .
moreover , a systematic study is required for exploring various possibilities of tuning fano effect with external fields . in this paper , we intend to provide with a natural yet simple explanation for the fano effect in the parallel - coupled dqd , and propose several ways to control the fano resonance in the conductance spectra by the electrostatic and magnetic approaches . the paper is organized as follows . in sec.[sec : model ] , a widely used two - level fano - anderson model is introduced with an inter - dot coupling term added . since the coupled quantum dots may be considered as an artificial qd - molecules,@xcite an effective hamiltonian in terms of the bonding and antibonding states of the qd - molecule may form an appropriate working basis .
thus with the help of the green function technique and the equation of motion method,@xcite the density of states is calculated in three asymmetric configurations classified according to the spatial symmetry of the dot - lead coupling.@xcite in sec .
[ sec : calculation ] , the conductance formula is derived for this system,@xcite whereby the fano lineshape in conductance spectra is calculated in the absence of the magnetic flux . in sec.[sec : explain ] , a simple mechanism explaining the fano lineshape in the dqd conductance spectrum is presented .
then , the possibilities of tuning the conductance lineshape by various electrostatic and magnetic methods are described in details , and several novel effects are predicted .
most importantly , by tuning the the total magnetic flux , or the flux difference between the left and right parts of the ab ring , the swap effect between two resonance peaks in the conductance spectra is predicted , which might be of potential application as a type of c - not gate in the quantum computation . finally , a brief summary is drawn and presented .
we start with the fano - anderson model for the parallel - coupled dqd where the discrete states in two quantum dots are coupled each other via tunneling .
( fig.[fig : model]a ) then the hamiltonian reads @xmath2 the @xmath3 in eq.([hamiltonian ] ) represents the noninteracting electron gas in the left(l ) and right(r ) leads , @xmath4 where , @xmath5 are the creation and annihilation operators for a continuum in the lead @xmath6 with energy @xmath7 .
the @xmath8 in eq.([hamiltonian ] ) describes the qd electrons and their mutual coupling in the dqd , _
i.e. _ @xmath9 the first term in eq.([hd1 ] ) , @xmath10 @xmath11 , represents the create ( annihilate ) operator of the electron with the energy @xmath12 in the dot @xmath13 .
the second and third terms in eq.([hd1 ] ) denote the inter - dot coupling , in which @xmath14 is the coupling strength taken as a real parameter , and @xmath15 denotes a phase shift related to the flux difference between the left and right sub - rings .
the @xmath16 in eq.([hamiltonian ] ) represents the tunneling coupling between the qd and lead electrons , @xmath17 where the tunneling matrix element @xmath18 , and @xmath19 . here for the sake of simplicity , @xmath20 is assumed to be independent of @xmath21 , and the phase shift due to the total magnetic flux threading into the ab ring , @xmath22 , is assumed to distribute evenly among 4 sections of the dqd - ab ring .
namely , @xmath23 , where the flux quantum @xmath24 .
thus , @xmath25 . in the following calculation
, we define the line - width matrix as @xmath26 ( @xmath27 and @xmath28 . according to fig.[fig : model](a ) , the line - width matrices in the qd representation read @xmath29 and @xmath30 where @xmath31 is short for @xmath32 . to make the physical picture clearer and formalism simpler
, it is attractive to introduce a qd - molecule representation by transforming two tunneling - coupled qd levels into the bonding and antibonding states of the qd - molecule .
the operator for a molecule state can be expressed as a linear superposition of the qd operators as @xmath33 where , @xmath34 and @xmath35 are referred to as the annihilation operators for the bonding and antibonding states of the artificial qd - molecule . and the parameter @xmath36 is defined as @xmath37 .
$ ] for mathematical simplicity , in the following only the symmetric case is studied , _
i.e. _ @xmath38 , and thus @xmath39 .
then the hamiltonian for coupled dots is decoupled as @xmath40 in the molecular - state representation , the tunneling hamiltonian between the leads and dqd is rewritten into @xmath41 where the effective tunneling matrix elements are @xmath42 now , the dqd system has been mapped into a system of two independent molecular states , which are connected to leads , respectively(_cf . _
fig.[fig : model]b ) . in the molecular - state representation ,
the linewidth matrices read @xmath43 i.e. @xmath44 and @xmath45 to estimate the broadening of the molecular level due to its couplings to leads , let us first calculate the density of states(dos ) for each state .
the retarded green function for the molecular state is defined as @xmath46 . with the equation of motion approach@xcite
, we have @xmath47 where the imaginary part of the self - energy is equal to @xmath48 the local density of states is defined as the imaginary part of the retarded green function as @xmath49 ^ 2+(\gamma_{\pm})^2},\ ] ] which is the lorentzian peaked at the molecular level .
three configurations for the dqd system according to the asymmetric couplings between two dots and leads@xcite that we focus on are shown in fig.[fig : structure ] : ( a)structure 1 , in which @xmath50 ; ( b ) structure 2 , in which @xmath51 ; and ( c ) structure 3 , where @xmath52 . here
the @xmath53 is taken as an energy unit .
these typical structures for the dqd system are basic , yet convenient to analyze theoretically .
, where @xmath54 which is taken as an energy scale.,scaledwidth=50.0% ] fig.[fig : dos ] shows how the calculated dos of two molecular states changes with the total magnetic flux @xmath22 in the structure 1 and 2 .
it is noticed that the structure 1 and 2 share the identical dos .
two points are worth to pointing out .
firstly , since the full width at half maximum @xmath55 for each molecular state depends on the parameters @xmath56 , @xmath22 , and @xmath15 , the broadenings of molecular states could be tuned not only by the dot - lead coupling strength and total magnetic flux , but also by the flux difference .
secondly , the broadening of one molecular state is always accompanied by the shrinking of the other molecular state because the trace of the matrix of @xmath57 is an invariant , as explicitly shown by eqs .
( [ mogammal ] ) and ( [ mogammar ] ) .
thirdly , in the absence of the magnetic flux ( @xmath58 ) , @xmath59 in the structure 3 , i.e. the antibonding state is totally decoupled from the leads and possesses an infinite lifetime .
the finite @xmath22 introduced by the magnetic flux in this structure results in the finite coupling between the antibonding state and leads , and thus a finite width for the antibonding state .
to study the transport through the dqd system , the conductance at zero temperature is derived , which , on the basis of the non - interacting molecular levels , can be reduced to the landauer - bttiker formula @xcite @xmath60 where , @xmath61 is the fermi energy at both leads at equilibrium . in the absence of coulomb interaction between the electrons on the dots ,
the transmission @xmath62 is expressed as@xcite @xmath63,\ ] ] where the retarded and advanced green functions , @xmath64 and @xmath65 , and the linewidth matrices @xmath66 can be expressed in molecular states representation or in dot levels representation , respectively .
the retarded green function in the dot level representation is defined as @xmath67 , and its fourier transformation @xmath68 satisfies @xmath69|d^{\dagger}_j\rangle\rangle,\ ] ] which generates a closed set of linear equations for @xmath70 .
the solution of @xmath64 is given by @xmath71 the advanced green function is the hermite conjugate of the retarded green function .
put eq.([grdot ] ) into eq.([transmission ] ) and after some algebra , the transmission probability through the dqd system is found to be @xmath72 ^ 2+\gamma^2_+\ } \{[\varepsilon-(\varepsilon_0-t_c)]^2+\gamma^2_-\}},\nonumber\\\end{aligned}\ ] ] where @xmath73 the conductance through the dqd system as expressed in eq.([conduct1 ] ) depends in general on the energy levels of two dots , dot - lead coupling , inter - dot coupling , and phase shift induced by the magnetic flux . without the magnetic flux ,
i.e. @xmath58 , and @xmath74 , eq.([conduct1 ] ) recovers to the result by guevara _
et al_.@xcite ; while in the limit of vanishing inter - dot coupling ( @xmath75 ) , the result of kubala _ et al_.@xcite is repeated . in the molecular state representation
, the total conductance can also be divided into @xmath76 where the transmission via each molecular state is @xmath77 ^ 2+\gamma_{\pm}^2},\end{aligned}\ ] ] and the interference term is given by @xmath78[\varepsilon-(\varepsilon_0- t_c)+i\gamma_{- } ] } \}.\nonumber\\\end{aligned}\ ] ]
as shown in fig.[fig : fano3 ] , the total conductance through the parallel - couple dqd system consists of a breit - wigner peak and a fano peak , which is quite different from the dos where two lorentzians are superposed . based on the molecular level representation formulated above , let us first explain how the fano interference is produced in somewhat details , then show how to tune it . although similar observations have been reported in refs .
, it seems that till now no simple and transparent explanation is available on the formation of the fano lineshape in this structure . , and @xmath54 .
notice that the conductance spectra in two structures have quite different interference pattern though the identical dos as in fig.[fig : dos ] .
, scaledwidth=50.0% ] when a discrete level is buried into a continuum , the coupling between the discrete and continuous states gives rise to the renormalization of the states of whole system .
the phase of the renormalized wave function varies by @xmath79 swiftly as the energy transverses an interval @xmath80 around the discrete level , where @xmath0 is the broadening of the discrete level due to coupling with the continua.@xcite then , if there is a reference channel whose phase changes little in the interval of @xmath0 around the discrete level , the quantum interference above and below the resonance level will be responsible for the fano lineshape in the conductance spectra , for example , the experimentally observed fano lineshape in the hybrid system of a qd and a reference arm.@xcite in the present dqd multi - path system , it is not simple to identify the resonant and the reference channel .
but , on the basis of the molecular level representation , it is straightforward to interpret the fano resonance in the dqd system in terms of the interference between two channels .
usually , two molecular levels are coupled to the leads unequally . in the absence of magnetic flux , eq.([gfmo5 ] ) simply reduces to @xmath81 i.e. , the broadening of one level is always accompanied by the shrinking of the other . the molecular level associated with a wider band
can be referred to as the strongly - coupled one , while that with narrow band is referred to as the weakly - coupled level .
suppose that the broadening of the strongly - coupled level entirely covers the band width with the weakly - coupled level , and the phase shift for the strongly - coupled channel is negligibly small around the weakly - coupled level , then a phase shift of @xmath1 across the weakly - coupled level can be detected with characteristic of the fano lineshape .
namely , the waves through two channels interfere constructively for electron with energy on one side of the weakly - coupled level ; while interfere destructively on the other side . as a result
, the fano lineshape shows up around the weakly - coupled state .
it is worth pointing out that the @xmath1 phase shift also happens around the strongly - coupled level , but on a energy scale much larger than that around the weakly - coupled level ( fig.[fig : dos ] ) .
that is why usually there is only one fano peak around the weakly - coupled level , and the other peak around the strongly - coupled level only exhibits little asymmetry .
compared with the fano effect in the hybrid system,@xcite where the nonresonant channel served by a quantum point contact detects the @xmath1 phase shift around the resonant tunneling channel through the qd , in the dqd system , the reference channel is the `` less resonant '' tunneling channel on one side of the strongly - coupled level , and the other `` more resonant '' channel through the weakly - coupled level is accompanied with a swift @xmath1 phase shift within a small energy region .
the fano lineshape in the present system can be tuned by the applied electrostatic and magnetic fields .
compared with the one dot and one arm case , where the magnetic field affects only the phase of the electron , but not the interaction strength,@xcite in the parallel - coupled dqd , not only the phase but also the magnitude of the effective coupling between the molecular states and leads can be tuned by the magnetic flux .
of course , one should keep in mind that the inter - dot coupling is the prerequisite to the tunable fano effect . , where the fermi energy is roughly at the antibonding level .
( b),(c),and ( d ) simulation of the normalized fano lineshape with different asymmetric factors @xmath82 as compared to the results obtained by green functions in ( a).,scaledwidth=50.0% ] the inter - dot coupling @xmath14 can be tuned by adjusting the height and thickness of the tunneling barrier between two dots through gate voltages .
so can be done the dot - lead coupling @xmath83 .
the interference between the resonance and the reference channels can be described as @xmath84 where @xmath85 and @xmath86 denote respectively the transmission amplitude via the resonance and the reference channels , the detuning @xmath87 , and the asymmetric factor @xmath88 .
it is well known that when the asymmetric factor @xmath82 is small , eq.([fanolineshape ] ) gives rise to the asymmetric fano lineshape ; while for large @xmath82 , it tends to the symmetric lorentzian .
[ fig : tuningtc](a ) depicts the conductance spectra for three different inter - dot coupling strengths in structure 2 . as for comparison , the normalized fano lineshape ( multiplied by a normalized factor @xmath89 )
is also plotted in fig.[fig : tuningtc](b),(c ) , and ( d ) , demonstrating how the fano lineshape evolves with increasing @xmath82 .
it is obvious that as the inter - dot coupling increases , the bonding and antibonding splitting increases . as a result ,
the amplitude through the channel associated with the strongly - coupled level but at the energy range of antibonding peak , @xmath86 , decreases , and since @xmath82 is inversely proportional to @xmath86 , the asymmetric factor @xmath82 increases .
this further confirms our explanation above about the origin of fano effect in this parallel - coupled dqd system . when tuning the dot - lead coupling strength @xmath83 by tuning the gate voltage ,
the structure studied is changed , and the fano lineshape is changed accordingly .
for example , when the strong and weak coupling in the linewidth matrix is adjusted such that the structure 1 is transformed into the structure 2 , ( see fig.[fig : fano3]b ) , noticeably , the tail direction of the fano peak is flipped .
this delicate change can be understood in terms of the product of the effective tunneling matrix elements @xmath90 in the wide band limit , a linewidth matrix element is proportional to the product of two dot - lead tunneling matrix elements . according to fig.[fig : structure ] ,
the expression above for the structure 1 differs from that for structure 2 by a minus sign , which implies that , compared to structure 2 , an extra flux of @xmath1 threads the loop in the structure 1(fig.[fig : jump ] ) .
thus , the fano lineshape in structure 1 is just opposite to that in structure 2 , i.e. , if two channels interfere with each other constructively in structure 1 , then destructively in structure 2 ; and vice versa .
.,scaledwidth=40.0% ] coupled quantum dot systems have been proposed to materialize quantum bit for quantum computation@xcite .
the swap operation is an important element to the controlled - not gate , which is the key to implementing the quantum computation .
@xcite recently , a new mechanism has been proposed to realize the swap operation in the parallel - coupled dqd system by using the time - dependent inter - dot spin superexchange j(t ) , which flips the singlet and triplet states formed by two localize electrons.@xcite in the following we will discuss a new type of swap operation in the dqd system , which flips two quantum states by tuning the magnetic flux , including tuning the total flux @xmath22 or the flux difference @xmath91 between the left and right sub - rings . for the purpose of comparison , only tuning one parameter , and letting the other alone .
tuned for three asymmetric configurations .
, scaledwidth=50.0% ] first , we tune the total flux @xmath22 and let @xmath74 .
when @xmath92 ( n is an integer ) , eq.([gfmo5 ] ) turns out to be @xmath93 compared with the case of null flux , or @xmath94 [ eq.([gamma_pm ] ) ] , the widths of the bonding and antibonding states are interchanged .
fig.[fig : tuningphi ] demonstrates the evolution of conductances with changing the total magnetic flux in three asymmetric configurations for zero @xmath15 . in this circumstance
the fano and breit - wigner peaks in the conductance spectra of the structures 1 and 2 have been exchanged each other , when changing @xmath22 from @xmath95 to @xmath94 . on the other hand , when @xmath96 , two identical peaks appear in conductance spectra symmetrically , which is quite similar to the dos spectra . in structure 3 ,
no fano peak exists without the magnetic flux according to eq.([rhopm ] ) and fig.[fig : tuningphi ] , because the antibonding state is totally decoupled from the leads .
however , if @xmath22 deviates a little from zero , or more generally , from a multiple of @xmath97 , the channel connecting the decoupled state and the leads is open , and the fano interference comes back again .
our model calculation indicates that , this flux - dependent fano effect takes place whenever @xmath98 and @xmath99 , regardless whether four dot - lead couplings are the identical.@xcite tuning @xmath15 , the half of the flux difference between the left and right sub - rings , can also lead to two resonance peak swap and the flux - dependent fano effect as shown in fig.[fig : tuningtheta ] .
it is noticed when @xmath100 , the results coincide with what obtained when @xmath101 . in three asymmetric configurations .
, scaledwidth=50.0% ] it is interesting to note that the conductance oscillation is a periodic function of @xmath22 with a period of @xmath102 , and a periodic function of @xmath15 with the period of @xmath97 . the difference in the period can be readily explained as the multi - pathway nature in this parallel - coupled dqd system .
if an electron transits from the left to the right , it gathers a phase factor @xmath103 via the upper arms and @xmath104 via the lower arms(fig.[fig : multipathway]a ) .
the interference of the two paths gives @xmath105 , which is associated with a period of @xmath102 . on the other hand , if two routes of an electron are illustrated as the arrows in fig.[fig : multipathway]b , the interference yields @xmath106 and the @xmath97 periodicity , because the first path from the left lead @xmath107 dot 1 @xmath107 dot 2 @xmath107 right lead accumulates a phase @xmath108 , and the second symmetric route through dot2 @xmath107 dot1 gives a phase @xmath109 . , and
( b ) the flux difference @xmath15.,scaledwidth=40.0% ] the quantum interference tuned by the magnetic flux in the dqd system makes the conductance to vary in some fancy way .
let us look at a peculiar instance , the continuous manipulation of the conductance with the lineshape retained .
in fig.[fig : switch ] , two cases of the flux tuning are : ( 1 ) @xmath96 , @xmath110 $ ] ; and ( 2 ) @xmath111 , @xmath112 $ ] . according to eq.([mogammal ] ) and eq.([mogammar ] ) ,
once @xmath111 or @xmath96 , the lineshape of the dos for two molecular levels is fixed . in our calculations ,
@xmath113 , only a single peak appears in fig.[fig : switch](a)(when increasing @xmath14 , the conductance will recover the double - peak feature ) .
in contrast , in fig.[fig : switch](b ) due to the destructive interference , the conductance at @xmath114 is always zero . besides
, a fully symmetric configuration with four identical dot - lead couplings is considered in this calculation , so that the maximum and minimum values of the conductance are precisely at @xmath115 and 0 , respectively .
, when @xmath22 is fixed at @xmath116 ; ( b ) the conductance is tuned by @xmath22 , when @xmath111.,scaledwidth=50.0% ]
in summary , the transport through the parallel - coupled dqd system has been studied , in which a particular attention is paid to the mechanism of the fano lineshape in conductance spectra as well as its tunability . due to the inter - dot coupling , a qd - molecule is formed and can be the proper representation for the present investigations . due to the coupling between the molecular levels and leads ,
two levels are broadened into two bands : one is wider which is associated with the strongly - coupled level , and the other is narrower band related to the weakly - coupled level . when the wider band covers the narrower band , and the phase shift of the wavefunction of the wider band is negligible around the weakly - coupled level , the @xmath1 phase shift at resonance of the wavefunction of the narrower band may be detected via the quantum interference , and shows up as the fano lineshape .
thus , both the reference and the resonance channels for fano interference are readily identified .
since the density of states and the effective couplings between the molecular levels and leads are tunable by making use of the magnetic flux and gate - voltages , several ways to control fano lineshape are proposed , including the total flux @xmath22 , the flux difference between two sub - rings @xmath15 , the inter - dot coupling strength @xmath14 , and the dot - lead coupling . in these ways
, we may realize the swap effect , the flipped tail direction of the fano lineshape , and the continuous lineshape - keeping magnetic switch in this dqd system , which might be of practical applications .
we would like to acknowledge hui zhai , zuo - zi chen and chaoxing liu for helpful discussions .
this work is supported by the natural science foundation of china ( grant no .
10374056 ) , the moe of china ( grant no .
2002003089 ) , and the program of basic research development of china ( grant no .
2001cb610508 ) .
kouwenhoven , c.m .
markus , p.l .
mceuen , s. tarucha , r.m .
westervelt , and n.s .
wingreen , in _ mesoscopic electon transport _ , edited by l.l .
sohn , l.p .
kouwenhoven , and g. sch@xmath117n , nato advanced study institutes , ser .
345 ( kluwer , dordrecht , 1997 ) .
a. yacoby , m. heiblum , d. mahalu , and h. shtrikman , phys .
lett . * 74 * , 4047 ( 1995 ) .
r. schuster , e. buks , m. heiblum , d. mahalu , v. umansky , and h. shtrikman , nature(london ) * 385 * , 417 ( 1997 ) .
e. buks , r. schuster , m. heiblum , d. mahalu , and v. umansky , nature(london ) * 391 * , 871 ( 1998 ) .
van der wiel , s. de franceschi , t. fujisawa , j.m .
elzerman , s. tarucha , and l.p .
kouwenhoven , science * 289 * , 2105 ( 2000 ) .
y. ji , m. heiblum , d. sprinzak , d. mahalu , h. shtrikman , science * 290 * , 779 ( 2000 ) .
k. kobayashi , h. aikawa , s. katsumoto , and y. iye , phys .
lett . * 88 * , 256806 ( 2002 ) .
k. kobayashi , h. aikawa , s. katsumoto , and y. iye , phys .
b * 68 * , 235304 ( 2003 ) .
j. gres , d. goldhaber - gordon , s. heemeyer , m.a .
kastner , h. shtrikman , d. mahalu , and u. meirav , phys .
b * 62 * , 2188 ( 2000 ) .
zacharia , d. goldhaber - gordon , g. granger , m.a .
kastner , y.b .
khavin , h. shtrikman , d. mahalu , and u. meirav , phys .
b * 64 * , 155311 ( 2001 ) .
johnson , c.m .
marcus , m.p .
hanson , and a.c .
gossard , phys .
. lett . * 93 * , 106803 ( 2004 ) r.k .
adair , c.k .
bockelman , and r.e .
peterson , phys . rev . * 76 * , 308 ( 1949 ) .
u. fano and j.w .
cooper , phys . rev . * 137 * , a1364 ( 1965 ) .
f. cerdeira , t.a .
fjeldly , and m. cardona , phys .
b * 8 * , 4734 ( 1973 ) .
j. faist , f. capasso , c. sirtori , k.w .
west , and l.n .
pfeiffer , nature ( london ) * 390 * , 589 ( 1997 ) . d. loss and d.p .
divincenzo , phys .
a * 57 * , 120 ( 1998 ) .
x. hu and s. das sarma , phys .
a * 61 * , 062301 ( 2000 ) .
divincenzo , cond - mat/9612126 ; in _ mesoscopic electron transport _ , edited by l.l .
sohn , l.p .
kouwenhoven , and g. schn , nato advanced study institutes , ser .
345 ( kluwer , dordrecht , 1997 ) .
clerk , x. waintal , and p.w .
brouwer , phys .
lett . * 86 * , 4636 ( 2001 ) .
k. kang and s.y .
cho , j. phys .
: condens .
matter * 16 * , 117 ( 2004 ) .
ladrn de guevara , f. claro , and pedro a. orellana , phys .
b * 67 * , 195335 ( 2003 ) .
bai , m .- f .
yang , and y .- c .
chen , j. phys.:condens.matter * 16 * , 2053 ( 2004 ) .
b. dong , i. djuric , h.l .
cui , and x.l .
lei , cond - mat/0403741 ( 2004 ) .
orelana , m.l .
ladrn de guevara , and f. claro , cond - mat/0404293 ( 2004 ) . c. livermore , c.h .
crouch , r.m .
westervelt , k.l .
campman , and a.c .
gossard , science * 274 * , 1332 ( 1996 ) .
h. haug and a .-
jauho , _ quantum kinetics in transport and optics of semiconductors _ ( springer - verlag , 1996 ) .
j. knig and y. gefen , phys .
b * 65 * , 045316 ( 2002 ) s. datta , _ electronic transport in mesoscopic systems _ ( cambridge univ . press , 1997 ) .
y. meir and n.s .
wingreen , phys .
* 68 * , 2512 ( 1992 ) .
b. kubala and j. knig , phys .
b * 65 * , 245301 ( 2002 ) .
divincenzo , phys . rev . a * 51 * , 1015 ( 1995 ) .
zhang , r. l , z.r .
liu , and l. yu , cond - mat/0403629 . | with the help of the green function technique and the equation of motion approach , the electronic transport through a parallel - coupled double quantum dot(dqd ) is theoretically studied .
owing to the inter - dot coupling , the bonding and antibonding states of the artificial quantum - dot - molecule may constitute an appropriate basis set .
based on this picture , the fano interference in the conductance spectra of the dqd system is readily explained .
the possibility of manipulating the fano lineshape in the tunneling spectra of the dqd system is explored by tuning the dot - lead coupling , the inter - dot coupling , the magnetic flux threading the ring connecting dots and leads , and the flux difference between two sub - rings .
it has been found that by making use of various tuning , the direction of the asymmetric tail of fano lineshape may be flipped by external fields , and the continuous conductance spectra may be magnetically manipulated with lineshape retained .
more importantly , by adjusting the magnetic flux , the function of two molecular states can be exchanged , giving rise to a swap effect , which might play a role as a qubit in the quantum computation . |
over forty lgrbs have been observed with _ hst _ at various times after outburst .
_ hst _ is unique in its capability to easily resolve the distant hosts of these objects . shown in figure 1 is a mosaic of _ hst _ images of the hosts of forty - two bursts .
these are all lgrbs with public data which had an afterglow detected with better than three - sigma significance and a position sufficiently well localized to determine a host galaxy .
a list of all the grbs used in this work can be found in tables 13 of the supplementary material .
the supernovae discussed in this _ article _ were all discovered as part of the _ hubble _ higher z supernova search@xcite , which was done in cooperation with the _ hst _ goods survey@xcite . the goods survey observed two @xmath1 sq .
arcminute patches of sky five times each in epochs separated by forty - five days .
supernovae were identified by image subtraction . in this paper
we discuss only the cc supernovae identified in this survey .
a list of the supernovae used is presented in table 4 of the supplementary material , and images of the supernovae hosts can be seen in figure 2 .
if lgrbs do in fact trace massive star formation , then in the absence of strong extinction we should find a close correlation between their position on their host galaxies and the blue light of those galaxies .
however , many of the grb hosts and quite a few of the supernovae hosts are irregular galaxies made up of more than one bright component . as a result the common astronomical procedure of identifying the centroid of the galaxy s light , and then determining the distance of the object in question from the centroid is not particularly appropriate for these galaxies the centroid of light may in fact lie on a rather faint region of the host ( examine grbs 000926 and 020903 in figure 1 for excellent illustrations of this effect ) .
we therefore have developed a method which is independent of galaxy morphology .
we sort all of the pixels of the host galaxy image from faintest to brightest and ask what fraction of the total light of the host is contained in pixels fainter than or equal to the pixel containing the explosion .
if the explosions track the distribution of light , then the fraction determined by this method should be uniformly distributed between zero and one .
( a detailed exposition of this method can be found in the supplementary materials ) .
as can be seen in figure 3 , the cc supernovae do track the light of their hosts as well as could be expected given their small number statistics . a kolmogornov - smirnov ( ks )
test finds that the distribution of the supernovae is indistinguishable from the distribution of the underlying light .
the situation is clearly different for lgrbs .
as can be seen in figure 3 , the grbs do not simply trace the blue light of the hosts , rather they are far more concentrated on the peaks of light in the hosts than the light itself .
a ks test rejects the hypothesis that grbs are distributed as the light of their hosts with a probability greater than 99.98% .
furthermore , this result is robust : it shows no dependence on grb host size or magnitude . and
in spite of the relatively small number of sn hosts on which a comparison can be made , the two populations are found by the ks test to be drawn from different distributions with @xmath2 certainty . in the next section of this paper we show that the surprising differences in the locations of these objects on the underlying light of their hosts may be due not only to the nature of their progenitor stars but also that of their hosts .
an examination of the mosaics of the grb and sn hosts ( figures 1 and 2 ) immediately shows a remarkable contrast only one grb host in this set of 42 galaxies is a grand - design spiral , while nearly half of the sn hosts are grand - design spirals .
one might wonder if this effect is due to a difference in redshift distribution the cc supernovae discovered by the goods collaboration all lie at @xmath3 , while lgrbs can be found at much larger redshifts where grand - design spirals are rare to non - existent . yet
if we restrict the grb population to @xmath3 ( and thus produce a population with a nearly identical mean and standard deviation in redshift space compared to the goods cc supernovae ) , the situation remains essentially unchanged : only one out of the eighteen grb hosts is a grand - design spiral .
( for a detailed comparison of grb hosts to field galaxies , rather than the sn selected galaxies shown here , see ref . ) .
were the difference in spiral fraction the only indication of a difference in the host populations , we could not rule out random chance given the small number statistics both populations are barely consistent with each other and a spiral fraction of @xmath4 .
however , the host populations differ strongly in ways other than morphology . in figure 4
we compare the 80% light radius ( @xmath5 ) and absolute magnitude distributions of the grb and supernovae hosts .
included in the comparison are all lgrbs with known redshifts @xmath6 at the time of submission and the 16 cc supernovae of goods with spectroscopic or photometric redshifts ( see the supplementary tables for a complete list of the grbs , supernovae and associated parameters used in this study ) . the small minority of grb hosts in this redshift range without _ hst _ imaging are compared only in absolute magnitude and not in size .
the absolute magnitudes have been derived from the observed photometry using a cosmology of @xmath7 , and @xmath8 , and the magnitudes have been corrected for foreground galactic extinction@xcite . for a technical discussion of the determination of the magnitude and size of individual objects , please see the supplementary materials . as can be readily seen
the two host populations differ substantially both in their intrinsic magnitudes and sizes .
the grb hosts are fainter and smaller than the sn hosts .
indeed ks tests reject the hypothesis that these two populations are drawn from the same population with greater than 98.6% and 99.7% certainty for the magnitude and size distributions , respectively .
although the evidence is now overwhelming that both cc supernovae and lgrbs are formed by the collapse of massive stars , our observations show that the distribution of lgrbs and cc supernovae on their hosts , and the nature of their hosts themselves are substantially different . how then can this be ?
we propose here that these surprising findings are the result of the dependence of the probability of grb formation on the state of the chemical evolution of massive stars in a galaxy .
even before the association of lgrbs with massive stars had been established , a number of theorists had suggested that these objects could be formed by the collapse of massive stars , which would leave behind rapidly spinning black holes .
an accretion disk about the black hole would power the grb jet .
these models , sometimes referred to as hypernovae " or
collapsar " models implicitly require very massive stars , since only stars greater than about @xmath9 solar masses form black holes .
but in fact it was widely suspected that even more massive stars would be required if only to provide the required large energies , and to limit the numbers of supernovae progressing to grbs .
we conclude that lgrbs do indeed form from the very most massive stars and this is the reason that they are even more concentrated on the blue light of their hosts than the light itself .
the most massive stars ( o stars ) are frequently found in large associations .
these associations can be extremely bright , and can indeed provide the peak of the light of a galaxy particularly if that galaxy is a faint , blue irregular , as are the grb hosts in general .
indeed , a connection of lgrbs with o - stars ( and perhaps wolf - rayet stars ) is a natural one given the strong emission lines ( including ne [ iii ] ) seen in many of these hosts@xcite and the evidence for possible strong winds off of the progenitors of the grbs seen in absorption in some lgrb spectra@xcite .
however , o stars are found in galaxies of all sizes .
indeed , studies of the magellanic clouds suggest the initial distribution of masses of stars at formation in these dwarf galaxies is essentially identical to that in our much larger spiral , the milky way@xcite . therefore , a difference in the initial mass function of stars is unlikely to be responsible for the differences between the hosts .
we propose that the fundamental differences between the lgrb and sn host populations is not their size or luminosity , but rather their metallicity , or chemical evolution .
some evidence of this already exists .
the hosts of seven lgrbs ( grbs 980425 ( p.
m. vreeswijk , personal communication ) , 990712@xcite , 020903@xcite , 030323@xcite , 030329@xcite , 031203@xcite and 050730@xcite ) have measurements of or limits on their metallicity , and in all cases the metallicity is less than one - third solar .
the small size and low luminosity of the grb hosts is then a result of the well known correlation between galaxy mass and metallicity ( see ref . and references therein ) . but why do lgrbs
choose low - metallicity galaxies ?
this may be a direct result of the evolution of the most massive stars .
it has recently been proposed that metal rich stars with masses of tens of solar masses have such large winds off their surfaces ( due to the photon pressure on their metal rich atmospheres ) that they lose most of their mass before they collapse and produce supernovae@xcite . as a result
they leave behind neutron stars , not the black holes necessary for lgrb formation .
ironically , stars of 15 - 30 solar masses may still form black holes , as they do not possess radiation pressure sufficient to drive off their outer envelopes .
direct evidence for this scenario comes from recent work showing that the galactic soft gamma - ray repeater , sgr 1820 - 06 , is in a cluster of extremely young stars of which the most massive have only started to collapse@xcite yet , the progenitor of sgr 1820 - 06 collapsed to a neutron star , not a black hole .
recent observations of winds from very massive ( wolf - rayet ) stars provide further support for this scenario : outlfows from the low - metallicity stars in the lmc are substantially smaller than those seen from more metal - rich galactic stars@xcite .
the possible importance of metallicity in lgrb formation has therefore not escaped the notice of theorists@xcite .
a preference for low - metallicity may also explain one of the most puzzling results of grb host studies .
none of the lgrb hosts is a red , sub - millimeter bright galaxy .
these highly dust - enshrouded galaxies at redshifts of @xmath10 are believed to be the site of a large fraction of the star formation in the distant universe@xcite . and while some lgrb hosts do show sub - mm emission , none has the red colors characteristic of the majority of this population .
however , it is likely that these red dusty galaxies have substantial metallicities at all redshifts .
the low - metallicity of hosts may also help explain the fact that a substantial fraction of high - redshift lgrb hosts display strong lyman - alpha emission@xcite .
all well classified supernovae associated with lgrbs are types ic , presumably because the presence of a hydrogen envelope about the collapsing core can block the emergence of a grb jet@xcite .
thus only those supernovae whose progenitors have lost some , but not too much mass , appear to be candidates for the formation of a grb . given the large numbers of type ic supernovae in comparison to the estimated numbers of lgrbs however , it is likely that only a small fraction of ic supernovae produce lgrbs .
indeed , even the number of unusually energetic type ib / ic supernovae appears to dwarf the lgrb population@xcite .
another process , perhaps the spin - up of the progenitor in a binary@xcite , may decide which type ic supernovae produce lgrbs .
interestingly , it was the similar distribution of supernovae on their hosts , and particularly the fact that type ib / ic were no more correlated than type ii supernovae with the uv bright regions of their hosts , that led ref to the conclusion that type ib / ic form from binaries .
lgrbs clearly track light differently than the general type ic population .
however the samples used by refs were from supernovae largely discovered on nearby massive galaxies dwarf irregular hosts are underrepresented in these samples
. it will be particularly interesting to see whether large unbiased sn surveys presently underway produce similar locations for their supernovae .
we do not know , however , what separates the small fraction of low - metallicity type ic supernovae which turn into lgrbs from the rest of the population .
potentially , the answer is the amount of angular momentum available in the core to form the jet . in this case , the preference for low - metallicity may indicate that single star evolution dominates over binary interaction in forming lgrbs .
deep , high spectral resolution studies of lgrb afterglows may provide insight here , by allowing a studies of the winds off of the progenitor and any binary companion .
only a small fraction of lgrbs are found in spiral galaxies , even in lgrbs with redshifts @xmath11 where spirals are much more common .
however , the local metallicity in spirals is known to be anti - correlated with distance from the center of the galaxy .
thus one might expect lgrbs in spirals to violate the trend we have seen for the general lgrb population and avoid the bright central regions of their hosts .
the present number of lgrbs known in spirals is still too small to test this prediction . but
a sample size a few times larger should begin to allow such a test . additionally , a survey of the metallicity of the hosts of the goods supernovae should find a higher average metallicity than that seen in grb hosts .
finally , if low - metallicity is indeed the primary variable in determining whether lgrbs are produced , then as we observe higher redshifts , where metallicities are lower than in most local galaxies , lgrbs should be more uniformly distributed among star - forming galaxies .
indeed , some evidence of this may already be present in the data@xcite .
lgrbs , however , are potentially visible to redshifts as high as @xmath12 . at significant redshifts , where the metallicities of even relatively large galaxies are expected to be low , we may find that lgrbs do become nearly unbiased tracers of star formation .
figure i : a mosaic of grb host galaxies imaged by hst .
each individual image corresponds to a square region on the sky @xmath13 on a side .
these images were taken with the space telescope imaging spectrograph ( stis ) , wide - field and planetary camera 2 ( wfpc2 ) and the advanced camera for surveys ( acs ) on _ hst_. in cases where the location of the grb on the host is known to better than @xmath14 , the position of grb is shown by a green mark .
if the positional error is smaller the the point spread function of the image ( @xmath15 for stis and acs , @xmath16 for wfpc2 ) the position is marked by a cross - hair , otherwise the positional error is indicated by a circle .
the stis images were all taken in white light ( no filter ) and in most cases the wfpc2 and acs image are in the f606w filter ( though in a few cases where images in this filter were not available we have used images in f555w or f775w ) .
the stis and f606w images can be thought of as broad `` v '' or visual images , and are , for galaxies exhibiting typical colors of grb hosts , the single most sensitive settings for these cameras .
f555w is close to the ground - based johnson v - band , and f775w corresponds to the ground - based johnson i - band .
due to the redshifts of the hosts , these images generally correspond to blue or ultra - violet images of the hosts in their rest frame , and thus detect light largely produced by the massive stars in the hosts .
figure 2 : a mosaic of cc sn host galaxies imaged with _ hst _ as part of the goods program .
each image in the mosaic has a width of @xmath17 on the sky , and thus two times the field - of - view of each image in the grb mosaic .
the position of each sn on its host galaxy is marked . in all cases ,
these positions are known to sub - pixel accuracy .
supernovae in the goods sample were identified by [ ] as either type ia or cc supernovae based on their colors , luminosities and light curves , as data allowed ( a sn going off near the beginning or end of one of the multi - epoch observing runs would have much less data , and sometimes poor color information ) .
thus bright type ib and ic supernovae , which have colors and luminosities similar to type ia supernovae , would have likely been classified as type ia ( unless a grism spectrum was taken
however only a small fraction of objects were observed spectroscopically ) . on the other
hand fainter type ib and ic supernovae ( @xmath18 ) could in principle be identified from photometric data ; however , in practice the data were rarely sufficient for a clear separation from other cc supernovae . based on surveys of nearby galaxies
, one might expect approximately 20% of the cc supernovae to be type ib or ic@xcite .
figure 3 : the locations of the explosions in comparison to the host light . for each object
an arrow indicates the fraction of total host light in pixels fainter than or equal to the light in the pixel at the location of the transient .
the cumulative fraction of grbs or supernovae found at a given fraction of the total light is shown as a histogram .
the blue arrows and histogram correspond to the grbs and the red arrows and histogram correspond to the supernovae .
were the grbs and supernovae to track the light identically , their histograms would follow the diagonal line . while the supernovae positions do follow the light within the statistical error , the grbs are far more concentrated on the brightest regions of their hosts .
thus while the probability of a sn exploding in a particular pixel is roughly proportional to the surface brightness of the galaxy at that pixel , the probability of a grb a given location effectively goes as a higher power of the local surface brightness .
figure 4 : a comparison of the absolute magnitude and size distributions of the grb and sn hosts . in the main panel ,
the cc sn hosts are represented as red squares and the lgrb hosts as blue circles .
the absolute magnitudes of the hosts are shown on the x - axis and the lengths of the semi - major axes of the hosts on the y - axis .
the plot is then projected onto the two side panels where a histogram is displayed for each host population in each of the dimensions - absolute magnitude and semi - major axis . shown as blue arrows are the absolute magnitudes of grb hosts with @xmath6 that have been detected from the ground but have not yet been observed by hst .
these hosts are only included in the absolute magnitude histogram .
the hosts of grbs are both smaller and fainter than those of supernovae .
support for this research was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
observations analyzed in this work were taken by the nasa / esa hubble space telescope under programs : 7785 , 7863 , 7966 , 8189 , 8588 , 9074 and 9405 ( pi : fruchter ) ; 7964 , 8688 , 9180 , and 10135 ( pi s. r. kulkarni ) ; 8640 ( pi : s. t. holland ) .
the authors wish to thank nino panagia , nolan walborn and alicia soderberg for informative conversations .
we also thank alex filippenko and collaborators for early - time images of grb 980326 , and josh bloom and collaborators for making their early observations of grb 020322 public . | when massive stars exhaust their fuel they collapse and often produce the extraordinarily bright explosions known as core - collapse supernovae . on occasion
, this stellar collapse also powers an even more brilliant relativistic explosion known as a long - duration @xmath0-ray burst .
one would then expect that long @xmath0-ray bursts and core - collapse supernovae should be found in similar galactic environments . here
we show that this expectation is wrong .
we find that the long @xmath0-ray bursts are far more concentrated on the very brightest regions of their host galaxies than are the core - collapse supernovae .
furthermore , the host galaxies of the long @xmath0-ray bursts are significantly fainter and more irregular than the hosts of the core - collapse supernovae .
together these results suggest that long - duration @xmath0-ray bursts are associated with the most massive stars and may be restricted to galaxies of limited chemical evolution .
our results directly imply that long @xmath0-ray bursts are relatively rare in galaxies such as our own milky way .
space telescope science institute , 3700 san martin dr . , baltimore , md 21218 , usa department of physics and astronomy , university of leicester , university road , leicester , le1 7rh , uk centre for astrophysics research , university of hertfordshire , college lane , hatfield , al10 9ab , uk physics & astronomy , tccw 246 , western kentucky university , 1 big red way , bowling green , ky 42101 , usa
european southern observatory , alonso de crdova 3107 , casilla 19001 , santiago , chile dept of astronomy & astrophysics , university of california , 1156 high st , santa cruz , ca 95064 , usa astrophysics research institute , liverpool john moores university , twelve quays house , egerton wharf , birkenhead , ch41 1ld norwegian meteorological institute , p.o .
box 43 , blindern , n-0313 oslo ,
norway dark cosmology centre , niels bohr institute , university of copenhagen , dk-2100 copenhagen , denmark instituto de astrofsica de andaluca ( csic ) , camino bajo de hutor , 50 , 18008 granada , spain .
california institute of technology , mail code 105 - 24 , pasadena , ca 91125 , usa school of physics and astronomy , university of nottingham , university park , united kingdom , ng7 2rd department of physics , stockholm university , se-106 91 stockholm , sweden physics department , university of notre dame , 225 nieuwland hall , notre dame , in 46556 , usa vanderbilt university , dept . of physics and astronomy , 6301 stevenson center , nashville , tn 37235 ,
usa code 667 extraterrestial planets and stellar astrophysics , exploration of the universe division , goddard space flight center , greenbelt , md 20771 , usa code 660.1 , nasa s gsfc , greenbelt , md 20771 , usa nasa / marshall space flight center , vp-62 , national space science & technology center , 320 sparkman drive , huntsville , al 35805 , usa renaissance technologies corporation , 600 route 25a , east setauket , new york 11733 lawrence berkeley national laboratory , m.s .
50f-1650 , 1 cyclotron road , berkeley , ca 94720 , usa inaf , osservatorio astronomico di trieste , via g.b .
tiepolo 11 , i-34131 trieste , italy astronomical institute ` anton pannekoek ' , university of amsterdam , kruislaan 403 , nl-1098 sj amsterdam , the netherlands _ this paper has been accepted for publication in nature . in accordance with that journal s editorial policy we ask that you do not discuss this work with the press until it appears in nature either online or in print
. thank you . _
+ it is an irony of astrophysics that stellar birth is most spectacularly marked by the deaths of massive stars .
massive stars burn brighter and hotter than smaller stars , and exhaust their fuel far more rapidly
. therefore a region of star formation filled with low mass stars still early in their lives , and in some cases still forming , may also host massive stars already collapsing and producing supernovae .
indeed , with the exception of the now famous type ia supernovae , which have been so successfully used for cosmological studies@xcite and which are thought to be formed by the uncontrolled nuclear burning of stellar remnants comparable in mass to the sun@xcite , all supernovae are thought to be produced by the collapse of massive stars .
the collapse of the very most massive stars ( tens of solar masses ) is thought to leave behind either black holes or neutron stars , depending largely on the state of chemical evolution of the material that formed the star , while the demise of stars between approximately 8 and 20 solar masses produces only neutron stars@xcite .
gamma - ray bursts ( grbs ) , like supernovae , are a heterogeneous population .
grbs can be divided into two classes : short , hard bursts , which last between milliseconds and about two seconds and have hard high - energy spectra , and long , soft bursts , which last between two and tens of seconds , and have softer high - energy spectra@xcite .
only very recently have a few of the short bursts been well localized , and initial studies of their apparent hosts suggest that these bursts may be formed by the binary merger of stellar remnants@xcite .
in contrast , the afterglows of over eighty long grbs ( lgrbs ) have been detected in the optical and/or radio . and
as a result of these detections , it has become clear that lgrbs , like core - collapse supernovae , are related to the deaths of young , massive stars .
it is these objects , born of the deaths of massive stars , that we study here .
lgrbs are generally found in extremely blue host galaxies@xcite which exhibit strong emission lines@xcite suggesting a significant abundance of young , very massive stars .
furthermore while the light curves of the optical transients ( ots ) associated with lgrbs are often dominated by radiation from the relativistic outflow of the grb , numerous lgrbs have shown late - time `` bumps '' in their light - curves consistent with the presence of an underlying sn@xcite .
in several cases spectroscopic evidence has provided confirmation of the light of a sn superposed on the ot@xcite .
indeed , given the large variations in the brightnesses of ots and supernovae , and the limited observations on some grbs , it seems plausible that _ all _ lgrbs have an underlying sn@xcite .
furthermore , while the energy released in a lgrb often appears to the observer to be orders of magnitude larger than that of a supernovae , there is now good evidence suggesting that most lgrbs are highly collimated and often illuminate only a few percent of the sky@xcite .
when one takes this into account , the energy released in lgrbs more closely resembles that of energetic supernovae
. however , not all core - collapse supernovae may be candidates for the production of grbs .
the supernovae with good spectroscopic identifications so far associated with grbs have been type ic that is cc supernovae which show no evidence of hydrogen or helium in their spectra .
( type ib supernovae , which are often studied together with type ic , have spectra which are also largely devoid of hydrogen lines but show strong helium features . )
a star may therefore need to lose its outer envelope if a grb is to be able to burn its way through the stellar atmosphere@xcite .
studies which have compared the locations type ib / c supernovae with the more numerous type ii supernovae ( cc supernovae showing hydrogen lines ) in local galaxies so far show no differences in either the type of host or the placement of the explosion on the host @xcite .
this result led ref .
to argue cc supernovae all come from the same mass range of progenitor stars , but that type ib / c supernovae may have had their envelopes stripped by interaction with a binary stellar companion . whether type ic supernovae come from single stars , or binary stars , or both , it is very likely that only small fraction of these supernovae produce grbs@xcite . given the common massive stellar origins of cc supernovae and lgrbs , one might expect that their hosts and local environments might be quite similar .
it has long been argued that cc supernovae should track the blue light in the universe ( the light from massive stars is blue ) , both in their distribution among galaxies and within their host galaxies themselves .
one would expect similar behavior from lgrbs , and indeed rough evidence for such a correlation has been reported@xcite . here
we use the high resolution available from hubble space telescope ( _ hst _ ) images , and an analytical technique developed by us that is independent of galaxy morphology , to study the correlation between these objects and the light of their hosts .
we also compare the sizes , morphologies and brightnesses of the lgrb hosts with those of the supernovae .
our results reveal surprising and substantial differences between the birth places of these cosmic explosions .
we find that while cc supernovae trace the blue - light of their hosts , grbs are far more concentrated on the brightest regions of their hosts .
furthermore , while the hosts of cc supernovae are approximately equally divided between spiral and irregular galaxies , the overwhelming majority of grbs are on irregulars , even when we restrict the grb sample to the same redshift range as the sn sample .
we argue that these results may be best understood if grbs are formed from the collapse of extremely massive , low - metallicity stars . |
for some time , the santa cruz linear collider r&d group has been interested in exploring the requirements that measuring forward selectron production places on the linear collider detector . with right - handed selectron and lightest neutralino masses of 143 and 95 gev , respectively , snowmass point sps1a produces a substantial forward ( @xmath2 ) population of electrons from selectron decay . over the past two years , the santa cruz group has developed techniques to isolate the forward selectron - decay signal from standard - model backgrounds , as well as techniques to extract the selectron mass from the observed selectron - decay electron spectrum .
this paper reports on the results of those studies .
backgrounds to the selectron - decay electron signal come from two primary sources : four - electron production for which two of the final - state fermions are close in angle to the beam trajectory , and thus escape detection , and the @xmath3-boson fusion process @xmath4 .
during the course of searches for supersymmetry at lep , several cuts were developed to isolate potential susy signals from standard model backgrounds .
* * fiducial region cut : * exactly one final - state positron and one final - state electron pair must be detected in the fiducial region . for this study ,
the fiducial region lies above 5 gev in momentum , and within @xmath5 for central - region studies , or within @xmath6 for studies including the forward region . *
* tagging cut : * no observable electron or positron in the low - angle ` tagging calorimetry ' ( with assumed coverage of 20mrad @xmath7 110mrad ) .
* * transverse momentum cut : * cuts events for which the vector sum of the observed @xmath8 pair transverse momenta is less than 500 gev @xmath9(20 mrad ) . for a signal region of @xmath5 ,
these cuts completely eliminate the four - electron background up to diagrams with additional radiation , and sufficiently select against the @xmath10 process . however , as one attempts to reconstruct the selectron decay signal beyond @xmath11 , both the radiative four - electron background and the @xmath10 backgrounds approach poles in their respective production cross sections , and begin to dominate the selectron signal . to suppress these backgrounds to a degree appropriate for use of the forward selectron decay signal , two additional cuts were applied .
* * photon cut : * remove event if there is an observed photon of 20 gev or greater in either the fiducial or tagging region . * * high - momentum cut : * remove even if the vector sum of the @xmath8 pair momenta has a magnitude of greater than 225 gev . with the application of these cuts to events produced at @xmath0 tev with 80% right - handed - polarized electron beam and unpolarized positrons , it was found that standard model processes provided a negligible background to the right - handed selectron signal .
in addition , the application of these selection criteria did not significantly alter the shape of the selectron - decay electron spectrum in either the central or forward region .
nevertheless , in the analysis that follows all of the selection cuts will be applied to the selectron signal ; however , standard model backgrounds will be ignored from this point on . finally , running with an 80% right - polarized electron beam ,
electrons from the decay of the more massive left - handed selectron provide a diffuse background to the right - handed selectron decay electron spectrum .
this contribution was also ignored in the right - handed selectron analysis .
the selectron signal was generated at @xmath0 tev using the snowmass point sps1a
@xcite parameters @xmath12 gev , @xmath13 gev , @xmath14 , @xmath15 , and sgn(@xmath16 ) = @xmath17 .
the sps1a specifications were implemented within the isajet @xcite package , and included the effects of initial state radiation , beamstrahlung ( @xmath18 ) , and three different values of the fractional beam energy spread ( assuming a gaussian distribution ) : 1.0% , 0.16% , and 0.0% .
for this point in susy parameter space , the endpoint of the selectron - decay electron energy distribution lies above 270 gev , making the precise measurement of the endpoint energy by the tracking system somewhat challenging . nonetheless , the projected @xmath19 resolution of the tracking system at this energy is somewhat better than that expected for the electromagnetic calorimetry .
the @xmath20 distribution of final - state electrons and positrons from the decay of right - handed selectrons is peaked towards @xmath21 , due to the sizeable t - channel contribution admitted by the relatively light selectron and neutralino .
approximately half of the signal lies in the forward region , beyond @xmath11 .
to focus on the accuracy of the selectron mass reconstruction , it was assumed for now that the lightest neutralino mass was known precisely from other measurements .
the santa cruz group plans to relax this assumption in further studies .
the selectron mass was determined by finding the best fit of the selectron energy spectrum to that of a series of ` template ' distributions generated with slightly varying right - handed selectron masses , but with all other aspects of the signal generation , including beam energy spread and detector smearing effects , the same as for the data simulation .
templates were produced for 15 right - handed selectron masses in a range of approximately @xmath22 gev about the nominal sps1a mass of 143.11 gev .
a total of 120 independent data sets , each corresponding to an integrated luminosity of 115 fb@xmath23 , were generated at the nominal sps1a mass .
for each of these data sets , a @xmath24 was formed against each of the templates points , according to @xmath25 where the sum is over energy bins in the selectron - decay electron spectrum , @xmath26 and @xmath27 are the data and template bin contents , respectively , and @xmath28 is the relative sample - size weighting factor . to minimize the statistical contributions of the template files ,
they were generated with an integrated luminosity of approximately 1000 fb@xmath23 each .
for each data set , the @xmath24 contour was fit with a quartic polynomial , which was then minimized to find the best - fit selectron mass .
this procedure was repeated for the 120 data sets , and the mean and root - mean - square deviation was calculated from the distribution of best - fit right - handed selectron masses . in forming the @xmath24 for each of the 120 independent data sets , it was found that the inclusion of energy - spectrum bins not sufficiently close to the upper and lower endpoints to provide useful information on the endpoint location introduced unacceptable scatter in the contour .
thus , for each scenario that was studied ( beam energy spread , detector model , polar angle reach ) , only bins near the upper and lower endpoint were used in forming the @xmath24 .
the exact energy ranges used in the @xmath24 calculation depended on the scenario under study , and was determined in a data - driven manner by examining the spectrum in the region of the endpoints in the generated data samples .
table i shows the regions used in the forming the @xmath24 for the different scenarios . to explore the effects of detector resolution ,
data sets and mass templates were generated assuming perfect detector resolution as well as that of the sdmar01 detector .
detector resolution effects were incorporated via gaussian smearing based on error matrices from billior - based tracking error calculations provided by the lcdtrk program @xcite .
two different ranges in @xmath29 were explored : between 0.0 and 0.8 and between 0.0 and 0.994 .
finally , three different values of the beam energy spread were explored : @xmath30 ( to allow benchmark comparisons to previous studies @xcite ) , @xmath31 ( approximately that expected for the superconducting rf design ) , and , for comparison , 0.0% .
.endpoint fit regions [ cols="<,^,^",options="header " , ] [ l2ea4-t1 ]
figure 1 shows the root - mean - square deviation observed for the 120 independent data sets for the different scenarios that were explored in the study ( in all cases , the mean value of the fit mass was reasonably consistent with the input vale of the right - handed selectron mass of 143.1 gev ) . for a perfect detector ( no measurement uncertainty ) ,
the inclusion of the forward region provides substantial improvement in the selectron mass measurement , independent of beam energy spread .
in fact , the improvement is better than one would expect from the factor - of - two increase in the number of signal electrons that is associated with the inclusion of the forward region .
a study of the two - dimensional frequency distribution of signal electrons as a function of energy and the cosine of the polar angle ( figure 2 ) reveals that the spectrum has a greater contribution at higher energy , where the sensitivity to the selectron mass is greatest , at high values of @xmath20 .
thus , in this case , most of the information on the selectron mass ( and on slepton masses in general ) is in the forward region . for large beam energy spread
, it is seen that the sensitivity to the selectron mass has little dependence on the detector resolution .
for this case , the sid design seems to be adequate to take advantage of the physics capabilities of the accelerator .
however , for smaller beam energy spread , this is not the case . for the energy
spread anticipated for the selected machine design , substantial improvement in the selectron mass determination can be achieved by improving the detector resolution , particularly in the forward region .
to examine the issue of measuring selectron masses in the forward region of the linear collider detector , we have simulated sps1a selectron production at @xmath0 tev . by developing two new selection criteria
, we have demonstrated that the selectron signal can be separated from standard model backgrounds through the entire forward tracking region @xmath32 .
due to the stiffening of the selectron - decay electron spectrum at higher values of @xmath29 , for a light selectron most of the information on selectron mass ( and slepton mass in general ) comes from the forward region . for large beam energy spread ( @xmath30 ) ,
we find that the sid detector design is adequate to exploit the potential of the linear collider ; however , for the expected beam energy spread of the chosen cold rf technology , gains in sensitivity to the selectron mass can be made by further improvements in momentum resolution , in both the central and forward regions .
although they appear as authors on this paper , one of us ( schumm ) would like to acknowledge the dedication and creativity of the other authors , who performed these studies as undergraduate physics majors at the university of california at santa cruz . | we have used isajet and the jas lcd fast simulation to explore the precision of snowmass point sps1a selectron mass reconstruction for the silicon detector concept . simulating collisions at @xmath0 tev
, we have found that most of the information constraining the selectron mass is carried in the forward ( @xmath1 ) region .
we have also found that , for a beam energy spread of 1% ( conventional rf design ) , detector resolution limitations compromise the selectron mass reconstruction only in the forward region .
however , for a beam energy spread of less than 0.2% ( superconducting rf design ) , the detector resolution compromises the selectron mass reconstruction over the full angular region .
1.5 cm * selectron mass reconstruction and the resolution of the linear collider detector * + sharon gerbode , heath holguin , troy lau , paul mooser , adam pearlstein , joe rose , bruce schumm + santa cruz institute for particle physics + university of california , santa cruz we have used isajet and the jas lcd fast simulation to explore the precision of snowmass point sps1a selectron mass reconstruction for the silicon detector concept . simulating collisions at @xmath0 tev , we have found that most of the information constraining the selectron mass is carried in the forward ( @xmath1 ) region .
we have also found that , for a beam energy spread of 1% ( conventional rf design ) , detector resolution limitations compromise the selectron mass reconstruction only in the forward region .
however , for a beam energy spread of less than 0.2% ( superconducting rf design ) , the detector resolution compromises the selectron mass reconstruction over the full angular region .
talk presented at the 2005 international linear collider workshop , stanford university , stanford , california , march 18 - 22 , 2005 |
let @xmath9 be a symplectic manifold with an action of a compact torus @xmath0 .
a _ moment map _ is an invariant map @xmath10 which intertwines the group action and the symplectic form by the _ moment map condition _
@xmath11 where @xmath12 , @xmath13 is the vector field on @xmath1 generated by @xmath14 , and @xmath15 is the pairing of @xmath16 with @xmath17
. we also write @xmath18 to indicate @xmath19 , the @xmath14-component of @xmath20 .
condition ( [ eq : momentmapcondition ] ) and the nondegeneracy of @xmath21 imply that singular points of @xmath20 occur when @xmath22 , or when a subtorus of g acts trivially .
if @xmath23 is a component of the fixed point set of @xmath0 on @xmath1 , then @xmath24 has constant value in @xmath16 . for @xmath1 a compact manifold ,
the image of @xmath20 is a convex polytope in @xmath16 @xcite .
let @xmath25 .
for any subtorus @xmath26 , denote by @xmath27 the fixed point set of @xmath1 under @xmath2 .
we say @xmath2 is _ generic _ if @xmath28 .
the terminology is appropriate as there are a finite number of subtori @xmath2 such that @xmath29 contains but does not equal @xmath30 .
if @xmath2 is generic , @xmath31 is a set of isolated points . in general , however , the images @xmath32 form _
codimension-@xmath33 walls _ of the polytope , where @xmath34 is the dimension of the subtorus of @xmath5 acting effectively .
these walls may be internal to the polytope .
let @xmath35 be a 1-dimensional subtorus .
if a component of @xmath36 has an effective @xmath37 action , then its image will be a codimension-1 wall .
these include but are not restricted to the facets of the polytope .
when @xmath38 is a regular value of @xmath20 , @xmath39 is a submanifold of @xmath1 and has a locally free @xmath0 action by the invariance of @xmath20 .
the quotient space @xmath40 is called the _
symplectic reduction _ and is denoted @xmath41 , where the parameter @xmath38 is suppressed when @xmath42 . this paper is concerned with the ordinary , rational cohomology of symplectic reductions and its relationship to the @xmath0-equivariant cohomology of @xmath1 . by definition ,
the @xmath0-equivariant cohomology of @xmath1 is @xmath43 where @xmath44 is a contractible space with a free @xmath0 action and @xmath45 indicates the product @xmath46 quotiented by the diagonal @xmath0 action . for more details ,
see @xcite . when @xmath0 acts locally freely on a manifold z , the @xmath0-equivariant cohomology of z is the ordinary cohomology of the quotient @xmath47 . in particular for @xmath38 a regular value of the moment map , @xmath48 [ th : surjective ] let @xmath1 be a compact symplectic manifold with a hamiltonian @xmath0 action , where @xmath0 is a compact torus .
let @xmath20 be a moment map for the @xmath0 action on @xmath1 .
for any regular value @xmath49 , the natural map @xmath50 induced from the inclusion @xmath51 is a surjection .
we generalize this theorem to the following equivariant version .
[ th : equvtsurjective ] let @xmath1 be a compact symplectic manifold with a hamiltonian @xmath0 action , where @xmath0 is a compact torus and @xmath2 a subtorus , not necessarily generic .
let @xmath52 and @xmath53 be moment maps for the respective actions . for a regular value @xmath54 of @xmath53 , the inclusion of the submanifold @xmath55 of @xmath1
induces a surjection in equivariant cohomology @xmath56 note that we use @xmath57 to indicate the map on cohomology induced by the restriction from @xmath1 to @xmath58 , and @xmath59 for the map on cohomology induced by the restriction from @xmath1 to @xmath60 .
these two theoreoms allow one to compute the cohomology of the symplectic reduction by a torus action , as long as one can compute the equivariant cohomology and the kernel of this map . in certain cases ,
the equivariant cohomology has been completely described @xcite , @xcite .
the kernel of the kirwan map in theorem [ th : surjective ] has also been described in very general terms , which we present below as theorem [ th : tw ] .
these results hinge on the key fact that for compact hamiltonian torus spaces , the equivariant cohomology injects into the equivariant cohomology of the fixed point set under the natural restriction map @xcite .
let @xmath1 be a compact hamiltonian @xmath0 space for @xmath0 a compact torus .
let @xmath61 be the collection of connected components of the fixed point set @xmath30 .
then @xmath62 is an injection . for a class @xmath63 and @xmath64
, we write @xmath65 to indicate the restriction to the fixed component @xmath66 .
the relationship between surjectivity onto the symplectic reduction and injectivity of the equivariant cohomology into that of the fixed point set is manifest in the following description of @xmath67 from theorem [ th : surjective ] .
[ th : tw ] let @xmath1 be a compact hamiltonian @xmath0-space , with moment map @xmath20 .
let @xmath68 be a regular value of @xmath20 .
the kernel of the kirwan map @xmath69 is the ideal @xmath70 in @xmath71 generated by @xmath72 where @xmath73 we prove an equivariant analogue of this theorem : [ th : kernel ] let @xmath1 be a compact hamiltonian @xmath0 space with moment map @xmath52 .
let @xmath2 be any subtorus of @xmath0 and @xmath53 the corresponding moment map for the @xmath2 action . for @xmath38
be a regular value of @xmath53 , the kernel of the equivariant kirwan map @xmath74 is the ideal @xmath75 generated by @xmath76 where @xmath77 is defined as in theorem [ th : tw ] .
the subindex @xmath0 on @xmath78 and @xmath79 indicates that these generate ideals in @xmath71 , while the superindices indicate the relevant set of vectors in the theorems .
we will use @xmath75 to indicate the ideal in @xmath71 generated by @xmath79 and we will suppress the parameter @xmath38 when @xmath42 . notice that the difference between the kernel of @xmath57 and that of @xmath59 is that the union in theorem [ th : kernel ] is taken only over @xmath80 .
the significance is that theorem [ th : tw ] can be recovered by the successive application of theorem [ th : kernel ] to one - dimensional subtori of @xmath0 . in the case
that @xmath42 , for each @xmath35 , the kernel is generated by @xmath81 and @xmath82 for a choice of generator @xmath83 .
it follows that the kernel of @xmath57 is generated by classes @xmath63 satisfying one of @xmath84 conditions , where @xmath25 .
the final contribution of this article is to find a small set @xmath85 of @xmath86 such that @xmath78 in theorem [ th : tw ] can be replaced by @xmath87 .
let @xmath88 be the ( finite ) set of unit vectors perpendicular to codimension-1 walls of the moment polytope .
for any @xmath89 , the annihilator @xmath90 is a hyperplane in @xmath16 through 0 and parallel to a codimension-1 wall of the moment polytope .
these hyperplanes , shifted to pass through @xmath38 , are the key to the kernel of @xmath57 .
the kernel of @xmath91 is generated by classes @xmath63 with the following property .
there exists an oriented hyperplane @xmath92 through @xmath38 and parallel to a codimension-1 wall of the moment polytope such that @xmath93 restricts to 0 on the set of all fixed points whose image under @xmath20 lie to the positive side of @xmath92 .
we prove this theorem in section [ se : wallstheorem ] .
first we state several basic facts about equivariant morse theory ( as developed in @xcite ) .
we then refine these ideas to gain equivariant homotopy information that is standard in the case that the morse function has only isolated critical points and there is no group action .
let @xmath94 be a smooth function on a compact manifold @xmath1 and @xmath66 a connected component of the critical set of @xmath94 on @xmath1 .
choose a riemannian metric on @xmath1 .
we say @xmath66 is a _ non - degenerate critical manifold _ for @xmath94 if 1 .
@xmath23 is a submanifold of @xmath1 such that @xmath95 along @xmath66 , and 2 .
the hessian @xmath96 ( the matrix of second derivatives of @xmath94 ) is non - degenerate on the normal bundle @xmath97 of @xmath66 in @xmath1 .
if every connected component of the critical set is non - degenerate , we say @xmath94 is _ morse - bott_. at every non - degenerate critical submanifold , we use the riemannian metric to identify a neighborhood of the zero - section in the normal bundle @xmath97 with a tubular neighborhood of @xmath66 in @xmath1 .
the hessian defines a splitting @xmath98 into positive and negative normal bundles .
the dimension of the fibres of @xmath99 is called the _
morse index _ of @xmath66 and is denoted @xmath100 . now assume that @xmath1 has a @xmath0 action , where @xmath0 is a compact torus , and that the metric and the function @xmath94 are invariant with respect to this action .
the splitting of @xmath97 into positive and negative bundles is equivariant .
this setting mimics that in which @xmath94 is a generic component of the moment map , and @xmath66 is a connected component of @xmath30 .
the proofs of the following two lemma are close to identical to those presented by milnor @xcite in the case that critical manifolds are points and there is no group action .
we note only the minor differences in this more general setting . to make the lemmas true for morse - bott functions , all local calculations must include extra coordinates along the critical manifolds . to make the lemmas equivariant , we choose an invariant riemannian metric ( so that the group acts by isometries ) and we equivariantly identify a neighborhood of the zero - section in the normal bundle of a critical submanifold with a tubular neighborhood of that submanifold ( see @xcite ) .
this makes all relevant maps and homotopies equivariant . for lemmas [ le : morselemma1 ] and [
le : morselemma2 ] , assume @xmath1 is a compact manifold with a @xmath0 action and a @xmath0-invariant riemannian metric , where @xmath0 is a compact torus .
let @xmath101 be an invariant morse - bott function .
let @xmath102 $ ] for @xmath103 .
[ le : morselemma1 ] suppose @xmath104 and @xmath105 $ ] is compact and contains no critical points of @xmath94 . then @xmath106 is a @xmath0-equivariant homotopy equivalence .
[ le : morselemma2 ] suppose there are @xmath107 connected components @xmath108 of the critical set of @xmath94 with the same critical value , and such that @xmath109 .
then @xmath110 is equivariantly homotopic to @xmath111 with a @xmath112-cell bundle over @xmath113 attached for each @xmath114 , where @xmath112 is the morse index of @xmath113 .
these homotopy theorems lead the following results about equivariant cohomology . for the sake of simplicity , we assume that @xmath94 has distinct values for distinct connected components of the critical set .
we note that the theorems proven below can be generalized ( using lemma [ le : morselemma2 ] ) to include the case that @xmath94 does not have distinct values for distinct critical sets , however the notation becomes cumbersome .
assume now that the @xmath0 action is hamiltonian , and that @xmath94 is a component of the moment map for the @xmath0 action . as mentioned , we assume also that @xmath94 separates the critical sets , i.e. one can order the critical sets @xmath115 of @xmath94 so that @xmath116 if and only if @xmath117 .
if @xmath94 is generic , these critical sets are the fixed points of the @xmath0 action .
the fundamental principle introduced by atiyah and bott @xcite is that an equivariant cohomology class on @xmath118 extends to a class on @xmath111 for any @xmath119 ; in particular , it extends to one on all of @xmath1 ( although not uniquely ) .
we prove this theorem following @xcite and @xcite .
define the sets @xmath120 where @xmath121 is small enough that @xmath113 is the only critical set in @xmath122 .
note that by lemma [ le : morselemma1 ] @xmath123 for each @xmath124 , there is a long exact sequence in @xmath0-equivariant cohomology @xmath125 as before , we equivariantly identify a tubular neighborhood of @xmath113 with a neighborhood of 0 in the normal bundle of @xmath113 .
this bundle splits @xmath126 into the positive and negative normal bundles of @xmath113 . by excision and homotopy equivalence ,
@xmath127 where @xmath128 and @xmath129 are the unit disk and sphere bundles , respectively , of @xmath130 .
the equivariant thom isomorphism states that @xmath131 where @xmath132 .
thus the exact sequence ( [ seq : relexact ] ) is equivalently @xmath133 [ le : cishortexact ] for @xmath134 , the sequence ( [ seq : cilongexact ] ) splits into the short exact sequence @xmath135 the composition @xmath136 , where the second map is induced by inclusion , restricts to the composition @xmath137 .
this latter composition is muliplication by the equivariant euler class of the negative normal bundle of @xmath113 .
atiyah and bott show in @xcite that , in the case that there is an @xmath35 which is acting on @xmath130 and fixing @xmath113 , this class is not a zero - divisor .
it follows that @xmath138 must be an injection .
furthermore , by the exactness of sequence ( [ seq : cilongexact ] ) , the image of @xmath139 is the kernel of @xmath140 , which is 0 by injectivity .
thus there is a surjection @xmath141 showing that the sequence splits .
an equivariant cohomology class on @xmath118 extends to one on @xmath142 . by the homotopy equivalence ( [ eq : homotopyequivalent ] ) , a class on @xmath143 extends to one on @xmath144 .
surjectivity in ( [ seq : cishortexact ] ) implies that a class on @xmath144 extends to one on @xmath145 .
thus by induction a class on @xmath118 extends to a class in @xmath71 .
one may ask the question of how unique these extensions are . by the injection ( @xmath146 ) , a class is distinguished by its restriction to the fixed point set .
as these fixed point sets are critical sets for morse - bott functions obtained from components of the moment map , we exploit their relationship among each other .
let @xmath147 be the gradient of @xmath94 with respect to a compatible riemannian metric .
for any critical subset @xmath66 , there is a cell - bundle of points @xmath148 which converge to @xmath66 under the flow of @xmath149 ( or @xmath147 ) , called the _ stable manifold _ ( or _ unstable manifold _ ) of @xmath66 .
furthermore , every point in @xmath1 converges to some @xmath66 under this flow .
thus , for any @xmath148 , there is a ( nonunique ) sequence of critical sets @xmath150 such that @xmath151 converges to @xmath152 , and there are points in the unstable manifold of @xmath153 which converge to @xmath154 for every @xmath155 .
define the _ extended stable manifold _ of a critical set @xmath66 to be the set of points @xmath148 whose flows along @xmath149 have an associated sequence including @xmath66 .
in the case that @xmath1 is a coadjoint orbit of a semi - simple lie group , the extended stable manifold of a critical point @xmath156 is just the closure of the stable manifold out of @xmath156 ; they are called ( permuted ) schubert varieties . [
le : restricteuler ] suppose @xmath63 restricts to 0 on all @xmath113 such that @xmath117 .
then @xmath157 is some multiple of @xmath158 , the equivariant euler class of the negative normal bundle of @xmath159 . by ( [ seq : cishortexact ] ) ,
a class @xmath93 such that @xmath160 is in the image of @xmath161 . as the map @xmath162 ( where the latter map is restriction ) is multiplication by @xmath163
, @xmath157 must be a multiple of this class .
[ le : flowupclass ] for every connected component @xmath66 of the critical set of @xmath94 , there is a class @xmath93 with the following restriction properties : 1 .
[ le : flowupclassrestriction1 ] @xmath164 if @xmath113 is not in the equivariant extended stable manifold of @xmath66 , and 2 .
[ le : flowupclassrestriction2 ] @xmath165 where @xmath166 is the equivariant euler class of the negative normal bundle ( defined by @xmath94 ) of @xmath66 .
let @xmath167 be the index such that @xmath168 .
note that @xmath169 are not in the extended stable manifold of @xmath159 , as @xmath117 implies @xmath116 .
using the short exact sequence ( [ seq : cishortexact ] ) we extend the class @xmath170 on @xmath118 to @xmath93 on @xmath171 such that @xmath93 restricts to 0 on @xmath172 .
as @xmath173 and the composition @xmath162 is multiplication by @xmath163 , we may extend @xmath93 to @xmath174 such that its restriction to @xmath159 is @xmath163 .
now suppose @xmath175 but @xmath113 is not in the extended stable manifold from @xmath159 .
since points in the unstable manifold of @xmath113 flow into points in @xmath176 , by lemma [ le : morselemma2 ] there is a cw - complex @xmath177 which is @xmath0-homotopic to @xmath171 with a @xmath178-cell bundle over @xmath113 attached .
@xmath177 has critical sets @xmath179 . by lemma [ le : restricteuler ]
@xmath180 further restricts to @xmath181 on @xmath113 where @xmath182 .
let @xmath183 be such that @xmath184 and @xmath185 .
@xmath186 exists by the first part of this proof , and @xmath187 restricts to 0 on @xmath179 and to @xmath163 on @xmath159 .
let @xmath188 where the sum is over all @xmath175 such that @xmath113 is not in the extended stable manifold of @xmath159 .
then @xmath189 has the desired restriction properties . instead of dealing with cw - complexes
, one might consider the closures of the stable manifolds and make the same argument as that above with these varieties .
however , the singularities require resolving to be sure there are well - defined classes , restricting one to the complex case only .
in this section we use equivariant morse theory to prove theorem [ th : equvtsurjective ] , that the equivariant kirwan map is surjective .
this proof follows very closely tolman and weitsman s rendition of kirwan s result ( theorem [ th : surjective ] ) in the case where @xmath190 .
the main elements of this proof are found in @xcite , modified to allow for a torus action commuting with the @xmath191 action and for the possibility of non - generic @xmath26 .
choose @xmath35 and let @xmath192 be a moment map for the @xmath191 action .
we first show that the restriction @xmath193 is surjective .
let @xmath194 generate the @xmath191 action .
consider the function @xmath195 where @xmath196 for an appropriate choice of norm on @xmath197 , we have @xmath198 . the critical set of @xmath195 consists of the minimum @xmath199 and the critical sets of @xmath18 . by the moment map condition ( [ eq : momentmapcondition ] ) , the critical points of @xmath18 are the fixed points of the @xmath191 action generated by @xmath14 . as kirwan notes in @xcite , the function @xmath195 may not be morse - bott function ; there may be degenerate critical sets .
however , this occurs only at the minimum @xmath200 where the short exact sequence ( [ seq : cishortexact ] ) holds trivially . as before , without loss of generality
we may suppose that @xmath201 separates the critical set .
we order them by @xmath116 if and only if @xmath117 and @xmath202 .
let @xmath203 and @xmath204 be as in ( [ def : mi+ ] ) , ( [ def : mi- ] ) .
we use lemma [ le : cishortexact ] to show that @xmath205 is surjective for each @xmath124 .
we noted that @xmath206 is an isomorphism .
now assume that we have a surjection @xmath207 for all @xmath208 . by the short exact sequence ( [ seq : cishortexact ] ) , there is a surjection @xmath209 but @xmath210 by the homotopy equivalence ( [ eq : homotopyequivalent ] ) and the latter ring surjects onto @xmath211 by assumption . thus @xmath212 is a surjection . by induction
there is a surjection @xmath213 as @xmath170 is a regular value of @xmath214 , @xmath215 and thus @xmath216 is a surjection .
there is a residual hamiltonian @xmath37 action on @xmath217 , which allows us to apply this technique inductively . by reduction in stages , for any commuting subgroups @xmath218 and @xmath219 of @xmath0 , @xmath220 . for any @xmath26 , choose a splitting @xmath221 .
successively apply the surjection ( [ eq : one - dimred ] ) to obtain a sequence of surjections @xmath222
here we prove theorem [ th : kernel ] , that @xmath223 , where @xmath224 is the equivariant kirwan map .
we first show that @xmath225 by directly restricting classes in @xmath226 to @xmath227 .
we then show the ideals are equal by a dimension count .
let @xmath228 be the inclusion map of lie algebras and @xmath229 the induced projection .
denote by @xmath230 and @xmath231 the natural pairings between @xmath16 and @xmath17 and between @xmath232 and @xmath233 , respectively .
then if @xmath234 is a moment map for the @xmath0 action , @xmath235 is a moment map for the restricted @xmath2 action .
choose @xmath236 a regular value of @xmath53 .
let @xmath237 for some @xmath80 . by definition @xmath238 for every connected component @xmath66 of @xmath30 such that @xmath239 .
let @xmath240 @xmath241 is a maximal dimension open @xmath0-invariant submanifold of @xmath1 .
thus the restriction of the injection @xmath242 to @xmath241 is an injection into the cohomology of those components @xmath66 of @xmath30 which lie in @xmath241 .
thus @xmath243 for all @xmath244 implies @xmath245 .
in particular @xmath246 but @xmath247 therefore , @xmath248 thus @xmath249 , or equivalently @xmath250 .
it follows that any class in the ideal generated by @xmath251 lies in @xmath252 . to show that the inclusion @xmath253 is an equality , we prove that @xmath254 as graded ideals .
we prove this in the case @xmath42 , as the more general case is identical but notationally more cumbersome .
as @xmath1 is a hamiltonian @xmath0-space and @xmath255 is a hamiltonian @xmath5-space , they are both equivariantly formal with respect to their group actions .
this implies that as graded vector spaces @xmath256 and @xmath257 recall that @xmath258 and @xmath259 are surjective ( theorems [ th : equvtsurjective ] and [ th : surjective ] ) .
thus there is a graded equality @xmath260 where the last equality follows from theorem [ th : tw ] .
we now show that @xmath261 has the same dimension . in degree @xmath33 for @xmath80 , @xmath262 where @xmath117 if and only if @xmath263 .
let @xmath264 be the surjective map which forgets the @xmath5 action . because @xmath265 for generic @xmath2 , if @xmath266 has the property that @xmath267 , then @xmath268 has the property that @xmath269 .
furthermore , for @xmath80 , @xmath270 so that such classes are precisely those in @xmath271 . as @xmath272 in all but degree 0 ,
we conclude that @xmath273
in this section , we prove an important refinement of theorem [ th : kernel ] .
it states that the collection of @xmath274 for a small number of @xmath194 are sufficient to generate the kernel of the map @xmath275 .
in particular , one can consider only such @xmath14 which are perpendicular to codimension-1 walls of the moment polytope .
we illustrate the main theorem of this section with an example .
let @xmath1 be a generic 6-dimensional coadjoint orbit of @xmath276 .
the maximal torus @xmath277 of @xmath276 acts on this orbit in a hamiltonian fashion .
the image of the moment map is a hexagon , and the codimesion-1 walls of the moment polytope are shown in figure [ fig : su3coadjointorbit](a ) .
consider the reduction at the point @xmath38 indicated . according to theorem [ th : tw ] , a generating set for the kernel of the map @xmath278 would include classes which restrict to 0 on fixed points whose images under the moment map ( indicated by vertices ) lie to one side of any hyperplane through @xmath38 .
in particular , the class @xmath93 whose restrictions to the fixed points is indicated by figure [ fig : su3coadjointorbit](b ) would be a generator of the kernel because it is 0 to one side of the hyperplane @xmath279 .
theorem [ th : wallssufficient ] states that such a class is redundant ; it will in fact be generated by classes which are 0 to one side of a hyperplane through @xmath38 _ parallel to a wall _ of the moment polytope . in figures [ fig : su3coadjointorbit](c ) and [ fig : su3coadjointorbit](d ) we see two classes which are in the kernel according to theorem [ th : wallssufficient ] and whose sum is @xmath93 .
more generally , let @xmath66 be a connected fixed point component of the @xmath0-action on @xmath1 , and @xmath194 generic so that @xmath66 is a critical manifold for @xmath18 .
let @xmath280 be the extended stable manifold of @xmath66 under @xmath18 .
the image @xmath281 of @xmath280 under the moment map is not _ a priori _ convex .
suppose @xmath282 .
consider the collection of codimension-1 walls of @xmath283 that lie in @xmath281 .
these are the images of @xmath191-fixed point sets in @xmath280 .
at least one of these walls , translated to pass through @xmath66 , has the property that @xmath281 lies entirely to one side .
if @xmath281 is not maximal dimension , then every codimension-1 wall containing @xmath281 has this property .
we are now ready to state and prove in what sense the walls of the moment polytope are sufficient information for calculating the kernel of @xmath57 .
[ th : wallssufficient ] let @xmath284 consist of hyperplanes through @xmath38 and parallel to codimension-1 walls of the moment polytope .
let @xmath285 be the ideal generated by classes @xmath63 which restrict to 0 on all connected components @xmath66 of @xmath30 whose images under @xmath20 lie to one side of some @xmath286
. then @xmath287 , where @xmath91 is the kirwan map
. choose any @xmath63 where @xmath288 .
by theorem [ th : tw ] , @xmath289 .
we want to show that @xmath93 can be written as a linear combination of elements in @xmath290 where @xmath291 are the annihilators of the hyperplanes through @xmath170 and parallel to codimension-1 walls of the moment polytope . without loss of generality ,
assume @xmath292 for some @xmath14 , where @xmath293 are classes restricting to 0 on fixed points whose image under @xmath20 lies to one side of @xmath294 .
order the connected components of the critical sets @xmath295 so that @xmath117 if and only if @xmath296 .
we prove that @xmath93 can be expressed as a sum of elements in @xmath297 , by induction on the index of the critical sets .
let @xmath152 be the first critical set such that @xmath298 .
then @xmath299 is some multiple @xmath300 of @xmath301 .
let @xmath302 be the extended stable manifold of @xmath152 and let @xmath303 be any class satisfying the properties of lemma [ le : flowupclass ] . in particular ,
@xmath304 where @xmath305 is perpendicular to a codimension-1 wall @xmath279 of @xmath283 such that @xmath306 lies to one side of @xmath279 shifted to pass through @xmath38 .
then @xmath307 is a class which restricts to 0 on @xmath308 .
now suppose that @xmath309 restricts to 0 on @xmath310 .
let @xmath311 be the first critical set on which this class is non - zero .
use lemma [ le : flowupclass ] to find a class @xmath312 supported on @xmath313 .
then @xmath314 is 0 on @xmath315 . in this manner
, we express @xmath316 , where each @xmath317 for some choice of @xmath318 perpendicular to a codimension-1 wall of the polytope .
let @xmath319 be the product of symplectic manifolds @xmath320 , each with a hamiltonian @xmath2 action .
theorem [ th : wallssufficient ] allows us to say a lot about the reduction of @xmath1 by the diagonal torus action .
we note that the diagonal torus @xmath321 is a subtorus of the product @xmath322 acting on @xmath1 . if the torus is just one - dimensional , theorem [ th : wallssufficient ] is not more useful than the original formulation of theorem [ th : tw ] .
it is when @xmath323 that one can significantly reduce the number of vectors needed to generate the kernel of @xmath324 .
we calculate the cohomology of the reduced space for a product of two copies of @xmath8s , where we have quotiented by the diagonal @xmath325 action . by the moment map condition ( [ eq : momentmapcondition ] ) , if we choose the symplectic forms @xmath21 on @xmath326 and @xmath327 , @xmath328 on @xmath329 , the image for the moment map for the @xmath2 action on @xmath329 is that for the action of @xmath2 on @xmath326 dilated by @xmath33 .
the image of the moment map for the diagonal @xmath2 action on @xmath330 is the sum of the moment maps for each component .
let @xmath331 and @xmath2 act on @xmath320 by @xmath332=[z_0 : e^{i\theta_1}z_1 : e^{i\theta_2}z_2]$ ] .
choose @xmath333 and let @xmath334 be a moment map for the diagonal action .
then @xmath335 with its walls is pictured in figure [ fig : cp2productmomentmap ] .
the image of fixed points are labeled and indicated by vertex dots , and the image of fixed point sets of codimension-1 tori are indicated by line segments , which in this case form codimension-1 walls of the moment polytope .
we find the cohomology of the symplectic reduction @xmath336 where @xmath38 is the point indicated in the figure by finding the kernel of @xmath337 using theorem [ th : wallssufficient ] .
we note that there are three distinct hyperplanes through @xmath38 parallel to walls of the moment polytope : horizontal , vertical , and diagonal with slope -1 .
consider first @xmath338 . by the equivariant knneth theorem , @xmath339 where @xmath340 is the diagonal torus .
@xmath341 is generated in degree 2 , by characteristic classes inherited from the module structure @xmath342 , and the equivaraint symplectic form on @xmath8 .
we use the chosen basis for @xmath343 and denote the characteristic classes by @xmath344 and @xmath345 .
these classes restrict to themselves on each fixed point of @xmath2 on @xmath8 .
let @xmath151 be ( the multiple of ) the equivariant symplectic class given by the restrictions indicated in figure [ fig : symplres ] .
similarly , @xmath344 and @xmath345 are degree 2 equivariant classes on @xmath346 which restrict to themselves at every fixed point of the @xmath321 action on the product . by ( [ eq : kunneth ] ) the classes @xmath347 and @xmath348 generate @xmath349 . here @xmath350 . note that a fixed point of @xmath346 is a pair @xmath351 .
the restriction of a class on the product space to a fixed point is @xmath352 where @xmath353 .
the algebraic structure of these classes is inherited by multiplication on each fixed point .
the betti numbers for @xmath349 are easy to compute . as graded vector spaces , @xmath354 .
it follows that the equivariant poincar polynomial for @xmath346 is @xmath355 we noted above that @xmath356 and @xmath357 are four linearly independent degree 2 classes .
a choice of 10 linearly independent degree 4 classes is @xmath358 and @xmath359
. the restrictions of these classes to the fixed point set @xmath360 is determined by the formula ( [ eq : productrestriction ] ) . as an example
, we compute the restriction of @xmath361 to the fixed point set
. see figure [ fig : x2restriction ] .
we have left to find the kernel of @xmath362 .
the reduced space @xmath363 has ( real ) dimension @xmath364 thus we expect all classes of degree 6 or higher to be in the kernel of @xmath365 . in degree 2
, we can easily see that no linear combination of the four classes above will restrict to 0 on one side of any of the three hyperplanes parallel to walls of the moment polytope . by theorem
[ th : wallssufficient ] , there are no degree 2 classes in @xmath366 . in degree 4 , however , we expect nine ( of ten ) linearly independent classes to be in the kernel , as the image of @xmath57 is ( a multiple of ) the volume form on the reduction .
the reader can verify using ( [ eq : productrestriction ] ) that the following classes restrict to 0 on one side of one of the three hyperplanes through @xmath38 parallel to a wall of the moment polytope : @xmath367 and @xmath368 .
they are all linearly independent , which can be verified rather tediously . as an example
, one can see in figure [ fig : x2restriction ] that @xmath361 restricts to 0 on one side of the diagonal hyperplane through @xmath38 . | let @xmath0 be a compact torus acting on a compact symplectic manifold @xmath1 in a hamiltonian fashion , and @xmath2 a subtorus of @xmath0 .
we prove that the kernel of @xmath3 is generated by a small number of classes @xmath4 satisfying very explicit restriction properties . our main tool is the _ equivariant kirwan map _ , a natural map from the @xmath0-equivariant cohomology of @xmath1 to the @xmath5-equivariant cohomology of the symplectic reduction of @xmath1 by @xmath2 .
we show this map is surjective .
this is an equivariant version of the well - known result that the ( nonequivariant ) kirwan map @xmath3 is surjective .
we also compute the kernel of the equivariant kirwan map , generalizing the result due to tolman and weitsman @xcite in the case @xmath6 and allowing us to apply their methods inductively .
this result is new even in the case that @xmath7 .
we close with a worked example : the cohomology ring of the product of two @xmath8s , quotiented by the diagonal 2-torus action . |
the theory of classical thermodynamics has been well supported by the methods of equilibrium statistical physics , which in turn is based on the boltzmann probability distribution for the states of a system in phase space . for systems that are slightly away from equilibrium ( where the perturbation to the system can be assumed to be linearly
coupled to the applied weak force field ) , the linear response theory characterizes the response of the system in terms of equilibrium correlation functions . however , there are only very few exact results that remain valid even when the system is driven arbitrarily out of equilibrium .
the fluctuation theorems have been one of them .
first introduced for thermostatted systems following dissipative but deterministic dynamics by evans and searles @xcite , it has produced several closely related equalities @xcite , some theorems holding in general for all systems while the others being satisfied for special cases ( for example , in the limit of a large observation time ) . in a nutshell , these theorems have the following general form : @xmath0 where @xmath1 is some observable that is to be measured and is in general a path function , @xmath2 is the probability density of this observable along a process ( usually parametrized by an externally controlled time - dependent protocol @xmath3 ) and @xmath4 is the probability density whose functional form is related to @xmath5 through a conjugate transformation which is not necessarily time - reversal .
its argument @xmath6 is the value assumed by the observable along the transformed trajectory in phase space .
some observables that have been shown to follow the fluctuation theorems are entropy , work and heat @xcite .
it is easy to see that the theorems are closely related to the second law , in the sense that the averaged observable @xmath7 always abides by the inequalities dictated by the second law .
for instance , the total entropy of the system plus environment can be shown to give rise to the inequality @xcite @xmath8 what this inequality says is that , even though some individual realizations of the experiment may observe a decrease in the total entropy , its _ average _ value must never decrease with time .
an issue that has recently attracted much interest is the use of feedback - controlled protocol to enhance the efficiency of such a process @xcite . instead of using a predetermined protocol
, we can perform intermediate measurements on some observable and modify our protocol accordingly .
we are in general not interested in the measuring device involved in the experiment ( which definitely belongs to the `` universe '' in which second law must be obeyed ) .
it is obvious that the form of the second law , now dealing with only a part of the universe under consideration , needs to be modified .
as expected , the correction term involves the properties of the ignored subpart , in the form of the mutual information gained through measurements and gives rise to the following inequality : @xmath9 here , the mutual information is defined as @xcite @xmath10 where the presence of measurement errors is assumed . in the arguments of the conditional probability @xmath11 , @xmath12 is the outcome when a measurement is performed , while @xmath13 is the actual value of the observable .
the averaging has been done over all possible phase space trajectories _ as well as _ over all possible protocol functions , which results due to the different feedback rules corresponding to different outcomes .
since @xmath14 is a relative entropy and consequently always non - negative @xcite , the above inequality provides us with the privilege of extracting work from or obtaining a decrease in entropy of the system , provided we can formulate a nice feedback algorithm . note that eq .
( [ law2_mod ] ) only says that the lower bound for the dissipated work ( or entropy ) _ can _ be decreased by an amount proportional to the average mutual information .
however , it does not provide any clue to what a good feedback algorithm should be . in this sense , it is reasonable to look for a quantity that actually provides a quantitative measure of how efficient an employed feedback procedure is .
this is the so - called _ efficacy parameter _
@xcite defined below . in this paper , we clarify the physical meaning and some of the properties of this parameter . the jarzynski equality @xcite has two different generalized forms in the presence of feedback .
the more commonly used form of the extended jarzynski equality ( eje ) is @xcite @xmath15 - i[x , m]}\ra = 1 .
\label{ejea}\ ] ] @xmath16 is the dissipated work defined through @xmath17 , where @xmath18 is the difference between free energies at the end and at the beginning of a particular protocol . here , the reverse trajectories are generated by simply reversing the sequence of one of the forward protocols and we have used upper case letters to denote the full path of successive phase points or measured outcomes : @xmath19 , @xmath20 . the time has been discretized as @xmath21 .
we will set @xmath22 and @xmath23 .
as we will show in this paper , @xmath24 $ ] will in general be replaced by a different physical quantity @xmath25 $ ] , if we choose to use feedback along the reverse process as well , as will be detailed below .
other than eq .
( [ ejea ] ) , there is yet another form of the extended jarzynski equality that has been introduced in the literature @xcite : @xmath15}\ra = \gamma .
\label{ejeb}\ ] ] the efficacy parameter @xmath26 is a functional of the feedback control used along the forward process , and that determines the extent to which the feedback is efficient ( i.e. , more work can be extracted from the system ) . if @xmath27 , then we would have the jarzynski equality in absence of feedback .
the efficacy parameter has been measured experimentally @xcite and equations ( [ ejea ] ) and ( [ ejeb ] ) have been verified . using the jensen s inequality , we have the dissipated work bounded from below through the relation @xmath28\ra \ge -\ln\gamma\ ] ] although this inequality looks similar to the one stated before , it may seem to be a trivial statement because it is a simple consequence of the definition of @xmath26 . however , the fact that this definition lends a very clear physical meaning to the efficacy parameter and that this meaning can be exploited to experimentally measure @xmath26 ( without using the definition eq .
( [ ejeb ] ) ) underlines the importance of the above bound .
we explicitly show that @xmath26 retains the simple meaning when we extend eq .
( [ ejeb ] ) to driven systems that make excursions among nonequilibrium steady states under arbitrary feedback - controlled protocols , whereas the physical meaning of @xmath25 $ ] depends on the design of feedback along the reverse process .
as shown in @xcite , in presence of information gain and feedback applied _ only _ along the forward trajectory , the jarzynski equality gets modified to eq .
( [ ejea ] ) .
this relation is easily derived from the detailed fluctuation theorem , in the case when the reverse protocol is the blind time - reversal of the corresponding protocol along the forward process .
then the ratio of the forward and reverse trajectories become @xcite @xmath29}{\tilde p[\tilde x;m ] } & = \frac{p[x]}{\tilde p[\tilde x ] } \times \frac{p(m_1|x_1)p(m_2|x_2)\cdots p(m_n|x_n)}{p(m_1,m_2,\cdots , m_n ) } , \nn\\ & = e^{\beta w_d[x]+i[x , m]}. \label{eje1 } \end{aligned}\ ] ] in the above equation , @xmath30 is the time - reversed trajectory @xmath31 , and @xmath32 $ ] is the probability density for this time - reversed trajectory ( for simplicity of notation , we would always use the forward time @xmath33 even during the reverse process , in place of using the actual time @xmath34 elapsed during the reverse process ) .
if the state is denoted only by the position variable , then we have @xmath35 .
@xmath36 $ ] is the probability path functional for the _ actual reverse process _
being carried out .
this is different from @xmath37 $ ] which gives the probability of obtaining the reverse phase space trajectory along with the reversed measured outcomes , provided measurements are performed at the same time instants as in the forward process .
the expression for the latter quantity has no explicit reference to the actual feedback procedure followed to generate the reverse process ( i.e. , its physical meaning does not change due to change in feedback procedure for reverse process ) .
a simple cross - multiplication followed by integration over @xmath38 and @xmath39 will give rise to the modified jarzynski equality , eq .
( [ ejea ] ) .
the reverse process can also be defined by designing a suitable feedback procedure to generate the time - reversed protocol , which does not violate causality @xcite .
the feedback procedure is as follows : we first measure system observable at time @xmath40 , and if the measurement outcome is equal to the forward outcome at time @xmath41 , then we drive the system using the reverse protocol @xmath42 ( note that the time instants for measurement along the backward process are not the same as the forward ones , but are shifted ) . continuing with this process , i.e. , changing the protocol according to measurement outcome @xmath43 at time @xmath44 , the reverse protocol is exactly reproduced without ever violating causality . this the new relation @xmath45}{\tilde p[\tilde x;m ] } = e^{\beta w_d[x]+\delta s_p[x , m ] } , \label{eje2}\ ] ] where @xmath46 \equiv \ln\frac{p(m_0|x_0)p(m_1|x_1)\cdots p(m_{n}|x_{n})}{p(m_n|\tilde
x_{n+1})p(m_{n-1}|\tilde x_{n})\cdots p(m_0|\tilde x_1)}.\ ] ] here , @xmath47 represents a disorder parameter , but is not the mutual information as defined in @xcite .
of course , equations ( [ eje1 ] ) and ( [ eje2 ] ) contain different information as well as provide different bounds for the average total entropy change for the system .
to @xmath48 , we have the feedback - controlled reverse protocol , while from @xmath49 to @xmath44 we have used the blindly time - reversed protocol . in the final part , up to @xmath50 , we revert to the feedback - controlled protocol . ]
[ case : iii ] the general form of the probability density for a forward trajectory in presence of feedback is given by @xcite @xmath51 = & p(m_0|x_0)p_{m_0}[x_0\to x_1]p(m_1|x_1)p_{m_1}[x_1\to x_2 ] \nn\\ & \cdots p(m_n|x_n)p_{m_n}[x_n\to x_\tau].\end{aligned}\ ] ] the general protocol to generate the reverse process would be to use both the protocols [ case : i ] and [ case : ii ] at random during the reverse process . let us take one simple case when , along the reverse process , up to time @xmath52 , we use the feedback - controlled protocol , then from time @xmath48 to @xmath44 , we use the blindly applied reverse protocol , and finally from @xmath44 to @xmath50 , we once again use the feedback - controlled reverse protocol ( see figure [ fig1 ] ) . here
, @xmath53 and @xmath54 can be any two integers chosen at random from the set @xmath55 . in this case , the reverse process becomes , @xmath56 = p(\tilde x_\tau ) p(m_n|\tilde x_\tau ) p_{m_n}[\tilde x_\tau\to \tilde x_{n } ] \nn\\ & \hspace{1 cm } p(m_{n-1}|\tilde x_n ) p_{m_{n-1}}[\tilde x_n\to \tilde x_{n-1 } ] \nn\\ & \cdots\hspace{0.5 cm } p(m_{n}|\tilde x_{n+1 } ) p_{m_n}[\tilde x_{n+1}\to \tilde x_n ] \nn\\ & \hspace{1 cm } p(m_{n-1},\cdots , m_k ) p[\tilde x_n \to \tilde x_k ] p(m_{k-1}|\tilde x_k ) \nn\\ & \hspace{1 cm } p[\tilde x_k \to \tilde
x_{k-1 } ] \cdots p(m_0|\tilde x_1 ) p[\tilde x_1\to \tilde x_0 ] .
\label{genrev } \end{aligned}\ ] ] therefore , we arrive at @xmath29}{\tilde p[\tilde x;m ] } = & \frac{p[x]}{\tilde p[\tilde x]}~ \frac{p(m_k|x_k)\cdots p(m_{n-1}|x_{n-1})}{p(m_{k},\cdots , m_{n-1 } ) } \nn\\ & \times \frac{p(m_0|x_0)\cdots p(m_{k-1}|x_{k-1})}{p(m_0|\tilde x_1)\cdots p(m_{k-1}|\tilde x_k ) } \nn\\ & \times \frac{p(m_n|x_n)\cdots p(m_n|x_n)}{p(m_n|\tilde x_{n+1})\cdots
p(m_n|\tilde x_\tau ) } \nn\\ = & \exp\left[{\beta w_d+i^1 + \delta s_p^1 + \delta s_p^2}\right ] , \label{eje3 } \end{aligned}\ ] ] where @xmath57 we are thus led to a different extended detailed fluctuation theorem where disorder parameters @xmath58 , @xmath59 and @xmath60 are different as they contain different information about the feedback process .
moreover , these are not unique and may depend on different reverse protocols not considered here .
thus , it is clear that the @xmath25 $ ] in eq .
( [ eje1 ] ) does not have a unique interpretation and clearly depends on the manner of feedback along the backward process . from eq .
( [ eje3 ] ) , one can obtain the related integral fluctuation theorem , namely @xmath61\ra = 1\ ] ] all the three results derived above , eqs .
( [ eje1 ] ) , ( [ eje2 ] ) and ( [ eje3 ] ) , can be written in compact form as @xmath45}{\tilde p[\tilde x;m ] } = e^{\beta w_d+\phi[x , m ] } , \label{eje_gen}\ ] ] where @xmath25 $ ] is a functional of the phase space space trajectory as well as of the measurement trajectory ( i.e. , sequence of measured outcomes ) .
we next turn our attention to the efficacy parameter .
in absence of feedback , the jarzynski equality is given by @xcite @xmath15}\ra = 1.\ ] ] in presence of feedback , the right hand side of the above relation will in general not be unity .
the efficacy parameter for the feedback in this case ( when system is initially at thermal equilibrium with the medium ) is defined as @xmath62}\ra .
\label{eff}\ ] ] it can be readily seen that lesser the amount of dissipated work , more is the magnitude of the efficacy parameter , as is desirable for a quantity that decides the efficiency of feedback .
now we use the most general case for obtaining the reverse trajectories , namely the case [ case : iii ] .
therefore , we get @xmath63 p[x , m ] e^{-w_d[x , m ] } \nn\\ & = \int \mathcal{d}[x , m ] \tilde p[\tilde x;m]e^{\delta s_p^1 + i^1 + \delta s_p^2 } \nn\\ & = \int \mathcal{d}[x , m ] \tilde p[\tilde x ] p(m_0|x_0 ) p(m_1|x_1 ) \nn\\ & \hspace{4cm}\cdots p(m_n|x_n ) \nn\\ & = \int \mathcal{d}[\tilde x,\tilde m ] \tilde p[\tilde x ] p(\tilde m_0|\tilde x_0 ) \cdots p(\tilde m_n|\tilde
x_n ) \nn\\ & = \int \mathcal{d}[\tilde x,\tilde m ] \tilde
p[\tilde x,\tilde m ] = \int \mathcal{d}[\tilde m ] \tilde p[\tilde m ] , \label{eff_je}\end{aligned}\ ] ] where we have used the detailed fluctuation theorem ( [ eje_gen ] ) , and the definitions ( [ sp1 ] ) , ( [ i1 ] ) and ( [ sp2 ] ) .
the time - reversibility of measurements has been assumed : @xmath64 @xcite .
once again we may recall that the two quantities @xmath36 $ ] and @xmath37 $ ] , both of which appear in the above derivation , are different . in more general cases ,
the steps 2 and 3 can be written as @xmath65 & e^{\phi[x , m ] } = \tilde p[\tilde x;m ] \exp\left(\sum_i\delta s_p^i + \sum_j i^j\right ) \nn\\ & = \tilde p[\tilde x ] p(m_1|x_1 ) \cdots p(m_n|x_n ) \nn\\ & = \tilde p[\tilde x ] \tilde p(\tilde m_1|\tilde x_1 ) \cdots \tilde p(\tilde m_n|\tilde
x_n ) \nn\\ & = \tilde p[\tilde x,\tilde m].\end{aligned}\ ] ] here , the summations @xmath66 and @xmath67 run over all the time intervals in which the reverse protocols have been executed using feedback and by blind time - reversal , respectively .
we find that although the form of @xmath36 $ ] contains detailed information about the actual feedback procedure used along the backward process , when it is multiplied by the factor @xmath68}$ ] , we obtain @xmath37 $ ] whose form does not contain any such information .
this is the reason behind the fact that the efficacy parameter retains the same physical meaning in each case , namely , it is the _ total probability to observe the time - reversed outcomes for the measurements performed along the reverse process _ @xcite .
this derivation of @xmath26 for the most general case is our central result .
it simply shows that it retains the same meaning as for the specific cases considered in the earlier literature @xcite .
we will be generalizing our treatment to the other detailed fluctuation theorems , which involve the non - adiabatic entropy production ( change in entropy due to shift between two different steady states ) and adiabatic entropy production ( change in entropy caused due to maintenance of a given steady state ) .
the total entropy change by definition is the sum of entropy changes in the system ( @xmath69 ) and in the medium ( @xmath70 ) : @xmath71 .
recently it has been observed that while generalizing the second law for systems making transitions between steady states , the total entropy production can also be split into two distinct parts such that each part , interestingly , follows a detailed fluctuation theorem @xcite : @xmath72 the averages of all these three quantities are always non - negative , thereby providing a new twist to the second law .
@xmath73 is related to the housekeeping heat @xmath74 , while @xmath75 is the sum of the entropy change of the system and the entropy produced due to excess heat @xmath76 @xcite . in the case of adiabatic and non - adiabatic entropy productions ,
the concept of dual dynamics is very helpful . under the dual dynamics ,
if the system is allowed to reach the corresponding steady state , then the steady - state distribution @xmath77 retains the same form , but the probability current reverses its sign @xcite . hatano and sasa had shown that the physical meaning of the nonadiabatic entropy becomes clear in the dual dynamics formalism @xcite .
these detailed fluctuation theorems are taken up in the following discussion .
when the initial state of the system is not at thermal equilibrium with the bath , then in absence of feedback , the following ratio is obtained between the forward and the reverse trajectories @xcite : @xmath78}{\tilde p[\tilde x ] } = e^{\delta s_{tot}[x]},\ ] ] from which the following integral fluctuation theorem can be obtained : @xmath79}\ra = 1.\ ] ] in presence of feedback , the right hand side will in general be different from unity .
for this general case , instead of eq .
( [ eje_gen ] ) , we would get the following ratio between the forward and reverse paths : @xmath45}{\tilde p[\tilde x;m ] } = e^{\delta s_{tot}[x]+\phi[x , m]}.\ ] ] we now consider the case with general reverse protocol .
we define the efficacy parameter as @xmath80}\ra \end{aligned}\ ] ] proceeding in exactly the same way as before , we find @xmath81 p[x , m ] e^{-\delta s_{tot}[x , m ] } \nn\\ & = \int \mathcal{d}[x , m ] \tilde p[\tilde x;m ] e^{\phi[x , m]}\nn\\ & = \int \mathcal{d}[\tilde m ] \tilde p[\tilde m].\end{aligned}\ ] ] thus , @xmath82 retains the same physical meaning as @xmath26 for the jarzynski equality , although here we do not have the constraint of sampling the initial state of the system from the boltzmann distribution . for transitions between nonequilibrium steady states
, we have the following detailed fluctuation theorem in absence of feedback @xcite : @xmath78}{\tilde p^\dagger[\tilde x ] } = e^{\delta s_{na}[x]}.\ ] ] the superscript @xmath83 implies dual dynamics , which implies that keeping the functional form of the protocol same , we are switching to a dynamics where the steady state current simply changes direction as compared to the original dynamics .
the tilde symbol over @xmath5 implies that the protocol for the forward process has been time - reversed after the system has been allowed to follow the dual dynamics . in other words ,
@xmath84 $ ] is the probability density for a trajectory along the process generated , in presence of dual dynamics , by the time - reversed protocol .
similar to the above cases , in presence of feedback , we have @xcite @xmath45}{\tilde p^\dagger[\tilde x;m ] } = e^{\delta s_{na}[x]+\phi[x , m]},\ ] ] where the form of @xmath25 $ ] depends on the way in which feedback is applied along the reverse trajectory , as given in section 2 .
the efficacy parameter in this case is given by @xmath85}\ra = \int \mathcal{d}[x , m ] p[x , m ] e^{-\delta s_{na}[x , m ] } \nn\\ & = \int \mathcal{d}[x , m ] \tilde p^\dagger [ \tilde x;m ] e^{\phi[x , m ] } \nn\\ & = \int \mathcal{d}[x , m ] \tilde p^\dagger [ \tilde x ] p(m_1|x_1)\cdots p(m_n|x_n ) \nn\\ & = \int \mathcal{d}[x , m ] \tilde p^\dagger [ \tilde x,\tilde m ] = \int \mathcal{d}[\tilde m]\tilde p^\dagger [ \tilde m ] .
\end{aligned}\ ] ] in the third step , we have used the algebra that has already been shown in the case of the most general protocol ( section [ gamma_je ] ) for extended jarzynski equality , which results in @xmath86 & e^{\phi[x , m ] } = \tilde p^\dagger[\tilde x ] \exp\left(\sum_i\delta s_p^i + \sum_j i^j\right ) \nn\\ & = \tilde p^\dagger[\tilde x ] p(m_1|x_1 ) \cdots p(m_n|x_n ) \nn\\ & = \tilde p^\dagger[\tilde x ] \tilde p^\dagger(\tilde m_1|\tilde x_1 ) \cdots \tilde p^\dagger(\tilde m_n|\tilde
x_n ) \nn\\ & = \tilde p^\dagger[\tilde x,\tilde m].\end{aligned}\ ] ] we have assumed that the measurement errors do not change on changing the dynamics , which is quite reasonable assumption .
thus , @xmath87 is the net probability for obtaining the time - reversed outcomes along the time - reversed dual dynamics ( i.e. , the process in which the protocol corresponding to the dual dynamics is time - reversed ) .
the dft for adiabatic entropy production is given by @xcite @xmath78}{p^\dagger[x ] } = e^{\delta s_{a}[x]}.\ ] ] @xmath88 $ ] is the probability density for the path followed by the system in phase space , when the system is evolving under the dual dynamics . in presence of feedback , we then have , @xmath45}{p^\dagger[x;m ] } = e^{\delta s_{a}[x]+\phi[x , m]}.\ ] ] since there is no time - reversal involved , the denominator can only consist of the following options : \(1 ) the same feedback procedure is used to generate the forward process in the dual dynamics as well , in which case we have @xmath25=0 $ ] ( since the error probabilities in the numerator cancel with those in the denominator ) .
once again , measurement errors are assumed to be independent of the dynamics followed by the system .
\(2 ) one of the forward protocols in the original dynamics is recorded , and this protocol is blindly executed in presence of the dual dynamics , in which case we simply have @xmath25 = i[x , m]$ ] .
\(3 ) we use the above two procedures at random while generating the forward trajectories in presence of dual dynamics , which is the most general case . in this case , however , @xmath25=\sum_j i^j$ ] , i.e. , the summation over @xmath89 will be absent , because the latter quantity never appears in this case .
the efficacy parameter is @xmath90}\ra,\ ] ] which leads to @xmath91 p[x , m ] e^{-\delta s_{a}[x , m ] } \nn\\ & = \int \mathcal{d}[x , m ] p^\dagger[x;m ] e^{\sum_j i^j } = \int \mathcal{d}[m]p^\dagger[m].\end{aligned}\ ] ] therefore , @xmath92 is the total probability for observing the same outcomes as the initial process with the same protocol , if the system follows the dual dynamics .
we thus find that the physical meaning of efficacy parameter can be very generally stated as follows : _ it is the total probability to observe the measured outcomes conjugate to those along the forward protocol , for the intermediate measurements along the process with the corresponding conjugate dynamics .
_ since the efficacy parameters are experimentally measurable , they would provide more meaningful bounds for the @xmath93 , and these bounds are in fact independent of whether or not feedback is performed along the conjugate process in the actual protocol , but only requires measurements to be performed along the conjugate process .
for the other bounds stated in section 2 , the values would depend sensitively on whether and how the feedback is performed along the conjugate dynamics , namely , the extended integral fluctuation theorems can be stated as @xmath94 for @xmath95 representing @xmath96 , @xmath75 and @xmath73 , respectively , and this would lead to the bounds @xmath97 as a consequence , arbitrary number of bounds can be computed for this latter case , which is not only confusing , but undermines the motivation behind such calculations .
thus , the efficacy parameter is a far more suitable experimentally measurable quantity that can characterize not only the performance of the system , but can also act as a useful lower bound for the averaged observables .
in this paper , we have shown that out of the two known forms of the modified fluctuation theorems in presence of feedback , one of the forms is heavily dependent on the way feedback is applied along the conjugate process , and thereby leads to arbitrary number of lower bounds for work done on the system or for the relevant entropy changes ( total entropy , nonadiabatic and adiabatic entropies ) taking place .
on the other hand , the second form , namely the fluctuation theorem expressed in terms of the efficacy parameter , provides a bound for work and entropy changes that carries a clear and consistent physical meaning , irrespective of the manner of application of feedback along the conjugate process .
this consistency is robust even when the conjugate process is not the time - reversed process .
this study would hopefully help in simpler experimental verification of the extended fluctuation theorems .
one of us ( amj ) thanks dst , india for financial support .
we thank dr .
anupam kundu for sending us the preprint of his manuscript prior to its publication .
d. j. evans , e. g. d. cohen and g. p. morriss , .
d. j. evans and d. j. searles , . c. jarzynski , . c. jarzynski , .
g. e. crooks , .
g. e. crooks , .
u. seifert , .
u. seifert , .
m. campisi , p. hnggi and p. talkner , .
u. seifert , arxiv:1205.4176 .
t. sagawa and m. ueda , .
t. sagawa and m. ueda , .
t. sagawa and m. ueda , .
j. m. horowitz and s. vaikuntanthan , .
m. ponmurugan , .
s. lahiri , s. rana and a. m. jayannavar , .
s. rana , s. lahiri and a. m. jayannavar , arxiv:1202.1097 .
s. rana , s. lahiri and a. m. jayannavar , arxiv:1111.6229 ; accepted for publication in pramana j. phys .
d. abrieu and u. seifert , .
f. j. cao and m. feito , .
t. m. cover and j. a. thomas , _ elements of information theory _ ,
2nd edn ( hoboken , nj : wiley - interscience ) .
s. toyabe et al , nature physics * 6 * , 988 ( 2010 ) .
a. kundu , private communication .
m. esposito and c. van den broeck , .
m. esposito and c. van den broeck , .
m. esposito and c. van den broeck , .
r. garcia - garcia , v. lecomte , a. b. kolton and d. dominguez , . v. y. chernayk , m. chertkov and c. jarzynski , .
t. hatano and s. sasa , . | the fluctuation theorems have remained one of the cornerstones in the study of systems that are driven far out of equilibrium , and they provide strong constraints on the fraction of trajectories that behave atypically in light of the second law .
they have mainly been derived for a predetermined external drive applied to the system .
however , to improve the efficiency of a process , one needs to incorporate protocols that are modified by receiving feedbacks about the recent state of the system , during its evolution .
in such a case , the forms of the conventional fluctuation theorems get modified , the correction term involving terms that depend on the way the reverse / conjugate process is defined , namely , the rules of using feedback in order to generate the exact time - reversed / conjugate protocols .
we show in this paper that this can be done in a large number of ways , and in each case we would get a different expression for the correction terms .
this would in turn lead to several lower bounds on the mean work that must be performed on the system , or on the entropy changes . here
we analyze a form of the extended fluctuation theorems that involves the efficacy parameter , and find that this form gives rise to a lower bound for the mean work that retains a consistent physical meaning regardless of the design of feedback along the conjugate process , as opposed to the case of the previously mentioned form of the modified fluctuation theorems .
[ 1 ] [ 2 ] [ 2 ] [ 3]phys . rev .
lett .
* # 1 * , # 2 ( # 3 ) [ 3]phys . rev .
e * # 1 * , # 2 ( # 3 ) [ 3]phys .
rev .
a * # 1 * , # 2 ( # 3 ) [ 2]j .
stat .
mech .
# 1 ( # 2 ) [ 3]eur .
phys .
j. b * # 1 * , # 2 ( # 3 ) [ 3]rev . mod .
phys . *
# 1 * , # 2 ( # 3 ) [ 3]europhys .
lett . *
# 1 * , # 2 ( # 3 ) [ 3]j .
stat .
phys . *
# 1 * , # 2 ( # 3 ) [ 3]prog .
theor .
phys .
suppl . *
# 1 * , # 2 ( # 3 ) [ 3]physics today * # 1 * , # 2 ( # 3 ) [ 3]adv .
phys . *
# 1 * , # 2 ( # 3 ) [ 3]j .
phys . condens .
matter * # 1 * , # 2 ( # 3 ) [ 3]j .
phys . a : math .
theor . *
# 1 * , # 2 ( # 3 ) [ 3]pram .
j. phys . *
# 1 * , # 2 ( # 3 ) [ 3]physica a * # 1 * , # 2 ( # 3 ) |
the expressions below are given in terms of scalar integral functions , @xmath74 and @xmath75 .
expressions for these functions are given in appendix d of ref .
@xcite .
@xmath17 : @xmath76 \\
\pi_{\gamma\gamma } \quad & = \quad -\frac{c_{bb}}{\lambda^2 } \frac{1}{16\pi^2 } \frac{g^2 q^2 \sin^2\theta_w}{m_h^2}\left[m_h^2 a_0\left(m_h^2\right ) + 3m_z^2 a_0\left(m_z^2\right ) \right .
\\ & \qquad \left .
+ 6m_w^2 a_0\left(m_w^2\right ) -4m_w^4 - 2m_z^4\right ] \\ \pi_{\gamma z } \quad & = \quad \frac{c_{bb}}{\lambda^2 } \frac{1}{16\pi^2 } \frac{g^2 \sin^3\theta_w}{m_h^2 \cos\theta_w } \left[m_h^2 m_z^2\left(m_h^2 - m_z^2 - q^2\right)b_0\left(q^2,m_h^2,m_z^2\right)\right . \\ & \qquad + m_z^2\left(m_h^2 + 3q^2\right ) a_0\left(m_z^2\right ) - m_h^2\left(m_z^2-q^2\right ) a_0\left(m_h^2\right ) \\ & \qquad \left .
+ 6 m_w^2 q^2 a_0\left(m_w^2\right ) - 4m_w^4 q^2 - 2m_z^4 q^2\right]\end{aligned}\ ] ] @xmath16 : @xmath77 \\
\pi_{zz } \quad & = \quad \frac{c_{ww}}{\lambda^2 } \frac{1}{16\pi^2 } \frac{g^2 \cos^2\theta_w}{m_h^2 } \left[2m_h^2 m_z^2\left(q^2 - m_h^2 + m_z^2\right ) b_0\left(q^2,m_h^2,m_z^2\right ) \right .
\\ & \qquad - m_z^2\left(3q^2 + 2m_h^2 + 3m_z^2\right ) a_0\left(m_z^2\right ) + m_h^2 \left(2m_z^2 - q^2\right ) a_0\left(m_h^2\right ) \\ & \qquad \left
- 6\left(m_z^4 + m_w^2 q^2\right ) a_0\left(m_w^2\right ) + 4 m_w^4 q^2 + 2 m_z^4 q^2 + 4m_w^2 m_z^4 + 2 m_z^6\right ] \\ \pi_{\gamma\gamma } \quad & = \quad -\frac{c_{ww}}{\lambda^2 } \frac{1}{16\pi^2 } \frac{g^2 q^2 \sin^2\theta_w}{m_h^2 } \left[m_h^2 a_0\left(m_h^2\right ) + 3 m_z^2 a_0\left(m_z^2\right ) \right . \\ & \qquad \left . + 6 m_w^2 a_0\left(m_w^2\right ) - 4m_w^4 - 2m_z^4\right ] \\
\pi_{\gamma z } \quad & = \quad -\frac{c_{ww}}{\lambda^2 } \frac{1}{16\pi^2 } \frac{g^2 \sin\theta_w \cos\theta_w}{m_h^2 } \left[m_h^2 m_z^2\left(m_h^2 - m_z^2 - q^2\right ) b_0\left(q^2,m_h^2,m_z^2\right ) \right . \\ & \qquad + m_z^2\left(m_h^2 + 3q^2\right)a_0\left(m_z^2\right ) -
m_h^2 \left(m_z^2 - q^2\right ) a_0\left(m_h^2\right ) \\ & \qquad \left .
+ 6 m_w^2 q^2 a_0\left(m_w^2\right ) - 4 m_w^4 q^2 - 2 m_z^4 q^2\right]\end{aligned}\ ] ]
@xmath78 99 s. weinberg , physica a * 96 * , 327 ( 1979 ) .
a. de rujula , m. b. gavela , p. hernandez and e. masso , nucl .
b * 384 * , 3 ( 1992 ) .
k. hagiwara , s. ishihara , r. szalapski and d. zeppenfeld , phys .
b * 283 * , 353 ( 1992 ) .
k. hagiwara , s. ishihara , r. szalapski and d. zeppenfeld , phys .
d * 48 * , 2182 ( 1993 ) .
d. c. kennedy and b. w. lynn , nucl .
b * 322 * , 1 ( 1989 ) .
m. e. peskin and t. takeuchi , phys .
d * 46 * , 381 ( 1992 ) .
n. greiner , s. willenbrock and c. zhang , phys .
b * 704 * , 218 ( 2011 ) [ arxiv:1104.3122 [ hep - ph ] ] . c. zhang , n. greiner and s. willenbrock , phys .
d * 86 * , 014024 ( 2012 ) [ arxiv:1201.6670 [ hep - ph ] ]
. s. weinberg , phys .
lett . * 43 * , 1566 ( 1979 ) .
s. alam , s. dawson and r. szalapski , phys .
d * 57 * , 1577 ( 1998 ) [ hep - ph/9706542 ] .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
j. alcaraz _ et al . _
[ aleph and delphi and l3 and opal and lep electroweak working group collaborations ] , hep - ex/0612034 .
r. barbieri , a. pomarol , r. rattazzi , a. strumia and , nucl .
b * 703 * , 127 ( 2004 ) [ hep - ph/0405040 ] .
m. c. gonzalez - garcia , s. m. lietti and s. f. novaes , phys .
d * 59 * , 075008 ( 1999 ) [ hep - ph/9811373 ] .
m. c. gonzalez - garcia , int .
j. mod .
a * 14 * , 3121 ( 1999 ) [ hep - ph/9902321 ] .
t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , phys .
d * 86 * , 075013 ( 2012 ) [ arxiv:1207.1344 [ hep - ph ] ] .
t. corbett , o. j. p. eboli , j. gonzalez - fraile , m. c. gonzalez - garcia and , phys .
d * 87 * , 015022 ( 2013 ) [ arxiv:1211.4580 [ hep - ph ] ] .
e. masso , v. sanz and , phys .
d * 87 * , 033001 ( 2013 ) [ arxiv:1211.1320 [ hep - ph ] ] .
w. -f .
chang , w. -p .
pan , f. xu and , arxiv:1303.7035 [ hep - ph ] .
g. passarino and m. j. g. veltman , nucl .
b * 160 * , 151 ( 1979 ) . | the one loop effects of two dimension - six operators on gauge boson self energies are computed within an effective field theory framework .
these self energies are translated into effects on precision electroweak observables , and bounds are obtained on the operator coefficients .
the effective field theory framework allows for the divergences that arise in the loop calculations to be properly handled , and for unambiguous bounds on the coefficients to be obtained .
we find that the coefficients are only weakly bounded , in contrast to previous calculations that obtained much stronger bounds .
we argue that the results of these previous calculations are specious .
+ mpp-2013 - 101 + cp3 - 13 - 13 * effective field theory of + precision electroweak physics at one loop *
+ + @xmath0department of physics , university of illinois at urbana - champaign + 1110 west green street , urbana , il 61801 + @xmath1max - planck - institut fr physik , fhringer ring 6 , 80805 mnchen , germany + @xmath2centre for cosmology , particle physics and phenomenology ( cp3 ) + universit catholique de louvain , b-1348 louvain - la - neuve , belgium in addition to searching for direct evidence of new physics , data can be probed for the indirect effects of new heavy particles .
the most general , model - independent framework for considering the indirect effects of new physics is effective field theory @xcite .
the new physics is parameterized by effective operators not present in the standard model .
the power of effective field theory is particularly on display when dealing with one - loop calculations .
the effective field theory framework provides a systematic means to deal with the divergences that arise in loop calculations , yielding a finite and unambiguous result . in this paper
, we continue an analysis begun in refs .
@xcite on the loop - level effects of effective operators on precision electroweak observables .
those papers focused on the divergent portions of the loop diagrams and a subset of the finite parts , and did not appreciate that unambiguous bounds could be obtained on the coefficients of the effective operators .
we use the full ( finite plus divergent ) expressions in order to obtain unambiguous bounds on a particular pair of effective operator coefficients .
these calculations involve only weak boson self energies , also called oblique corrections .
methods for organizing and applying such corrections to observables are well known @xcite and will be used throughout the analysis . because we are dealing with loops of gauge or higgs bosons , which are not significantly heavier than the @xmath3 mass , the @xmath4 , @xmath5 , @xmath6 , parametrization of ref .
@xcite may not be accurate .
this is in contrast to an analysis of top - quark loops in which the @xmath4 , @xmath5 , @xmath6 , parameters are useful due to the large mass of the top quark @xcite .
here we instead use the `` star '' formalism of ref .
@xcite , as in ref .
@xcite which involves both top and bottom loops .
when used to extend the standard model , a general effective field theory can be written in the form @xmath7 where the @xmath8 are dimensionless coefficients , @xmath9 is the energy scale of new physics , and the @xmath10 are effective operators with mass dimension six .
we have not included a dimension - five term in the above expression because there is only one such operator , and it does not involve weak bosons @xcite . in the basis of ref .
@xcite , there are nine dimension - six operators involving only weak bosons and/or higgs doublets that affect precision electroweak measurements at tree or one - loop level .
four of these operators affect weak boson self energies at tree level . of the remaining five , three affect triple gauge couplings and
can be bounded at tree level from weak boson pair production at high - energy colliders .
this paper will focus on the remaining two operators , @xmath11 where @xmath12 , @xmath13 ( @xmath14 are the pauli matrices ) , and @xmath15 is the standard model higgs doublet .
the above two operators affect precision electroweak observables only through oblique corrections .
when the higgs field takes its vacuum expectation value , both operators appear to affect the weak boson self energies at tree level ; however , because the operators have the same form as the standard model gauge kinetic terms , all corrections generated by these operators that involve only gauge bosons can be absorbed into the standard model through field and coupling redefinitions .
thus , the only observable corrections from these operators arise from effective interactions involving gauge bosons and at least one higgs boson or goldstone boson . for this reason
, these operators can not be bounded from processes involving only vector bosons .
explicitly , the new interactions generated by the operators @xmath16 and @xmath17 that contribute at the one - loop level are @xmath18 these interactions induce several contributions to the gauge boson self energies at the one - loop level . the general structure of the relevant diagrams appears in figure [ fig : loops ] , and the explicit self energies can be found in appendix [ apdx : corrections ] . and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] + and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] + and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] + and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] + and @xmath17 .
wavy lines represent gauge bosons , dashed lines represent higgs or goldstone bosons , dotted lines represent ghost fields , and the black dots represent effective operator interactions.,title="fig:",width=163 ] the self energies contain divergences that must be eliminated in order to arrive at meaningful results . because of the gauge - invariant structure of the effective field theory ,
divergences arising from operators of a given dimension can always be absorbed by some other operator of the same dimension . as shown in ref .
@xcite , the operator @xmath19 contributes to gauge boson self energies at tree level and is able to absorb all oblique divergences arising from the operators @xmath16 and @xmath17 .
thus the operator @xmath20 must be included in our analysis .
an analogous situation arises for top - quark loops @xcite .
electroweak boson self energies contribute to precision electroweak data through corrections to the input variables @xmath21 , @xmath22 , and @xmath23 . the correction to @xmath21 depends upon the type of vertex ; these corrections will be labeled @xmath24 , @xmath25 , or @xmath26 , depending on the mediating boson .
the self energy between bosons @xmath27 and @xmath28 is denoted @xmath29 in the expressions below : @xmath30\end{aligned}\ ] ] where @xmath31 .
explicit expressions for the self energies are given in appendix [ apdx : corrections ] .
the correction to any electroweak observable @xmath27 measured at an energy at or above the @xmath32-pole is given by @xmath33 in contrast , low - energy observables are affected by corrections to @xmath23 and by changes to the @xmath34 parameter @xmath35 where @xmath36 and the standard model value of @xmath34 is unity . in our calculations , we used the following values for input parameters : @xmath37 .precision electroweak quantities .
data taken from @xcite.[tab : observables ] [ cols="^,^,^",options="header " , ] we now apply equations ( [ dx1 ] ) and ( [ dx2 ] ) to the electroweak observables listed in table [ tab : observables ] . in order to obtain a bound on the coefficients @xmath38 and @xmath39 , we use the @xmath40 statistic @xmath41 where @xmath42 and @xmath43 is the error matrix , related to the errors for each observable , @xmath44 , and the error correlation matrix , @xmath45,@xcite @xmath46 we calculate the bounds by first setting @xmath47 to the value ( as a function of @xmath48 and @xmath49 ) which minimizes @xmath40 , @xmath50 .
thus we make no assumptions about @xmath47 , but rather allow its value to float .
we then write this new @xmath40 in the following way @xmath51 where @xmath52 is the value of @xmath40 minimized with respect to all coefficients , @xmath53 is the best fit value of the coefficient @xmath8 , and @xmath54 is a symmetric matrix .
we arrive at bounds by solving the equation @xmath55 we can diagonalize @xmath54 to find two statistically independent combinations of our two operators and obtain a bound on those .
we find @xmath56 the contributions of the two coefficients are essentially decoupled due to the fact that the net effect of @xmath16 is significantly larger than that of @xmath17 .
while both coefficients are consistent with zero , the central value of @xmath16 differs from zero by just over one standard deviation . if we instead compute bounds for each coefficient separately , setting the other coefficient to zero in each case , we obtain @xmath57 we again see that the net effect of @xmath16 is significantly larger than that of @xmath17
. we can attempt to gain some understanding of the very different numerical impact of these two operators by considering the oblique parameters @xmath58 of ref .
@xcite .
these parameters are defined in terms of a taylor expansion of the self energies about @xmath59 , and are therefore useful if the values of @xmath60 probed by the experiments are within the radius of convergence of the expansion .
the nearest singularities of the self energies are branch points at @xmath61 and @xmath62 from the first two diagrams in fig . 1 .
since all precision electroweak data are at values of @xmath60 less than these values , the taylor expansion should converge . in a sense , the higgs boson is playing the role of the `` heavy '' particle in the formalism , analogous to the role of the top quark in ref .
@xcite .
analytic expressions for the oblique parameters @xmath58 are given in appendix [ apdx : oblique ] .
we find that the coefficient @xmath39 contributes only to @xmath63 , @xmath64 , and @xmath65 .
because @xmath63 contains a contribution at tree level from @xmath66 , it can not be used to constrain @xmath39 ; we conclude that @xmath64 and @xmath65 must be the oblique parameters that determine the bound on @xmath39 .
unfortunately , @xmath64 and @xmath65 are not parametrically suppressed with respect to the other oblique parameters , so there is no obvious reason why the bound on @xmath39 is so much weaker than the bound on @xmath38 .
this marks the first time bounds on these operators have been obtained from precision electroweak data using the full power of the effective field theory framework .
the bounds obtained are unambiguous and independent of any assumptions about the coefficient of the operator @xmath20 , which was allowed to float .
an analogous result for top - quark loops is given in refs .
@xcite .
let us compare this calculation with the previous calculations of refs . @xcite . as mentioned above
, these papers did not appreciate that unambiguous bounds could be obtained on the operators from a one - loop analysis of precision electroweak data . in ref .
@xcite , only the divergent portion of the one - loop diagrams , and a subset of the finite parts ( those that are proportional to @xmath67 or @xmath68 ) , were calculated .
a bound on the coefficients of @xmath16 and @xmath17 was obtained by assuming that the coefficient of the operator @xmath20 vanishes .
this is an unjustified assumption , because the coefficient of the operator @xmath20 is renormalized by @xmath16 and @xmath17 .
explicitly , in the @xmath69 scheme , @xmath70 where @xmath71 is the renormalized coefficient and @xmath72 is the bare coefficient . there is no reason why the renormalized coefficient should vanish .
making this unjustified assumption allows one to extract much stronger bounds on @xmath48 and @xmath49 , but these stronger bounds are specious . in contrast ,
our bounds are obtained by letting @xmath71 float .
the resulting bounds on @xmath48 and @xmath49 are unambiguous and do not depend on the renormalization scale @xmath73 .
the bounds we obtained on the coefficients of the operators @xmath16 and @xmath17 at one loop from precision electroweak data are so weak that they are easily eclipsed by bounds obtained by tree - level processes involving the higgs boson @xcite .
thus we have shown that , when done properly , the bounds on operators obtained from a one - loop analysis of precision electroweak data can not compete with bounds obtained from tree - level processes , as one would naturally suspect . |
in this paper , we study the existence of positive non - monotone travelling waves for a family of delayed reaction - diffusion equations which includes , as a particular case , the diffusive nicholson s blowflies equation @xmath7 @xmath8 .
this problem was suggested in @xcite , where numerical simulations indicated a loss of monotonicity of the wave profile caused by the delay . eq .
( [ 1s ] ) was introduced in @xcite and it generalizes the famous nicholson s blowflies equation @xmath9 intensively studied for the last decade ( e.g. see our list of references ) . after a linear rescaling of both variables @xmath10 and @xmath11 , we can assume that @xmath12 .
( [ 1s ] ) takes into account spatial distribution of the species , and the mentioned problems reflect the interest in understanding the spatial spread of the growing population @xcite .
relevant biological discussion can be found in @xcite , where various modifications of ( [ 1s ] ) were proposed and studied . here
, however , we will concentrate mainly on the mathematical aspects of the dynamics in ( [ 1s ] ) .
for the sake of simplicity , we will consider the case of a single discrete delay , but extensions for more general functionals ( which additionally can take into account non local space effects ) are possible ( cf .
since the biological interpretation of @xmath10 is the size of an adult population , we will consider _ only _ non - negative solutions for ( [ 1s ] ) and for other population models .
actually , our approach allows us to study a more general family of scalar reaction - diffusion equations @xmath13 related to the mackey - glass type delay differential equations , @xmath14 with exactly two non - negative equilibria @xmath15 ( so that @xmath16 ) .
in particular , with @xmath17 in ( [ 17a ] ) , we obtain the equation proposed in 1977 by mackey and glass , to model hematopoiesis ( blood cell production ) .
the non - linearity @xmath5 is called _ the birth function _ and thus it is non - negative , and generally non - monotone and bounded . due to these properties of @xmath5 and the simple form of dependence on the delay in ( [ 17 ] ) , the cauchy problem @xmath18 , \
x \in { { \bbb r}}^m,\ ] ] for eq . ( [ 17a ] ) has a unique eventually positive global solution for every @xmath19 taken from an appropriately chosen functional space ( e.g. see @xcite ) .
recently , the existence of travelling fronts connecting the trivial and positive steady states in @xmath20 was studied in @xcite and @xcite ( see also @xcite for other methods which eventually can be applied to analyze this problem ) . in @xcite ,
the authors use a monotone iteration procedure coupled with the method of upper and lower solutions .
this approach ( proposed in @xcite ) works well if @xmath21 since in this case the function @xmath5 is increasing on @xmath22\supset [ 0,k]$ ] , thus @xmath23 satisfies the quasimonotonicity condition in @xcite .
this allows one to establish the existence of _ monotone _ wave front solutions @xmath24 for every @xmath25 $ ] and @xmath26 ( cf .
@xcite ) . moreover , as it was proved in @xcite , every solution of @xmath27 with @xmath28 $ ] converges exponentially to some travelling wave provided that @xmath29 is sufficiently close ( in a weighted @xmath30 norm ) to this wave at the very beginning of the propagation . for the case
@xmath31 , clearly @xmath5 is not monotone on @xmath32 $ ] , and wu and zou s method @xcite is no longer applicable . in @xcite ,
the lyapunov - schmidt reduction was used to study systems of delayed reaction - diffusion equations with non - local response .
we observe that eqns .
( [ 1s ] ) and ( [ 17 ] ) fit into the framework developed in @xcite .
this approach requires a detailed analysis of an associated fredholm operator and the existence of heteroclinic solutions of ( [ 17a ] ) ( in @xcite , the latter was established with the use of the monotone semiflows approach developed by h. smith and h. thieme @xcite ) . as a result , it was proved in @xcite that , even when @xmath33 , ( [ 1s ] ) possesses a family of travelling waves if @xmath34 for some @xmath35 ( which is given explicitly ) . the rather restrictive condition
@xmath36 from @xcite was considerably weakened in @xcite by invoking a schauder s fixed - point argument to find heteroclinic solutions of ( [ 17a ] ) .
unfortunately , the main results of @xcite do not answer the question about the existence ( and shape ) of _ positive _ travelling fronts of ( [ 1s ] ) or ( [ 17 ] ) .
we recall here that only non - negative solutions to ( [ 17 ] ) are biologically meaningful . in this paper , inspired by @xcite , for a broad family of nonlinearities @xmath5 ( which includes eq .
( [ 1s ] ) with @xmath37 ) , we prove that eq .
( [ 17 ] ) has a continuous family of positive travelling wave fronts @xmath38 , indexed by the speed parameter @xmath39 , provided that @xmath40 and @xmath41 is sufficiently large : @xmath42 .
furthermore , we show that these fronts generally are not monotone : in fact , they can oscillate infinitely about the positive steady state . on the other hand , for large negative values of @xmath43 , the wave profile
@xmath44 is asymptotically equivalent to an increasing exponential function .
condition ( [ gsc ] ) assures the global attractivity of the positive equilibrium of ( [ 17a ] ) , which is required by our approach . it should be noted that this condition is rather satisfactory in the sense that ( [ gsc ] ) determines a domain of parameters approximating very well the maximal region of local stability for the positive steady state in ( [ 17a ] ) or ( [ 1ss ] ) ( cf .
@xcite ) . before announcing the main results of the present work , we state our basic hypothesis : * ( h )
* : : eq .
( [ 17a ] ) has exactly two steady states @xmath45 and @xmath46 , the second equilibrium being exponentially asymptotically stable and the first one being hyperbolic .
furthermore , @xmath47 and is @xmath48- smooth in some vicinity of the equilibria , with @xmath49 .
the latter implies that the solution @xmath50 of ( [ 17a ] ) is unstable for all @xmath51 . in the sequel
, @xmath52 denotes the minimal positive root of the characteristic equation @xmath53 for sufficiently large @xmath41 , and @xmath54 the unique positive root of the equation @xmath55 , where @xmath56 . as shown later , @xmath57 .
now we are ready to state our main result : [ main ] assume * ( h)*. if the positive equilibrium @xmath58 of eq .
( [ 17a ] ) is globally attracting , then there is @xmath59 such that , for each @xmath60 , equation ( [ 17 ] ) has a continuous family of positive travelling waves @xmath61 .
furthermore , for some @xmath62 , we have @xmath63 as @xmath64 , so that @xmath65 on some semi - axis @xmath66 $ ] . finally , if @xmath67 then the travelling profile @xmath68 oscillates about @xmath58 on every interval @xmath69 .
in order to apply theorem [ main ] , one needs to find sufficient conditions to ensure the global attractivity of the positive equilibrium of ( [ 17a ] ) .
some results in this direction were found in @xcite for a family of nonlinearities having negative schwarz derivative ( or , more generally , satisfying a generalized yorke condition @xcite ) . in particular , ( * ? ? ?
* corollary 2.3 ) implies the following useful version of theorem [ main ] : [ nasa ] assume * ( h ) * and ( [ gsc ] ) .
in addition , suppose that @xmath70 has only one critical point @xmath71 ( maximum ) and that the schwarz derivative @xmath72 is negative for all @xmath73 , @xmath74 .
then all conclusions of theorem 1 hold true .
notice that corollary [ nasa ] applies to both the nicholson s blowflies equation and the mackey - glass equation with non - monotone nonlinearity , see @xcite . to prove our main results , we need a detailed analysis of heteroclinic solutions of ( [ 17a ] ) .
this study is presented in section 2 , and is crucial for the selection of an appropriate functional space where a lyapunov - schmidt reduction is realized .
the existence of positive travelling waves is proven in the third section .
the main result of section 3 is given in theorem [ mr ] , which is essentially theorem [ main ] without its non - monotonicity statement . finally , in the last short section
, we show that these waves have non - monotonic profiles when the delay is over some critical value .
in this section , we study the existence and properties of heteroclinic solutions to the scalar functional equation @xmath75 where @xmath76 , { { \bbb r}}_+ ) \to { { \bbb r}}_+$ ] is a continuous functional which takes closed bounded sets into bounded subsets of @xmath77 . here
@xmath78 , { { \bbb r}}_+)$ ] is the metric space equipped with the norm @xmath79}|\phi(s)|$ ] . throughout this paper
, we suppose that eq .
( [ 1 ] ) has exactly two steady states @xmath80 and @xmath81 , the second equilibrium being asymptotically stable and globally attractive .
thus , if ( [ 1 ] ) has a heteroclinic solution @xmath82 , it must satisfy @xmath83 .
we start by proving a general existence result which is valid for the abstract setting of dynamical systems .
let @xmath84 be a continuous semidynamical system defined in a complete metric space @xmath85 .
first , we mention the following fact ( see e.g. @xcite ) : suppose that @xmath86 is a complete orbit of @xmath87 .
if the closure @xmath88 is compact for some @xmath89 , then the @xmath90limit set @xmath91 of @xmath92 is nonempty , compact and invariant ( this means that for every @xmath93 there exists at least one full trajectory @xmath94 with @xmath95 ) .
for every @xmath96 and @xmath97 , let @xmath98 denote the set of right endpoints of all orbit segments @xmath99}z = \{s^uz : u\in [ 0,h]\}$ ] which are completely contained in @xmath100 : @xmath101}z \subset a , \
{ \rm for \ some \ } z \in a\}.\ ] ] next statement shows clearly how to relate the global attractivity property of the positive equilibrium of ( [ 17a ] ) to the problem concerning the existence of travelling fronts for ( [ 17 ] ) : [ 123 ] assume that @xmath102 is either empty or pre - compact , for all bounded sets @xmath100 and some @xmath97 .
suppose that there exist two disjoint compact invariant subsets @xmath103 of @xmath104 such that @xmath105 as @xmath106 for every @xmath107 .
if the set @xmath108 is not empty for every sufficiently small @xmath109 , then there exists at least one complete orbit @xmath94 with @xmath110 and @xmath111 .
let @xmath112 and , for every @xmath113 , take some @xmath114 .
due to the compactness of @xmath115 , we can assume that @xmath116 for some @xmath117 . in consequence , if @xmath118 is the minimal real number such that @xmath119 , then @xmath120 .
set @xmath121 . due to the compactness condition imposed on @xmath87
, we can suppose that @xmath122 .
let now @xmath123 .
we have @xmath124 for every @xmath125 .
since , for every integer @xmath126 the sequence @xmath127 has a convergent subsequence ( say , @xmath128 ) , we can assume that @xmath129 converges uniformly on @xmath130 $ ] to @xmath131 .
moreover , we have that @xmath132 and @xmath133 for all @xmath134 . in this way , taking @xmath135 , we can use @xmath136 to construct a continuous function @xmath137 , such that @xmath138 for every @xmath139 .
such @xmath94 defines the complete orbit we are looking for . since @xmath140 is a subset of the bounded set @xmath141
, we conclude that @xmath140 is pre - compact .
furthermore , because of @xmath142 , we have @xmath143 .
this means that @xmath144 so that @xmath145 .
@xmath146 a direct application of lemma [ 123 ] to eq . (
[ 1 ] ) gives the following let @xmath76 , { { \bbb r}}_+ ) \to { { \bbb r}}_+$ ] be a continuous functional which takes closed bounded sets into bounded subsets of @xmath77
. assume further that every non - negative solution of ( [ 1 ] ) admits a unique extension on the right semi - axis .
if @xmath147 @xmath148 and @xmath149 attracts every solution of ( [ 1 ] ) with non - negative and nontrivial initial function , then there exists a positive complete solution @xmath82 to ( [ 1 ] ) such that @xmath150 .
with some additional conditions on @xmath151 , we can say more about such an orbit @xmath94 : [ 4 ] assume @xmath152 for some @xmath153 ( so that @xmath154 ) .
assume that @xmath155 , and let @xmath156 be such that @xmath157 .
let @xmath158 be the unique positive real root of the equation @xmath159 ( @xmath160 ) , so we have @xmath161 . then for every heteroclinic solution @xmath82 of the equation @xmath162 there exist @xmath163 such that @xmath164 choose @xmath165 sufficiently small such that @xmath166 for all @xmath167 , and let @xmath168 be such that @xmath169 for all @xmath170 .
we claim that , for every @xmath171 @xmath172 } \psi(s+u ) \geq \frac{\exp(-h)}{p_2}\max_{u \in [ -h,0 ] } \psi(s+u ) = \frac{\exp(-h)}{p_2}\psi(t_m).\ ] ] indeed , if @xmath173 , then , by the variation of constants formula , @xmath174 finally , suppose that @xmath175 so that @xmath176
. then ( [ es ] ) holds since @xmath177 next , for every @xmath171 and @xmath178 $ ] , we have that @xmath179 from the inequalities above and since additionally @xmath166 for all @xmath167 , then for @xmath180 and @xmath181 $ ] we have @xmath182 \leq \psi ( t_m ) e^{\lambda_2 ( u+2h ) } , \nonumber\end{aligned}\ ] ] and @xmath183 \geq \psi ( t_m ) e^{\lambda_1 ( u+h ) } .\nonumber\end{aligned}\ ] ] by repeating the above procedure over intervals of length @xmath4 , the step by step method implies that , for all @xmath184 , @xmath185 in particular , @xmath186 thus , for every @xmath171 , @xmath187 in what follows , we shall assume that @xmath47 , @xmath188 , and use several times the following simple assertion . [ char3 ] suppose that @xmath189 and @xmath190 .
then the characteristic equation @xmath191 has only one real root @xmath192 .
moreover , all roots @xmath193 of ( [ char ] ) are simple and we can enumerate them in such a way that @xmath194 the last inequality follows from @xmath195 . @xmath196
[ psi ] suppose that @xmath188 and that @xmath94 is a heteroclinic solution to ( [ 2 ] ) .
let @xmath54 be the positive root of ( [ char ] ) .
if there exists @xmath197 , then , for each @xmath198 and some @xmath199 , we have that @xmath200 at @xmath201 , so that @xmath202 on some semi - axis @xmath203 $ ] .
moreover , if there exists @xmath204 and is bounded for some @xmath205 , then @xmath82 is unique up to a shift in @xmath11 .
since @xmath206 is finite , @xmath207 as @xmath208 . from lemma 6 ,
given @xmath165 small , for @xmath209 we have @xmath210 at @xmath211 , where @xmath212 is the unique positive root of @xmath213 .
it is easy to see that @xmath214 .
in fact , let @xmath215 . for @xmath216 , we have @xmath217 if and only if @xmath218 .
on the other hand , @xmath219+\delta/2[\exp(-(\lambda-\delta/2)h)-1]<0.$ ] hence , @xmath220 where @xmath221 as @xmath222 .
now , consider the linear inhomogeneous delay differential equation @xmath223 the change of variables @xmath224 transforms it into @xmath225 the spectra @xmath226 of the linear parts of eqns .
( [ llps ] ) and ( [ lps ] ) are related by @xmath227 , therefore the linear part of ( [ llps ] ) has not pure imaginary eigenvalues for all sufficiently small @xmath198 ( equivalently , the linearization of eq . ( [ llps ] ) about zero is hyperbolic ) . in this case , ( [ llps ] ) has a bounded solution @xmath228 at @xmath229 ( e.g. see ( * ? ? ? * lemma 3.2 , p. 246 ) or
* section 10.1 ) .
note that eqns .
( [ lps ] ) and ( [ llps ] ) are not autonomous .
nevertheless , the results for autonomous equations near hyperbolic equilibria in @xcite are valid in this setting , since the linearized equation near zero for eq .
( [ llps ] ) has an exponential dichotomy , cf .
thus eq . ( [ lps ] ) has a solution @xmath230 in consequence , @xmath231 solves the linear homogenous equation @xmath232 and is bounded at @xmath201 .
this is possible if and only if @xmath233 where @xmath234 is a finite set of roots having non - negative real parts of the characteristic equation ( [ char ] ) .
notice that @xmath235 and @xmath236 ( see lemma [ char3 ] ) .
in this way @xmath237 on the other hand , from lemma [ 4 ] we know that @xmath238 . since @xmath236 , this implies immediately that all @xmath239 and that @xmath240 by ( [ ps ] ) , @xmath241 observe also that @xmath242 defines another heteroclinic solution of ( [ 1 ] ) .
finally , suppose that @xmath243 are two heteroclinic solutions to ( [ 1 ] ) such that @xmath244 applying the lagrange mean value theorem twice , we get @xmath245 for @xmath246 close to @xmath247 . since @xmath248 we obtain that @xmath249 therefore @xmath250 satisfies @xmath251 from which , applying the same procedure as above , we deduce that @xmath252 . in this way
, we can show that @xmath253 for every integer @xmath254 .
this leads us to the conclusion that @xmath255 has superexponential decay at @xmath229 ( equivalently , @xmath255 is a _ small solution _ at @xmath256 , see @xcite ) .
we will finalize our proof showing that only the trivial solution of the linear asymptotically autonomous homogeneous equation @xmath257 can have superexponential decay at @xmath258 ( notice that @xmath250 satisfies ( [ se ] ) with @xmath259 ) . indeed ,
if @xmath260 on some semi - axis @xmath66 $ ] , then we can repeat the arguments in the proof of lemma [ 4 ] to find an exponential lower bound for @xmath261 , in contradiction to our assumption of superexponential decay of @xmath261 .
consider now the case of @xmath261 oscillatory on every semi - axis @xmath66 $ ] , and take @xmath262 such that @xmath263 for all @xmath264 $ ] .
let @xmath265 be a point of the global maximum of @xmath266 : we can assume that @xmath267 .
then @xmath268 , so that @xmath269 .
analogously , @xmath270 .
thus @xmath271 so that @xmath261 can not decay superexponentially as @xmath272 .
@xmath146 now , assume * ( h ) * and the global attractivity of @xmath273 for eq .
( [ 2 ] ) , and then take @xmath274 satisfying ( [ char ] ) and the unique ( up to a shift in time ) heteroclinic solution @xmath94 described in lemma [ psi ] .
let @xmath275 be sufficiently close to @xmath54 and such that the equation @xmath276 is hyperbolic .
note that this latter equation is obtained by effecting the change of variables @xmath277 to the linear equation @xmath278 . for a fixed @xmath279
, we will consider the seminorms @xmath280 , @xmath281 , @xmath282 and the following banach spaces : @xmath283 equipped with the norms @xmath284 and @xmath285 respectively ( in order to simplify the notation , we shall often write @xmath286 instead of @xmath284 , etc ) . notice that , due to lemma [ psi ] , we have @xmath287 .
we shall also need the following integral operator @xmath288 where @xmath289 with @xmath290 .
observe that @xmath291 is well defined , since @xmath292 and , for @xmath293 , @xmath294 [ 6 ] if * ( h ) * is assumed , then @xmath295 is an isomorphism of banach spaces .
we first prove that @xmath296 .
indeed , if @xmath297 and @xmath298 , then @xmath299 therefore @xmath300 is a bounded solution of the linear delay differential equation @xmath301 since @xmath302 at @xmath303 and @xmath304 at @xmath305 , we conclude that @xmath306 thus @xmath307 can be viewed as a bounded solution of the inhomogeneous equation @xmath308 since @xmath309 at @xmath211 , with @xmath310 close to @xmath54 , the procedure which has been used before to prove the uniqueness of the heteroclinic @xmath82 allows us to conclude that @xmath311 and that @xmath312 . on the other hand , we know that @xmath313 satisfies ( [ veq ] ) .
thus we must have @xmath314 , @xmath315 constant , a contradiction
. therefore @xmath316 and @xmath317 .
we now establish that @xmath318 is an epimorphism .
take some @xmath319 and consider the following integral equation @xmath320 if we set @xmath321 , this equation is transformed into @xmath322 hence we have to prove the existence of at least one @xmath323-solution of the equation @xmath324 first , notice that all solutions of ( [ bso ] ) are bounded on the positive semi - axis @xmath77 due to the boundedness of @xmath325 and the exponential stability of the homogeneous @xmath326-limit equation @xmath327 here we use the persistence of exponential stability under small bounded perturbations ( e.g. see ( * ? ? ?
* section 5.2 ) or ( * ? ? ?
* chapter vi ( 9c ) ) ) and the fact that @xmath328 .
furthermore , since every solution @xmath329 of ( [ bso ] ) satisfies @xmath330 with @xmath331 , we get @xmath332 .
next , by effecting the change of variables @xmath333 to eq .
( [ bso ] ) , we get a linear inhomogeneous equation of the form @xmath334y(t - h ) + \epsilon_2(t),\ ] ] where @xmath335 and @xmath336 at @xmath256 . since the @xmath337-limit equation @xmath338 to the homogeneous part of ( [ bso0 ] ) is hyperbolic , due to the above mentioned persistence of the property of exponential dichotomy , we again conclude that eq .
( [ bso0 ] ) also has an exponential dichotomy on @xmath339 .
thus ( [ bso0 ] ) has a solution @xmath340 which is bounded on @xmath341 so that @xmath342 is a @xmath323-solution of eq .
( [ bso ] ) .
now , it is evident that @xmath343 solves ( [ bso ] ) for each @xmath344 . in consequence ,
@xmath345 , if we take @xmath346 . @xmath347 [ re1 ] for @xmath165 small , consider @xmath348 , where @xmath349 is defined by @xmath350 ( recall here the discussion after formula ( [ llps ] ) ) . replacing @xmath291 by @xmath349 in the proof of lemma [ 6 ] ,
we establish similarly that @xmath351 is an isomorphism of the banach space @xmath352 onto itself .
since the linear equation @xmath353 is hyperbolic , this situation is actually simpler than the one considered in lemma [ 6 ] .
in this section , we are looking for travelling waves for ( [ 17 ] ) , that is , solutions @xmath354 , where @xmath355 is the wave speed , connecting the two equilibria of ( [ 17 ] ) .
we will suppose that @xmath356 is sufficiently small .
this leads us to the question about the existence of heteroclinic solutions to the singularly perturbed equation @xmath357 with @xmath358 .
being a bounded function , each travelling wave should satisfy the following integral equation @xmath359 where @xmath360 for solutions in @xmath323 with @xmath361 close to @xmath54 , this equation can be written in the shorter form @xmath362 where @xmath363 are defined by @xmath364 and @xmath365 is the nemitski operator .
( for the sake of simplicity , we write @xmath366 instead of @xmath367 ) . throughout all this section
, we will suppose that the @xmath368-smooth function @xmath5 is defined and bounded on the whole real axis @xmath369 . clearly , this assumption does not restrict the generality of our framework , since it suffices to take any smooth and bounded extension on @xmath339 of the nonlinearity @xmath5 described in * ( h)*. notice that , since there exists finite @xmath370 , we have @xmath371 for a bounded @xmath372 .
set @xmath373 .
as it can be easily checked , @xmath374 so that actually @xmath375 is well - defined . furthermore , we have the following lemma : [ 7 ] assume that @xmath376 . then @xmath375 is frchet continuously differentiable on @xmath323 with differential @xmath377 by the taylor formula , @xmath378 @xmath379 fix some @xmath380 , then we have @xmath381 clearly , it holds that @xmath382 .
since functions in @xmath323 are bounded and @xmath383 is uniformly continuous on bounded sets of @xmath369 , for any given @xmath165 there is @xmath384 such that for @xmath385 we have @xmath386 now , we consider the integral operators @xmath387 defined as @xmath388 [ 8 ] set @xmath389 and @xmath390 . if @xmath391 , then @xmath392 in the operator norm .
moreover , both operator families @xmath393 are continuous in the operator norm .
we prove only that @xmath394 as @xmath395 , the proof of the continuous dependence of @xmath396 on @xmath356 being completely analogous .
we first establish that @xmath397 uniformly as @xmath395 .
in fact , for all @xmath398 , we obtain @xmath399 furthermore , since @xmath400 for all @xmath401 , for @xmath402 we have @xmath403 hence , for @xmath404 , we obtain that @xmath405 .
next , we prove that @xmath406 uniformly as @xmath395 .
we have @xmath407 thus , for @xmath408 , we obtain that @xmath409 and , for all @xmath11 , @xmath410 thus @xmath411 , and the proof of the lemma is complete .
@xmath196 to prove the main result of this section , stated below as theorem [ mr ] , we will make use of the following proposition : [ lem ] let @xmath412 , where @xmath413 , denote the ( countable ) set of roots to the equation @xmath414 if @xmath415 then ( [ char2 ] ) has exactly two real roots @xmath416 such that @xmath417 moreover : ( i ) there exists an interval @xmath418 such that , for every @xmath419 , all roots @xmath420 of ( [ char2 ] ) are simple and the functions @xmath421 are continuous ; ( ii ) we can enumerate @xmath422 , in such a way that there exists @xmath423 for each @xmath424 , where @xmath425 are the roots of ( [ char ] ) , with @xmath426 ; ( iii ) for all sufficiently small @xmath356 , every vertical strip @xmath427 contains only a finite set of @xmath428 roots ( if @xmath429 , then @xmath428 does not depend on @xmath356 ) @xmath430 to ( [ char2 ] ) , while the half - plane @xmath431 contains only the root @xmath432 . the existence of real roots @xmath433 satisfying @xmath434 is obvious when @xmath435 . on the other hand , if @xmath436 is a real root of ( [ char2 ] ) , then @xmath437 hence @xmath438 , from which it can be checked easily that @xmath439 we also notice that every multiple root @xmath262 has to satisfy the system @xmath440 which implies @xmath441 the first equation of ( [ ep1 ] ) implies that @xmath262 is real while the second equation of ( [ ep ] ) says that @xmath442 . since @xmath262 is positive , from the first equation of ( [ ep1 ] ) we obtain @xmath443 ( we recall that @xmath444 ) .
let @xmath445 be the maximal positive root of the second equation of ( [ ep1 ] ) . if @xmath446 is so small that @xmath447 , system ( [ ep ] ) can not have any positive solution . in consequence ,
the second assertion of this lemma holds if we set @xmath448 .
finally , we prove that the half - plane @xmath449 contains only the root @xmath450 of ( [ char2 ] ) . for this ,
let us evaluate @xmath451 on the boundary of some rectangle @xmath452\times[-c , c ] \subset { { \bbb c}}$ ] , with @xmath453 being sufficiently large .
for @xmath454 the ( real ) roots of @xmath455 , we have that @xmath456 , we obtain @xmath457 if @xmath458 , then @xmath459 similarly , for @xmath460 fixed , we get @xmath461 thus , by rouch s theorem , @xmath462 and @xmath455 have the same number of roots in the half - plane @xmath431 , that is exactly one root . therefore , for all @xmath463 with @xmath464 $ ] and @xmath465 , we get @xmath466 so that @xmath467 . hence , applying rouch s theorem to the functions @xmath468 and @xmath469 along an appropriate rectangle inside @xmath470\times [ -3pe^{-\xi h } , 3pe^{-\xi h } ] \subset { { \bbb c}}$ ] , we prove the last assertion of lemma [ lem ] .
@xmath146 [ mr ] assume * ( h ) * and that the positive equilibrium of eq .
( [ 17a ] ) is globally attractive .
let @xmath471 be some heteroclinic orbit of eq .
( [ 17a ] ) : @xmath472 @xmath473 .
then , for every @xmath165 there is an interval @xmath474 and a continuous family of positive heteroclinic orbits @xmath475 of eq .
( [ twe ] ) satisfying the additional conditions @xmath476 and @xmath477 .
furthermore , for every @xmath478 we have that @xmath479 at @xmath272 for some @xmath480 , and that @xmath481 on some semi - axis @xmath66 $ ] .
we represent the mentioned orbit @xmath94 of ( [ 17a ] ) as @xmath482 , where @xmath483 for @xmath165 small , consider @xmath484 . in virtue of lemmas [ 6 ] , [ 7 ] and [ 8 ] , we can apply the implicit function theorem ( e.g. , see @xcite or @xcite ) to eq .
( [ be ] ) written as @xmath485 , where @xmath486 , @xmath487 observe that @xmath488 and @xmath489 . in this way
, we establish the existence of an interval @xmath490 and a continuous family @xmath491 of solutions to @xmath492 . notice that @xmath493 , @xmath494 satisfy eq .
( [ be ] ) , so that , as it can be checked directly , @xmath495 .
thus @xmath496 and @xmath497 satisfies all conclusions of the third sentence of the theorem , except its positivity , which is proved below .
assume now that @xmath498 is sufficiently small so that @xmath499 for all @xmath500 .
since @xmath501 as @xmath502 , and since there exists a constant @xmath503 such that @xmath504 , we get @xmath505 where @xmath506 . moreover , @xmath507 for some @xmath508 which does not depend on @xmath356
now , @xmath509-solutions @xmath510 to @xmath511 are solutions to the equation @xmath512 . due to remark [ re1 ] and lemma [ 8 ]
, for @xmath513 close to @xmath54 the operator @xmath514 is invertible in @xmath509 for all sufficiently small @xmath356 .
moreover , lemma [ 8 ] implies that there exists a subinterval @xmath515 such that @xmath516 for all @xmath517 .
hence , we obtain @xmath518 for all @xmath517 . therefore eq . ( [ lpse ] )
has a bounded solution @xmath510 such that @xmath519 .
consequently , @xmath520 solves the linear homogenous equation @xmath521 and is bounded as @xmath201 .
this is possible if and only if @xmath522 where @xmath523 and @xmath524 are the roots with non - negative real parts of the characteristic equation ( [ char2 ] ) , @xmath525 , @xmath526 . in consequence ,
@xmath527 it follows from lemma [ lem ] that @xmath528 , provided @xmath356 is small ( say , @xmath529 ) and @xmath530 is sufficiently close to @xmath54 . since @xmath531 , this implies immediately that @xmath532 and @xmath533 to prove the positivity of @xmath497 for @xmath356 small , we first establish that @xmath534 is finite , from which we deduce that the constants @xmath535 are positive ; in fact , we will find that @xmath536 .
let us suppose already that @xmath537 and @xmath538 for all @xmath539 .
since @xmath540 , for all @xmath541 , we get @xmath542 . in particular , taking @xmath543 and @xmath544 , we obtain @xmath545 hence @xmath546 for @xmath547 . noting that @xmath548 as @xmath395
, we deduce that there is @xmath549 such that @xmath550 for @xmath551 , so that @xmath552 for @xmath553 set @xmath554 . by lemma [ psi ] ,
we have @xmath555 with @xmath556 at @xmath557 . since @xmath558
, we now conclude that there is @xmath559 such that for @xmath560 and @xmath561 we have @xmath562 , there exists @xmath563 $ ] such that , for @xmath564 , we have @xmath565 taking @xmath566 we obtain for @xmath567 @xmath568 hence @xmath569 .
therefore , for all @xmath570 , @xmath571 thus , for @xmath572 and @xmath560 we get @xmath573>0.$ ] since @xmath574 uniformly on @xmath369 and @xmath94 is bounded from below by a positive constant on @xmath575 , we conclude that @xmath497 is positive on @xmath369 , for all @xmath576 small . finally , for every fixed @xmath577 , we have that @xmath578 @xmath579 where @xmath580 and @xmath581 is bounded on @xmath582 $ ] . hence , @xmath583 at @xmath211 and @xmath584 where @xmath585 is bounded : @xmath586 . differentiating ( [ psea ] )
, we obtain @xmath587 hence , @xmath588 so that @xmath589 in ( [ pse ] ) .
therefore , in view of ( * ? ? ?
* proposition 7.1 ) and the inequality @xmath590 , we get from ( [ pse ] ) that @xmath591 and @xmath592 . @xmath593
as it was noticed in @xcite , various investigators have studied numerically the case of a large delay in the nicholson s blowflies equation , and noted a loss of monotonicity of the wave front as the delay increases , _ with the front developing a prominent hump " _ whose height _ is bounded above by a bound that does not depend on the delay " _ , see @xcite from which the above citation was taken .
it is not difficult to explain the second phenomenon , since at every point of local maximum @xmath255 of @xmath594 we have @xmath595 so that @xmath596 . here
we explain also the first phenomenon , easily getting the oscillation of the travelling waves about @xmath58 as @xmath597 stated in theorem [ main ] from the next two lemmas . [
15 ] let @xmath598 and @xmath599 .
then the equation @xmath600 has no negative real roots , for all sufficiently small @xmath356 .
moreover , if the equilibrium @xmath58 of ( 4 ) is hyperbolic , then , for all small @xmath356 , there are no roots of ( 28 ) on the imaginary axis
. set @xmath601 .
we first prove that the lemma is valid for @xmath602 ( see also @xcite ) .
let @xmath262 be the maximum point of @xmath603 on @xmath369 , i.e. , @xmath604 is such that @xmath605 .
note that @xmath606 and @xmath607 for @xmath608 . if there is a negative zero of @xmath603 , then @xmath609 and @xmath610 , implying that @xmath611 , which contradicts the hypothesis @xmath599 .
if @xmath612 , then @xmath613 for all @xmath614 and @xmath615 , hence @xmath616 for @xmath617 .
now , let @xmath618 , so that @xmath609 . for @xmath619 small , by the implicit function theorem we conclude that there is a negative root @xmath620 of the equation @xmath621 with @xmath622 ; moreover , @xmath620 is the absolute maximum point of @xmath623 on @xmath624 $ ] .
since @xmath625 depends continuously on @xmath356 and @xmath626 , for @xmath627 small we have @xmath628 for all @xmath629 .
we now prove that ( [ char5 ] ) has no roots on the imaginary axis .
first , notice that @xmath630 if @xmath631 .
for @xmath602 , eq .
( [ char5 ] ) does not have roots on the imaginary axis , therefore @xmath632 .
hence , @xmath633 for all @xmath634 small and @xmath635 , which implies the hyperbolicity of eq .
( [ char5 ] ) .
@xmath196 the next lemma can be considered as an extension of the linearized oscillation theorem from @xcite to the second order delay differential equation @xmath636 [ lll ] assume * ( h ) * and that @xmath637 . for small @xmath638 set @xmath639 .
then every non constant and bounded solution @xmath640 of @xmath641 such that @xmath642 oscillates about @xmath58 .
consider some non - constant solution @xmath640 of ( [ twee ] ) such that @xmath643 .
if we suppose for a moment that , for some @xmath644 , it holds @xmath645 identically for all @xmath646 , then we obtain easily that @xmath645 for all @xmath647 $ ] .
hence @xmath648 should be a constant solution , in contradiction with our initial assumption .
therefore either @xmath649 oscillates about zero or is eventually non - constant and non - negative , or non - positive . in order to get a contradiction ,
assume that @xmath255 is not oscillatory .
notice that @xmath255 satisfies the following linear asymptotically autonomous delay differential equation @xmath650 where @xmath651 and @xmath652 for some @xmath653 given by the mean value theorem .
since @xmath261 is bounded on @xmath369 and @xmath654 as @xmath106 , we can use the integral representations ( [ psea ] ) and ( [ 2f ] ) to prove that @xmath655 . from lemma
[ 15 ] , it follows that the equation @xmath656 is hyperbolic , hence the equilibrium @xmath657 of the system @xmath658 is hyperbolic for all sufficiently small @xmath356 .
thus the trajectory of @xmath261 belongs to the stable manifold of the hyperbolic equilibrium @xmath58 of ( [ twee ] ) , so that we can find @xmath659 such that @xmath660 at @xmath661 .
therefore we have @xmath662 at @xmath663 . from (
* proposition 7.2 ) ( see also ( * ? ? ?
* proposition 2.2 ) ) , we conclude that : @xmath664 either there are @xmath665 and @xmath666 a nontrivial eigensolution of the limiting equation @xmath667 on the generalized eigenspace associated with the ( nonempty ) set @xmath668 of eigenvalues with @xmath669 , such that @xmath670 + @xmath671 or @xmath250 decays superexponentially at @xmath663 .
however , this latter condition is not possible : as it was established in ( * ? ? ?
* lemma 3.1.1 ) , if @xmath672 then every eventually nontrivial and nonnegative solution of ( [ tweee ] ) does not decay superexponentially ( see also ( * ? ? ? * lemma a.1 ) for the case @xmath673 ) . on the other hand , from lemma 15
we know that there are no real negative eigenvalues of ( [ 30 ] ) : hence @xmath674 for all @xmath675 . from
* lemma 2.3 ) , we conclude that @xmath676 is oscillatory .
@xmath196 finally , we observe that due to the exponential stability of the positive steady state , which implies fast convergence , numerical heteroclinic solutions @xmath594 exhibit only one or two well pronounced humps , see ( * ? ? ?
this research was supported by fondecyt ( chile ) , projects 7040044 ( teresa faria ) and 1030992 ( sergei trofimchuk ) , by fct ( portugal ) , program pocti/ feder , under cmaf ( teresa faria ) , and by conicyt ( chile ) through pbct program act-05 ( sergei trofimchuk ) .
this work was partially written while t. faria was visiting the university of talca , and she thanks the university for its kind hospitality . | we prove the existence of a continuous family of positive and generally non - monotone travelling fronts for delayed reaction - diffusion equations @xmath0 , when @xmath1 has exactly two fixed points : @xmath2 and @xmath3 .
recently , non - monotonic waves were observed in numerical simulations by various authors . here , for a wide range of parameters , we explain why such waves appear naturally as the delay @xmath4 increases . for the case of @xmath5 with negative schwarzian ,
our conditions are rather optimal ; we observe that the well known mackey - glass type equations with diffusion fall within this subclass of @xmath6 . as an example , we consider the diffusive nicholson s blowflies equation .
time - delayed reaction - diffusion equation , heteroclinic solutions , non - monotone positive travelling fronts , single species population models .
+ _ 2000 mathematics subject classification _ : 35k57 , 35r10 , 92d25 |
since the discovery of high-@xmath6 copper - oxide superconductors , extensive studies on strongly electron correlated system have been in progress on the basis of physical properties of two - dimensional ( 2d ) cuo@xmath0 planes .
moreover , from the viewpoint of low - dimensional physics , particular attention is paid to the physical role of one - dimensional ( 1d ) cuo chains included in some families of high-@xmath6 copper oxides such as y - based superconductors with the transition temperature @xmath6=@xmath9 92k .
it is well known that the pr - substitution for y - sites in yba@xmath0cu@xmath10o@xmath11 ( y123 ) and yba@xmath0cu@xmath1o@xmath12 ( y124 ) compounds dramatically suppresses @xmath6 and superconductivity in cuo@xmath0 planes disappears beyond the critical value of pr , @xmath13=0.5 and 0.8 , respectively .
@xcite such a suppression effect due to pr - substitution on superconductivity has been explained in terms of the hybridization model with respect to pr-4@xmath14 and o-2@xmath15 orbitals.@xcite y124 compound with double chains is thermally stable up to 800 @xmath16c , while in y123/7-@xmath17 oxygen deficiencies are easily introduced at lower annealing temperatures .@xcite intermediate between prba@xmath0cu@xmath10o@xmath11 ( pr123 ) with single chains and prba@xmath0cu@xmath1o@xmath12 ( pr124 ) with double chains is the pr@xmath0ba@xmath1cu@xmath2o@xmath18 ( pr247 ) compound with an alternative repetition of the cuo single chain and double chain blocks along the @xmath19-axis , isostructural with superconducting y@xmath0ba@xmath1cu@xmath2o@xmath18 ( y247 ) .
@xcite in contrast to y123 , y247 retains its orthorhombic structure and its superconducting properties , even after oxygen along the single chains is completely depleted.@xcite this finding is considered to be due to the presence of the double cuo chain block in y247 . for pr247 , it is possible to examine physical properties of metallic double chains in terms of varying oxygen contents along single chains .
recently , matsukawa et al . , discovered that oxygen removed polycrystalline pr@xmath0ba@xmath1cu@xmath2o@xmath3 with cuo metallic double chain shows superconductivity around @xmath20 k. @xcite matsushita et al .
, revealed that hall coefficient of oxygen defect pr247 is varied from positive sign to negative one upon decreasing temperatures , indicating an electron doped superconductor.@xcite the nuclear quadrapole resonance experiment indicated that the superconductivity of reduced pr247 is realized at the cuo double chains.@xcite for the appearance of pr247 superconductor , it is essential to remove oxygen from the as sintered sample of pr247.@xcite through several works on synthesis of higher @xmath21 samples in nominal pr247 composition , it has been made clear that the heterogeneous structure containing both pr123 and pr124 phases in pr247 system plays a crucial role on attaining higher @xmath21 .
@xcite high - resolution transmission electron microscopy ( tem ) analyses of the heterogeneous pr247 sample exhibits that there exists an irregular long - period stacking structure along the @xmath19 axis such as \{-s - s - s - s - d- } sequence , where s and d denote single cuo chain and double cuo chain blocks , respectively.@xcite in this paper , we have reported the magnetic , electronic and thermal transport properties of the oxygen removed pr@xmath0ba@xmath1cu@xmath2o@xmath3 . in particular , the @xmath5 sample exhibits higher superconductivity attaining @xmath4 k.
the inset of ( a ) shows the low angle diffraction data enlarged near @xmath22 . here ,
three arrows point to the peaks corresponding to the miller indexes ( 002 ) , ( 004 ) , and ( 001 ) of pr124 , pr247 , and pr123 , respectively upon increasing @xmath23 .
, title="fig:",width=302 ] ba@xmath1cu@xmath2o@xmath3 .
the inset of ( a ) shows the low angle diffraction data enlarged near @xmath22 . here ,
three arrows point to the peaks corresponding to the miller indexes ( 002 ) , ( 004 ) , and ( 001 ) of pr124 , pr247 , and pr123 , respectively upon increasing @xmath23 .
, title="fig:",width=302 ] polycrystalline samples of the nominal composition pr@xmath0ba@xmath1cu@xmath2o@xmath3(pr247 ) were synthesized using sol - gel technique .
high purity powders of praseodymium oxide ( pr@xmath24o@xmath25 ) , barium carbonate ( baco@xmath10 ) , and copper carbonate ( cuco@xmath10 ) were mixed to the stoichiometric composition , they were then added to a suitable amount of citric acid solution and finally their mixture was magnetically stirred in a container .
the solution was condensed on a hot plate at 80 @xmath26c for 30 min and the resultant citrate gel was pyrolyzed at 450 @xmath26c for 2 h , to remove the organic compounds .
the obtained precursor was ground into a powder by a glass rod .
the precursor was calcined in air at 800 @xmath26c for 24 h. the resultant powder was pressed into a pellet and then sintered at 875@xmath27876 @xmath26c for a long time over 180 h under ambient oxygen pressure ( as - sintered sample ) . for the sintered process ,
the electric furnace equipped with three temperature controllers was used to reduce thermal gradient along the axis of furnace tube .
the resultant sample was post annealed in vacuum at 500 @xmath26c for 48@xmath2872 h ( reduced sample ) .
x - ray powder diffraction measurement was carried out with cu@xmath29 radiation at diffraction angle range of @xmath30 .
the electron probe micro analyzer ( epma ) analysis on the superconductive sample annealed in vacuum revealed that the atomic ratio of pr : ba : cu = 1.99:4.18:6.72 , through an average of composition analyses at several points within its sample .
the oxygen deficiency was estimated to be @xmath31 and 0.94 from gravimetric analysis for 48 h and 72 h reduction treatments , respectively .
the electric resistivity was measured with dc four terminal method .
seebeck coefficient @xmath32 was determined from both measurements of a thermoelectric voltage and temperature difference along the longitudinal direction of the measured sample .
the seebeck coefficient of copper lead was subtracted from the measured values using the published data .
the thermal conductivity data were collected with a conventional heat flow method .
the dc magnetization was performed using the commercial superconducting quantum interference device ( squid ) magnetometers both at iwate university and national institute for materials science . in particular , to remove the remanent magnetic field below 1 moe , the fluxgate device was utilized in the zero - field cooled ( zfc ) and field cooled ( fc ) magnetization measurements under @xmath33 1 oe .
hydrostatic pressure in magnetization was applied by using a clamp - type cube cell up to 1.2 gpa .
fluorinert was used as a pressure transmitting medium .
the magnitude of pressure was calibrated by the pressure dependence of the critical temperature of lead . .
, width=377 ] .
( b ) the pressure dependence of @xmath8 of the pr247 sample with the oxygen defect @xmath34 as a function of the applied pressure up to 1.2 gpa . for comparison ,
the data of pure pr247 with @xmath35 and 0.62 prepared by a high oxygen pressure technique are cited.@xcite , width=377 ]
first of all , let us show in fig.[xray ] x - ray powder diffraction pattern of the superconductive sample with the nominal pr@xmath0ba@xmath1cu@xmath2o@xmath3 . the low angle diffraction data enlarged near @xmath22 are presented in the inset of fig.[xray](a ) , where three arrows point to the peaks corresponding to the miller indexes of pr124 , pr247 , and pr123 upon increasing @xmath23 .
these findings indicate that the nominal phase of pr247 obtained coexists with both pr124 and pr123 phases . under ambient pressure condition
, it seems to be difficult to synthesize high quality polycrystalline sample with nominal composition .
however , to our knowledge , the heterogeneous samples have shown good superconductive properties in comparison to pure pr247 with lower @xmath20 k.@xcite the scanning electron microscope ( sem ) micrographs of the superconducting sample with @xmath34 displayed in fig.[xray](b ) reveal that its typical grain size is estimated to be several @xmath36 m .
figure [ mt ] ( a ) shows the temperature variation of zero field cooled ( zfc ) and field cooled ( fc ) magnetization of the oxygen removed pr247 sample under an applied field of 1 oe . in the inset of fig .
[ mt ] , the magnified data are plotted , to clarify the definition of onset @xmath6 . the as - sintered sample exhibits
no superconducting anomaly as previously reported .
however , the reduction treatment on its as - sintered one in vacuum results in diamagnetic character from @xmath37 k for 48 h up to @xmath38 k for 72 h , which reaches highest value over the previous works . @xcite the magnetic susceptibility data taken under 1 oe using the fluxgate device are presented for the 72 h reduction sample . from the zfc magnetization at 2 k , the superconducting volume fraction is estimated to be @xmath39 , indicating bulk superconductivity .
next , let us examine the effect of external pressure on the low - t magnetization in the oxygen removed pr247 .
first of all , application of hydrostatic pressure on the superconducting sample enhances @xmath4 k at 0 gpa up to 36 k at 1.2 gpa as shown in fig .
[ mtp ] , where the data included the magnetization of the pressure cell are normalized by the value of magnetization at 40 k. in the inset of fig .
[ mtp](a ) , the enlarged plots are displayed to emphasize the variation of @xmath8 with pressure . in fig.[mtp ] ( b ) , the effect of pressure on @xmath8 of pure pr247 with @xmath35 prepared by a high pressure oxygen technique cited from ref.@xcite shows a rapid rise of @xmath8 from 15 k at ambient pressure up to 24 k at 1 gpa , indicating the stronger dependence of @xmath8 on pressure in comparison to the present result .
recently , ishikawa has reported that the application of pressure on the onset of @xmath6 in pr@xmath0ba@xmath1cu@xmath2o@xmath3 ( @xmath17=0.62 ) increases from 20.6 k at 0 gpa up to 33.1k at 2.5 gpa .
furthermore , increasing pressure above 3 gpa , the superconductivity disappeared and under high pressure above 5 gpa the temperature variation of the resistivity showed a semiconducting like behavior.@xcite these findings suggest that if the pure pr247 sample is located in the underdoped region , the present pr247 sample is near the optimum one .
second , the magnitude of negative magnetization ( the volume of the superconducting state ) is suppressed with the applied pressure of 1.2 gpa , which is contrast to the pressure effect of @xmath8 .
sano et al discussed the suppression of the superconductivity in the cuo double chain under pressure on the basis of tomonaga - luttinger liquid theory.@xcite if the volume shrinkage due to the application of pressure gives rise to a decrease of the distance between the two chains of a cuo double chain , we then expect that an increase of the hopping term between the nearest neighbor oxygen sites occupying 2@xmath15 orbitals , resulting in a suppression of the superconducting corerelation
. their calculation does not take into account for the pressure dependence of carrier concentration , giving no explanation for the overall change of the superconducting state with pressure .
the @xmath17 dependence of @xmath8 shows a monotonous increase from 15 k at @xmath40 up to 30 k at @xmath41 , which is qualitatively in good agreement with a theoretical prediction by nakano.@xcite comparing the effect of pressure on @xmath8 with the @xmath42 dependence of it , we expect that the application of pressure gives rise to an increase in the superconducting carrier concentration . for several reduction conditions , the temperature variation of electrical resistivity of the present sample pr247
is given in fig .
[ rt ] ( a ) . for 72h reduced case , @xmath43(t )
curve exhibits its dramatic drop near 30 k followed by zero resistance at low temperatures .
we notice that the resistivity in the normal state for except lower temperature region rises gradually with increasing the reduction time .
this tendency seems to be not anomalous but common behavior in the present system because an increase of the resistivity in cuo@xmath0 plane , @xmath44 , due to the reduction treatment is predominant over a decrease of that along cuo double chain , @xmath45 at higher temperatures . finally , we check thermal transport properties of the superconducting and non - superconducting pr247 ( fig . [
rt](b)).(thermal conductivity data are not shown here ) seebeck coefficient @xmath46 is a sensitive probe to examine the electronic state of highly inhomogeneous materials since the thermal current is hard to be disturbed by the grain boundaries in contrast to the electric current . upon decreasing @xmath47 ,
seebeck coefficient @xmath32 of the superconducting sample reaches a maximum peak , then shows a rapid decrease below 200 k , and finally followed by a substantial drop below @xmath48 .
it is true that the behavior of @xmath32 of the as - sintered non - superconducting sample is , both in its temperature dependence and magnitude , similar to that of pr124 with metallic double chain.@xcite the magnitude of @xmath46 of pr124 reaches a small value less than 1 @xmath49v / k below 100 k , which is explained on the basis of the one- to two - dimensional ( 1d-2d ) crossover . for conventional metal , seebeck coefficient is expressed in terms of @xmath50ln@xmath51 , where @xmath52 is the density of electronic state.@xcite in our rough estimation , if we assume that the magnitude of s is proportinal to the energy derivative of @xmath52 , the density of state for 2d metal is independent of the energy , giving @xmath53 .
however , the @xmath46 value of the superconducting pr247 shows a finite value of at most a few @xmath49v / k down to @xmath48 .
this finding is probably ascribed to a variation of the electronic state of the cuo double chain associated with carrier doping into the double - chain block due to the reduction treatment . upon decreasing @xmath47 , the hall coefficient @xmath54 of the 24
h reduced sample changes its sign across 100 k and then exhibits a rapid increases with negative sign .
we expect that the disagreement in sign between @xmath54 and @xmath46 in the oxygen removed sample is ascribed to the dramatic variation of the density of states with reduction treatment since @xmath55 near the fermi energy is related to the sign of seebeck coefficient in addition to the sign of charge carrier .
a previous work@xcite on role of oxygen in pr123 compound revealed that the oxygen deficient pr123 phase remains highly insulating state , causing neither metallic or superconducting sign .
double chain conduction in single crystalline pr124 exhibits a strongly metallic character but no superconducting transition down to 2 k.@xcite we concentrate on single - chain rich microdomains present in the nominal pr247 as observed by tem measurements@xcite , which is probably close to the appearance of higher @xmath8= 30 k accompanied by the larger oxygen defect @xmath34 .
it is , thus , expected that a reduction treatment removes oxygen from cuo single chain rich sequence such as \{s - d - s - s - s - d } and results in higher carrier doping into its surrounding cuo double chain in comparison to that in the regular sequence \{s - d - s - d - s - d}. in summary , we have demonstrated the pressure effect of @xmath56 , and the transport properties of the nominal composition pr@xmath0ba@xmath1cu@xmath2o@xmath3 .
magnetization and resistivity measurements of the @xmath17=0.94 sample reduced in vacuum exhibit @xmath4 k. the pressure dependence of @xmath8 of the @xmath34 sample strongly suggests the nearly optimum doping of the superconducting carries . comparing the effect of pressure on @xmath8 with the @xmath42 dependence of it ,
we notice that the application of pressure gives rise to an increase in the carrier concentration .
we believe that the heterogeneous structure of the nominal pr247 is closely related to the appearance of higher @xmath57 .
seebeck coefficient of the @xmath17=0.94 sample shows a metallic conduction below 100 k and is in its temperature dependence quite different from the data of the as - sintered one , leading to a dramatic change of the density of state along the cuo metallic double chain .
the authors are grateful for dr .
k. nonaka for his assistance in epma experiments .
this work was partially supported by a grant - in - aid for scientific research from japan society of the promotion of science .
30 l. soderholm , k. zhang , d. g. hinks , m. a. beno , j. d. jorgensen , c. u. segre , i. k. schuller , nature 328 ( 1987 ) 604 . s. horii , y. yamada , h. ikuta , n. yamada , y. kodama , s. katano , y. funahashi , s. morii , a. matsushita , t. matsumoto , i. hirabayashi , and u. mizutani , physica c 302 ( 1998 ) 10 .
r. fehrenbacher and t. m. rice , phys .
70 ( 1993 ) 3471 .
j. d. jorgensen , m. a. beno , d. g. hinks , l. sonderholm , k. j. volin , r. l. hitterman , j. d. grace , ivan k. schuller , c. u. segre , k. zhang , and m. s. kleefisch , phys.rev.b .
36 ( 1987 ) 3608 .
p. bordet , c. chaillout , j. chenavas , j. l. hodeau , m. marezio , j. karpinski , and e. kaldis , nature 334 ( 1988 ) 596 . y. yamada , s. horii , n. yamada , z. guo , y. kodama , k. kawamoto , u. mizutani , and i. hirabayashi , physica c 231(1994)131 .
j. y. genoud , t. graf , g. triscone , a. junod , and j. muller , physica c 192 ( 1992 ) 137 . m. matsukawa , y. yamada , m. chiba , h. ogasawara , t. shibata , a. matsushita , and y. takano , physica c 411 ( 2004 ) 101 .
a. matsushita , k. fukuda , y. yamada , f. ishikawa , s. sekiya , m. hedo , and t. naka , physica c 463 - 465 ( 2007 ) 161 .
s. watanabe , y. yamada and s. sasaki , physica c 426 - 431 ( 2005 ) 473 .
y. yamada and a. matsushita , physica c 426 - 431 ( 2005 ) 213 .
m. hagiwara , t. shima , s. tanaka , k. nishio , t. isshiki , t. saito , and k. koyama , physica c 463 - 465 ( 2007 ) 161 .
m. hagiwara , s. tanaka , t. shima , k. gotoh , s. kanda , t. saito , and k. koyama , physica c 468 ( 2008 ) 1217 .
f. ishikawa , k. fukuda , s. sekiya , a. kaeriyama , y. yamada and a. matsushita , j. phys .
, 76 ( 2007 ) suppl .
f. ishikawa , k. fukuda , y. hiura , y. onoda , a. nakayama , y. yamada , a. matsushita , and t. kageyama , j. phys .
series , 150 ( 2009 ) 052079 .
s. horii , y. yamada , t. kosaka , a. matsushita , t. matsumoto , n. mori , i. hirabayashi , and u. mizutani , advances in superconductivity viii ( 1996 ) 93 .
k. sano , y. ono , and y. yamada , j. phys .
jpn . , 74 ( 2005 ) 2885 .
t. nakano and k. kuroki , phys .
b76 ( 2007 ) 014515 .
i. terasaki , n. seiji , s. adachi , and h. yamauchi , phys .
b54 ( 1996 ) 11993 .
j.m.ziman , principles of the theory of solids ( the cambridge university press , 1976 ) .
m.e.lopez-morales , d. rios - jara , j. taguena , and r. escudero , phys .
b41 ( 1990 ) 6655 .
s. horii , u. mizutani , h. ikuta , y. yamada , j. h. ye , a. matsushita , n. e. hussey , h. takagi , and i. hirabayashi , phys .
b61 ( 2000 ) 6327 . | we have reported the effect of pressure on the magnetization , and transport properties in the nominal composition pr@xmath0ba@xmath1cu@xmath2o@xmath3 synthesized by a sol - gel technique .
a reduction treatment of the as - sintered sample in vacuum causes higher superconductivity achieving @xmath4 k for @xmath5 .
application of hydrostatic pressure on the oxygen depleted sample enhances its onset temperature up to 36 k at 1.2 gpa , indicating the nearly optimum doping level of the charge carrier in comparison to the pressure dependence of lower @xmath6 samples with @xmath7 .
seebeck coefficient of the superconducting sample shows a metallic conduction , followed by a clear drop below @xmath8 and is in its temperature dependence below 100 k quite different from that of the non - superconducting one .
this finding strongly suggests a dramatic change of the electronic state along the cuo double chain due to the reduction treatment for the appearance of superconductivity . |
an interesting problem of 4-dimensional topology is to determine whether @xmath5 has an exotic smooth structure .
the gluck twisting operation @xcite has been regarded as a potential method to produce an exotic @xmath5 ( cf .
this is the operation of removing an embedded @xmath6 from a 4-manifold and regluing it via the non - trivial diffeomorphism of @xmath7 . in @xcite by gluck twisting operation , an exotic pair of non - orientable 4-manifolds was constructed . on the other hand ,
many families of knotted @xmath1-spheres do not change @xmath5 by gluck twisting ( e.g. @xcite , @xcite , @xcite , @xcite , @xcite ) .
indeed , no exotic pair of orientable 4-manifolds , which are related each other by gluck twist , known to exits .
it is thus interesting to see whether gluck twists can produce an exotic pair of orientable 4-manifolds . for a smooth @xmath0-manifold @xmath4 and a
smoothly imbedded @xmath1-sphere with trivial normal bundle @xmath2 , let @xmath8 denote the tubular neighborhood of @xmath3 .
let @xmath9 be the manifold obtained by surgering @xmath4 along @xmath3 , that is @xmath10 we call a second homology class @xmath11 of @xmath4 is sherical , if @xmath11 is representd by a continuous map of the 2-sphere . in this paper ,
by using handlebody techniques and plug twisting operation , we give an alternative proof of the known result ( corollary [ intro : cor ] ) .
in fact , this approach gives the slightly stronger result below .
[ intro : thm ] let @xmath4 be any compact connected smooth @xmath0-manifold , and @xmath2 be a smoothly embedded @xmath1-sphere with trivial normal bundle .
then , gluck twisting @xmath4 along @xmath3 does not change the diffeomorphism type of @xmath4 , provided there is a @xmath1-dimensional spherical homology class in @xmath12 with odd self intersection .
note that this theorem is the best possible result about gluck twisting odd manifolds .
this is because @xcite implies that one can get an example of a @xmath1-sphere in an odd manifold @xmath2 , where gluck twisting @xmath4 along @xmath3 can change its diffeomorphism type ( consequently @xmath13 does not have any odd spherical class ) .
note that the orientibility is not part of the hypothesis of our theorem .
also note that the condition of having a spherical class in @xmath14 with odd self intersection is equivalent to having a class of odd self intersection in the cohomology group @xmath15 with coefficients , where @xmath16 is the group ring @xmath17 $ ] , since @xmath18 where @xmath19 denotes the universal cover , and @xmath20 cohomology with compact support .
[ intro : cor ] let @xmath4 and @xmath3 be as in theorem [ intro : thm ] .
if @xmath21 has a simply connected codimension 0 submanifold with odd intersection form , then gluck twisting @xmath4 along @xmath3 does not change the diffeomorphism type of @xmath4 .
* acknowledgements .
* this work was done during the second author s stay at michigan state university in march 2012 and march 2013 .
he would like to thank their hospitality .
the authors would like to thank robert gompf for pointing out an error in an earlier draft of this paper .
in this paper we will use the dotted circle notation of 1-handles . now let us recall the definition of the gluck twisting operation :
let @xmath22 be the self - diffeomorphism difined by @xmath23 , where @xmath24 denotes the @xmath25 rotation of the unit 2-sphere about the axis through its poles .
[ def : gluck ] let @xmath4 be a smooth @xmath0-manifold , and let @xmath3 be a 2-sphere in @xmath4 such that its self - intersecion number is zero .
let @xmath26 be the smooth @xmath0-manifold obtained from @xmath4 by removing the regular neighborhood @xmath8 of @xmath3 , and then regluing it via @xmath27 .
this construction @xmath28 is called the gluck twisting of @xmath4 along @xmath3 .
there are several ways to draw a handle diagram of gluck twists , here we recall the simple method pointed out in @xcite
. the usefulness of this new method will be evident in the proof of theorem [ intro : thm ] let @xmath4 and @xmath3 be as in definition [ def : gluck ] .
consider a handle decomposition @xmath29 where we represent @xmath8 by a @xmath30-framed unknot in the handle diagram .
introduce a canceling 1- and 2-handle pair locally as in the right picture of figure [ fig1 ] .
surger the introduced @xmath31 to @xmath6 , and surger @xmath32 to @xmath31 ( i.e. exchange the dot and the @xmath30 as in the bottom picture ) .
note that this is a plug twist along @xmath33 ( see @xcite ) .
this operation in fact describes the gluck twist : along @xmath3,width=336 ] [ gluck_diagram ] the resulting @xmath0-manifold is diffeomorphic to the gluck twist @xmath26 .
if @xmath4 has boundary , then we may assume that the diffeomorphism is the identity on the boundary .
one can easily see that the above operation is removing @xmath8 and regluing it via a self - diffeomorphism ( say @xmath34 ) of @xmath7 . a theorem of gluck @xcite ( cf .
@xcite ) on the diffeotopy group of @xmath7 implies that , we only need to check @xmath34 does not extend to a self - homeomorphism of @xmath6 .
this can be seen by applying the above procedure to the standard handle decomposition of @xmath35 and @xmath36 : the resulting manifold @xmath37 is not homeomorphic to @xmath35 .
we can similarly prove that the gluck twist is obtained by the plug twist along @xmath38 for any integer @xmath39 .
for a 4-dimentional handlebody @xmath40 , we denote by @xmath41 the subhandlebody of @xmath40 consisting of its handles with their indices @xmath42 .
we first prepare the lemma below .
let @xmath40 be a smooth compact connected 4-dimentional handlebody , and let @xmath43 be a continuous map .
we denote by @xmath44 $ ] the second homology class of @xmath40 given by @xmath45 .
suppose @xmath44\neq 0 $ ] .
then , if necessary after introducing canceling 2- and 3-handle pairs and sliding 2-handles , @xmath44 $ ] is represented by a 2-handle of @xmath40 whose attaching circle is null - homotoic in @xmath46 .
we may assume that @xmath45 is an immersion by homotopy .
furthermore , we may assume that the image of @xmath45 is contained in @xmath47 , since the cocore of a 3-handle is codimention 3 .
let @xmath48 be the transverse intersection points of @xmath49 and the cocores of 2-handles of @xmath40 .
let @xmath50 @xmath51 be a small 2-disk neighborhoods of @xmath52 in @xmath53 .
we may assume that each @xmath54 is a core of a 2-handle of @xmath40 and that @xmath55 is contained in @xmath46 . for each @xmath56
, take an arc @xmath57 connecting @xmath50 and @xmath58 , with @xmath59 its tubular neighborhood , so that @xmath60 is a 2-disk in @xmath53 and that each @xmath61 is an embedding .
we can isotope @xmath45 so that each @xmath62 is contained in @xmath63 .
now we obtain a 2-handle of @xmath40 whose attaching circle is @xmath64 , by introducing canceling 2- and 3-handle pairs and sliding 2-handles along the band @xmath65 s .
this 2-handle represents the class @xmath44 $ ] . since @xmath66 is a 2-disk , the attaching circle @xmath67 bounds an immersed disk in @xmath46 .
hence this attaching circle is null - homotopic in @xmath46 .
we next prove the main theorem .
let @xmath68 be the attaching circle of the handle representing the spherical homology class of @xmath12 with odd self intersection . according to the above lemma , we can assume @xmath68 is a null homotopic loop in the codimension zero submanifold @xmath69 of @xmath12 consisting of its @xmath30- and @xmath70-handles .
hence after ( self crossing ) clasp moves @xmath68 can be made to be an unknot . in paricular ,
if there is another handle represented by a small circle @xmath71 linking @xmath68 with zero framing , then by sliding @xmath68 over @xmath71 we can make @xmath68 an unknotted circle as shown in figure [ figa1 ] . having this observation in mind , let us perform the gluck twisting operation to @xmath4 along @xmath3 as shown in figure [ figa2 ] .
now by sliding the middle @xmath70-framed handle over @xmath68 , we get a 2-handle @xmath72 with even framing . by using the above observation
, we can isotope @xmath72 to the standard middle @xmath30-framed circle of the last picture of figure [ figa2 ] . here
consider the subhandlebody of @xmath26 described in the upper left diagram of figure [ fig4 ] .
we reverse the gluck twisting operation applied to @xmath26 .
lemma [ gluck_diagram ] shows that this reversing operation is equivalent to exchanging @xmath30 and the dot as shown in the left procedure of figure [ fig4 ] .
the right procedure of figure [ fig4 ] shows that this is a composition of undone operations ( recall that every self - diffeomorphism of @xmath73 extends to a self - diffeomorphism of @xmath74 . ) . hence @xmath26 is diffeomorphic to @xmath4 .
s. akbulut and k. yasui , _ corks , plugs and exotic structures _ , journal of gkova geometry topology , volume * 2 * ( 2008 ) , 4082 .
f. ding and h. geiges , _ the diffeotopy group of @xmath75 via contact topology _ , compos .
* 146 * ( 2010 ) , no . 4 , 10961112 . h. gluck , _ the embedding of two - spheres in the four - sphere _ ,
soc . * 104 * ( 1962 ) , 308333 .
r. gompf , _ more cappell - shaneson spheres are standard _ , algebr . geom . topol . *
10 * ( 2010 ) , no . 3 , 16651681 .
gompf and a.i .
stipsicz , _ @xmath0-manifolds and kirby calculus _ , graduate studies in mathematics , * 20*. american mathematical society , 1999 .
c. mca .
gordon , _ knots in the 4 -sphere _ , comment . math . helv .
* 51 * ( 1976 ) , no . 4 , 585596 . d. nash and a.i .
stipsicz , _ gluck twist on a certain family of 2-knots _ , arxiv:1103.5571 . | we give a simple criterion when a gluck twisting an odd smooth @xmath0-manifold along a @xmath1-sphere @xmath2 does not change its diffeomorphism type .
we obtain this by handlebody techniques and plug twisting operation , getting a slightly stronger version of the known fact that gluck twisting of a @xmath1-sphere @xmath2 of a compact smooth @xmath0-manifold with an odd spherical class , in the complement of @xmath3 , does not change the diffeomorphism type of @xmath4 .
this is the best possible result on gluck twisting manifolds with odd homology classes . |
since the discovery of the first planets around other stars @xcite , the search for exoplanets has blossomed to become one of the most exciting fields of modern astronomical research . the great majority of the hundreds of exoplanets that have been discovered over the past two decades have been found orbiting sun - like stars by dedicated international radial velocity programs . among these programmes
are harps ( the high accuracy radial velocity planetary search project , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , aaps ( the anglo - australian planet search , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , california ( e.g. * ? ? ?
* ; * ? ? ?
* ) , lick - carnegie ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and texas ( e.g. * ? ? ?
* ; * ? ? ?
the other main method for exoplanet detection is the transit technique , which searches for the small dips in the brightness of stars that result from the transit of planets across them .
ground - based surveys such as wasp @xcite and hat @xcite have pioneered such observations , and resulted in the discovery of a number of interesting planetary systems . in the coming years , such surveys using space - based observatories will revolutionise the search for exoplanets .
indeed , a rapidly growing contribution to the catalogue of known exoplanets comes from the _ kepler _ spacecraft ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which will likely result in the number of known exoplanets growing by an order of magnitude in the coming years . in recent years , a number of new exoplanet discoveries have been announced featuring host stars that differ greatly from the sun - like archetype that make up the bulk of detections .
the most striking of these are the circumbinary planets , detected around eclipsing binary stars via the periodic variations in the timing of observed stellar eclipses .
a number of these unusual systems feature cataclysmic variable stars , interacting binary stars composed of a white dwarf primary and a roche lobe filling m star secondary .
sharply - defined eclipses of the bright accretion spot , with periods of hours , can be timed with a precision of a few seconds . in these systems , the eclipse timings are fitted with a linear ephemeris , and the residuals ( @xmath0 ) are found to display further , higher - order variations .
these variations can be attributed to the gravitational effects of distant orbiting bodies which tug on the eclipsing binary stars , causing the eclipses to appear slightly early or late . this light - travel time
( ltt ) effect can then be measured and used to infer the presence of planetary - mass companions around these highly unusual stars .
some examples of circumbinary companions discovered in this manner include uz for @xcite , nn ser @xcite , dp leo @xcite , hu aqr @xcite and sz her @xcite the first circumbinary planets to be detected around hosts other than pulsars were those in the hw vir system @xcite , which features a subdwarf primary , of spectral class b , and a red dwarf companion which display mutual eclipses with a period of around 2.8 hours @xcite .
the detection of planets in this system was based on the timing of mutual eclipses between the central stars varying in a fashion that was best fit by including two sinusoidal timing variations .
the first , attributed to a companion of mass @xmath1 , had a period of 15.8 years , and a semi - amplitude of 77 s , while the second , attributed to a companion of mass @xmath2 , had a period of 9.1 years and semi - amplitude of 23 s. whilst these semi - amplitudes might appear relatively small , the precision with which the timing of mutual eclipses between the components of the hw vir binary can be measured means that such variations are relatively easy to detect . over the last decade ,
a number of studies have shown that , for systems that are found to contain more than one planetary body , a detailed dynamical study is an important component of the planet discovery process that should not be overlooked ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , despite these pioneering works , the great majority of exoplanet discovery papers still fail to take account of the dynamical behaviour of the proposed systems .
fortunately , this situation is slowly changing , and recent discovery papers such as @xcite , @xcite and @xcite have shown how studying the dynamical interaction between the proposed planets can provide significant additional constraints on the plausible orbits allowed for those planets .
such studies can even reveal systems in which the observed signal can not be explained by the presence of planetary companions .
the planets proposed to orbit hu aqr are one such case , with a number of studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) showing that the orbital architectures allowed by the observations are dynamically unstable on astronomically short timescales .
in other words , whilst it is clear that the observed signal is truly there , it seems highly unlikely that it is solely the result of orbiting planets .
given these recent studies , it is clearly important to consider whether known multiple - planet exoplanetary systems are truly what they seem to be . in this work ,
we present a re - analysis of the 2-planet system proposed around the eclipsing binary hw vir . in section 2 ,
we briefly review the hw vir planetary system as proposed by @xcite . in section 3 , a detailed dynamical analysis of the planets proposed in that work is performed . in section 4
, we present a re - analysis of the observations of hw vir that led to the announcements of the exoplanets , obtaining a new orbital solution for those planets which is dynamically tested in section 5 .
finally , in section 6 , we present a discussion of our work , and draw conclusions based on the results herein .
the hw vir system consists of a subdwarf b primary and an m6 - 7 main sequence secondary .
the system eclipses with a period of 2.8 hours @xcite , and the stars have masses of @xmath3 and @xmath4 @xcite .
changes in the orbital period of the eclipsing binary were first noted by @xcite ; further observations led other authors to suggest that the period changes were due to ltt effects arising from an orbiting substellar companion @xcite . @xcite obtained a further 8 years of photometric observations of hw vir . those data , in combination with the previously published eclipse timings spanning 24 years , indicated that the period changes consisted of a quadratic trend plus two sinusoidal variations with periods of 15.8 and 9.1 years .
@xcite examined alternative explanations for the cyclical changes , ruling out apsidal motion and magnetic period modulation via the applegate mechanism @xcite .
they concluded that the most plausible cause of the observed cyclic period changes is the light - travel time effect induced by two companions with masses 19.2 and 8.5 .
the parameters of their fit can be found in table [ lee09_params ] .
formally , the planets are referred to as hw vir ( ab)b and hw vir ( ab)c , but for clarity , we refer to the planets as hw vir b and hw vir c. | in 2009 , the discovery of two planets orbiting the evolved binary star system hw virginis was announced , based on systematic variations in the timing of eclipses between the two stars .
the planets invoked in that work were significantly more massive than jupiter , and moved on orbits that were mutually crossing - an architecture which suggests that mutual encounters and strong gravitational interactions are almost guaranteed . in this work ,
we perform a highly detailed analysis of the proposed hw vir planetary system .
first , we consider the dynamical stability of the system as proposed in the discovery work . through a mapping process involving 91,125 individual simulations , we find that the system is so unstable that the planets proposed simply can not exist , due to mean lifetimes of less than a thousand years across the whole parameter space .
we then present a detailed re - analysis of the observational data on hw vir , deriving a new orbital solution that provides a very good fit to the observational data .
our new analysis yields a system with planets more widely spaced , and of lower mass , than that proposed in the discovery work , and yields a significantly greater ( and more realistic ) estimate of the uncertainty in the orbit of the outermost body . despite this
, a detailed dynamical analysis of this new solution similarly reveals that it also requires the planets to move on orbits that are simply not dynamically feasible .
our results imply that some mechanism other than the influence of planetary companions must be the principal cause of the observed eclipse timing variations for hw vir .
if the system does host exoplanets , they must move on orbits differing greatly from those previously proposed .
our results illustrate the critical importance of performing dynamical analyses as a part of the discovery process for multiple - planet exoplanetary systems .
binaries : close , binaries : eclipsing , stars : individual : hw vir , planetary systems , methods : n - body simulations |
adhesion between elastic bodies was relatively unexplored until the last few decades , and this is reflected in the very marginal role it has in the otherwise very comprehensive book of k.l .
johnson ( 1985 ) , despite johnson himself is one of the authors of one of the most important papers on adhesion ( on adhesion of elastic spheres , the jkr theory , johnson et al . , 1971 , which has over 5000 citations at present ) .
this is obviously because until sufficiently accurate and high - resolution technique were developed , adhesion was hard to measure , because roughness , it was commonly observed and explained , destroys the otherwise very strong field of attraction between bodies , which should in principle make them stuck to each other at the theoretical strength of the material .
jkr theory itself was developed having in mind the special cases where adhesion can indeed be measured at the macroscopic scales , using very soft materials like rubber and gelatin spheres , clean and with extremely smooth surfaces .
today , there is however interest in both scientific and technological areas also at small scale , where very smooth surfaces for example in information storage devices result in adhesive forces playing a more crucial role than in more conventional tribological applications . on the other hand ,
when people have started to study adhesion in geckos , which adhere to just about any surface , being it wet or dry , smooth or rough , hard or soft , with a number of additional extraordinary features ( self - cleaning , mechanical switching ) , interest is emerging on how to reproduce these capabilities in `` gecko inspired synthetic adhesives '' .
the interest stems on the fact that adhesion can not be produced on hard rough surfaces , and therefore only the strikingly complex hierarchical structure of the gecko attachment can produce the macroscopic values of load that gecko can sustain .
the hierarchical structure of the gecko attachment ( about three million microscale hairs ( setae ) which in turn each branch off into several hundreds of nanoscale spatula , totalling perhaps a billion spatula ) makes one wonder why the multiscale nature of surface roughness also could not show an effect of adhesion enhancement . indeed , at least one model of adhesion of solid bodies ( that of persson and tosatti , 2001 , pt in the following ) , does show adhesion persistence and even enhancement .
there seems to be a qualitative difference for surfaces with fractal dimensions below @xmath2 , which turns out to be the case in most if not the totality of the commonly observed rough surfaces ( persson , 2014 ) . in general , it is hard to measure strong adhesion , despite the van der waals interactions in principle are orders of magnitude larger than atmospheric pressure this adhesion paradox ( pastewka and robbins , 2014 , persson _ et al_. , 2005 ) has been linked to surface roughness , but the explanations of the paradox have been different , the latest very interesting one being due to pastewka and robbins ( 2014 ) , which is a very promising parameter - free theory that shows how adhesion changes contact area and when surfaces are sticky but mostly in a regime near small contact areas .
pastewka and robbins ( 2014 ) conclude that " _ for most materials , the internal cohesive interactions that determine elastic stiffness are stronger than adhesive interactions , and surfaces will only stick when they are extremely smooth .
tape , geckos , and other adhesives stick because the effect of internal bonds is diminished to make them anomalously compliant"_. this conclusion seems in qualitative agreement with the classical asperity theory , except that pastewka and robbins use in their model quantities related to slopes and not to heigths and therefore are in quantitative disagreement .
persson ( 2002a , 2002b ) introduced more elaborate version of the theory , which solves the partial contact problem also , and the coupling of the two aspects ( effective energy due to roughness in full contact , and its use in a partial contact with a diffusion model ) makes the limit behaviour for very short wavelengths difficult to capture , and motivated us to search a possibly simpler , more traditional picture . the traditional asperity model of fuller and tabor ( 1975 ) ,
today is not considered to be adequate because of its many assumptions on geometry and absence of interaction , showed that adhesion and pull - off force is reduced very easily at macroscopic scale by roughness .
even extremely tiny amounts of roughness , of the order of the pull - off distance for the highest asperities in contact , make the pull - off force orders of magnitude lower than the nominal value .
ft theory seemed to be in good agreement with the experiments , within the limits of their accuracy .
the only case where it seemed contradicted by some experimental evidence , was in some measurements of adhesion in highly viscoelastic solids ( fuller and roberts , 1981 , briggs and briscoe 1977 ) .
these experiments indeed showed an enhancement of adhesion with roughness , which was not expected in the pure elastic ft model .
more recent evidence comes from the cleverly designed experiments using a two - scale axisymmetric problem with roughness between gelatin and perspex flat rough plates , by guduru and his group ( guduru ( 2007 ) , guduru _ et al _ ( 2007 ) , waters _
et al _ ( 2009 ) ) .
they showed clearly that an elastic jkr analysis explains the strong increment of pull - off forces observed ( an order of magnitude increase ) , and that this comes with irreversible energy dissipated in many jumps of the force - area curve .
in this paper , we shall try therefore a new model for a rough surface , completely different from either asperity models , and pt model ( or persson , 2002a , 2002b ) .
the model is based on the very simple idea johnson used several times in analyzing contacts near full contact , and which in turn could be attributed to bueckner ( 1958 ) : namely , that the gaps in an otherwise full contact are cracks that can not sustain finite stress intensity factors in the case of pure mechanical contact without adhesion , or that can sustain the appropriate stress intensity factor corresponding to the toughness @xmath3(in terms of surface energy , @xmath4 ) , in the case of adhesion .
further , it was used more recently by xu _
( 2014 ) ( xjm theory ) for a random rough surface near full contact but without adhesion , whose model in fact inspired the present extension to the case with adhesion .
before embarking into the full rough surface case , it is crucially important to understand qualitatively the mechanics of adhesion near full contact . the best strategy is to start from the relatively simple behavior of a single sinusoidal contact , as studied quantitatively by ken johnson under the jkr regime assumption ( johnson , 1995 ) .
taking therefore a sinusoid ( in either 1d or full 2d ) with @xmath5 wavelength , @xmath6 amplitude , and considering the limit case without adhesion @xmath7 is the compressive mean pressure to flatten the sinusoid and achieve full contact , the adhesive case follows curves of area - load described in fig.1 , where we have considered the case of a 1d profile for simplicity because it is fully analytical , whereas probably the 2d case can not be solved in closed form , except near full contact and near pull - off . starting from the case of `` low adhesion '' , @xmath8 , we can describe the behavior during loading as follows .
the curve has two extremes , a minimum and a maximum : under zero load , the contact jumps into a state of contact given by the intersection of the curve with the load axis . upon further increase of the load
, it follows the stable curve , until it jumps into full contact at the maximum . at this point
the strength is theoretically infinite ( more precisely , the theoretical stress of the material , which is very high ) unless we postulate the existence of some flaw of trapped air , as johnson suggests , and which gives a bounded tensile pressure for returning on the curve at the maximum . upon further unloading ,
the curve is followed stably until the minimum is reached ( therefore we have a new part of the curve that is now stable , that under negative loads ) , where pull - off , or jump out - of - contact , is obtained . for a `` critical '' value of adhesion , which depends on modulus of the material as well as the two length scales in the problem
, the surfaces will spontaneously snap into contact at zero load .
this occurs for johnson s parameter @xmath9 where @xmath10 is the surface energy .
what matters in particular to the present investigation is that the original contact curve ( that without adhesion ) changes sharply shape when adhesion is introduced , since the negative pressure region appears which is crucial to understand pull - off loads , and the transition towards infinite tension is also introduced , rather than having full contact at specific value of ( compressive ) pressure . however ,
for what concerns the condition of jump into full contact , this is essentially given by the maximum of the curve which is a perturbation of the contact solution the curve for negative tension and pull - off requires a detailed analysis of the regime of low contact , and we believe for this part the model of fuller and tabor ( 1975 ) is a good starting point , as the hertzian jkr solution is a good starting point to study the sinusoidal indenter .
notice finally that for @xmath11 we simply have that the pull - off force will be load dependent .
* figure 1 .
the relationship between * @xmath12 * * and contact area ratio * * @xmath13 * * for johnson s jkr solution of the single 1d sinusoidal adhesive contact problem .
the change from the pure contact case * * @xmath14 , * to adhesive case * @xmath15 .
in the classical random process theory , the pressure to cause full contact , @xmath16 , is a random variable , whose variance is easily related in terms of the power spectrum density ( psd ) of the profile or of the surface ( see manners and greenwood , 2006 , persson , 2001 , persson and tosatti , 2001 ) @xmath17 where @xmath18 is the variance of the slopes .
@xmath19 here is the combined plane strain modulus of the contact materials .
the distribution of pressures in full contact is also a gaussian distribution , namely@xmath20 \label{pfullcontact}\ ] ] this means that , strictly speaking , there is always a tail of negative ( tensile ) pressures , and indeed persson s solution for the adhesionless contact problem simply truncates this distribution by subtracting from it a specular distribution , with negative mean pressure , obtaining in fact a rayleigh distribution . for typical self - affine surfaces
, eliminating the so called `` roll - off '' wavelength , the power spectrum is a power law above a certain long wavelength cut - off ( wavenumber @xmath21)@xmath22{cc}0 & \text{for } q < q_{0}\\ \frac{h}{2\pi}\left ( \frac{h_{0}}{\lambda_{0}}\right ) ^{2}\left ( \frac { q}{q_{0}}\right ) ^{-2\left ( h+1\right ) } & \text{for } q > q_{0}\end{array } \right.\ ] ] for any finite short wavelength cutoff @xmath23 , the moments are bounded , but in the limit , @xmath24 only is bounded ( the variance of the heights ) , whereas @xmath25 diverges as well as the other higher order ones .
for this reason , it is already well known that full contact can not occur for any finite pressure , in the fractal limit , as it is for determinist fractal profiles like those defined with the weierstrass series in ciavarella et al .
( 2000 ) . in the random case , persson s solution ,
which is approximate but qualitatively correct ( wolf et al , 2014 ) shows that the contact area is only complete for infinite applied compression . according to bueckner s principle ( 1958 ) , as used by johnson ( 1995 ) and xjm ( xu _ et al .
_ 2014 ) , we need to look carefully at the tensile stresses of the full contact pressure solution , which are applied by superposition to the gaps area , and compute the stress intensity factors .
considering this contact pressure solution as a random process , it is clear that its moments will correspond to the original surface moments , shifted by a factor 2 and multiplied by @xmath26@xmath27 according to taylor series expansion therefore , as manners and greenwood ( 2006 ) , and xjm ( xu _ et al .
_ 2014 ) also suggest , the tensile part of the full contact solution can be approximated with quadratic equations near the pressure tensile summits , similarly to what is done for asperity theories for the real geometry of the surface .
this generates a model of isolated `` non - contact '' areas . the xjm model ( xu _ et al .
_ 2014 ) shows that this leads to the following:- * the individual tensile area is @xmath28 where @xmath29 is the radius of the asperity full contact pressure distribution , and @xmath30 is the pressure on the asperity ( the full contact solution in the tensile regions , with a change in sign due to the bueckner s superposition ( 1958 ) ) ; * the non - contact area can be exactly shown , according to bueckner s principle , to correspond to a pressurized crack . * for the pure contact case , the stress intensity factor ( sif ) has to be zero along the boundary , axysimmetric by assumption ( consideration of elliptical form do not change the results significantly)@xmath31
= 0\ ] ] leading to the conclusion that the non - contact area is larger than the tensile stress area , and in particular , it is @xmath32 in size of the original tensile area , @xmath33 notice however that the condition is also satisfied trivially by the solution @xmath34 since clearly the fact that the size of the gap goes to zero is also a solution is problematic for studying that limit .
the point is we also have another condition and not just lefm : namely , that there can not be any tension , which means actually the strength is zero .
if we consider adhesion , there can be tension , up to the theoretical strength .
* for the case where there is in fact surface energy , the condition becomes more elaborate @xmath31
= k_{ic } \label{kic}\ ] ] and for example if @xmath35 , we would have an open crack @xmath36 that would tend to propagate .
we expect naturally @xmath37 as an effect of adhesion .
* there can not be solutions below the size where we need to take into account the transition towards a strength criterion @xmath38 .
however , this is of concern only if there is a solution of full contact with finite pressure , and in any case , the suggestion of johnson to consider at this point the presence of trapped air and we shall return later on this point . substituting ( [ area_tension ] ) , into ( [ ct_noadhesion ] ) , we have an equation for @xmath39 @xmath40 = k_{ic } \label{kic2}\ ] ] it is clear that we can not solve this equation easily in a rigorous sense , since @xmath30 is a random variable .
this equation ( [ kic2 ] ) is an implicit equation which defines @xmath41 and acts like the basic function defining the local area as a function of the `` separation '' @xmath42 in the equivalent asperity model created by the tensile full contact pressure `` surface '' .
it is useful to derive separately this case , as done by xu et al ( 2014 ) . from ( [ kic2 ] ) , with @xmath43@xmath44 there is no minimum tensile tension that can be sustained ( unlike with adhesion ) and the integration proceeds simply as suggested by eqt.38 of xu et al ( 2014)@xmath45 where @xmath46 is the asperity density of the full contact pressure surface , @xmath47 is the distribution of the pressures summits in this surface , and hence it can be solved easily . notice the integral is simply of the same mathematical form as in the standard greenwood and williamson s ( 1966 ) theory , where mean separation is replaced by mean pressure , and the geometrical surface is replaced by the pressure surface .
in the present form , we obtain@xmath48 -\frac{1}{2}\frac{\overline{p}}{\sqrt{v}}erfc\left ( \frac{\overline{p}}{\sqrt{2v}}\right ) \right)\ ] ] further , at sufficiently large magnifications , @xmath49 , and @xmath50 , while also the density of asperities @xmath51 , which gives @xmath52 where @xmath53 is a prefactor of the order 1 ( the exact prefactor to make the area of contact zero at zero pressure is @xmath54 , but this is not the correct value at large pressures where we are concentrating our efforts)@xmath55 -\frac{1}{2}\frac{\overline{p}}{\sqrt{v}}erfc\left ( \frac { \overline{p}}{\sqrt{2v}}\right)\ ] ] this suggests that if we want to keep a constant value of given area of gap , upon increasing magnification , we need to keep @xmath56 constant , i.e. increase the mean pressure without limit .
this is the well know behavior of pure contact problem , and it is confirmed here .
it is important to discuss in details equation ( [ kic2 ] ) , as it governs the basic behavior of the gaps during the loading process .
first of all , it is easier to manipulate it in terms of the tension on the `` pressurized cracks '' , as ( notice that @xmath29 being the radius of the pressure surface , has dimensions @xmath57 ) , by rewriting it as @xmath58 where the second term cancels out obviously in the adhesion - less case , but also with adhesion at very large @xmath39 .
a few curves could be plotted to show the general trend of introducing a minimum which clearly corresponds to the maximum in the case in johnson s sinusoidal case ( see fig.1 ) near the ( unstable ) transition to full contact .
we prefer however to arrive at a cleaner plot , which will be in fig.2 , to include @xmath3 in a unique curve .
notice that as we increase the mean compression in the contact , the actual value of tension in the gaps decreases therefore , the loading progresses here by reducing the pressure on the y - axis .
the minimum occurs at @xmath59 , which gives @xmath60 in which case , for @xmath61 @xmath62 @xmath63 where @xmath29 is a radius of pressure `` asperity '' .
hence , we can rewrite ( [ kic3 ] ) as@xmath64 the result comes clean defining @xmath65 hence , we can rewrite ( [ eqz - intermedia ] ) as @xmath66 and this is indeed the curve plotted in fig .
2 . in the adhesion - less case , @xmath67 , and
@xmath68 and hence the curve diverges to infinity and hence this simple unique curve for the adhesive case represents a discontinuity and the dimensional quantities should be plotted to see more clearly the transition from the adhesion - less case to the adhesive one .
* figure 2 .
the relationship between peak dimensionless tension * @xmath69 * applied on a pressurized isolated gap and the size of the dimensionless gap radius squared * @xmath70 * * , together with a linear asymptotic approximation which turns out to be reasonably accurate also at low tensions .
the minimum in the curve , which leads to the transition to full contact , is obviously overestimated by the approximate curve , so there will be a minor spurious effect of increase of area of the gaps .
the way the plot is constructed does nt permit to see the limit of pure contact ( adhesive - less case ) , since both * * @xmath71 * * and * * @xmath72 * * are zero in that limit , and hence the adhesive - less curve becomes the very tail of the present one at infinity * * further , considering that we want to make estimates of the area of the gap that remains in contact , it is clear that a very good approximate solution could be @xmath73{cc}\frac{1}{5}\widehat{c}_{t}^{2}+\frac{1}{2 } & \text{for } \widehat{c}_{t}^{2}>\frac{5}{2}\\ 1 & \text{for } 0<\widehat{c}_{t}^{2}<\frac{5}{2}\end{array } \right .
\label{p - approx}\ ] ] since the other branch of the equation is unstable .
the only significant approximation this introduces , it is to overestimate the size of the gap region radius just before jumping into contact . however , while this will only change some prefactors by small amounts , it simplifies the study of the problem enormously .
now , supposing we start from a contact which does nt jump into contact completely and with an applied compressive load but that gaps area do exist .
it is clear that the ( [ p - approx ] ) solution holds for each gap , depending on the local tension arising from cancelling the tension in the full contact pressure , and at any given mean applied load , each gap will have a certain level of pressure and therefore a certain equilibrium size as per ( [ p - approx ] ) , which may include some gaps closing .
we need an integration process to establish the total area of gaps , and therefore the complementary remaining contact area . also , upon loading , the mean compressive pressure increases , and hence the tensile pressure on each gap decreases , so that more gaps will tend to reach the condition @xmath69 @xmath74 .
therefore , a certain number of gaps will close , and the others will reduce their size . to understand if the final state is full contact or not , we should consider the adhesion - less case for reference . in this case , approximately , persson s solution indicates that only @xmath75 is the parameter ruling the contact area size .
if we increase the short wavelength content , increasing @xmath75 , for a given contact area , we have to increase the mean pressure in proportion .
given @xmath75 grows unbounded , the pressure to obtain any value of contact area ( in fact , not just full contact ) , grows unbounded . repeating this reasoning here , we need to observe if , for a given condition with adhesion , the contact area depends only on the ratio @xmath76 or not . rewriting and inverting the equation ( [ p - approx ] ) in terms of the gap radiuses@xmath77{cc}5\left ( \widehat{p}-\frac{1}{2}\right ) & \text{for } \widehat{p}>1\\ 0 & \text{for } \widehat{p}<1 \end{array } \right.\ ] ] and in dimensional terms@xmath78{cc}5\left ( \frac{\left ( p-\overline{p}\right ) } { p_{0}}-\frac{1}{2}\right ) & \text{for } \frac{\left ( p-\overline{p}\right ) } { p_{0}}>1\\ 0 & \text{for } \frac{\left ( p-\overline{p}\right ) } { p_{0}}<1 \end{array } \right .
\label{ct - final}\ ] ] or@xmath79 \text { , for } \left ( p-\overline{p}\right ) > p_{0}\ ] ] we shall neglect the variation of @xmath29 with height , as otherwise the integration becomes too cumbersome .
we simply assume a mean value given by random process theory , and develop an integration of the type@xmath80 @xmath81 where the function for @xmath39 is now given in ( [ ct - final ] ) , which results in @xmath82 -\frac{\overline { p}+p_{0}}{2\sqrt{v}}erfc\left ( \frac{\overline{p}+p_{0}}{\sqrt{2v}}\right ) \right]\ ] ] which agrees with the adhesiveless case when @xmath83 . in this format
, it is clear that the effect of @xmath72 is exactly analogous to an increase of the mean pressure .
the non - contact area would tend to stay constant with magnification @xmath84 now if we keep increase , instead of the applied pressure proportionally to @xmath49 , only @xmath72 , and this is simple to study . at sufficiently large magnifications , looking at the parameter ( [ pmin ] ) , and considering the usual scaling arguments on the psd and its moments@xmath80 @xmath85 where the exponent @xmath86 is positive for @xmath87 ( low fractal dimensions ) , when the dimensionless @xmath88 increases with magnification , or negative otherwise .
hence , for low fractal dimensions , and a given applied mean pressure ( including zero ) , the non - contact area tends to decrease without limit , implying the tendency to full contact .
naturally , the tendency will be stronger the higher @xmath87 i.e. the farther from the limit case of @xmath89 .
there seems to be some connection to the conclusions and the parameters involved in the `` effective adhesion energy '' in persson and tosatti ( 2001 ) , which leads them to suggest that adhesion persists for low fractal dimensions @xmath0 ( the real range of surfaces , see persson ( 2014 ) ) . in their theory
, they obtain from the elastic energy associated to the deformation in balance with surface energy of a full spectrum of frequencies , a parameter which seems related to @xmath90 where @xmath91 is the first order moment of the psd .
this has the same qualitative power law behaviour ( besides the @xmath92factor ) , but it should be emphasized different from in pt coming from the first moment of the surface , and in our case from a combination of roots of the second and 6-th moment . * figure 3 .
the area of gaps * @xmath93 * for zero applied pressure decreases rapidly with the dimensionless adhesion pressure term * @xmath88 * * and when * * @xmath94**we can assume full contact holds . * * fig.3 shows a plot of the non contact area at zero applied mean pressure , @xmath95 , which decays very rapidly with the dimensionless ratio @xmath88 , indicating there is chance of spontaneous full contact for @xmath87 , although the model is clearly approximate in that range since we are far from full contact . for @xmath96 and hence large fractal dimensions
, the dimensionless ratio @xmath88 will decrease with magnification , and hence we return at large magnifications to the case of pure adhesion - less contact , for which we expect at zero load simply zero area . the situation is clear therefore also with applied compressive forces to the contact .
contrary to the case without adhesion , where no mean pressure is sufficient to squeeze the contact flat , here for large hurst exponent @xmath87 , the non - contact area decreases asymptotically to zero and the trends of the zero applied pressure are confirmed , as clearly seen in fig.4 , where we assume the power law scaling @xmath97 with @xmath98 .
@xmath99{cc}\frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig4a.eps } } & a\\ \frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig4b.eps } } & b\\ \frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig4c.eps } } & c \end{array } $ ] * figure 4 .
the area of gaps ( complementary to the real contact area ) for different * @xmath100 * * as a function of the dimensionless applied compression * * @xmath101 * * shows that the contact becomes practically full at finite values of the * * @xmath101 * * contrary to the purely mechanical case for * * @xmath87 . *
vice versa , for * @xmath89 * * , this tendency is very similar to the adhesive - less contact and for * * @xmath102 * there is no significant difference with the adhesive - less case .
we assume power law scaling for * @xmath88 * with * @xmath103 * *
we obtained a gw - like asperity full contact pressure surface model for adhesion and more in general , in the adhesive - less version of this ( obtained putting @xmath83 ) , xu et al ( 2014 ) noticed the close connection to the results with the persson ( 2001 ) theory of contact .
xu et al ( 2014 ) introduced varius advanced asperity models .
in particular , the most advanced theories which recognize the variation of mean radius with heigth , and for large separations they are considered to be exact large separation being here , in the equivalent pressure `` surface '' model , large pressures .
a further attempt is of interest , since it brings very close comparison with persson s theory . as recognized by the advanced theories of bgt or carbone ( 2009 ) , the contact area
can be connected the bearing area ratio @xmath104 at large separations to the contact area , and hence , here a given separation @xmath105 , it depends only on a single parameter of the random process surface ( @xmath24 is the zero - th order moment of the spectral density of the surface ) : @xmath106 where separation is made non - dimensional by the ratio @xmath107 . when translating this into the `` pressure surface '' model , this should give an area of `` non - contact '' as a function of mean pressure @xmath108 , made non - dimensional by dividing it by the @xmath109 order moment of the pressures @xmath110 defined above ( [ variance ] ) , which in fact translates onto @xmath111 . here , we recognize already some features of the persson s equation , contrary to the more `` accurate '' asperity models in xu _
( 2014 ) which additionally depend on bandwidth parameter .
now , in the limit when the mean pressure is zero , we know that by definition , the non - contact area should be 1 , so that the real contact area is zero .
therefore , in order for the non - contact area to be 1 at zero separation , we need it to be double of the bearing area ( [ bearing - area ] ) at zero separation . hence , we multiply ( [ bearing - area ] ) by factor 2 , and use ( [ m0 ] ) , to get @xmath112 writing in terms of the complementary term , the area of contact , we get@xmath113 remarkably , this is _ exactly _ persson s equation , in the entire range of pressures , including the low pressure end . therefore , the use of the bearing area assumption is not limited by extremely large separations in the asymptotic version of the area - separation relationship in bgt s model . turning back on the adhesion case
, we reached the conclusion that the integration for a gw - equivalent model was of the type @xmath80 @xmath114 \text { } \phi\left ( p\right ) dp\ ] ] and hence , this is exactly equivalent to an adhesiveless contact problem where the mean pressure has been replaced by the sum @xmath115 hence , in order to obtain a result consistent to adhesionless persson s theory in the adhesive case , we need to multiply by a factor 4/3 , resulting in @xmath80@xmath116 resulting in @xmath117 comparison with the previous theory shows that the results are qualitatively similar . however , it is easier to plot them in terms of actual contact area , rather than non - contact , because we forces the prefactors to be such , as in the original persson s theory , to produce zero contact area at zero load , in the absence of adhesion .
fig.5 therefore plots eve the range of negative applied load , where in some case the contact is nearly full and hence the assumptions made of isolated gaps in the present model may be fulfilled .
the results loose sense if the contact area is small , viceversa .
@xmath99{cc}\frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig5a.eps } } & a\\ \frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig5b.eps } } & b\\ \frame{\includegraphics[width=120 mm , keepaspectratio = true]{new_theory_fig5c.eps } } & c \end{array } $ ] * figure 5 .
the real contact area for different * @xmath100 * * as a function of the dimensionless applied compression * * @xmath101 * * shows that the contact becomes practically full at finite values of the * * @xmath101 * * ( and even zero load for * * @xmath118 * * , within the limits of our model ) . for * * @xmath89 * * , the magnification effect is neutral , whereas for * * @xmath102 * there is no significant difference with the adhesive - less case , and the curves are close to the adhesive - less case .
we assume power law scaling for * @xmath88 * with * @xmath103 * * let us now discuss the case of pull - off in qualitative terms .
this may occur in a range with small contact area , if either the contact does nt proceed spontaneously into large fractions of the nominal contact area , or if the applied loading does nt push toward this range . in that case contact
is so clearly isolated on asperities , than theories like fuller and tabor ( 1975 ) would be approximately true .
the problem and limit of the present theory is , insteed , that it can not deal with the low contact areas , where gaps are interacting and certainly the bends in the area - load curves which would appear in a set of isolated asperity theories are not reproduced here .
the only estimate we can make is therefore to extrapolate pull - off from the point where the contact area is expected to be zero , and this gives . @xmath119 which leads to a new meaning for the pressure @xmath72 @xmath120 which always increases with magnification .
notice however that in absolute terms , @xmath121 is always increasing for all parameters @xmath122 and this result is not easy to beleive , and in opposite constrast to asperity theories but even pastewka and robbins ( 2014 ) who find stick surfaces only those that are smooth enough ( in terms of surface slopes ) to have the cohesive energy in the bulk giving up against the adhesion forces at the interface .
a new model of adhesion has been discussed and shown to lead to very simple and clear results : there can not be spontaneous jump into contact for any surface having sufficiently multiscale content , no matter its fractal dimension .
the model is devised near the full contact regime , so that the contact consists of a set of isolated gaps whose surfaces are then loaded by bueckner s principle by the tensile pressures of the `` linear '' full contact solution , which are approximated by parabola since , by taylor s expansion , they must have this form near full contact when gaps are closing , and it is easy to write this for a gaussian surface . the stable branch of the curve of the gap radius vs applied pressure in the gaps
are then found imposing the stress intensity factor to be constant along the edge , and a very good approximation turns out to be a linear law with an offset , which permits extremely simple integration , which resemble those of asperity models , and indeed can be considered as a `` pressure asperity '' model for the pressurized gaps . in the case of no adhesion ,
as already noticed by xu et al ( 2014 ) , the result turn out to be extremely similar to those of persson s contact theory ( 2001 ) which is a widely recognized as a good approximate solution near full contact , and which gives us confidence the results are also very similarly accurate where it tends to be exact in the limit of full contact ( for infinite mean pressure applied ) .
it is shown that a dimensionless ratio governs the contact and a pressure @xmath72 can be defined which is scale dependent and includes the energy of adhesion .
this pressure has a role equivalent to the mean applied pressure in the equation of the non - contact area and hence since it grows without limit for low fractal dimensions , permits full contact to be achieved for those surfaces .
the conclusions can not be complete of unloading and pull - off force , which require further investigation .
an equation which looks like an extension of persson s theory for contact mechanics has been derived .
briggs g a d and briscoe b j 1977 the effect of surface topography on the adhesion of elastic solids j. phys .
phys . 10 24532466 ciavarella , m. , et al .
`` linear elastic contact of the weierstrass profile . '' proceedings of the royal society of london .
series a : mathematical , physical and engineering sciences 456.1994 ( 2000 ) : 387 - 405 .
persson , b n j , albohr , o , tartaglino , u , volokitin a i and tosatti .
e _ _ _ _ 2005 . on the nature of surface roughness with application to contact mechanics , sealing , rubber friction and adhesion j. phys . : condens .
matter 17 r1 , _
yastrebov , v.a . ,
anciaux , g. , molinari , j.f . , 2014 .
from infinitesimal to full contact between rough surfaces : evolution of the contact area .
j. solids struct .
available from : @xmath123http://arxiv.org / abs/1401.3800@xmath124 . | recently , there has been some debate over the effect of adhesion on the contact of rough surfaces .
classical asperity theories predict , in agreement with experimental observations , that adhesion is always destroyed by roughness except if the amplitude of the same is extremely small , and the materials are particularly soft .
this happens for all fractal dimensions .
however , these theories are limited due to the geometrical simplification , which may be particularly strong in conditions near full contact .
we introduce a simple model for adhesion , which aims at being rigorous near full contact , where we postulate there are only small isolated gaps between the two bodies .
the gaps can be considered as `` pressurized cracks '' by using ken johnson s idea of searching a corrective solution to the full contact solution .
the solution is an extension of the adhesive - less solution proposed recently by xu , jackson , and marghitu ( xjm model ) ( 2014 ) .
this process seems to confirm recent theories using the jkr theory , namely that the effect of adhesion depends critically on the fractal dimension . for @xmath0 , the case which includes the vast majority of natural surfaces
, there is an expected strong effect of adhesion . only for large fractal dimensions , @xmath1 ,
seems for large enough magnifications that a full fractal roughness completely destroys adhesion .
these results are partly paradoxical since strong adhesion is not observed in nature except in special cases
. a possible way out of the paradox may be that the conclusion is relevant for the near full contact regime , where the strong role of flaws at the interfaces , and of gaps full of contaminant , trapped air or liquid in pressure , needs to be further explored .
if conditions near full contact are not achieved on loading , probably the conclusions of classical asperity theories may be confirmed .
roughness , adhesion , fuller and tabor s theory , fractals |
@xmath0 boson , electrically neutral vector boson ( its spin equals 1 ) with mass @xmath28 gev and width @xmath29 gev , .
] occupies a unique place in physics .
this heavy analog of the photon was experimentally discovered in 1983 , practically simultaneously with its charged counterparts @xmath30 bosons with mass @xmath31 gev and width @xmath32 gev @xcite .
the discovery was crowned by nobel prize to carlo rubbia ( for the bosons ) and to simon van der meer ( for the cern proton - antiproton collider , which was specially constructed to produce @xmath33 and @xmath0 bosons ) @xcite .
the extremely short - lived vector bosons ( @xmath34 sec ) were detected by their decays into various leptons and hadrons .
the detectors , in which these decay products were observed , were built and operated by collaborations of physicists and engineers the largest in the previous history of physics .
the discovery of @xmath33 and @xmath0 bosons was a great triumph of experimental physics , but even more so of theoretical physics .
the masses and widths of the particles , the cross - sections of their production turned out to be in perfect agreement with the predictions of electroweak theory of sheldon glashow , abdus salam and steven weinberg @xcite .
the theory was so beautiful that its authors received the nobel prize already in 1979 @xcite , four years before its crucial confirmation .
the electroweak theory unified two types of fundamental interactions : electromagnetic and weak .
the theory of electromagnetic interaction quantum electrodynamics , or qed , was cast in its present relativistically covariant form in the late 1940 s and early 1950 s and served as a role model " for the relativistic field theories of two other fundamental interactions : weak and strong .
the main virtue of qed was its renormalizability .
let us explain this technical " term by using the example of interaction of photons with electrons .
one can find systematic presentation in modern textbooks @xcite . in the lowest approximation of perturbation theory ( the so called tree approximation in the language of feynman diagrams ) all electromagnetic phenomena can be described in terms of electric charge and mass of the electron ( @xmath35 , @xmath36 ) .
the small parameter of perturbation theory is the famous @xmath37 .
the problem with any quantum field theory is that in higher orders of perturbation theory , described by feynman graphs with loops , the integrals over momenta of virtual particles have ultraviolet divergences , so that all physical quantities including electric charge and mass of the electron themselves become infinitely large . to avoid infinities an ultraviolet cut - off @xmath38
could be introduced .
another , more sophisticated method is to use dimensional regularization : to calculate the feynman integrals in momentum space of @xmath39 dimensions .
these integrals diverge at @xmath40 , but are finite , proportional to @xmath41 in vicinity of @xmath40 , where by definition @xmath42 ( see appendix a ) .
the theory is called renormalizable , if one can get rid of this cut - off ( or @xmath41 ) by establishing relations between observables only . in the case of electrons and photons such basic observables are physical ( renormalized ) charge and mass of the electron .
this allows to calculate higher order effects in @xmath43 and compare the theoretical predictions with the results of high precision measurements of such observables as e.g. anomalous magnetic moments of electron or muon .
the renormalizability of electrodynamics is guaranteed by the dimensionless nature of the coupling constant @xmath43 and by conservation of electric current . after this short description of qed
let us turn to the weak interaction .
the first manifestation of the weak interaction was discovered by henri becquerel at the end of the xix century .
later this type of radioactivity was called @xmath44-decay .
the first theory of @xmath44-decay was proposed by enrico fermi in 1934 @xcite .
the theory was modelled after quantum electrodynamics with two major differences : first , instead of charge conserving ,
neutral " , electrical current of the type @xmath45 there were introduced two charge changing , charged " , vector currents : one for nucleons , transforming neutron into a proton , @xmath46 , another for leptons , transforming neutrino into electron or creating a pair : electron plus antineutrino , @xmath47 .
( here @xmath48 denotes operator , which creates ( annihilates ) electron and annihilates ( creates ) positron .
the symbols of other particles have similar meaning ; @xmath49 are four dirac matrices , @xmath50 . )
the second difference between the fermi theory and electrodynamics was that the charged currents interacted locally via four fermion interaction : @xmath51 where summation over index @xmath43 is implied ( in this summation we use feynman s convention : @xmath52 for @xmath53 and @xmath54 for @xmath55 ) ; h.c .
means hermitian conjugate .
the coupling constant @xmath56 of this interaction was called fermi coupling constant . from simple dimensional considerations
it is evident that dimension of @xmath56 is ( mass)@xmath57 and therefore the four - fermion interaction is not renormalizable , the higher orders being divergent as @xmath58 , @xmath59 ... .
why these divergent corrections still allow one to rely on the lowest order approximation , remained a mystery .
but for many years the lowest order four - fermion interaction served as a successful phenomenological theory of weak interactions .
it is in the framework of this phenomenological theory that a number of subsequent experimental discoveries were accommodated .
first , it turned out that @xmath44-decay is one of the large family of weak processes , involving newly discovered particles , such as pions , muons and muonic neutrinos , strange particles , etc .
second , it was discovered that all these processes are caused by selfinteraction of one weak charged current , involving leptonic and hadronic terms .
later on , when the quark structure of hadrons was established , the hadronic part of the current was expressed through corresponding quark current .
third , it was established in 1957 that all weak interactions violate parity conservation @xmath60 and charge conjugation invariance @xmath61 .
this violation turned out to have a universal pattern : the vector form of the current @xmath62 , introduced by fermi , was substituted @xcite by one half of the sum of vector and axial vector , @xmath63 , which meant that @xmath49 should be substituted by @xmath64 .
in other words one can say that fermion @xmath65 enters the charged current only through its left - handed chiral component @xmath66 from such structure of the charged current it follows that the corresponding antifermions interact only through their right - handed components .
attempts to construct a renormalizable theory of weak interaction resulted in a unified theory of electromagnetic and weak interactions
the electroweak theory @xcite with two major predictions .
the first prediction was the existence alongside with the charged weak current also of a neutral weak current .
the second prediction was the existence of the vector bosons : @xmath0 coupled to the weak neutral current and @xmath67 and @xmath68 coupled to the charged current ( like @xmath69 ) and its hermitian conjugate current ( like @xmath70 ) .
the vector bosons were massive analogues of the photon @xmath71 ; their couplings to the corresponding currents , @xmath72 and @xmath73 , were the analogues of the electric charge @xmath35 .
thus @xmath74 and @xmath75 were dimensionless like @xmath76 , which was a necessary ( but not sufficient ) condition of renormalizability of the weak interaction .
the first theory , involving charged vector bosons and photon , was proposed by oscar klein just before world war ii @xcite .
klein based his theory on the notion of local isotopic symmetry : he considered isotopic doublets @xmath77 and @xmath78 , and isotopic triplet @xmath79 . where @xmath80 denoted what we call now @xmath30 , while @xmath81 was electromagnetic field .
he also mentioned the possibility to incorporate a neutral massive field @xmath82 ( the analogue of our @xmath83 ) . in fact
this was the first attempt to construct a theory based on a non - abelian gauge symmetry , with vector fields playing the role of gauge fields .
the gauge symmetry was essential for conservation of the currents .
unfortunately , klein did not discriminate between weak and strong interaction and his paper was firmly forgotten .
the non - abelian gauge theory was rediscovered in 1954 by c.n.yang and r.mills @xcite and became the basis of the so - called standard model with its colour @xmath84 group for strong interaction of quarks and gluons and @xmath85 group for electroweak interaction ( here indices denote : @xmath86 colour , @xmath87 weak isospin of left - handed spinors , and @xmath88 the weak hypercharge ) . the electric charge @xmath89 , where @xmath90 is the third projection of isospin .
@xmath91 for a doublet of quarks , @xmath92 for a doublet of leptons . as for the right - handed spinors , they are isosinglets , and hence @xmath93 thus
, parity violation and charge conjugation violation were incorporated into the foundation of electroweak theory . out of the four fields (
three of @xmath94 and one of @xmath95 , usually denoted by @xmath96 and @xmath97 , respectively ) only two directly correspond to the observed vector bosons : @xmath67 and @xmath68 .
the @xmath83 boson and photon are represented by two orthogonal superpositions of @xmath98 and @xmath97 : @xmath99 where @xmath100 , @xmath101 , while the weak angle @xmath102 is a free parameter of electroweak theory .
the value of @xmath102 is determined from experimental data on @xmath0 boson coupling to neutral current .
the @xmath0-charge " , characterizing the coupling of @xmath0 boson to a spinor with definite helicity is given by @xmath103 where the values of the corresponding charges at @xmath2-scale , while @xmath35 , @xmath72 , @xmath73 refer to values at vanishing momentum transfer .
the same refers to @xmath22 , @xmath104 , @xmath105 and @xmath43 , @xmath106 , @xmath107 ( see below eqs .
( [ 13 ] ) - ( [ 18 ] ) ) . ]
@xmath108 note that the @xmath0-charge " is different for the right- and left - handed spinors with the same value of @xmath109 because they have different values of @xmath90 .
the coupling constant of @xmath33-bosons is also expressed in terms of @xmath110 and @xmath102 : @xmath111 the theory described above has many nice features , the most important of which is its renormalizability .
but at first sight it looks absolutely useless : all fermions and bosons in it are massless .
this drawback can not be fixed by simply adding mass terms to the lagrangian .
the mass terms of fermions would contain both @xmath112 and @xmath113 and thus explicitly break the isotopic invariance and hence renormalizability .
the gauge invariance would also be broken by the mass terms of the vector bosons .
all this would result in divergences of the type @xmath114 , @xmath115 , etc .
the way out of this trap is the so called higgs mechanism @xcite . in the framework of the minimal standard model
the problem of mass is solved by postulating the existence of the doublet @xmath116 and corresponding antidoublet @xmath117 of spinless particles .
these four bosons differ from all other particles by the form of their self - interaction , the energy of which is minimal when the neutral field @xmath118 has a nonvanishing vacuum expectation value .
the isospin of the higgs doublet is @xmath119 , its hypercharge is @xmath120 .
thus , it interacts with all four gauge bosons .
in particular , it has quartic terms @xmath121 , @xmath122 , which give masses to the vector bosons when @xmath123 acquires its vacuum expectation value ( _ vev _ ) @xmath124 : @xmath125 the magnitude of @xmath124 can easily be derived from that of the four - fermion interaction constant @xmath1 in muon decay : @xmath126 in the born approximation of electroweak theory this four - fermion interaction is caused by an exchange of a virtual @xmath33-boson ( see figure 1 ) .
hence holds , see below , eqs .
( [ 14 ] ) - ( [ 25 ] ) . ]
@xmath127 such mechanism of appearance of masses of @xmath33 and @xmath0 bosons is called spontaneous symmetry breaking .
it preserves renormalizability @xcite .
( as a hint , one can use the symmetrical form of lagrangian by not specifying the _ vev _ @xmath124 . )
the fermion masses can be introduced also without explicitly breaking the gauge symmetry . in this case
the mass arises from an isotopically invariant term @xmath128 , where @xmath129 is called yukawa coupling .
the mass of a fermion @xmath130 .
there is a separate yukawa coupling for each of the known fermions .
their largely varying values are at present free parameters of the theory and await for the understanding of this hierarchy .
let us return for a moment to the vector bosons .
a massless vector boson ( e.g. photon ) has two spin degrees of freedom two helicity states .
a massive vector boson has three spin degrees of freedom corresponding , say , projections @xmath131 on its momentum . under spontaneous symmetry breaking three out of four spinless states ,
@xmath132 become third components of the massive vector bosons . thus , in the minimal standard model there must exist only one extra particle : a neutral higgs scalar boson , or simply , higgs , representing a quantum of excitation of field @xmath133 over its _
the discovery of this particle is crucial for testing the correctness of msm .
the first successful test of electroweak theory was provided by the discovery of neutral currents in the interaction of neutrinos with nucleons @xcite .
further study of this deep inelastic scattering ( dis ) allowed to extract the rough value of the @xmath134 : @xmath135 and thus to predict the values of @xmath136 gev and @xmath137 gev , which served as leading lights for the discovery of these particles .
a few other neutral current interactions have been discovered and studied : neutrino - electron scattering @xcite , parity violating electron - nucleon scattering at high energies @xcite , parity violation in atoms @xcite .
all of them turned out to be in agreement with electroweak theory .
a predominant part of theoretical work on electroweak corrections prior to the discovery of the @xmath33 and @xmath0 bosons was devoted to calculating the neutrino - electron @xcite and especially nucleon - electron @xcite interaction cross sections . after the discovery of @xmath33 and @xmath0 bosons it became evident that the next level in the study of electroweak physics must consist in precision measurements of production and decays of @xmath0 bosons in order to test the electroweak radiative correction . for such measurements
special electron - positron colliders slc ( at slac ) and lep - i ( at cern ) were constructed and started to operate in the fall of 1989 .
slc had one intersection point of colliding beams and hence one detector ( sld ) ; lep - i had four intersection points and four detectors : aleph , delphi , l3 , opal . in connection with the construction of lep and slc , a number of teams of theorists carried out detailed calculations of the required radiative corrections .
these calculations were discussed and compared at special workshops and meetings .
the result of this work was the publication of two so - called ` cern yellow reports ' @xcite , @xcite , which , together with the ` yellow report ' @xcite , became the ` must ' books for experimentalists and theoreticians studying the @xmath0 boson .
the book @xcite ( which should be published in 1999 ) summarizes results of theoretical studies .
more than 2.000 experimentalists and engineers and hundreds of theorists participated in this unique collective quest for truth !
the sum of energies of @xmath138 was chosen to be equal to the @xmath0 boson mass .
lep - i was terminated in the fall of 1995 in order to give place to lep - ii , which will operate in the same tunnel till 2001 with maximal energy @xmath139 gev .
slc continued at energy close to @xmath140 gev .
the reactions , which has been studied at lep - i and slc may be presented in the form ( see figure 2 ) : @xmath141 where @xmath142 about 20,000,000 @xmath0 bosons have been detected at lep - i and 550 thousands at slc ( but here electrons are polarized , which compensates the lower number of events ) .
experimental data from all five detectors were summarized and analyzed by the lep electroweak working group and the sld heavy flavour and electroweak groups which prepared a special report a combination of preliminary electroweak measurements and constraints on the standard model " @xcite .
these data were analyzed in @xcite by using zfitter code ( see below section 6.1 ) and independently by j.erler and p.langacker @xcite .
fantastic precision has been reached in the measurement of the @xmath0 boson mass and width @xcite : @xmath143 of special interest is the measurement of the width of invisible decays of @xmath0 : @xmath144 by comparing this number with theoretical predictions for neutrino decays it was established that the number of neutrinos which interact with @xmath0 boson is three ( @xmath145 ) .
this is a result of fundamental importance .
it means that there exist only 3 standard families ( or generations ) of leptons and quarks . ) with the data on @xmath146 @xcite and @xmath147 @xcite scattering allowed to establish that @xmath148 , @xmath149 and @xmath150 have equal values of couplings with @xmath0 boson @xcite . ]
extra families ( if they exist ) must have either very heavy neutrinos ( @xmath151 ) , or no neutrinos at all .
this review is devoted to the description of the theory of electroweak radiative corrections in @xmath0 boson decays and to their comparison with experimental data @xcite .
our approach to the theory of electroweak corrections differs somewhat from that used in @xcite-@xcite .
we believe that it is simpler and more transparent ( see section 6.1 ) . in section 2
we introduce the basic input parameters of the electroweak born approximation . in section 3
we present phenomenological formulas for amplitudes , decay widths , and asymmetries of numerous decay channels of @xmath0 bosons .
the main subject of our review is the calculation of one - electroweak loop radiative corrections to the born approximation . in section 4
they are calculated to the hadronless decays and the mass of the @xmath33 boson , while in section 5 to the hadronic decays . in section 6 the results of the one - electroweak loop calculations
are compared with the experimental data .
section 7 gives a sketch of two - electroweak loop corrections and of their influence on the fit of experimental data .
section 8 discusses possible manifestations of new physics ( extra generations of fermions and supersymmetry ) .
section 9 contains conclusions . in order to make easier the reading of the main text , technical details and derivations
are collected in appendices .
the first step in the theoretical analysis is to separate genuinely electroweak effects from purely electromagnetic ones , such as real photons emitted by initial and final particles in reaction ( [ 10 ] ) and virtual photons emitted and absorbed by them .
the electroweak quantities extracted in this way are called sometimes @xcite pseudoobservables , but for the sake of brevity we will refer to them as observables . a key role among purely electromagnetic effects is played by a phenomenon which is called the running of electromagnetic coupling constant " @xmath152 .
the dependence of the electric charge on the square of the four - momentum transfer @xmath153 is caused by the photon polarization of vacuum , i.e. by loops of charged leptons and quarks ( hadrons ) ( see figure 3(a ) ) .
as is well known ( see e.g. @xcite ) @xmath154^{-1 } \;\ ; . \label{13}\ ] ] it has a very high accuracy and is very important in the theory of electromagnetic processes at low energies . as for electroweak processes in general and
@xmath0 decays in particular , they are determined by @xcite @xmath155^{-1 } \;\ ; , \label{14}\ ] ] the accuracy of which is much worse .
it is convenient to denote @xmath156 where @xmath157 ( for the value of @xmath158 see ref .
@xcite , while for the value of @xmath159 see @xcite ) .
it is obvious that the uncertainty of @xmath160 and hence of @xmath22 stems from that of hadronic contribution @xmath159 .
while @xmath152 is running electromagnetically fast , @xmath25 and @xmath24 are crawling " electroweakly slow for @xmath161 : @xmath162 @xmath163 the small differences @xmath164 and @xmath165 are caused by electroweak radiative corrections .
therefore one could and should neglect them when defining the electroweak born approximation .
( we used this recipe ( @xmath166 ) when deriving the relation ( [ 9 ] ) between @xmath1 and @xmath124 . ) the theoretical analysis of electroweak effects in this article is based on three most accurately known parameters @xmath1 , @xmath22 ( eq.([14 ] ) ) and @xmath2 ( eq.([11 ] ) ) .
@xmath167 this value of @xmath1 @xcite is extracted from the muon life - time after taking into account the purely electromagnetic corrections ( including bremsstrahlung ) and kinematical factors @xcite : @xmath168 \;\ ; , \label{20}\ ] ] where @xmath169 and @xmath170 now we are ready to express the weak angle @xmath102 in terms of @xmath1 , @xmath22 and @xmath2 . starting from eqs.([9 ] ) , ( [ 5 ] ) and ( [ 6 ] ) , we get in the electroweak born approximation : @xmath171 wherefrom : @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 the angle @xmath102 was introduced in mid-1980s @xcite .
however its consistent use began only in the 1990s @xcite . using @xmath102 automatically takes into account the running of @xmath152 and makes it possible to concentrate on genuinely electroweak corrections as will be demonstrated below .
( in this review we consistently use @xmath2 as defined by ewwg in accord with our eq .
( [ e.5 ] ) .
note that a different definition of the @xmath0 boson mass @xmath177 is known in the literature , related to a different parametrization of the shape of the @xmath0 boson peak @xcite . )
the introduction of the born approximation described above differs from the traditional approach in which @xmath178 is treated as the largest electroweak correction , masses @xmath179 and @xmath2 are handled on an equal footing , and the angle @xmath180 , defined by @xmath181 is considered as one of the basic parameters of the theory .
( note that the experimental accuracy of @xmath180 is much worse than that of @xmath102 . ) after discussing our approach and its main parameters we are prepared to consider various decays of @xmath0 bosons .
phenomenologically the amplitude of the @xmath0 boson decay into a fermion - antifermion pair @xmath182 can be presented in the form : @xmath183 where coefficient @xmath184 is given by eq.([22 ] ) .
boson couplings are diagonal in flavor unlike that of @xmath33 boson , where cabibbo - kobayashi - maskawa mixing matrix @xcite should be accounted for in the case of couplings with quarks . ] in the case of neutrino decay channel there is no final state interaction or bremsstrahlung .
therefore the width into any pair of neutrinos is given by @xmath185 where neutrino masses are assumed to be negligible , and @xmath186 is the so - called standard width : @xmath187 for decays to any of the pairs of charged leptons @xmath188 we have : @xmath189 \;\ ; .
\label{30}\ ] ] the qed radiator " ( @xmath190 ) is due to bremsstrahlung of real photons and emission and absorption of virtual photons by @xmath191 and @xmath192 .
note that it is expressed not through @xmath43 , but through @xmath22 . for the decays to any of the five pairs of quarks
@xmath193 we have @xmath194 \;\ ; . \label{31}\ ] ] here an extra factor of 3 in comparison with leptons takes into account the three colours of each quark .
the radiators @xmath26 and @xmath27 contain contributions from the final state gluons and photons . in the crudest approximation @xmath195 where @xmath196 is
the qcd running coupling constant : @xmath197 ( for additional details on @xmath198 and radiators see appendix e. ) the full hadron width ( to the accuracy of very small corrections , see sections 5 and 7 ) is the sum of widths of five quark channels : @xmath199 the total width of the @xmath0 boson : @xmath200 the cross section of annihilation of @xmath201 into hadrons at the @xmath0 peak is given by the breit - wigner formula @xmath202 finally the following notations for the ratio of partial widths are widely used : @xmath203 ( note that @xmath204 in the numerator of @xmath205 refers to a single charged lepton channel , whose lepton mass is neglected . )
parity violating interference of @xmath206 and @xmath207 leads to a number of effects : forward - backward asymmetries @xmath208 , longitudinal polarization of @xmath209-lepton @xmath210 , dependence of the total cross - section at @xmath0-peak on the longitudinal polarization of the initial electron beam @xmath211 , etc .
let us define for the channels of charged lepton and light quark ( @xmath212 ) whose mass may be neglected the quantity @xmath213 for @xmath214 : @xmath215 where @xmath216 is the velocity of the @xmath217-quark : @xmath218 here @xmath219 is the value of the running mass of the @xmath217-quark at scale @xmath2 calculated in @xmath220 scheme @xcite .
the forward - backward charge asymmetry in the decay to @xmath182 equals : @xmath221 where @xmath222 the number of events with @xmath72 going into forward ( backward ) hemisphere ; @xmath223 refers to the creation of @xmath0 boson in @xmath224-annihilation , while @xmath225 refers to its decay in @xmath182 .
the longitudinal polarization of the @xmath209-lepton in the decay @xmath226 is @xmath227 .
if , however , the polarization is measured as a function of the angle @xmath102 between the momentum of a @xmath228 and the direction of the electron beam , this allows the determination of not only @xmath229 , but @xmath223 as well : @xmath230 the polarization @xmath210 is found from @xmath231 by separately integrating the numerator and the denominator in eq.([42 ] ) over the total solid angle .
the relative difference between total cross - section at the @xmath0-peak for the left- and right - polarized electrons that collide with non - polarized positrons ( measured at the slc collider ) is @xmath232 the measurement of parity violating effects allows one to determine experimentally the ratios @xmath233 , while the measurements of leptonic and hadronic widths allow to find @xmath206 and @xmath198 . * table 1 * [ cols="<,>,>,>",options="header " , ] _ results of fit along the valley of minimum of @xmath234 for fixed value @xmath235 gev and @xmath236 gev . _
the main lesson of this subsection is the following .
the fit of the precision data on electroweak observables ( i.e. of @xmath0-boson decay parameters from lep and slc and the values of @xmath179 and the @xmath5 from tevatron ) in the framework of susy extension of the sm assuming small value of @xmath237 , the absence of @xmath238 mixing and @xmath239 gev leads to the growth of @xmath234 value .
for heavy squarks the susy sector of the theory decouples from low - energy observables and the results of standard model fit are reproduced . on the other hand even for light sbottom and for small mass of one of two stops
, one can find the values of @xmath240 mixing where supersymmetric corrections appear to be small and not excluded by experimental data . in this case
the quality of the fit ( i.e. the value of @xmath234 ) is almost the same as in the standard model .
the comparison of lep - i and slc precision data on @xmath0 boson decays with calculations based on the minimal standard model has confirmed the predictive power of the latter : 1 .
it was proved that there exist only three generations of quarks and leptons with light neutrinos .
the @xmath0 boson couplings of quarks , charged leptons and neutrinos are in accord with the theory .
3 . from the analysis of the radiative corrections
the mass of the top quark had been correctly predicted before this particle was discovered at tevatron .
4 . all electroweak observables ( except for the mass of the higgs ) are perfectly fitted by one loop electroweak corrections ( with virtual and `` free '' gluons being taken into account ) .
the dependence of the radiative corrections on the mass of the higgs is feeble when higgs is heavy .
therefore the value of the higgs mass extracted from lep - i and slc data has rather large error bars .
within one standard deviation the central fitted value of the higgs mass becomes smaller when the leading two - electroweak - loop corrections are taken into account . in this case
it is close to 90 gev
its direct lower limit from the lep - ii search .
however the non - leading two loop corrections may change this result .
calculation of these corrections is a challenge to theorists .
better understanding of systematic discrepancies between various asymmetries in @xmath0-decays is a challenge to experimentalists .
one of the main conclusions of the one - electroweak - loop case is that in this case the value of the leading and non - leading corrections are comparable and even may cancel each other ( in the case of leptonic parity violating parameter @xmath241 ) .
the remarkable agreement between the minimal standard model and experimental data on @xmath0-decays puts strong limits on the hypothetical `` new physics '' , such as extra generations of heavy quarks and leptons and/or properties of supersymmetric particles .
the discovery of the higgs , more precise measurements of the mass of the @xmath33 boson at lep - ii and tevatron , and more accurate prediction of the value of electric charge at the scale of @xmath2 , may substantially improve the sensitivity of the @xmath0-decay data to the possible manifestations of the new physics .
l.b.o . is grateful to alexander von humboldt stiftung for an award , a.n.r .
is grateful to + cppm / in2p3/cnrs for supporting this work .
the grants of rfbr no . 96 - 15 - 96578 , + 98 - 07 - 90076 , 98 - 02 - 17372 and 98 - 0217453 and of intas - rfbr 95 - 05678 are acknowledged .
the authors would like to thank d.bardin , a.olshevski and p.renton for correspondence .
we are grateful to our collaborators on various topics on @xmath0-boson decays .
( 0,150)(160,20 ) ( 80,165)(0,0)introduce bare coupling constants in the framework of msm ( 80,160)(0,0)(@xmath246 ) , bare masses ( @xmath247 ) ( 80,155)(0,0)and the vacuum expectation value ( _ vev _ ) of the higgs field , @xmath124 . ( 80,145)(0,1)5 ( 0,105)(160,20 ) ( 80,120)(0,0)express one - electroweak - loop corrections to all electroweak ( 80,115)(0,0)observables in terms of @xmath251 , and hence , in terms of ( 80,110)(0,0)of @xmath252 .
check cancellation of the terms @xmath253 .
( 80,100)(0,1)5 ( 0,75)(160,25 ) ( 80,95)(0,0)introduce gluon corrections to quark loops and qed ( and qcd ) ( 80,90)(0,0)final state interactions , as well as the two - electroweak - loop ( 80,85)(0,0)corrections , calculated by other authors , in terms of ( 80,80)(0,0)@xmath254 .
( 80,70)(0,1)5 ( 0,35)(160,35 ) ( 80,65)(0,0)compare the predictions of the successive approximations ( 80,60)(0,0)(born , one loop , two loops ) with the experimental data ( 80,55)(0,0)on @xmath0-decays .
perform the global fits for @xmath255 ( 80,50)(0,0)for one and two loops .
derive theoretical predictions ( 80,45)(0,0)of the central values for all electroweak observables ( 80,40)(0,0)and of the corresponding uncertainties ( ` errors ' ) .
( 80,30)(0,1)5 integrals corresponding to diagrams with loops formally diverge and thus need regularization .
note that there does not exist yet a consistent regularization of electroweak theory in all loops .
a dimensional regularization can be used in the first several loops ; this corresponds to a transition to a @xmath39-dimensional spacetime in which the following finite expression is assigned to the diverging integrals : @xmath257 where @xmath258 is a parameter with mass dimension , introduced to conserve the dimension of the original integral .
this formula holds in the range of convergence of the integral . in the range of divergence , a formal expression ( [ a.1 ] ) is interpreted as the analytical continuation .
obviously , the integral allows a shift in integration variable in the convergence range as well .
therefore , a shift @xmath259 for arbitrary @xmath39 can also be done in ( [ a.1 ] ) .
this factor is decisive in proving the gauge invariance of dimensional regularization . as for the dimensionality of spinors ,
different approaches can be chosen in the continuation to the @xmath39-dimensional space .
one possibility is to assume that the @xmath71 matrices are @xmath266 matrices , so that @xmath267 the @xmath39-dimensional regularization creates difficulties when one has to define the absolutely antisymmetric tensor and ( or ) @xmath268 matrix . for calculations in several first loops , a formal definition of @xmath268
, @xmath269 @xmath270 does not lead to contradictions .
if we eliminate bare quantities and express some physical observables in terms of other physical observables , then all pole terms cancel out .
the general property of renormalizability guarantees this cancellation .
( we have verified this cancellation directly in @xcite . ) the five steps " described in section 4.6 are based on this renormalization procedure . in order to avoid divergences in intermediate expressions
, one can agree to subtract from each feynman integral the pole terms @xmath272 , since they will cancel out anyway in the final expressions .
depending on which constant terms ( in addition to pole terms ) are subtracted from the diagrams , different subtraction schemes arise : the @xmath220 scheme corresponds to subtracting the universal combination @xmath273 here the fine structure constant @xmath275 is a physical quantity .
it can be measured as a residue of the coulomb pole @xmath276 in the scattering amplitude of charged particles . as for the running coupling constant @xmath152 ,
it can be measured from the scattering of particles with large masses @xmath36 at low momentum transfer : @xmath277 . in the standard model we have the @xmath0-boson , and the contribution of the photon
can not be identified unambiguously if @xmath278 .
therefore , the definition of the running constant @xmath152 becomes dependent on convention and on details of calculations .
one - loop expression for the self - energy of the photon can be rewritten as @xcite : @xmath283 - \\ & - & ( \alpha/4\pi ) [ 3s\delta_w + ( 3s + 4m^2_w ) f ( s , m_w , m_w)]\;\ ; , \label{b.4}\end{aligned}\ ] ] where @xmath284 , the subscript @xmath72 denotes fermions , the sum @xmath285 runs through lepton and quark flavors , and @xmath286 is the number of colors . the contribution of fermions to @xmath287 is independent of gauge .
the last term in ( [ b.4 ] ) refers to the gauge - dependent contribution of @xmath33-bosons ; the t hooft feynman gauge was used in equation ( [ b.4 ] ) .
the singular term @xmath288 is : @xmath289 where @xmath290 ( @xmath39 is the variable dimension of spacetime , @xmath291 ) , @xmath292 is the euler constant and @xmath258 is an arbitrary parameter .
both @xmath293 and @xmath258 vanish in relations between observables .
the function @xmath294 is defined by the contribution to self - energy of a scalar particle at @xmath295 , owing to a loop with two scalar particles ( with masses @xmath296 and @xmath297 ) and with the coupling constant equal to unity : @xmath298 the function @xmath299 is normalized in such a way that it vanishes at @xmath300 , which corresponds to subtracting the self - energy at @xmath300 : @xmath301 the following formula holds for @xmath302 : @xmath303 \;\ ; , & 4\tau > 1 \;\ ; , \\ 2\left [ 1-\sqrt{1
- 4\tau}\log \frac{1+\sqrt{1 - 4\tau}}{\sqrt{4\tau}}\right ] \;\ ; , & 4\tau < 1
\;\ ; , \end{array } \right .
\label{b.77}\end{aligned}\ ] ] where @xmath304 . as a result
we obtain @xmath309 - \nonumber \\ & - & \frac{\alpha}{4\pi } [ 3\delta_w + ( 3 + 4c^2 ) f(c^2)]\;\ ; , \label{b.11}\end{aligned}\ ] ] where @xmath310 , and @xmath311 @xmath312 + \left [ \left ( \frac{3}{4 } + c^2 \right )
f(c^2 ) - \frac{1}{6 } \right ] \right\}\;\;. \label{b.13}\end{aligned}\ ] ] therefore , @xmath313 is found as a sum of four terms , @xmath314 loops with top quarks give : @xmath317 where we have used that @xmath318 gev . note
that @xmath319 is negligible and has the antiscreening sign ( the screening of the @xmath13-quark loops in qed begins at @xmath320 , while in our case @xmath321 .
finally , the @xmath33 loop gives @xmath322 = 0.00050\;\;. \label{b.17}\ ] ] the value of @xmath323 depends on gauge @xcite ; here we give the result of calculations in the t hooft feynman gauge .
traditionally , the definition of @xmath22 takes into account the contributions of leptons and five light quarks only . the terms @xmath319 and @xmath323 are taken into account in the electroweak radiative corrections . in our approach
, these terms give the corrections @xmath324 . in the same way the loops of not yet discovered heavy new particles ( new physics " )
should be accounted for .
the effect of ` running ' of electromagnetic coupling constants @xmath152 ( logarithmic dependence of the effective charge on momentum transfer @xmath153 ) is known for more than four decades ) @xcite .
in contrast to @xmath152 , the effective constants of @xmath33 and @xmath0 bosons @xmath24 and @xmath25 in the region @xmath325 ` crawl ' rather than run @xcite . if we define the effective gauge coupling constant @xmath326 in terms of the bare charge @xmath327 and the bare mass @xmath328 , and sum up the geometric series with the self - energy @xmath329 inserted in the gauge boson propagator , this gives the expression @xmath330 here @xmath36 is the physical mass , and @xmath329 contains the contribution of fermions only , since loops with @xmath33 , @xmath0 and @xmath331 bosons do not contain large logarithms in the region @xmath332 .
if @xmath337 , equations ( [ c.2 ] ) and ( [ c.3 ] ) define the logarithmic running of charges owing to leptons and quarks , and @xmath338 and @xmath339 represent the contribution of fermions to the first coefficient of the gell - mann low function : @xmath340 + n_d 3[1+(-1 + \frac{4}{3}s^2)^2 ] + n_l[2+(1+(1 - 4s^2)^2)]\ } \;\ ; , \nonumber \\ b_w & = & \frac{1}{16\pi } \ [ 6n_q + 2n_l ] \;\ ; , \label{c.5}\end{aligned}\ ] ] where @xmath341 are the numbers of quarks and leptons with masses that are considerably lower than @xmath342 . for @xmath161 ,
the numerical values of the coefficients @xmath343 are @xcite : @xmath344 @xmath345 the massive propagator @xmath346 in ( [ c.1 ] ) greatly suppresses the running of @xmath24 and @xmath25 .
thus , according to ( [ c.2 ] ) and ( [ c.3 ] ) , the constant @xmath25 grows by 0.85% from @xmath347 to @xmath348 , @xmath349 @xmath350 and the constant @xmath24 grows by 0.95% , @xmath351 @xmath352 while the electromagnetic constant @xmath152 increases by 6.34% : @xmath353 the bare quantities are marked by the subscript ` 0 ' . in the electroweak theory ,
three bare charges @xmath356 , @xmath357 and @xmath358 that describe the interactions of @xmath71 , @xmath0 and @xmath33 are related by a single constraint : @xmath359 the fine structure constant @xmath361 is related to the bare charge @xmath356 by the formula @xmath362 where @xmath363 . in the feynman gauge @xmath364 , where the dimension of spacetime is @xmath365 . in the unitary gauge @xmath366 .
the simplest way to verify the presence of the term @xmath367 is by considering the interaction of a photon with the right - handed electron @xmath368 .
note that in this case there are no weak vertex corrections due to the @xmath33-boson exchange .
( note also that the left - handed neutrino remains neutral even when loop corrections are taken into account , since the diagram with the @xmath369 interaction is compensated for by the vertex diagram with the @xmath33 exchange ) .
the relation between @xmath370 and @xmath371 has the following form : @xmath372 \;\ ; , \label{e.4}\ ] ] where @xmath373 , @xmath374 , while @xmath375 mean that contributions of @xmath33-boson and @xmath13-quark are not accounted for .
it is convenient to introduce in ( [ e.4 ] ) explicit expression for @xmath376 : @xmath377 \;\ ; , \label{e.44}\ ] ] where in accordance with eq.([b.3 ] ) @xmath378 finally , the third basic equation is @xmath385 \;\ ; , \label{e.7}\ ] ] where @xmath386 comes from the propagator of @xmath33 , while @xmath39 is the contribution of the box and the vertex diagrams ( minus the electromagnetic corrections to the four - fermion interaction ) to the muon decay amplitude .
according to sirlin @xcite , @xmath387 where @xmath388 the next step is to express @xmath17 , @xmath391 and @xmath392 through @xmath86 , @xmath393 and loop corrections .
let us start with @xmath17 . from ( [ e.6 ] ) and ( [ e.5 ] )
we get : @xmath394 \;\ ; . \label{e.12}\ ] ] substituting @xmath395 given by ( [ e.11 ] ) we obtain : @xmath396+\pi_w(m_w^2)-\pi_w(0 ) - \right .
\nonumber \\ & - & \left .
\pi_{\gamma}(m_z^2 ) -2\frac{s}{c}\pi_{\gamma z}(0 ) -d-\delta\alpha_w -\delta\alpha_t \right ) \;\ ; . \label{e.13}\end{aligned}\ ] ] in order to obtain expression for @xmath391 we should recall that it is proportional to @xmath357 and take into account the @xmath0 boson wave function renormalization and @xmath397 vertex loop correction : @xmath398 \;\ ; , \label{e.14}\ ] ] where @xmath399 originates from the vertex correction .
the last quantity is the ratio @xmath19 .
one - loop corrections come from @xmath400 ( eq.([e.11 ] ) ) , from vector and axial @xmath397 vertices and from @xmath401 transition which contributes to @xmath392 only : @xmath402 -4csf_v + 4csf_a(1 - 4s^2)- \nonumber \\ & - & 4cs \pi_{\gamma z}(m_z^2 ) \;\ ; . \label{e.15}\end{aligned}\ ] ] formulas ( [ e.13 ] ) , ( [ e.14 ] ) and ( [ e.15 ] ) are derived in this appendix according to the five steps " procedure described in section 4.6 .
they describe finite one - loop corrections to hadronless observables .
it is easy to evaluate the contribution of @xmath13-quark to physical observables in the approximation @xmath403 . in this approximation @xmath404 , @xmath405 , @xmath406 and from eq.([e.13])-([e.15 ] ) we get : @xmath407 @xmath408 @xmath409 the corrections proportional to @xmath410 were first pointed out by veltman @xcite , who emphasized the appearance of such corrections for a large difference @xmath411 which violates the isotopic symmetry . in this review
the coefficients in front of the factors @xmath13 in equations ( [ e.18])-([e.20 ] ) are used as coefficients for normalized radiative corrections @xmath14 ( see sections 4.2 and 4.3 ) .
for decays to light quarks @xmath412 , we neglect the quark masses and take into account the gluon exchanges in the final state up to terms @xmath413 @xcite - @xcite , and also one - photon exchange in the final state and the interference of the photon and the gluon exchanges @xcite .
these corrections are slightly different for the vector and the axial channels .
for decays to quarks we have @xmath414\gamma_0 \label{f.1}\ ] ] where the factors @xmath415 are responsible for the interaction in the final state .
according to @xcite - @xcite : @xmath416(\frac{\hat{\alpha}_s } { \pi})^2 - \nonumber \\ & - & 12.77(\frac{\hat{\alpha}_s}{\pi})^3 + 12 \frac{\hat{m}^2_q}{m^2_z } \frac{\hat{\alpha}_s}{\pi } \delta_{vm } \label{f.2}\end{aligned}\ ] ] @xmath417 -\ ] ] @xmath418 where @xmath419 is the running quark mass ( see below ) , @xmath420 @xmath421 @xmath422 @xmath423 @xmath424 terms of the order of @xmath425 caused by the diagrams with three gluons in intermediate state were calculated in @xcite .
for @xmath27 they are numerically very small , @xmath426 ; for this reason , we dropped them from formula ( [ f.2 ] ) . for the @xmath427 decay ,
the @xmath217-quark mass is not negligible ; it reduces @xmath428 by about 1 mev ( @xmath429 ) . the gluon corrections result in a replacement of the pole mass @xmath430 gev by the running mass at @xmath431 .
we express @xmath432 in terms of @xmath6 , @xmath433 and @xmath434 using standard two - loop equations in the @xmath220 scheme ( see @xcite ) . for the @xmath435 decay ,
the running mass @xmath436 is of the order of @xmath437 gev and the corresponding contribution to @xmath438 is of the order of @xmath439 mev .
we have included this tiny term in the leptop code , since it is taken into account in other codes ( see , for example , @xcite ) .
we need to remark in connection with @xmath438 that the term @xmath440 , given by equation ( [ f.7 ] ) , contains interference terms @xmath441 .
these terms are related to three types of final states : one quark pair , a quark pair and a gluon , two quark pairs .
this last contribution comes to about 5% of @xmath442 and is below the currently achievable experimental accuracy .
nevertheless , in principle these terms require special consideration , especially if these quark pairs are of different flavors , for example , @xmath443 .
such mixed quark pairs must be discussed separately . in @xcite
when deriving equations for physical observables we systematically took into account corrections which contained first power of polarization operators and neglected terms @xmath446 .
this procedure was correct at one loop , but since @xmath447 contain terms of the order of @xmath448 we evidently lost @xmath16 terms . to restore them
let us repeat the procedure implemented in @xcite this time taking squares of @xmath449 and @xmath450 ( reducible two - loop diagrams ) into account .
our starting point are three basic equations for quantities @xmath2 , @xmath1 and @xmath451 . since terms @xmath452
are absent in @xmath453 , @xmath454 and @xmath39 functions , we will not consider these functions in this section .
equation for @xmath2 is the same as in appendix d , eqs.([e.5 ] ) , ( [ e.2 ] ) : @xmath455 \;\ ; , \label{4.1}\ ] ] while for @xmath1 we have : @xmath456 } = \frac{1}{\sqrt{2}\eta^2(1-\pi_w(0 ) ] } \;\ ; , \label{4.2}\ ] ] and we keep @xmath457 in the denominator to avoid loosing of the @xmath458 term ( compare with eq.([e.7 ] ) ) . from these two equations
we get : @xmath459 where we use equality @xmath460 which is valid for the leading term @xmath452 . considering @xmath464 as a small parameter and solving equation ( [ 4.5 ] ) perturbatively ,
we get : @xmath465 \;\ ; , \label{4.7}\ ] ] where we keep terms linear and quadratic in @xmath464 . for definitions of @xmath86 and @xmath393
see eq.([23 ] ) .
our next step should be the calculation of the @xmath466 corrections to the functions @xmath14 . but first let us discuss the expression for @xmath464 as given by eq.([4.6 ] ) which contains factor @xmath467 in denominator . at one loop , corrections proportional to @xmath464 appear in physical observables .
they should be carefully calculated in order not to induce extra @xmath16 terms .
fortunately this can be done straightforwardly , using the following chains of equalities : @xmath468 = 4\sqrt{2}g_{\mu}(1-\pi_w(0 ) ) [ \sigma_z(0)/f_0 ^ 2 ] \;\ ; , \label{4.9}\ ] ] @xmath469 = 4\sqrt{2}g_{\mu}(1-\pi_w(0 ) ) [ \sigma_w(0)/g_0 ^ 2 ] \;\ ; , \label{4.10}\ ] ] where expressions in square brackets contain self - energies coupling constants ( @xmath470 and @xmath471 respectively ) and eq.([4.2 ] ) is used to express @xmath124 through @xmath1 . substituting eq.([4.9 ] ) and eq.([4.10 ] ) into eq.([4.6 ] ) we get : @xmath472 = 4\sqrt{2}g_{\mu}m_z^2 \frac{[\sigma_z(0)/f_0 ^ 2 - \sigma_w(0)/g_0 ^ 2 ] } { m_z^2 } = \nonumber \\ & = & \frac{3\bar{\alpha}}{16\pi s^2 c^2}(\frac{m_t}{m_z})^2 \;\ ; . \label{4.11}\end{aligned}\ ] ] now everything is prepared for the calculation of @xmath466 corrections to physical observables .
let us start from the @xmath33-boson mass .
for the ratio of the squares of vector boson masses we have : @xmath473 the next step is the correction to axial coupling of @xmath0 boson to charged leptons .
axial coupling is proportional to @xmath357 , and from eq.([4.3 ] ) and eq.([4.14 ] ) we immediately obtain : @xmath476 the final formula for the correction to @xmath477 is presented in section 7.1 . for the degenerate case the contribution of additional quark and lepton to @xmath479 are given by ( @xcite ) : @xmath480 + 2 s^2 [ ( 1-l / c^2 ) f(l / c^2)- \right .
\nonumber \\ & -&\left.(1 + 2l)f(l)]+ 4s^4 ( q^2_u + q^2_d)[(1 + 2l)f(l ) -\frac{1}{3}]\right\ } \label{1.3}\end{aligned}\ ] ] @xmath481 +
[ 4 s^4(q_u^2 + q_d^2 ) - \right.\nonumber \\ & -&\left.2 s^2][2l+1 - 12l^2 f(l)]\right\}/(1 - 4l ) \;\ ; , \label{1.4}\end{aligned}\ ] ] @xmath482\right\ } \;\ ; , \label{1.5}\ ] ] where @xmath483 , @xmath484 , @xmath485 for quark doublet ; @xmath486 , @xmath487 , @xmath488 for lepton doublet ; @xmath489 and the function @xmath490 is defined in appendix b , eqs.(b8 ) , ( b.10 ) . for different
up- and down- quark ( and lepton ) masses analytical expressions for @xmath479 are given in by @xmath491 + \nonumber \\ & + & \frac{8}{9}[(1-u)f(u)+(1-d)f(d)-\frac{2}{3}]+\frac{4}{3}\frac{s^2}{c^2 } [ u+d-\frac{2ud}{u - d}\log\frac{u}{d } ] + \nonumber \\ & + & \frac{8}{9}(1-\frac{s^2}{c^2})[\frac{u - d}{2}\log\frac{u}{d}+(u+d)+(c^2 -\frac{u+d}{2})\frac{u+d}{u - d}\log\frac{u}{d}-\frac{4}{3}c^2 - \nonumber \\ & -&(2c^2 -u - d-\frac{(u - d)^2}{c^2})f(m_w^2 , m_u^2 , m_d^2 ) ] \;\ ; ; \label{1.9}\end{aligned}\ ] ] @xmath492 + \right .
\nonumber \\ & + & \left .
3[u+d -\frac{2ud}{u - d}\log\frac{u}{d } -f'(u ) -f'(d)]\right\ } \;\ ; ; \label{1.10}\end{aligned}\ ] ] @xmath493 + \frac{64}{27}s^2 c^2 [ ( 1 + 2u)f(u)+(1 + 2d)f(d ) -\frac{2}{3 } ] \;\ ; , \label{1.11}\ ] ] where @xmath494 is the number of generations and @xmath495 , @xmath496 , @xmath497 , @xmath498 ; @xmath499 and @xmath500 is defined in appendix b , eq.(b.6 ) . here
we will present formulas which connect our functions @xmath14 with two other sets of parameters widely used in the literature to describe electroweak radiative corrections .
all formulas of this appendix are valid at one electroweak loop approximation . a set of three parameters @xmath507 , @xmath508 , @xmath509 has been suggested by altarelli , barbieri and jadach @xcite for phenomenological analysis of new physics : @xmath510 @xmath511 @xmath512 where @xmath513 describes the correction to @xmath391 , @xmath514 to @xmath392 and @xmath515 to @xmath17 : @xmath516 @xmath517 @xmath518 \;\ ; . \label{t.6}\ ] ] as is evident from the last two formulas , the virtue of @xmath508 and @xmath509 is that they do not contain the term @xmath13 .
so , at the time when top - quark mass was not measured at tevatron the corresponding uncertainties in @xmath508 and @xmath509 were diminished .
another set of parameters , @xmath528 , @xmath529 , @xmath530 was introduced a few years earlier by peskin and takeuchi @xcite .
these parameters were proposed to describe only the so - called oblique corrections due to the physics beyond the standard model . using the definitions of @xmath528 , @xmath529 , @xmath530 from @xcite and designating new physics contributions to @xmath531 as @xmath532
we obtain : @xmath533 @xmath534 @xmath535 according to @xcite @xmath540=16\pi [ \sigma'_a(0)-\sigma'_v(0 ) ] \;\ ; , \label{t.19}\ ] ] @xmath541 \;\ ; , \label{t.20}\ ] ] @xmath542 \;\ ; , \label{t.21}\ ] ] where @xmath543 and @xmath544 s are defined by the corresponding currents ( isotopic , 1 and 3 , and electromagnetic , @xmath109 , vector , @xmath62 , and axial , @xmath63 ) with coupling constants being extracted .
thus @xmath528 characterizes the degree of chiral symmetry breaking , while both @xmath529 and @xmath530 that of isotopic symmetry .
note that in eqs.([t.19 ] ) - ( [ t.21 ] ) only the contribution of new physics should be considered . since
new particles should be heavy it is reasonable to take into account only values of self - energies at @xmath347 and their first derivatives ( higher derivatives are power suppressed ) @xcite .
altogether we have eight parameters ( @xmath545 , @xmath546 , @xmath547 , @xmath548 , @xmath549 , @xmath550 , @xmath551 , @xmath552 ) , two of which are equal zero ( @xmath548 and @xmath547 ) , while three combinations can be absorbed in the definition of @xmath43 , @xmath1 and @xmath2 .
the remaining three combinations enter @xmath528 , @xmath529 and @xmath530 ( or @xmath532 , @xmath553 ) .
99 g. arnison et al .
( ua1 ) , phys . lett . *
126b * ( 1983 ) 398 ; + p. bagnaia et al .
( ua2 ) , phys . lett . * 129b * ( 1983 ) 310 ; + g. arnison et al .
( ua1-collaboration ) , phys . lett .
* 122b * ( 1983 ) 103 ; + m. banner et al .
( ua2-collaboration ) , phys . lett .
* 122b * ( 1983 ) 476 . s. van der meer , rep .
* 57 * ( 1985 ) 689 ; + c. rubbia , rep .
* 57 * ( 1985 ) 699 .
glashow , nucl .
* 22 * ( 1961 ) 579 ; + s. weinberg , phys .
* 19 * ( 1967 ) 1264 ; + a. salam , in elementary particle theory , ed .
n. svartholm ( almquist and wiksells , stockholm , 1968 ) .
glashow , rev .
* 52 * ( 1980 ) 539 ; + s. weinberg , rev .
* 52 * ( 1980 ) 515 ; + a. salam , rev .
* 52 * ( 1980 ) 525 .
peskin , d.v .
shroeder , _ an introduction to quantum field theory _ , addison - wesely , 1995 ; + s. weinberg , _ the quantum theory of fields _ , cambridge university press , 1995 , 1996 .
e. fermi , zeitschrift f. physik * 88 * ( 1934 ) 161 .
feynman , m. gell - mann , phys .
* 109 * ( 1958 ) 193 ; + r.e .
marshak , e.c.g .
sudarshan , phys . rev . * 109 * ( 1958 ) 1860 ; + j.j .
sakurai , nuovo cimento * 7 * ( 1958 ) 649
. o. klein , _ on the theory of charged fields _ , new theories in physics .
warsaw ( 1938 ) 66 .
( reprinted in oskar klein memorial lectures .
vol.1 . ed .
g. ekspong .
world sci . , singapore , 1991 . ) c.n .
yang , r.l.mills , phys . rev .
* 96 * ( 1954 ) 191 .
higgs , phys .
* 12 * ( 1964 ) 508 ; phys .
* 145 * ( 1966 ) 1156 ; + f. englert , r. brout , phys .
* 13 * ( 1964 ) 321 ; + g.s .
guralnik , c.r .
hagen , t.w .
kibble , phys .
* 13 * ( 1964 ) 585 ; + t.w .
kibble , phys . rev .
* 155 * ( 1967 ) 1554 .
g.t hooft , nucl . phys .
* b33 * ( 1971 ) 173 ; nucl . phys .
* b35 * ( 1971 ) 167 .
f. hasert et al .
, nucl . phys . *
b73 * ( 1974 ) 1 ; + a. benvenuti et al .
* 32 * ( 1974 ) 800 .
a.m. cnops et al .
* 41 * ( 1978 ) 357 ; + f. bergsma et al . , phys .
* 147b * ( 1984 ) 481 .
prescott et al . , phys . lett .
* b77 * ( 1978 ) 347 ; + c.y .
prescott et al . , phys . lett .
* b84 * ( 1979 ) 524 .
barkov , m.s .
zolotorev , pisma v zhetf * 26 * ( 1978 ) 544 ; + l.m .
barkov , m.s .
zolotorev , phys . lett . * b85 * ( 1979 ) 308 ; + i.b .
khriplovich , _ parity nonconservation in atomic phenomena _ , harwood academic publishers , 1991 ; + s.a .
blundell , w.r .
johnson , and j. sapirstein , phys .
* 65 * ( 1990 ) 1411 ; + m.a .
bouchiat , l. pottier , science * 234 * ( 1986 ) 1203 ; + j. rosner , w. marciano , phys
* 65 * ( 1990 ) 2963 .
bardin , v.a .
dokuchaeva , nucl . phys .
* b246 * ( 1984 ) 221 ; + s. sarantakos , a. sirlin , w.j .
marciano , nucl .
b227 * ( 1983 ) 84 ; + r.g .
stuart , z. phys . *
c34 * ( 1987 ) 445 .
a. sirlin , w.j .
marciano , phys . rev .
* d22 * ( 1980 ) 2695 ; nucl . phys . *
b189 * ( 1981 ) 442 ; + j.f .
wheater , c.h .
llewellynn - smith , nucl . phys .
* b208 * ( 1982 ) 27 ( errata : ibid , * b226 * ( 1983 ) 547 ) ; + c.h .
llewellynn - smith , nucl .
* b228 * ( 1983 ) 205 ; + d.yu .
bardin , v.a .
dokuchaeva , preprint jinr , e2 - 86 - 260 ( 1986 ) , dubna , unpublished .
physics at lep , eds .
j. ellis , r. peccei , cern 86 - 02 ( 1986 ) .
physics at lep - i , eds . g. altarelli , r. kleiss , c. verzegnassi , cern 89 - 08 ( 1989 ) .
reports of the working group on precision calculations for the @xmath0 resonance , eds .
d. bardin , w. hollik , g. passarino , cern 95 - 03 ( 1995 ) .
d. bardin , g. passarino , _ the standard model in the making : precision study of electroweak interactions _ , oxford university press ( to be published in 1999 ) .
lep ewwg and sld hfewg , cern - ep/99 - 15 , 8 febr .
j. erler , p. langacker , preprint upr-0816-t ( 1998 ) , hep - ph/9809352 , v.2 , 2 oct 1998 .
charm ii collab . , p. vilain et al .
, phys . lett . *
b335 * ( 1994 ) 246 .
charm collab . , j.dorenbosch et al . ,
b180 * ( 1986 ) 303 ; + lampf collab .
allen et al . , phys . rev .
* d47 * ( 1993 ) 11 . v.a .
novikov , l.b .
okun , m.i .
vysotsky , phys . lett .
* b298 * ( 1993 ) 453 ; + p. vilain et al . ,
charm ii collaboration , phys .
b320 * ( 1994 ) 203 .
review of particle physics .
the european physical journal c * 3 * ( 1998 ) number 1 - 4 .
m. steinhauser , phys . lett .
* b429 * ( 1998 ) 158 .
s. eidelman and f. jegerlehner , z. phys .
* c67 * ( 1995 ) 585 .
m. davier and a. hcker , phys . lett .
* b419 * ( 1998 ) 419 .
berman , a. sirlin , ann .
( n. y. ) * 20 * ( 1962 ) 20 ; + a. sirlin , rev .
* 50 * ( 1978 ) 573 .
lynn and m.e .
peskin , report slac - pub-3724 ( 1985 ) ( unpublished ) ; + b.w .
lynn , m.e . peskin and r.g .
stuart , in : _ physics at lep _ , ( report cern 86 - 02 ) ( cern , geneva , 1986 ) .
v. novikov , l. okun , m. vysotsky , nucl . phys .
* b397 * ( 1993 ) 35 .
s. willenbrock , g. valencia , phys
b259 * ( 1991 ) 573 ; + a. leike , t. riemann , j. rose , phys .
b273 * ( 1991 ) 513 ; + l3 collaboration ( o. adriani et al . ) , phys . lett .
* b315 * ( 1993 ) 494 ; + h. veltman , z. phys . *
c62 * ( 1994 ) 35 .
n. cabibbo , phys .
* 10 * ( 1963 ) 531 ; + m. kobayashi , t. maskawa , progr .
* 49 * ( 1973 ) 652 .
n. gray et al . , z.phys .
* c48 * ( 1990 ) 673 ; + l.r .
surguladze , univ . of oregon report
oits543 , hep - ph/9405325 ( 1994 ) .
m. veltman , nucl . phys .
* b123 * ( 1977 ) 89 ; acta phys . pol .
, * b8 * ( 1977 ) 475 .
cdf collaboration : f. abe et al .
* 74 * ( 1995 ) 2626 .
d0 collaboration : s. abachi et al .
* 74 * ( 1995 ) 2632 .
politzer , phys .
lett . * 30 * ( 1973 ) 1346 ; + d.j .
gross , f. wilczek , phys .
30 * ( 1973 ) 1343 ; + d.j .
gross , f. wilczek , phys .
rev . * d8 * ( 1973 ) 3633
vysotsky , v.a .
novikov , l.b .
okun , a.n .
rozanov , usp .
nauk * 39 * ( 1996 ) 503 .
v. novikov , l. okun , a. rozanov , m. vysotsky , preprint itep 19 - 95 , preprint cppm-1 - 95 ; + http://cppm.in2p3.fr./leptop/[email protected] a. djouadi , c. verzegnassi , phys . lett .
* b195 * ( 1987 ) 265 ; + b.a .
kniehl , nucl . phys .
* b347 * ( 1990 ) 86 ; + f. halzen and b.a .
kniehl , nucl . phys .
* b353 * ( 1991 ) 517 .
v. novikov , l. okun , m. shifman , a. vainshtein , m. voloshin and v. zakharov , phys . rep .
* 41c * ( 1978 ) 1 .
chetyrkin , j.h .
khn , m. steinhauser , phys . lett .
* b351 * ( 1995 ) 331 ; phys .
* 75 * ( 1995 ) 3394 ; + l. avdeev , j. fleisher , s. mikhailov , o. tarasov , phys
b336 * ( 1994 ) 560 ; * b349 * ( 1995 ) 597 ( e ) . a. czarnecki and j.h .
khn , phys .
* 77 * ( 1996 ) 3955 .
akhundov , d.yu .
bardin , t. riemann , nucl . phys .
* b276 * ( 1986 ) 1 ; + j. bernabeu , a. pich , a. santamaria , phys . lett . *
b200 * ( 1986 ) 569 ; + w. beenaker , w. hollik , z. phys . *
c40 * ( 1988 ) 141 . j. fleisher , f. jegerlehner , p. raczka , o.v .
tarasov , phys .
b293 * ( 1992 ) 437 ; + c. buchalla , a. buras , nucl
b398 * ( 1990 ) 285 ; + g. degrassi , nucl . phys .
* b407 * ( 1993 ) 271 ; + k.g .
chetyrkin , a. kwiatkowski , m. steinhauser , mod .
a8 * ( 1993 ) 2785 .
r. harlander , t. seidensticker and m. steinhauser , phys . lett .
* b426 * ( 1998 ) 125 .
d. bardin et al . , program zfitter 4 . 9 , nucl . phys .
* b351 * ( 1991 ) 1 ; z. phys . *
c44 * ( 1989 ) 493 ; phys . lett .
* b255 * ( 1991 ) 290 ; cern - th .
6443 - 92 ( 1992 ) .
j. ellis , g. fogli , phys .
* 213b * ( 1988 ) 189 , 526 ; * 232b * ( 1989 ) 139 ; + j. ellis , g. fogli , phys . lett . *
b249 * ( 1990 ) 543 .
w. hollik , fortschr . phys .
* 38 * ( 1990 ) 3 , 165 ; + m. consoli , w. hollik and f. jegerlehner , proceedings of the workshop on @xmath0 physics at lep - i , cern report 89 - 08 vol .
i , 7 ; + g. burgers , f. jegerlehner , b. kniehl and j.h .
khn , cern report 89 - 08 vol .
g. montagna , o. nicrosini , g. passarino , f. piccinini and r. pittau , nucl . phys .
* b401 * ( 1993 ) 3 ; + g. montagna , o. nicrosini , g. passarino , f. piccinini and r. pittau , program topazo , comput .
* 76 * ( 1993 ) 328 .
j. erler , p. langacker , the european journal of phys .
* 3 * ( 90 ) 1998 .
v. novikov , l. okun , a. rozanov , m. vysotsky , preprint itep 19 - 95 , cppm-1 - 95 .
r. barbieri , m. beccaria , p. ciafaloni , g. curci and a. vicere , phys .
b288 * ( 1992 ) 95 ; nucl . phys .
* b409 * ( 1993 ) 105 .
j. fleischer , o.v .
tarasov and f. jegerlehner , preprint bi - tp-93/24 and psi - pr-93 - 14 ; phys .
* b319 * ( 1993 ) 249 .
g. degrassi , p. gambino and a. vicini , phys . lett .
* b383 * ( 1996 ) 219 .
g. degrassi , p. gambino and a. sirlin , phys . lett .
* b394 * ( 1997 ) 188 .
g. degrassi , p. gambino , m. passera and a. sirlin , phys . lett .
* b418 * ( 1998 ) 209 .
m. chanowitz , phys .
+ m. chanowitz , preprint lbnl-42103 ( 1998 ) ; hep - ph/9807452 .
t. inami , t. kawakami , and c.s .
lim , mod .
. lett . * a10 * ( 1995 ) 1471 . v. novikov , l. okun , a. rozanov , m. vysotsky , and v. yurov , + mpl * a10 * ( 1995 ) 1915 .
a. masiero , f. feruglio , s. rigolin , and r. strocchi , phys . lett .
* b355 * ( 1995 ) 329 .
nilles , phys.rep .
* 110 * ( 1984 ) 1 ; + h.e .
haber and g.l .
kane , phys.rep .
* 117 * ( 1985 ) 75 ; + m. peskin , in : n. ellis and m. neubert , eds . , proc .
1996 european school of high - energy physics , carry - le - rouet , france , cern report 97 - 03 , p.49 , hep - ph/9705479 ; + j. ellis , in : proc
. 1998 european school of high - energy physics , st.andrews , scotland , cern - th/98 - 329 , hep - ph/9812235 .
p. chankowski , a. dabelstein , w. hollik et al .
, nucl . phys . *
b417 * ( 1994 ) 101 .
d. garcia , r.j .
jimenez and j. sola , phys .
b347 * ( 1995 ) 309 , 321 .
boer , a. dabelstein , w. hollik et al . , z.phys . *
c75 * ( 1997 ) 627 .
j. erler , d.m .
pierce , nucl .
b526 * ( 1998 ) 53 . l. alvarez - gaume , j. polchinski and m. wise , nucl
b211 * ( 1983 ) 495 ; + r. barbieri and l. maiani , nucl . phys . *
b224 * ( 1983 ) 32 .
m. boulware , d. finnell , phys . rev .
* d44 * ( 1991 ) 2054 .
gaidaenko , a.v .
novikov , v.a .
novikov , a.n .
rozanov , + m.i .
vysotsky , jetp lett .
* 67 * ( 1998 ) 761 ; + i.v .
gaidaenko , a.v .
novikov , v.a .
novikov , a.n .
rozanov , + m.i .
vysotsky , hep - ph/9812346 .
k. hagiwara , h. murayama , phys .
246b * ( 1990 ) 533 .
s. lammel , fermilab - conf-98 - 055-e ( 1998 ) .
p. chankowski , proc .
quantum effects in the mssm , barcelona 1997 , p.87 , preprint ift/97 - 18 , hep - ph/9711470 .
haber and r. hempfling , phys .
* 66 * ( 1991 ) 1815 ; + y. okada , m. yamaguchi , t. yanagida , progr .
* 85 * ( 1991 ) 1 ; + j. ellis , g. ridolfi , f. zwirner , phys . lett .
* b257 * ( 1991 ) 83 .
landau , a.a .
abrikosov , i.m .
khalatnikov , doklady akad .
nauk sssr * 95 * ( 1954 ) 497 , 1177 ; + m. gell - mann and f. low , phys
* 95 * ( 1954 ) 1300 . in _ physics at lep - i
_ vol.1 ( eds .
g. altarelli , r. kleiss , c. verzegnassi ) ( cern 89 - 08 , geneva , 1989 ) p. 7 .
( conveners : m. consoli , w. hollik ; working group : f.jegerlehner ) .
r. nevzorov , a. novikov , yad . fiz .
* 59 * ( 1996 ) 540 .
novikov , l.b . okun and m.i .
vysotsky , mod .
a9 * ( 1994 ) 1489 .
a. sirlin , phys . rev .
* d22 * ( 1980 ) 971 .
gorishny , a.l .
kataev and s.a .
larin , phys . lett .
* b259 * ( 1991 ) 144 ; + l.r .
surguladze and m.a .
samuel , phys .
* 66 * ( 1991 ) 560 .
chetyrkin , j.h .
khn , phys . lett .
* b248 * ( 1990 ) 359 ; + k.g .
chetyrkin , j.h .
khn and a. kwiatkowski , phys . lett .
* 282 * ( 1992 ) 221 .
chetyrkin and a. kwiatkowski , phys . lett .
* b305 * ( 1993 ) 288 ; + k.g .
chetyrkin and a. kwiatkowski , phys . lett .
* b319 * ( 1993 ) 307 ; + k.g .
chetyrkin , phys . lett . *
b307 * ( 1993 ) 169 .
larin , t.van ritberger , j.a.m .
vermaseren , phys . lett .
* b320 * ( 1994 ) 159 ; + k.g .
chetyrkin , o.v .
tarasov , phys .
b327 * ( 1994 ) 114 .
kataev , phys . lett .
* 287 * ( 1992 ) 209 .
g. altarelli , r. barbieri , phys . lett .
* b253 * ( 1991 ) 161 ; + g. altarelli , r. barbieri , s. jadach , nucl . phys . *
b369 * ( 1992 ) 3 .
m. peskin , t. takeuchi , phys .
* 65 * ( 1990 ) 964 ; phys . rev .
* d46 * ( 1992 ) 381 .
limits on higgs boson masses from combining the data of the four lep experiments at @xmath555 gev , aleph , delphi , l3 and opal collaborations , preprint cern - ep ( 1999 ) , in preparation .
figure 3 : photon polarization of the vacuum , resulting in the logarithmic running of the electromagnetic charge @xmath35 and the ` fine structure constant ' @xmath556 , as a function of @xmath153 , where @xmath557 is the 4-momentum of the photon ( a ) .
some of the diagrams that contribute to the self - energy of the @xmath33-boson ( b)-(g ) . some of the diagrams that contribute to the self - energy of the @xmath0-boson ( h)-(n ) . some of the diagrams that contribute to the @xmath558 transition ( o)-(r ) .
figure 4 : vertex triangular diagrams in the @xmath559 decay ( a ) , ( b ) , ( c ) . loops that renormalize the lepton wavefunctions in the @xmath559 decay .
( of course , antilepton have similar loops . ) ( d ) , ( e ) .
types of diagrams that renormalize the @xmath0 boson wavefunction in the @xmath559 decay ( f ) , ( g ) .
virtual particles in the loops are those that we were discussing above .
@xmath521 as a function of @xmath5 for three values of @xmath4 .
the dotted parabola corresponds to veltman approximation : @xmath560 .
solid horizontal line traces the experimental value of @xmath521 while the dashed horizontal lines give its upper and lower limits at the @xmath561 level .
@xmath519 as a function of @xmath5 .
the dotted parabola corresponds to veltman approximation : @xmath562 .
solid horizontal line traces the experimental value of @xmath519 while the dashed horizontal lines give its upper and lower limits at the @xmath561 level .
figure 9 : @xmath520 as a function of @xmath5 .
the dotted parabola corresponds to veltman approximation : @xmath563 .
solid horizontal line traces the experimental value of @xmath520 while the dashed horizontal lines give its upper and lower limits at the @xmath561 level .
figure 14 : the 2-dimensional exclusion plot for the case of @xmath569 extra generations and for the choice @xmath570 gev the lowest allowed value for new quark mass from tevatron search @xcite , using @xmath571 gev at 95 % c.l . from lep-2
little cross corresponds to @xmath234 minimum ; lines show one sigma , two sigma , etc allowed domains . | the precision data on @xmath0 boson decays from lep - i and slc colliders are compared with the predictions based on the minimal standard theory .
the born approximation of the theory is based on three most accurately known observables : @xmath1 the four fermion coupling constant of muon decay , @xmath2 the mass of the @xmath0 boson , and @xmath3 the value of the running fine structure constant " at the scale of @xmath2 .
the electroweak loop corrections are expressed , in addition , in terms of the masses of higgs , @xmath4 , of the top and bottom quarks , @xmath5 and @xmath6 , and of the strong interaction constant @xmath7 .
the main emphasis of the review is focused on the one - electroweak - loop approximation .
two electroweak loops have been calculated in the literature only partly .
possible manifestations of new physics are briefly discussed .
* contents . * 1 . * introduction . * 2 .
* basic parameters of the theory . * 3 .
* amplitudes , widths , and asymmetries . * 4 . *
one - loop corrections to hadronless observables .
* + 4.1 four types of feynman diagrams .
+ 4.2 the asymptotic limit at @xmath8 .
+ 4.3 the functions @xmath9 , @xmath10 and @xmath11 .
+ 4.4 corrections @xmath12 .
+ 4.5 accidental ( ? ) compensation and the mass of the @xmath13-quark .
+ 4.6 how to calculate @xmath14 ? ` five steps ' . 5 .
* one - loop corrections to hadronic decays of the @xmath0 boson .
* + 5.1 the leading quarks and hadrons .
+ 5.2 decays to pairs of light quarks .
+ 5.3 decays to @xmath15 pair .
6 .
* comparison of one - electroweak - loop results and experimental lep - i and slc data . * + 6.1 leptop code .
+ 6.2 one - loop general fit
. 7 .
* two - loop electroweak corrections and theoretical uncertainties . * + 7.1 @xmath16 corrections to @xmath17 , @xmath18 and @xmath19 from reducible diagrams .
+ 7.2 @xmath20 corrections from irreducible diagrams .
+ 7.3 @xmath21 corrections and the two - loop fit of experimental data .
+ 8 .
* extensions of the standard model . *
+ 8.1 sequential heavy generations in the standard model .
+ 8.2 susy extensions of the standard model .
+ 9 . * conclusions .
* * appendix a. regularization of feynman integrals .
* + + * appendix b. relation between @xmath22 and @xmath23 . * + + * appendix c. how @xmath24 and @xmath25 ` crawl ' . * + + * appendix d. general expressions for one - loop corrections to hadronless observables . *
+ + * appendix e. radiators @xmath26 and @xmath27 . *
+ + * appendix f. @xmath16 corrections from reducible diagrams .
* + + * appendix g. oblique corrections from new generations and susy . * + + * appendix h. other parametrizations of radiative corrections
. * + + * references .
* + + * figure captions . * |
sdss j1029 + 2623 @xcite is a unique quasar lens system with three images of the @xmath8 quasar produced by a foreground cluster of galaxies at @xmath9 .
such `` naked cusp '' configurations are predicted to comprise a significant fraction of cluster - scale quasar lenses @xcite .
all three quasar images are also detected at radio @xcite and x - ray wavelengths @xcite , and show mini - broad absorption line ( mini - bal ) features in the ly@xmath10 , siiv , and civ emission lines .
these earlier observations already implied that the lensing cluster is likely complex .
the optical images clearly show two bright central galaxies in the lensing cluster .
follow - up spectroscopy found that the two galaxies have a velocity difference of @xmath11 @xcite , which might imply a high - speed collision of two massive haloes .
chandra _ x - ray observations show that the x - ray surface mass distribution is centred on one of the central galaxies , with an additional x - ray subpeak @xmath12 north - west of the main peak that has no optical counterpart @xcite .
this complex x - ray morphology also suggests that the lensing cluster is in a merger .
it is crucial to construct a reliable mass model from a lensing analysis in order to draw any conclusion on the dynamical state of the cluster .
another interesting feature of this system is the anomalous flux ratios seen in the quasar images .
simple mass models predict the relative brightnesses of the three quasar images @xmath13 , @xmath14 , and @xmath15 should be @xmath16 , whereas optical , radio , and x - ray observations all found relative brightnesses of @xmath17 .
the fact that this `` anomaly '' is seen in radio @xcite and absorption - corrected x - ray @xcite flux ratios indicates that it is not due to extinction by any intervening matter .
in addition , the rough agreement of the flux ratios between three different wavelengths implies that it is not entirely due to stellar microlensing ( but see also * ? ? ? * ) , particularly since microlensing generally leads to large differences between optical and x - ray flux ratios ( e.g. , * ? ? ?
thus the most plausible explanation for the anomalous flux ratio is millilensing by ( dark halo ) substructures @xcite .
an accurate macro lens model is crucial for breaking the degeneracy between the macro model uncertainties and the amount of substructure required to account for the flux ratios . in this paper
, we present a new lensing analysis of sdss j1029 + 2623 utilizing deep multiband _ hubble space telescope ( hst ) _ observations of this system .
several multiply imaged galaxies , which are newly identified from the _ hst _ images , are used to construct an accurate mass model of the core of the lensing cluster .
strong lensing results are then combined with weak lensing results from the _ hst _ images and our earlier wide - field weak lensing analysis based on the subaru / suprime - cam images @xcite .
we then use the mass model to interpret the _
chandra _ x - ray surface brightness profiles @xcite and the anomalous flux ratios between the quasar images .
the structure of this paper is as follows .
after briefly describing the observations in section [ sec : obs ] , we present the strong lensing analysis in section [ sec : slens ] and the weak lensing analysis in section [ sec : wlens ] .
we then combine the various observations in section [ sec : comb ] to discuss the properties of the lensing cluster .
section [ sec : flux ] is devoted to discussions of the anomalous flux ratio between the quasar images .
finally we summarize the results on section [ sec : conclusion ] . throughout this paper
we assume a flat cosmological model with matter density @xmath18 , cosmological constant @xmath19 , and dimensionless hubble constant @xmath20 @xcite .
all photometry is in the ab system without correcting for galactic extinction .
diameter aperture magnitudes . the dotted line indicates the colour cut used to eliminate cluster member galaxies for weak lensing analysis presented in section [ sec : wlens ] .
[ fig : colmag ] ] we observed sdss j1029 + 2623 with the advanced camera for surveys ( acs ) and the wide field camera 3 ( wfc3 ) on _ hst _
( go-12195 ; pi oguri ) .
the observation consisted of 2 orbits ( total exposure 5276 sec ) of acs / f475w imaging on 2011 may 15 , 3 orbits ( total exposure 8010 sec ) of acs / f814w imaging on 2011 april 17 , and 2 orbits ( total exposure 5223.5 sec ) of wfc3/f160w imaging on 2011 april 15 .
we mostly follow the data reduction procedure described in @xcite , which uses multidrizzle @xcite for cosmic ray removal , distortion removal and stacking .
a pixel scale of @xmath21 is adopted for both the acs and wfc3 images .
the acs ( wfc3 ) frames are drizzled using the lanczos3 ( square ) kernel , with a pixfrac of 1 ( 0.8 ) .
the acs data are preprocessed using the pipeline calacs , which also removes bias striping and provides a pixel - level charge transfer inefficiency correction .
we use sextractor @xcite to construct a photometric catalogue of the galaxies in the _ hst _ images . for each galaxy , we derive a total magnitude from the mag_auto parameter and a @xmath22 diameter aperture magnitude .
we show the colour - magnitude diagram of the sdss j1029 + 2623 field from the two acs images in figure [ fig : colmag ] .
the figure shows a clear red - sequence of passively evolving galaxies in the lensing cluster at @xmath23 .
we also use wide - field subaru suprime - cam @xcite images of sdss j1029 + 2623 for the analysis .
these images consist of 1200 sec @xmath24-band , 2700 sec @xmath25-band , and 1920 sec @xmath26-band images , with a seeing fwhm of @xmath27 in @xmath25-band .
the weak lensing analysis of the subaru data was presented in @xcite , which we adopt in this paper .
sdss j1029 + 2623 was also observed with _
chandra _ x - ray observatory ( obsid 11755 ; pi oguri ) on 2010 march 11 for 60 ksec .
the x - ray image clearly reveals x - ray emission both from the lensing cluster and the three quasar images .
the data reduction and basic spectral and image analysis were detailed in @xcite .
the main objective of the _ hst _ imaging observation was to constrain the central mass distribution of the lensing cluster using strong lensing information . for this purpose , we identify multiply imaged galaxies in addition to the three quasar images based on their colours and morphologies combined with matching multiple image candidates while iteratively refining the mass models ( see below for details of our mass model ) . in total
, we identified 5 additional sets of multiply imaged galaxies ( i d 1 - 5 ) , which are shown in figures [ fig : sdss1029rgb_all ] and [ fig : zoom_all ] , and summarized in table [ tab : img ] .
in addition , the _ hst _ image shown in figure [ fig : sdss1029rgb_all ] exhibits a prominent host galaxy of the lensed quasar which is highly elongated due to lensing . in order to take account of the shape of the lensed host galaxy
, we include an additional set of multiple images ( i d 6 ) which roughly correspond to the edge of the host galaxy .
to summarize , in this paper we use the positions of 27 multiple images of 7 systems as constraints .
.list of multiple images for strong lensing analysis .
[ tab : img ] [ cols="^,^,^,^",options="header " , ] we now combine the constraints from strong lensing ( section [ sec : slens ] ) , _ hst _ weak lensing ( section [ sec : wlens ] ) , and subaru weak lensing @xcite to derive constraints on the overall radial density profile of the lensing cluster . for the strong lensing result , we conservatively use the constraint on the einstein radius of @xmath28 for the source redshift of @xmath8 . for the _ hst _ and subaru weak lensing data , we use the tangential shear profiles as constraints .
we use these data to constrain the virial mass @xmath29 and the concentration parameter @xmath30 , assuming the nfw profile .
the fit was performed for the mass range of @xmath31 and the concentration parameter range of @xmath32 .
we show constraints in the @xmath29-@xmath30 plane in figure [ fig : cont_mc ] , and summarize the resulting constraints on @xmath29 and @xmath30 in table [ tab : mc ] .
the individual constraints are quite degenerate in this plane , but along slightly different directions so that we obtain a tighter constraint when combining the three constraints .
we find @xmath33 and @xmath4 from the combined analysis .
the large concentration parameter value for the virial mass of @xmath34 is in line with a recent study the @xmath29-@xmath30 relation for strong lensing clusters @xcite , which can be explained by the effect of baryon cooling and central galaxies ( * ? ? ?
* but see also @xcite ) .
however , the high concentration for this cluster should be interpreted cautiously given the complex core structure with two density peaks .
we compare the tangential shear profiles with the best - fit model prediction in figure [ fig : profile_all ] .
the three observational constraints are complementary and they are consistent with each other where they overlap .
the best - fit nfw profile reproduces the observations for a wide range in radii .
predicted by the best - fit strong lens model ( _ filled triangles _ ) for @xmath35 .
the best - fit shear profile is computed for @xmath35 , which is close to the effective source redshifts of @xmath36 and @xmath37 for the _ hst _ and subaru weak lensing data .
the shaded region indicates the @xmath38 range .
the lower panel shows the radial profile of the @xmath39 rotated component @xmath40 .
[ fig : profile_all ] ] here we discuss several systematic effects that can affect our results .
one is the redshift distribution of source galaxies for weak lensing analysis .
since the area of the field used for our weak lensing analysis is small , the redshift distribution in this field can be different from that of a mean distribution .
another effect is a multiplicative shear measurement bias , which is estimated to be @xmath41% for our subaru weak lensing measurements .
also the _ hst _ weak lensing measurement probes the region @xmath42 , where weak lensing measurements are less tested and therefore less reliable . to estimate a possible impact of these systematic errors
, we consider an extreme situation where both the _ hst _ and subaru weak lensing measurements are offset by @xmath43% , and find the resulting shifts of the best - fit virial mass to be @xmath44 .
the systematic error for this case is still comparable to the @xmath38 statistical error , implying that these systematic errors are not significant . from the _ chandra _ x - ray analysis and from the combined lensing analysis ( _ solid line _
; see section [ sec : raden ] ) , assuming a spherically symmetric mass distribution . for the _ chandra _ x - ray analysis
, we show the results for the isothermal @xmath45-model from ( * ? ? ?
* _ open circle _ ) and a @xmath45-model with the temperature profile of ( * ? ?
* _ open squares _ ) .
the errors in the x - ray masses come from the statistical errors in the x - ray temperature measurement . for reference
, we also show the cluster radii @xmath46 , @xmath47 , and @xmath48 , which are computed from the best - fit lensing mass profile , as well as the einstein radius @xmath49 .
[ fig : m3d ] ] @xcite presented the _
chandra _ x - ray analysis of sdss j1029 + 2623 , and derived a mass profile assuming hydrostatic equilibrium and isothermality .
figure [ fig : m3d ] compares the x - ray mass profile from @xcite with the result of the combined lensing analysis in section [ sec : raden ] .
we find that the mass profiles derived from lensing and x - ray differ significantly with each other .
while the enclosed masses agree at the radius of @xmath50 that roughly corresponds to the einstein radius of this system , the enclosed masses inferred from the x - ray data are a factor of @xmath5 larger than those inferred from the combined lensing analysis at radii @xmath51 , where @xmath48 is defined by the radius within which the average density is 2500 times the critical density at the cluster redshift .
we note that the lensing derived mass implies an x - ray temperature and luminosity of @xmath52 kev and @xmath53 erg s@xmath54 based on x - ray scaling relations ( e.g. , * ? ? ?
* ) , which are significantly lower than the observed x - ray temperature of @xmath55 kev and @xmath56 erg s@xmath54 @xcite .
it is also worth noting that recent systematic weak lensing analysis of clusters at @xmath57 found somewhat smaller weak lensing masses for a given x - ray temperature @xcite , which is qualitatively similar to our result , although the difference appears to be much smaller than found here .
there are several effects that can lead to a systematic difference between the x - ray and lensing masses .
one of the most significant effects for x - ray cluster mass measurements is any violation of the assumption of hydrostatic equilibrium due the presence of non - thermal pressure support .
however , this effect typically leads to an underestimate of the x - ray mass , particularly in the outskirts of clusters ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and would only make the discrepancy larger .
another possibility is our assumption of isothermality .
in fact , x - ray temperatures generally decrease at large radii , but our _ chandra _ data were not sensitive enough to detect such a decrease . to examine this effect
, we adopt a temperature profile obtained from hydrodynamic simulations @xcite , which appears to be consistent with recent x - ray observations of cluster outskirts ( e.g. , * ? ? ?
* ) , and recalculate the x - ray mass profile . while this reduces the difference at the center , it can not explain the overall difference between the x - ray and lensing masses .
it is also difficult to explain the difference by a triaxial halo , because the large @xmath30 implies that the major - axis of the cluster is more likely to be aligned with the line - of - sight direction , in which case the lensing mass should be overestimated @xcite .
the most likely explanation for the mass discrepancy is shock heating of the intracluster gas during a merger .
numerical simulations show significant boosts of x - ray luminosity and temperature @xmath58 gyr after mergers , which can lead to significant overestimates of x - ray masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
observationally , there are several clusters showing signs of ongoing mergers that also have significantly higher x - ray masses than lensing masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
indeed the lensing cluster of sdss j0129 + 2623 shows hints of an ongoing merger , including the bimodal nature of cluster cores , the complex x - ray morphology , a possible offset between mass and x - ray centroids , and the large g1-g2 velocity difference . in addition , a line - of - sight merger can naturally explain the high concentration estimates for this cluster ( e.g. , * ? ? ?
spectroscopy of many more cluster member galaxies is needed to understand this complex cluster further .
if this interpretation is correct , the agreement between x - ray and lensing masses near the einstein radius might just be a coincidence , in that both the masses are overestimated by merger shock heating and halo elongation along the line - of - sight , respectively .
another useful cross - check of our interpretation is provided by the gas - to - mass ratio , @xmath59 , because it is expected to roughly match the cosmic baryon fraction @xmath60 @xcite for massive clusters , and also because @xmath61 appears to be the most promising cluster mass proxy @xcite .
we use the gas mass profile implied by the isothermal @xmath45-model from @xcite to estimate a gas mass within the radius @xmath62mpc from the combined lensing analysis .
we find a gas mass fraction of @xmath63 , which is significantly smaller than the cosmic baryon fraction .
the gas mass fraction is slightly increased to @xmath64 when the temperature profile of @xcite is used to derive the total mass , but the value is still low . on the other hand , when the lensing mass estimate is used for the total mass , we obtain a gas mass fraction of @xmath65 , which is in better agreement with @xmath66 estimates in other clusters ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this further supports our interpretation that the x - ray derived mass is significantly overestimated due to shock heating of the intracluster gas during a merger .
] our strong lens models in section [ sec : slens ] confirm the earlier claims @xcite that a substructure is required to explain the anomalous flux ratios between the three quasar images . in order to search for possible substructures ,
we subtract the quasars from each _ hst _ image and carefully examine the residual images . in the acs / f814w image , we find a probable galaxy in the vicinity of image c , as shown in figure [ fig : sdss1029bc_sub ] .
we fit the galaxy , named gx , assuming a sersic profile with index @xmath67 using _ galfit _
@xcite , and find the centroid of the galaxy to be @xmath68 from image c ( r.a.=157.309482 , decl.=26.394158 ) , and that the magnitude of the galaxy is @xmath69 .
these estimates have large uncertainties due to the significant residuals from the subtraction of image c ( see figure [ fig : sdss1029bc_sub ] ) . in the strong lens models we estimate that galaxy g2 with @xmath70 has a velocity dispersion of @xmath71 , which implies a velocity dispersion of @xmath72 for gx assuming @xmath73
. the inferred velocity dispersion of gx is large enough to mean that it will significantly perturb the flux of image c. ] we next explore the impact of gx on the strong lens models .
here we assume the same mass model as used in section [ sec : slens ] , but include an additional component modelled by pseudo - jaffe model at the location of gx with the velocity dispersion and the cutoff radius as free parameters .
we find that the best - fit model has @xmath74 including the flux ratios , which is significantly smaller than the @xmath75 of the original best - fit model ( @xmath76 ) and the large improvement is due to improvements in fitting the flux ratios .
this confirms the argument by @xcite that a substructure in the vicinity of image c can explain the anomalous flux ratio .
the velocity dispersion of galaxy gx in the best - fit model is @xmath77 , which is broadly consistent with the velocity dispersion expected from the magnitude .
we summarize the flux ratios of the best - fit mass models , as well as the observed flux ratios at various wavelengths , in figure [ fig : fratio ] .
the best - fit models with and without galaxy gx have quite different total magnification factors . for the model without galaxy gx , the magnifications for the quasar images are @xmath78 , @xmath79 , and @xmath80 , whereas for the model with galaxy gx , they are @xmath81 , @xmath82 , and @xmath83 .
the total magnifications are @xmath84 for the model without gx and @xmath85 for the model with gx .
substructures near merging image pairs can have large effects on time delays between the images @xcite .
this was indeed the case for sdss j1004 + 4112 @xcite , another large - separation lensed quasar produced by a massive cluster of galaxies , in which a time delay between the smallest image pair was found to be affected by the perturbation from nearby cluster member galaxies @xcite .
we find that our best - fit mass models predict the time delay between image b and c as @xmath86 day for the model without gx , and @xmath87 day for the model with gx . in both cases ,
image b leads image c as expected from the image geometry . on the other hand
, gx has a negligible impact on the ab time delay .
the large difference of @xmath88 implies that measuring the bc time delay will greatly help constrain the substructure that affects these quasar images . an interesting feature of sdss j1029 + 2623 is the presence of a highly elongated host galaxy , which is most prominent in the _ hst _ wfc3/f160w image .
here we examine whether our best - fit models reproduce the morphology of the lensed host galaxy .
to do so , we include the host galaxy modelled by a sersic profile , and optimise the parameters ( the total flux , size , ellipticity and position angle ) of the sersic profile assuming the best - fit mass models , both with and without galaxy gx .
we fix the centroid of the host galaxy in the source plane to the best - fit source quasar position .
the optimisation is conducted on a pixel - by - pixel basis using a binned _ hst _ wfc3/f160w image with a pixel scale of @xmath89 .
the model - predicted images are convolved with a psf assuming a moffat profile with a fwhm of @xmath90 and @xmath91 , before comparing with the observed image .
the result is shown in figure [ fig : hostcomp ] .
we find that our best - fit models reproduce the morphology of the lensed host galaxy quite well .
the match is better for the model with galaxy gx , which further supports its presence and the significant impact of galaxy gx on this lens system .
in particular , we note that the observed host galaxy shows signs of a kink near image c ( see also figure [ fig : sdss1029bc_sub ] ) which is reproduced in our best - fit model with galaxy gx .
we estimate the total magnitude of the lensed host galaxy as @xmath92 and the lensing - corrected original magnitude as @xmath93 .
the best - fit effective radius and sersic index are @xmath94 and @xmath95 , respectively .
the presence of galaxy gx has a negligible impact on the total magnification of the host galaxy .
the highly magnified host galaxy provides a very unique opportunity to study the relation between supermassive black holes and host galaxies in a distant quasar ( e.g. , * ? ? ?
* ; * ? ? ?
we have presented a detailed gravitational lensing analysis of sdss j1029 + 2623 , the largest - separation lensed quasar system known ( image separation @xmath96 , lens redshift @xmath9 , and source redshift @xmath8 ) , based on new _ hst _ acs and wfc3 observations . with the help of the high - resolution _ hst _ images ,
we have identified 6 new multiply imaged galaxies in addition to the three - image lensed quasar . based on the newly identified multiple images , we construct an accurate mass model at the cluster core .
we find dark haloes with roughly similar masses should be associated with both of the bright central galaxies .
the strong lens models allow us to estimate the einstein radius of the overall lens as @xmath1 .
our best - fit strong lens model successfully reproduces the observed time delay between quasar image a and b which was recently measured by @xcite .
we have also detected a significant weak lensing signal in the _ hst _ acs image .
the mass map obtained from the strong and weak lensing analysis appears to be different from the x - ray surface brightness profile @xcite , suggesting an ongoing merger .
we have constrained the radial mass profile of the cluster from the _ hst _ strong and weak lensing analysis as well as the subaru weak lensing result @xcite .
the combination of the three lensing constraints enabled a robust determination of the radial mass profile , with a best - fit virial mass of @xmath97 and a concentration parameter of @xmath4 .
the lensing derived mass profile differs significantly from the mass profile inferred from the _ chandra _ x - ray data assuming hydrostatic equilibrium and isothermality , such that the x - ray mass is a factor of @xmath5 larger at outer radii of @xmath98 .
we ascribe the mass discrepancy to shock heating of the intracluster gas during a merger , which leads to significant overestimates of x - ray cluster masses .
our interpretation is supported by the small cluster gas mass fraction implied by the x - ray estimates of total masses , with @xmath63 , as compared to @xmath65 using the lensing estimate of the total mass .
a merger along the line - of - sight can also explain the high concentration parameter value found for the cluster , and is consistent with the large velocity difference of @xmath99 between the bright central galaxies g1 and g2 @xcite . the merger interpretation for this strong lens selected cluster , in turn , may imply a large impact of cluster mergers on strong lensing cross sections ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if the merger interpretation is correct , one question is whether the large velocity difference is compatible with the theoretical expectations .
there are several works examining the distribution of velocity differences between close halo pairs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which indicate that a relative velocity of @xmath11 for haloes with @xmath100 is quite rare .
therefore , it is important to make more quantitative comparisons taking account of the redshift and mass of sdss j1029 + 2623 in order to check whether this large velocity difference can be explained in the framework of the standard @xmath101-dominated cold dark matter cosmology .
we confirm the earlier claims @xcite that the observed flux ratios between the quasar images can not be reproduced by strong lens models without any substructure in the vicinity of the quasar images . from a careful inspection of the _ hst _ images ,
we have identified a probable galaxy gx which is only @xmath68 from quasar image c. a mass model including gx successfully reproduces the observed flux ratios , thereby resolving the flux ratio anomaly .
while gx has little effect on the ab time delay measured by @xcite , the bc time delay is a sensitive probe of its effects .
we have also checked the morphology of the lensed host galaxy and found a very good agreement with the model prediction .
the best - fit model predict the total magnification of the quasar as @xmath102 and that of the lensed quasar host galaxy as @xmath103 .
finally we note that additional observations are essential to better characterise this unique quasar - cluster lens system .
for instance , the measurement of the time delay between quasar images b and c will be important for constraining the properties of galaxy gx .
the merger interpretation for the lensing cluster should be checked by extensive spectroscopy of cluster member galaxies .
spectroscopy of the multiply imaged galaxies identified in this paper is important both for more robust mass models and for possible cosmological constraints ( e.g. , * ? ? ?
furthermore , the highly elongated host galaxy makes this lens system an ideal site to conduct detailed explorations of a quasar host galaxy at @xmath104 ( e.g. , * ? ? ?
we thank an anonymous referee for useful suggestions . this work was supported in part by the first program
`` subaru measurements of images and redshifts ( sumire ) '' , world premier international research center initiative ( wpi initiative ) , mext , japan , and grant - in - aid for scientific research from the jsps ( 23740161 ) .
ts acknowledges support from nsf through grant ast-0444059 - 001 , and the smithsonian astrophysics observatory through grant go0 - 11147a .
csk is supported by nsf grant ast-1009756 .
xd acknowledges financial support from nasa / sao award go011147b and nasa / stsci award hst - go-12915.07-a .
this paper is based on observations with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
this research has made use of data obtained from the chandra data archive and the chandra source catalog , and software provided by the chandra x - ray center ( cxc ) in the application packages ciao , chips , and sherpa .
this paper is based on data collected at subaru telescope , which is operated by the national astronomical observatory of japan . | we present _ hubble space telescope ( hst ) _ advanced camera for surveys ( acs ) and wide field camera 3 ( wfc3 ) observations of sdss j1029 + 2623 , a three - image quasar lens system produced by a foreground cluster at @xmath0 .
our strong lensing analysis reveals 6 additional multiply imaged galaxies in addition to the multiply imaged quasar .
we confirm the complex nature of the mass distribution of the lensing cluster , with a bimodal dark matter distribution which deviates from the _ chandra _ x - ray surface brightness distribution . the einstein radius of the lensing cluster is estimated to be @xmath1 for the quasar redshift of @xmath2 .
we derive a radial mass distribution from the combination of strong lensing , _
weak lensing , and subaru / suprime - cam weak lensing analysis results , finding a best - fit virial mass of @xmath3 and a concentration parameter of @xmath4 .
the lensing mass estimate at the outer radius is smaller than the x - ray mass estimate by a factor of @xmath5 .
we ascribe this large mass discrepancy to shock heating of the intracluster gas during a merger , which is also suggested by the complex mass and gas distributions and the high value of the concentration parameter . in the _ hst _ image
, we also identify a probable galaxy , gx , in the vicinity of the faintest quasar image c. in strong lens models , the inclusion of gx explains the anomalous flux ratios between the quasar images .
the morphology of the highly elongated quasar host galaxy is also well reproduced .
the best - fit model suggests large total magnifications of @xmath6 for the quasar and @xmath7 for the quasar host galaxy , and has an ab time delay consistent with the measured value . -
.5 in [ firstpage ] dark matter galaxies : clusters : individual : sdss j1029 + 2623 gravitational lensing : strong gravitational lensing : weak quasars : individual : sdss j1029 + 2623 x - rays : galaxies : clusters |
the recent experimental confirmation of the violation of bell inequalities @xcite in systems of electron spins @xcite , entangled photons @xcite , etc .
has made a compelling case for the non - locality " of quantum mechanics .
quantum phenomena exhibit correlations between space - like separated measurements that appear to be inconsistent with any local hidden variable explanation .
the spooky action at a distance " of quantum non - locality is now embraced and utilized in fundamental applications such as device - independent cryptography and randomness generation @xcite and reductions in communication complexity @xcite . moreover ,
this non - locality has also been used to show that even a tiny amount of free - randomness can be amplified @xcite and that extensions of quantum theory which incorporate a particular notion of free - choice can not have better predictive power than quantum theory itself @xcite .
the quantum non - local correlations are known to be fully compatible with the no - signaling principle , i.e. , the space - like separated parties can not use the non - local correlations to communicate superluminally @xcite .
since the proposal of popescu and rohrlich @xcite , it has been realized that non - local correlations might take on a more fundamental aspect .
not only quantum theory but any future theory that might contain the quantum theory as an approximation is now expected to incorporate non - locality as an essential intrinsic feature . this program has led to the formulation of device - independent information - theoretic principles @xcite that attempt to _ derive _ the set of quantum correlations from amongst all correlations obeying the no - signaling principle . in parallel , cryptographic protocols
have been devised based on the input - output statistics in bell tests such that their proof of security only relies on the no - signaling principle .
when one considers such post - quantum cryptography @xcite , randomness amplification @xcite , etc .
the eavesdropper eve is assumed to be limited to the preparation of boxes ( input - output statistics ) obeying a set of constraints collectively referred to as the no - signaling constraints .
the general properties of no - signaling theories have been investigated @xcite in a related program to formulate an information - theoretic axiomatic framework for quantum theory . on the other hand ,
quantum theory does not provide a mechanism for the non - local correlations .
several theoretical proposals have been put forward to explain the phenomenon of non - local correlations between quantum particles via superluminal communication between them .
these models go beyond quantum mechanics but reproduce the experimental statistical predictions of quantum mechanics , the most famous of these models being the de broglie - bohm pilot wave theory @xcite . in all relativistic theories , causality , " is imposed i.e. , the requirement that causes must precede effects in all space - time rest frames . before going further , two remarks are in order here .
first , by an effect we mean any possible event , even if it has been affected by other events ( causes ) indirectly .
second , we shall use as general a correlation point of view as possible , _ regardless _ of the physical theory from which the correlations arise . in this context , let us also note that given an arbitrary space - time structure , the question of causal order for any two measures has been formalized using intuitions from optimal transport theory @xcite . furthermore it has been proven that in any hyperbolic space - time , casuality of the evolution of measures supported on time - slices is an observer independent concept @xcite .
both formalisms have also been applied in the study of the evolution of the quantum wavepackets showing in particular the natural consistency of the causality concept with the relativistic continuity equation @xcite . from the perspective of communication
, the requirement of relativistic causality strictly demands that no faster - than - light ( ftl ) transmission of information takes place between a sender and a receiver . the no - signaling principle being in ubiquitous use ( in device - independent cryptography , axiomatic formulations , etc . as explained earlier ) , a natural question is to explore whether the no - signaling constraints that are currently in use precisely capture the constraints imposed by relativistic causality , i.e. , to derive the no - signaling constraints from relativistic causality . in this paper
, we investigate this question and find that , surprisingly , the answer is that the multi - party no - signaling principle requires a modification to capture the notion of causality . in particular ,
the actual no - signaling constraints that one must impose in a multi - party bell experiment are dictated by the space - time configuration of the measurement events in the experiment .
this modification of the no - signaling constraints naturally leads to a number of implications in tasks of device - independent cryptography against relativistic eavesdroppers and to explanations of quantum correlations via finite - speed superluminal influences .
this paper is therefore , propaedeutic to a larger project undertaken by the authors to explore the relativistic causality constraints and their implications on the foundations of quantum theory .
the structure of the paper is as follows .
we initially establish the setup of the bell experiment and recall the assumptions in the bell theorem .
we then define the notion of relativistic causality that we use in this paper ( and that is commonly accepted , i.e. , that there be no causal loops in spacetime ) and revisit the derivation of the two - party no - signaling constraints from causality .
we then show that , in the multi - party scenario , only a restricted subset of the no - signaling constraints is required to ensure that no causal loops appear .
we explicitly identify a region of space - time for the measurement events in a bell scenario where the usual no - signaling constraints fail . in this regard ,
we extend a particular framework of jamming " non - local correlations by grunhaus , popescu and rohrlich in @xcite based upon an earlier suggestion of shimony in @xcite .
we then examine the implications of the restricted subset of no - signaling constraints for device - independent cryptographic tasks against an eavesdropper constrained only by the laws of relativity .
we detail explicit attacks on known protocols for randomness amplification based on the ghz - mermin inequalities using boxes that obey the new relativistic causality conditions .
we show that from that perspective the security theory needs revision , and - in a way - to the some degree collapses .
we also explore the implications on some of the known features of no - signaling theories @xcite , in particular we find that the phenomenon of monogamy of correlations is significantly weakened in the relativistically causal theories and that the monogamy of chsh inequality violation @xcite disappears in certain spacetime configurations .
the notions of freedom - of - choice and no - signaling are known to be intimately related @xcite .
we re - examine how the notion of free - choice as proposed by bell and formalized by colbeck and renner @xcite can be stated mathematically within the structure of a space - time configuration of measurement events .
a breakthrough result in @xcite was a claim that any finite superluminal speed explanation of quantum correlations could lead to superluminal signaling and must hence be discarded .
we re - examine this question in light of the modified relativistic causality and free - will conditions . both non - relativistic quantum theory and relativistic quantum field theory @xcite
are well - known to obey a no superluminal signaling condition , proposals to modify quantum theory by introducing non - linearities have been shown to lead to signaling @xcite .
we end with discussion and open questions concerning feasible mechanisms for the point - to - region superluminal influences .
let us first establish the setup of a typical bell experiment . in the bell scenario denoted @xmath1 , we have @xmath0 space - like separated parties , each of whom chooses from among @xmath2 possible measurement settings and obtains one of @xmath3 possible outcomes . in general , the number of inputs and outputs for each party may vary , but this will not concern us in this paper .
the inputs of the @xmath4-th party will be denoted by random variable ( r.v . )
@xmath5 taking values @xmath6 in @xmath7 = \{1 , \dots , m\}$ ] and the outputs of this party will be denoted by r.v @xmath8 taking values @xmath9 in @xmath10 $ ] .
accordingly , the conditional probability distribution of the outputs given the inputs will be denoted by @xmath11 following @xcite , we also consider the notion of a spacetime random variable ( srv ) , which is a random variable @xmath12 together with a set of spacetime coordinates @xmath13 in some inertial reference frame @xmath14 at which it is generated . a measurement event @xmath15
is thus modeled as an input srv @xmath16 together with an output srv @xmath17 . as in typical studies of bell experiments
, here we consider the measurement process as instantaneous , i.e. , @xmath16 and @xmath17 share the same spacetime coordinates .
denote a causal order relation between two srv s @xmath16 and @xmath17 by @xmath18 if @xmath19 in all inertial reference frames , i.e. , @xmath17 is in the future light cone of @xmath16 ( so that @xmath16 may cause @xmath17 ) .
a pair @xmath20 of srvs is spacelike separated if @xmath21 . in this paper , we will have occasion to distinguish the specific spacetime location at which _ correlations _ between random variables manifest themselves , i.e. , the particular spacetime location at which the correlations are registered , from the spacetime locations at which the random variables themselves are generated .
accordingly , we label by @xmath22 the srv denoting the correlations between the output srvs @xmath23 and @xmath24 with its associated spacetime location @xmath25 being at the earliest ( smallest @xmath26 ) intersection of the future light cones of @xmath23 and @xmath24 in the reference frame @xmath14 .
consider the two - party bell experiment with two spacelike separated parties alice and bob .
the inputs of alice and bob are denoted by srv s @xmath27 and the outputs by @xmath28 .
if the measurement process is considered to be instantaneous , @xmath29 and @xmath30 share the same space - time coordinates as do @xmath31 . in the two - party bell scenario ,
the results of the experiment are described by the set of conditional probability distributions @xmath32 .
let @xmath33 denote a set of underlying variables describing the state of the system under consideration , these could in general be local or non - local .
we have evidently @xmath34 in this scenario , the bell theorem @xcite is based on the following assumptions
@xcite : 1 . _
outcome independence _ : the statistical correlations between the outputs arise from ignorance of the underlying variable @xmath33 @xmath35 this is evidently true for deterministic models ( those that output deterministic answers @xmath36 for inputs @xmath37 ) as well as for stochastic models and is motivated by a notion of realism , i.e. , that the measurements merely reveal pre - existing outcomes encoded in @xmath33 .
parameter - independence _ :
for each so - called microstate " @xmath38 , the probability of an outcome on alice s side is assumed to be ( stochastically ) independent of the experimental setting ( the parameters of the device ) on bob s side , @xmath39 the justification for this assumption comes from special relativity , which imposes that spacelike separated measurements do not influence each other s underlying outcome probability distributions .
measurement independence _ : the measurement inputs are uncorrelated with the underlying variable @xmath33 @xmath40 this is also sometimes called the _ free - choice _ assumption , i.e. , that the inputs @xmath37 are chosen freely , independent of @xmath33 . we elaborate on the notion of freeness in this two - party bell experiment in section [ subsec : two - party - freewill ] @xcite .
there are other assumptions such as : fair sampling , no backward causation , reality being single valued , etc . @xcite but these will not concern us in this paper .
substituting eqs.([eq : two - party - oi ] , [ eq : two - party - pi ] , [ eq : two - party - mi ] ) in eq.([eq : two - party - bi ] ) , we obtain the description of a local hidden variable model : @xmath41 figure [ fig : two - party - caus - struc ] shows the causal structure represented by the two - party bell experiment represented in terms of a directed acyclic graph ( dag ) @xcite .
remark that the local hidden variable model in eq.([eq : lhv - model ] ) can also be arrived at starting from other postulates such as local causality @xcite . from the assumptions of parameter - independence and measurement - independence
, one can deduce the _ no - signaling constraints _ : @xmath42 the no - signaling conditions in this two - party bell experiment formally state that probability distribution of the outcomes of any party is independent of the input of the other party .
the assumption of measurement - independence or free - choice can be formally expressed in terms of spacetime random variables . in the two - party bell experiment ,
one imposes the free - choice constraints @xmath43 one formal way to define the freedom - of - choice condition in terms of spacetime random variables was formulated by colbeck and renner ( cr ) in @xcite .
recall that we denote a causal order relation by @xmath18 if @xmath19 in all inertial reference frames , i.e. , @xmath17 is in the future light cone of @xmath16 so that @xmath16 cause @xmath17 .
cr formulate the notion of free - choice ( formalising bell s notion from @xcite ) as follows : [ def : freewill ] a spacetime random variable @xmath16 is said to be free if @xmath16 is uncorrelated with every spacetime random variable @xmath17 such that @xmath44 , i.e. , for all such @xmath17 , we have @xmath45 . in other words , @xmath16 is free if it is uncorrelated with any @xmath17 that it could not have caused , where the requirement for @xmath16 causing @xmath17 is that @xmath17 lies in the future light cone of @xmath16 .
clearly , eq.([eq : two - party - freewill ] ) follows if one adopts this notion of freedom - of - choice , although note that the def .
[ def : freewill ] is strictly stronger than just the conditions ( [ eq : two - party - freewill ] ) imposed in the bell theorem . in this paper , our focus is on the no - signaling and free - will constraints above which are intimately related to each other .
in particular , we consider the generalization of the no - signaling conditions to the multi - party scenario .
the generalized multi - party no - signaling constraints are usually stated as follows ( see for example @xcite ) : @xmath46 , \{a_1 , \dots , a_n\ } \setminus a_j , \{x_1 , \dots , x_j , x'_j , \dots , x_n\ } \end{aligned}\ ] ] in words , the above constraints state that the outcome distribution of any subset of parties is independent of the inputs of the complementary set of parties ( while eq.([eq : multi - party - ns ] ) imposes this for subsets of @xmath47 parties , one can straightforwardly show that this also implies that the marginal distribution for smaller sized subsets is well - defined @xcite ) .
now , given that as stated earlier the justification of the two - party no - signaling constraint came from the causality constraints of special relativity , the natural question which we investigate in this paper is to what extent the multi - party no - signaling constraints in eq.([eq : multi - party - ns ] ) are imposed by the causality constraints of special relativity .
in other words , while the constraints in eq.([eq : multi - party - ns ] ) are clearly sufficient to ensure that no superluminal signaling takes place , we derive the set of necessary and sufficient constraints that ensure that no causal loops appear in the theory .
we will see that the multi - party no - signaling conditions in fact depend on the space - time coordinates of the measurement events in the bell experiment and we propose an appropriate modification of these constraints .
the main results of the paper are as follows . 1 .
we derive in prop . [
prop : two - party - ns ] the no - signaling constraints in the two - party bell experiment from relativistic causality and show that these constraints are both necessary and sufficient in this scenario .
we show that in general , depending on the exact space - time coordinates of the measurement events of the parties participating in a bell experiment , relativistic causality strictly only imposes a subset of the usual no - signaling conditions .
in particular , we show that in the three - party bell experiment , two sets of relativistic causal constraints are possible : @xmath48 the usual no - signaling conditions and @xmath49 a subset of the no - signaling conditions in prop .
[ prop : rel - cau - constraints ] which ensure causality is preserved in certain measurement configurations ( such as in fig .
[ fig : three - party - meas - config ] ) .
we geometrically analyze in prop.[prop : spacetime - region ] the exact space - time region in the three - party bell scenario where the constraints in prop .
[ prop : rel - cau - constraints ] are necessary and sufficient to prevent causal loops .
we propose in definition [ lem : mod - freewill ] a rigorous definition of the free - choice constraints in the multi - party bell experiment and show in prop .
[ prop : caus - eq - freewill ] how these are compatible with the relativistic causality conditions .
we examine the implications of the causality constraints on device - independent cryptography against an adversary constrained only by the laws of relativity . in the task of randomness amplification ,
we demonstrate in prop .
[ prop : ra - rel - caus ] an explicit attack that renders a family of multi - party protocols based on the @xmath0-party mermin inequalities insecure , when the measurement events of the honest parties conform to certain spacetime configurations .
in the task of key distribution , we show in prop . [
prop : qkd - sec - rel - caus ] that even two - party protocols can be rendered insecure unless assumptions are made concerning the spacetime location of the eavesdropper s measurement event or shielding of the honest parties to any possible superluminal influences , even those respecting causality
. we then examine properties that were previously considered to be common to all no - signaling theories . in particular , we show in prop .
[ prop : rel - caus - mono ] that the paradigmatic phenomenon of monogamy of correlations violating the chsh inequality no longer holds in relativistically causal theories in certain measurement configurations
we re - examine , in section [ sec : v - causal - exp ] the question whether quantum correlations may admit explanations by finite speed superluminal influences propagating between the spacelike separated parties , by modifying the argument from @xcite against such @xmath50-causal models to show that they would lead to causal loops in certain measurement configurations .
finally , we investigate the phenomenon of multiparty non - locality in light of the new considerations and propose in def .
[ def : rel - caus - bilocal ] a new class of relativistically causal bilocal models in the three - party scenario .
we formulate a new svetlichny - type inequality in lemma [ lem : rcbl - ineq ] that is satisfied by this class of models and examine its violation within quantum theory .
in this paper , we work in the regime of special relativity , i.e. , flat spacetime with no gravitational fields , this regime is valid within small regions of spacetime where the non - uniformities of any gravitational forces are too small to measure .
we consider the causal structure of measurement events occurring at fixed spacetime locations @xmath51 . within this regime ,
the relativistic causality constraint we consider is simply that : * no _ causal loops _ occur , where a causal loop is a sequence of events , in which one event is among the causes of another event , which in turn is among the causes of the first event .
causality implies that for two causally related events taking place at two spatially separated points , the cause always occurs before the effect , and this sequence can not be changed by any choice of a frame of reference .
relativistic causality is a consequence of prohibiting faster - than - light transmission of information from one spacetime location @xmath17 to another space - like separated location @xmath52 in any inertial frame of reference .
a violation of this condition would lead to the well - known grandfather - paradoxes " .
in other words , if an effect @xmath52 that occurs at a space - time location @xmath53 in an inertial reference frame @xmath14 _ precedes _ its cause @xmath17 that occurs at space - time location @xmath54 in @xmath14 ( i.e. , @xmath55 ) then an observer at @xmath52 may in turn affect the cause at @xmath17 and lead to a paradox .
an explicit example of a closed causal loop is shown in the proof of prop .
[ prop : two - party - ns ] .
a few remarks are in order .
firstly , note that faster - than - light propagation in one privileged frame of reference alone does not lead to causal loops .
secondly , note that the `` principle of causality '' , i.e. , the invariance of the temporal sequence of causally related events is also directly related to the macroscopic notion of arrow of time from the second law of thermodynamics @xcite .
finally , we remark that in the general theory of relativity , the field equations allow for solutions in the form of closed timelike curves , and there has been much debate over these , with proposals such as the chronology protection conjecture and a self - consistency principle @xcite suggested to prevent time - travel paradoxes .
let us first revisit the derivation of the no - signaling constraint from relativistic causality in the typical two - party bell experiment . in any run of the bell experiment , alice _
freely _ chooses at spacetime location @xmath56 in @xmath14 , her measurement input @xmath57 $ ] and obtains ( instantaneously ) the output @xmath58 $ ] .
similarly , bob who is at a space - like separated location , in the same run , freely chooses at @xmath59 his input @xmath60 $ ] and obtains an output @xmath61 $ ] .
the requirement of space - like separation , i.e. , @xmath62 , ensures that the measurement events fall outside each other s light cone .
the conditional probability distributions of outputs given the inputs @xmath63 that they obtain in the bell experiment jointly constitute a box " @xmath64 .
the causality constraint is the requirement that faster - than - light information transmission from alice to bob or bob to alice is forbidden , i.e. , alice and bob can not use their local measurements to signal to one another .
note that here , alice and bob choose their measurement inputs _ freely _ , i.e. , we assume the free - choice constraints in eq.([eq : two - party - freewill ] ) . and @xmath65 respectively .
the inputs are chosen freely so no arrows are pointed towards @xmath66 .
the correlations between the outputs @xmath65 may be explained as caused by a shared random variable @xmath33 which may be local or non - local depending on the theory under consideration .
the input of each party is prevented from influencing the output distribution of the other party due to the spacelike separation between them.,scaledwidth=50.0% ] [ prop : two - party - ns ] in the two - party bell experiment , the usual no - signaling constraints @xmath67 are necessary and sufficient to ensure that relativistic causality is not violated .
suppose by contradiction that one of the constraints in eq.([eq : two - party - ns ] ) was violated , for definiteness , let us suppose that the marginal distribution of bob s outputs depended upon the input of alice , i.e. , suppose @xmath68 note that by assumption , alice chooses her measurement freely , i.e. , @xmath69 so suppose that in some specific run alice chooses the input @xmath70 at spacetime location @xmath54 .
since alice and bob s measurement events are spacelike separated , eq.([eq : alice - signal - bob ] ) would imply physically that a superluminal influence propagated from alice s system to bob s system informing bob of alice s input @xmath71 .
explicitly , consider that alice and bob share many copies of a system obeying ( [ eq : alice - signal - bob ] ) , when alice chooses input @xmath71 on all her subsystems , bob guesses @xmath71 with a probability strictly larger than uniform by inspecting his local statistics @xmath72 .
this is illustrated in fig.[fig : two - party - ccl ] where in a particular frame of reference @xmath73 labeled by axes @xmath74 ( for concreteness , this can be taken to be the laboratory frame at which alice and bob s systems are at rest ) , the superluminal influence is shown to be instantaneous , i.e. , the signal travels parallel to the time axis @xmath26 and reaches bob s system at spacetime location @xmath53 .
now , consider the inertial reference frame @xmath75 labeled by @xmath76 of charlie and dave who are moving uniformly at some speed @xmath50 relative to @xmath73 . at the space - time location @xmath77 charlie and bob s world - lines intersect , suppose bob informs charlie of the value @xmath71 at this point .
charlie immediately transmits this information via the same superluminal mechanism to dave . in the frame @xmath75 ,
this superluminal influence again travels parallel to the time axis @xmath78 .
the information about @xmath71 thus reaches dave at spacetime location @xmath79 which is in the causal past of @xmath54 as shown in fig .
[ fig : two - party - ccl ] .
alice has thus managed to transmit the information about @xmath71 to her causal past .
dave may then transmit @xmath71 by a sub - luminal signal that reaches alice at @xmath54 and alice could now decide _ freely _ to not measure @xmath71 and measure @xmath80 instead .
therefore , we see that eqs.([eq : alice - signal - bob ] ) and ( [ eq : alice - freewill ] ) have resulted in a causal loop , which would lead to grandfather - style paradoxes . in order to prevent causal loops and preserve the notion of relativistic causality
while still keeping the notion of free will in eq.([eq : alice - freewill ] ) , we therefore impose the no - signaling condition @xmath81 analogous reasoning to prevent a closed causal loop starting from bob s measurement gives the other no - signaling condition @xmath82 we have therefore derived the necessity of the two - party no - signaling constraints from the requirement that there be no closed causal loops .
that these constraints are also sufficient is clear , since these ensure that the causal structure in fig .
[ fig : two - party - caus - struc ] is maintained .
note that while the outputs can be correlated with each other , this correlation is attributed to the underlying ( local or non - local ) hidden variable @xmath33 rather than due to any superluminal influence propagating from one party to another . ) is violated as in eq.([eq : alice - signal - bob ] ) . in the figure
, we consider the violation in terms of bob s output distribution depending on alice s input .
since alice s input @xmath71 was chosen freely at @xmath17 @xmath54 , this dependence implies a superluminal signal containing the information @xmath71 was transmitted from @xmath17 to @xmath52 at @xmath53 , here the signal is shown to travel at infinite speed , i.e. parallel to the @xmath26-axis .
the world - lines of charlie and dave who move at uniform velocity relative to alice and bob are also shown . at @xmath77 , charlie s world line interesects bob who informs charlie of @xmath71 .
charlie then uses the same superluminal signal to send @xmath71 to dave at @xmath79 , this signal travels parallel to the @xmath78-axis as shown .
note that @xmath79 is in the causal past of alice , so dave can send @xmath71 via a subluminal signal to alice .
alice can then freely chose to not measure @xmath71 and measure @xmath80 instead .
thus , we have a closed causal loop resulting in a violation of causality.,scaledwidth=85.0% ] the two ( spatially separated ) events @xmath17 and @xmath52 appear to occur simultaneously , i.e. , @xmath83 . in the inertial reference frame @xmath76 ,
event @xmath52 occurs _ before _
@xmath17 , i.e. , @xmath84 . in another inertial reference frame @xmath85
on the other hand , event @xmath17 occurs before @xmath52 , i.e. , @xmath86 . to maintain causality
, we must have that the output distribution @xmath87 at @xmath17 must be independent of the input @xmath88 at @xmath52 , and similarly @xmath89 must be independent of the input @xmath71 at @xmath17.,scaledwidth=50.0% ] this situation is illustrated in the space - time diagram of the measurement process in fig .
[ fig : two - party - ns ] . as seen in the figure ,
while causal relationships are manifestly lorentz invariant , the specific time sequence of events changes in different inertial frames of reference , in particular spacelike separated events have no absolute time ordering between them . in the @xmath90 reference frame ,
the measurement events of alice and bob occur simultaneously . in the @xmath91 inertial reference frame , bob s measurement precedes that of alice ( @xmath84 ) , while in the @xmath92 reference frame , alice s measurement precedes that of bob ( @xmath86 ) .
we see that the imposition of the constraints in eq.([eq : two - party - ns ] ) prevents any causal loop ( and resulting grandfather paradoxes " ) , or in other words relativistic causality is guaranteed by the two - party no - signaling principle .
note that the freedom - of - choice " condition that alice and bob are allowed to choose their measurements freely ( using a private random number generator , for example ) is necessary in the argument above to identify the choice of measurement by alice as cause and observation by bob of his output distribution as effect .
furthermore , remark that it is essential to consider different inertial reference frames to make the argument , no causality violation would occur if faster - than - light propagation occurred in only one reference frame .
as shown by eberhard in @xcite , the no - signaling requirements are satisfied in both non - relativistic quantum theory and in the relativistic quantum field theory . in quantum field theory
, this requirement would be ensured by the vanishing of commutators of field operators @xmath93 and @xmath94 representing alice and bob s observables respectively , i.e. , @xmath95 = 0 $ ] @xcite . in the non - relativistic quantum theory that is usually used to analyze bell experiments , the description of the alice - bob composite system by a density operator @xmath96 in the tensor product space @xmath97 , the description of quantum operations by local kraus operators acting on the respective hilbert spaces and the partial trace rule ensure that the local statistics only depend on the reduced density matrices of the respective party , and no superluminal communication even using entangled states is possible .
the causal structure of the three - party bell experiment is shown in fig .
[ fig : three - party - caus - struc ] .
alice s spacetime random variables corresponding to her input @xmath16 and output @xmath17 are generated at spacetime location @xmath54 in inerial reference frame @xmath14 , similarly bob s input - output @xmath98 are generated at @xmath53 and charlie s input - output @xmath99 are generated at @xmath100 .
we now investigate the question : * _ what are the necessary and sufficient conditions in the three - party bell scenario that ensure that no causal loops appear _ ?
we shall see that the answer depends upon the exact spacetime locations of the measurement events in the bell experiment .
[ prop : rel - cau - constraints ] consider the three - party measurement configuration shown in fig .
[ fig : three - party - meas - config ] , where in an inertial frame @xmath14 , the spacetime locations of the measurement events @xmath54 , @xmath53 and @xmath100 are such that the intersection of the future light cones of @xmath17 and @xmath101 is contained within the future light cone of @xmath52 . the necessary and sufficient constraints to ensure that no causality violation occurs in this configuration are given by @xmath102 the constraints in eq.([eq : rel - caus-3-party-1 ] ) guarantee that the marginal distributions @xmath103 , @xmath104 , @xmath105 and @xmath106 are well - defined .
firstly , notice that the fact that the marginal @xmath107 is also well - defined for all @xmath108 is guaranteed as a consequence of these constraints by the relation @xmath109 where we used the first and second equalities from eq.([eq : rel - caus-3-party-1 ] ) successively .
the constraint that each party s marginal distribution is well - defined is necessary in order not to violate causality by the two - party result from prop .
[ prop : two - party - ns ] , i.e. , if either alice s or charlie s output statistics exhibited a dependence on bob s input ( @xmath110 or @xmath111 ) then by the prop .
[ prop : two - party - ns ] , bob could signal his input @xmath112 to his causal past and a causal loop would result .
we now move to the two - party distributions . here , compared to the usual no - signaling constraints in eq.([eq : multi - party - ns ] ) , the constraints in eq.([eq : rel - caus-3-party-1 ] ) only ensure that the a - b joint distribution @xmath113 and the b - c joint distribution @xmath114 are well - defined independent of the remaining party s input .
the fact that this is necessary can be seen as follows . in the measurement configuration in fig .
[ fig : three - party - meas - config ] , the spacetime random variable @xmath115 corresponding to the correlations between the outcomes of @xmath17 and @xmath52 manifests itself at the intersection of the future light cones of @xmath17 and @xmath52 , and in particular , @xmath115 can be verified outside the future light cone of @xmath101 .
any dependence of the joint distribution @xmath113 on the input @xmath116 at @xmath101 can then result in a causal loop , following an analogous reasoning to the two - party result in prop .
[ prop : two - party - ns ] .
similarly , we see that it is necessary for @xmath114 to be well - defined independent of the input @xmath117 at @xmath17 .
we now move to see the sufficiency of the four constraints in eq.([eq : rel - caus-3-party-1 ] ) . to see this
, we note that as opposed to the usual sufficient no - signaling constraints ( [ eq : multi - party - ns ] ) , these constraints do not ensure that the marginal @xmath118 is well - defined . in other words , the joint output distribution of alice and charlie
can explicitly depend upon bob s input @xmath119 as @xmath120 even though the marginal distribution of each party does not depend on @xmath121 .
this implies that a superluminal influence propagated from bob s system to the joint system of alice and charlie altering their joint distribution while keeping their marginal distributions unaffected .
the spacetime random variable @xmath122 corresponding to the correlations between the outputs of @xmath17 and @xmath101 only manifests itself at the intersection of the future light cones of @xmath17 and @xmath101 .
briefly , when this intersection is contained within the future light cone of @xmath52 , @xmath122 is timelike separated from @xmath52 and hence any influence of @xmath122 by the choice of input at @xmath52 does not lead to signaling .
more formally , the argument is stated as follows .
we want that the choice of input @xmath123 at @xmath52 does not signal to any space - time location @xmath124 via the change of correlations @xmath122 .
now , two possibilities ( attempts to signal to @xmath124 via @xmath122 ) arise . 1 .
alice at @xmath17 and charlie at @xmath101 transmit their outputs by a light signal at speed @xmath125 to @xmath124 . in this case
, @xmath124 must be contained in the intersection of the future light cones of @xmath17 and @xmath101 .
now , the crucial property of the measurement configuration of fig .
[ fig : three - party - meas - config ] is that the intersection of the future light cones of @xmath17 and @xmath101 is contained within the future light cone of @xmath52 .
therefore , @xmath124 is timelike separated from @xmath52 .
formally , the sum of the time taken for a superluminal influence ( at any arbitrary speed @xmath126 ) to move from @xmath52 to @xmath17 and the time taken for a light signal ( at speed @xmath125 ) to move from @xmath17 to @xmath124 is less than the time taken for a light signal at speed @xmath125 to travel from @xmath52 to @xmath124 directly .
a similar condition holds for the influence traveling via @xmath101 .
mathematically , these constraints are captured by the following equations @xmath127 where @xmath128 denotes the time taken for a signal at speed @xmath129 to travel from @xmath52 to @xmath17 , @xmath130 denotes the time taken for a light signal ( at speed @xmath125 ) to travel from @xmath17 to @xmath124 , and so on .
the fact that @xmath124 is always in the future light cone of @xmath52 ensures that no superluminal transmission of information takes place .
2 . alternatively , alice ( or charlie or both ) transmits her output @xmath17 via a subsequent measurement @xmath131 and a similar superluminal influence as bob did , i.e. , alice superluminally influences the measurement events at two spacelike separated locations @xmath132 and @xmath133 ( with the property that the intersection of the future light cones of @xmath132 and @xmath133 is contained within the future light cone of @xmath17 ) in such a way as to change the joint distribution @xmath134 while retaining the marginal distributions @xmath135 and @xmath136 . if @xmath132 and @xmath133 subsequently send a light signal , then as in the previous argument , the output at @xmath17 is available only in a location @xmath137 in the future light cone of @xmath17 .
the intersection of the future light cones of @xmath137 and @xmath101 is directly seen to be contained within the future light cone of @xmath17 and @xmath101 which is in turn contained within the future light cone of @xmath52 , so that the information about the outputs at @xmath17 and @xmath101 is again only available at a timelike separation from @xmath52 . similarly ,
if @xmath132 and @xmath133 subsequently send a superluminal influence as bob did , then we repeat the argument to see that any concentration of information only occurs within the future light cone of @xmath52 .
we therefore deduce that in measurement configurations such as in fig .
[ fig : three - party - meas - config ] , the constraints in eq.([eq : rel - caus-3-party-1 ] ) are necessary and sufficient as opposed to the usual three - party no - signaling constraints from eq.([eq : multi - party - ns ] ) which while being sufficient are not strictly necessary in order to prevent a violation of causality . as a simple illustrative example of the above , consider a box @xmath138 with binary inputs and binary outputs for the three parties of the form @xmath139 this box deterministically outputs @xmath140 for both inputs @xmath141 and has marginals @xmath142 of the form @xmath143 here @xmath144 is a local box that returns correlated outcomes @xmath145 for any input @xmath146 with uniform marginals @xmath147 for @xmath148 ( such a local box is generated by shared randomness between @xmath17 and @xmath101 ) , and @xmath149 is the popescu - rohrlich box @xcite that returns outcomes satisfying @xmath150 also with uniform marginals @xmath151 and @xmath152 .
we see that such a box @xmath138 explicitly satisfies the relativistic causality constraints from prop .
[ prop : rel - cau - constraints ] since the local marginals @xmath151 , @xmath153 and @xmath152 are well - defined ( independent of the inputs of the other party ) , and the two - party marginals @xmath113 and @xmath114 are given explicitly as @xmath154 which are also well - defined , independent of the input of the remaining party .
this relativistically causal box maximally saturates the three - party mermin - type expression @xmath155 ( even beyond the customary promise on the input set @xmath156 ) , since we have for @xmath157 that @xmath158 and for @xmath159 that @xmath160 .
however , despite this maximal violation , notably in this box , the outputs @xmath52 are always deterministic .
are correlated via a common @xmath33 .
the inputs @xmath161 are chosen freely according to a notion of free - choice ( see section [ sec : freewill ] ) .
the input @xmath16 can not signal to change the distribution of a remote party s output @xmath52 or @xmath101 , analogously for inputs @xmath112 and @xmath162 . on the other hand , depending upon the spacetime configuration of the measurement events , the input may influence the correlations between remote parties outputs without violating causality . in particular , in the measurement configuration shown in fig .
[ fig : three - party - meas - config ] , the input @xmath112 may influence the correlations @xmath122.,scaledwidth=75.0% ] labeled by axes @xmath51 are given by @xmath54 , @xmath53 and @xmath100 respectively .
the correlations between the outputs of alice and bob is denoted by a spacetime random variable @xmath115 which manifests itself ( is checked by alice and bob ) at a spacetime location within the intersection of the future light cones of alice and bob .
similarly , the correlations between the outputs of the other pairs of parties is denoted by @xmath122 and @xmath163 .
the crucial property of this measurement configuration is that the intersection of the future light cones of @xmath17 and @xmath101 is contained within the future light cone of @xmath52.,scaledwidth=85.0% ] having seen that in certain measurement configurations , the set of constraints in eq.([eq : rel - caus-3-party-1 ] ) as opposed to the full set of no - signaling constraints is already necessary and sufficient to ensure no violation of causality , we now identify the spacetime configurations where this occurs . to do this
, we consider the situation where the superluminal influence propagates at some fixed speed @xmath129 in an inertial reference frame @xmath14 .
we slightly alter the notation and consider the situation from a cryptographic perspective with bob @xmath52 replaced by eve @xmath133 with input - output @xmath164 and alice - charlie @xmath165 replaced by alice - bob @xmath166 .
thus , we fix the spacetime coordinates @xmath54 and @xmath53 of alice and bob , and investigate from which spacetime location @xmath167 eve is able to influence the correlations @xmath168 without affecting the marginal distributions @xmath151 and @xmath153 , i.e. , in which measurement configurations ( [ eq : rel - caus-3-party-1 ] ) is necessary and sufficient to prevent the violation of causality .
accordingly , our result is then the following proposition [ prop : spacetime - region ] whose proof is in the supplemental material . [
prop : spacetime - region ] consider measurement events @xmath169 with corresponding spacetime coordinates @xmath170 , @xmath53 in a chosen inertial reference frame @xmath14 .
then a measurement event @xmath133 can superluminally influence the correlations between @xmath17 and @xmath52 at speed @xmath129 without violating causality in @xmath171 if and only if its space coordinate @xmath172 satisfies @xmath173 for any circle @xmath174 with @xmath175 as a chord and having angle @xmath176 as the angle in the corresponding minor segment , where @xmath177 and @xmath178 , and if its time coordinate @xmath179 satisfies @xmath180.\ ] ] outside of the spacetime region in proposition [ prop : spacetime - region ] , the usual no - signaling constraints from ( [ eq : multi - party - ns ] ) turn out to be necessary and sufficient to ensure causality is preserved .
figure [ fig : three - party - caus - struc ] shows the causal structure represented by the three - party bell experiment represented in terms of a directed acyclic graph ( dag ) @xcite .
we can now consider the implications of the spacetime structure on the causal relationship , namely we can consider the study of bayesian networks of spacetime random variables ( srvs ) in light of the considerations of this section , a study which we pursue in future work . recall that a bayesian network , or probabilistic directed acyclic graphical model is a probabilistic graphical model ( a type of statistical model ) that represents a set of random variables and their conditional dependencies via a directed acyclic graph ( dag ) .
now , in the dag that represents a causal structure in the three - party scenario , an additional ingredient of a new type of edge representing a causal link between the spacetime variable @xmath112 and the effects @xmath181 would be added depending on the spacetime location of the measurement events .
this edge represents the causal link that @xmath112 does not influence the marginal distributions of @xmath17 and @xmath101 themselves but changes the joint distribution of @xmath182 .
the additional ingredient when considering spacetime variables is that the random variable @xmath122 representing the correlations between @xmath17 and @xmath101 is only created in the future light cone of @xmath112 provided that the intersection of the future light cones of @xmath17 and @xmath101 is contained within the future light cone of @xmath112 .
the causal network corresponding to a particular spacetime arrangement of measurement events would incorporate the additional edges as possible influences that do not lead to superluminal signaling and respect causality .
we now extend to the general @xmath0-party scenario the considerations of the previous subsections . in the mutli - party case ,
different subsets of the usual no - signaling constraints are sufficient to preserve causality , depending on the spacetime locations of the measurement events of the @xmath0 parties . here
, we focus on an extension in the simplest @xmath183-dimensional scenario , when the parties are arranged in a line in some reference frame .
consider @xmath47 parties with respective fixed laboratory space coordinates @xmath184 in frame @xmath185 .
let us identify the possible space - time region @xmath186 from which a party @xmath0 can influence the correlations between the systems of the @xmath47 parties .
clearly , the time coordinates @xmath187 of the @xmath47 parties at which the influence traveling at @xmath188 in frame @xmath185 can be felt are given by @xmath189.\ ] ] by the results of the previous section , we know that to influence two - party correlations for parties @xmath4 and @xmath190 where @xmath191 $ ] , @xmath192 must belong to @xmath193 all points in the intersection of @xmath194 over all pairs @xmath195 $ ] , i.e. , @xmath196 satisfy the two conditions : ( i ) @xmath197 is contained in the future @xmath188-cone of @xmath198 for all @xmath199 $ ] , and ( ii ) the intersection of the future light cones of the @xmath47 parties is entirely contained within the future light cone of the @xmath0-th party . in particular , when the @xmath0 parties are in a line , the relativistic causality constraints ( the subset of the usual no - signaling constraints needed to ensure that causality is not violated ) are given as follows .
let @xmath200 $ ] denote a contiguous subset of @xmath201 = \{1 , \dots , n\}$ ] of size @xmath3 with initial element @xmath2 , i.e. , @xmath202 for some @xmath203 . clearly , the number of such contiguous subsets for fixed @xmath3 is @xmath204 , and the total number of contiguous subsets of all possible sizes @xmath3 is @xmath205 . for a string of outputs @xmath206 of the @xmath0 parties , let @xmath207 denote the substring of outputs of the parties belonging to the set @xmath208 ,
i.e. , @xmath209 with @xmath210 and let @xmath211 denote the outputs of the complementary set of parties . similarly , let @xmath212 denote the inputs of the parties in the set @xmath208 and @xmath213 denote the inputs of the complementary set .
then the relativistic causality constraints for the @xmath0 parties stationed in @xmath214-d are given by @xmath215 for all @xmath216 with @xmath217 and for all @xmath218 with @xmath219 .
in other words , the marginal distribution of the outputs for a contiguous subset of parties inputs is independent of the complementary set of parties inputs . on the other hand , for a non - contiguous subsets of parties , the joint probability distribution of their outputs can depend on the inputs of the complementary set , i.e. , the parties in between can change the marginal distributions by their choice of inputs while still respecting relativistic causality .
manifest themselves at the intersection of the future light cones of @xmath220 .
for instance , as shown in the figure , the correlations @xmath221 are seen in the intersection of @xmath30 and @xmath222 which in this measurement configuration lies entirely within the future light cone of @xmath223 .
as such , any dependence of the joint distribution of outcomes @xmath30 and @xmath222 on the measurement input at @xmath223 does not lead to any causal loops as shown in the proposition [ prop : rel - cau - constraints].,scaledwidth=85.0% ] the causal structure of the @xmath0-party bell experiment is depicted in fig .
[ fig : n - party - caus - struc ] , where the state of the system is denoted by @xmath33 , the input and output of the @xmath4-th party by @xmath224 for @xmath225 $ ] . -party bell experiment .
the outputs @xmath8 for @xmath225 $ ] are correlated via a common @xmath33 , while the inputs @xmath5 are chosen freely according to a notion of free - choice.,scaledwidth=50.0% ]
before we proceed , let us examine the relationship of the present work with previous studies on relativistic causality in the foundations of quantum theory .
the paradigmatic example of superluminal explanations for correlations is the de broglie - bohm pilot - wave theory @xcite . here ,
the wave function of the system is supplemented by the specification of hidden particle positions which evolve according to a _ guidance equation_. the velocity of any particle can depend upon the positions of other distant particles , and the theory is manifestly nonlocal .
besides positing a privileged reference frame , the theory also contradicts our usual notion of free choice @xcite ( as for example formalised in @xcite ) . in @xcite
, the authors studied the possibility of explaining quantum correlations using influences propagating at a finite speed @xmath226 ( denoted as @xmath50-causal models ) defined with respect to a privileged reference frame .
it was shown that in the multi - party bell scenario , if one assumes the usual no - signaling conditions eq.([eq : multi - party - ns ] ) and the corresponding free - choice conditions , any such @xmath50-causal model allows correlations which lead to superluminal signaling .
recently , @xcite looked at a new framework for multi - party correlations without a pre - defined global causal structure ( no explicit space - time structure as considered here ) but with the validity of quantum mechanics assumed locally . using _
process matrices _
( a generalization of the usual density matrices ) as fundamental entities , it was shown that there exist process matrices that are compatible with local quantum mechanics which are however causally nonseparable , i.e. , incompatible with a global causal order .
this paper , in contrast , follows a suggestion of shimony @xcite , followed by grunhaus , popescu and rohrlich @xcite of jamming " non - local correlations , i.e. , the possibility of non - local correlations that go beyond the usual quantum formalism , and yet do not lead to any contradictory causal loops , that therefore might appear in post - quantum theories .
recall that the lorentz symmetry is a fundamental principle of nature stating that the laws of physics stay the same in all inertial reference frames , or more formally , that the laws must be expressed in terms of lorentz covariant quantities .
now , the hypothetical influences that we considered in the derivation of the relativistic causal constraints propagate at speeds @xmath129 in some chosen inertial reference frame .
one might wonder if the resulting theory can still be lorentz invariant .
firstly , as we have argued , the superluminal influence does not lead to _ superluminal transmissions of information _ , which are what relativity theory actually prohibits , since they can lead to causal loops and grandfather paradoxes .
secondly , observe that the conditions imposed by relativistic causality in proposition [ prop : rel - cau - constraints ] are manifestly lorentz covariant , i.e. , _ if the intersection of the future light cones of @xmath17 and @xmath101 is contained within the future light cone of @xmath52 in one inertial reference frame , then this intersection is contained in the future light cone of @xmath52 in all inertial reference frames_. indeed , this is the reason why these constraints do not allow for a violation of causality .
the speed @xmath129 of the superluminal influence is specified in the chosen inertial frame @xmath14 , however the lorentz transformations do not transform into superluminal frames . the fact that special relativity does not prohibit such faster - than - light influences has been known since the seminal papers by bilaniuk , deshpande and sudarshan @xcite and feinberg @xcite .
the standard approach is to not consider transformations into the rest frame of a particle moving at speed @xmath129 , so long as all the predictions for the usual subluminal inertial reference frames are lorentz covariant as considered here . in this respect ,
the theory resembles the paradigmatic non - local theory , namely the de broglie - bohm theory which also only makes lorentz covariant predictions .
however , the de broglie - bohm theory allows for instantaneous action - at - a - distance throughout all spacetime and is a non - local deterministic theory .
in contrast , a theory incorporating the relativistic causality constraints manifestly allows for a consistent notion of free - choice ( and consequently intrinsic randomness ) which is detailed in the following section [ sec : freewill ] .
finally , we remark that if one wishes to also consider an extended lorentz transformation into superluminal frames of reference , one might follow the approach of @xcite who show that by re - deriving lorentz - like transformations for the superluminal case ( rather than merely substituting into the usual subluminal transformations ) one arrives at a lorentz factor @xmath227 which makes all proper physical quantities real - valued for superluminal frames . also note that very recently , hill and cox @xcite have developed an
extended version of special relativity that uses such a real - valued lorentz factor for @xmath226 .
recall the freedom - of - choice assumption as stated in def .
[ def : freewill ] .
this notion of free - choice was captured by the requirement that each party s input @xmath5 be uncorrelated with any random variable @xmath162 that is not in its causal future ( denoted by @xmath228 ) , where @xmath17 is said to be in the causal future of @xmath52 if @xmath229 in all inertial reference frames . in other words , when @xmath228 , we impose @xmath230 .
consider the causal structure of the @xmath0-party bell experiment in fig .
[ fig : n - party - caus - struc ] , where the state of the system is denoted by @xmath33 , the input and output of the @xmath4-th party by @xmath224 for @xmath225 $ ] . in @xcite
, it is claimed that if for all @xmath225 $ ] , @xmath5 are free according to def .
[ def : freewill ] , then @xmath231 obeys the usual no - signaling constraints in eq.([eq : multi - party - ns ] ) . with @xmath232 and @xmath233 , the formulation of the condition that @xmath5 are free in @xcite is as follows : [ lem : cr - freewill ] consider the @xmath0-party bell experiment , where the state of the system is denoted by @xmath33 , the input and output of the @xmath4-th party by @xmath224 for @xmath225 $ ] .
then @xmath5 is free according to def .
[ def : freewill ] if the following condition is satisfied : @xmath234 but this notion of freedom - of - choice in @xcite misses the slight subtlety raised in this paper which def .
[ def : freewill ] allows for .
namely , the correlation between two ( or more ) output spacetime random variables is a spacetime random variable @xmath235 in its own right , with the associated spacetime location being in the intersection of the future light cones of the two ( or more ) output spacetime random variables @xmath236 themselves .
then , a party s input @xmath5 does not need to be independent of the joint distribution ( of correlations ) of output srv s if the correlation only manifests itself within the future light cone of @xmath5 , even though @xmath5 is required to be independent of the individual marginal distributions of each output srv themselves .
accordingly , one can have a non - local effect precede a cause in the sense that one ( or even both ) of the outputs @xmath23 or @xmath24 may occur before the @xmath5 was chosen .
this has important consequences as we shall see below .
we propose the following equivalent reformulation ( different from the formulation in @xcite ) of the free - choice definition [ def : freewill ] in the multi - party bell scenario .
[ lem : mod - freewill ] consider the @xmath0-party bell experiment , where the state of the system is denoted by @xmath33 , the input and output of the @xmath4-th party by @xmath224 for @xmath225 $ ] .
let @xmath237 denote a set of outputs @xmath23 such that the correlation srv @xmath238 between _ all _ the outputs @xmath23 is generated outside the future light cone of @xmath5 .
then @xmath5 is free according to def .
[ def : freewill ] if the following condition is satisfied : @xmath239 let us illustrate the definition [ lem : mod - freewill ] by explicitly stating the free - choice conditions in the three - party bell scenario with causal structure shown in fig .
[ fig : three - party - caus - struc ] and measurement configuration shown in fig.[fig : three - party - meas - config ] . in the three - party bell scenario with causal structure shown in fig .
[ fig : three - party - caus - struc ] and measurement configuration shown in fig.[fig : three - party - meas - config ] with the intersection of @xmath17 and @xmath101 s future light cones falling within the future light cone of @xmath52 , we have the following free - choice constraints @xmath240 we can now deduce a formal condition following @xcite stating that incorporating the three - party free - choice conditions in eq.([eq : three - party - freewill ] ) in a theory requires that the theory obey the three - party causality constraints in eq.([eq : rel - caus-3-party-1 ] ) . [
prop : caus - eq - freewill ] consider the causal structure of the @xmath241-party bell experiment in fig .
[ fig : three - party - caus - struc ] , where the state of the system is denoted by @xmath33 , the input and output of alice , bob and charlie are denoted by @xmath242 , @xmath243 and @xmath244 respectively .
if the input variables @xmath245 and @xmath162 are free according to definition [ def : freewill ] , i.e. , obey eq.([eq : three - party - freewill ] ) , then @xmath246 obeys the relativistic causality constraints in eq.([eq : rel - caus-3-party-1 ] ) .
the proof follows by rewriting @xmath247 here , the first equality follows by bayes rewriting and the second equality uses the free - choice constraints from eq.([eq : three - party - freewill ] ) that @xmath248 .
analogously , we can use the free - choice constraints @xmath249 to show that @xmath250 .
this recovers the two - party marginals in eq.([eq : rel - caus-3-party-1 ] ) .
similarly , the single - party marginals can also be deduced as follows .
@xmath251 here again the first equality follows by bayes rewriting and the second equality uses the free - choice constraints that @xmath252 and @xmath253 .
analogously , one can deduce that @xmath254 .
the fact that the single - party marginal @xmath255 is well - defined follows as in eq.([eq : three - party - b - marginal ] ) .
note that the distribution @xmath256 does not in general equal @xmath257 , with @xmath258 this is because in general @xmath259
since the correlation @xmath260 only manifests itself within the future light cone of the measurement event at @xmath52 . as such , the amount by which the free - choice condition is violated @xmath261 is exactly equal to the amount by which the no - signaling constraint is violated @xmath262 . the considerations of the three - party scenario in prop .
[ prop : caus - eq - freewill ] can also be extended straightforwardly to the @xmath0-party scenario .
consider the causal structure of the @xmath0-party bell experiment in fig .
[ fig : n - party - caus - struc ] for the measurement configuration in fig.[fig : n - party - meas - config ] , where the state of the system is denoted by @xmath33 , the input and output of the @xmath4-th party by @xmath224 for @xmath225 $ ] . if , for all @xmath225 $ ] , @xmath5 are free according to def .
[ def : freewill ] , then @xmath231 obeys the relativistic causality constraints in eq.([eq : rel - caus ] ) . in @xcite , reichenbach proposed a principle of common cause in an attempt to characterize the asymmetry of time by exploiting a statistical distinction between cause and effect .
principle @xcite ( in the classical setting ) states that if two events @xmath17 and @xmath101 are statistically correlated , then either @xmath17 causes @xmath101 or @xmath101 causes @xmath17 or they have a cause operative in their common past , where a classical common cause is a shared random variable @xmath112 from which the correlations derive .
in other words , we have that @xmath263 , i.e. , @xmath17 and @xmath101 are independent conditional upon the common cause @xmath112 so that this feature could be used to define the distinction between past and future . the notion of common cause
can also be extended to the non - classical setting with the shared random variable @xmath112 replaced by a quantum state @xmath264 .
this is akin to the notion of outcome independence from section [ subsec : two - party - bell - theorem ] , where the common cause of the correlations in the two - party bell experiment is denoted by @xmath33 .
more formally , we have the following definition [ def : rccp ] for a classical probability measure space @xmath265 , let @xmath266 and @xmath267 be two positively correlated events in @xmath268 , i.e. , @xmath269 .
@xmath270 is said to be a common cause of the correlations between @xmath17 and @xmath101 if the following conditions hold : @xmath271 the condition can also be extended to include a _ system _ of cooperating common causes .
a partition @xmath272 in @xmath268 is said to be the ( reichenbachian ) common cause system of the correlation between @xmath17 and @xmath101 if the screening condition @xmath273 holds for all @xmath274 . in this case ,
the common causes @xmath275 are in the common causal past of @xmath17 and @xmath101 , i.e. , in the intersection of the past light cones of @xmath17 and @xmath101 .
now , consider that @xmath17 and @xmath101 are the output srv s of measurements by alice and charlie , and let @xmath112 be the measurement input of bob .
as we have seen , relativistic causality allows , in certain spacetime configurations , for @xmath112 to influence the correlations between @xmath17 and @xmath101 without affecting their marginal distributions .
still in the classical setting , denoting this influence by @xmath276 , we obtain the screening conditions @xmath277 with the restriction on the marginals @xmath278 this extension of reichenbach s principle from def .
[ def : rccp ] allows the srv s @xmath279 to lie in a region that is even outside the common causal past of @xmath17 and @xmath101 so long as the causality constraints in ( [ eq : rel - caus-3-party-1 ] ) are obeyed .
the considerations of the previous sections have important implications for post - quantum cryptographic tasks in the device - independent scenario , where the honest parties are not assumed to know the exact internal workings of their device and the eavesdropper is only constrained by the laws of relativity , see for example @xcite .
in particular , two important considerations appear . 1 .
firstly , note that the set of boxes obeying the relativistic causality considerations forms a larger dimensional polytope than the usual no - signaling polytope .
this confers a larger set of attack strategies for an eavesdropper who may prepare boxes for the honest parties from this larger set .
as shown in proposition [ prop : ra - rel - caus ] , in certain known device - independent protocols for the cryptographic task of randomness amplification @xcite , this larger set of attack strategies can severely compromise the security of the protocol , the honest parties were to perform the required bell test in a spacetime measurement configuration where the superluminal influence can take effect .
secondly , and crucially , the security of cryptographic protocols ( even relying on a two - party bell test ) can be compromised when the measurement event of eve s system happens in a suitable spacetime location . in particular , as shown in proposition [ prop : qkd - sec - rel - caus ] , the property of monogamy of non - local correlations can break down under the relativistic causality constraints so that such an eavesdropper can obtain full information about the output of the honest parties in the protocol .
remark that the first type of attack strategy above may be circumvented if the honest parties perform their measurements in a carefully chosen measurement configuration where the usual no - signaling constraints ( [ eq : multi - party - ns ] ) are both necessary and sufficient .
in contrast , the second type of attack can only be avoided if certain assumptions are made about the space - time location of the eavesdropper s measurement event , or alternatively if the honest parties systems are assumed to be sufficiently shielded from all influences , even those respecting causality .
the implications manifest as an adversarial attack strategy in device - independent protocols for randomness generation , amplification and key distribution when the adversary is only subject to the laws of relativity such as considered in @xcite as opposed to the scenario when the adversary is in addition assumed to obey the laws of quantum theory @xcite .
the key step in such a device - independent protocol is a test for the violation of a bell inequality .
multiple observers , who are spatially separated , choose random inputs on their parts of a device and check that the outcomes conform to the violation of some specific bell inequality .
the notion of device - independence then stems from the black - box scenario , the parties do not have access to the inner workings of the device and the security of the protocol is assessed solely based on the input - output statistics of the device , so that the device may have been prepared by the adversary eve herself .
mathematically , the device in any single run of the protocol is described by the set of conditional probability distributions ( box ) @xmath280 . here
@xmath281 and @xmath282 denote the outputs and inputs of the device for the @xmath0 honest parties in the protocol .
the parties check the input - output statistics to assess the value of a bell expression and make an inference about @xmath283 , where @xmath14 is an indicator vector for the bell expression that picks out the suitable input - output pairs .
the proof of security of such protocols relies on the inference drawn from the observed bell violation regarding the amount of randomness present in the output .
this is achieved by checking that for all non - signaling boxes @xmath284 that achieve this bell value , for some function @xmath285 of the outputs for some set of inputs @xmath286 , it happens that @xmath287 , where @xmath288 @xcite . as an example , in the protocol of barrett , hardy and kent @xcite for quantum key distribution , the observation of correlations between alice and bob s outputs as in the braunstein - caves - pearle chained bell inequality @xcite leads to the conclusion that in the limit of a large number of runs , each party s output was necessarily uniformly random .
this is because all no - signaling boxes that maximally violate the chained bell inequality possess the property of uniformly random marginals for each party .
similar considerations also hold for the protocols considered in @xcite where the maximal violation of a multi - party inequality belonging to the ghz - mermin family @xcite , leads to the conclusion that a particular function of the outputs ( the majority of a subset of the parties output bits in this case ) gives rise to randomness . now , from the considerations of the previous sections , we see that there is a larger set of boxes that respect relativistic causality than those belonging to the usually considered no - signaling polytope , depending on the spatial locations of the parties .
this implies that an adversary who is aware of the spatial locations of the parties involved in the protocol may provide them with a box that satisfies only the relativistic causality conditions stated in this paper .
in particular , when the @xmath0 parties are in a line , the relativistic causality constraints ( the subset of the usual no - signaling constraints needed to ensure that causality is not violated ) are given from section [ subsec : multi - party - rel - caus ] as @xmath289 for all @xmath216 with @xmath217 and for all @xmath218 with @xmath219 .
recall that here @xmath200 $ ] denotes a contiguous subset of @xmath201 = \{1 , \dots , n\}$ ] of size @xmath3 with initial element @xmath2 , i.e. , @xmath202 for some @xmath203 and for a string of outputs @xmath206 of the @xmath0 parties , @xmath207 denotes the substring of outputs of the parties belonging to the set @xmath208 , i.e. , @xmath209 with @xmath210 and let @xmath211 denotes the outputs of the complementary set of parties .
in other words , the marginal distribution of the outputs for a contiguous subset of parties inputs is independent of the complementary set of parties inputs . on the other hand , for a non - contiguous subsets of parties , the joint probability distribution of their outputs can depend on the inputs of the complementary set , i.e. , the parties in between can change the marginal distributions by their choice of inputs while still respecting relativistic causality .
the boxes @xmath290 are thus only required to obey the restricted set of constraints imposed in eq.([eq : rel - caus-2 ] ) as opposed to the usual no - signaling constraints ( which posit that the marginal distribution of _ every _ subset of parties outputs is independent of the input of the complementary set of parties ) .
the boxes @xmath290 under the reduced set of constraints constitute a set of enhanced attack strategies for an eavesdropper assumed to only obey the causality constraints imposed by relativity in a device - independent cryptographic protocol .
note that the set of boxes still forms a polytope in a larger dimensional space ( the number of relativistic causality constraints is smaller than the number of usual no - signaling constraints ) .
furthermore , note that not all the constraints in eq.([eq : rel - caus-2 ] ) are independent .
we illustrate this attack strategy with a fairly generic example , namely we will show that this enhanced set of attack strategies allows the adversary to ensure that the parties can not extract any randomness from the outputs ( for inputs appearing in the inequality ) , from the violation of the quintessential multi - party bell inequalities , namely the ghz - mermin inequalities @xcite .
the mermin inequality is set in the bell scenario when each of @xmath0 parties ( for @xmath0 odd , @xmath291 ) measures one of two inputs @xmath292 and obtains one of two outputs @xmath293 .
the inequality is given as the following set of constraints on the @xmath0-party correlators @xmath294 for @xmath295 , with @xmath296 , @xmath297 @xmath298 here , the @xmath0-party correlation function @xmath294 is defined as @xmath299 when @xmath300 is even , no constraints are imposed on the corresponding correlator .
measurements by each of the @xmath0 parties of the pauli @xmath301 and @xmath302 operators ( for @xmath303 respectively ) on the @xmath0-qubit ghz state @xmath304 result in a violation of the inequality up to its maximum value , i.e. , satisfies all the constraints in eq.([eq : mermin - corr ] ) .
the inputs @xmath286 for which constraints are imposed on the correlator are said to appear in the inequality , this set of inputs is denoted by @xmath305 . in @xcite , it was shown that satisfying eq.([eq : mermin - corr ] ) implies , in the asymptotic setting of an infinite number of parties @xmath306 , that a particular function of the output bits is fully random . in particular , the following function of the outputs @xmath307 was considered for any input @xmath286 that appears in the mermin inequality .
@xmath308 as the number of parties @xmath306 it was shown that @xmath309 , implying that for all boxes satisfying the usual multi - party no - signaling conditions and the mermin constraints eq.([eq : mermin - corr ] ) , the bit defined by the function @xmath310 possesses full intrinsic randomness and defines a process where full randomness amplification takes place .
we show in the following proposition that this conclusion no longer holds when in place of the usual no - signaling conditions , only the relativistic causality conditions are taken into account .
furthermore , we show that when considering the inputs appearing in the mermin inequality , _ no function _ of the output bits possesses _ any _ randomness for all odd @xmath291 .
we leave as an open question whether there exists any multi - party bell inequality with the property of maximum violation such that all the boxes obeying the new relativistic causality conditions admit a hashing function that defines a ( partially ) random bit .
[ prop : ra - rel - caus ] consider the @xmath0-party ghz - mermin bell inequality , for odd @xmath291 .
suppose that in some inertial reference frame , the @xmath0 space - like separated parties are arranged in @xmath214-d , with @xmath311 and perform their measurements simultaneously , i.e. , @xmath312 .
then for any input @xmath313 appearing in the inequality , i.e. , @xmath314 , there exists a box @xmath315 violating the mermin inequality maximally and obeying the relativistic causality constraints in eq.([eq : rel - caus ] ) , such that no randomness can be extracted from the outputs @xmath206 of the box under input @xmath313 .
in other words , we have @xmath316 for some fixed output bit string @xmath317 .
see the supplemental material .
the considerations of the previous subsection also extend to multi - party protocols for device - independent key distribution against relativistic eavesdoppers , namely that the boxes that the honest parties use in such protocols are only subject to the subset of the no - signaling constraints imposed by relativistic causality .
here , we focus on the second important consideration highlighted at the beginning of this section , namely the compromised security of two - party protocols due to the failure of monogamy of non - local correlations under the relativistic causality constraints when the eavesdropper s measurement event is at an appropriate spacetime location . in a qkd protocol , we model the initial correlations between alice , bob and an eavesdropper eve by a distribution @xmath138 . in the device - independent setting , this distribution is a priori unknown and may have been chosen by the adversary .
we consider the situation when the security claims are supposed to hold for any possible initial distribution compatible with the constraints imposed by relativity .
alice and bob have access to a public authenticated communication channel so that any information sent through this channel is available to eve . in each step of the protocol , alice and bob either access correlated data , perform local operations or exchange messages @xmath318 over the public authenticated channel and in the final step , they generate the keys @xmath319 .
the protocol is said to be secure if the resulting distribution @xmath320 is indistinguishable from an ideal one @xmath321 , where @xmath322 is the marginal distribution obtained from @xmath320 and @xmath323 indicates that alice and bob s versions of the secret key are identical and uniformly distributed independent of @xmath324 . in the notion of universally composable security , the distinguishability of these two distributions , quantified by @xmath325 ,
can be made arbitrarily small for large @xmath326 , so that when the secret key is used in another subsequent cryptographic task , the composed protocol is as secure as if an ideal secret key was used . here
@xmath327 denotes the total variation distance @xmath328 . here , we focus on the key distribution protocols from @xcite that were proven secure against eavesdroppers obeying the usual no - signaling constraints and examine what happens when we modify these to the relativistically causal constraints .
these protocols are based on the quantum violation of the braunstein - caves - pearle chain inequalities @xcite .
the chained bell inequalities are a family of two - party correlation bell inequalities ( xor games ) with @xmath329 inputs and two outputs per party , that generalize the well - known chsh inequality . as usual , labeling the inputs of alice and bob by @xmath330 $ ] respectively , and the outputs by @xmath331 , the chain bell expression @xmath332 is explicitly written as @xmath333 \\ x = y \ ; \vee \ ; x = y + 1 } } p_{a , b|x , y}(a \oplus b = 1 | x , y ) \right ) + p_{a , b|x , y}(a \oplus b = 0 | 1 , m ) \geq 1.\ ] ] the minimum value of the chain expression in eq.([eq : m - chain ] ) within classical theories is @xmath214 . in general
no - signaling theories obeying eq.([eq : two - party - ns ] ) , the algebraic minimum of @xmath334 can be achieved .
measurements on a quantum state can give rise to correlations that achieve the minimum quantum value of @xmath335 , which approaches the algebraic minimum of @xmath334 for large @xmath2 .
this value is achieved when alice and bob share the maximally entangled state @xmath336 and perform the measurements corresponding to the basis @xmath337 where @xmath338 \nonumber \\
\theta_y = \frac{\pi(2y-1)}{2 m } \ ; \ ; \ ; \forall y \in [ m]\end{aligned}\ ] ] denote alice and bob s measurement angles .
the significance of the quantum violation of the chained bell inequalities for generating secure keys is that , in the limit of large @xmath2 , the correlations that achieve the minimum quantum value become monogamous and uniform . in other words , consider any distribution @xmath138 obeying the usual tripartite no - signaling constraints ( where @xmath339 denote the input and output of eve on her system ) for which @xmath340 is close to @xmath334 .
then , for any choice of input @xmath341 by eve , her output @xmath101 is virtually uncorrelated with @xmath17 and @xmath103 is virtually indistinguishable from uniform , for all @xmath342 .
this fact that the violation of the chain inequality leads to private randomness and secure key in the outputs of alice and bob for large number of inputs @xmath2 is captured by the following proposition [ prop : chain - sec - ns ] .
[ prop : chain - sec - ns ] for any distribution @xmath343 obeying the usual multi - party no - signaling constraints ( [ eq : multi - party - ns ] ) in which @xmath17 and @xmath52 are binary , there holds @xmath344 for all @xmath345 $ ] , and @xmath346 , and @xmath347 for all @xmath342 , @xmath345 $ ] and @xmath346 . here
@xmath348 denotes the violation of the chain inequality by the marginal box @xmath113 and @xmath349 denotes the uniform distribution , i.e. , @xmath350 for all @xmath342 , @xmath351 $ ] .
this is also sometimes equivalently stated as the classical mutual information between the outputs @xmath17 and @xmath101 of alice and eve being bounded by the violation of the chain inequality @xmath352 .
we now show that the above proposition [ prop : chain - sec - ns ] does not hold for relativistic causal distributions @xmath353 in the measurement configuration of fig .
[ fig : three - party - meas - config ] , thus implying that the security of qkd protocols such as @xcite that are based on this proposition may be compromised when eve happens to perform her measurement at an appropriate spacetime location .
[ prop : qkd - sec - rel - caus ] consider a two - party qkd protocol where alice and bob perform a test of the braunstein - caves - pearle chained bell inequality @xmath354 ( [ eq : m - chain ] ) with @xmath329 of inputs per party , and eve measures a single input @xmath355 .
suppose that in some inertial reference frame , the three space - like separated parties are arranged in the measurement configuration of fig .
[ fig : three - party - meas - config ] , i.e. , in @xmath214-d with @xmath356 and @xmath357 with the intersection of @xmath17 and @xmath101 s future light cones contained within the future light cone of @xmath52 . for fixed @xmath2 , and any chosen input @xmath121 of bob
, there exists a relativistic causal box @xmath358 such that @xmath359 . \ ] ] the proof is by explicit construction of the box @xmath358 that satisfies the conditions in ( [ eq : bc - rel ] ) .
simply , this box is defined by the entries @xmath360 by construction , the marginal box @xmath361 outputs correlated answers for all input pairs @xmath362 except for @xmath363 for which it outputs anti - correlated answers so that @xmath364 . also , by construction each marginal @xmath365 is uniform and the marginals @xmath361 and @xmath366 are well - defined , so that the box obeys the relativistic causality constraints for this scenario .
finally , bob s output is perfectly correlated with the output of eve , so that her guessing probability of his seemingly random output is @xmath214 inspite of the fact that the chain inequality is maximally violated . also note that during the key run , as usual in a qkd protocol @xcite , alice and bob announce their inputs whence eve is able to guess alice s output for any of her inputs also with unit probability .
we remark that while the attack in proposition [ prop : qkd - sec - rel - caus ] is allowed within the laws of relativity in that it does not lead to any causal loops , in a cryptographic scenario one may additionally assume that the honest parties perform their bell test within shielded laboratories , where the assumption of shielding ( which ensures that no influence , superluminal or otherwise , is transmitted from their systems to eve s and vice versa ) is formally equivalent to imposing the no - signaling constraints on the distributions @xmath138 .
additionally , the honest parties may also vary ( randomly ) the timing of their measurement events to attempt to ensure that eve s measurement happens outside the space - time region in proposition [ prop : spacetime - region ] .
we also remark that in such applications , the boxes should be labeled by the set of space - time coordinates of the measurement events @xmath367 to deduce the relativistic causality constraints obeyed by the box ( in each run of the bell test ) .
the general properties of theories that obey the no - signaling constraints ( [ eq : multi - party - ns ] ) were studied in @xcite .
a number of properties that were considered to be quintessentially quantum such as monogamy of correlations , no - cloning , randomness , secrecy etc . were found to be present in such theories .
as we have seen , in general the set of relativistically causal correlations is a larger set than the set of no - signaling correlations .
paradoxically , just as the relaxation to the minkowski causality conditions lead to a larger set of attack strategies for an eavesdropper ( and a consequent loss in extractable randomness ) , we find that the paradigmatic property of monogamy of non - local correlations which was hitherto thought to be generic in all theories compatible with relativity may be circumvented in specific spacetime measurement configurations .
we now show that the property of monogamy of non - local correlations is significantly weakened under the relativistic causality constraints and the feature of monogamy of correlations violating the chsh inequality even disappears in certain spacetime configurations . specifically , we consider the three - party bell scenario where the parties alice , bob and charlie perform two binary outcome measurements @xmath368 and obtain outcomes @xmath369 respectively .
we label the corresponding binary observables of each party by @xmath370 and @xmath371 respectively .
we consider the well - known chsh inequality @xcite between alice - bob and bob - charlie .
the chsh expression @xmath372 reads as @xmath373 where as usual @xmath374 . in a local hidden variable theory ,
the value @xmath372 is bounded by @xmath375 .
the well - known popescu - rohrlich ( pr ) box is a two - party no - signaling box that achieves a value of @xmath376 for the expression . in the three - party scenario , under the usual no - signaling constraints , the non - local correlations exhibit a phenomenon of monogamy that is captured by the relation @xcite @xmath377 in other words , when alice - bob observe maximum violation of the chsh inequality , no correlations can occur between the observables of bob and charlie .
this phenomenon of monogamy of correlations has found application as the underlying feature that is responsible for the security of many device - independent cryptographic protocols . by observing a sufficiently high violation of the bell inequality ,
alice and bob are able to ensure that their systems are not highly correlated with any system held by a third party such as an eavesdropper .
we find however that under the relativistically causal constraints that occur when the three parties are arranged in @xmath214-dimension , the phenomenon of monogamy is considerably weakened in general and in the above mentioned bell scenario , it completely disappears .
this is captured by the following proposition .
[ prop : rel - caus - mono ] consider a three - party bell scenario , with alice , bob and charlie each performing two measurements @xmath368 of two outcomes @xmath369 respectively .
suppose that in some inertial reference frame , the three space - like separated parties are arranged in @xmath214-d , with @xmath378 and perform their measurements simultaneously , i.e. , @xmath356 .
then , there exists a three - party relativistically causal box @xmath379 such that @xmath380 see the supplemental material .
in @xcite , the question was considered whether quantum correlations somehow arise from outside spacetime or can the correlations be explained by causal ( even if superluminal ) influences propagating continuously in space .
the possibility of
@xmath50-causal theories " was considered , i.e. , theories incorporating a finite superluminal influence propagating at speed @xmath226 in a privileged reference frame . for two - party bell experiments ,
the best one can hope to do is to experimentally lower - bound such a @xmath50 by testing bell inequality violations between systems in laboratories farther apart and better synchronized .
however , in the multi - party scenario , the authors of @xcite showed that surprisingly the strength of multi - party quantum nonlocality is large enough that any @xmath50-causal model that attempts to explain the quantum correlations also gives rise to predictions that can lead to superluminal communication . the argument used to reach
this conclusion relied on a special kind of bell inequality , termed hidden influence " inequalities . here , we re - examine this argument in light of the formulation of relativistic causality and free - choice constraints in this paper .
in particular , two measurement configurations were considered : a four - party bell experiment in @xcite and a three - party bell experiment in @xcite . in the four - party bell experiment
@xcite with spacelike separated parties @xmath17 , @xmath52 , @xmath101 and @xmath132 , the measurement configuration was as shown in figure [ fig : four - party - v - caus ] .
here , in addition to the usual light cones , the @xmath50-causal model also gives rise to a past and future @xmath50-cone in a privileged reference frame . in these models ,
two events that are causally related by a @xmath50 signal could be correlated while correlations between events that are outside each other s @xmath50 cones are assumed to be local .
a particular dichotomy theorem was proven : namely that either the statistics of @xmath381 is local or the no - signaling conditions are violated .
this was based on the following hidden influence " inequality @xcite @xmath382 which is satisfied by all no - signaling boxes @xmath383 for which the conditional boxes @xmath384 are local . the hidden influence inequality ( [ eq : hidden - inf - ineq ] ) is violated by quantum boxes which achieve a value @xmath385 giving rise to the conclusion that quantum correlations can not be explained by finite - speed @xmath50-causal influences .
now , by eq .
( [ eq : rel - caus ] ) , the set of constraints that are necessary and sufficient to preserve causality in the measurement configuration of fig .
[ fig : four - party - v - caus ] is given by the condition that the marginal distribution of the outputs for any contiguous subset of parties is well - defined , i.e. , that @xmath386 , @xmath387 , @xmath388 , @xmath389 , @xmath390 , @xmath391 , @xmath392 , @xmath393 and @xmath394 are well - defined .
@xmath395 for the particular sequence of measurement events in the fig .
[ fig : four - party - v - caus ] , with @xmath396 , the @xmath50-causal model predicts that the correlations @xmath381 of systems @xmath52 and @xmath101 conditioned upon the input - output of systems @xmath17 and @xmath132 are local .
this implies by the result of @xcite that the usual no - signaling conditions which impose that @xmath397 , @xmath398 , @xmath399 and @xmath387 are well - defined is violated . in @xcite ,
the authors argue that it is the @xmath397 or the @xmath387 that is not well - defined .
the argument is that since the @xmath400 measurement events happen _ outside _ the future @xmath50-cone of @xmath52 in fig .
[ fig : four - party - v - caus ] , and @xmath52 chooses his measurement _ freely _ at that location ( with the notion of free will according to definition [ lem : cr - freewill ] ) , @xmath52 s input @xmath401 can not cause ( be correlated to ) the output distribution of @xmath400 .
an analogous argument gives that the joint output distribution of @xmath402 can not depend on @xmath101 s input @xmath403 . while the above argument was in a single privileged reference frame , to make the argument for causality violation by the @xmath50-causal model ( modulo the assumptions of @xcite )
, one must take into account the following . 1 .
a consideration of measurement events in different inertial reference frames is necessary for superluminal influences to lead to causal loops .
this is for example seen in figure [ fig : two - party - ccl ] where the introduction of charlie and dave moving with uniform velocity relative to alice and bob was necessary to create a closed causal loop and the consequent violation of causality .
superluminal signals within a single reference frame do not lead to any causal paradoxes .
note also that , when investigating @xmath50-causal models in different reference frames , one can always find a reference frame in which two space - like separated events are simultaneous , in which case the non - locality seen from that reference frame requires an explanation in terms of influences propagating at infinite speed ( in that frame ) .
2 . an influence propagating faster - than - light in one frame of reference , can travel back in time in another frame .
explicitly , consider a superluminal @xmath50-speed influence propagating in an inertial reference frame @xmath14 from @xmath404 to the event at @xmath405 with @xmath406 where @xmath407 and @xmath408 so that @xmath132 happens after @xmath52 in frame @xmath14 .
now , consider another reference frame @xmath409 moving at speed @xmath410 relative to @xmath14 .
according to the lorentz transformations , we have @xmath411 now with @xmath226 , we may find a reference frame @xmath409 moving at speed @xmath410 and obeying @xmath412 so that @xmath413 . in this case , we get @xmath414 and the event @xmath132 precedes the event @xmath52 in reference frame @xmath409 . in particular , the above argument shows that the fact that the @xmath400 measurement events are outside the future @xmath50-cone of @xmath52 in a privileged reference frame does not imply that they will remain outside the future @xmath50-cone of @xmath52 in every inertial reference frame , since it is @xmath125-causality that is preserved by the lorentz transformation .
therefore , one may find inertial frames in which a superluminal influence propagates backwards in time and influences the @xmath400 distribution . that such an influence would not lead to causal loops in the measurement configuration of fig .
[ fig : four - party - v - caus ] is guaranteed by the argument of proposition [ prop : rel - cau - constraints ] since the intersection of the future @xmath125-cones of @xmath17 , @xmath101 and @xmath132 lies entirely within the future @xmath125-cone of @xmath52 .
analogously , the @xmath402 correlations may also depend on @xmath101 s input .
remark however , that in order for the argument of @xcite to work , it is necessary to assume that the @xmath381 correlations are local , despite this fact .
3 . in order to incorporate superluminal influences within a theory with the aim to preserve causality , it is necessary to consider the free - choice definition in def .
[ lem : mod - freewill ] .
in particular , since such superluminal influences allow correlations between measurement inputs and joint distributions of sets of measurement outcomes as long as they do not lead to causal paradoxes , the imposition of the free - choice condition in def . [
lem : cr - freewill ] is too strong a requirement in a theory with superluminal influences . a modification of the measurement configuration
is then required to ensure that the argument stands even with the notion of free - will in def .
[ lem : mod - freewill ] . as an example , let us impose a requirement that the parties measurement events be performed at the coordinates given by @xmath415 , @xmath416 , @xmath417 and @xmath418 . for @xmath419 , these coordinates ensure that the usual no - signaling restrictions are both necessary and sufficient to preserve causality , while still ensuring that the @xmath50-causal order @xmath420 is maintained . in other words
, one can check that for any subset of the parties , there exists a point in the intersection of their future light cones that is outside the future light cone of the remaining party .
an explanation of the quantum violation of the hidden influence inequality in this configuration would then lead to the conclusion ( modulo the assumptions above and in @xcite ) that the @xmath50-causal model results in a causal loop .
for the three - party measurement configuration of @xcite , an analogous argument as above shows that since the intersection of @xmath17 and @xmath101 s future light cones is contained within @xmath52 s future light cone , by the argument in proposition [ prop : rel - cau - constraints ] no causal loops would result if the correlations @xmath122 depend upon @xmath52 s input .
a suitable modification of the measurement configuration would ensure that the argument is maintained even under the notion of free - choice in ( [ lem : mod - freewill ] ) .
, @xmath53 , @xmath100 and @xmath421 respectively .
the yellow cones denote the future light cones of the four measurement events in the chosen reference frame while the grey cones denote the @xmath50-like future cones , for @xmath226 .
the argument in @xcite is that due to a randomness in the choice of measurement time and freedom in the choice of measurement setting , either @xmath52 or @xmath101 is unable to influence the correlations @xmath400 , @xmath402 respectively.,scaledwidth=75.0% ]
compared with the scenario of two - party nonlocality where @xmath113 is either local or non - local , in the multi - party scenario , different kinds of nonlocality can be distinguished . in the tripartite scenario ,
the local correlations are defined as usual as those @xmath138 that can be expressed as @xmath422 with @xmath423 . while correlations not of the form above are non - local , different kinds of tripartite nonlocality can be distinguished @xcite . in pioneering work @xcite ,
svetlichny introduced the notion of genuine @xmath241-way nonlocal or svetlichny nonlocal correlations which are those @xmath138 that _ can not _ be expressed in the form @xmath424 with @xmath425 , @xmath426 and @xmath427 . here ,
the bipartite distributions are _ not _ required to obey the bipartite no - signaling conditions , so these can be arbitrary signaling correlations .
correlations of the form in eq.([eq : bilocal ] ) are referred to as _ bilocal _ ( bl ) or
@xmath428-local " .
svetlichny introduced an inequality , the violation of which guarantees that the correlations are svetlichny nonlocal , i.e. , not of the form in ( [ eq : bilocal ] ) .
the inequality is expressed in terms of the correlation functions @xmath429 with @xmath430.\end{aligned}\ ] ] in terms of the correlators , svetlichny s inequality is explicitly written as @xmath431 suitable measurements on @xmath432 and @xmath433-class states lead to a violation of the inequality showing that quantum correlations are genuinely tripartite non - local according to this notion . alternatively , when the bipartite distributions in ( [ eq : bilocal ] ) obey the two - party no - signaling constraints , the resulting correlations @xmath138 are said to be _ no - signaling bilocal _
( nsbl ) . in the intermediate situation ,
when the bipartite distributions allow for signaling in one direction , the resulting correlations are termed _ time - ordered bilocal _ ( tobl ) i.e. , when @xmath434 and the other bipartite distributions are of the form @xmath435 or @xmath436 where these obey only the unidirectional no - signaling restrictions @xmath437 clearly , @xmath438 . in formulating an operational framework for nonlocality @xcite , i.e.
, terming genuine tripartite nonlocality as that which can not be created by local operations and allowed classical communication when two parties collaborate , the authors of @xcite showed that surprisingly being outside the set @xmath439 is sufficient but not necessary for genuine tripartite nonlocality , i.e. , there exist correlations @xmath440 which would also be genuinely tripartite nonlocal under an operational consideration .
in other words , there are correaltions in @xmath439 which under wirings and classical communication prior to the inputs between @xmath52 and @xmath101 can result in nonlocality in the cut @xmath441 , so that violation of svetlichny s inequality is sufficient but not necessary for genuine multiparty nonlocality .
a fundamental reason why the seemingly natural definition of svetlichny ( [ eq : bilocal ] ) does not capture the operational notion of genuine multi - way nonlocality is the lack of the no - signaling restrictions on the bipartite distributions in ( [ eq : bilocal ] ) .
violation of an inequality separating @xmath442 correlations from the rest by @xmath138 indicates that these correlations are genuinely @xmath241-way ns nonlocal @xcite .
here we review the notion of genuine @xmath241-way nonlocality in light of the considerations of previous sections .
in particular , we introduce a notion of relativistically causal bilocal ( rcbl ) correlations and propose an inequality the violation of which suggests that the resulting correlations @xmath138 are genuinely relativistically causal @xmath241-way nonlocal .
[ def : rel - caus - bilocal ] suppose that @xmath246 can be written in the form @xmath443 with @xmath425 , @xmath426 and @xmath427 , where the terms obey the relativistic causality constraints eq.([eq : rel - caus-3-party-1 ] ) , i.e. , each of the marginals @xmath151 , @xmath153 and @xmath152 is well - defined independently of the other parties inputs while the two - party term @xmath444 exhibits explicit dependence on @xmath112
. then the correlations @xmath246 are said to be relativistically causal bi - local ( rcbl ) .
otherwise , we say that they are genuinely tripartite relativistically causal non - local . in the final statements of this section
, we will show @xmath445 correlations that violate the svetlichny inequality and are hence outside @xmath439 ( @xmath446 ) and also we exhibit bilocal @xmath439 correlations that are outside the @xmath445 set ( @xmath447 ) showing that these two sets are incomparable and give rise to two distinct notions of genuine tripartite non - locality .
we have @xmath448 evidently , @xmath449 since the relativistic causality constraints eq.([eq : rel - caus-3-party-1 ] ) are a subset of the usual no - signaling constraints ( [ eq : multi - party - ns ] ) . to see that @xmath446 , we exhibit a box @xmath450 that violates the svetlichny inequality ( [ eq : svetlichny - ineq ] ) .
this box is simply ( and non - uniquely ) given by the correlators @xmath451 as an explicit example , we consider the box @xmath452 this achieves the algebraic value of @xmath453 for the svetlichny bell expression ( [ eq : svetlichny - ineq ] ) , maximally violating the inequality . evidently , any inequality based only on the three - body correlators @xmath429 can not be used to separate @xmath445 from genuine tripartite causal non - local boxes .
as such , an interesting question is to find a separating hyperplane to separate the set of @xmath445 from the set @xmath454 of quantum correlations .
while a detailed investigation of this is carried out in future work @xcite , here we introduce an inequality that accomplishes this in the tripartite scenario .
let us first write the probablilities @xmath138 in terms of the correlation functions .
the difference here is that the correlations @xmath455 exhibit a dependence on bob s measurement input @xmath123 .
we thus have @xmath456,\end{aligned}\ ] ] with the correlator @xmath457 exhibiting explicit dependence on @xmath121 in contrast to eq.([eq : prob - corrfn - old ] ) .
[ lem : rcbl - ineq ] consider the three - party bell scenario , with each party performing one of two dichotomic measurements . in measurement configurations such as fig.[fig : three - party - meas - config ] , the following inequality holds for all relativistically causal bi - local boxes @xmath450 .
@xmath458 suitably chosen measurements on the @xmath432 state @xmath459 attain a value @xmath460 violating the inequality , showing that quantum correlations are genuinely tripartite relativistically causal non - local .
furthermore , these quantum correlations belong to the set @xmath439 , hence @xmath447
. one can directly check that the inequality ( [ eq : rcbl - ineq ] ) is valid for all relativistically causal bilocal ( rcbl ) correlations by linear programming ( recall that the set @xmath445 is a convex polytope ) .
the measurements on @xmath461 are given by @xmath462 these measurements on the @xmath432 state give the value @xmath463 violating the inequality showing that quantum correlations are genuinely tripartite relativistically causal nonlocal .
interestingly , these measurements on the @xmath432 state do not lead to a violation of svetlichny s original inequality ( [ eq : svetlichny - ineq ] ) , and one can in fact find an explicit decomposition of the resulting box in the form ( [ eq : bilocal ] ) , showing that @xmath447 .
a central point of departure in deriving the relativistic causality constraints as opposed to the usual no - signaling constraints is that in the former , we also account for influences that propagate faster than light as long as they do not lead to causal loops .
we discuss in this section why this is not any cause for alarm .
firstly , remark that resonance - induced superluminal _ group _ velocities ( with values up to @xmath464 times @xmath125 ) have been known for a while @xcite , these group velocities do not create any problem of principle , since the _ signal _ velocity still has @xmath125 as an upper bound ( as in this paper ) .
the group velocity of a wave may exceed the speed of light and can even become negative , but in such cases , no energy or information actually travels faster than @xmath125 .
experiments showing group velocities greater than @xmath125 include that of wang et al .
@xcite , who produced a laser pulse in atomic cesium gas with a group velocity index of @xmath465 .
the observed superluminal propagation of the wave is not at odds with causality , and is instead a consequence of classical interference between its constituent frequency components in a region of anomalous dispersion @xcite .
secondly , as mentioned before , the justification that special relativity does not necessarily exclude faster - than - light propagation has also been known for a while @xcite .
a main objection to the existence of these particles was that they would lead to violations of causality and as such must be unphysical @xcite , which as we have seen can be overcome by the considerations of this paper .
thirdly , recall that the notion of point - to - region signaling considered in this paper is formulated as the nonlocal influence of a field rather than of a superluminal particle .
evidence in favor of a superluminal field emerges from the fact that spin - zero ftl particles ( satisfying the klein - gordon equation ) _ can not _ be localized as pointlike particles , contrary to all other known fundamental constituents of matter @xcite .
a brief overview of the argument is as follows , for details see @xcite .
consider the wavefunction @xmath466 of a free tachyon of spin zero obeying the klein - gordon equation in @xmath183 dimensions @xmath467 with solutions that are superpositions of @xmath468 with the energy - momentum relation @xmath469 .
for the particle moving with velocity @xmath226 , we have that @xmath470 is imaginary and @xmath471 is negative , so that we can consider the solutions with @xmath472 as well as those with @xmath473 .
if the initial conditions for the wave equation , @xmath466 and @xmath474 at time @xmath475 are zero outside some interval @xmath476 $ ] , then they remain outside the interval @xmath477 $ ] for any @xmath478 by the standard property of the wave equation . in other words
, the support of the evolved field configuration lies inside the light - cone of the initial data .
so that localized ( particle - like ) point - to - point influences do not move faster than the speed of light in vacuum , as considered in this paper . on the other hand ,
if we ignore the solution with @xmath473 ( which has energy @xmath133 imaginary ) , then it becomes impossible to solve the equation for initial conditions that are zero outside some interval @xmath479 $ ] @xcite so that we do not have a localized influence but only a delocalized one which is then allowed to be superluminal , consistent with the considerations of this paper .
finally , note that in algebraic quantum field theory , as mentioned earlier , to every bounded open subset @xmath480 of spacetime , one can associate observables @xmath481 and observables in spacelike separated regions @xmath482 are required to commute , i.e. , for all @xmath483 and @xmath484 we have @xmath485 = 0 $ ] thus ensuring that the causal constraints are satisfied . in linear field theories , there is a complete duality between particles and fields , via the well - known fock space construction .
however as haag pointed out @xcite , the one - to - one correspondence between particles and fields holds only in the asymptotic sense , i.e , as time goes to @xmath486 ( in the heisenberg picture ) , the limit fields obey the free field equation and canonical commutation relations so as to describe free particles ; so that it may well be that a quantum field theory does not describe any localized particles at all @xcite .
in this paper , we have examined the no - signaling constraints from the point of view of causality and proposed a modification to these constraints to a subset that ensures preservation of causality depending on the spacetime configuration of measurement events . as a consequence
, we find that boxes in the post - quantum scenario , where one only restricts to the constraints imposed by relativistic causality , should be labeled by the space - time locations of the parties performing the measurement .
the correlations that the parties observe can then exhibit various dependencies while still being consistent with relativistic causality .
several questions are open .
one important question is to formulate protocols ( under suitable assumptions ) for cryptographic tasks such as randomness amplification , expansion and key distribution against an adversary only limited by the laws of relativity ( from creating closed causal loops ) . to do this ,
an intermediate task is to investigate the existence of multi - party bell inequalities which allow for certification of randomness against such a relativistically causal adversary .
a geometric question is to characterize the subset of multi - party relativistically causal constraints and the space - time regions where they are necessary and sufficient .
the notion of genuine multi - party nonlocality as evidenced by the violation of a relativistically causal bilocal ( @xmath445 ) inequality calls for further investigation @xcite , in particular in light of studies such as @xcite .
a pressing open question is to investigate whether years after special relativity was formulated , such superluminal influences can be tested experimentally , and what new effects they could lead to .
both non - relativistic quantum theory and relativistic quantum field theory @xcite are well - known to obey a no superluminal signaling condition , proposals to modify quantum theory by introducing non - linearities have been shown to lead to signaling @xcite .
an important open question is to investigate feasible mechanisms for the point - to - region superluminal influences within the paradigm of non - linear modifications to quantum theory . _
acknowledgments.- _ we acknowledge stimulating discussions with andrzej grudka and jakub rembieliski . r.r . acknowledges useful discussions with piotr mironowicz , karol horodecki and roberto salazar and thanks stefano pironio for valuable comments on an earlier version of the manuscript .
this work was made possible through the support of a grant from the john templeton foundation .
the opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the john templeton foundation .
it was also supported by the erc adg grant qolaps .
r.r . acknowledges support from the grant `` causality in quantum theory : foundations and applications '' funded by la fondation philippe wiener - maurice anspach .
s. pironio , a. acin , s. massar , a. boyer de la giroday , d.n .
matsukevich , p. maunz , s. olmschenk , d. hayes , l. luo , t.a .
manning and c. monroe , _ random numbers certified by bell s theorem _ , nature * 464 * , 1021 ( 2010 ) .
s. popescu and d. rohrlich , _ causality and nonlocality as axioms for quantum mechanics _ , in the _ proceedings of the symposium on causality and locality in modern physics and astronomy : open questions and possible solutions _ ( york university , toronto , 1997 ) .
a. bers , r. fox , c. g. kuper and s. g. lipson , _ the impossibility of free tachyons _ , in _ relativity and gravitation _ , eds . c. g. kuper and asher peres , new york , gordon and breach science publishers , 41 ( 1971 ) .
r. colbeck and r. renner , _ free randomness can be amplified _ , nature physics * 8 * , 450 - 454 ( 2012 ) .
r. gallego , ll .
masanes , g. de la torre , c. dhara , l. aolita and a. acn , _ full randomness from arbitrarily deterministic events _ , nature communications * 4 * , 2654 ( 2013 ) .
p. h. eberhard and r. r. ross ,
_ quantum field theory can not provide faster - than - light communication _
, foundations of physics letters , vol .
2 , no . 2 ( 1989 ) . k. a. peacock , b. s. hepburn , _ begging the signalling question : quantum signalling and the dynamics of multiparticle systems _ , arxiv : quant - ph/9906036 ( 1999 ) .
* proposition 5*. _ consider measurement events @xmath169 with corresponding space coordinates @xmath487 in a chosen inertial reference frame @xmath14 .
then a measurement event @xmath133 can superluminally influence the correlations between @xmath17 and @xmath52 at speed @xmath129 without violating causality in @xmath171 if and only if its space coordinate @xmath172 satisfies @xmath173 for any circle @xmath174 with @xmath175 as a chord and having angle @xmath176 as the angle in the corresponding minor segment , where @xmath177 and @xmath178 , and if its time coordinate @xmath179 satisfies @xmath180.\ ] ] _
we observe that in order for eve to signal to a space - time point @xmath124 via her influence of the correlations at @xmath17 and @xmath52 the following constraints must hold .
the sum of the time taken for an influence ( at speed @xmath126 ) to move from @xmath133 to @xmath17 and the time taken for a signal ( at speed @xmath125 ) to move from @xmath17 to @xmath124 must be less than the time taken for a signal at speed @xmath125 to travel from @xmath133 to @xmath124 directly . to elaborate , @xmath133 may superluminally influence the measurement at @xmath17 which subsequently signals to the event @xmath124 or alternatively , @xmath133 may signal to @xmath124 ( signaling being by definition at speed @xmath125 ) .
a similar condition must also hold for the influence traveling via point @xmath52 . .
@xmath133 superluminally influences the correlations between @xmath17 and @xmath52 while ensuring that the marginal distributions are unaffected . in order to ensure that @xmath133 does not send a superluminal signal to a spacetime location @xmath124 via the influence of the correlations between @xmath175
, we impose the constraints @xmath489 and @xmath490 .
lemma [ lem : dual - signal ] shows that for fixed locations @xmath491 and @xmath492 , no superluminal signaling takes place iff the cones shown in the figure @xmath493 and @xmath494 either share a common tangent plane or intersect only at @xmath133.,scaledwidth=60.0% ] [ lem : single - signal ] given measurement events @xmath495 and @xmath17 with corresponding space - time co - ordinates @xmath496 , @xmath497 and @xmath498 in a chosen inertial reference frame @xmath171 and an influence propagating from @xmath133 to @xmath17 at speed @xmath129 , @xmath133 superluminally signals to @xmath124 via @xmath17 if and only if the space coordinates satisfy @xmath499 where @xmath500 denotes a sphere with center at @xmath501 and radius @xmath502 , @xmath503 , @xmath504 and @xmath505 .
our task is to show that for any @xmath124 with space coordinate @xmath508 , the first signal does not reach @xmath124 before the second .
consider a ball @xmath509 of radius @xmath510 with centre @xmath133 , where @xmath511 corresponds to the distance between @xmath133 and @xmath17 in the reference frame @xmath171 .
we also consider the ball @xmath512 with centre @xmath124 and radius @xmath513 .
the constraint we impose is @xmath488 this is equivalent to @xmath514 now recall that @xmath515 so that the above condition may be rewritten as @xmath516 this in turn is equivalent to the criterion that the spheres @xmath509 and @xmath517 do not intersect , @xmath518 in other words , @xmath133 is able to superluminally signal to a point @xmath124 via the superluminal influence on @xmath17 if and only if the two spheres so constructed do not intersect .
we now proceed to analyze the question whether @xmath133 is able to superluminally signal to any point @xmath124 via her superluminal influence of the correlations @xmath166 and a subsequent signal ( at speed @xmath125 ) propagating from both @xmath17 and @xmath52 to @xmath124 .
let @xmath491 and @xmath492 be the fixed laboratory locations .
we propose a geometric condition identifying for given @xmath487 the positions @xmath172 for which _ no _ @xmath124 exists such that @xmath133 superluminally signals to @xmath124 via the above mentioned route .
let @xmath519 denote the tangent cone to the sphere @xmath500 emanating from point @xmath318 .
for a given tangent cone to a sphere @xmath500 , let @xmath520 denote the cone with apex point @xmath501 and passing through the points of intersection of the sphere and its tangent cone , i.e. , through @xmath521 .
[ lem : dual - signal ] for fixed @xmath491 and @xmath492 , there does not exist any @xmath124 such that @xmath133 superluminally signals to @xmath124 via @xmath17 and @xmath52 if and only if the cones @xmath493 and @xmath494 either share a common tangent plane or intersect only at @xmath133 . from lemma [ lem : single - signal ] , for a fixed inertial reference frame ,
we know that for any @xmath522 , @xmath133 is able to superluminally signal to any point @xmath124 with corresponding space coordinate @xmath508 via @xmath133 s superluminal influence of the event @xmath17 if and only if @xmath523 . in order for @xmath133 to superluminally signal to @xmath124 via her influence of the correlations between @xmath17 and @xmath52 , a signal
( at speed @xmath125 ) must reach @xmath124 from _ both _ @xmath17 and @xmath52 before the signal from @xmath133 at speed @xmath125 reaches @xmath124 .
now , in order for the condition in eq .
( [ eq : single - signal ] ) to hold for the signal from @xmath17 , we show that @xmath124 must obey @xmath524.\ ] ] points in space that are outside this cone do not perceive any superluminal signal from @xmath133 in that for these points @xmath525 .
on the other hand , we have the following .
consider a far enough point @xmath526 on the ray emanating from @xmath133 and passing through the interior of the cone such that the sphere @xmath528 does not intersect @xmath529 .
let @xmath530 denote the point at which the ray intersects the sphere @xmath528 .
by construction , we have @xmath531 so that such a sphere can always be drawn and we see that @xmath530 does not lie on @xmath529 . similarly , for the signal from @xmath52 to reach @xmath124 before the one from @xmath133
, we must have @xmath532.\ ] ] therefore , in order to exclude every point in space from a superluminal influence from @xmath133 , one must ensure that either @xmath533 or that the two cones only intersect at a common tangent plane . in this case , no point @xmath124 different from @xmath133 exists such that @xmath133 superluminally signals to @xmath124 via her influence of @xmath17 and @xmath52 .
alternatively , if the intersection @xmath534 is a three - dimensional region , then by claim [ cl : sign - point ] one can explicitly find an @xmath124 in this region obeying both eqs .
( [ eq : signal - a ] ) and ( [ eq : signal - b ] ) , so that superluminal signaling to such an @xmath124 occurs . the constraint in lemma.([lem : dual - signal ] )
is illustrated in fig [ fig : signaling - region - cond ] .
having lemma [ lem : dual - signal ] we can proceed to identify for given @xmath487 the region of allowable @xmath172 such that no causality condition is violated ( in the chosen reference frame ) . given
the points @xmath169 let @xmath535 denote a circle containing @xmath175 as a chord and @xmath536 as the angle in the minor segment of the circle .
note that there are an infinite number of such circles , one for each plane containing the line segment @xmath175 .
let us also denote by @xmath537 the corresponding minor segment of the circle .
we now show that the possible space coordinates @xmath172 lie within the shaded region ( in green ) in fig .
[ fig : aposteriori - ns - region ] . we have seen in lemma [
lem : dual - signal ] that in order for @xmath133 to influence the correlations between @xmath17 and @xmath52 at speed @xmath188 without violating causality , we must have that the cones @xmath493 and @xmath494 at most share a common tangent plane , in particular that the intersection of their interiors is empty .
let us first consider the case that @xmath493 and @xmath494 share a common tangent plane passing through the points @xmath538 where @xmath30 is a point of intersection of the sphere @xmath539 and its tangent plane from @xmath17 , @xmath540 and similarly @xmath541 is a point of intersection of the sphere @xmath542 and @xmath543 , such that @xmath544 lie on one line . by the property of the tangent we have that @xmath545 .
also by definition we have that @xmath546 so that @xmath547 .
consequently , we obtain that any point @xmath133 for which the cones @xmath493 and @xmath494 share a common tangent plane has space coordinate in the chosen inertial reference frame satisfying @xmath548 by the fact that angles on the same segment of a circle are equal , we have that all such points lie on a minor segment of a circle ( the segment is minor since @xmath549 for @xmath550 ) having @xmath175 as a chord and with @xmath176 as the corresponding angle in the segment .
this shows the statement for points on the boundary of the region @xmath551 .
for all such points the cones @xmath493 and @xmath494 share a common tangent plane .
now consider the points that lie within this region .
for any such point , we have that @xmath552 . by construction we know that @xmath547 so that by the perpendicularity of the tangent with the radius of the sphere , we have @xmath553 .
therefore , for points that lie within the region , the cones @xmath493 and @xmath494 intersect only at @xmath133 , this is the case illustrated in fig . [
fig : signaling - region - cond ] . for points outside this region
we readily see that @xmath493 and @xmath494 neither intersect only at @xmath133 nor share only a common tangent plane .
we see by lemma [ lem : dual - signal ] that such points violate causality and therefore can not be the space coordinates of @xmath133 in the chosen reference frame . , @xmath492 from which @xmath133 is able to superluminally influence the correlations @xmath175 without changing the marginals while preserving causality .
the figure illustrates the algebraic condition found in prop .
[ prop : spacetime - region].,scaledwidth=45.0% ] * proposition 12 .
* _ consider the @xmath0-party ghz - mermin bell inequality , for odd @xmath291 .
for any input @xmath313 appearing in the inequality , i.e. , @xmath314 , there exists a box @xmath315 violating the mermin inequality maximally and obeying the relativistic causality constraints in eq.([eq : rel - caus ] ) , such that no randomness can be extracted from the outputs @xmath206 of the box under input @xmath313 .
in other words , we have @xmath316 for some fixed output bit string @xmath317 .
_ the proof is by explicit construction of the @xmath0-party box @xmath290 satisfying the relativistic causality constraints eq.([eq : rel - caus ] ) and violating the mermin inequality maximally that returns a deterministic output for the fixed input @xmath314 .
let us first list the set of constraints that @xmath290 must obey .
split the set of inputs @xmath554 appearing in the mermin inequality into two subsets @xmath555 then for the inputs @xmath286 in the set @xmath556 , the box @xmath290 is required to give outputs such that the value of the @xmath0-party correlation function is @xmath557 , i.e. , @xmath558 similarly for the inputs @xmath559 , the outputs from the box are required to yield a value of @xmath214 for the @xmath0-party correlation function , i.e. , @xmath560 for the given fixed input @xmath314 , the box @xmath290 is required to return a deterministic input @xmath317 , this is the constraint from eq.([eq : det - hash ] ) .
the box @xmath290 is also required to obey all of the relativistic causality constraints from eq.([eq : rel - caus ] ) . finally , as usual
we have the non - negativity and normalization constraints for the probability distributions , i.e. , @xmath561 we will show a binary - tree algorithmic construction of @xmath315 obeying all the constraints eqs.([eq : rel - caus ] , [ eq : det - hash ] , [ eq : constraint-1 ] , [ eq : constraint-2 ] , [ eq : norm - nonneg ] ) for given @xmath562 .
the construction for @xmath563 will follow along analogous lines . let @xmath562 be given .
initiate as step @xmath334 , @xmath564 ( the all-@xmath334 bit string ) . at the @xmath565-th step , @xmath566 , @xmath567 @xmath568 ,
if @xmath569 define @xmath570 if on the other hand , @xmath571 define @xmath572 at the @xmath573-th step @xmath574 , @xmath567 @xmath575 , if @xmath576 define @xmath577 if on the other hand , @xmath578 define @xmath579 @xmath580 , set @xmath581 the proof that the binary tree algorithm yields a box with the desired properties is done by induction .
first consider the base steps .
we are given @xmath562 , and a fixed output bit string of even parity @xmath317 ( the all-@xmath334 string ) .
the algorithm sets @xmath582 and @xmath583 for @xmath584 .
this assignment clearly satisfies both eqs .
( [ eq : det - hash ] ) and the mermin constraint @xmath585 . in particular , since the output distribution is deterministic for the chosen input @xmath313 , no hash function chosen by the parties can extract any randomness from the outputs for this input .
the next base step in the construction of box @xmath315 is to define the conditional probability distribution of the outputs for the set of inputs @xmath286 that are at hamming distance @xmath214 from @xmath313 , i.e. , the inputs obtained from @xmath313 by a bit flip at position @xmath4 .
explicitly , we consider the set of inputs @xmath586 where @xmath587 , @xmath588 denotes the all-0 bit string with a single bit value @xmath214 at position @xmath4 , and @xmath589 denotes the bit - wise @xmath590 of the two bit strings . for @xmath591 ,
the algorithm sets @xmath592 , and @xmath593 otherwise .
observe that inputs from @xmath594 have even parity and do not enter the mermin inequality , i.e. , @xmath595 . from the above assignment
, we have @xmath596 for @xmath591 .
furthermore , the assignment satisfies the relativistically causal constraints for the contiguous subsets to the left ( @xmath597 ) and right ( @xmath598 ) of bit @xmath4 , i.e. , having shown the base steps , we now prove by induction that the above algorithmic construction works as advertised , i.e. , that the box @xmath290 it produces satisfies all the constraints in eqs.([eq : rel - caus ] , [ eq : det - hash ] , [ eq : constraint-1 ] , [ eq : constraint-2 ] , [ eq : norm - nonneg ] ) .
clearly , eq.([eq : det - hash ] ) and the normalization constraints ( [ eq : norm - nonneg ] ) are satisfied by construction .
so it remains to show that the box satisfies the relativistic causality constraints ( [ eq : rel - caus ] ) and the mermin inequality constraints ( [ eq : constraint-1 ] ) and ( [ eq : constraint-2 ] ) .
let us firstly consider the mermin inequality constraints ( [ eq : constraint-1 ] ) and ( [ eq : constraint-2 ] ) .
these constraints appear at the @xmath573-th ( even ) steps with @xmath600 of the algorithm .
assume as the inductive hypothesis that the mermin constraints are satisfied at the @xmath601-th step , i.e. , that for all @xmath602 , we have @xmath603 now , consider the scenario when @xmath604 .
we have two cases : either in the case ( 1 ) , we have by the inductive hypothesis eq.([eq : induct - mermin - corr ] ) that @xmath609 , i.e , both @xmath610 and @xmath611 have even parity . then when @xmath612 we get by first applying eq.([eq : odd - step-0 ] ) and then eq.([eq : even - step-1 ] ) that @xmath613 so that both outputs @xmath614 have even parity and we obtain @xmath615 as required . similarly when @xmath616 we get by first applying eq.([eq : odd - step-1 ] ) and then eq.([eq : even - step-0 ] ) that @xmath617 so that again both outputs have even parity . in the case
( 2 ) , we have by the inductive hypothesis eq.([eq : induct - mermin - corr ] ) that @xmath618 , i.e , both @xmath610 and @xmath611 have odd parity .
then when @xmath619 we get by first applying eq.([eq : odd - step-0 ] ) and then eq.([eq : even - step-0 ] ) that @xmath620 so that a bit flip occurs on both outputs @xmath614 which now have even parity and we obtain @xmath615 as required .
finally , when @xmath621 we get by first applying eq.([eq : odd - step-1 ] ) and then eq.([eq : even - step-1 ] ) that @xmath622 so that again the outputs have even parity as required .
now , consider the scenario when @xmath623 .
we again have two cases : either as in the previous scenario , an exhaustive check of the output parities in the case ( 1 ) and ( 2 ) shows that the mermin constraints are satisfied . for clarity , we explicitly provide these here . in the case
( 1 ) , we have by the inductive hypothesis eq.([eq : induct - mermin - corr ] ) that both @xmath610 and @xmath611 have even parity .
then when @xmath619 , applying first eq.([eq : odd - step-0 ] ) and then ( [ eq : even - step-0 ] ) we get that @xmath620 so that now both the outputs @xmath614 have odd parity .
analogous results are obtained when @xmath621 , i.e. , applying first eq.([eq : odd - step-1 ] ) and then ( [ eq : even - step-1 ] ) we get that @xmath622 so that the resulting outputs have again odd parity . in the case
( 2 ) , we have by the inductive hypothesis eq.([eq : induct - mermin - corr ] ) that both @xmath610 and @xmath611 have odd parity . then when @xmath612 , applying first eq.([eq : odd - step-0 ] ) and then ( [ eq : even - step-1 ] ) we get that @xmath624 so that the resulting outputs have again odd parity . similarly , when @xmath616 , applying first eq.([eq : odd - step-1 ] ) and then ( [ eq : even - step-0 ] ) we get that @xmath617 so that the resulting outputs have again odd parity
. having exhaustively checked all the cases , we conclude by the inductive hypothesis that the mermin constraints ( [ eq : constraint-1 ] ) and ( [ eq : constraint-2 ] ) are satisfied at all the even steps of the algorithm . it remains to show that the box @xmath290 satisfies the relativistic causality constraints ( [ eq : rel - caus ] ) .
the relativistic causality constraints ( [ eq : rel - caus ] ) effectively say that the output probability distribution for any contiguous set of parties is independent of the complementary set of parties inputs .
@xmath625 the structure of the algorithm is such that from step @xmath2 to step @xmath626 ( @xmath2 here is arbitrary ) , in going from input @xmath627 to input @xmath628 , a bit flip is applied to either @xmath629 or @xmath630 _ at position @xmath631 _ and depends only on @xmath632 .
this implies that the probability distribution of the outputs to the left and right of position @xmath626 are unaffected , i.e. , the probability distribution for the parties in the set @xmath633 and those in the set @xmath634 remain fixed under this operation ( recall that @xmath202 ) .
this ensures that the relativistic causality constraints are satisfied , since for any two input strings @xmath635 and @xmath636 that share a common contiguous bit sequence @xmath637 , both @xmath638 and @xmath639 only depend explicitly on the sequence of bits @xmath640 for @xmath641 .
we have thus shown the box @xmath290 that satisfies the mermin constraints and the relativistic causality constraints in addition to returning a deterministic output @xmath317 for input @xmath313 .
* proposition 18 . *
_ consider a three - party bell scenario , with alice , bob and charlie each performing two measurements @xmath368 of two outcomes @xmath369 respectively .
suppose that in some inertial reference frame , the three space - like separated parties are arranged in @xmath214-d , with @xmath378 and perform their measurements simultaneously , i.e. , @xmath356 .
then , there exists a three - party relativistically causal box @xmath642 such that @xmath643 _ it can be readily verified that the box @xmath290 as defined above satisfies all the relativistic causality constraints in eq.([eq : rel - caus-3-party ] ) .
furthermore , we also see that when alice - bob or bob - charlie measure input @xmath646 , they get anti - correlated outputs , while for other inputs , they get correlated outcomes .
this gives that the values of the chsh bell expressions @xmath372 and @xmath647 are both maximal ( equal to the algebraic value of @xmath376 ) .
in other words , the phenomenon of monogamy for the chsh violating correlations does not hold under the relativistic causality conditions in certain measurement configurations . | the ubiquitous no - signaling constraints state that the probability distribution of the outputs of any subset of parties is independent of the inputs of the complementary set of parties ; here we re - examine these constraints to see how they arise from relativistic causality .
we show that while the usual no - signaling constraints are sufficient , in general they are not necessary to ensure that a theory does not violate causality . depending on the exact space - time coordinates of the measurement events of the parties participating in a bell experiment
, relativistic causality only imposes a subset of the usual no - signaling conditions .
we first revisit the derivation of the two - party no - signaling constraints from the viewpoint of relativistic causality and show that the they are both necessary and sufficient to ensure that no causal loops appear .
we then consider the three - party bell scenario and identify a measurement configuration in which a subset of the no - signaling constraints is sufficient to preserve relativistic causality .
we proceed to characterize the exact space - time region in the tripartite bell scenario where this phenomenon occurs .
secondly , we examine the implications of the new relativistic causality conditions for security of device - independent cryptography against an eavesdropper constrained only by the laws of relativity .
we show an explicit attack in certain measurement configurations on a family of randomness amplification protocols based on the @xmath0-party mermin inequalities that were previously proven secure under the old no - signaling conditions .
we also show that security of two - party key distribution protocols can be compromised when the spacetime configuration of the eavesdropper is not constrained .
thirdly , we show how the monogamy of non - local correlations that underpin their use in secrecy can also be broken under the relativistically causal constraints in certain measurement configurations .
we then inspect the notion of free - choice in the bell experiment and propose a definition of free - choice in the multi - party bell scenario .
we re - examine the question whether quantum correlations may admit explanations by finite speed superluminal influences propagating between the spacelike separated parties .
finally , we study the notion of genuine multiparty non - locality in light of the new considerations and propose a new class of causal bilocal models in the three - party scenario .
we propose a new svetlichny - type inequality that is satisfied by the causal bilocal model and show its violation within quantum theory . |
the space density of distant clusters of galaxies is a measurable quantity whose theoretical value is highly sensitive to the physical and cosmological parameters of models of structure formation and evolution ( e.g. , oukbir & blanchard , 1992 ; bahcall & cen , 1992 ; viana & liddle , 1996 ; carlberg et al . , 1997 ; oukbir & blanchard , 1997 ; eke et al . , 1998 ) .
a large number of independent measurements of the cluster x - ray luminosity function ( xlf ) have been performed in the past decade .
given the diversity of the original observations used in these studies and of the data analysis techniques applied , the good agreement of the results is impressive .
virtually all studies agree that the abundance of clusters of low to intermediate x - ray luminosity ( @xmath4 , @xmath2 kev ) does not change significantly out to @xmath5 ( gioia et al .
, 1990b ; henry et al . , 1992 ; burke et al . , 1997 ; ebeling et al . , 1997 ;
vikhlinin et al . , 1998a ; jones et al .
, 1998 ; rosati et al . , 1998 ; degrandi et al . 1999 ; nichol et al . , 1999 ) . at higher x - ray luminosities
, however , a consistent picture has yet to emerge .
reports of strong negative evolution at @xmath6 ( @xmath7 kev ) already at moderate redshifts just beyond @xmath8 ( gioia et al . , 1990b ; henry et al . , 1992
; see also nichol et al .
, 1997 , for a contrary result ) are supported by the findings of vikhlinin et al . ( 1998a ) , although the latter rest on a statistically less secure basis .
if these results are correct , a much more pronounced dearth of x - ray luminous clusters is expected at yet higher redshift , unless cluster evolution is a strongly non - linear , almost discontinuous function of x - ray luminosity and redshift .
however , luppino & gioia ( 1995 ) show that the cluster xlf at @xmath9 is consistent with the one found in the emss for the redshift range @xmath10 ( median @xmath11 ) , i.e. , there appears to be no further significant evolution of luminous clusters beyond @xmath12 ( although such evolution is not ruled out ) . with only a handful of x - ray luminous clusters currently known at @xmath13 , the key
to understanding these apparently conflicting results lies clearly in new discoveries , and more detailed observations , of x - ray luminous ( and , by inference , massive ) clusters at high redshift .
any additional detection of a massive cluster at high redshift ( and certainly at @xmath14 ) is thus of paramount importance as it brings us one step closer to an accurate measurement of the cluster abundance at very high redshift , where its sensitivity to evolutionary effects is greatest .
with the exception of the bright sharc survey ( nichol et al . , 1999 ) , all of the present deep pspc cluster surveys provide sufficient depth to detect a cluster of @xmath15 ( @xmath7 kev ) throughout ] out to @xmath16 and beyond .
when the cumulative high - redshift emss xlf ( median @xmath11 ) is scaled to the comoving volume corresponding to the redshift range @xmath17 ( i.e. , assuming no evolution between @xmath18 and @xmath19 ) , it predicts about 15 clusters with @xmath15 ( @xmath7 kev ) per steradian , i.e. @xmath20 per square degree .
since the mentioned cluster surveys cover only from 18 square degrees ( sharc - s ) to 160 square degrees ( cfa cluster survey ) , the detection of only very few x - ray luminous clusters in any of these surveys places significant constraints on the evolution of clusters and large scale structure in general . note that this is true only for the most luminous systems : if the x - ray luminosity criterion is relaxed and clusters down to @xmath21 ( @xmath7 kev ) are considered , the expected cluster density in the same redshift range rises by almost an order of magnitude and any individual cluster detection becomes much less significant .
we emphasize that , although any detection ( x - ray , optical , infra - red ) of massive clusters at very high redshifts is an important discovery in its own right , it is clusters detected in the course of statistically complete surveys that bear the most weight . only the latter allow the space density of such systems to be quantified and compared to predictions from theoretical models .
any systematic effects in the data analysis and interpretation that could cause such clusters to be missed or misidentified need to be thoroughly understood and corrected for before conclusions about the physical or cosmological parameters governing cluster evolution are drawn from derived statistics such as the cluster xlf . in the rest of this paper we summarize the current observational status ( section [ sofar ] ) and give a short overview of the warps serendipitous cluster survey ( section [ warps ] ) .
we then describe the warps discovery of clj0152.71357 , a very x - ray luminous cluster at @xmath22 , and discuss and summarize the results from all available x - ray observations of this system ( section [ disc_obs ] ) . prompted by our finding that clj0152.71357 was missed in the emss , we take a closer look at how deviations from spherical symmetry in a cluster s x - ray emission may affect the emss cluster sample ( and possible other cluster samples ) as a whole ( section [ emss ] ) . trying to assess the importance of biases caused by complex cluster morphology we investigate the prevalence of substructure in distant clusters ( section [ morph ] ) before , finally , discussing the implications of our findings for attempts to constrain cosmological parameters using x - ray flux limited cluster samples ( section [ summary ] ) .
very few clusters of galaxies have been detected at redshifts greater than 0.8 , and even fewer can be called x - ray luminous . prior to the discovery of clj0152.71357
, only two x - ray selected clusters were known at @xmath23 : ms1054.40321 ( @xmath24 , @xmath25 in the @xmath7 kev band , donahue et al . , 1998 ) and , much less x - ray luminous , rxj1716.6 + 6708 ( @xmath26 , @xmath27 in the @xmath2 kev band , henry et al .
1997 , gioia et al.1999 ) .
slightly closer than @xmath28 ( @xmath29 , gioia & luppino 1994 ) , but distant and x - ray luminous enough to be noteworthy in this context , is ms1137.5 + 6625 ( @xmath30 in the @xmath7 kev band ) .
the discovery of an even more distant x - ray emitting cluster at @xmath31 was recently reported by rosati et al .
( 1999 ) ; however , at @xmath32 ( @xmath2 kev ) this system is even less x - ray luminous than rxj1716.6 + 6708 .
all other presently known clusters at very high redshift have been optically selected as projected galaxy overdensities in deep ccd images ( gunn , hoessel & oke , 1986 , gho ; postman et al . , 1996 , pdcs , scodeggio et al .
1999 ) or were originally detected at radio or infrared wavelengths ( e.g. crawford & fabian 1996 , deltorn et al . , 1997 , stanford et al . , 1997 ) .
although there are now an impressive number of optically selected clusters at @xmath23 ( postman and coworkers alone list a dozen clusters at @xmath33 in their pdcs sample ) , it ought to be emphasized that , for the majority of these possibly very distant optical clusters , the published ` redshifts ' are estimated statistically and are not the result of actual spectroscopic measurements .
the physical reality of many of these systems thus remains to be confirmed through either x - ray or extensive spectroscopic observations .
the difficulties inherent to the optical approach are evidenced by , e.g. , the work of oke , postman & lubin ( 1998 ) who obtained 892 redshifts in the fields of nine distant cluster candidates selected from the gho and pdcs catalogues .
three of their nine candidate clusters showed no significant peak in the observed redshift histogram , and three others showed between two and four equally significant peaks at very different redshift , illustrating the severity of projection effects in optically selected cluster samples . by way of contrast , castander et al .
( 1994 ) demonstrate how x - ray observations can be used efficiently to test whether optically selected distant clusters are indeed gravitationally bound , massive systems .
castander and coworkers analyzed pspc observations of five gho clusters which had spectroscopic redshifts ranging from 0.7 to 0.92 .
they detected only two of the five , and found all five to have measured x - ray luminosities , or upper limits , of less than @xmath34 erg s@xmath35 ( @xmath2 kev ) ; i.e. they are , at best , poor clusters which , unless detected in very large numbers , do not provide stringent constraints on either the rate of cluster evolution or the cosmological parameters of structure formation models .
the goal of the wide angle rosat pointed survey ( warps ) is to compile a complete and unbiased , x - ray selected sample of clusters of galaxies from serendipitous detections of x - ray sources in deep rosat pspc pointings . a comprehensive overview of the scientific goals of the project , the x - ray source detection algorithm employed ( voronoi tesselation and percolation : vtp ) , the sample selection and flux corrections techniques , as well as first results , are presented in paper i. vtp is particularly well suited for the detection and characterization of low - surface brightness emission ( ebeling & wiedenmann 1993 ; ebeling et al . , 1996 ; paper i ) and is likely to recognize even very distant clusters as extended x - ray sources ( paper i ) .
however , in order to reduce possible incompleteness in our cluster sample due to erroneous classification of distant clusters as point sources , our optical follow - up observations are not limited to extended x - ray sources but also include likely point sources without obvious optical counterparts ( paper ii ) .
paper ii also discusses the warps @xmath36 distribution of poor clusters of galaxies and its implications for cluster evolution .
the first two warps papers focus on results for a complete sample of clusters compiled over a geometric solid angle of 14.7 square degrees during the first phase of the project . in may 1997 ,
the warps project went into its second phase which increases the total solid angle to 73 deg@xmath37 and will yield a statistically complete sample of more than 70 x - ray selected galaxy clusters at @xmath38 . in this second phase
, cluster candidates without obvious optical counterpart on the poss plates ( as provided by the digitized sky survey ) were imaged at the michigan - dartmouth - mit 1.3 m and university of hawaii 2.2 m telescopes in preparation for spectroscopic follow - up observations at larger telescopes .
although observations of a few very distant cluster candidates have yet to be performed ( scheduled for spring 2000 ) , the warps cluster sample is already complete at @xmath39 over the full solid angle .
in 1996 and 1997 , the cluster clj0152.71357 was discovered independently in the rdcs and warps surveys .
later , clj0152.71357 was also detected in the bright sharc survey ( nichol et al . , 1999 ) . in this section
we describe the discovery of clj0152.71357 in the warps survey and discuss and summarize previous and subsequent x - ray observations of this system . the standard warps x - ray analysis detected clj0152.71357 as a very extended source 14.2 arcmin off - axis in a 20ks pspc pointed observation of ngc720 ; the poss-2 digitized sky survey image is blank at the position of the source .
figure [ iband ] shows an i band image of clj0152.71357 , taken with the uh 2.2 m telescope on aug 4 , 1997 , with adaptively smoothed pspc x - ray flux contours overlaid .
a blow - up of the central cluster region is shown in figure [ iband_core ] . based on the x - ray source extent and the observed overdensity of faint galaxies at and around the position of the x - ray source ,
it was classified as a likely distant cluster of galaxies .
the x - ray emission from clj0152.71357 shows a high degree of substructure and a pronounced elongation along a position angle of about 40@xmath40 which follows roughly the distribution of galaxies in the cluster core ( see section [ morph ] for a discussion of the dynamical state of clj0152.71357 ) . on aug 11 , 1997
we observed a total of 14 distant cluster candidates , among them clj0152.71357 , with the low - resolution spectrograph lris ( oke et al .
1995 ) on the keck - ii 10 m telescope on mauna kea .
using a longslit of @xmath41 width and the 300/5000 grating which provides 2.4 / pixel resolution and spectral coverage from 5000 to 10000 , we obtained spectra of six galaxies ( see fig .
[ iband_core ] ) close to the peak of the x - ray emission from clj0152.71357 and found redshifts as listed in table [ rstab ] .
the spectra are shown in fig .
[ spectra ] .
all redshifts are accordant and consistent with a cluster redshift of @xmath22 .
all spectra show absorption features typical of old stellar populations in elliptical galaxies , and none shows emission lines that would suggest agn contamination .
ngc720 was observed not only with the rosat pspc in 1992 , but also with the einstein ipc in 1980 , and with the rosat hri in 1994 .
we examine and compare the images from all three observations in fig .
[ xcomp ] .
the images are shown in chronological order and have been registered using the astrometry solutions in the respective fits headers . to allow an assessment of the quality of the raw data as well as of the presence and morphology of any extended emission ,
we show contours of the smoothed emission ( a gaussian smoothing kernel with @xmath42 arcsec was used ) overlaid on the observed raw photon data .
the latter are binned such that they slightly oversample the point spread function of the respective instrument which is represented by the fwhm bar in the lower left corner of each image . in the following
we discuss the serendipitous hri and ipc observations of clj0152.71357 in more detail before summarizing briefly the results of a recent targeted observation of the cluster with the bepposax satellite .
the relatively high angular resolution of the rosat hri ( @xmath43 arcsec at an off - axis angle of 14 arcmin ) allows us to investigate whether contaminating point sources might contribute to the observed pspc flux of clj0152.71357 . at an off - axis angle of 14 arcmin
a point source with a flux of about one third of the flux detected from clj0152.71357 would be detectable with the hri at greater than @xmath44 significance .
however , a secure detection of diffuse emission from clj0152.71357 with the hri would require an exposure time in excess of 100 ks . as can be seen in the rightmost panel of fig .
[ xcomp ] , no point sources are detected within the contours shown in fig .
[ iband ] but there is marginal evidence of low - surface - brightness excess emission at the position of the cluster , indicating that the overwhelming majority of the emission detected with the pspc originates from the cluster
. moreover , the emission detected with the hri shows a clear elongation along the same position angle of about 40@xmath40 as the one found in the pspc data .
although the southwestern extension of the emission detected with the pspc does not coincide with any prominent galaxy overdensity in the uh2.2 m image ( cf .
[ iband ] ) , we note that , if any of the major x - ray surface brightness peaks in the pspc image were due to a single , unvarying point source , they would have been detected with the hri .
it is noteworthy that ours is in fact not the first detection of clj0152.71357 at x - ray wavelengths . as mentioned before ,
ngc720 was not only observed with rosat but was , in 1980 , also a pointing target of observations with the einstein observatory .
a source at @xmath45 , @xmath46 ( j2000 ) ( i.e. within one arcmin of the pspc position of clj0152.71357 ) is clearly detected with the imaging proportional counter ( ipc , sequence number of pointing i5769 ) ; the einstein source catalogue assigns this source the number 496 and quotes a significance of detection of 4.8@xmath47 in the ipc broad band and within the detect cell ( harris et al . ,
we show the ipc broad band data around this position in the leftmost panel of fig .
[ xcomp ] .
the position angle ( approximately zero ) of the apparent elongation of the source is different from the one found from the rosat pspc and hri data .
however , since the ipc point spread function has a fwhm of about 90 arcsec fwhm in the broad band , the source elongation found in this short ipc observation is only marginally significant .
the same ipc source is also listed as exss 0150.21411 in the catalogue of extended einstein detections compiled by oppenheimer , helfand & gaidos ( 1997 ) who find the source significance ( presumably in the broad band ) to be 4.7 and 5.7@xmath47 within circular apertures of 1.25 and 2.35 arcmin radius .
although this source therefore appears to be sufficiently significant to be included in the emss catalogue , it remained un - identified until its re - discovery in the rdcs and warps surveys in 1996/97 .
we will come back to the ipc detection of clj0152.71357 in section [ emss_0152 ] .
a recent pointed bepposax observation of clj0152.71357 allowed the temperature and metallicity of the intra - cluster gas to be measured : della ceca et al .
( 1999 ) report values of k@xmath48 kev and @xmath49 .
this gas temperature is consistent both with the temperature estimate of @xmath50 kev obtained by us from the pspc data ( cf .
fairley et al . , 1999 )
and with the cluster x - ray luminosity temperature relation as determined by allen & fabian ( 1998 ) which predicts k@xmath51 kev .
the relatively poor angular resolution of the bepposax telescope does not allow any conclusions to be drawn about the possibility of point source contamination . using the galactic neutral hydrogen column density in the direction of the cluster of @xmath52 @xmath53 ( dickey & lockman 1990 ) as well as a metallicity of 0.5 and a gas temperature of 6.5 kev ( della ceca et al . , 1999
) , we convert the total pspc count rate of @xmath54 ct s@xmath35 ( pha channels 50 to 200 ) measured in the warps analysis into a total , unabsorbed flux of @xmath55 @xmath53 ( 0.52.0 kev ) , corresponding to an x - ray luminosity of @xmath56 \times 10^{44}$ ] in the 0.52.0 kev [ 0.33.5 kev , bolometric ] band . thus , clj0152.71357 is slightly more luminous than ms1054.40321 , making it the most x - ray luminous distant cluster detected so far .
it is also worth noting that both clj0152.71357 and ms1054.40321 are more x - ray luminous than any other known cluster at @xmath3 .
we find our measurement of the total , unabsorbed cluster flux of clj0152.71357 in the 0.52.0 kev band to be in good agreement with all other results obtained from the available x - ray observations of this system : einstein ipc : : : @xmath57 @xmath53 ( this work ) rosat pspc ( rdcs ) : : : @xmath58 @xmath53 ( della ceca et al . , 1999 ) rosat pspc ( warps ) : : : @xmath59 @xmath53 ( this work ) rosat pspc ( bright sharc ) : : : @xmath60 @xmath53 ( romer et al . , 1999 ) rosat hri : : : @xmath61 @xmath53 ( this work ) bepposax : : : @xmath62 @xmath53 ( della ceca et al . , 1999 )
all rosat flux measurements agree within the errors . by comparison ,
the ipc result is high and the bepposax result is low ; compared directly , the discrepancy between these two measurements is significant at the @xmath63 confidence level . as noted before in section [ beppo ] , the measurements of the cluster gas temperature obtained independently from the pspc data ( this work ) and the bepposax data ( della ceca et al . , 1999 )
also agree well within their errors .
although we can not rule out that some of the observed emission originates from one or more variable point sources , the overall good agreement of the source positions , x - ray fluxes and cluster gas temperatures measured for clj0152.71357 between 1980 and 1998 makes major contamination unlikely .
table [ partab ] summarizes the optical and x - ray properties of clj0152.71357 .
as mentioned in section [ ipc_disc ] , clj0152.71357 was detected with the einstein ipc at 4.8@xmath47 significance ( eoscat , harris et al . , 1990 ) ; however , the emss source catalogue ( gioia et al . , 1990a ) lists the respective ipc field ( i5769 ) as containing no serendipitous detections that would be significant at the greater than 4@xmath47 level .
since this discrepancy has been the subject of some debate , we investigate the issue in detail in the following . specifically , we address three questions : firstly
, how can the two catalogues , using ( apparently ) the same data , arrive at substantially different significances of detection for the same source ?
secondly , what are the implications of the absence of clj0152.71357 from the emss catalogue for the overall completeness of the emss cluster sample ? and thirdly , what are the consequences of our findings for the clusters included in the emss ? both the einstein ipc source catalogue ( eoscat , harris et al . , 1990 ) and the emss sample ( gioia et al .
, 1990a ) were compiled using the same source detection algorithm .
it combines a sliding cell detection algorithm ( cell geometry : @xmath64 arcmin@xmath37 ) with a maximum likelihood ( ml ) peak finding algorithm which fits a gaussian model of the instrumental point spread function ( the size of which varies with the chosen energy range ) to the data inside the detect cell .
the final source positions are taken from the ml results . while this approach is adequate for the detection of point sources , the use of a peak finding algorithm can clearly lead to non - optimal results in the case of extended sources with internal structure . while the eoscat and emss results for clj0152.71357 are obtained from the same data , the compilation procedures of the two catalogues are not entirely identical .
eoscat computes the source significance within a detect cell centred on the ml source position measured in the ipc _ broad band _
( @xmath65 kev ) , whereas the emss uses the ml source position determined in the ipc _ hard band _ ( @xmath66 kev ) .
however , both catalogues use the _ broad band _
photons within the detect cell to compute the source significance that is used as the final criterion for the inclusion of sources in the respective catalogue .
the rationale behind the two - band approach chosen by the emss team is to take advantage of the higher resolution of the ipc in the hard band without sacrificing the better photon statistics of the broad band data ( maccacaro and gioia , private communication ) .
the energy dependence of the instrumental resolution means , however , that a narrower point spread function will be used by the ml algorithm in the hard band which , as we shall see , is part of the reason why the emss missed clj0152.71357 .
we re - analyze the ipc data for field i5769 in both the hard and the broad band using the same sliding cell algorithm employed by the eoscat and emss teams . for the broad band data our analysis yields results similar to those listed in eoscat for source # 496 : at the position maximizing the source significance in the broad band we find the detect cell to contain 44 photons of which 9.9 are expected to be background .
the resulting signal - to - noise ratio ( snr ) in the broad band is 5.1 . at the ml source position quoted by eoscat we measure 38 counts in the detect cell (
eoscat : 44 ) of which 9.8 are attributed to background ( eoscat : 11.7 ) ; the resulting snr value is 4.6 ( eoscat : 4.8 ) .
our results are also in good agreement with those of oppenheimer , helfand & gaidos ( 1997 ) who , in their independent re - analysis of the einstein ipc data , find the significance of their source exss 0150.21411 to be 4.7 and 5.7@xmath47 ( presumably in the broad band ) within circular apertures of 1.25 and 2.35 arcmin radius .
the left panel in figure [ emss_det ] shows contours of the smoothed x - ray emission in the ipc broad band at the position of clj0152.71357 with both ours and the eoscat source position marked .
also shown is the contour within which our analysis finds the signal - to - noise ratio in the hard band and within the detect cell to exceed the threshold value of four .
although the astrometry used to create this image is taken directly from the einstein events list , the offset of the marked eoscat source position from the peak of the emission suggests that the satellite attitude solution used in the original eoscat analysis may have differed by some 10 to 20 arcseconds . note that the non - sphericity of the emission causes the position of the x - ray peak to lie only marginally within the @xmath67 contour .
the results of the same analysis in the ipc hard band are shown in the right panel of figure [ emss_det ] .
again we show the @xmath67 ( broad band ) contour as well as our best estimate of the source position in the broad band .
also shown is the source position returned by the ml algorithm from the emss analysis of the hard band data ( kindly provided by isabella gioia ) . due to the more pronounced bimodality of the source in the hard band ,
the ml algorithm , using a narrower model of the point spread function than in the ipc broad band , now centres on an apparent peak more than one arcmin north of the position that maximizes the source significance in the broad band . at the emss source position we measure a value of 4.1 for the broad band snr ;
slight differences in the astrometric solution caused the original emss snr measurement at this position to fall just below the threshold value of four .
consequently , the emss rejected clj0152.71357 as not sufficiently significant to be included in the emss catalogue , and thus missed what would have been the most distant and most x - ray luminous cluster in the emss sample .
as demonstrated in the previous section and illustrated in fig .
[ emss_det ] , the use of a peak - finding ml detection algorithm in the einstein ipc data analysis leads to significant offsets between the ml source position and the one maximizing the broad - band snr of clj0152.71357 .
moreover , in the presence of emission with apparent substructure the ml source positions in different energy bands can differ significantly .
if clj0152.71357 were a spherically symmetric , relaxed system with a radial surface brightness profile following a beta model , the peak centering algorithm would very likely have come closer to returning the maximal possible source significance of @xmath68 ( using the pspc count rate and assuming a core radius of 250 kpc ) .
this leads us to investigate whether the failure of the emss to include clj0152.71357 can be regarded as symptomatic of a general bias against unrelaxed clusters .
before we attempt to assess the importance of cluster substructure for the efficiency of the emss point source detection algorithm ( or , more generally , any algorithm that explicitly or implicitly assumes a unimodal source geometry ) , it should be stressed that this assumption is not vital to the source detection process .
the choice of source detection algorithm is crucial though , as the algorithm s biases can have a significant impact on the statistical quality of the resulting sample .
the emss and warps surveys , for instance , are inherently different due to differences in the source detection process .
the emss is _ x - ray surface brightness limited _ ( the selection criterion is the significance of the flux in a detect cell of fixed angular size , and the survey flux limit refers to the flux in the same detect cell , gioia et al . , 1990a ) while warps is almost completely _ x - ray flux limited _ ( the detection procedure uses a very low surface brightness threshold
see paper i and the limiting flux is the total flux of the cluster including the fraction that has escaped direct detection ) .
we investigate the redshift and luminosity dependence of the emss detection efficiency for morphologically complex sources by simulating ipc observations of two kinds of unrelaxed clusters : firstly , mergers of two similarly extended components ( akin to clj0152.71357 ) and , secondly , extended systems containing a compact but off - centre core ( similar to ms@xmath69 , see section [ morph ] ) .
table [ simtab ] gives an overview of the model parameters used in the simulations . in all simulations we assume a uniform background of @xmath70 ct s@xmath35 within the detect cell and an exposure time of 2.5 ks , values typical of the average ipc pointing ; we also blur all simulated images by convolving them with a gaussian of 33 arcsec width ( @xmath71 ) thereby accounting for the ipc point spread function ( lea & henry 1988 ) .
finally we scale the total emission from both components such that the maximal significance in the broad band is always constant at the emss threshold value of @xmath72 within the detect cell .
since we are investigating a systematic effect , no poisson noise is added to the simulated data .
figure [ sim ] summarizes the results of our simulations by showing , for a range of projected subcluster separations , the signal to noise ratio ( snr ) for a detect cell centred on the overall peak of the emission as a function of redshift . for the merger scenario we find no evidence for a systematic underestimation of the source significance at any redshift as long as the projected separation of the two cluster components remains less than about 400 kpc .
this is not surprising : at redshifts greater than 0.3 such small separations are simply not resolved by the ipc . for projected subcluster separations of more than 400 kpc ,
however , the source significance is systematically underestimated when measured around the position of the highest peak within the emission region .
the effect is small ( @xmath73 ) but redshift dependent .
the underestimation is most severe for x - ray luminous systems ( @xmath74 , @xmath7 kev ) at intermediate to high redshift , although it takes pronounced substructure on the scale of more than 700 kpc ( in projection ) to produce a noticeable effect at @xmath5 . for the offset - core scenario
we find the redshift and luminosity dependence to be reversed : now it is nearby clusters ( @xmath75 ) of moderate luminosity ( @xmath76 , @xmath7 kev ) that are most strongly affected .
these trends are consistent with the mentioned observations of clusters at @xmath5 : while clj0152.71357 is missed by the emss , ms1137.5 + 6625 and ms1054.40321 ( the first apparently relaxed , the latter a case of substructure akin to our second simulated scenario , see section [ morph ] ) are both detected . taken together , our results thus indicate that , in the presence of different kinds of substructure , the emss peak finding algorithm tends to underestimate the significance both of nearby clusters of low to moderate x - ray luminosity , and of distant clusters of very high x - ray luminosity .
while figure [ sim ] suggests that the underestimation of the snr within the detect cell is small ( 0.10.2 ) , we note that the real effect will be magnified by photon noise ( not included in our simulations ) which will cause the peak position found by the emss ml algorithm to deviate from the true position .
the resulting positional error is considerable : for the photon statistics of our simulated example we find a radius of 20 arcsec for the 90% confidence error circle of the ml peak position . in most cases measuring
the source significance around this ml fit position will yield values that are lower than those in fig .
this is underlined by the very case of clj0152.71357 , a distant , x - ray luminous cluster with substructure on the scale of 600 kpc .
its source position as determined by the emss peak finding algorithm in the ipc hard band lies so far off the x - ray centroid that the source significance in the ipc broad band is underestimated by more than @xmath71 far more than what is implied by figure [ sim ] .
although the above arguments suggest that the emss detection bias against unrelaxed clusters could be severe , a re - analysis of the einstein ipc data or numerical simulations beyond the scope of this paper would be required to accurately quantify the effect . to be conservative , the values from our simple simulation
may be taken at face value , in which case the smallness of the amplitude of the bias might cause one to believe that its impact on the emss cluster sample will be negligible .
this is , however , not necessarily true .
the emss catalogue as used for the definition of the emss cluster sample ( gioia et al .
1990b ) comprises 733 sources , 93 of which were identified as clusters of galaxies . from the distribution of source significances we estimate the number of sources with significances between 3.8 and 4@xmath47 to be about 60 ; 5 of these are expected to be clusters at redshifts greater than 0.2 . since the fraction of significantly unrelaxed clusters at these redshifts is almost certainly non - negligible ( see section [ morph ] ) , and considering the inherent uncertainties of our crude analysis , we are left with the conclusion that the number of distant and x - ray luminous , but unrelaxed , clusters missed by the emss is likely to be of the order of a few . while not immediately alarming ,
this estimate is still disconcertingly high given that the emss cluster sample contains only a handful of distant x - ray luminous clusters to begin with .
in addition to the cosmological relevance of the sheer existence of a distant cluster as x - ray luminous as clj0152.71357 , the complex optical and x - ray morphology of this cluster provides further important clues . as can be seen from fig .
[ iband ] , clj0152.71357 consists of at least two pronounced subclusters which are ( in projection ) about 600 kpc apart and are likely to merge within a few gyr ( assuming a true spatial separation of one to a few mpc and equal masses of a few @xmath77 for the two main cluster components ) . the fact that clj0152.71357 is still in the process of formation has several interesting implications .
firstly , the subclusters observed today are likely to have existed as separate clusters of @xmath78 erg s@xmath35 ( 0.52.0 kev ) at a redshift considerably greater than unity , and , secondly , the x - ray luminosity of clj0152.71357 is bound to increase as the merger proceeds , possibly rendering clj0152.71357 more x - ray luminous than any cluster observed to date .
thirdly , clj0152.71357 is the third x - ray selected cluster ( out of five ) detected at @xmath14 that shows pronounced substructure and is distinctly non - virialized , in contrast to the morphologically much more diverse local cluster population .
the last point is illustrated by figure [ morphfig ] which shows adaptively smoothed x - ray flux contours of all three @xmath5 clusters for which high - resolution x - ray images are currently available .
our hri data reduction corrects for particle background as well as exposure time variations using software kindly provided by steve snowden . for each cluster
we align and merge all available observations which yields total exposure times as follows .
ms1137.5 + 6625 ( @xmath29 ) : 98.0 ks ; ms1054.40321 ( @xmath24 ) : 186.6 ks ; rxj1716.6 + 6708 ( @xmath26 ) : 167.2 ks .
the final images use a pixel size of @xmath79 arcsec@xmath37 thus slightly oversampling the hri point - spread function . using asmooth ( ebeling et al.1999 ) the hri counts image
is then adaptively smoothed with a gaussian kernel the size of which is adjusted such that the local significance of the signal within the kernel exceeds 99% .
the boxy thick contours in fig .
[ morphfig ] mark the regions within which the signal is high enough for this criterion to be met and within which all structure apparent in the contour plots is thus significant at greater than 99% confidence .
the dashed boxes illustrate the effect of a placement of the emss detect cell on the highest peak in the emission region . according to fig .
[ morphfig ] , the only relaxed cluster of the three is ms1137.5 + 6625 while both ms1054.40321 and rxj1716.6 + 6708 exhibit significantly nonspherical emission with off - centre cores .
although this high - redshift sample is still too small to allow more quantitative conclusions , the rarity of relaxed systems is intriguing and may indicate that we are beginning to actually observe the epoch of formation of the majority of massive clusters .
the discovery of the x - ray luminous , unrelaxed galaxy cluster clj0152.71357 in the warps cluster survey has important implications for our understanding of the evolution of clusters as a function of x - ray luminosity and redshift .
clj0152.71357 was previously detected in a pointed observation with the einstein ipc ; however , due to an underestimation of its significance the source is missing from the emss catalogue .
clj0152.71357 would have been the most distant and most x - ray luminous cluster in the emss sample .
simulations of ipc observations of unvirialized clusters show that the absence of clj0152.71357 from the emss cluster sample may reflect a general bias of the emss against unrelaxed , distant clusters .
we can not currently quantify accurately the amplitude of such a bias ; however , conservative estimates suggest that of the order of a few x - ray luminous clusters may have been missed at @xmath38 .
we attempt to assess the frequency of significant substructure in distant x - ray luminous clusters by comparing the x - ray morphology of all such systems observed to date with the rosat hri .
although the resulting sample is small , we find tentative evidence that highly unrelaxed systems such as clj0152.71357 may indeed be common at high redshift .
an important implication of our findings is that quantitative cosmological conclusions based on measurements of the abundance of x - ray luminous , distant clusters ought to be regarded with caution .
any comparison of cluster space densities with the predictions of structure formation models assumes that the clusters used satisfy the collapse criteria specified in those models ( e.g.press-schechter ) . in the light of our morphological observations
we add a cautionary note that it is possible that many of these distant systems do not yet meet these conditions .
clearly this would seriously complicate the measurement of cosmological quantities using cluster counts .
however , it could also offer a new means to tackle these questions through detailed observation and a dynamical analysis of merger rates in statistically selected ` proto - clusters ' .
as far as the representative nature of current cluster samples is concerned , the dynamical state of a cluster could complicate matters beyond the detection bias discussed in section [ emss_bias ] .
numerical simulations by ricker ( 1998 ) indicate that shock fronts created in the primary collision of two merging clusters can increase the total x - ray luminosity of the merging system by up to an order of magnitude compared to the combined x - ray luminosity of the progenitor clusters .
while this effect is expected to be prominent only for less than half the sound crossing time ( typically a few times @xmath80 yr ) , it may still , to some extent , counteract any detection bias against merging clusters ( see section [ emss_bias ] ) by causing such systems to be preferentially detected in x - ray flux limited surveys . if cluster mergers are indeed common at high redshift and the net x - ray emission of these systems does not adequately distinguish between formed and forming systems , we may be forced to develop much more sophisticated models and data analysis strategies in order to draw secure conclusions about the physical mechanisms and cosmological implications of cluster evolution .
deeper observations and more detailed analyses of a sizeable , representative sample of distant x - ray luminous clusters are required to conclusively address these issues .
observing time with chandra s acis - i imaging spectrometer is scheduled in cycle 1 for a high - resolution x - ray study of clj0152.71357 ; the cluster is also a gto ( guaranteed time observation ) target of xmm . in combination with ongoing observations at optical and infrared wavelengths from the ground
these x - ray observations will allow in - depth studies of the internal dynamics and mass distribution of this system .
a detailed optical study of the cluster galaxy population with the keck-2 telescope is underway and first results will be presented shortly .
for now we only mention that our recent multi - slit spectroscopy observations yielded more than 20 accordant redshifts for this system .
llllll a & @xmath81 & @xmath82 & @xmath83 & @xmath84 & ca h&k + b & @xmath85 & @xmath86 & @xmath87 & @xmath88 & ca h&k , g + c & @xmath89 & @xmath90 & @xmath91 & @xmath92 & 4000 break + d & @xmath93 & @xmath94 & @xmath95 & @xmath96 & ca h&k , g + e & @xmath97 & @xmath98 & @xmath99 & @xmath100 & ca h&k , g + f & @xmath101 & @xmath102 & @xmath103 & @xmath104 & ca h&k , g + ll redshift @xmath105 & 0.8325 + @xmath106 & @xmath52 @xmath53 + pspc exposure time & 19,912 s + detected vtp count rate ( pha 50200 ) & @xmath107 s@xmath35 + total count rate ( pha 50200 ) & @xmath108 s@xmath35 + total unabsorbed flux ( 0.52.0 kev ) & @xmath109 erg @xmath53 s@xmath35 + observed icm temperature & @xmath110 kev ( rosat pspc ) + & @xmath111 kev ( bepposax ) + rest frame luminosity ( 0.52.0 kev ) & @xmath112 h@xmath113 erg s@xmath35 + rest frame luminosity ( 0.33.5 kev ) & @xmath114 h@xmath113 erg s@xmath35 + rest frame luminosity ( bolometric ) & @xmath115 h@xmath113 erg s@xmath35 + we thank the telescope time allocation committees of the university of hawaii , uco / lick , the kitt peak national observatory , and the michigan - dartmouth - mit observatory for their generous support of the warps optical follow - up programme .
several members of the emss team provided us with advice and much useful information about details of the emss data processing and source selection procedure .
their help is gratefully acknowledged .
special thanks to pat henry for many fruitful discussions that led to substantial improvements in the presentation and interpretation of our results .
alexey vikhlinin kindly proofread an early version of the manuscript and made a number of helpful suggestions .
thanks also to steve snowden for kindly providing the latest version of his @xmath116 package .
this work has made use of data obtained through the www interfaces to the gsfc / heasarc and mpe rosat public data archives , as well as the stsci digitized sky survey .
he acknowledges financial support from sao contract sv4 - 64008 and nasa ltsa grant nag 5 - 8253 .
lrj thanks the uk pparc for financial support .
allen s.w . &
fabian a.c .
1998 , mnras , 297 , l57 bahcall , n.a . & cen , r. 1992 , apj , 398 , l81 burke , d. et al .
, 1997 , apj , 488 , l83 carlberg , r.g .
1997 , apj , 479 , l19 castander , f.j . ,
ellis , r.s . ,
frenk , c.s . , dressler , a. , gunn , j.e .
1994 , , 424 , l79 collins c.a .
1997 , apj , 479 , l120 crawford c.s . & fabian a.c . 1996 , mnras , 282 , 1483 degrandi et al . 1999 ,
apj , 513 , l17 della ceca r. , scaramella r. , gioia i.m . , rosati p. , fiore f. , squires g. , 1999 , a&a , accepted deltorn j .-
, le fvre o. , crampton d. , dickinson m. 1997 , apj , 483 , l21 dickey , j.m . & lockman , f.j .
1990 , ara&a , 28 , 215 donahue m. , voit g.m . , gioia i. , luppino g. , hughes j.p .
, stocke j.t . 1998 , apj , 502 , 550 ebeling h. , voges w. , bhringer h. , edge a.c .
, huchra j.p . , briel u.g . 1996 , mnras , 281 , 799 ebeling h. et al .
1997 , apj , 479 , l101 ebeling h. , edge a.c . , allen s.w .
, crawford c.s . , fabian a.c . ,
huchra j.p . 1999 ,
mnras , submitted ebeling h. , white d.a . &
rangarajan f.v.n .
1999 , mnras , accepted ebeling h. & henry j.p .
1999 , in preparation eke v.r .
, cole s. , frenk c.s . ,
henry p.h . , 1998 , mnras , 298 , 114 ellingson e. , yee h.k.c .
, abraham r.g . ,
morris s.l .
, carlberg r.g .
1998 , apjs , 116 , 247 fairley b.w .
1999 , mnras , submitted gehrels n. 1986 , , 303 , 336 gioia i.m .
, maccacaro t. , schild r.e . ,
wolter a. , stocke j.t .
, morris s.l .
, henry j.p .
1990a , apjs , 72 , 567 gioia i.m .
, henry j.p . , maccacaro t. , morris s.l . , stocke j.t . , wolter a. 1990b , , 356 , l35 gioia i.m . & luppino g.a .
1994 , apjs , 94 , 583 gioia i.m .
, henry j.p . , mullis c.r . , ebeling h. , wolter a. 1999 , aj , 117 , 2608 gunn j.e . , hoessel j.g . , oke j.b .
1986 , apj , 306 , 30 harris d.e .
1990 , _ the einstein observatory catalog of ipc x - ray sources _ , smithsonian astrophysical observatory , cambridge , mass .
1 - 4 henry j.p . , gioia i.m . , maccacaro t. , morris s.l . ,
stocke j.t . , wolter a. 1992 , , 386 , 408 henry j.p .
1997 , aj , 114 , 1293 henry j.p . 1999 , in preparation jones l.r .
, scharf c. , ebeling h. , perlman e. , wegner g. , malkan m. , horner d. 1998 , , 495 , 100 lea s.m . & henry j.p .
, 1988 , 332 , 81 luppino g.a . &
gioia i.m .
1995 , apj , 445 , l77 mullis c.r .
, gioia i.m . &
henry j.p . 1998 , in _ the hot universe _ , koyama , kitamoto & itoh ( eds ) , dordrecht , kluwer academic , sym .
no.188 , p473 nichol r.c .
1997 , , 481 , 644 nichol r.c .
1999 , , 521 , l21 oke j.b .
1995 , pasp , 107 , 375 oke j.b .
, postman m. & lubin l.m .
1998 , apj , 116 , 549 oppenheimer b.r .
, helfand d.j . &
gaidos e.j . 1997 , aj , 113 , 2134 oukbir j. & blanchard a. 1992 , a&al , 262 , 21 oukbir j. & blanchard a. 1997 , a&a , 317 , 10 osterbrock d.e . , martel a. 1992 , pasp , 104 , 76 postman m. , lubin l.m .
, gunn j.e . , oke j.b . , hoessel j.g . , schneider d.p . , christensen j. 1996 , aj , 111 , 615 rector t.a .
, stocke j.t . & perlman e.s . , 1999 ,
apj , in press ( astro - ph/9903295 ) ricker p.m. 1998 , apj , 496 , 670 romer a.k .
et al . , 1999 , apjs , accepted rosati p. , della ceca r. , burg r. , norman c. , giacconi r. 1995 , , 445 , l11 rosati p. , della ceca r. , norman c. , giacconi r. 1998 , , 492 , l21 rosati p. , stanford s.a .
, eisenhardt p.r .
, elston r. , spinrad h. , stern d. , dey a. 1999 , aj , 118 , 76 scharf c.a .
, jones l.r .
, ebeling h. , perlman e. , malkan m. , wegner g. 1997 , , 477 , 79 , paper i scodeggio m. , olsen l.f . , da costa l. , slijkhuis r. , benoist c. , deul e. , erben t. , hook r. , nonino m. , wicenec a. , zaggia s. 1999 , a&as , 137 , 83 stanford s.a .
, elston r. , eisenhardt p.r . , spinrad h. , stern d. , dey a. 1997 , aj , 1997 , 114 , 2232 viana p.t.p . &
liddle a.r .
1996 , mnras , 281 , 323 vikhlinin a. , mcnamara b.r . , forman w. , jones c. , quintana h. , hornstrup a. , 1998a , apj , 498 , l21 vikhlinin a. , mcnamara b.r . , forman w. , jones c. , quintana h. , hornstrup a. , 1998b , apj , 502 , 558 vikhlinin a. , mcnamara b.r . , hornstrup a. , quintana h. , forman w. , jones c. , way m. 1999 , apj , 520 , l1 | the warps ( wide angle rosat pointed survey ) team reviews the properties and history of discovery of clj0152.71357 , an x - ray luminous , rich cluster of galaxies at a redshift of @xmath0 . at @xmath1
( @xmath2 kev ) clj0152.71357 is the most x - ray luminous cluster known at redshifts @xmath3 .
the high x - ray luminosity of the system suggests that massive clusters may begin to form at redshifts considerably greater than unity .
this scenario is supported by the high degree of optical and x - ray substructure in clj0152.71357 , which is similarly complex as that of other x - ray selected clusters at comparable redshift and consistent with the hypothesized picture of cluster formation by mass infall along large - scale filaments .
x - ray emission from clj0152.71357 was detected already in 1980 with the einstein ipc . however , because the complex morphology of the emission caused its significance to be underestimated , the corresponding source was not included in the cluster sample of the einstein extended medium sensitivity survey ( emss ) and hence not previously identified .
simulations of the emss source detection and selection procedure performed by us suggest a general , mild bias of the emss cluster sample against x - ray luminous clusters with pronounced substructure .
if highly unrelaxed , merging clusters are common at intermediate to high redshift ( as is suggested by the current data ) they could create a bias in some samples as the morphological complexity of mergers may cause them to fall below the flux limit of surveys that make the implicit or explicit assumption of a unimodal spatial source geometry .
conversely , the enhanced x - ray luminosity of mergers might cause them to , temporarily , rise above the flux limit .
either effect could lead to erroneous conclusions about the evolution of the comoving cluster space density .
a high fraction of morphologically complex clusters at high redshift would also call into question the validity of evolutionary studies ( and , specifically , cosmological conclusions ) which implicitly or explicitly assume that the systems under investigation are virialized . |
linear matrix inequalities ( lmis ) are versatile modeling objects in the context of convex programming , with many engineering applications @xcite .
an @xmath8-dimensional lmi set is defined as @xmath9 where the @xmath10 are given symmetric matrices of size @xmath11 and @xmath12 means positive semidefinite . from the characteristic polynomial @xmath13 it follows from e.g. (
* theorem 20 ) that @xmath14 hence the lmi set @xmath15 is basic semialgebraic : it is the intersection of polynomial sublevel sets . from linearity of @xmath16 and convexity of the cone of positive semidefinite matrices
, it also follows that @xmath15 is convex .
hence lmi sets are _ convex basic semialgebraic_. , title="fig:",width=566 ] + one may then wonder whether all convex basic semialgebraic sets are lmi . in @xcite , helton and vinnikov answer by the negative , showing that in the plane ( @xmath1 ) some convex basic semialgebraic sets can not be lmi .
an elementary example is the so - called tv screen set defined by the fermat quartic @xmath17 see figure [ tv_screen ] .
assume that the set @xmath15 defined in ( [ lmi ] ) has a non - empty interior , and choose a point @xmath18 in this interior , i.e. @xmath19 where @xmath20 means positive definite .
a segment starting from @xmath18 attains the boundary of @xmath15 when the determinant @xmath21 vanishes .
the remaining polynomial inequalities @xmath22 , @xmath23 only isolate the convex connected component containing @xmath18 .
this motivated helton and vinnikov @xcite to study semialgebraic sets defined by a single polynomial inequality @xmath24 the set @xmath25 is called an algebraic interior with defining polynomial @xmath3 , and it is equal to @xmath26 when @xmath5 is convex . with these notations , the question addressed in @xcite is as follows : what are the conditions satisfied by a polynomial @xmath3 so that @xmath5 is an lmi set ? for notational simplicity we will assume , without loss of generality , that @xmath27 , so that @xmath5 contains the origin , and hence we can normalize @xmath3 so that @xmath28 .
if @xmath29 for some matrix mapping @xmath16 we say that @xmath3 has a determinantal representation .
in particular , the polynomial @xmath30 in ( [ basic ] ) has a symmetric linear determinantal representation .
consider an lmi set @xmath15 as in ( [ lmi ] ) and define @xmath31 as the determinant of the symmetric pencil @xmath16 .
note that @xmath32 , the dimension of @xmath16 .
define the algebraic variety @xmath33 and notice that the boundary of @xmath15 is included in @xmath6 . indeed , a point @xmath34 along the boundary of @xmath15 is such that the rank of @xmath35 vanishes .
since the origin belongs to @xmath15 it holds @xmath36 .
now consider a line passing through the origin , parametrized as @xmath37 where @xmath38 is a parameter and @xmath39 is any vector with unit norm .
for all @xmath40 , the symmetric matrix @xmath41 has only real eigenvalues as a pencil of @xmath42 , and its determinant @xmath43 has only real roots .
therefore , a given polynomial level set @xmath5 as in ( [ poly ] ) is lmi only if the polynomial @xmath44 has only real roots for all @xmath40 , it must satisfy the so - called _ real zero condition _ @xcite .
geometrically it means that a generic line passing through the origin must intersect the variety ( [ var ] ) at @xmath45 real points .
the set @xmath5 is then called _ rigidly convex _ , a geometric property implying convexity .
a striking result of @xcite is that rigid convexity is also a sufficient condition for a polynomial level set to be an lmi set in the plane , i.e. when @xmath1 .
for example , it can be checked easily that the tv screen set ( [ tv ] ) is not rigidly convex since a generic line cuts the quartic curve only twice . in the litterature on partial differential equations , polynomials satisfying real zero condition
are also called hyperbolic polynomials , and the corresponding lmi set is called the hyperbolicity cone , see @xcite for a survey , and @xcite for connections between real zero and hyperbolic polynomials . in passing ,
note the fundamental distinction between an lmi set ( as defined above ) and a semidefinite representable set , as defined in @xcite .
a semidefinite representable set is the projection of an lmi set : @xmath46 where the variables @xmath47 , sometimes called liftings , are instrumental to the construction of the set through an extended pencil @xmath48 .
such a set is called a lifted lmi set .
it is convex semialgebraic , but in general it is not basic .
however , it can be expressed as a union of basic semialgebraic sets . in the case of the tv screen set ( [ tv ] )
a lifted lmi representation follows from the extended pencil @xmath49\ ] ] obtained by introducing two liftings .
it seems that the problem of knowing which convex semialgebraic sets are semidefinite representable is still mostly open , see @xcite for recent developments .
once rigid convexity of a plane set , or equivalently the real zero property of its defining polynomial , is established , the next step is constructing an lmi representation .
algebraically , given a real zero bivariate polynomial @xmath50 of degree @xmath11 , the problem consists in finding symmetric matrices @xmath51 , @xmath52 and @xmath53 of dimension @xmath11 such that @xmath54 and @xmath36 .
if the @xmath55 are symmetric complex - valued matrices , this is a well - studied problem of algebraic geometry called _ determinantal representation _ , see @xcite for a classical reference and @xcite for more recent surveys and extensions to trivariate polynomials .
if one relaxes the dimension constraint ( allowing the @xmath55 to have dimension larger than @xmath11 ) and the symmetry constraint ( allowing the @xmath55 to be non - symmetric ) , then results from linear systems state - space realization theory ( in particular linear fractional representations , lfrs ) can be invoked to design computer algorithms solving constructively the determinantal representation problem .
for example , the lfr toolbox for matlab @xcite is a user - friendly package allowing to find non - symmetric determinantal representations : .... > > lfrs x1 x2 > > f=1/(1-x1
^ 4-x2 ^ 4 ) .. lfr - object with 1 output(s ) , 1 input(s ) and 0 state(s ) .
uncertainty blocks ( globally ( 8 x 8) ) : name dims type real / cplx full / scal bounds x1 4x4 lti r s [ -1,1 ] x2 4x4 lti r s [ -1,1 ] .... the software builds a state - space realization of order 8 of the transfer function @xmath56 .
this indicates that a non - symmetric real pencil @xmath16 of dimension 8 could be found that satisfies @xmath57 , as evidenced by the following script using the symbolic math toolbox ( matlab gateway to maple ) : .... > > syms
x1 x2 > > d = diag([ones(1,4)*x1 ones(1,4)*x2 ] ) ; > > f = eye(8)-f.a*d f = [ 1 , -x1 , 0 , 0 , 0 , 0 , 0 , 0 ] [ 0 , 1 , -x1 , 0 , 0 , 0 , 0 , 0 ] [ 0 , 0 , 1 , -x1 , 0 , 0 , 0 , 0 ] [ -x1 , 0 , 0 , 1 , -x2 , 0 , 0 , 0 ] [ 0 , 0 , 0 , 0 , 1 , -x2 , 0 , 0 ] [ 0 , 0 , 0 , 0 , 0 , 1 , -x2 , 0 ] [ 0 , 0 , 0 , 0 , 0 , 0 , 1 , -x2 ] [ -x1 , 0 , 0 , 0 , -x2 , 0 , 0 , 1 ] > > det(f ) ans = -x2 ^ 4 + 1-x1 ^ 4 .... note that lfr and state - space realization techniques are not restricted to the bivariate case , but they result in pencils of large dimension ( typically much larger than the degree of the polynomial ) , and there is apparently no easy way to reduce the size of a pencil .
if one insists on having the @xmath55 symmetric , then results from non - commutative state - space realizations can be invoked to derive a determinantal representation , at the price of relaxing the sign constraint on @xmath51 .
an implementation is available in the nc mathematica package @xcite . here too , these techniques may produce pencils of large dimension .
now if one insists on having symmetric @xmath55 of minimal dimension @xmath11 , then two essentially equivalent constructive procedures are known in the bivariate case to derive hermitian complex - valued @xmath55 from a defining polynomial @xmath50 of degree @xmath11 .
real symmetric solutions must then be extracted from the set of complex hermitian solutions . the first one is based on the construction of a basis for the riemann - roch space of complete linear systems of the algebraic plane curve @xmath6 given in ( [ var ] ) .
the procedure is described in @xcite : one needs to find a curve of degree @xmath58 touching @xmath6 at each intersection point , i.e. the gradients must match .
the algorithm is illustrated in @xcite .
it is not clear however how to build a touching curve ensuring @xmath36 .
the second determinantal representation algorithm is sketched in @xcite and in much more detail in @xcite .
it is based on complex riemann surface theory @xcite .
explicit expressions for the @xmath55 matrices are given via theta functions .
numerically , the key ingredient is the computation of the period matrix of the algebraic curve and the corresponding abel - jacobi map .
the period matrix of a curve can be computed numerically with the algcurves package of maple , see @xcite and the tutorial @xcite for recent developments , including new algorithms for explicit computations of the abel - jacobi map . a working computer implementation taking @xmath50 as input and producing the @xmath55 matrices as output is still missing however .
the focus of this paper is mostly on computational methods and numerical algorithms .
the contribution is twofold .
first in section [ rigid ] we describe an algorithm for detecting rigid convexity in the plane . given a bivariate polynomial @xmath50 , the algorithm uses a hybrid symbolic - numerical method to detect whether the connected component of the sublevel set ( [ poly ] ) containing the origin is rigidly convex .
the problem boils down to deciding whether a univariate polynomial matrix is positive semidefinite .
this is a well - known problem in linear systems theory , for which numerical linear algebra algorithms are available ( namely eigenvalue decomposition ) , as well as a ( more expensive but more flexible ) semidefinite programming formulation .
then in section [ rational ] we describe an algorithm for solving the determinantal representation problem for algebraic plane curves of genus zero .
the algorithm is essentially symbolic , using bzoutians , but it assumes that a rational parametrization of the curve is available . the idea behind the algorithm is not new , and can be traced back to @xcite , as surveyed recently in @xcite .
an algorithm for detecting rigid convexity of a connected component delimited by such curves readily follows .
extensions to positive genus algebraic plane curves and higher dimensional sets are mentioned in section [ extensions ] .
in particular we survey the case of cubic plane curves and cubic surfaces which are well understood .
the case of quartic ( and higher degree ) curves seems to be mostly open , and computer implementations of determinantal representations are still missing .
similarly , checking rigid convexity in higher dimensions seems to be computational challenging since it amounts to deciding whether a multivariate polynomial matrix is positive semidefinite .
in this section we design an algorithm to assess whether the connected component delimited by a bivariate polynomial around the origin is rigidly convex .
the idea is elementary and consists in formulating algebraically the geometric condition of rigid convexity of the set @xmath5 defined in ( [ poly ] ) : a line passing through the origin cuts the algebraic curve @xmath6 defined in ( [ var ] ) a number of times which is equal to the total degree @xmath11 of the defining bivariate polynomial @xmath59 a line passing through the origin can be parametrized as : @xmath60 where @xmath61 and @xmath62 . along this line
, we define @xmath63 as a univariate polynomial of degree @xmath11 which vanishes on @xmath6 .
moreover @xmath64 is monic since @xmath65 .
the remaining coefficients are laurent polynomials @xmath66 with real coefficients , also called trigonometric cosine polynomials .
set @xmath5 is rigidly convex if and only if this polynomial has only real roots , i.e. if the number of intersections of the line with the curve @xmath6 is maximal
. a well - known result of real algebraic geometry ( * ? ? ?
* theorem 4.57 ) states that a univariate polynomial @xmath64 of degree @xmath11 has only real roots if and only if its hermite matrix is positive semidefinite .
the hermite matrix is the @xmath11-by-@xmath11 moment matrix of a discrete measure supported with unit weights on the roots @xmath67 of polynomial @xmath64 ( note that these roots are not necessarily distinct ) .
it is a symmetric hankel matrix whose entries @xmath68 are newton sums @xmath69 .
the newton sums are elementary symmetric functions of the roots that can be expressed explicitly as polynomial functions of the coefficients of @xmath64 .
recursive expressions are available to compute the @xmath70 , or equivalently , @xmath71 where @xmath72 is a companion matrix of polynomial @xmath64 , i.e. a matrix with eigenvalues @xmath73 , see e.g. ( * ? ? ?
* proposition 4.54 ) . recall that coefficients of the polynomial @xmath64 given in ( [ pt ] ) are laurent polynomials .
it follows that the hermite matrix of @xmath64 is a symmetric trigonometric polynomial matrix of dimension @xmath11 , that we denote by @xmath74 .
we have proved the following result .
[ rh ] the bivariate polynomial @xmath3 is rigidly convex if and only if its hermite matrix @xmath74 is positive semidefinite along the unit circle .
coefficients of @xmath74 are explicit polynomial expressions of the coefficients of @xmath3 . the problem of checking positive semidefiniteness of a polynomial matrix on the unit circle is generally referred to as ( discrete - time ) spectral factorization .
it is a well - known problem of systems and circuit theory @xcite .
the positivity condition can also be defined on the imaginary axis ( continuous - time spectral factorization ) or the real axis .
various numerical methods are available to solve this problem @xcite .
several algorithms are implemented in the polynomial toolbox for matlab @xcite . in increasing order of complexity
, we can distinguish between * newton - raphson algorithms : the spectral factorization problem is formulated as a quadratic polynomial matrix equation which is then solved iteratively @xcite . at each step , a linear polynomial matrix equation must be solved @xcite .
quadratic ( resp . linear ) convergence is ensured locally if the polynomial matrix is positive definite ( resp .
semidefinite ) ; * polynomial operations : a sequence of elementary operations is carried out in the ring of polynomials to reduce the polynomial matrix to some canonical form , see @xcite and @xcite for a recent survey .
these algorithms are cheap computationally but their numerical behavior ( stability ) is unclear ; * algebraic riccati equation : using state - space realization , the problem is formulated as a quadratic matrix equation , which in turn can be solved via a matrix eigenvalue decomposition with a particular structure @xcite ; * semidefinite programming : polynomial matrix positivity is formulated as a convex semidefinite program , see @xcite and the recent surveys @xcite .
the particular structure of this semidefinite program can be exploited in interior - point schemes , in particular when forming the gradient and hessian .
general purpose semidefinite solvers can be used as well . the semidefinite programming formulation of discrete - time polynomial matrix factorization , a straightfoward transposition of the continuous - time case studied in @xcite , is as follows . the symmetric trigonometric polynomial matrix @xmath75 of size @xmath11 is positive semidefinite along the unit circle if and only if there is a symmetric matrix @xmath76 of size @xmath77 such that @xmath78 + \left[\begin{array}{ccc } i \\ \hline & \ddots \\ & & i \\ 0 & \cdots & 0 \end{array}\right ] p \left[\begin{array}{c|ccc } i & & & 0 \\ & \ddots & & \vdots \\ & & i & 0 \end{array}\right ] \\ & & - \left[\begin{array}{ccc } 0 & \cdots & 0 \\ \hline i \\ & \ddots \\ & & i \\ \end{array}\right ] p \left[\begin{array}{c|ccc } 0 & i \\ \vdots & & \ddots\\ 0 & & & i \end{array}\right ] \\ & = & \left[\begin{array}{c|c } h_0 & h_{01 } \\
\hline h_{01}^t & 0 \end{array}\right ] + \left[\begin{array}{c } b^t \\ a^t \end{array}\right ] p \left[\begin{array}{c|c } b & a \end{array}\right ] - \left[\begin{array}{c } d^t \\ c^t \end{array}\right ] p \left[\begin{array}{c|c } d & c \end{array}\right ] \succeq 0 .
\end{array}\ ] ] notice that the columns and rows of the above matrix are indexed w.r.t .
increasing powers of @xmath40 in such a way that @xmath79^t l(p ) \left[\begin{array}{c } i \\ z \\ \cdots \\
z^d \end{array}\right ] = h(z).\ ] ] positive semidefiniteness of @xmath80 then amounts to the existence of a polynomial sum - of - squares decomposition of @xmath74 . from the schur decomposition @xmath81 with
@xmath82 $ ] it follows that @xmath83 polynomial matrix @xmath84 is called a spectral factor .
if the lmi problem ( [ spf ] ) is feasible , then it admits a whole family of solutions . assuming that @xmath85 , maximizing the trace of @xmath76 subject to the lmi constraint ( [ spf ] ) yields a particular solution @xmath86 such that @xmath87 .
it follows that the schur complement of @xmath88\ ] ] w.r.t .
@xmath89 vanishes , where symmetric entries are denoted by @xmath90 .
this means that @xmath86 satisfies the quadratic matrix equation @xmath91 called the ( discrete - time ) algebraic riccati equation . in this case , the spectral factor @xmath92 in ( [ pspf ] ) is square non - singular .
consider the component of set ( [ poly ] ) around the origin delimited by the cubic polynomial @xmath93 , see figure [ cubic_curve ] .
, title="fig:",width=566 ] + using the substitution ( [ trigo ] ) we obtain @xmath94 from the companion matrix @xmath95\ ] ] we build ( symbolically ) the hermite matrix @xmath96.\ ] ] solving ( numerically ) the lmi ( [ spf ] ) with sedumi interfaced with yalmip yields the spectral factorization ( [ pspf ] ) with factor ( in matlab notation ) @xmath97\ ] ] which certifies numerically that @xmath3 is rigidly convex .
let us apply the algorithm to test rigid convexity of the tv quartic level set ( [ tv ] ) with @xmath98 , see figure [ tv_screen ] .
we obtain @xmath99 and the hermite matrix @xmath100.\ ] ] from the zero diagonal entries and the non - zero entries in the corresponding rows and columns we conclude that @xmath74 can not be positive semidefinite and hence that the tv quartic level set is not rigidly convex .
since the hermite matrix @xmath74 has a hankel structure , and positive definite symmetric hankel matrices have a conditioning ( ratio of extreme eigenvalues ) which can be bounded below by an exponential function of the matrix size @xcite , it may be appropriate to apply a congruence transformation on matrix @xmath74 , also called scaling . for example , if @xmath101 is positive definite for some @xmath102 ( say @xmath103 , but other choices are also possible ) , it admits a schur factorization @xmath104 with @xmath105 orthogonal and @xmath106 diagonal non - singular .
if @xmath106 is reasonably well - conditioned , we can test positive semidefiniteness of the modified trigonometric polynomial matrix @xmath107 along the unit circle , which is such that @xmath108 is the identity matrix .
if @xmath106 is not well - conditioned , we can still use @xmath109 which is such that @xmath108 is a diagonal matrix .
the impact of this data scaling on the numerical behavior of the semidefinite programming or algebraic riccati equation solvers is however out of the scope of this paper .
in the case that the algebraic curve @xmath6 in ( [ var ] ) has genus zero , i.e. the curve is rationally parametrizable , an alternative algorithm can be devised to test rigid convexity of a connected component delimited by @xmath6 .
the algorithm is based on elimination theory .
it uses a particular symmetric form of a resultant called the bzoutian . as a by - product
, the algorithm also solves the determinantal representation problem in this case . as surveyed recently in @xcite , the key idea of using bzoutians in the context of determinantal representations
can be traced back to @xcite .
starting from the implicit representation @xmath110 of curve @xmath6 , with @xmath3 a bivariate polynomial of degree @xmath11 , we apply a parametrization algorithm to obtain an explicit representation @xmath111 with @xmath112 univariate polynomials of degree @xmath11 .
algorithms for parametrizing an implicit algebraic curve are described in @xcite .
an implementation by mark van hoeij is available in the algcurves package of maple .
the coefficients of @xmath112 are generally found in an algebraic extension of small degree over the field of coefficients of @xmath3 . with the help of resultants
, we can eliminate the variable @xmath113 in parametrization ( [ param ] ) and recover an implicit equation ( [ implicit ] ) , see ( * ? ? ?
* section 3.3 ) . to address this implicitization problem ,
we make use of a particular resultant , the bzoutian , see ( * ? ? ?
* section 5.1.2 ) . given two univariate polynomials @xmath114 of the same degree @xmath11 (
if the degree is not the same , the smallest degree polynomial is considered as a degree @xmath11 polynomial with zero leading coefficients ) build the following bivariate polynomial @xmath115 called the bzoutian of @xmath116 and @xmath117 , and the corresponding symmetric matrix @xmath118 of size @xmath119 with entries @xmath120 bilinear in coefficients of @xmath116 and @xmath117 .
as shown e.g. in ( * ? ? ?
* section 5.1.2 ) , the determinant of the bzoutian matrix is the resultant , so we can use it to derive the implicit equation ( [ implicit ] ) of a curve from the explicit equations ( [ param ] ) .
[ pencil ] given polynomials @xmath121 in ( [ param ] ) , a polynomial @xmath122 in ( [ implicit ] ) is given by @xmath123 where @xmath124 is a symmetric pencil of size @xmath11 .
* proof : * rewrite the system of equations ( [ param ] ) as @xmath125 and use the bzoutian resultant to eliminate indeterminate @xmath113 and obtain conditions for a point @xmath126 to belong to the curve . the bzoutian matrix is @xmath127 .
linearity in @xmath128 follows from bilinearity of the bzoutian and the common factor @xmath129.@xmath130 lemma [ pencil ] provides an implicit equation of curve ( [ implicit ] ) in symmetric linear determinantal form .
once polynomial @xmath3 is in symmetric linear determinantal form as in lemma [ pencil ] , checking rigid convexity of the connected component containing the origin @xmath131 amounts to testing positive definiteness of @xmath132 .
[ definite ] the bzoutian matrix @xmath133 is positive semidefinite if and only if polynomial @xmath134 and @xmath135 have only real roots that interlace . *
proof : * the signature ( number of positive eigenvalues minus number of negative eigenvalues ) of the bzoutian of @xmath134 and @xmath135 is the cauchy index of the rational function @xmath136 , the number of jumps of the function from @xmath137 to @xmath138 minus the number of jumps from @xmath138 to @xmath137 , see ( * ? ? ?
* definition 2.53 ) or ( * ? ? ?
* theorem 9.4 ) .
it is maximum when @xmath133 is positive definite .
this occurs if and only if the roots of @xmath134 and @xmath135 are all real and interlace.@xmath130 [ rigidbez ] the connected component around the origin delimited by curve ( [ implicit ] ) is rigidly convex if and only if @xmath139 .
* proof : * since @xmath36 , the set admits the lmi representation @xmath140 , which is equivalent to being rigidly convex . @xmath130 if the connected component around the origin is not ridigly convex , it may happen that there is another rigidly convex connected component elsewhere . to find it
, it suffices to determine a point @xmath34 such that @xmath141 .
this is equivalent to solving a bivariate lmi problem
. we can apply primal - dual interior - point methods @xcite to solve this semidefinite programming problem , since the function @xmath142 is a strictly convex self - concordant barrier for the interior of the lmi set .
if the lmi set is bounded , minimizing @xmath143 yields the analytic center of the set .
if the lmi set is empty , the dual semidefinite problem yields a farkas certificate of infeasibility .
however , in the bivariate case a point @xmath34 satisfying @xmath141 can be found more easily with real algebraic geometry and univariate polynomial root extraction .
a first approach consists in identifying the local minimizers of function @xmath143 .
they are such that the gradient @xmath144 of @xmath3 vanishes , i.e. they are such that @xmath145 we can characterize these minimizers by eliminating one variable , say @xmath146 , from the system @xmath147 , and solving for the other variable @xmath148 via polynomial root extraction .
resultants can be used for that purpose .
a second approach consists in finding points on the boundary of the lmi set , which are such that @xmath149 and either @xmath150 or @xmath151 . here too , resultants can be applied to end up with a polynomial root extraction problem . from the points generated by these two procedures ,
we keep only those satisfying @xmath152 , an inequality that can be certified by testing the signs of the coefficients of the characteristic polynomial of @xmath16 , as explained in the introduction .
let @xmath153 . with the parametrization function of the algcurves package of maple
, we obtain a rational parametrization @xmath154 with the bezoutmatrix function of the linearalgebra package , we build the corresponding symmetric pencil @xmath155\ ] ] whose determinant ( up to a constant factor ) is equal to @xmath3 .
the eigenvalues of @xmath156 are equal to @xmath157 ( double ) and @xmath158 .
they are all non - negative which indicates that the origin lies on the boundary of an lmi region defined by @xmath152 .
values of @xmath148 at local optima satisfying the system of cubic equations ( [ grad ] ) can be found with maple as follows : .... > p:=x1
^ 2*(x1 ^ 2+x2 ^ 2)-2*(x1 ^ 2+x2 ^ 2-x2)^2 : > solve(resultant(diff(p , x1),diff(p , x2),x1 ) ) ; 0 , 0 , 0 , 1 , 1/2 , 3+sqrt(5 ) , 3-sqrt(5 ) , 3+sqrt(5 ) , 3-sqrt(5 ) .... from which it follows that , say , the point @xmath159 , @xmath160 is such that @xmath161 .
, title="fig : " ] + the corresponding lmi region together with the quartic capricorn curve are represented on figure [ capricorn_curve ] .
, title="fig:",width=566 ] + let @xmath162 . with the bzoutians we obtain the following pencil @xmath163.\ ] ] we can check that @xmath156 has eigenvalues @xmath164 and @xmath157 ( triple ) and this is the only point for which @xmath152 .
it follows that the convex set delimited by the curve @xmath149 is not lmi , see figure [ bean_curve ] .
in this paragraph we outline some potential extensions of the results to algebraic plane curves of positive genus and varieties of higher dimensions . the case of cubic plane algebraic curves
is well understood , see e.g. @xcite or @xcite .
singular cubics ( genus zero ) can be handled via bzoutians as in section [ rational ] .
smooth cubics ( genus one ) , also called elliptic curves , can be handled via their hessians .
let @xmath50 be a cubic polynomial that we homogeneize to @xmath165 .
define its 3-by-3 symmetric hessian matrix @xmath166 with entries @xmath167 and the corresponding hessian @xmath168 .
the elliptic curve @xmath149 has 9 inflection points , or flexes , satisfying @xmath169 , and 3 of them are real .
since @xmath3 and @xmath170 share the same flexes and the hessian matrix yields a symmetric linear determinantal representation for @xmath170 , we can use homotopy to find a determinantal representation for @xmath3 .
for real @xmath42 define the parametrized hessian @xmath171 and find @xmath172 satisfying @xmath173 at a real flex @xmath34 by solving a cubic equation . as a result , we obtain three distinct symmetric pencils not equivalent by congruence transformation .
one of the may be definite hence lmi .
for example , let @xmath174 .
build the hessian @xmath175 and the parametrized hessian @xmath176 .
polynomial @xmath177 matches @xmath144 at flex @xmath178 for @xmath179 yielding the following three representations @xmath180 \\ \\
f^2(x ) & = & 4^{-\frac{1}{3}}\left[\begin{array}{ccc } 1 + 3x_1 & \star & \star \\ -x_2 & -1-x_1 & \star\\ -1+x_1 & -x_2 & 1-x_1 \end{array}\right ] \\ \\
f^3(x ) & = & 4^{-\frac{1}{3}}\left[\begin{array}{ccc } 1 - 3x_1 & \star & \star\\ -x_2 & 1-x_1 & \star \\ 1+x_1 & x_2 & 1+x_1 \end{array}\right ] \end{array}\ ] ] such that @xmath181 for all @xmath182 .
only the first one generates an lmi set @xmath183 .
the case of algebraic plane curves of positive genus and degree equal to four ( quartic ) or higher is mostly open .
whereas rigid convexity of higher degree polynomials can be checked with the proposed approach , there is no known implementation of an algorithm that produces symmetric linear determinantal ( and hence lmi ) representations in this case . for quartics ,
contact curves can be recovered from bitangents . in @xcite complex
symmetric linear determinantal representations of the quartic @xmath184 could be derived from the equations of the bitangents found previously by cayley for this particular curve .
bzoutians can be generalized to the multivariate case , as surveyed in @xcite . in lemma [ pencil ]
we derived a symmetric linear determinantal representation by eliminating the variable @xmath113 in the system of equations @xmath185 corresponding to a rational parametrization @xmath186 , @xmath187 of the curve @xmath188 . in the positive genus case , such a rational parametrization is not available , but we can still define a system of equations @xmath189 describing the curve @xmath188 after eliminating variables @xmath190 and @xmath191 .
define the discrete differentials @xmath192 and the quadratic form @xmath193 = \det \left[\begin{array}{ccc } x_1-u_1 & -1 & 0 \\ x_2-u_2 & 0 & -1 \\
p(u_1,u_2 ) & \partial_1 p(u , v ) & \partial_2 p(u , v ) \end{array}\right ] = \sum_{\alpha,\beta } f_{\alpha,\beta } u^{\alpha}v^{\beta}\ ] ] using bi - indices @xmath194 and @xmath195 . then the matrix @xmath16 of the quadratic form is a symmetric pencil satisfying @xmath196 where @xmath197 is an extraneous factor .
we hope that @xmath197 does not depend on @xmath128 , even though this can not be guaranteed in general .
for example , in the case of the fermat curve @xmath198 whose genus is three , using the multires package for maple @xcite , we could obtain @xmath199\ ] ] which is such that @xmath200 , i.e. @xmath201 .
the case @xmath202 , i.e. cubic surfaces , is well understood , see @xcite for a full constructive development .
all the self - adjoint linear determinantal representations can be obtained from the tritangent planes .
the number of non - equivalent representations depends on the number and class of real lines among the 27 complex lines of the surface .
see @xcite for a nice survey on cubic surfaces .
, title="fig:",width=566 ] + a well - known example is the cayley cubic @xmath203 whose algebraic equation is @xmath204 under involutary linear mapping @xmath205 = \frac{1}{2 } \left[\begin{array}{cccc } 1 & 1 & 1 & 1 \\ 1 & 1 & -1 & -1 \\ 1 & -1 & 1 & -1 \\ 1 & -1 & -1 & 1 \end{array}\right ] \left[\begin{array}{c } u_0 \\ u_1 \\ u_2 \\ u_3 \end{array}\right]\ ] ] the dehomogenized ( @xmath206 ) algebraic equation becomes @xmath207 = \det f(x)\ ] ] which is the determinant of the 3x3 moment matrix of the maxcut lmi relaxation .
the surface @xmath149 is represented on figure [ cayley_cubic ] , using the surf visualization package . in particular
, we can easily identify the convex connected component containing the origin , described by the lmi @xmath152 .
the component has four vertices , or singularities , for which the rank of @xmath16 drops down to one .
in general , only curves and cubic surfaces admit generically a determinantal representation .
when @xmath208 or @xmath209 and no lifting is allowed , the hypersurface @xmath149 must be highly singular to have a determinantal representation @xcite , and hence , a fortiori , an lmi representation .
this leaves however open the existence of alternative algorithms consisting in constructing symmetric linear determinantal representations of modified polynomials @xmath210 , with @xmath197 globally nonnegative , say @xmath211 or @xmath212 for @xmath213 large enough .
finally , let us conclude by remarking that , as a by - product of the proof leading to lemma [ rh ] , checking numerically rigid convexity of a scalar polynomial when @xmath7 amounts to checking positivity of a multivariate hermite matrix .
see e.g. @xcite for recent developments on the use of semidefinite programming for multivariate trigonometric polynomial matrix positivity .
this work benefited from feedback by ( in alphabetical order ) daniele faenzi , leonid gurvits , frdric han , bill helton , toma koir , salma kuhlmann , jean - bernard lasserre , adrian lewis , bernard mourrain , pablo parrilo , jens piontkowski , mihai putinar , bernd sturmfels , jean valls and victor vinnikov . | given a polynomial @xmath0 in @xmath1 variables , a symbolic - numerical algorithm is first described for detecting whether the connected component of the plane sublevel set @xmath2 containing the origin is rigidly convex , or equivalently , whether it has a linear matrix inequality ( lmi ) representation , or equivalently , if polynomial @xmath3 is hyperbolic with respect to the origin .
the problem boils down to checking whether a univariate polynomial matrix is positive semidefinite , an optimization problem that can be solved with eigenvalue decomposition . when the variety @xmath4 is an algebraic curve of genus zero , a second algorithm based on bzoutians is proposed to detect whether @xmath5 has an lmi representation and to build such a representation from a rational parametrization of @xmath6 .
finally , some extensions to positive genus curves and to the case @xmath7 are mentioned . * keywords * + polynomial , convexity , linear matrix inequality , real algebraic geometry . |
Subsets and Splits