article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
during recent years , the experimental observation of a large number of so - called @xmath9 states has inspired tremendous effort to unravel their nature beyond the conventional quark model @xcite . in @xmath10 ,
the belle collaboration first reported the observation of @xmath2 in the @xmath11 invariant mass spectrum of @xmath12 @xcite which was subsequently confirmed by the babar collaboration @xcite in the same channel .
it was also discovered in proton - proton / antiproton collisions by the d0 @xcite , cdf @xcite , and lhcb collaborations @xcite .
the @xmath2 is the first , and perhaps the most renowned exotic candidate ; however , its nature is still ambiguous due to its peculiar properties .
first , only an upper bound of the total width has been measured experimentally : @xmath13 mev @xcite , which is tiny compared to typical hadronic widths .
second , the mass is located close to the @xmath14 threshold , @xmath15 mev @xcite , which leads to speculation that the @xmath2 is presumably a meson - meson molecular state @xcite .
many studies on the production and decay of the @xmath2 have been carried out in order to understand its nature .
the discrimination of a compact multiquark configuration and a loosely bound hadronic molecule configuration is one important aspect . from this point , it is also valuable to look for the counterpart in the bottom sector , denoted as @xmath1 following the notation in ref .
@xcite , as states related by heavy quark symmetry may have universal behaviors .
since the @xmath1 is very heavy and its quantum numbers @xmath16 , it is less likely discovered at the current electron - positron collision facilities , though the super kekb may provide an opportunity in @xmath17 radiative decays @xcite .
the production of @xmath1 at the lhc and the tevatron @xcite and other exotic states at hadron colliders @xcite have been extensively investigated .
on one hand , it is shown that the production rates at the lhc and the tevatron are sizeable @xcite .
on the other hand , the search for @xmath1 also depends on reconstructing the @xmath1 , which motivates us to study the @xmath1 decays . in ref .
@xcite , we have studied the radiative decays of @xmath18 ( @xmath19 ) , with @xmath1 being a candidate for the meson - meson molecular state , and found that the partial widths into @xmath20 are about @xmath21 kev . since this meson is expected to be far below threshold and the mass difference between neutral and charged @xmath22 mesons is very small , the isospin - violating decay mode , for instance @xmath23 , is highly suppressed , and this may explain the escape of @xmath1 in the recent cms search @xcite . as a consequence ,
the isospin - conserved decays @xmath24 , on which we will focus this paper , will be of high priority . to calculate @xmath24 , we investigate the intermediate meson - loop ( iml ) contributions .
as is well known , iml transitions have been one of the important nonperturbative transition mechanisms in many processes , and their impact on the heavy quarkonium transitions has been noticed for a long time @xcite .
recently , this mechanism has been applied to study @xmath22 decays @xcite , the production and decays of exotic states @xcite and a global agreement with experimental data were obtained .
thus , this may be an effective approach to deal with the @xmath1 hidden bottomonium decays .
the rest of this paper is organized as follows . in sec .
[ sec : formula ] , we introduce the formalism used in this work .
numerical results are presented in sec .
[ sec : results ] , and the summary is given in sec . [
sec : summary ] .
with the @xmath25 as the intermediate states.,scaledwidth=70.0% ] a state of the heavy meson pair can be decomposed in terms of the total spin of the @xmath26 pair @xmath27 and the total angular momentum @xmath28 of the rest degrees of freedom , for which the latter includes both the total spin of the light quark pair @xmath29 and any orbital momentum @xcite . in the s - wave @xmath25 molecular picture of @xmath1 ,
the heavy quark spin structure is mandated to be @xmath30 @xcite . however , in the tetraquark picture of @xmath1 , the total spin @xmath27 can be either @xmath31 or @xmath21 . as a result ,
the dominant hidden bottomonium decay channel is @xmath32 in the molecule picture of @xmath1 , while the hidden bottomonium decay channels of @xmath1 can be those channels with @xmath33 or @xmath21 final bottomonium . from this point of view
, it seems that a molecule @xmath1 may have a bigger branching ratio of @xmath34 than a tetraquark @xmath1 . generally speaking
, we should include all the possible intermediate meson loops in the calculation .
in reality , the breakdown of the local quark - hadron duality allows us to pick up the leading contributions as a reasonable approximation @xcite . since we consider the isoscalar @xmath1 as an @xmath35-wave molecular state with @xmath36 given by the superposition of @xmath37 and @xmath38 hadronic configurations as @xmath39 meson loops here .
the coupling of @xmath1 to the bottomed meson pair is based on the following effective lagrangian , @xmath40 + h.c.,\end{aligned}\ ] ] where @xmath41 denotes the coupling constant . for the @xmath1 below the @xmath35-wave @xmath42 threshold , the effective coupling of this state
is related to the probability of finding the two - hadron component in the physical wave function of the bound states and the binding energy , @xmath43 @xcite @xmath44 where @xmath45 , @xmath46 is the reduced mass . here , the coupling constant @xmath41 in eq .
( [ eq : coupling - xb ] ) is based on the assumption that @xmath1 is a shallow bound state where the potential binding the mesons is mostly contact range .
based on the heavy quark symmetry , the leading - order effective lagrangian for the @xmath47 can be expressed as @xcite @xmath48 where @xmath49 and @xmath50 correspond to the bottom meson isodoublets .
@xmath51 is the antisymmetric levi - civita tensor and @xmath52 . due to the heavy quark symmetry
, the following relationships of the couplings can be used @xcite @xmath53 where @xmath54 , @xmath55 and @xmath56 denote the mass and decay constant of @xmath47 , respectively .
the decay constant @xmath56 can be extracted from the @xmath57 , @xmath58 where @xmath59 is the electromagnetic fine - structure constant . using the mass and leptonic decay width of the @xmath47 state , @xmath60 kev @xcite
, one can obtain @xmath61 mev .
for the lagrangians relevant to the light vector mesons used in this work , we can write them in the chiral and heavy quark limits @xcite , @xmath62 where the heavy meson couplings to light vector mesons have the following relationships , @xmath63 with the parameters @xmath64 , @xmath65 , and @xmath66 @xcite . based on the relevant lagrangians given above , the loop transition amplitudes in fig .
[ fig : loops ] can be expressed in a general form as follows , @xmath67 where @xmath68 and @xmath69 are the vertex functions and the denominators of the intermediate meson propagators , respectively .
for example , in fig .
[ fig : loops ] ( b ) , @xmath70 are the vertex functions for the initial @xmath1 , final bottomonium and final @xmath71 , respectively .
@xmath72 are the denominators for the intermediate @xmath73 , @xmath74 and @xmath75 propagators , respectively . in order to take care of the off - shell effects of the exchanged particles @xcite
, we adopt a dipole form factor @xmath76 with @xmath77 , and the qcd energy scale @xmath78 mev .
many phenomenological studies have suggested the cutoff parameter @xmath79 .
the explicit expression of transition for @xmath34 amplitudes is given in appendix [ appendix - a ] .
.predicted partial widths ( in units of kev ) of the @xmath1 decays .
the parameter in the form factor is chosen as @xmath80 , @xmath81 , and @xmath82 , respectively . [
cols="^,^,^,^",options="header " , ] on the @xmath83 using dipole form factors with @xmath80 ( solid lines ) , @xmath84 ( dashed lines ) , and @xmath85 ( dotted lines ) , respectively.,scaledwidth=70.0% ] both the tetraquark model @xcite and hadronic molecular calculations @xcite have predicted the existence of the @xmath1 . in ref .
@xcite , a. ali _ et al .
_ predicted the mass of the lowest - lying @xmath86 @xmath87 tetraquark to be @xmath88 mev , while the mass of the @xmath3 molecule is a few tens of mev higher @xcite . in ref .
@xcite , the mass was predicted to be @xmath89 mev , which corresponds to a binding energy @xmath90 mev .
these studies can provide a range for the binding energy , and in the following we will choose a few illustrative values : @xmath91 mev . in table [ tab : results-1 ] , we list the predicted partial widths by choosing the dipole form factor and three values for the cutoff parameter @xmath92 . from this table , we can see that the widths for the @xmath34 are about tens of kevs .
the particle data group gives an upper bound for the total width of @xmath2 to be @xmath93 mev @xcite .
if we consider the similarity between @xmath1 and @xmath2 , the full width for @xmath1 should be narrow .
hence , our results would indicate a sizeable branching fraction for @xmath34 . in fig .
[ fig : widthonexb ] , we present the partial widths for the @xmath0 in terms of the @xmath83 with the dipole form factors @xmath94 ( solid lines ) , @xmath81 ( dashed line ) , and @xmath82 ( dotted lines ) , respectively . as shown in this figure , there is no cusp structure in the curve .
this is because the mass of @xmath1 lies below the intermediate @xmath42 threshold .
the uncertainties caused by the form factors indicate our limited knowledge on the applicability of the effective lagrangian .
however , fortunately , the dependence of the partial widths are not drastically sensitive , which indicates a reasonable cutoff of the ultraviolet contributions by the empirical form factors .
both the coupling strength of @xmath1 in eq .
( [ eq : coupling - xb ] ) and the threshold effects influence the binding energy dependence of the partial width . on the one hand , as shown in eq .
( [ eq : coupling - xb ] ) , the coupling strength of @xmath1 increases with the binding energy @xmath83 . on the other hand
, the threshold effects decrease with @xmath83 .
both the coupling strength of @xmath1 and the threshold effects change quickly in the small @xmath83 region and slowly in the large @xmath83 region . as a result
, the behavior of the partial width is relatively sensitive at small @xmath83 , while it becomes smooth at large @xmath83 . as mentioned above , the coupling constant @xmath41 in eq .
( [ eq : coupling - xb ] ) is based on the assumption of a shallow bound state , so our determination of this coupling has uncertainties which are mainly due to the nonperturbative pion exchange . for the interaction between two heavy hadrons forming a bound state , one can formulate an effective field theory ( eft ) description @xcite of hadronic molecules , which is analogous to that of the eft formulation of the nucleon - nucleon interaction @xcite .
there are two main simplifications for heavy hadrons : the first one is that heavy quark symmetry severely constrains the low - energy interactions among them @xcite ; the second is that pion exchanges are usually perturbative and only produce small effects @xcite .
the only exception is the isoscalar in the bottom sector where the pions might be nonperturbative due to the large masses of the bottom mesons @xcite . in ref .
@xcite , the studies indicate that the relative errors of @xmath95 are about 20% for the isoscalar @xmath1 .
so even if we consider the impact of this effect , the estimated order of the magnitude for the branching ratio @xmath24 may also be sizeable and can be measured in experiments . in the heavy quark symmetry limit
, one may correlate the transition matrix element of the @xmath0 with that of @xmath96 . generally speaking , with the experimental measurements of the masses of @xmath2 and the experimental lower limit branching ratio of @xmath97 @xcite
, one can first use the heavy quark symmetry to determine the cutoff parameter value @xmath92 and then give a round estimation of the possible experimental measurements on @xmath34 .
however , we should also notice that the phase space dependence behavior of the partial widths may be quite different for these two decay channels . for the process @xmath97
, the allowed phase space is very small , while the phase space is large enough for the process @xmath34 .
maybe systematic experimental and theoretical studies on these decay processes can test this point . in the case of hidden bottomonium decays of @xmath1 ( such as @xmath34 and @xmath98 ) with @xmath1 being an @xmath35-wave @xmath99 molecule candidate , since the heavy quarks in the @xmath99 meson pair have to recombine to form the final bottomonium , the transition from the @xmath99 meson pair into the @xmath100 plus @xmath71 ( or photon ) occurs at a distance much smaller than both the size of the @xmath1 as a hadronic molecule and the range of forces between the @xmath22 and @xmath101 mesons . as a consequence ,
these processes are not sensitive to the @xmath99 wave function at long distances which is governed by the binding energy , but rather they are determined by the short distance part of the @xmath1 . as shown in ref .
@xcite , the radiative decays @xmath102 , and especially their ratio @xmath103 , are not sensitive to the long - range structure of the @xmath1 , and thus they can not be used to rule out the picture that the @xmath1 is dominantly a hadronic molecule . in order to further investigate the structure of @xmath1 , it is necessary to explore other decay channels , and the radiative decay @xmath104 might give the best hint about the molecular nature of @xmath1 . in the case of @xmath104 , one of the constituent hadrons ( @xmath22 ) is in the final state and the rest of the final particles are products of the decay of the other constituent hadron ( @xmath105 ) of the @xmath1 molecule . in these processes , the relative distance between the @xmath3 pair can be as large as allowed by the size of the @xmath1 , since the final state is produced by the decay of @xmath105 instead of the rescattering transition .
so this process is sensitive to the long - distance part of the wave function of a hadronic molecule and can be used to probe the long structure of @xmath1 .
the experimental observation of @xmath1 , hunting for hidden bottomonium decays ( @xmath6 , @xmath106 ) and @xmath104 can help us understand the structure of @xmath1 .
in this work , we have investigated the isospin - conserved hidden bottomonium decay @xmath0 , where @xmath1 is taken to be the counterpart of @xmath2 in the bottomonium sector as a meson - meson molecule candidate .
since the mass of this state may be far below the @xmath3 threshold , the isospin - violating decay channel @xmath23 would be highly suppressed and stimulate the importance of the isospin - conserved decay channel @xmath0 .
we explore the rescattering mechanism with the effective lagrangian based on the heavy quark symmetry .
the calculated partial width of @xmath0 is about tens of kevs .
taking into account the fact that the total width of @xmath1 may be smaller than a few mev like @xmath2 , the calculated branching ratios may reach to orders of @xmath5 .
this study of hidden bottomonium decay along with the previous work on production rates in hadron - hadron collisions @xcite and radiative decays of @xmath1 @xcite have indicated a promising prospect to find the @xmath1 at hadron colliders , in particular , the lhc , and we suggest that our experimental colleagues perform an analysis .
also , the associated studies of hidden bottomonium decays @xmath6 , @xmath7 , and @xmath8 may help us investigate the structure of @xmath1 systematically .
such attempt will likely lead to the discovery of the @xmath1 and enrich the exotics spectrum in the heavy quarkonium sector .
the authors thank xiao - hai liu , qian wang , and wei wang for useful discussions .
this work is supported by the national natural science foundation of china ( grant no .
11275113 ) .
[ appendix - a ]
in this appendix , we give the transition amplitudes for the intermediate heavy meson loops in fig . [
fig : loops ] in the framework of the ela .
@xmath107 , @xmath108 , and @xmath109 are the four - vector momenta for initial state @xmath1 , final heavy quarkonium @xmath47 , and final light vector meson @xmath71 , respectively .
@xmath110 , @xmath111 and @xmath112 are the four - vector momenta for the intermediate heavy mesons .
@xmath113 , @xmath114 and @xmath115 are the polarization vectors for initial state , final quarkonium , and final light vector meson , respectively .
@xmath116}&= & \int \frac { d^4q_2 } { ( 2\pi)^4}[g_{x_{b } } \varepsilon_{1\mu } ] [ g_{\upsilon(1s ) bb}(q_1-q_2)_\rho \varepsilon_2^{*\rho } ] [ -2f_{b^*bv } \varepsilon_{\lambda \theta \phi\kappa } p_3^\lambda \varepsilon_3^ { * \theta } ( q_2+q_3)^\phi ] \nonumber \\ & & \times \frac { i } { q_1 ^ 2-m_1 ^ 2 } \frac { i } { q_2 ^ 2-m_2 ^ 2 } \frac { i(-g^{\mu\kappa } + q_3^\mu q_3^\kappa / m_3 ^ 2 ) } { q_3 ^ 2-m_3 ^ 2 } { \cal f}(m_{2 } , q_2
^ 2 ) , \nonumber \\ { \cal a}_{bb^ * [ b^ * ] } & = & \int \frac { d^4q_2 } { ( 2\pi)^4}[g_{x_{b } } \varepsilon_{1\mu } ] [ -g_{\upsilon(1s ) b^*b}\varepsilon_{\rho \sigma \xi\tau}p_2^\rho \varepsilon_2^{*\sigma}q_2^\xi ] [ 2g_{b^*b^*v } q_{2\theta } \varepsilon_3^ { * \theta } g_{\phi\kappa } + 4f_{b^*b^*v } ( p_{3}^{\theta}\varepsilon_{3\phi}^ * - p_{3\phi}\varepsilon_{3}^{*\theta})g_{\kappa \theta } ] \nonumber \\ & & \times \frac { i } { q_1 ^ 2-m_1 ^ 2 } \frac { i(-g^{\tau\phi } + q_2^\tau q_2^\phi / m_2 ^ 2 ) } { q_2 ^ 2-m_2 ^ 2 } \frac { i(-g^{\mu\kappa } + q_3^\mu q_3^\kappa / m_3 ^ 2 ) } { q_3 ^ 2-m_3 ^ 2 } { \cal f}(m_{2 } , q_2 ^ 2 ) , \nonumber \\
{ \cal a}_{b^*b [ b ] } & = & \int \frac { d^4q_2 } { ( 2\pi)^4}[g_{x_{b } } \varepsilon_{1\mu } ] [ -g_{\upsilon(1s ) b^*b}\varepsilon_{\rho \sigma \xi\tau}p_2^\rho \varepsilon_2^{*\sigma}q_1^\xi ] [ -g_{bbv } ( q_2+q_3)_\lambda \varepsilon_3^{*\lambda } ] \nonumber \\ & & \times \frac { i(-g^{\mu\tau } + q_1^\mu q_1^\tau / m_1 ^ 2 ) } { q_1 ^ 2-m_1 ^ 2 } \frac { i } { q_2 ^ 2-m_2 ^ 2 } \frac { i } { q_3 ^ 2-m_3 ^ 2 } { \cal f}(m_{2 } , q_2 ^ 2 ) \nonumber \\ { \cal a}_{b^*b [ b^*]}&= & \int \frac { d^4q_2 } { ( 2\pi)^4}[g_{x_{b } } \varepsilon_{1\mu } ] [ g_{\upsilon(1s ) b^*b^*}\varepsilon_2^{*\rho } ( ( q_2-q_1)_\rho g_{\sigma\xi}+p_{2\xi } g_{\rho\sigma}+p_{2\sigma } g_{\rho\xi})][2f_{b^*bv } \varepsilon_{\lambda \theta \phi\kappa } p_3^\lambda \varepsilon_3^ { * \theta } ( q_2+q_3)^\phi ] \nonumber \\ & & \times \frac { i(-g^{\mu\sigma } + q_1^\mu q_1^\sigma / m_1 ^ 2 ) } { q_1 ^ 2-m_1 ^ 2 } \frac { i(-g^{\xi\kappa } + q_2^\xi q_2^\kappa / m_2 ^ 2 ) } { q_2 ^ 2-m_2 ^ 2 } \frac { i } { q_3 ^ 2-m_3 ^ 2 } { \cal f}(m_{2 } ,
q_2 ^ 2 ) , \nonumber \\\end{aligned}\ ] ] n. brambilla , s. eidelman , b. k. heltsley , r. vogt , g. t. bodwin , e. eichten , a. d. frawley and a. b. meyer _ et al . _ ,
j. c * 71 * , 1534 ( 2011 ) [ arxiv:1010.5827 [ hep - ph ] ] .
e. s. swanson , phys .
rept . * 429 * , 243 ( 2006 ) [ hep - ph/0601110 ] .
e. eichten , s. godfrey , h. mahlke and j. l. rosner , rev .
phys . * 80 * , 1161 ( 2008 ) [ hep - ph/0701208 ] .
m. b. voloshin , prog .
phys . * 61 * , 455 ( 2008 ) [ arxiv:0711.4556 [ hep - ph ] ] .
s. godfrey and s. l. olsen , ann .
nucl . part .
* 58 * , 51 ( 2008 ) [ arxiv:0801.3867 [ hep - ph ] ] .
n. drenska , r. faccini , f. piccinini , a. polosa , f. renga and c. sabelli , riv .
nuovo cim . * 033 * , 633 ( 2010 ) [ arxiv:1006.2741 [ hep - ph ] ] .
g. t. bodwin , e. braaten , e. eichten , s. l. olsen , t. k. pedlar and j. russ , arxiv:1307.7425 .
s. k. choi _ et al . _
[ belle collaboration ] , phys .
lett . * 91 * , 262001 ( 2003 ) [ hep - ex/0309032 ] . c. bignamini , b. grinstein , f. piccinini , a. d. polosa and c. sabelli , phys .
lett . * 103 * , 162001 ( 2009 ) [ arxiv:0906.0882 [ hep - ph ] ] .
a. esposito , f. piccinini , a. pilloni and a. d. polosa , j. mod .
phys . * 4 * , 1569 ( 2013 ) [ arxiv:1305.0527 [ hep - ph ] ] .
f. k. guo , u. g. meiner and w. wang , commun .
phys . * 61 * , 354 ( 2014 ) [ arxiv:1308.0193 [ hep - ph ] ] .
g. li and w. wang , phys .
b * 733 * , 100 ( 2014 ) [ arxiv:1402.6463 [ hep - ph ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 727 * , 57 ( 2013 ) [ arxiv:1309.0250 [ hep - ex ] ] .
p. moxhay , phys .
d * 39 * , 3497 ( 1989 ) .
h. -y .
cheng , c. -k .
chua and a. soni , phys .
d * 71 * , 014030 ( 2005 ) [ hep - ph/0409317 ] . c. -d .
lu , y. -l .
shen and w. wang , phys .
d * 73 * , 034005 ( 2006 ) [ hep - ph/0511255 ] .
q. wang , c. hanhart and q. zhao , phys .
* 111 * , 132003 ( 2013 ) [ arxiv:1303.6355 [ hep - ph ] ] .
g. li , c. s. an , p. y. li , d. liu , x. zhang and z. zhou , arxiv:1412.3221 [ hep - ph ] .
x. -h .
liu and g. li , phys .
d * 88 * , 014013 ( 2013 ) [ arxiv:1306.1384 [ hep - ph ] ] .
g. li and x. -h .
liu , phys .
d * 88 * , 094008 ( 2013 ) [ arxiv:1307.2622 [ hep - ph ] ] . c. -w .
zhao , g. li , x. -h .
liu and f. -l .
shao , eur .
j. c * 73 * , 2482 ( 2013 ) .
m. b. voloshin , phys .
d * 87 * , no . 7 , 074011 ( 2013 ) [ arxiv:1301.5068 [ hep - ph ] ] . f. -k .
guo , c. hanhart and u. -g .
meiner , phys .
lett . * 103 * , 082003 ( 2009 ) [ erratum - ibid . *
104 * , 109901 ( 2010 ) ] [ arxiv:0907.0521 [ hep - ph ] ] .
d. -y .
chen , x. liu and s. -l .
zhu , phys .
d * 84 * , 074016 ( 2011 ) [ arxiv:1105.5193 [ hep - ph ] ] .
f. -k .
guo , c. hanhart , g. li , u. -g .
meissner and q. zhao , phys .
d * 82 * , 034025 ( 2010 ) [ arxiv:1002.2712 [ hep - ph ] ] .
f. -k .
guo , c. hanhart , g. li , u. -g .
meiner and q. zhao , phys .
d * 83 * , 034013 ( 2011 ) [ arxiv:1008.3632 [ hep - ph ] ] .
d. -y .
chen , x. liu and t. matsuki , phys .
d * 84 * , 074032 ( 2011 ) [ arxiv:1108.4458 [ hep - ph ] ] .
d. y. chen , x. liu and t. matsuki , chin .
c * 38 * , 053102 ( 2014 ) [ arxiv:1208.2411 [ hep - ph ] ] .
d. -y .
chen , x. liu and t. matsuki , phys .
d * 88 * , 036008 ( 2013 ) [ arxiv:1304.5845 [ hep - ph ] ] .
d. -y .
chen , x. liu and t. matsuki , phys .
d * 88 * , 014034 ( 2013 ) [ arxiv:1306.2080 [ hep - ph ] ] .
m. b. voloshin , phys .
d * 84 * , 031502 ( 2011 ) [ arxiv:1105.5829 [ hep - ph ] ] .
a. e. bondar , a. garmash , a. i. milstein , r. mizuk and m. b. voloshin , phys .
d * 84 * , 054010 ( 2011 ) [ arxiv:1105.4473 [ hep - ph ] ] .
x. li and m. b. voloshin , phys .
d * 87 * , no .
9 , 094033 ( 2013 ) [ arxiv:1303.2949 [ hep - ph ] ] .
x. li and m. b. voloshin , phys .
d * 88 * , no . 3 , 034012 ( 2013 ) [ arxiv:1307.1072 [ hep - ph ] ] .
q. wang , phys .
d * 89 * , 114013 ( 2014 ) [ arxiv:1403.2243 [ hep - ph ] ] .
m. b. voloshin , phys .
b * 604 * , 69 ( 2004 ) [ hep - ph/0408321 ] .
m. b. voloshin , phys .
d * 85 * , 034024 ( 2012 ) [ arxiv:1201.1222 [ hep - ph ] ] .
s. weinberg , phys .
rev . * 137 * , b672 ( 1965 ) .
v. baru , j. haidenbauer , c. hanhart , y. kalashnikova and a. e. kudryavtsev , phys .
b * 586 * , 53 ( 2004 ) [ hep - ph/0308129 ] .
p. colangelo , f. de fazio and t. n. pham , phys .
d * 69 * , 054023 ( 2004 ) [ hep - ph/0310084 ] .
r. casalbuoni , a. deandrea , n. di bartolomeo , r. gatto , f. feruglio and g. nardulli , phys .
* 281 * , 145 ( 1997 ) [ hep - ph/9605342 ] . c. isola , m. ladisa , g. nardulli and p. santorelli , phys .
d * 68 * , 114001 ( 2003 ) [ arxiv : hep - ph/0307367 ] .
d. becirevic , b. blossier , e. chang and b. haas , phys .
b * 679 * , 231 ( 2009 ) [ arxiv:0905.3355 [ hep - ph ] ] .
x. -q .
li , d. v. bugg and b. -s .
zou , phys .
d * 55 * , 1421 ( 1997 ) .
m. karliner and s. nussinov , jhep * 1307 * , 153 ( 2013 ) [ arxiv:1304.0345 [ hep - ph ] ] .
e. epelbaum , h. w. hammer and u. g. meissner , rev .
* 81 * , 1773 ( 2009 ) [ arxiv:0811.1338 [ nucl - th ] ] .
t. mehen and j. w. powell , phys . rev .
d * 84 * , 114013 ( 2011 ) [ arxiv:1109.3479 [ hep - ph ] ] .
m. t. alfiky , f. gabbiani and a. a. petrov , phys .
b * 640 * , 238 ( 2006 ) [ hep - ph/0506141 ] .
s. fleming , m. kusunoki , t. mehen and u. van kolck , phys .
d * 76 * , 034006 ( 2007 ) [ hep - ph/0703168 ] .
m. p. valderrama , phys .
d * 85 * , 114037 ( 2012 ) [ arxiv:1204.2400 [ hep - ph ] ] . | in this work , we study the isospin conserved hidden bottomonium decay of @xmath0 , where @xmath1 is taken to be the counterpart of the famous @xmath2 in the bottomonium sector as a candidate for the meson - meson molecule .
since it is likely that the @xmath1 is below the @xmath3 threshold and the mass difference between the neutral and charged bottom meson is small compared to the binding energy of the @xmath1 , the isospin violating decay mode @xmath4 would be greatly suppressed .
we use the effective lagrangian based on the heavy quark symmetry to explore the rescattering mechanism of @xmath0 and calculate the partial widths .
our results show that the partial width for the @xmath0 is about tens of kevs . taking into account the fact that the total width of @xmath1 may be smaller than a few mev like @xmath2
, the calculated branching ratios may reach to orders of @xmath5 .
these hidden bottomonium decay modes are of great importance in the experimental search for the @xmath1 particularly at the hadron collider .
also , the associated studies of hidden bottomonium decays @xmath6 , @xmath7 , and @xmath8 may help us investigate the structure of @xmath1 deeply .
the experimental observation of @xmath1 will provide us with further insight into the spectroscopy of exotic states and is helpful to probe the structure of the states connected by the heavy quark symmetry . |
globular clusters are touchstones of astrophysics.galaxies .
the oldest of them witnessed the formation and early evolution of their host galaxies and of their substructures , and their study has historically coloured the different scenarios of galaxy formation .
however , as a collective population in a galaxy , they present unsolved problems .
in particular their origin is not firmly established despite the large amount of work devoted to the analysis of correlations among their observable properties .
it has long been realised that part of the difficulty arises from sheer dynamical evolution undergone by these objects since the time of their formation . indeed , any star cluster is the subject of a long list of erosive mechanisms that operate at different rates depending on the cluster s location and orbit within the galaxy , and on its initial mass ( @xcite , @xcite ) . related to these difficulties , the search for a
second parameter " , beyond metallicity , to explain the distribution of stars along the horizontal branch , has met with a limited success .
@xcite suggested that the halo globular clusters break down into two groups according to their horizontal - branch properties : the two groups have different ages , kinematics and radial distributions .
@xcite found that an age difference can explain different horizontal - branch morphologies at a given metallicity . but
@xcite , following the seminal work by @xcite on the impact of stellar mass loss on the horizontal - branch morphology , has shown that considering @xmath0-element enhancement and metallicity - dependent mass loss along the red - giant phase produces similar effects .
on the other hand @xcite have found that globular clusters with extended horizontal branch are more massive than normal clusters and are dominated by random motions with no correlation between kinematics and metallicity .
multivariate analyses have been performed by @xcite who find that more concentrated clusters have bluer and longer horizontal - branch , and by @xcite who find that more massive clusters have a horizontal branch that extends to higher temperatures .
we can point to several reasons why previous studies have not been completely successful .
for example , empirically separating the clusters into disc and halo populations solely on the basis of one parameter ( metallicity ) can not be satisfactory , as the environment at birth must influence other cluster parameters .
this was already recognized by @xcite , who combined metallicity and kinematics to establish two major groups ( halo versus bulge / disk ) .
the properties of the horizontal branch have been used , but its usual characterisation by the parameter hbr = ( b - r)/(b+v+r ) is also unsatisfactory . in this respect ,
the parameter @xmath1 ( maximum effective temperature on the horizontal branch ) introduced by @xcite is a welcome innnovation as shown in the present paper .
the classical paradigm describing globular clusters as fairly simple systems of coeval stars of homogeneous chemical composition has been seriously challenged recently , and this may bring a crucial piece to the puzzle .
one fundamental characteristic of these systems is their metallicity [ fe / h ] ( identified as the so - called first - parameter " ( @xcite ) ) that is generally inferred from their integrated photometric colours , and that varies strongly from cluster to cluster ; in our galaxy , globular clusters have [ fe / h ] ranging between @xmath2 -2.2 and 0 ( @xcite ) .
spectroscopy reveals that , within individual clusters , stars present very homogeneous contents in fe , but also in @xmath3 and s - elements , indicating that protoclusters formed from gas pre - enriched in heavy metals ( @xcite ) .
this is in agreement with the predictions of quantitative models that rule out stochastic self - enrichment in most globular clusters as a significant contributor to their heavy metals , leaving pre - enrichment as the dominant contributor to metallicity [ fe / h ] ( @xcite ) .
however globular cluster stars exhibit extremely scattered light - element ( li , f , c , n , o , na , mg , and al ) contents that are not seen among their field counterparts ( see e.g. @xcite ) .
this points to early internal chemical evolution ( i.e. , self - enrichment ) in the globular cluster driven by first - generation massive and fast evolving stars which polluted with their hydrogen - burning products ( among which helium in very important quantities ) the intracluster gas out of which second generation stars formed ( @xcite , @xcite ) .
recent findings of double or even multiple stellar populations in the colour - magnitude diagrams of several globular clusters , as well as the complexity of the horizontal branch morphology ( namely the wide colour distribution , i.e. , effective temperature , of the stars presently burning helium in their core ) constitute further evidence for internal evolution ( @xcite ) .
all these features can indeed be related to the presence of a second generation of he - enriched stars .
importantly , the star formation history depicted by these features seems to vary from cluster to cluster ( @xcite ) in a way which is still far from being understood .
however we have now firm evidence that galactic globular clusters have undergone internal chemical evolution and complex star formation histories during their infancy that shaped their properties and in particular their present total mass ( @xcite ) .
this new paradigm has opened a novel route for a better understanding of the origin and history of globular clusters .
it thus appeared to us that a multivariate analysis which simultaneously takes into account any cosmic variance due to evolving physical conditions and groups objects according to environment of formation would be very valuable .
cladistics provides such a methodology .
it differs from other clustering analyses in that it focuses on evolution within and between groups rather than on similarities between objects ( @xcite ) .
cladistics is very commonly used in evolutionary biology and has been pioneered in astrophysics by @xcite and successfully applied to the dwarf galaxies of the local group ( @xcite ) .
this paper presents a multivariate analysis based on the method of cladistics of a large sample of galactic globular clusters . after presenting the data and the method of analysis ( sect .
[ data ] ) , we describe the three groups found by the cladistic analysis ( sect .
[ 3groups ] ) , and discuss two important results , evidence for self - enrichment ( sect .
[ enrichment ] ) and a possible luminosity - metallicity relation ( sect . [ mv - fe ] ) .
we then compare the properties of the three groups with those of galactic halo field stars ( sect . [ comp ] ) , before proposing a scenario for the formation of the three groups ( sect .
[ origin ] ) .
the choice of parameters is a crucial step in any multivariate analysis . @xcite
have shown that the manifold of galactic globular cluster properties has a dimension larger than 4 , but that a subset of parameters linked to morphology and dynamics forms a three - dimensional family . including properties of the stellar populations ( e.g. a horizontal - branch parameter or metallicity ) will increase the manifold by 1 or 2 dimensions ( @xcite ) . finally , using a large number of photometric and structural parameters
, @xcite found that 4 eigenvectors account for 79% of the total sample variance .
taking advantage of this indication , we selected the following four parameters for analysis : relative ages , metallicity ( [ fe / h ] ) , absolute v magnitude ( @xmath4 ) , and maximum effective temperature ( @xmath1 ) on the horizontal branch .
the age parameter is related to the secular evolution of the stellar populations .
[ fe / h ] reflects the chemical composition of the environment when and where globular clusters formed and is the first parameter " for the horizontal branch morphology @xmath4 is a structural parameter that measures the present total baryonic mass of the globular clusters of globular clusters in the milky way spans a vast range ( @xmath5 , @xcite ) , and reflects a large mass range ( @xmath6 , @xcite ) . ] .
finally @xmath1 is a measure of both the pristine chemical composition of the protocluster ( [ fe / h ] being the first parameter ) and of the helium enrichment during early internal chemical evolution , since stars with higher helium content are expected to reach higher effective temperatures on the horizontal branch ( @xcite ) .
the last three quantities describe truly intrinsic properties of globular clusters .
as the age parameter evolves in all clusters , it can not be used to classify them at the same level as the other three parameters .
we thus gave it a lower weight in the cladistic analysis ( see appendix a ) .
we performed our analysis using the large sample ( 54 objects ) of @xcite based on homogeneous hubble space telescope photometry .
we used the @xmath1 values obtained uniformly from this database by @xcite , as well as the relative ages and @xmath4 values they adopted ( i.e. , taken respectively from @xcite and the 2003 on - line revision of @xcite ) .
we did not include more parameters in the cladistic analysis as we preferred to avoid the unwanted effect of redundancies , which give more weight to correlated parameters .
note also that we did not use any kinematical information in the cladistic analysis .
however we used other parameters _ a posteriori _ to characterize the different groups found by the cladistic analysis : the radial velocities and structural parameters were taken from the 2003 on - line revision of @xcite , the orbital parameters from @xcite . in this respect , the distinct orbital properties of group 3 found a posteriori ( see 3 ) are independent of the methodology .
more recent age estimation was published by @xcite after most of this project was completed , but for only 35 out of 54 globular clusters of our sample . using these produces an inhomogenous data set , from different sources , relying on different values of fe / h .
the ages of @xcite have been determined from colour - magnitude diagrams and values of fe / h which are different from those of @xcite .
@xcite themselves point out the importance of using a homogeneous set of fe / h to derive the ages .
nevertheless , it is instructive to perform analyses using both sets and compare the results , so as to determine how sensitive they are to the specific choice of parameter values .
one then has the problem of combining two sets of ages , and it is not obvious how this should be done .
it turns out that 9 out of 11 globular clusters that calibrate relative ages of @xcite are in common with @xcite , and using the relative ages of @xcite for these 9 globular clusters gives a mean value of 1.0055 instead of 1.00 , which is fine .
however , comparing the ages of all the globular clusters in common suggests a non - linear systematic effect , which should perhaps be taken into account .
furthermore , we can not simply convert each set of relative ages to absolute ages with the zero - point of each set , because the two zero - points are rather different ( 11.2gyr for @xcite and 12.8gyr for @xcite ) .
we thus simply used the relative ages without any attempt at homogenizing them , and the zero - point of one author for all ages .
this additional analysis is compared to the main one in sect .
[ addanal ] .
the multivariate analysis was performed using the method of cladistics . in short ,
the method works as follows .
one first builds a matrix with values of the four parameters for each clusters .
the values must be discretized , and the number of bins ( here 10 ) depends on the resolution one wants for the analysis .
one then chooses a cluster which represents the most unevolved state , in the present case the metal - poorest cluster ( ngc 6934 ) , and the software classifies all the other clusters in order of increasing diversification of properties ( in other words , by increasing distance in the manifold of parameters ) .
clusters which are diverging from the original cluster in the same direction are put on the same branch .
we refer to appendix a and @xcite for more details on the principles of the method .
the main result of our analysis is presented in the form of a tree structure , a usual form of representation in graph theory .
the properties of the sample can be read from the structure of the tree shown on fig .
[ treefig ] .
the tree has been rooted with group 2 which has the lowest metallicity on average , and as such is supposedly made of more primitive ( or `` ancestral '' ) material .
the tree divides into three main branches , which define three groups with quite distinct properties .
age increases roughly monotonically along each branch , as expected .
there are some subbranches sharing similar values of the adopted parameters , such as ngc 5904 , 2808 , 6388 and 6441 , and which set them apart in the four - dimensional space of parameters , implying that they might in fact belong to a fourth small group with properties similar to those of group 1 .
specialists will immediately remark that the latter three clusters indeed share peculiar properties , in particular a very helium - rich stellar population ( @xcite ) .
we emphasize that the helium abundance is not one of the parameters included in the cladistic analysis : it thus must influence one way or another the four parameters used in the analysis .
.properties of the three groups of globular clusters .
no age estimate is available for ngc 6139 , 6229 , 6304 , 6388 , 6441 , 6569 and 6642 .
the middle part of the table gives the correlations : + means a correlation , - means an anticorrelation , x no correlation .
the orbital properties presented in the lower part of the table are only available for a subset of each group .
numbers in brackets are rms dispersions [ cols= " < ,
< , < , < " , ]
it is generally assumed that the galaxy assembled in a hierarchical fashion from collapsing haloes of dark matter ( @xcite . small proto - galactic clumps formed first , from initial small - scale density fluctuations , and collapsed in a dissipationless way ( because their gas was quickly consumed or blown away ) , or else they merged and grew in size to form larger clumps which spiralled toward the inner regions of the galaxy by dynamical friction and experienced a dissipational collapse .
this scenario , or variants of it , first proposed by @xcite and verified by cosmological simulations(@xcite ) , has repeatedly been invoked to explain the properties of the galactic globular clusters and of the stellar halo .
furthermore , cosmological simulations have shown that globular clusters can form in the densest regions of collapsing subhaloes of dark matter and giant molecular clouds , when the clouds reach a critical density and are under a high external pressure .
the mass distribution function of the clusters is similar to that of the clouds ( @xcite ) .
it has often been proposed that some clusters , in particular those identified as young halo clusters by @xcite , which tend to be in the outer halo and counter - rotating , were formed by accretion and disruption of satellite galaxies .
but the chemical homogeneity of the halo , as well as substantial differences in chemical composition between field stars in the halo and dwarf spheroidal galaxies , argue against the accretion scenario ( @xcite ) .
the properties of our three groups of clusters can be interpreted in the following way , without resorting to an external origin for any of the groups : the clusters of the outer halo ( group 2 ) formed during the initial dissipationless collapse of the proto - galaxy , from material already polluted by earlier generations of stars , but not well homogenized and thus inhomogeneous in @xmath0-elements .
contrary to the outer halo stars , they lost their initial average retrograde rotation by dynamical friction and gravitational encounters .
as suggested by the referee , this group could also have originated in some `` pregalactic dwarfs '' ( i.e. metal - poor , gas - rich satellites that soon afterward began hierarchical merging ) . the clusters of the inner halo ( group 1 ) formed later , during the dissipational phase of galactic collapse , which continued in the halo after the formation of the thick disc and its globular clusters . since the formation of group 1 occurred later , the molecular clouds from which they formed had time to grow by accretion of smaller clumps .
these clouds were already enriched at the same level in @xmath0-elements .
thanks to the strong potential well in the clusters ( as evidenced by their high central velocity dispersions ) , the he - rich ejecta of first generation massive stars were not blown away and found their way into a more strongly helium - enriched second generation of stars , favouring the production of hot horizontal - branch and cepheid stars .
as indicated by their short range in age ( 1.5 gyr , see fig .
[ age_fe2fig ] ) , the disc clusters ( group 3 ) formed in a more rapid fashion than the two other groups , before many clusters of group 1 .
this could presumably be due to the higher densities and external pressure in the thick disc .
this group shows significant average prograde rotation , because the dissipational collapse of the disc conserved angular momentum .
the metal - poor disc clusters seem to rotate more slowly and have larger eccentricities and inclinations than the metal - rich disc clusters .
since there is no significant age difference between the two subgroups , we assume that the metal - poorer clusters formed further away from the galactic plane , and thus retained a larger vertical velocity component .
paired t - tests were performed with the vostat software ( http://vo.iucaa.ernet.in/ voi / vostat.html ) .
we thank asis chattopadhyay for advice .
we would like to thank the referee , w. harris , for his constructive comments and suggestions .
bailin j. , harris w.e .
, 2009 , apj 695 , 1082 bekki k. , chiba m. , 2001 , apj 558 , 666 bertschinger e. , 1998 , araa 36 , 599 brodie j.p .
, strader j. , 2006 , araa 44 , 193 ( astro - ph/0602601 ) buchanan b. , collard m. , 2008 , journal of archaeological science 35 , 1683 caloi v. , dantona f. , 2008 , apj 673 , 847 carretta e. , 2006 , aj 131 , 1766 ( astro - ph/0511144 ) carretta e. , recio - blanco r. , gratton r.g . et al . , 2007 ,
apj 671 , l125 carollo d. , beers t.c . ,
lee y.s . ,
et al . , 2007 , nature 450 , 1020 ( http://arxiv.org/abs/0706.3005 ) casetti - dinescu d.i . ,
girard t.m . , herrera d. , van altena w.f . ,
lopez c.e . ,
castillo d.j . , 2007 , aj 134 , 195 ( http://arxiv.org/abs/0705.3438 ) clement c.m . , muzzin a. , dufton q. , et al . , 2001 , aj 122 , 2587 dantona , f. , caloi v. , 2008 , mnras 390 , 693 ( astro - ph/0209331 ) dantona , f. , caloi , v. , montalbn , j. , ventura , p. , gratton , r. , 2002 , a&a 395 , 69 ( astro - ph/0209331 ) de angeli f. , piotto g. , cassisi s. , busso g. , recio - blanco a. , salaris m. , aparicio a. , rosenberg , a. , 2005 , aj 130 , 116 decressin , t. , charbonnel , c. , meynet , g. , 2007 , a&a , 475 , 859 ( http://arxiv.org/abs/0709.4160 ) dekel a. , silk j. , 1986 , apj 303 , 39 dinescu d.i . , girard t.m . , van altena w.f
lopez c.e . , 2003 , aj 125 , 1373 ( astro - ph/0212279 ) dinescu d.i . ,
girard t.m . , van altena w.f . ,
1999 , aj 117 , 1792 djorgovski s. , meylan g. , 1994 , aj 108 , 1292 dotter a. , 2008 , apj 687 , l21 fraix - burnet , d. , in _ evolutionary biology .
concept , modelization and application_. pontarotti , p. ( ed . ) , springer , in press fraix - burnet d. , choler p. , douzery e.j.p .
, verhamme a. , 2006a , journal of classification , 23 , 31 - 56 ( astro - ph/0602581 ) fraix - burnet d. , douzery e.j.p . ,
choler p. , verhamme a. , 2006b , journal of classification , 23 , 57 - 78 ( astro - ph/0602580 ) fraix - burnet d. , choler p. , douzery e.j.p .
, 2006c , a&a , 455 , 845 - 851 ( astro - ph/0605221 ) frenk c.s . ,
white s.d.m . , 1980 ,
mnras 193 , 295 fusi pecci f. , ferraro f.r . , bellazini m. , et al . , 1993 aj 105 , 1145 fusi pecci f. , bellazini m. , buzzoni a. , et al .
, 2005 , aj 130 , 554 geisler d. , wallerstein g. , smith v.v .
, casetti - dinescu d.i . , 2007 ,
pasp 119 , 939 gnedin o.y . ,
ostriker j.p . , 1997 , apj 474 , 223 ( astro - ph/9603042 ) goloboff p.a . , mattoni c.i . ,
quinteros a.s . , 2006 , cladistics 22 , 589 gratton r.g . , carretta e. , desidera s. , et al . , 2003 , a&a 406 , 131 ( astro - ph/0305358 ) harris , w. , 1996 , aj 112 , 1487 ( http://physwww.physics.mcmaster.ca/@xmath2harris/mwgc.dat , 2003 on - line revision ) harris w.e
. , whitmore b.c . , karakla d. , okon w. , baum w.a . , hanes d.a . , kavelaars j.j
. , 2006 , apj 636 , 90 harris w. , 2001 , saas - fee advanced course 28 .
lecture notes 1998 swiss society for astrophysics and astronomy .
carney b.w .
, harris w.e .
( eds ) , springer harris , w. , 2009 , apj in press ( arxiv:0904:4208 ) james g. , franois p. , bonifacio p. , carretta e. , gratton r.g . , spite f. , 2004 , .
427 , 825 ( astro - ph/0408330 ) .
kravtsov a.v , gnedin o.y . , 2005 , apj 623 , 650 ( astro - ph/0305199 ) lee y - w .
, gim h.b .
, casetti - dinescu d.i . , 2007 ,
apjl 661 , l49 ( http://arxiv.org/abs/0704.0486 ) maddison w.p . ,
maddison d.r . , 2004 , mesquite : a modular system for evolutionary analysis , version 1.05 ( http://mesquiteproject.org ) marn - franch a. , aparicio a. , piotto g. , et al .
2009 , apj 694 , 1498 mclaughlin d.e .
, van der marel r.p .
, 2005 , apjs 161 , 304 mieske s. , jordn a. , ct p. , et al . 2006 , apj 653 , 193 milone a. p. , bedin , l. r. , piotto , g. , et al . , 2008 , a pj 673 , 241 ( http://arxiv.org/abs/0709.3762 ) morrison h.l . , harding p , perrett k , hurley - keller d. , 2004 , apj 603 , 87 piotto g. , 2009 , iau symp .
258 ( in press ) , astro - ph/0902.1422 prantzos n. , charbonnel c. , 2006 , a&a 458 , 135 pritzl b.j .
, venn k.a . , irwin m. , 2005 , aj 130 , 2140 ( astro - ph/0506238 ) pumo m.l . , dantona f. , ventura p. , 2008
, apj 672 , 25 ( http://arxiv.org/abs/0803.0235 ) recio - blanco a. , aparicio a. , piotto g. , de angeli f. , djorgovski s.g . , 2006 , a&a 452 , 875 ( astro - ph/0511704 ) rey s .- c . ,
yoon s .- j . ,
lee y .- w .
, 2001 , aj 122 , 3219 rood r.t . , 1973 , apj 184 , 815 searle l. , zinn r. , 1978 , apj 225 , 357 stephens a. , boesgaard a.m. , 2002 , aj 123 , 1647 swofford d. l. , 2003 , paup * : phylogenetic analysis using parsimony ( * and other methods ) , sinauer associates , sunderland , massachusetts ( http://paup.csit.fsu.edu/ ) thuillard m. , fraix - burnet d. , 2009 , evolutionary bioinformatics 5 , 33 ( http://fr.arxiv.org/abs/0905.2481 ) van den bergh s. , 1967 , aj 72 , 70 venn k.a . , irwin m. , shetrone m.d . , et al .
aj 128 , 1177 ( astro - ph/0406120 ) wallerstein , g. , 2002 , pasp 114 , 689 wiley e.o . , siegel - causey d. , brooks d.r .
, funk v.a . _
the compleat cladist : a primer of phylogenetic procedures_. the university of kansas , museum of natural history special publication n@xmath7 19 , 2007 zinn r. , 1985 , apj 293 , 424 zinn r. , 1993 , _ the globular cluster - galaxy connection _ , asp conf .
multivariate clustering methods compare objects with a given measure and then gather them according to a proximity criterion .
distance analyses are based on the overall similarity derived from the values of the parameters describing the objects .
the choice of the most adequate distance measure for the data under study is not unique and remains difficult to justify a priori .
the way objects are subsequently grouped together ( this is called the linkage ) is also not uniquely defined .
cladistics uses a specific measure that is based on characters ( a trait , a descriptor , an observable , or a property , that can be given at least two states characterizing the evolutionary stages of the object for that character ) and compares objects in their evolutionary relationships . here , the `` distance '' is an evolutionary cost .
groupings are then made on the basis of shared or inherited characteristics , and are most conveniently represented on an evolutionary tree .
character - based methods like cladistics are better suited to the study of complex objects in evolution , even though the relative evolutionary costs of the different characters is not easy to assess .
distance - based methods are generally faster and often produce comparable results , but the overall similarity is not always adequate to compare evolving objects . in any case , one has to choose a multivariate method , and the results are generally somewhat different depending on this choice ( e.g. @xcite ) .
however , the main goal is to reveal a hidden structure in the data sample , and the relevance of the method is mainly provided by the interpretation and usefulness of the result . in the present paper , the use of cladistics is justified a priori by the evolutionary nature of globular clusters , and a posteriori by the strong astrophysical significance of the grouping found . in astrophysics , cladistics
has already been applied to galaxies because they can be shown to follow a transmission with modification process when there are transformed through assembling , internal evolution , interaction , merger or stripping ( @xcite ) . for each transformation event ,
stars , gas and dust are transmitted to the new object with some modification of their properties . for globular clusters , interactions and mergers
are probably rare .
it was previously thought that once they assembled , only the stellar ageing would affect their properties .
nowadays , we have firm evidence that internal evolution can create another generation of stars , and clusters can lose mass .
basically , the properties of a globular cluster strongly depend on the environment in which it formed ( chemical composition and dynamics ) , and also on the internal evolution which includes at least the ageing of its stellar populations . to compare globular clusters ,
it is thus necessary to take into account the different stages of evolution of both the objects and their environments of formation . since the clusters form in a very evolving environment ( evolution of the universe and the dynamical environment of the parent galaxy )
, the basic properties of different clusters are related to each other by some evolutionary pattern . in particular , the dust and gas from which the stars of the globular clusters form have been `` polluted '' ( enriched in heavy elements ) by more ancient stars , being field stars or belonging to other globular clusters .
this results in a kind of transmission with modification process , which justifies a priori the use of cladistics .
it must be clear that this is not a `` descent with modification '' in the sense that there is no replication .
but evolution does nevertheless create diversity .
we are dealing with phylogeny ( relationships between species ) , not with genealogy ( relationships between individuals ) . since a multivariate classification of globular clusters
is not yet available , we assume in the present work that each cluster represents a `` species '' that will have to be defined later on . as our work on galaxies has shown us , it is important to remove parameters that are redundant . since
previous studies of the manifold of galactic globular clusters have shown that 4 parameters are sufficient to describe their diversity , we selected 4 parameters , 3 of which are intrinsic characteristics of the environment of formation .
the fourth one , age , is particular in the sense that it does not inform on the conditions when the clusters formed , and is not discriminant for clustering because it evolves similarly for any cluster ( parallel evolution ) .
however , age is useful to rank the clusters within each group . consequently , we applied to age a weight half that of the other three parameters .
in addition , a stepmatrix was employed to impose the irreversibility constraint on the age parameter ( age can only increase ) .
in contrast to multivariate distance methods , undocumented values are not a problem in cladistics analyses . this is why the seven galaxies that have no age determination ( see fig.[age_fe2fig ] and table [ tableprop ] ) have not been excluded in our work . in this paper , we use parsimony as the optimisation criterion .
this works as follows .
one first builds a matrix with values of the four parameters for all clusters .
the values for each parameter are discretized into 10 bins representing supposedly evolutionary states .
discretization of continuous variables is quite a complex problem , especially in the evolutionary context ( see e.g. @xcite ) .
the choice of the number of bins can not be made in a simple objective way . here , we took equal - width bins , and considered a compromise between an adequate sampling of continuous variables and the uncertainties on the measurements . the first constraint is given by the software ( 32 in this case )
the second one would a priori give a lower limit of something like total range / uncertainty , but shannon s theorem would multiply this by 2 . hence , 10 bins would account for about 20% measurement errors , which is rather large , but @xcite do not provide precise error estimates , especially for logte .
even so , border effects always imply that some objects could belong to a bin or its neighbour , a process that add some more artificial noise .
the best way to avoid this effect is to make several analyses with different number of bins and check that the result does not depend on this number .
we have done this for 3 , 5 , 8 , 10 , 12 , 15 and 20 , and the result is identical , to at most one misplaced cluster , for numbers higher than 8 . for 3 ,
no structure is found , and for 5 bins the groups are not well defined .
then , all possible arrangements of clusters on a tree are constructed , and using the discretized matrix , the total number of state changes is computed for each tree .
the most parsimonious tree is finally selected .
if several such trees are found , then a consensus ( strict or majority rule ) tree is built .
the whole procedure is computerised since the number of arrangements is here very large .
the result is a diversification scenario that should be confronted to other knowledge and parameters .
maximum parsimony heuristic searches were performed using the paup*4.0b10 ( @xcite ) package .
the results were interpreted with the help of the mesquite software ( @xcite ) . the tree presented in fig .
1 is a majority rule consensus tree of 20000 trees , the strict consensus tree showing exactly the same three groups but with group 1 and 2 slightly less resolved . to further assess the robustness of the tree , it was not possible to make bootstrapping due to the irreversibility constraint on the age parameter , and it would not have been very significant with only 4 parameters .
we performed other analyses using only 3 parameters , excluding the age .
they all gave essentially the same three groups , but they were individually slightly less resolved , as expected .
all these convergences yield strong confidence on the tree shown in fig .
1 . in the end ,
the most important point is the astrophysical interpretation we are able to give of the results . on fig .
1 , the tree is rooted with group 2 as an outgroup .
this is not strictly necessary in cladistics , and we find here the same three groups whatever the root chosen or even on the unrooted tree .
but we know that a low metallicity is an ancestral state for stars in general , this is why we have chosen group 2 a posteriori because it has a homogeneously low metallicity . | we perform an evolutionary multivariate analysis of a sample of 54 galactic globular clusters with high - quality colour - magnitude diagrams and well - determined ages .
the four parameters adopted for the analysis are : metallicity , age , maximum temperature on the horizontal branch , and absolute v magnitude .
our cladistic analysis breaks the sample into three novel groups .
an _ a posteriori _ kinematical analysis puts groups 1 and 2 in the halo , and group 3 in the thick disc .
the halo and disc clusters separately follow a luminosity - metallicity relation of much weaker slope than galaxies .
this property is used to propose a new criterion for distinguishing halo and disc clusters .
a comparison of the distinct properties of the two halo groups with those of galactic halo field stars indicates that the clusters of group 1 originated in the inner halo , while those of group 2 formed in the outer halo of the galaxy .
the inner halo clusters were presumably initially the most massive one , which allowed the formation of more strongly helium - enriched second generation stars , thus explaining the presence of cepheids and of very hot horizontal - branch stars exclusively in this group .
we thus conclude that the second parameter " is linked to the environment in which globular clusters form , the inner halo favouring the formation of the most massive clusters which subsequently become more strongly self - enriched than their counterparts of the galactic outer halo and disc . [ firstpage ]
globular clusters : general galaxy : formation galaxy : evolution galaxy : halo method : statistical |
despite much study of the formation and evolution of young t tauri stars , many questions remain concerning some of their most fundamental characteristics .
understanding the few million years of evolution between the protostellar and post t tauri stages , including circumstellar disk accretion , planet formation , and disk dissipation , is crucial to understanding the formation of the solar system and extra - solar planetary systems .
one key approach to resolving these issues is striking a balance between in - depth studies of individual star - forming regions and observations of a variety of star - forming regions that differ in age , distance , stellar density , binary fraction , etc .
the presence of t tauri stars in a region indicates an age of 1@xmath410 million years , the timescale for disk dissipation and the formation of gas giant planets around low mass stars ( e.g. , * ? ? ?
t tauri stars are primarily identified by h@xmath7 emission , ir and uv continuum excess , variability , and surrounding nebulosity ( e.g. , * ? ? ?
* ; * ? ? ?
after 50 years of study , the young , low mass multiple star system as 353 , in the constellation aquila , remains enigmatic .
a @xmath05 pair , the system appears to be a mixed t tauri binary , with the more massive component , as 353a , showing signatures of a classical t tauri star ( ctts ) , which is defined as having ew(h@xmath7 ) @xmath8 10 and excess infrared emission above the continuum level @xcite .
the secondary , as 353b , was classified as a weak - lined t tauri star ( wtts , ew(h@xmath7 ) @xmath9 10 ) , but has recently been resolved into a subarcsecond binary @xcite . as 353ba was reclassified by @xcite as a ctts ( but see 4.5.1 for further discussion ) . as 353 is the exciting source of herbig - haro object hh 32 , a highly collimated bipolar outflow ( see * ? ? ?
* and references therein ) and one of the only herbig - haro objects detected in 3.6 and 6 cm radio continuum @xcite . radio continuum emission has also been detected from the stars themselves , most recently attributed to the secondary rather than the primary @xcite .
the age of the as 353 system was most recently determined by @xcite based on the location of as 353b relative to the pre@xmath4main - sequence evolutionary tracks of @xcite .
this analysis resulted in an age of 5@xmath1010@xmath11 years / l@xmath12 ) for as 353b .
data in the tables and text are accurate . ] , or 1@xmath1010@xmath13 years if the binary nature of as 353b is taken into account ( see appendix b of @xcite ) .
in this current paper we provide an updated discussion of the age of the as 353 system as well as best estimates for the other objects in our sample ( 4.2 ) .
in addition to as 353b , there are five t tauri stars listed in the herbig and bell catalog ( 1988 , hereafter hbc ) that are within @xmath015 of as 353a , all between galactic latitudes of @xmath14 = @xmath41 to @xmath46 .
their projected location in the crowded galactic plane complicates the study of these stars , as well as the search for additional young stars in the region .
few of the hbc aquila objects have radial velocities or @xmath15 listed in the catalog , but from spectra with resolution lower than the observations presented here and/or using emission lines instead of absorption lines , with results that have uncertainties on the order of tens of km s@xmath5 or more @xcite .
the original motivation for this project was to search for radial velocity variations in a sample of ctts and wtts in the vicinity of as 353a as a means of identifying spectroscopic binaries . for double - lined spectroscopic binary systems , the relative velocity of each component ,
the mass ratio , and the center of velocity of the system can be determined .
obtaining dynamical masses of young stars is critical to the study of star formation because mass is the most fundamental of all stellar parameters , and evolutionary models that relate young star masses and ages with the observables l and t@xmath16 are not yet calibrated .
furthermore , absolute radial velocities are important for determining if the young stars in the sample are actually related , i.e. if they were formed in the same molecular cloud .
if they were , then it is expected that they have a low velocity dispersion relative to one another as well as with respect to the associated cloud ( e.g. , * ? ? ?
high resolution infrared spectra , even a single echelle order , can reveal much more than radial velocities . in this study ,
what were assumed to be typical classical or weak - lined t tauri stars were found to be surprisingly active systems with high rotational velocities , strong veiling , and unusual emission lines .
thus the goals of this project broadened to include the measurement of @xmath15 , determination of spectral type , and characterization of the circumstellar environments of the targets , in addition to the radial velocity measurements .
the sample is introduced in 2.1 , and the nirspec observations and their reduction are described in 2.2 and 2.3 , respectively .
the data analysis , including line identification , determination of spectral type and @xmath15 , and measurements of radial velocity variability and absolute radial velocity , are described in3 .
the results are discussed in 4 , including details of circumstellar environment , age , membership , and multiplicity .
a summary appears in 5 along with suggestions for further studies of the aquila star - forming region .
the targets were selected from the third catalog of emission - line stars of the orion population ( hbc ) because of their spatial proximity to the well - studied t tauri star as 353a ( hbc 292 ) and its companion , as 353b ( hbc 685 ) . the names , coordinates , @xmath1-band magnitudes , and equivalent width of h@xmath7 emission for each object are listed in table [ table : coords ] .
coordinates are from simbad unless otherwise noted .
the objects all lie within a region from 19h 00 m to 19h 40 m right ascension and @xmath45 to @xmath1711 declination ( equinox 2000.0 ) , placing them in the constellation aquila .
we adopt a distance of 200 pc @xmath6 30 pc ( 4.2 ) , which is consistent with both the most recent distance estimate to as 353 ( 150@xmath650 pc , @xcite ) and to the dark cloud complex associated with hbc 294 ( 230@xmath630 , @xcite ) .
the physical size of this region on the plane of the sky is estimated to be 32@xmath65 pc by 58@xmath69 pc at this distance .
targets with previously determined spectral types range from early k to early m @xcite ; several have been revised as a result of the work described in this paper . as originally selected the sample consisted of six objects : hbc 292 , 294 , 681 , 682 , 683 , and 684 . during the course of the observations and subsequent analysis ,
the sample has grown to nine objects .
hbc 682 was discovered to be a 1 visual binary upon the first observation with nirspec . as 353b was added to the sample in order to strengthen its association , using radial velocity measurements , with as 353a .
because of its close proximity to hbc 681 and 682 , a ninth star , fg aql / g3 @xcite , was also observed .
data were obtained with the cryogenic , cross - dispersed echelle spectrometer nirspec installed on the nasmyth platform of the keck ii telescope at the w.m .
keck observatory on mauna kea , hawaii @xcite .
nirspec employs a 1024 @xmath10 1024 aladdin insb array detector operating at 30 k and sensitive to the range 0.95 to 5.4 @xmath18 .
all observations were taken in high - resolution mode with a slit size of 0.288@xmath1024 for non - ao observations and [email protected] for ao observations . the echelle mode and
two pixel slit width provided a resolving power ( r=@xmath19/@xmath20 ) of r@xmath030,000 for non - ao observations and r@xmath035,000 for observations behind ao , corresponding to a velocity resolution of @xmath010 km s@xmath5 per two - pixel resolution element in non - ao mode and @xmath08.6 km s@xmath5 in ao mode .
integration times ranged from two to five minutes per exposure for the aquila objects . for each observation , a series of four spectra was taken in an abba pattern , nodding @xmath010 along the slit between the a and b positions .
typical seeing was 06 .
multiple abba sets were taken for fainter objects .
integration times ranged from 120 s to 300 s ( table [ table : obslog ] ) .
@xmath1-band observations using the n5 filter ( 1.413 - 1.808 @xmath18 ) were taken at several epochs between 2002 july and 2004 november to search for radial velocity variations .
the intention was to observe each object a minimum of three times , however , it was possible to observe hbc 683 only twice and as 353b and fg aql / g3 only once . on 2004
may 27 and 28 ( ut ) the observations were made using nirspec behind the keck adaptive optics system @xcite . a @xmath2-band spectrum was obtained for each object on either 2003 september 07 ( hbc numbers 294 , 681 , 682a&b , 683 , and 684 ) or 2004
november 22 ( as 353a ) . as 353b and fg
aql / g3 were not observed in @xmath2-band .
slightly different echelle angles resulted in different spectral coverage for the two sets of @xmath2-band observations .
@xmath2-band spectra obtained on 2003 september 07 ( not shown ) suffered from poor subtraction of telluric absorption lines .
table [ table : obslog ] provides an observing log ; instrument settings for all observations are summarized in table [ table : inst ] . for data reduction we used the redspec code , a package of idl procedures created specifically for the reduction of nirspec spectra by s. kim , l. prato , and i. mclean .
median - filtered dark and flat fields were created from multiple flat lamp and dark exposures for non - ao @xmath1-band observations . for ao observations and @xmath2-band observations ,
only a single flat and dark exposure were taken and differenced to create a final flat field .
we determined the dispersion solution with a target star frame using oh night sky emission lines @xcite or an arc lamp frame .
the latter approach was necessary for the ao data because the night sky oh emission lines are too faint in the target frames .
based on the error analysis reported in @xcite , we assign an uncertainty of 1 km s@xmath5 to the dispersion of each target .
we used @xmath1-band order 49 ( 1.545 - 1.567 @xmath18 ) for the radial velocity measurements because of the presence of many narrow atomic and molecular absorption lines and the lack of telluric absorption lines .
therefore , no division by a telluric a0 star was needed for order 49 spectra . in our @xmath2-band analysis
we focused on the br@xmath21 line in order 35 ( 2.159 - 2.190 @xmath18 ) ; orders 33 , 36 and 37 were also reduced .
reduction of order 35 was complicated by the presence of the broad br@xmath21 absorption in the telluric a0 star spectrum . to remove this intrinsic spectral feature
while retaining the telluric features , we reduced the order 35 a0 star spectrum leaving the broad br@xmath21 and narrow terrestrial features intact . in a second reduction , we interpolated over the narrow telluric absorption lines only , leaving the br@xmath21 feature intact .
the a0 star spectra resulting from these two reductions were then subtracted and the difference spectrum was renormalized to a continuum of 1.0 , creating a spectrum of telluric features only .
target spectra were then divided by this corrected a0 star spectrum to remove the telluric lines .
the spectra were transformed onto a heliocentric reference frame and the continua were flattened and normalized to a value of 1.0 . the signal
to noise ratio ( snr ) of the spectra was estimated by determining the maximum peak - to - peak variation in an order and dividing this by a factor of 4 to estimate the noise .
the snr appears in the last column of table [ table : obslog ] .
figures [ fig:08sep03 ] through [ fig : hbc684 ] present the high - resolution spectra that will be discussed further in 4 . absorption and emission lines were identified using atlases of the solar spectrum and sunspot umbral spectrum in the near - infrared @xcite .
@xmath2-band line identifications are from table 4 of @xcite .
individual objects are discussed in detail in 4.5 .
almost 40 absorption lines ( some blended ) were identified in the order 49 spectra of the object with the lowest rotational velocity , hbc 683 , and therefore the sharpest lines .
fewer individual lines are identified in spectra with higher rotational velocity because nearby lines become increasingly blended as @xmath15 increases .
the vast majority of the lines are atomic : mainly neutral fe , some neutral ti , ni , and si , and possibly s and c are identified .
intrinsic oh absorption appears in most spectra and a possible detection of cn emission in hbc 684 .
significant @xmath1-band veiling in all of the absorption line object spectra is evident when these are compared to template spectra at similar @xmath15 .
strong emission lines of very different nature were detected in order 49 for two objects : br16 in as 353a and multiple atomic , and possible molecular , lines in hbc 684 . in the @xmath2-band spectra for the sample , neutral atomic lines , such as na ,
he , al , and mg , are observed in orders 33 , 34 , 36 , and 37 ( e.g. , figure [ fig : as353a_k ] ) .
br @xmath21 appears in order 35 .
molecular lines of co ( orders 32 and 33 ) and h@xmath22 ( order 36 ) are also present .
the equivalent widths ( ew ) of the br@xmath21 line ( n=7@xmath44 ,
@xmath19=2.1661 @xmath18 ) were calculated for all objects observed in the @xmath2-band . for as 353a , the br@xmath21 ew , @xmath231 , is consistent with the br@xmath21 ew of @xmath24 from @xcite , derived from low spectral resolution ( r@xmath0800 ) observations made in 1996 .
ew was also calculated for the br16 ( n=16@xmath44 , @xmath19=1.5661 @xmath18 ) line observed in @xmath1-band order 49 of as 353a .
line fluxes were calculated using the equivalent widths and @xmath2-band magnitude from _ 2mass_. the uncertainty of the ew calculations is approximately @xmath61 , corresponding to an uncertainty of @xmath25 0.5 w@xmath4m@xmath26 for the br@xmath21 line fluxes and @xmath00.3 w@xmath4m@xmath26 for the br16 line fluxes .
brackett line ews and fluxes appear in table [ table : ews ] .
the lines that were primarily used to determine spectral type were two fe lines at 1.56259 @xmath18 and 1.56362 @xmath18 and a stellar oh doublet between them at 1.56314 @xmath18 in order 49 . for a k1 star ,
the fe lines are prominent and the oh line is nearly absent .
for later k spectral types , the fe lines monotonically weaken while the oh line strengthens ; all three lines are of approximately equal strength for spectral type m1 .
@xcite convolved rotation profiles from @xcite with the spectra of observed spectral type standard stars , resulting in template spectra , for a range of spectral types , with @xmath15=0 , 2 , 4 , 6 , 8 , 10 , 12 , 15 , 20 , 25 , 30 , 40 , and 50 km s@xmath5 .
examples of the template spectra with @xmath15=10 km s@xmath5 are presented in figure [ fig : templates ] ; these objects are listed in table [ table : templates ] .
uncertainties in the radial velocities of the template spectra are on the order of 1 km s@xmath5 @xcite .
the best fit spectral types were estimated by comparison with these standard star template spectra from @xcite and then confirmed with cross - correlation .
the spectral types are accurate to at least two subclasses ; there are templates for about every second subclass , but the difference between the template spectra is generally well - defined , especially in the fe and oh lines mentioned above . because of veiling of the object spectra by continuum emission from reprocessing in circumstellar disks and accretion - induced excesses , cross - correlation yields an artificial best fit for high values of @xmath15 .
therefore , the @xmath15 of each object was determined by visual inspection and comparison to rotated spectral templates alone .
the spectral type and @xmath15 for each object in the sample appear in table [ table : results ] .
radial velocities were measured by cross - correlating the target spectra with the best fit template spectrum from @xcite .
spectra of hbc 684 were inverted because the typical absorption lines appear in emission for this object ( see 4.5.5 ) . before cross - correlation , spectra were converted to the heliocentric reference frame , interpolated onto the same wavelength basis as the matching template spectrum , and rebinned by a factor of ten .
because of the strong br16 emission line in order 49 , a different approach was used to measure the radial velocity of as 353a .
absorption lines are more reliable for the measurement of radial velocity of an object because emission lines are produced by accretion flows or stellar winds , which have their own characteristic radial velocities .
therefore , we avoided the strong br16 emission line in the spectra of as 353a and used only the absorption lines shortward and longward of br16 in the cross - correlation . to estimate the uncertainties in the radial velocity measurements , the target spectra were cross - correlated with five template spectra of varying rotational velocity and spectral type , including the best - fit template .
the standard deviation of the resulting radial velocity measurements was less than 2 km s@xmath5 for all targets other than hbc 682a and hbc 684 . for several targets the standard deviation of measurements was less than 1 km s@xmath5 .
we therefore adopt an uncertainty in the measurement of 2 km s@xmath5 for all targets .
this is consistent with the more formal uncertainties for the same method derived by @xcite and @xcite .
we measured each object s radial velocity at every epoch observed .
table [ table : radvels ] lists the objects in column ( 1 ) , the ut dates of observation in column ( 2 ) , and the radial velocities measured for each object at each epoch in column ( 3 ) . for each object , we averaged the radial velocities over all epochs .
these average values are reported in column ( 4 ) of table [ table : results ] .
the standard deviation and the number of epochs for each target appear in columns ( 5 ) and ( 6 ) of table [ table : results ] .
no average radial velocity is reported for hbc 682a because of the large variations between epochs .
standard deviations are not given for fg aql / g3 and hbc 685 because only one observation of each object was made .
targets were initially examined for radial velocity variability by cross - correlating spectra taken at different epochs . for all objects except hbc 682a ,
radial velocity variations were less than @xmath04 km s@xmath5 , with typical values around 1@xmath42 km s@xmath5 .
this is within the estimated uncertainty of the radial velocity measurements .
radial velocity variations for hbc 682a were as large as 12 km s@xmath5 .
velocity shifts between spectra from ao and non - ao observations were systematically higher than the typical velocity shifts between non - ao spectra from different epochs .
this may be the result of the different approach to the dispersion solution necessary with ao observations .
the oh night sky lines used for the non - ao dispersion are distributed regularly throughout order 49 , allowing for an excellent solution .
the arc lamp lines required for the ao observations dispersion solution are only four and their distribution across order 49 is sparse .
to determine the general circumstellar properties of the sample we plot @xmath27 versus @xmath28 colors from the _
2mass _ photometry ( figure [ fig : ccd ] ) .
five of the objects have infrared excess according to this analysis , consistent with the same five objects that have ew(h@xmath7 ) @xmath8 10 ( table [ table : coords ] ) .
visual extinctions were estimated by dereddening the objects to the ctts locus @xcite using @xmath29 from @xcite .
the _ 2mass _ colors and the derived extinctions appear in table [ table : colors ] .
the average visual extinction of the sample is 1.6 magnitudes , not including as 353a , which falls below the ctts locus , possibly as the result of scattered light or a blue component to the color from continuum excess emission produced in accretion flows .
to estimate the ages of the aquila stars , we place all of the sample objects on an h - r diagram with the pre@xmath4main - sequence evolutionary tracks of @xcite ( figure [ fig : hrd ] ) .
luminosities were calculated using the method described in 3.3 of @xcite .
_ 2mass _ photometry was used for all objects except hbc 294 , for which the angularly resolved photometry of @xcite was used .
@xmath30-band bolometric corrections were from ( table 4 in * ? ? ? * ) .
temperatures were taken from @xcite based on the spectral types listed in table [ table : results ] .
our adopted distance , 200 pc , was selected based on previous distance estimates to stars in this region .
@xcite used 150@xmath650 pc as the distance to as 353b , citing distance estimates to the aquila rift and gould s belt .
@xcite estimated the distance to the dark cloud l694 to be 230@xmath630 pc based on star counts .
we adopt a distance that is consistent with both of these estimates , 200@xmath630 pc .
the bolometric luminosities for the components of hbc 682 have been corrected for binarity based on their near - infrared flux ratio of 2.25 and 2.15 , calculated from @xmath1- and @xmath2-band unpublished nirc2 images , respectively . for as 353b
the flux ratio of the components is @xmath01 , and for hbc 682 the flux ratio was calculated in @xmath1- and @xmath2-band from unpublished nirc2 images .
based on a distance of 150 pc , the age of as 353b was most recently determined to be 5@xmath1010@xmath11 years @xcite .
however , this calculation assumed that as 353b was a single star ; @xcite has since identified the system as a close binary comprised of two m1.5 stars with a flux ratio of @xmath01 . in placing as 353b on the hr diagram
, we adopt our spectral type of m0 ( table [ table : results ] and 4.5.1 ) , a distance of 200 pc , and one - half the luminosity determined from the unresolved system .
our resulting improved age for each component of as 353b is @xmath01 myr .
although the vertical error bars are large because of the uncertain distances to the objects , the location of the ensemble on the h - r diagram suggests that most of the objects are approximately 1 myr old .
the large scatter in age could be attributed to the uncertainty in extinction down to the photospheres as well as uncertainty in the amount of infrared excess , both of which could lead to over- or under - estimation of bolometric luminosity for a particular object .
disk signatures in the young set of objects are consistent with an age of less than a few myr .
hbc 682a and b and especially fg aql / g3 appear older , @xmath06@xmath412 myr , consistent with their lack of emission lines and ir excesses , although the spectrum of fg aql / g3 appears heavily veiled ( figure [ fig:08sep03 ] ) . the standard deviation of the radial velocities of all the stars in the sample ( @xmath02 km s@xmath5 ) , combined with many of the objects signatures of youth and spatial proximity , are consistent with most if not all of the targets being members of a young stellar association . the average radial velocity for each object ranges from @xmath46.8 to @xmath411.4 km s@xmath5 ; all but two of the objects have radial velocities within one standard deviation of the mean of @xmath48.6 km s@xmath5 . the only object for which a relatively constant radial velocity was not measured is hbc 682a , indicating that it may be a spectroscopic binary
higher sampling frequency observations are needed to determine its mass ratio and and center of mass velocity .
however , we speculate that hbc 682a is a probable member of the aquila association because of its proximity to hbc 682b ( @xmath01 ) .
thus , it is likely that all nine stars in the sample belong to a single association that is part of the aquila star - forming region .
the multiplicity among the aquila association members is comparable to that of taurus , the well - studied star - forming region with the highest binary fraction ( e.g. * ? ? ?
* ; * ? ? ?
of seven systems that we observed , as many as thirteen components were found .
systems hbc 681 and hbc 682 were identified as multiple candidates for the first time .
two systems were known binaries ( hbc 294 and hbc 682 ) and at least one is triple ( as 353 ) .
additionally , hbc 682a may be a spectroscopic binary ( 4.5.3 ) and hbc 681 may have a low - mass companion ( 4.5.2 ) .
the order 49 spectra from the three observations of as 353a are presented in figure [ fig : as353a ] . as 353a is best fit by a k5 template spectrum with @xmath15=10 km s@xmath5 , somewhat later than the k2 spectral type previously reported in the literature @xcite .
the order 49 spectrum from the observation of as 353b presented in figure [ fig:08sep03 ] is best fit by an m0 template spectrum with @xmath15=20 km s@xmath5 , which is similar to the m1.5 spectral types found for both components of this subarcsecond binary by @xcite using low resolution @xmath2-band spectra .
strong br16 ( n=16@xmath44 ) emission dominates the spectrum of as 353a in order 49 .
a rise at the short wavelength end of the 2004 july 22 ( ut ) spectrum also suggests the presence of br17 , although the majority of the line falls outside the spectral coverage of the order .
these strong brackett lines are consistent with the strength of the h@xmath7 detected in the optical @xcite and confirmed by us in a visible light echelle spectrum ( not shown ) .
however , it is unusual to see what are presumably photospheric absorption lines superimposed on the emission line .
the emission line appears to shift relative to the absorption lines between epochs .
the veiling also appears variable ; for example , the absorption lines are markedly deeper in the ut 2004 july 22 spectrum than in earlier spectra ( figure [ fig : as353a ] ) . selected orders of the @xmath2-band spectra of as 353a along with probable feature identifications
are presented in figure [ fig : as353a_k ] .
order 35 shows strong br@xmath21 emission , which is again consistent with the strong h@xmath7 emission . also apparent in the @xmath2-band spectra
are atomic mg , al , and he , the co(2@xmath40 ) bandhead , and possibly h@xmath22 , all in emission .
this rich emission spectrum is also apparent at low resolution ( r@xmath0760 ) @xcite and in the optical @xcite . both as 353a and b occupy unusual locations on the color - color diagram .
the position of as 353a indicates significant ir excess , consistent with the origin of the strong hydrogen emission in a circumstellar disk .
the detection of as 353 by the _ iras _ satellite ( _ iras _ 19181@xmath171056 ; the components are unresolved in the _ iras _ beam ) also points to the presence of warm dust in the system @xcite . however , as 353a lies _ below _ the ctts locus , which may be the result of abundant scattered light in the circumstellar environment or a blue component of excess continuum emission from accretion flows . as 353b
does not show ir excess on the color - color diagram , which is consistent with its classification as a wtts based on the strength of its h@xmath7 emission .
however , the extinction is larger than that of other cluster members if it is dereddened to the ctts locus .
the prominent neutral hydrogen emission from as 353a is likely a combination of accretion and stellar winds .
emission is the primary characteristic used to identify circumstellar accretion @xcite .
however , a strong stellar wind is indicated by other characteristics of this system , namely , that it is the exciting source of hh 32 , a system of bright , compact emission line regions that are produced by a highly collimated bi - polar outflow @xcite .
the h@xmath7 emission from as 353a also displays a p cygni profile with absorption in the blue wing of the emission line @xcite , which indicates outflow from the star .
the similar radial velocity measurements of as 353a and b ( @xmath411.4 and @xmath410.7 km s@xmath5 , respectively ) and similar proper motions measured by @xcite make it likely that they are indeed a bound pair .
therefore , as 353 is a mixed t tauri system in which the more massive component is a ctts while the secondary is a wtts , which is thought to be more evolved than a ctts .
the slight infrared excess implied by the @xmath31 color of as 353ba by @xcite , 0.56 magnitudes , suggests the possible presence of circumstellar material around at least one of the as 353b components .
however , if the @xmath2- and @xmath32-band magnitudes for as 353ba are dereddened by an a@xmath33 of 2 magnitudes @xcite , the resultant k@xmath4l color is 0.44 magnitudes . for an a@xmath33 of 3 magnitudes ( this paper ) , the dereddened k@xmath4l color is 0.37 magnitudes . therefore , if circumstellar material is present at all around as 353ba , it is likely to be found at an advanced stage of dispersion .
possibly the binarity of as 353b has played a role in the relatively early dissolution of its circumstellar disks .
it has been suggested that as 353a is also a close binary @xcite .
if true , this would complicate the idea that binarity drives accelerated disk evolution , as as 353a harbors one of the most substantial and active circumstellar disks known .
@xcite find no evidence for a companion object down to a separation of 0.1@xmath34 , corresponding to 20 au for a distance of 200 pc to the system .
if we assign as 353a a mass of 1.2 m@xmath12 ( figure [ fig : hrd ] ) , the velocity signature induced by a companion object with a mass of 0.8 m@xmath12 in a circular orbit is about 7.5 km s@xmath5 .
any higher mass companion would result in a larger amplitude of radial velocity variability in as 353a , whereas any inclination of the system s orbit out of the plane of our line of sight , a larger distance to the system than 200 pc , or a smaller mass companion , would result in smaller amplitude velocity variations . over an interval of two years ,
however , no radial velocity variation greater than 1 @xmath3 was measured between our three observations . our observations
, however , may be insensitive to highly eccentric spectroscopic binary orbits ( e.g. * ? ? ?
hbc 294 is best fit by a k7 template spectrum with @xmath15=25 km s@xmath5 .
this agrees with the spectral type reported in the hbc .
the radial velocity is well - determined , with a standard deviation of @xmath9 0.5 km s@xmath5 based on measurements from three different epochs .
hbc 294 , also known as v536 aql , is a typical ctts , with br@xmath21 emission detected in our @xmath2-band observations and h@xmath7 , pa@xmath35 , and optical forbidden emission lines reported in the literature , indicative of accretion from a circumstellar disk .
@xcite describe evidence for a bipolar outflow from the system as well .
@xcite first discovered hbc 294 to be a binary with an 052 separation at a position angle of 17 .
the unresolved near - ir excess observed in the system indicates warm dust surrounding at least one of the stars .
the _ iras _ psc flux measurements of hbc 294 ( _ iras _ 19365 + 1023 ) reveal even greater mid- to far - ir fluxes than those of as 353a at all but 100 @xcite .
it is unclear whether the circumstellar material is located around one or both components of this binary .
the lower 100 @xmath36 m flux suggests that circumstellar material is truncated by the binary companion .
the extinction required to deredden hbc 294 to the ctts locus is comparable to the average extinction of the sample . a k5 template with @xmath15=20 km s@xmath5 provided the best fit to the spectrum of hbc 681 .
the h@xmath7 emission of hbc 681 , ew=35 , is relatively low for the ctts in this sample , but still well above the 10 ctts limit .
a clear ir excess is indicated by the location of hbc 681 on the color - color diagram .
we identified a faint potential companion to hbc 681 at separation @xmath012 and a position angle of @xmath0195@xmath37 .
the @xmath28 color of this object is @[email protected] , typical of an unreddened m6 star , which , at the age of the aquila association , would be substellar .
however , the proper motion of this object must be observed to determine if the two objects are indeed related as hbc 681 is located in the direction of the galactic plane , and hence the field is crowded with faint background objects . for the spectrum of hbc 683 , we found a best fit using a k7 template spectrum with @xmath15=15 km s@xmath5 .
this is later than the spectral type of k1 reported in the hbc .
the 2003 september 08 ( ut ) order 49 spectrum of hbc 683 is one of the best for line identification because of the high signal to noise ratio and low @xmath15 relative to the rest of the sample .
hbc 683 is a typical ctts and has the strongest h@xmath7 and br@xmath21 emission in our sample after as 353a .
hbc 683 also shows a clear ir excess in the color - color diagram ( figure [ fig : ccd ] ) .
these properties are all consistent with its ctts classification and the presence of an actively accreting circumstellar disk . as a result of the observations presented in this paper , hbc 682
was discovered to be a visual binary with a separation of @xmath01 .
hbc 682a is best fit by a k5 template spectrum with @xmath15=50 km s@xmath5 , and hbc 682b by a k7 template spectrum with @xmath15=30 km s@xmath5 . both of these spectral types are earlier than that of m0 reported in the hbc .
although the spectra of both components of hbc 682 have similar @xmath1- and @xmath2-band features , there are striking rotational and radial velocity differences between the two .
hbc 682a has the broadest absorption lines in our sample , while hbc 682b is more typical .
a consistent radial velocity was not measured for hbc 682a because of the large variations between epochs ( table [ table : radvels ] ) .
this may be in part because of the wide and shallow absorption lines that make precise measurement of radial velocity more difficult .
however , the radial velocity variations measured in 5 different epochs , were significantly larger than the typical variation for other objects , e.g. , 1@xmath42 km s@xmath5 .
hbc 682a is therefore a candidate spectroscopic binary .
we speculate that hbc 682a and b are bound because of their close proximity .
verification by component proper motion measurements and by determination of the center of mass velocity of the hbc 682a spectroscopic binary is needed , particularly in that this system is located in projection on to the galactic plane and hence confusion with background sources is possible .
the angularly unresolved hbc 682 system shows no ir excess , consistent with its weak h@xmath7 and lack of br@xmath21 emission .
however , the classification of hbc 682 is complicated by the fact that it has only recently been resolved into a visual binary , and may in fact be a hierarchical triple .
the components lack separate h@xmath7 emission measurements and resolved near - ir photometry , both of which might provide additional information about their individual circumstellar environments . an m0 template with @xmath15=15
km s@xmath5 provided the best fit to the fg aql / g3 system .
the only observations of this object since @xcite are from _ 2mass _ and _ denis _ ( deep near infrared survey ) , which combined provide @xmath38- , @xmath30- , @xmath1-and @xmath2-band photometry .
no ir excess is detected .
there are as yet no known signatures of youth associated with fg aql / g3 ; @xcite did not detect h@xmath7 emission from this star and we did not observe it in the @xmath2-band . the @xmath1-band spectrum , however , appears highly veiled .
the location within 20 of three cluster members and fg aql / g3 s measured radial velocity are consistent with cluster membership .
however , further evidence is needed to confirm the youth of this object .
the spectra from the three observations of hbc 684 are presented in figure [ fig : hbc684 ] .
the inverted order 49 spectrum of hbc 684 is best fit by a k5 template spectrum with @xmath15=40 km s@xmath5 .
the source of the emission lines in the spectra of this star is enigmatic .
an examination of the fits files revealed that the emission lines appeared in the raw data .
furthermore , emission lines were apparent in all orders of the raw @xmath1- and @xmath2-band data .
although emission lines associated with accretion are common for young stars , they are usually manifested in strong hydrogen lines and only occasionally in other species , such as co and neutral na @xcite .
hbc 684 shows only modest h@xmath7 ( 11 in the hbc and @xmath013 in our unpublished visible light spectrum ) and br@xmath21 emission , while most of the atomic lines in order 49 , mainly fe but possibly ti , si , and ni as well , all appear in emission .
we have tentatively identified lines of oh and cn also in emission ; see figure [ fig : hbc684 ] and table [ table : lines ] for line identifications .
in addition to these molecular lines in order 49 , many emission lines probably arising from oh were identified in orders 46 , 47 , 48 , and 50 , and several emission lines from cn in orders 48 and 50 .
all of the typical ctts is this sample ( hbc 294 , hbc 681 , and hbc 683 ) show stronger h@xmath7 emission , and all but hbc 294 show stronger br@xmath21 emission as well ( table [ table : ews ] ) . @xcite reported a rich near - ir emission line spectrum of the class i young stellar object _ iras _
04239@xmath172436 in taurus .
although the resolution was modest ( r@xmath0500 ) , over a dozen emission features were identified in the observed spectral range of 1.1 to 2.5 @xmath18 .
probable features included several co bandheads , br@xmath21 , br10 , br11 , br13 , br14 , pa@xmath35 , h@xmath22 , and neutral he .
only the neutral ca triplet at @xmath19=2.26 @xmath18 was observed in absorption .
we know of no other spectra , however , with low - excitation atomic and possibly molecular species in emission as they appear in hbc 684 s @xmath1-band spectra .
the palomar observatory sky survey and _ 2mass _ images show some nebulosity to the north and east of hbc 684 ; however , this diffuse emission was not evident in more recent @xmath1-band images ( in short , 0.1s exposures ) taken with the nirspec slit - viewing camera ( scam ) in 2004 november .
hbc 684 is an _ iras _ source ( _ iras _
19046@xmath170508 ) with psc fluxes at 12 , 25 , 60 and 100 @xmath18 even greater than those of as 353a , which indicates a considerable amount of warm dust around the star .
the near - ir excess of hbc 684 is the strongest in our sample ( figure [ fig : ccd ] ) .
we suggest four mechanisms that might produce spectra such as those observed in hbc 684 .
first , accretion could be creating an oblique shock close to the photosphere of the star , which occurs when the accretion speed equals the escape velocity of the star .
however , the lack of strong h@xmath7 emission limits the amount of circumstellar material that could be accreting from a circumstellar disk .
furthermore , our h@xmath7 spectrum shows a clear p cygni profile , characteristic of outflow , but not infall .
hbc 684 s emission lines could also be produced by an inversion layer resembling the chromosphere of the sun .
if some blocking mechanism such as a circumstellar disk , similar to the function of the moon during total solar eclipse , could be construed , a phenomenon resembling a solar flash spectra might be produced ( see * ? ? ?
stellar chromospheres are heated by stellar coronae ; the x - ray emission frequently observed in young stars is thought to originate in the corona ( * ? ? ? * e.g. ) .
however , the nearest known x - ray source in the vicinity of hbc 684 , 1rxs j190722.4@xmath17051231 in the _ rosat _ all - sky survey faint source catalog , is 3.2 away and is therefore probably not associated with hbc 684 .
a transient impact on the photosphere could provide sufficient energy to excite the atomic metals that are observed in emission .
in 1994 july fragments of the comet shoemaker - levy 9 plunged into the atmosphere of jupiter .
@xcite observed the plume of the most energetic impact and found atomic emission lines from neutral na , fe , ca , li in the optical spectrum .
infrared spectra of another impact show molecular band / line emission , including ch@xmath39 and co(2@xmath40 ) @xcite .
given the few million year age of this association and the evidence for circumstellar material around hbc 684 , it is plausible that planetessimals are forming around and may occasionally fall into the star . indeed , the planet plunging phenomenon proposed to account for the high metallicity observed in extra - solar planetary systems @xcite may be invoked in this case .
the slow motion merger of a giant planet or brown dwarf could create the requisite energy for the fe i emission lines ( on the order of 6@xmath47 ev ) without exciting significant hydrogen emission .
such a merger would require a substantial circumstellar disk in order to abet angular momentum transfer ; the presence of a disk is supported by the strong near- and mid - ir excesses observed in this system .
the p cygni profile in the h@xmath7 emission line implies an outflow of 200 km s@xmath5 .
a planet plunging event might stimulate the velocity of such a jet .
the duration of the effects of a merger is not known ; we have observed consistent emission line spectra of hbc 684 on three separate occasions over a 14 month period . in 1981 , a broad , strong p cygni profile was already present ( notes in hbc ) .
a final possibility we consider is the stimulation of line emission by fluorescence .
we found that the upper energy level terms for the order 49 fe i lines represent too broad a range of states for the production of all the @xmath1-band lines to be consistent with fluorescence originating in one particular pumping line .
furthermore , all the lines expected in a normal k5 star are present in the infrared spectra of hbc 684 , except in emission .
compare this , for example , to hbc 682b , which has a similar spectral type and value of @xmath15 .
hbc 684 s emission lines also display similar line ratios as the corresponding absorption lines in a template spectrum or in hbc 682b .
some of the unusual features in the hbc 684 spectra may be the result of fluorescence , but we are skeptical that this is the explanation for the entire set of emission lines observed .
this paper presents the results of multi - epoch observations of a sample of nine young , late - type stars in the aquila star - forming region .
sample objects were selected from @xcite based on their proximity to as 353a and classification as a wtts or ctts . the high - resolution ( r@xmath030,000 )
@xmath1-and @xmath2-band observations were made between 2002 july and 2004 november with nirspec on keck ii .
the initial goal of this project was to examine one @xmath1-band order of the multiple - epoch spectra for radial velocity variations in order to detect spectroscopic binaries .
one radial velocity variable object was discovered and absolute radial velocities were measured for the remaining eight objects .
circumstellar properties of the stars were also examined via high - resolution @xmath2-band spectra and _ 2mass _ photometry .
eight of the nine sample objects appear to be members of the same association because of similar radial velocities .
the spatial extent of the known members , using our adopted distance of 200@xmath630 pc , is 32@xmath65 pc by 58@xmath69 pc .
the ages of the observed objects exhibit a large scatter , with the youngest , most active members all of age @xmath01@xmath42 myr .
two systems appear to have ages between 6 and 12 myr , and the oldest of these ( fg aql / g3 ) has not been confirmed to show signatures of youth .
therefore , the aquila association of stars is apparently one of the youngest nearby star - forming regions .
the visual extinction to the association is estimated to be about 1@xmath42 magnitudes .
the circumstellar environments of associations members are varied , ranging from extremely active ( as 353a , whose winds drive hh 32 ) to nearly devoid of circumstellar material ( as 353b , hbc 682 , and fg aql / g3 , which all lack ir excess and strong h@xmath7 emission ) , with several typical ctts in between ( hbc 294 , hbc 681 , and hbc 683 ) .
the source of the atomic and molecular emission lines in the spectra of hbc 684 is as yet unknown , but may be caused by an inversion layer , increased accretion , fluorescence , or a transient impact .
for the seven systems in this study , hbc 681 , hbc 682 , hbc 683 , hbc 684 , hbc 294 , fg aql / g3 , and as 353 , there is a high total number of stars , thirteen at least , yielding @xmath02 stars per system .
this multiplicity is significantly higher than even that of the taurus region ( @xmath01.6 stars per system ; @xcite ) , although it applies to only a small sample and may be significantly impacted by observational biases .
the extremely interesting stars identified to date in this association certainly warrant further investigation .
much work remains to characterize this unusual star - forming region .
the most pressing task is to confirm the existence of a larger association by searching for other young stars in the region and identifying radial velocities and proper motions of the new members .
obtaining a more accurate distance determination to the association will also be critical to unraveling the age - distance discrepancy .
for the non - thermal radio source as 353b , this could be accomplished with vlbi ( e.g. * ? ? ?
once more complete membership is determined , the aquila star - forming region can be characterized in terms of mass function , multiplicity fraction , and other parameters .
further observations in the millimeter , sub - millimeter , infrared , radio , and x - ray will also provide valuable information about the circumstellar environments of these objects , which will improve our understanding of key phases of pre - main - sequence stellar evolution .
we thank the expert observing assistants and staff of the keck observatory for their capable help .
insightful discussions with c. mccabe , d. hunter , m. simon , s. kenyon , e. becklin , m. jura , b. zuckerman , s. strom , g. herbig , and b. reipurth improved the presentation of this paper .
we are grateful to s. zoonematkermani for providing some of the idl procedures used in the analysis , to c. mccabe and g. duchne for obtaining several spectra for us in 2004 may , to m. huerta for speedy reduction of the as 353a and hbc 684 h@xmath7 spectra taken at mcdonald observatory in 2005 november , and to q. konopacky for several nirc2 images .
we thank the anonymous referee for helpful comments .
l. prato was supported in part by nsf grant ast 04 - 44017 .
data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the universities of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
this work made use of the simbad reference database , the vizier service , the nist atomic spectra database , the nasa astrophysics data system , and the nasa / ipac infrared service archive .
mclean , i. s. , beckin , e. e. , bendiksen , o. , brims , g. , canfield , j. , figer , d. f. , graham , j. r. , hare , j. , lacayanga , f. , larkin , j. e. , larson , s. b. , levenson , n. , magnone , n. , teplitz , h. , & wong , w. 1998 , proc .
spie , 3354 , 566 tokunaga , a. t. , reipurth , b. , gssler , w. , hayano , y. , hayashi , m. , iye , m. , kanzawa , t. , kobayashi , n. , kamata , y. , minowa , y. , nedachi , k. , oya , s. , pyo , t. , saint - jacques , d. , terada , h. , takami , h. , & takato , n. 2004 , , 127 , 444 wizinowich , p. , acton , d. s. , shelton , c. , stomski , p. , gathright , j. , ho , k. , lupton , w. , tsubota , k. , lai , o. , max , c. , brase , j. , an , j. , avicola , k. , olivier , s. , gavel , d. , macintosh , b. , ghez , a. , & larkin , j. 2000 , , 112 , 315 hbc 681 & fg aql / g1 & 19 02 22.20 & @xmath405 36 20.4 & 29.17 , @xmath44.98 & 9.429 & 35 + fg aql / g3 & _ 2mass _ 190222.6@xmath4053522 & 19 02 22.60 & @xmath405 36 22.1 & 29.17 , @xmath44.98 & 12.019 & + hbc 682 & fg aql / g2 & 19 02 22.84 & @xmath405 36 15.8 & 29.18 , @xmath44.98 & 9.492 & 2.3 + hbc 683 & fh aql & 19 02 23.22 & @xmath405 36 37.3 & 29.17 , @xmath44.98 & 10.380 & 105 + hbc 684 & _ iras _
19046@xmath170508 & 19 07 09.87 & @xmath1705 13 10.8 & 39.37 , @xmath41.11 & 8.555 & 11 + as 353a & hbc 292 & 19 20 30.99 & @xmath1711 01 54.7 & 46.05 , @xmath41.33 & 9.161 & 150 + as 353b & hbc 685 & 19 20 31.04 & @xmath1711 01 49.1 & 46.05 , @xmath41.33 & 9.154 & 4.4 + hbc 294 & v536 aql & 19 38 57.41 & @xmath1710 30 16.1 & 47.75 , @xmath45.57 & 8.102 & 52 + [ table : coords ] llcccc 2002 july 18 & as 353a & n5 & 300 & 1.03 & 170 + & as 353b & n5 & 300 & 1.04 & 200 + 2003 august 10 & hbc 681 & n5 & 300 & 1.11 & 350 + & fg aql / g3 & n5 & 220 & 1.11 & 250 + & hbc 682 ( a & b ) & n5 & 180 & 1.11 & 280 + 2003 september 07 & hbc 681 & k & 240 & 1.12 & 150 + & hbc 682 ( a & b ) & k & 200 & 1.13 & 180 + & hbc 683 & k & 240 & 1.16 & 170 + & hbc 684 & k & 120 & 1.11 & 170 + & hbc 294 & k & 120 & 1.05 & 180 + 2003 september 08 & hbc 681 & n5 & 240 & 1.13 & 250 + & hbc 682 ( a & b ) & n5 & 240 & 1.19 & 270 + & hbc 683 & n5 & 220 & 1.40 & 250 + & hbc 684 & n5 & 300 & 1.49 & 280 + & hbc 294 & n5 & 300 & 1.40 & 420 + 2004 may 27 ( ao ) & hbc 681 & n5 & 300 & 1.25 & 210 + & as 353a & n5 & 300 & 1.02 & 170 + 2004 may 28 ( ao ) & hbc 682a & n5 & 300 & 1.22 & 250 + 2004 july 22 & hbc 682a & n5 & 240 & 1.28 & 300 + & hbc 681 & n5 & 180 & 1.36 & 300 + & hbc 684 & n5 & 180 & 1.30 & 340 + & as 353a & n5 & 300 & 1.05 & 210 + & hbc 294 & n5 & 120 & 1.06 & 320 + 2004 november 21 & hbc 683 & n5 & 240 & 1.56 & 80 + & hbc 684 & n5 & 180 & 1.53 & 330 + & hbc 294 & n5 & 120 & 1.39 & 270 + 2004 november 22 & hbc 682 ( a & b ) & n5 & 180 & 1.57 & 210 + & as 353a & k & 240 & 1.49 & 340 + [ table : obslog ] n5 & 2002 jul . - 2004 nov . &
63.04 & 36.30 & 0.288 @xmath10 24 & 30,800 + n5 ( ao ) & 2004 may 27 & 28 & 63.04 & 36.30 & 0.027 @xmath10 2.26 & 35,900 + k & 2003 september 07 & 62.15 & 35.59 & 0.288
@xmath10 24 & 24,500 + k & 2004 november 22 & 62.84 & 35.59 & 0.288 @xmath10 24 & 29,000 + [ table : inst ] lrrc & br@xmath21 & & + hbc 681 & 2003 september 07 & @xmath45 & [email protected] + hbc 682a & 2003 september 07 & 1 & + hbc 682b & 2003 september 07 & 1 & + hbc 683 & 2003 september 07 & @xmath45 & [email protected] + hbc 684 & 2003 september 07 & @xmath43 & [email protected] + hbc 294 &
2003 september 07 & 0 & [email protected] + as 353a & 2004 november 22 & @xmath422 & [email protected] + & br16 & & + as 353a & 2002 july 18 & @xmath45 & [email protected] + as 353a & 2004 may 27 & @xmath47 & [email protected] + as 353a & 2004 july 22 & @xmath44 & [email protected] + lccccc hbc 681 & k5 & 20 & @xmath47.4 & 0.9 & 4 + fg aql / g3 & m0 & 15 & @xmath48.9 & & 1 + hbc 682a & k5 & 50 & & 5.6 & 5 + hbc 682b & k7 & 30 & @xmath46.8 & 1.5 & 3 + hbc 683 & k7 & 15 & @xmath46.8 & 0.8 & 2 + hbc 684 & k5 & 40 & @xmath47.2 & 2.2 & 3 + as 353a & k5 & 10 & @xmath411.4 & 1.1 & 3 + as 353b & m0 & 20 & @xmath410.7 & & 1 + hbc 294 & k7 & 25 & @xmath49.8 & 0.3 & 3 + [ table : results ] llr hbc 681 & 2003 august 10 & @xmath48.1 + & 2003 september 08 & @xmath47.7 + & 2004 may 27 & @xmath46.4 + & 2004 july 22 & @xmath47.6 + fg aql / g3 & 2003 august 10 & @xmath49.4 + hbc 682a & 2003 august 10 & -3.0 + & 2003 september 08 & @xmath413.2 + & 2004 may 28 & @xmath46.9 + & 2004 july 22 & @xmath46.8 + & 2004
november 22 & @xmath42.1 + hbc 682b & 2003 august 10 & @xmath45.9 + & 2003 september 08 & @xmath45.5 + & 2003
november 22 & @xmath48.5 + hbc 683 & 2003 september 08 & @xmath47.4 + & 2004 november 21 & @xmath46.2 + hbc 684 & 2003 september 08 & @xmath45.5 + & 2004 july 22 & @xmath47.2 + & 2004
november 21 & @xmath49.8 + as 353a & 2002 july 18 & @xmath412.1 + & 2004 may 27 & @xmath410.1 + & 2004 july 22 & @xmath412.1 + as 353b & 2002 july 18 & @xmath410.7
+ hbc 294 & 2003 september 08 & @xmath49.4 + & 2004 july 22 & @xmath49.8 + & 2004
november 21 & @xmath49.8 + [ table : radvels ] hbc 681 & 0.925 & 0.636 & 0.1 + fg aql / g3 & 0.858 & 0.212 & 2.7 + hbc 682 & 0.752 & 0.161 & 1.6 + hbc 683 & 1.015 & 0.712 & 0.7 + hbc 684 & 1.342 & 1.156 & 1.6
+ as 353a & 0.733 & 0.723 & @xmath43.3 + as 353b & 1.029 & 0.396 & 3.5 + hbc 294 & 1.042 & 0.706 & 1.1 + [ table : colors ] hbc 681 & possible low - mass companion & @xmath01.2 & @xmath0195 + fg aql / g3 & single & & + hbc 682 & binary , probable triple & @xmath01 & @xmath0160 + a & possible sb & & + hbc 683 & single & & + hbc 684 & single & & + as 353 & triple & 5.6 & 174 + b & binary & 0.24 & 107 + hbc 294 & binary & 0.52 & 17 + [ table : mult ] llc 1.54562 & fe & 0 + 1.54635 & fe & 1 + 1.54740 & s & 2 + 1.54797 & fe / s & 3 + 1.54829 & s / fe & 4 + 1.54897 & fe & 5 + 1.54951 & fe & 6 + 1.55009 & fe & 7 + 1.55036 & fe & 8 + 1.55053 & fe & 9 + 1.55103 & c & 10 + 1.55185 & fe & 11 + 1.55236 & fe & 12 + 1.55360 & fe & 13 + 1.55385 & fe & 14 + 1.55448 & ni & 15 + 1.55463 & fe & 16 + 1.55480 & ti & 17 + 1.55552 & fe & 18 + 1.55596 & ni & 19 + 1.55620 & si & 20 + 1.55650 & fe & 21 + 1.55710 & fe & 22 + 1.55760 & fe & 23 + 1.55833 & fe & 24 + 1.55925 & fe & 25 + 1.55958 & fe & 26 + 1.56071 & ti & 27 + 1.56085 & fe & 28 + 1.56099 & ni & 29 + 1.56154 & fe & 30 + 1.56179 & fe & 31 + 1.56259 & fe & 32 + 1.56314 & oh & 33 + 1.56336 & fe & 34 + 1.56362 & fe & 35 + 1.56528 & fe & 36 + 1.56571 & fe & 37 + 1.56663 & fe & 38 + 1.56695 & fe & 39 + [ table : lines ] | we present the properties of a group of young stars associated with the well - studied t tauri star system as 353 , located in the aquila star - forming region .
the association is identified using radial velocity measurements of sample objects selected from the herbig and bell catalog based on their spatial proximity to as 353 .
radial velocities of nine objects were measured from multi - epoch high - resolution ( r@xmath030,000 ) @xmath1-band spectra obtained with nirspec on keck ii .
high - resolution @xmath2-band spectra were also obtained for most of the sample objects .
spectral types and rotational velocities are determined for all objects in the sample .
the multi - epoch @xmath1-band spectra were examined for radial velocity variations in order to detect possible spectroscopic binaries .
eight of the nine objects have radial velocities that are consistent within the 1-@xmath3 scatter of the sample . from their mean of @xmath48
km s@xmath5 these eight objects have a standard deviation of 2 km s@xmath5 , which suggests that the sample stars are related .
the ninth object shows significant radial velocity variations between epochs , characteristic of a spectroscopic binary .
the overall multiplicity of the sample is high ; we observed 13 stars in seven systems , identifying three new candidate binary components in this project .
many of the spectra reveal hydrogen emission lines typical of strong accretion processes , indicating that most of these objects harbor circumstellar disks and are less than a few million years old . based on previous estimates ,
we adopt a distance of 200@xmath630 pc to the young stars in aquila in order to calculate luminosities and place the stars on an h - r diagram .
we discuss possible interpretations of the enigmatic pure emission line spectrum of hbc 684 .
this work represents the highest spectral resolution infrared observations to date of these intriguing , nearby young stars . |
i would like to thank dr sudhir .
r. jain and dr abhas mitra for critical reading of the manuscript .
1 . a. messiah , _ quantum mechanics _ ( north - holland publishing .
co. , amsterdam , 1962 ) vol .
i , pp . 91 - 114 . 2 .
e. merzbacher , _ quantum mechanics _ ( john wiley & sons , new york , 1970 ) ch . 6 , pp .
80 - 114 3 .
s. flgge , _ practical quantum mechanics _ ( springer - verlag , berlin , 1971 ) vol .
4 . d. rapp , _ quantum mechanics _ ( holt , rinhart & wilson inc .
, new york , 1970 ) , part iii , ch .
125 - 200 ; g. barton , `` quantum mechanics of inverted harmonic oscillator potential '' , ann .
( n.y . ) * 166 * , 322 ( 1986 ) ; z. ahmed , `` tunneling through asymmetric parabolic barriers '' , j. phys .
a : math . gen .
( 1997 ) ) 3115 - 3116 . 5 .
landau and e. m. lifshitz , _ quantum mechanics _
( pergamon press , london,1958 )
p.m. morse and h. feshbach , _ methods of theoretical physics _ ( mcgraw - hill book company , ltd .
, new york , 1953 ) pp.1650 - 1660 . 7 .
ginocchio , `` a class of exactly solvable 1 : one - dimensional schrdinger equation '' , ann . phys . * 152 * , 203 ( 1984 ) ; b.sahu , s.k . agarwalla and c.s .
shastry , `` above - barrier resonances : analytic expression for energy and width '' , j. phys .
a : math . gen . * 35 * , 4349 - 4358 ( 2002 ) .
a. khare and u. p. sukhatme , `` scattering amplitudes for supersymmetric shape - invariant potential by operator method '' , j. phys .
a : math . gen .
* 21 * l501-l508 ( 1988 ) . 9 .
z.ahmed,``tunneling through the morse barrier '' , phys .
a * 157 * , 1 - 5 , ( 1991 ) . 10 .
z.ahmed , `` tunneling through a one dimensional potential barrier '' , phys . rev .
a , * 47 * 4761 - 4757 ( 1993 ) . 11 .
arfken and h.j .
weber , _ mathematical methods for physicists _ , ( academic press inc .
, san diego,1995 ) pp .
627 - 664 ; m. abramowitz and a. s. stegun , _ handbook of mathematical functions _ ( dover publications , inc . , new york , 1970 )
pp.358 - 365
b.f . buxton and m.v .
berry , `` bloch wave degeneracies in systematic high energy electron diffraction '' , phil .
a * 282 * , 485 - 525 ( 1976 ) ( see p.534 ) ; g. barton , `` levinsion theorem in one dimension : heuristics '' , j. phys .
a * 18 * 497 ( 1985 ) ; y. nogami and c.y .
ross , `` sacattering from a non - symmetric potential in one dimension : as a coupled channel problem '' , am .
j. phys . ,
* 64 * , 923 ( 1996 ) . 13 .
z. ahmed , `` schrdinger transmission through one - dimensional complex potentials '' , phys .
a * 64 * 042716-(1 - 4 ) ( 2001 ) . 14 .
e.g. , h. a. enge,_introduction to nuclear physics _
( addison wesley publishing co. , reading , mass . , 1966 ) ,
pp . 274 - 295 . 15 .
e.g. , a.s .
davydov , _ quantum mechanics _ ( pergamon press , new york , 1965 ) . | we present the case of quantal transmission through a smooth , single - piece exponential potential , @xmath0 , in contrast to the piece - wise continuous potentials , as a pedagogic model to demonstrate the analytic extraction of transmission ( reflection ) co - efficient .
= 100000 the schrdinger transmission through a one - dimensional potential well / barrier constitutes an essential topic in the courses and text - books [ 1,2 ] on quantum mechanics .
the main interest is to find the probability of transmission , @xmath1 , and that of reflection , @xmath2 , when a particle is incident with an energy @xmath3 at the potential from one side . for a real potential
these two probanilities are such that @xmath4 , ensuring the conservation of the incident flux of the particles . due to simplicity of treatment , usually a rectangilar well / barrier or a sharp step - potentials
are included as examples .
these potentials are both piece - wise constant and piece - wise continuous , the former is a three - piece and the latter is a two - piece function . for the case of three - piece potentials :
@xmath5 and @xmath6 , where the schrdinger equation does not possess analytic form , the numerical computation of @xmath1 is performed by letting @xmath7 , @xmath8 and @xmath9 [ 3 ] .
the functions @xmath10 and @xmath11 denote two linearly indedependent solutions of the schrdinger equation with initial values as @xmath12 , for numerical integration of the schrdinger from @xmath13 upto @xmath14 . here
the prime denotes differentiation with respect to @xmath15 . for the case of a rectangular well / barrier of depth / height , @xmath16 ,
these functions are @xmath17 and @xmath18 , where @xmath19 . the + matching of the wavefunctions and their derivatives at @xmath14 , leads to the elimination of @xmath20 and @xmath21 and we can get the complex transmission amplitude as @xmath22 and complex reflection amplitude as @xmath23 the energy - dependent quantities @xmath24 and @xmath25 are called reflection phase - shift and transmission phase - shift , respectively .
the energy derivative of the phase - shift is interpreted as the time spent by a particle in these processes [ 1,2 ] . however , in practical situations the potentials are smooth ( continuous and diferentiable everywhere ) and single piece which may or may not vanish asymptotically .
these potentials are the parabolic barrier [ 4 ] , the eckart barrier [ 3,4,5 ] , the fermi - step barrier[3,5 ] , the rosen - morse barrier [ 6 ] , the ginocchio barrier [ 7 ] , the scarf barrier [ 8 ] , the morse barrier [ 9 ] and a potential which interpolates between morse and eckart barriers [ 10 ] .
the transmission co - efficients for these models have been obtained analytically . in all such cases [ 3 - 10 ] ,
a single function is obtained as the solution of the schrdinger equation which behaves as a transmitted wave on one side of the potential at asymptotically large distances ( say , @xmath26 ) . on the other side of the potential
the same wavefunction behaves asymptotically ( say , @xmath27 ) as a combination of oppositely travelling incident and reflected waves .
moreover , these asymptotic forms of the wavefunctions may or may not be plane - waves depending on whether the potential converges to zero or not at the asymptotic distance .
this dual asymptotic behaviour of a single wave - solution on two sides of the potential barrier which is absent in the piece - wise constant / continuous potentials is worthwhile to demonstrate through a simple do - able example
. it will be even more valuable if one can extract simple forms of @xmath1 and @xmath2 analytically . in this paper
, we present a smooth single - piece potential ( see fig.1 ) : @xmath28 which vanishes asymptotically as @xmath27 but it diverges to @xmath29 as @xmath30 .
we feel that the attractive exponential potential ( 1 ) presents a simple instance containing the essential features of analytic extraction of transmission co - efficient for a one piece smooth potential . in
the following we proceed to find @xmath1 and discuss various aspects of schrdinger transmission arising thereby .
the time - independent schrdinger equation with ( 1 ) is written as @xmath31\psi(x)=0,\ ] ] where @xmath32 and @xmath33 denote the mass and energy of the incident particle and @xmath34 is the planck - constant .
the time - dependent wave function is written as @xmath35 let us substitute @xmath36 in eq .
( 2 ) and write @xmath37 where @xmath38 and @xmath39 .
the transformed equation is the standard cylindrical bessel equation [ 11 ] whose linearly independent solutions for any positive energy are the bessel functions : @xmath40 and @xmath41 or the henkel functions : @xmath42 and @xmath43 . noting the asymptotic property that @xmath44 as @xmath45 is large for large positive values of @xmath15 , we admit @xmath46 @xmath47,~~\mbox{when}~x\rightarrow \infty,\ ] ] @xmath48.\ ] ] this represents a wave which is exponentially decaying yet travelling towards @xmath30 .
the condition of constancy of the phase i.e. , @xmath49 yields positive phase velocity : @xmath50 which is positive definite and does not change sign and hence the direction of the wave is unchanged .
thus eq .
( 6a ) represents a transmitted wavefunction , @xmath51 .
the other degenerate and linearly independent solution of the schrdinger equation ( 2 ) , @xmath43 has been dropped .
it is instructive to note that a linear combination of these solutions can not produce the boundary condition of a transmitted wave , traveling towards @xmath30 .
next , we note valuable identities : @xmath52/\sin(\nu \pi ) , \mbox{and } h^{(2)}_\nu(z)=i[j_{\nu}(z)-e^{i \pi \nu}j_{-\nu}(z)]/\sin(\nu \pi).\ ] ] we can also write eq .
( 6a ) as @xmath53 e^{-iet \over \hbar}.\ ] ] the asymptotic behaviour of bessel function for small @xmath54-values is given by [ 11 ] @xmath55}.\ ] ] consequently , we can write by combining eqs .
( 8) and ( 9 ) @xmath56 } e^{i(kx - et/\hbar)}- { ( p/2)^{-iq } \over \sinh q\pi \gamma[1-iq]}e^{-i(kx+e t/\hbar)},~~ x \rightarrow -\infty.\ ] ] when @xmath27 , @xmath54 tends to a very small value , for the result ( 9 ) to hold .
the right side of the above equation represents a linear combination of two oppositely travelling plane waves .
the phase - velocities of the first and second part are @xmath57 and @xmath58 , respectively .
since the transmitted wave has been identified to have positive phase - velocity travelling from left to right so as to leave the barrier , the first part represents the incident wave , @xmath59 travelling from left to right impinging upon the barrier .
similarly , the second part represents a reflected wave , @xmath60 , travelling from right to left .
it is instructive to note here that the asymptotic form of @xmath61 as @xmath27 is a linear combination of the plane - waves since the potential ( 1 ) actually vanishes at a large distance on the left side of the barrier ( see fig .
1 ) .
the wave - function @xmath61 in eqs .
( 8) and ( 10 ) are nothing but exact and asymptotic forms of the sum of incident wave , @xmath59 , and reflected wave , @xmath60 , wavefunctions , respectively .
having found the incident , reflected and transmitted wavefunctions , we now find the flux , @xmath62 arising due to @xmath63 .
yet another instructive point is in order here .
the flux can also be written as a wronskian : @xmath64 $ ] .
since @xmath65 and @xmath66 are the linearly independent solution of ( 2 ) , the wronskian , @xmath67 $ ] , will be constant ( independent of @xmath15 ) of motion ( see e.g. , p. 48 in [ 2 ] ) . therefore , if the ( approximate ) asymptotic forms of @xmath68 in eqs .
( 6b ) and ( 10 ) are used they will yield the same value of flux @xmath69 as the exact wavefunctions will do .
one can check this point for the present problem by recalling the wronskians [ 11 ] , @xmath70=-{2\sin\nu \pi \over \pi z},~ w[h^{(1)}_{\nu}(z),h^{(2)}(z)_{\nu}]=-{4i \over \pi z}~\ ] ] and @xmath71^\ast=- e^{\pi q } h^{(2)}_{iq } ( z).\ ] ] thus , the various fluxes work out to be @xmath72 in obtaining the above results we have used the formula : @xmath73 [ 11 ] .
finally , the transmission and reflection co - efficients are found as @xmath74,~r(e)={{\cal j}_r \over { \cal j}_i}=e^{-2\pi q},~ q=\sqrt { e\over \delta},~~\delta={\hbar^2 \over 8ma^2}.\ ] ] these co - efficients satisfy @xmath4 .
let us find the special values of @xmath1 and @xmath2 for consistency .
we get @xmath75 ; @xmath76 which are physically feasible . interestingly
, these co - efficients ( 14 ) do not depend up on the parameter , @xmath16 , and rightly so . since any finite positive value of @xmath16 is expressible as @xmath77 , and it would merely result in shifting the whole potential on the @xmath78axis by a distance @xmath79 : @xmath80 . in mathematical terms , when @xmath81 , the classical limits are obtained as @xmath82 , and @xmath83 from eq .
( 14 ) .
however , practically the classical limit is reached when an effective parameter , for example @xmath84 ( 14 ) , becomes extremely small .
this in turn is materialized when the tunneling system becomes massive or / and a length scale found in the interaction ( e.g. , @xmath85 here ) becomes extremely large . in the cases [ 3 - 10 ] , where a potential offers a barrier of a height , @xmath16 , the classical limit of @xmath1 is a step function of energy : @xmath86 .
since the exponential potential ( 1 ) does not offer any barrier to an incident particle in a classical sense , we have got @xmath82 .
let us divide eqs .
( 10 ) and ( 6b ) by the co - efficent of @xmath87 in ( 10 ) to re - write @xmath61 respectively as @xmath88 \over \gamma[1-iq]}\right ) e^{-i(kx+e t/\hbar ) } , ~~ x \rightarrow -\infty,\ ] ] @xmath89 e^{-q\pi/2 } \right ) e^{-{x \over 4a } } \exp [ i(pe ^{x \over 2a}-e t/\hbar)],~~\mbox{when}~x\rightarrow \infty.\ ] ] the multiplicative factor of @xmath90 in eq .
( 15a ) is called as reflection amplitude , @xmath91 and the factor written in big brackets in eq .
( 15b ) is called transmission amplitude .
@xmath92 generally , @xmath93 , however , @xmath94 holds only if the potential converges to the same asymptotic value on both the sides of the barrier .
since here it is not the case ( see fig .
1 ) , the expression of @xmath1 as obtained in eq .
( 14 ) by using the ratio of transmitted to incident flux , is more fundamental and one finds that @xmath95 is proportional to ( if not identical ) to @xmath1 .
let us define arg@xmath96=\alpha$ ] and arg@xmath97=\beta$ ] .
for the exponential potential barrier , we get @xmath98 when a particle is transmitted through a potential from one side to the other , it sees the total potential in every detail .
eventually , detailed balancing takes place to give rise to reciprocity of transmission : it does not matter if the particle is incident from left of right the transmission amplitude remains the same ( invariant ) , i.e. , @xmath99.$ ] this implies the invariance of both transmission probability i.e. , @xmath100 and transmission phase - shift i.e. , @xmath101 .
further , due to the conservation of flux invariance of reflection probability also takes place , namely , @xmath102 the behaviour of @xmath103 with respect to the side of the incidence the particle is known ( p. 109
[ 1 ] ) as .
@xmath104 = ( 2n+1)\pi-[\phi_{right}(e)-\theta_{right}(e)],~n=0,1,2, ... \ ] ] interestingly , it turns out that if the potential is real and symmetric , we have @xmath105 [ 12 ] , and hence @xmath106 .
the present potential being non - symmetric provides an opportunity for a simple pedagogic demonstration of the result in eq . (
18 ) and the invariance of @xmath107 and @xmath2 with respect to the side from which the particle is incident on the potential . in the following ,
we proceed to repeat the calculations when the side of the incidence has been changed from left to right ( see fig .
1(b ) ) . using the standard relations ( 7 )
: we find that @xmath108 this identity suggests that @xmath109 which in turn has an analytic behaviour for @xmath110 , @xmath111 } e^{-i(kx+et/\hbar)}.\ ] ] this represents a wave travelling to the left after being transmitted from the potential ( see fig .
1(b ) ) . using eq.(9 ) and the asymptotic properties of @xmath112 as in eq .
( 5 ) , we find that @xmath113 in the asymptotic form of the wave function ( 21 ) , the first part denotes the incident wave , @xmath59 , entering from the right side of the potential ( see fig.1(b ) ) and travelling to the left .
the second part denotes oppositely travelling reflected wave , @xmath60 . the transmitted wave , @xmath51 , is represented by eq.(19 ) .
one can recover , the results found in eq .
( 14 ) by evaluating the flux arising due to these wave - forms when the side of incidence has been changed .
the recovery of @xmath1 and @xmath2 despite the change of the side of the incidence of the particle on the potential ( see fig .
1(b ) ) demonstrates their invariance with respect to the side of incidence of the particle whether it is left or right .
once again , let us divide eqs .
( 19b ) and ( 20 ) by the co - efficient written outside the big bracket in eq .
( 20 ) to get @xmath114 } e^{-i(kx+et/\hbar ) } , ~~ \mbox { when}~~x\rightarrow -\infty,\ ] ] @xmath115 the amplitudes can be extracted as earlier : @xmath116}.\ ] ] further , we find the phase - shifts as @xmath117 to compare with phase - shifts given in eq .
( 17 ) , and this completes the demonstration of the general results given in eq .
( 18 ) .
it is further relevant to know that when the potential is complex , the reciprocity of the transmission amplitude with respect to the side of the incident particle still holds .
however , reflection and absorption amplitudes and co - efficients depend on the side ( left / right ) of incident particle , if only , the total complex potential is non - symmetric , namely @xmath118 [ 13 ] .
the utility of the schrdinger transmission in many physical processes can be appreciated as follows .
look at one body , @xmath119 , at rest in fig .
1(b ) and let other body , @xmath120 , come from a side with an energy , @xmath3 .
assume that the interactions between @xmath119 and @xmath120 are of two types : short - ranged and long - ranged .
thus , the interaction barrier ( e.g. eq.1(b ) ) due to long - ranged interaction has to be overcome by the body @xmath120 in order to appear on the other side to have any worthwhile contact / reaction with @xmath119 , characterized by their short - ranged interaction .
the probability of the appearance of the body , @xmath120 , on the other side is clearly proportional to the probability of transmission through the potential barrier .
this suggests that the reaction rates of @xmath119 and @xmath120 , would depend on the transmission probability , @xmath1 , apart from other factors .
in fact , barrier penetration method of determining the fusion rates ( see [ 10 ] and refs . therein ) of nuclei is one such instance where fusion rates are determined by finding the penetrability @xmath121 of the inter - nuclear potential barrier formed due to short - ranged nuclear attraction and coulomb plus centrifugal repulsion . in other kind of instances , a bound system ( e.g. @xmath122-particles in a nucleus [ 14 ] or electrons of a metal in the field emission [ 15 ] )
faces a barrier and the emission rates are then determined by the transmission co - efficient of the barrier .
the analytically solvable , smooth and single - piece potential barriers mentioned in the introduction find applications in several physical processes .
the extraction of transmission co - efficient for these barriers involves higher order functions . in this regard ,
the exponential barrier ( 1 ) presented here , despite being simple allows one to have the actual experience of extracting transmission ( reflection ) amplitude / co - efficent from a smooth , single - piece potential analytically .
this problem may be practised right after doing rectangular well / barrier and before attempting the `` higher - order '' problems found in refs .
[ 3 - 10 ] .
the exponential potential model also illustrates , with simplicity , several other aspects of schrdinger transmission from one - dimensional , real potentials . |
the su(2)@xmath4u(1)@xmath5 structure of the electroweak interactions has been consolidated by an enormous wealth of experimental data during the past three decades .
the canonical way to generate masses for the fermions and intermediate bosons without violating this gauge symmetry in the lagrangian is by the higgs mechanism of spontaneous symmetry breaking . in the minimal standard model ( sm ) , this is achieved by introducing one complex su(2)@xmath6-doublet scalar field @xmath7 with @xmath8 .
the three massless goldstone bosons which emerge via the electroweak symmetry breaking are eaten up to become the longitudinal degrees of freedom of the @xmath9 and @xmath10 bosons , _
i.e. _ , to generate their masses , while one @xmath11-even higgs scalar boson @xmath12 remains in the physical spectrum .
the higgs potential @xmath13 contains one mass and one self - coupling .
since the vacuum expectation value is fixed by the relation @xmath14 gev , where @xmath15 is fermi s constant , there remains one free parameter in the higgs sector , namely @xmath16 .
in fact , one has @xmath17 where @xmath18 .
the higgs boson has the quantum numbers of the vacuum , namely electric charge @xmath19 , spin , parity , and charge conjugation @xmath20 .
it has tree - level couplings to all massive particles with strengths that are determined by their masses , viz .
@xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 , where @xmath26 denotes a generic fermion and @xmath27 . at a future @xmath0 linear collider ( lc ) , an important experimental task will be to determine of the higgs quantum numbers and couplings in order to distinguish between the minimal sm and possible extensions .
in particular , the measurement of the higgs self - couplings will allow one to directly test the higgs mechanism .
roughly speaking , the requirement that the running higgs self - coupling @xmath28 , where @xmath29 is the renormalization scale , stays finite ( positive ) for all values @xmath30 , where @xmath31 is the cutoff beyond which new physics operates , leads to the triviality upper bound ( vacuum - stability lower bound ) on @xmath16 @xcite .
assuming the sm to be valid up to the grand - unified - theory scale @xmath32 gev , one thus obtains @xmath33 gev @xcite [ see fig . [
fig : blue](a ) ] .
this range comfortably lies between the lower bound on @xmath16 from direct searches at cern lep2 , 113 gev , and the 95% confidence level upper bound from electroweak precision tests @xcite , 212 gev , based on @xmath34 @xcite , and it is compatible with the @xmath35 range @xmath36 gev resulting from the latter @xcite [ see fig . [
fig : blue](b ) ] . [ cols="^,^ " , ]
we reviewed theoretical results that are relevant for the phenomenology of the sm higgs boson at a future @xmath0 lc , putting special emphasis on radiative corrections to its partial decay widths and production cross sections , and on the logistics of extracting its quantum numbers and couplings from the analysis of appropriate final states .
it is fair to say that theoretical predictions for partial decay widths and production cross sections are generally in good shape .
however , the precision on the partial decay widths is limited by parametric uncertainties , mainly by those in @xmath37 and the quark masses .
the strategies for the determination of the higgs profile are also well elaborated .
the list of urgent tasks left to be done includes the calculation of the full @xmath38 corrections for important @xmath39 processes , such as @xmath40 fusion , @xmath41 fusion , and @xmath42 associated production , and the inclusion of background processes and detector simulation .
we thank matthias steinhauser for carefully reading this manuscript .
this work was supported in part by the deutsche forschungsgemeinschaft through grant no .
kn 365/1 - 1 , by the bundesministerium fr bildung und forschung through grant no .
05 ht1gua/4 , and by the european commission through the research training network _
quantum chromodynamics and the deep structure of elementary particles _ under contract no .
erbfmrx - ct98 - 0194 .
the lep collaborations aleph , delphi , l3 , opal , the lep electroweak working group and the sld heavy flavour and electroweak groups , d. abbaneo _ et al .
_ , report no . lepewwg/2001 - 01 , aleph 2001 - 041phys 2001 - 015 , delphi 2001 - 109 phys 897 , l3 note 2669 , and opal technical note tn 692 ( 31 may 2001 ) .
m. davier and a. hcker , phys .
b * 435 * , 427 ( 1998 ) ; j. h. khn and m. steinhauser , phys . lett .
b * 437 * , 425 ( 1998 ) ; s. groote , j. g. krner , k. schilcher , n. f. nasrallah , phys .
b * 440 * , 375 ( 1998 ) ; j. erler , phys .
d * 59 * , 054008 ( 1999 ) ; a. d. martin , j. outhwaite , and m. g. ryskin , phys .
b * 492 * , 69 ( 2000 ) .
r. n. cahn , m. s. chanowitz , and n. fleishon , phys .
lett . * 82b * , 113 ( 1979 ) ; l. bergstrm and g. hulth , nucl .
* b259 * , 137 ( 1985 ) ; * b276 * , 744(e ) ( 1986 ) ; a. barroso , j. pulido , and j. c. romo , nucl .
* b267 * , 509 ( 1986 ) ; j. c. romo and a. barroso , nucl .
phys . * b272 * , 693 ( 1986 ) ; j. f. gunion , g. l. kane , and j. wudka , nucl .
b299 * , 231 ( 1988 ) .
f. wilczek , phys .
39 * , 1304 ( 1977 ) ; h. m. georgi , s. l. glashow , m. e. machacek , and d. v. nanopoulos , phys .
lett . * 40 * , 692 ( 1978 ) ; j. r. ellis , m. k. gaillard , d. v. nanopoulos , and c. t. sachrajda , phys .
* 83b * , 339 ( 1979 ) ; t. g. rizzo , phys .
d * 22 * , 178 ( 1980 ) ; * 22 * , 1824(a ) ( 1980 ) .
d. yu . bardin , b. m. vilenski , p. kh .
khristova , yad .
* 53 * , 240 ( 1991 ) [ sov . j. nucl .
* 53 * , 152 ( 1991 ) ] ; b. a. kniehl , nucl .
phys . * b376 * , 3 ( 1992 ) ; a. dabelstein and w. hollik , z. phys .
c * 53 * , 507 ( 1992 ) .
e. braaten and j. p. leveille , phys .
d * 22 * , 715 ( 1980 ) ; n. sakai , phys .
d * 22 * , 2220 ( 1980 ) ; t. inami and t. kubota , nucl .
* b179 * , 171 ( 1981 ) ; m. drees and k. hikasa , phys .
b * 240 * , 455 ( 1990 ) ; * 262 * , 497(e ) ( 1991 ) .
s. g. gorishny , a. l. kataev , s. a. larin , and l. r. surguladze , mod .
lett . a * 5 * , 2703 ( 1990 ) ; phys . rev .
d * 43 * , 1633 ( 1991 ) ; a. l. kataev and v. t. kim , mod .
a * 9 * , 1309 ( 1994 ) ; l. r. surguladze , phys . lett .
b * 341 * , 60 ( 1994 ) ; b. a. kniehl , phys .
b * 343 * , 299 ( 1995 ) ; k. g. chetyrkin and a. kwiatkowski , nucl .
* b461 * , 3 ( 1996 ) .
h. zheng and d. wu , phys .
d * 42 * , 3760 ( 1990 ) ; a. djouadi , m. spira , j. j. van der bij , and p. m. zerwas , phys .
b * 257 * , 187 ( 1991 ) ; s. dawson and r. p. kauffman , phys .
d * 47 * , 1264 ( 1993 ) ; a. djouadi , m. spira , and p. m. zerwas , phys .
b * 311 * , 255 ( 1993 ) ; k. melnikov and o. i. yakovlev , phys .
b * 312 * , 179 ( 1993 ) ; m. inoue , r. najima , t. oka , and j. saito , mod .
lett . a * 9 * , 1189 ( 1994 ) ; j. fleischer and o. v. tarasov , z. phys .
c * 64 * , 413 ( 1994 ) .
m. steinhauser , in _ proceedings of the ringberg workshop : the higgs puzzle what can we learn from lep2 , lhc , nlc and fmc ?
_ , ringberg castle , germany , 1996 , edited by b. a. kniehl ( world scientific , singapore , 1997 ) , p. 177 .
a. i. vanshten , m. b. voloshin , v. i. zakharov , and m. a. shifman , yad .
* 30 * , 1368 ( 1979 ) [ sov . j. nucl .
* 30 * , 711 ( 1979 ) ] ; a. i. vanshten , v. i. zakharov , and m. a. shifman , usp .
nauk * 131 * , 537 ( 1980 ) [ sov . phys .
* 23 * , 429 ( 1980 ) ] ; m. b. voloshin , yad .
fiz . * 44 * , 738 ( 1986 ) [ sov .
. phys . *
44 * , 478 ( 1986 ) ] ; m. a. shifman , usp .
nauk * 157 * , 561 ( 1989 ) [ sov . phys .
* 32 * , 289 ( 1989 ) ] ; b. a. kniehl and m. spira , z. phys .
c * 69 * , 77 ( 1995 ) ; w. kilian , z. phys .
c * 69 * , 89 ( 1995 ) ; m. spira , a. djouadi , d. graudenz , and p. m. zerwas , nucl .
* b453 * , 17 ( 1995 ) .
a. djouadi and c. verzegnassi , phys .
b * 195 * , 265 ( 1987 ) ; a. djouadi , nuovo cim .
* 100a * , 357 ( 1988 ) ; b. a. kniehl , j. h. khn , and r. g. stuart , phys .
b * 214 * , 621 ( 1988 ) ; b. a. kniehl , nucl .
phys . * b347 * , 86 ( 1990 ) ; a. djouadi and p. gambino , phys .
d * 49 * , 3499 ( 1994 ) ; * 53 * , 4111(e ) ( 1996 ) ; * 49 * , 4705 ( 1994 ) .
r. barbieri , m. beccaria , p. ciafaloni , g. curci , and a. vicer , phys .
b * 288 * , 95 ( 1992 ) ; * 312 * , 511(e ) ( 1993 ) ; nucl .
* b409 * , 105 ( 1993 ) ; j. fleischer , o. v. tarasov , and f. jegerlehner , phys .
b * 319 * , 249 ( 1993 ) .
g. degrassi , s. fanchiotti , and p. gambino , int .
phys . a * 10 * , 1377 ( 1995 ) ; g. degrassi , s. fanchiotti , f. feruglio , p. gambino , and a. vicini , phys .
b * 350 * , 75 ( 1995 ) ; g. degrassi , p. gambino , and a. vicini , phys .
b * 383 * , 219 ( 1996 ) .
j. m. cornwall , d. n. levin , and g. tiktopoulos , phys .
d * 10 * , 1145 ( 1974 ) ; * 11 * , 972(e ) ( 1975 ) ; c. e. vayonakis , lett .
nuovo cimento * 17 * , 383 ( 1976 ) ; m. s. chanowitz and m. k. gaillard , nucl .
* b261 * , 379 ( 1985 ) ; g. j. gounaris , r. kgerler , and h. neufeld , phys .
d * 34 * , 3257 ( 1986 ) ; y .- p .
yao and c .- p .
yuan , phys .
d * 38 * , 2237 ( 1988 ) ; j. bagger and c. schmidt , phys .
d * 41 * , 264 ( 1990 ) ; h. veltman , phys .
d * 41 * , 2294 ( 1990 ) ; h .- j .
he , y .- p .
kuang , and x. li , phys .
* 69 * , 2619 ( 1992 ) ; phys .
d * 49 * , 4842 ( 1994 ) ; h .- j .
he , y .- p .
kuang , and c .- p .
yuan , phys .
d * 51 * , 6463 ( 1995 ) ; d. espriu and j. matias , phys .
d * 52 * , 6530 ( 1995 ) .
l. durand , b. a. kniehl , and k. riesselmann , phys .
rev . lett . * 72 * , 2534 ( 1994 ) ; * 74 * , 1699(e ) ( 1995 ) ; phys . rev .
d * 51 * , 5007 ( 1995 ) ; a. ghinculov , phys .
b * 337 * , 137 ( 1994 ) ; * 346 * , 426(e ) ( 1995 ) ; v. borodulin and g. jikia , phys . lett .
b * 391 * , 434 ( 1997 ) .
j. a. aguilar - saavedra _ et al .
_ , in _ tesla technical design report , part iii : physics at an @xmath0 linear collider _ , edited by r .- d .
heuer , d. miller , f. richard , and p. zerwas , report no .
desy 2001 - 011 , ecfa 2001 - 209 , tesla report 2001 - 23 , tesla - fel 2001 - 05 ( march 2001 ) , p. 11 .
v. telnov , in _ physics and experiments with linear colliders _ , saariselk , finland , 914 september 1991 , edited by r. orava , p. eerola , and m. nordberg ( world scientific , singapore , 1992 ) , vol .
ii , p. 739 ; d. bowser - chao and k. cheung , phys .
d * 48 * , 89 ( 1993 ) ; o. j. p. boli , m. c. gonzalez - garcia , f. halzen , and d. zeppenfeld , phys . rev .
d * 48 * , 1430 ( 1993 ) ; d. l. borden , d. a. bauer , and d. o. caldwell , phys . rev .
d * 48 * , 4018 ( 1993 ) ; h. veltman , z. phys .
c * 62 * , 235 ( 1994 ) .
a. djouadi and b. a. kniehl , in _ proceedings of the workshop on @xmath0 collisions at 500 gev : the physics potential _
, part c , munich , annecy , hamburg , november 1992 to april 1993 , edited by p. m. zerwas , report no .
desy 93 - 123c ( december 1993 ) , p. 51 | the higgs boson is the missing link of the standard model of elementary particle physics .
we review its decay properties and production mechanisms at a future @xmath0 linear collider and its @xmath1 , @xmath2 , and @xmath3 modes , with special emphasis on the influence of quantum corrections .
we also discuss how its quantum numbers and couplings can be extracted from the study of appropriate final states .
16.cm 22.cm @=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta @=12 |
in a parameterized problem , we are given an instance @xmath9 , where @xmath3 is a secondary measurement , called as the _
parameter_. the central question in parameterized complexity is whether a parameterized problem admits an algorithm with running time @xmath10 , called a _ fixed parameter tractable algorithm _ ( shortly , an _ fpt algorithm _ ) , where @xmath5 is a function depending on the parameter @xmath3 alone and @xmath11 is the input size . a parameterized problem admitting such an algorithm is said to be _ fixed - parameter tractable _ , or _ fpt _ in short . as we study a parameterized problem when its unparameterized decision version is np - hard , the function @xmath5 is super - polynomial in general . for many natural parameterized problems ,
the function @xmath5 is overwhelming @xcite or even non - explicit @xcite , especially when the algorithm is indicated by a meta - theorem .
therefore , a lot of research effort focus on designing an fpt algorithm with affordable super - exponential part in the running time .
we are especially interested in solving a parameterized problem in _ single - exponential _ fpt time , that is , in time @xmath12 for some constant @xmath13 .
one of techniques to handle parameterized problems is the _
kernelization algorithm_. a kernelization algorithm takes an instance @xmath14 and outputs an instance @xmath15 in time polynomial in @xmath16 satisfying that ( 1 ) @xmath9 is a yes - instance if and only if @xmath17 is a yes - instance , ( 2 ) @xmath18 , and @xmath19 for some function @xmath20 . the reduced instance is called a _ kernel _ and the function @xmath20 is called the _ size _ of the kernel .
a parameterized problem is said to admit a _ polynomial kernel _ if there is a kernelization algorithm that reduces the input instance into an instance with size bounded by a polynomial function @xmath21 in @xmath3 .
graph modification problems are typically formulated as follows : given an input graph @xmath2 and a fixed set @xmath22 of elementary operations and a graph property @xmath23 , the objective is to transform @xmath2 into a graph @xmath24 by applying at most @xmath3 operations from @xmath22 .
vertex deletion , edge deletion / addition or contraction are examples of such elementary operations .
the graph property @xmath23 having treewidth or pathwidth at most @xmath25 has received in - depth attention as many problems become tractable on graphs of small treewidth .
the celebrated courcelle s theorem @xcite implies that every graph property expressible in monadic second order logic of the second type ( @xmath26 ) can be verified in time @xmath27 , when the input @xmath1-vertex graph has treewidth at most @xmath25 .
furthermore , having small treewidth frequently facilitates the design of a dynamic programming algorithm whose running time is much faster than that of the all - round algorithm from the courcelle s meta - theorem .
therefore , it is reasonable to measure how close an instance is from `` an island of tractability within an ocean of intractable problems '' @xcite . in the context of treewidth
, the vertex deletion problems for the special cases of @xmath28 and @xmath29 correspond to the well - known vertex cover and feedback vertex set problems respectively .
generally , for fixed @xmath25 , the treewidth-@xmath25 vertex deletion can be solved in time @xmath30 implied by courcelle s meta - theorem @xcite .
as the function @xmath5 subsumed in the meta theorem is gigantic , it is natural to ask whether the exponential function in the running time can be rendered realistic .
recent endeavor pursuing this question culminated in establishing that for fixed @xmath25 , the treewidth-@xmath25 vertex deletion can be solved in single - exponential fpt time @xcite . as for pathwidth
, pathwidth-@xmath0 vertex deletion was first studied by philip , raman , villanger @xcite , and later cygan , pilipczuk , pilipczuk , wojtaszczyk @xcite showed that pathwidth-1 vertex deletion can be solved in time @xmath31 and it admits a quadratic kernel . using the general method developed for treewidth-@xmath25 vertex deletion @xcite
, the pathwidth-@xmath25 vertex deletion problem also admits a single - exponential fpt algorithm .
[ [ linear - rankwidth . ] ] linear rankwidth .
_ rankwidth _ was introduced by oum and seymour @xcite for efficiently approximating _
cliquewidth_. compared to cliquewidth , there are some containment relations , called _ vertex - minors _ and _ pivot - minors _
@xcite , where the rankwidth of a graph does not increase when taking those relations . with these relations ,
rankwidth has been intensively studied to extend results for treewidth and graph minors @xcite .
_ linear rankwidth _ is a linearized variation of rankwidth as pathwidth is the linearized variant of treewidth .
while treewidth and pathwidth are small only on sparse graphs , dense graphs may have small rankwidth or linear rankwidth .
for instance , complete graphs , complete bipartite graphs , and threshold graphs @xcite have linear rankwidth at most @xmath0 even though they have unbounded treewidth .
linear rankwidth is deeply related to _
matroid pathwidth _ , also known as _ trellis - width _ , introduced by kashyap @xcite .
matroid pathwidth has been studied in some matroid theory literature @xcite .
kashyap @xcite showed that it is np - hard to compute the matroid pathwidth of a binary matroid given with its matrix representation . from the relation between a binary matroid and its fundamental graph due to oum @xcite
, one can also deduce that it is np - hard to compute the linear rankwidth of a graph .
recently , jeong , kim , and oum @xcite showed that for fixed @xmath3 , there is a cubic - time algorithm to test whether an input graph has linear rankwidth at most @xmath3 or not , and output such an ordering if one exists .
ganian @xcite pointed out that some np - hard problems , such as computing pathwidth , can be solved in polynomial time on graphs of linear rankwidth at most @xmath0 .
generally , the meta - theorem by courcelle , makowsky , and rotics @xcite states that for every graph property @xmath23 expressible in monadic second order logic of the first type ( @xmath32 ) and fixed @xmath3 , there is a cubic - time algorithm for testing whether a graph of rankwidth at most @xmath3 has property @xmath23 .
as rankwidth is always less than or equal to linear rankwidth , those problems are tractable on graphs of bounded linear rankwidth as well .
in the same context , it is natural to ask whether there is an fpt algorithm for the ( linear ) rankwidth-@xmath25 vertex deletion problem , that is a problem asking whether for a given graph @xmath2 and a positive integer @xmath3 , @xmath2 contains a vertex subset of size at most @xmath3 whose deletion makes @xmath2 a graph of ( linear ) rankwidth at most @xmath25 .
it is only known that for fixed @xmath25 , both problems are fpt from the meta - theorem on graphs of bounded rankwidth @xcite .
we discuss it in more detail in section [ sec : remark ] .
however , as the function of @xmath3 obtained from the meta - theorem is enormous , it is interesting to know whether there is a single - exponential fpt algorithm for both problems , like treewidth-@xmath25 vertex deletion .
also , to the best of our knowledge , there was no known previous result whether the ( linear ) rankwidth-@xmath25 vertex deletion problem admits a polynomial kernel for fixed integer @xmath25 .
[ [ our - contributions . ] ] our contributions . in this paper
, we show that the linear rankwidth-@xmath0 vertex deletion problem admits a single - exponential fpt algorithm and a polynomial kernel .
this is a first step towards a goal of investigating whether the ( linear ) rankwidth-@xmath25 vertex deletion problem admits a single - exponential fpt algorithm or has a polynomial kernel .
[ thm : main1 ] the lrw1-vertex deletion problem can be solved in time @xmath33 , and also can be solved in time @xmath7 .
[ thm : main2 ] the lrw1-vertex deletion problem has a kernel with @xmath34 vertices .
we note that several graph classes with a certain path - like structure have been studied for parameterized vertex deletion problems .
such classes include graphs of pathwidth-@xmath0 @xcite , proper interval graphs @xcite , unit interval graphs @xcite , and interval graphs @xcite . a common approach in the previous work was to use the characterization of the structures obtained after removing small obstructions .
we also characterize graphs excluding small obstructions for graphs of linear rankwidth at most @xmath0 .
we investigate a new class of graphs , called _ necklace graphs _ , which are close to graphs of linear rankwidth at most @xmath0 .
briefly speaking , necklace graphs , when viewed locally , are graphs of linear rankwidth at most @xmath0 , but they may have long induced cycles . in section [ sec : necklacegraph ] , we show that every connected graph having no obstructions of size at most @xmath35 for graphs of linear rankwidth at most @xmath0 is either a graph of linear rankwidth at most @xmath0 or a necklace graph ( theorem [ thm : mainlrw ] ) . combining a simple branching algorithm and a polynomial - time algorithm to find a minimum deleting set on necklace graphs , we obtain an fpt algorithm for lrw1-vertex deletion with running time @xmath36 in the beginning of section [ sec : fptthreaddel ] .
one might ask whether the polynomial factor @xmath37 can be reduced .
this running time appears as we start with finding obstructions of size at most @xmath35 .
indeed , we can improve it using a dynamic programming algorithm to find an induced subgraph of fixed size in a graph of bounded cliquewidth .
if the rankwidth of a given graph is more than @xmath38 , then the instance is trivially a no - instance because rankwidth can be decreased by at most @xmath0 when removing a vertex . using the approximation algorithm due to oum @xcite
, we can decide whether a given graph has rankwidth at most @xmath38 and if so , outputs a rank - decomposition of width at most @xmath39 and also a @xmath40-cliquewidth expression , in time @xmath41 .
then we develop a branching algorithm using the cliquewidth expression , and finally achieve an fpt algorithm with running time @xmath41 .
in section [ sec : lowerbound ] , we prove that the running time of our algorithms can not be reduced to @xmath8 under a reasonable assumption .
there is no @xmath8-time algorithm for lrw1 vertex deletion , unless exponential time hypothesis ( eth ) fails . in section [ sec : polykerthreaddel ]
, we obtain a polynomial kernel for the lrw1-vertex deletion problem .
we start with hitting obstructions of size at most @xmath35 using the sunflower lemma , and taking a minimum deleting set on the remaining necklace graph .
the union of two sets will have size bounded by a polynomial function in @xmath3 , and its removal makes an input graph into a graph of linear rankwidth at most @xmath0 .
graphs of linear rankwidth at most @xmath0 can be seen as graphs obtained by connecting certain blocks , called _
thread blocks _ , like a path ( theorem [ thm : structurethread ] ) .
the main difficulty for reducing the remaining part is to shrink a large thread block , and we can resolve this issue using the set obtained by the sunflower lemma .
we remark that a similar idea was used by fomin , saurabh , and villanger @xcite to obtain a polynomial kernel for the proper interval vertex deletion problem .
we conclude the paper with further discussions in section [ sec : remark ] .
in this paper , all graphs are finite and undirected , if not mentioned . for a graph @xmath2 ,
we denote by @xmath42 and @xmath43 the vertex set and the edge set of a graph @xmath2 , respectively .
let @xmath2 be a graph .
for @xmath44 , let @xmath45 denote the neighborhood of @xmath46 .
let @xmath47 be a subset of @xmath42 .
we denote by @xmath48 $ ] the subgraph of @xmath2 induced on @xmath47 and we define @xmath49 $ ] . for short
we write @xmath50 instead of @xmath51 for @xmath44 . a vertex @xmath52 of @xmath2
is called a _ pendant vertex _ if @xmath53 .
the subset of vertices @xmath54 is the set of all vertices of @xmath47 that have a neighbor in @xmath55 .
a graph @xmath56 is an _ induced subgraph _ of a graph @xmath2 if @xmath57 $ ] for some @xmath58 . for a set @xmath59 of graphs ,
a graph @xmath2 is _ @xmath59-free _ if @xmath2 has no induced subgraph isomorphic to a graph in @xmath59 .
a vertex @xmath52 of @xmath2 is called a _ cut vertex _ if the removal of @xmath52 from @xmath2 strictly increases the number of connected components . a maximal connected subgraph of a graph without a cut vertex
is called a _ block_. note than an edge can be a block .
the path on the vertex set @xmath60 and the edge set @xmath61 will be denoted by @xmath62 , and the cycle on the vertex set @xmath60 and the edge set @xmath63 will be denoted by @xmath64 .
the length of a path is defined as the number of edges in the path . for @xmath65 , we denote by @xmath66 the chordless cycle of length @xmath1 .
a graph @xmath2 is _ distance - hereditary _ if for every connected induced subgraph @xmath56 of @xmath2 and @xmath67 , the distance between @xmath52 and @xmath25 in @xmath56 is the same as the distance between @xmath52 and @xmath25 in @xmath2 .
for instance the cycle @xmath68 of length @xmath69 is not distance - hereditary , as @xmath70 and @xmath71 have distance @xmath72 in the graph , but they have distance @xmath73 in @xmath74 .
a _ star _ is a tree with a distinguished vertex adjacent to all other vertices .
complete graph _ is a graph with all possible edges .
an ordering on a finite set @xmath47 is a bijective mapping @xmath75 , and we write @xmath76 if @xmath77 , and @xmath78 as the inverse bijective mapping . for an @xmath79-matrix @xmath80 and @xmath81 , let @xmath82 $ ] be the submatrix of @xmath80 whose rows and columns are indexed by @xmath83 and @xmath84 , respectively .
when we analyze the running time of an algorithm , we agree that @xmath85 and @xmath86 if @xmath2 is an input graph .
the _ adjacency matrix _ of a graph @xmath2 , which is a @xmath87-matrix over the binary field , will be denoted by @xmath88 .
the _ width _ of an ordering @xmath89 of the vertex set of a graph @xmath2 is @xmath90),\ ] ] where the rank of a matrix is computed over the binary field .
the _ linear rankwidth _ of a graph @xmath2 is defined as the minimum width over all orderings of @xmath42 .
ganian @xcite first characterized graphs of linear rankwidth at most @xmath0 , and he called them _
thread graphs_. later , adler , farley , and proskurowski @xcite gave an easier way to define a thread graph , using a notion of _ thread blocks_. we follow the definition of thread blocks given by adler , farley , and proskurowski , and provide a unified way to define classes of graphs including thread graphs .
a graph @xmath2 with at least @xmath72 vertices is called a _ thread block _ if there is an ordering @xmath89 of @xmath42 and a function @xmath91 satisfying that 1 .
@xmath92 and @xmath93 , 2 . for @xmath94 with @xmath95 , @xmath96 if and only if @xmath97 and @xmath98 .
a thread block is a _
canonical thread block _ if it satisfies on the top of the previous two conditions and the following third condition : 1 .
@xmath99 if @xmath100 .
the third condition implies that every canonical thread block has no pendant vertex adjacent to its first vertex @xmath101 .
it will guarantee a unique decomposition of a thread graph into thread blocks .
we say that @xmath89 and @xmath102 are the _ ordering _ and _ labeling _ of @xmath2 respectively , and say that @xmath101 and @xmath103 are the _ first and last _ vertices of @xmath2 , respectively
. see figure [ fig : threadblock ] for an example .
let @xmath104 be a digraph .
a set @xmath105 of thread blocks is said to be _ mergeable with @xmath106 _ if 1 . for every arc @xmath107 of @xmath108
, @xmath109 is a thread block whose first and last vertices are @xmath46 and @xmath110 , respectively , and 2 . for two distinct arcs @xmath111 of @xmath108 , @xmath112 . for a digraph @xmath104 and a set of thread blocks @xmath113 mergeable with @xmath106
, we define @xmath114 as the graph @xmath2 with the vertex set @xmath115 and the edge set @xmath116 .
we say that @xmath106 is the _ underlying digraph _ of @xmath2 and that @xmath117 is a _ thread decomposition _ of @xmath2 .
we say that @xmath117 is a _ canonical thread decomposition _ of @xmath2 if every thread block of @xmath118 is a canonical thread block .
a connected graph @xmath2 is a _ thread graph _ if @xmath2 is either an one vertex graph or @xmath119 for some directed path @xmath120 and some set of thread blocks @xmath121 mergeable with @xmath120 .
a graph is a _ thread graph _ if each of its connected components is a thread graph .
see figure [ fig : threadgraph ] for an example of a connected thread graph , and its canonical thread decomposition .
it is not hard to observe that every connected thread graph admits a canonical thread decomposition ; one can obtain a canonical thread decomposition from any thread decomposition by rearranging pendant vertices adjacent to a vertex in the underlying digraph . in lemma [ lem :
splittreetothreadblock ] , we will give a polynomial - time algorithm that given a connected thread graph , outputs its canonical thread decomposition .
the following structural properties of thread graphs will be used in later sections .
[ lem : structure ] let @xmath65 be an integer , and let @xmath2 be a connected thread graph such that @xmath2 admits a canonical thread decomposition @xmath122 where @xmath123 and @xmath124
. then the following are satisfied . 1 . for every @xmath125
, @xmath52 has a neighbor in @xmath120 .
vertices @xmath126 are cut vertices of @xmath2 . moreover , every cut vertex of @xmath2 is contained in @xmath120 .
3 . for each @xmath127
, @xmath128 is exactly the union of the set of all pendant vertices adjacent to @xmath129 and the vertex set of the block of @xmath2 containing @xmath130 .
\(1 ) let @xmath131 be the thread block containing @xmath52 .
depending on the label of @xmath52 in @xmath131 , @xmath52 is adjacent to at least one of @xmath132 .
\(2 ) the first statement came from the definition of a connected thread graph that for @xmath133 with @xmath134 , @xmath135 .
that is , for each @xmath136 , all paths from @xmath137 to @xmath129 must pass through @xmath138 , and it implies that @xmath138 is a cut vertex of @xmath2 . if @xmath125 , then by the statement ( 1 ) , every vertex of @xmath139 has a neighbor in @xmath120 and thus @xmath139 is connected .
it implies that every cut vertex of @xmath2 is contained in @xmath120 .
\(3 ) let us fix @xmath140 and let @xmath89 and @xmath102 be the ordering and labeling of @xmath131 , respectively .
let @xmath58 be the vertex set containing all pendant vertices adjacent to @xmath129 and the vertex set of the block of @xmath2 containing @xmath141 and @xmath129 .
we need to prove that @xmath142 . to prove that @xmath143 , let @xmath144 .
observe that by construction of @xmath2 and the fact that @xmath145 , if @xmath52 is not a pendant vertex and satisfies @xmath146 , then @xmath147 .
so , let us assume that @xmath52 is not a pendant vertex and has a neighbor in @xmath120 other than @xmath138 and @xmath129 .
if @xmath52 is adjacent to some vertex @xmath148 of @xmath120 with @xmath149 , then it contradicts the fact that @xmath129 separates @xmath138 and @xmath148 in @xmath2 by ( 2 ) . similarly ,
if @xmath52 is adjacent to some vertex @xmath148 of @xmath120 with @xmath150 , then it contradicts the fact that @xmath138 separates @xmath148 and @xmath129 in @xmath2 by ( 2 ) .
thus , we have @xmath146 and @xmath147 . finally by construction , as @xmath151 is a canonical thread decomposition , the pendant vertices adjacent to @xmath129 belong to @xmath131 .
we conclude that @xmath143 .
we verify that @xmath152 .
let @xmath153 . if @xmath154 , then @xmath52 is contained in @xmath47 .
if @xmath155 , then by definition of a canonical thread block , there exists @xmath156 with @xmath157 .
so , @xmath52 is contained in a cycle of length @xmath158 together with @xmath132 , and thus , @xmath145 .
if @xmath159 and there exists @xmath95 with @xmath98 , then similarly we have @xmath145 , and if there are no such a vertex @xmath25 , then @xmath52 is a pendant vertex adjacent to @xmath129 , that is also contained in @xmath47 . that are distance - hereditary .
] let @xmath160 be the house , gem , domino graphs respectively , which are depicted in figure [ fig : obsdh ] . the induced subgraph obstructions for graphs of linear rankwidth at most @xmath0 consist of the set of induced subgraph obstructions for distance - hereditary graphs @xcite , that are @xmath161 , @xmath162 , @xmath163 , and induced cycles of length at least @xmath69 , and the set of 14 induced subgraph obstructions for graphs of linear rankwidth at most @xmath0 that are distance - hereditary , depicted in figure [ fig : obslrw1 ] @xcite .
we define that * @xmath164 is the union of @xmath165 and the set of @xmath166 graphs in figure [ fig : obslrw1 ] . [ thm : structurethread ] for a graph @xmath2 , the following are equivalent .
* @xmath2 has linear rankwidth at most @xmath0 .
* @xmath2 is a thread graph .
* @xmath2 has no induced subgraph isomorphic to a graph in @xmath164 .
we often use the term ` thread graphs ' for graphs of linear rankwidth at most @xmath0 . for a graph @xmath2 and @xmath58 , @xmath47 is called a _
lrw1-deletion set _ if @xmath167 is a graph of linear rankwidth at most @xmath0 .
it is known that one can recognize graphs of linear rankwidth at most @xmath0 in time @xmath168 using split decompositions @xcite .
furthermore , we can easily obtain a canonical thread decomposition of a graph of linear rankwidth at most @xmath0 from its split decomposition , but for our knowledge , it was not stated anywhere . in this subsection
, we clarify a procedure to obtain a thread decomposition .
this will be especially used in the kernelization algorithm in section [ sec : polykerthreaddel ] .
we use graph - labelled trees introduced by gioan and paul @xcite , which are convenient forms of split decompositions . a triple @xmath169 of a tree @xmath170 and a set of graphs @xmath171 and a set of functions @xmath172
is called a _
graph - labelled tree _
if * for every node @xmath52 of degree @xmath3 in @xmath170 , @xmath173 is a connected graph on @xmath3 vertices , called _ marker vertices _ , * @xmath174 is a bijection from the edges of @xmath170 incident with @xmath52 to the marker vertices of @xmath173 .
let @xmath52 be a leaf of @xmath170 .
a node or a leaf @xmath175 different from @xmath52 is called _ @xmath52-accessible _ if for every edges @xmath107 and @xmath176 on the path from @xmath175 to @xmath52 in @xmath170 , @xmath177 is adjacent to @xmath178 in @xmath179 .
the _ accessibility graph _ of a graph - labelled tree @xmath180 is the graph @xmath181 whose vertex set is the set of all leaves of @xmath170 , and @xmath182 if and only if @xmath110 is @xmath46-accessible .
we say that @xmath180 is a graph - labelled tree of @xmath181 .
we give two operations on a graph - labelled tree to define a _ reduced graph - labelled tree_. a _ split _ of a graph @xmath2 is a vertex partition @xmath183 of @xmath2 such that @xmath184 , and there exist @xmath185 where the set of edges incident with both @xmath186 and @xmath187 is exactly @xmath188
. a connected graph is _ degenerate _ if every vertex partition @xmath183 with @xmath184 is a split .
it is known that every degenerate graph is either a complete graph or a star graph .
a node in a graph - labelled tree is called a _ clique node _ ( or a _ star node _ ) if a complete graph ( or a star graph , respectively ) is assigned to the node . a graph without splits is called a _
prime graph_. let @xmath180 be a graph - labelled tree of a graph @xmath2 .
let @xmath52 be a node of @xmath170 such that @xmath173 admits a split @xmath189 . for each @xmath190
, let @xmath191 be the graph obtained from @xmath192 $ ] by adding a new vertex @xmath193 that is adjacent to all vertices in @xmath194 .
then the _ node - split _ operation on the node @xmath52 with respect to @xmath189 consists of substituting @xmath52 by two adjacent nodes @xmath195 , respectively labelled by @xmath196 , such that for each @xmath190 , @xmath197 the _ node - join _ operation is the reverse operation of the node - split operation ; if @xmath198 is an edge of @xmath170 , then the _ node - join _ operation on @xmath198 consists of contracting @xmath198 into a new node @xmath175 labelled by the graph @xmath199 where @xmath199 is the graph obtained from the disjoint union of @xmath173 and @xmath200 by deleting @xmath201 and @xmath202 and adding all edges between @xmath203 and @xmath204 .
we say that a star node @xmath52 is _ oriented towards a node @xmath205 _ of @xmath170 if the edge @xmath206 such that @xmath207 is the center of @xmath173 is on the path in @xmath170 between @xmath205 and @xmath52 . a graph - labelled tree is _ reduced _ if 1 .
every node is either prime or degenerate , and 2 .
it contains no edge that connects two degenerate nodes where the node - join operation on this edge results in another degenerate node .
cunningham showed the uniqueness of a reduced graph - laballed tree of a connected graph .
moreover , it can be computed in time @xmath208 .
[ thm : cunningham ] every connected graph @xmath2 admits a unique reduced graph - labelled tree , and it can be computed in time @xmath208 . for a connected graph @xmath2 ,
we denote by @xmath209 the unique reduced graph - labelled tree obtained from theorem [ thm : cunningham ] .
we call it the _ split tree _ of @xmath2
. the following characterization of graphs of linear rankwidth @xmath0 is crucial .
we give an example of the split tree of a connected thread graph in figure [ fig : splittree ] .
= [ circle , draw , solid , fill = black , inner sep=0pt , minimum width=3pt ] ( 30 - 40,40 ) ellipse ( 14 and 15 ) ; ( a1 ) [ v ] at ( 30 - 40,50 ) ; ( a3 ) [ v ] at ( 20 - 40,40 ) ; ( a4 ) [ v ] at ( 40 - 40,40 ) ; ( a1)(a3)(a4)(a1 ) ; ( a5 ) [ v ] at ( 20 - 50,40 ) ; ( a6 ) [ v ] at ( 30 - 40,60 ) ; ( a5)(a3 ) ; ( a6)(a1 ) ; ( 30 - 40 , 67 ) node@xmath210 ; ( 30,40 ) ellipse ( 14 and 15 ) ; ( b1 ) [ v ] at ( 30,50 ) ; ( b3 ) [ v ] at ( 20,40 ) ; ( b4 ) [ v ] at ( 40,40 ) ; ( b1)(b3)(b4 ) ; ( b6 ) [ v ] at ( 30,60 ) ; ( b6)(b1 ) ; ( 30 , 67 ) node@xmath211 ; ( 30 + 40,40 ) ellipse ( 14 and 15 ) ; ( c1 ) [ v ] at ( 30 + 40,50 ) ; ( c3 ) [ v ] at ( 20 + 40,40 ) ; ( c4 ) [ v ] at ( 40 + 40,40 ) ; ( c1)(c3)(c4)(c1 ) ; ( c6 ) [ v ] at ( 30 + 40,60 ) ; ( c6)(c1 ) ; ( 30 + 40 , 67 ) node@xmath210 ; ( 30 + 80,40 ) ellipse ( 14 and 15 ) ; ( d1 ) [ v ] at ( 30 + 80,50 ) ; ( d3 ) [ v ] at ( 20 + 80,40 ) ; ( d4 ) [ v ] at ( 40 + 80,40 ) ; ( d3)(d1)(d4 ) ; ( d6 ) [ v ] at ( 30 + 80,60 ) ; ( d6)(d1 ) ; ( 30 + 120,40 ) ellipse ( 14 and 15 ) ; ( e1 ) [ v ] at ( 30 + 120,50 ) ; ( e3 ) [ v ] at ( 20 + 120,40 ) ; ( e4 ) [ v ] at ( 40 + 120,40 ) ; ( e1)(e4)(e3 ) ; ( e6 ) [ v ] at ( 30 + 120,60 ) ; ( e6)(e1 ) ; ( 30 + 120 , 67 ) node@xmath212 ; ( 30 + 160,40 ) ellipse ( 14 and 15 ) ; ( f1 ) [ v ] at ( 30 + 160,50 ) ; ( f3 ) [ v ] at ( 20 + 160,40 ) ; ( f4 ) [ v ] at ( 40 + 160,40 ) ; ( f1)(f3)(f4)(f1 ) ; ( f6 ) [ v ] at ( 30 + 160,60 ) ; ( f6)(f1 ) ; ( 30 + 160 , 67 ) node@xmath210 ; ( 30 + 200,40 ) ellipse ( 14 and 15 ) ; ( g1 ) [ v ] at ( 30 + 200,50 ) ; ( g3 ) [ v ] at ( 20 + 200,40 ) ; ( g4 ) [ v ] at ( 40 + 200,40 ) ; ( g1)(g3)(g4 ) ; ( g6 ) [ v ] at ( 30 + 200,60 ) ; ( g6)(g1 ) ; ( 30 + 200 , 67 ) node@xmath211 ; ( 30 + 240,40 ) ellipse ( 14 and 15 ) ; ( h1 ) [ v ] at ( 30 + 240,50 ) ; ( h2 ) [ v ] at ( 30 + 240,30 ) ; ( h3 ) [ v ] at ( 20 + 240,40 ) ; ( h4 ) [ v ] at ( 40 + 240,40 ) ; ( h3)(h1)(h4 ) ; ( h1)(h2 ) ; ( h6 ) [ v ] at ( 30 + 240,60 ) ; ( h7 ) [ v ] at ( 30 + 240,20 ) ; ( h6)(h1 ) ; ( h7)(h2 ) ; ( 30 + 240 , 10 ) node@xmath212 ; ( a4)(b3 ) ; ( b4)(c3 ) ; ( c4)(d3 ) ; ( d4)(e3 ) ; ( e4)(f3 ) ; ( f4)(g3 ) ; ( g4)(h3 ) ; ( 30 + 265 , 40 ) node@xmath213 ; [ thm : charlrw1split ] a connected graph @xmath2 with the split tree @xmath180 is a graph of linear rankwidth at most @xmath0 if and only if every graph in @xmath171 is degenerate and the tree obtained from @xmath170 by removing all its leaves is a path .
thus , one can recognize graphs of linear rankwidth at most @xmath0 in time @xmath208 .
[ lem : splittreetothreadblock ] given a connected graph @xmath2 of linear rankwidth at most @xmath0 , we can output a canonical thread decomposition @xmath214 in time @xmath208 . using the algorithm in theorem [ thm : cunningham ] , we compute a split tree of @xmath2 in time @xmath168 .
let @xmath215 be the obtained split tree of a connected thread graph @xmath2 . by theorem [ thm : charlrw1split ] ,
the tree obtained from @xmath170 by removing all its leaves is a path and each @xmath173 is degenerate , that is , either a star node or a clique node .
let @xmath216 be the path @xmath217 obtained from @xmath170 by removing all its leaves , and we give a direction on edges so that it is the directed path from @xmath70 to @xmath218 .
note that for each leaf @xmath52 in @xmath170 with the neighbor @xmath25 , there exists a unique marker vertex @xmath175 in @xmath200 such that @xmath219 .
we denote by @xmath220 .
we choose the sequence @xmath221 of all star nodes such that * @xmath222 , * for each @xmath223 , there is a leaf @xmath224 of @xmath170 such that @xmath225 is the center of @xmath226 .
we choose a leaf @xmath227 of @xmath170 that is adjacent to @xmath70 , and choose a leaf @xmath228 of @xmath170 that is adjacent to @xmath218 .
it is not hard to check that @xmath229 .
also , from the accessibility among nodes of @xmath170 , one can observe that @xmath230 are cut vertices of @xmath2 , and thus @xmath231 is an induced path of @xmath2 .
let @xmath232 , and we regard it as a directed path from @xmath233 to @xmath234 .
now , we construct a set of thread blocks @xmath235 where @xmath236 is a canonical thread decomposition of @xmath2 . for each node @xmath237 , let @xmath238 be the set of all leaves in @xmath170 that are adjacent to @xmath52 . for convenience ,
let @xmath239 and @xmath240 .
for each @xmath241 , we define the following : 1 .
let @xmath242 $ ] .
we take an ordering @xmath243 of @xmath244 such that @xmath224 and @xmath245 are the first and last vertices , and @xmath246 if there is a directed path from the neighbor of @xmath52 to the neighbor of @xmath25 in @xmath216 .
we take an arbitrary ordering for the leaves that have the same neighbor in @xmath170 , except @xmath245 .
this can be done in time @xmath247 .
3 . from the definition of @xmath248 , it is not hard to check that for @xmath249 with @xmath246 , @xmath96 if and only if * @xmath250 , or the neighbor of @xmath52 is either a clique node or a star node oriented towards @xmath218 , and * @xmath251 , or the neighbor of @xmath25 is either a clique node or a star node oriented towards @xmath70 .
+ note that the vertices in @xmath252 are pendant vertices adjacent to @xmath245 in @xmath2 . following the above observations
, we give a labeling @xmath253 on @xmath244 such that for @xmath254 with the neighbor @xmath255 in @xmath170 , @xmath256 it takes @xmath247 time .
it is not hard to verify that @xmath243 and @xmath253 are proper ordering and labeling of @xmath257 using accessibility among nodes of @xmath170 .
moreover , as @xmath258 does not contain a vertex in @xmath259 , which is a pendant vertex adjacent to @xmath224 if exists , @xmath257 is a canonical thread block whose first and last vertices are @xmath224 and @xmath245 , respectively , with the ordering @xmath243 and the labeling @xmath253 . + we return @xmath260 .
note that @xmath230 are cut vertices of @xmath2 .
thus , it is straightforward to check that 1 .
@xmath261 for every pair @xmath262 with @xmath134 , 2 .
@xmath263 and @xmath264 .
we conclude that @xmath236 is the canonical thread decomposition of @xmath2 , and it can be computed in time @xmath168 .
we generalize the construction of thread graphs from directed paths to directed cycles .
a connected graph @xmath2 is called a _ necklace graph _ if @xmath265 for some directed cycle @xmath266 and some set of thread blocks @xmath267 mergeable with @xmath266 .
see figure [ fig : necklace ] for an example of a necklace graph .
let @xmath268 the main result of this section is the following .
[ thm : mainlrw ] every connected @xmath269-free graph is either a connected thread graph or a necklace graph whose underlying cycle has length at least @xmath270 . assuming that @xmath2 is not a thread graph , to prove theorem [ thm : mainlrw ]
, we recursively find an underlying cycle @xmath266 of length @xmath271 and a set of thread blocks @xmath272 such that @xmath273 . at each recursion step
, we assume that @xmath274 and prove that 1 . if @xmath45 contains two vertices of distance at least @xmath73 in @xmath2 , then @xmath2 contains a graph in @xmath269 or an induced cycle of length @xmath102 with @xmath275 , 2 .
if every pair of vertices of @xmath45 are at distance at most @xmath72 , then @xmath45 is contained in the union of two consecutive thread blocks of @xmath276 , and then either @xmath2 contains a graph in @xmath269 or it is a necklace graph with @xmath266 as the underlying cycle .
we need some preliminary lemmas .
[ lem : excludecycle8 ] let @xmath2 be a necklace graph whose underlying cycle @xmath266 has length @xmath277 .
the distance in @xmath2 between any pair of vertices is at most @xmath278 . observe that when merging the ( directed ) cycle @xmath266 with a set of thread blocks @xmath272 to obtain @xmath273 , the distance between any two vertices of @xmath266 in @xmath266 is the same as the distance in @xmath2 ( viewed as an undirected cycle ) .
as the distance between any pair of vertices in an induced cycle of length @xmath277 is at most @xmath279 , the statement follows from the fact that every vertex in @xmath2 either belongs to @xmath266 or has a neighbor in @xmath266 .
[ lem : pathtoobs ] let @xmath280 be an integer .
let @xmath2 be a graph and let @xmath281 such that @xmath139 is an induced path @xmath282 , and @xmath52 is adjacent to both @xmath283 and @xmath284 in @xmath2
. then @xmath2 contains an induced subgraph isomorphic to @xmath161 , @xmath162 , @xmath163 , or an induced cycle of length at least @xmath69 .
let @xmath285 be the sequence of integers such that @xmath286 are all neighbors of @xmath52 on @xmath120 . if @xmath287 for some @xmath288 , then @xmath289 is an induced cycle of length at least @xmath69 .
we may assume that @xmath290 for all @xmath288 .
if @xmath291 for some @xmath292 , then @xmath293 $ ] is isomorphic to @xmath163 .
if one of @xmath294 and @xmath295 is @xmath0 and the other value is @xmath72 for some @xmath292 , then @xmath293 $ ] is isomorphic to @xmath161 .
so we may assume that @xmath296 for all @xmath288 .
since @xmath280 , @xmath2 has an induced subgraph isomorphic to @xmath162 .
[ lem : extendthread ] let @xmath2 be a connected thread graph and @xmath281 such that @xmath139 admits a canonical thread decomposition @xmath297 where @xmath298 and @xmath299 . if @xmath300 , then there exists a set @xmath301 of thread blocks mergeable with @xmath120 such that @xmath302 is a canonical thread decomposition of @xmath2 . since @xmath2 is a connected thread graph
, @xmath2 admits a canonical thread decomposition @xmath303 with a directed path @xmath304 and a mergeable set of thread blocks @xmath305 .
we first show that @xmath306 should be contained in @xmath304 . by lemma [ lem : structure ]
, it is enough to show that they are cut vertices of @xmath2 .
note that @xmath307 and @xmath308 are still cut vertices of @xmath2 as @xmath300 .
we claim that @xmath71 is also a cut vertex of @xmath2 .
suppose that @xmath71 is not a cut vertex of @xmath2 .
this implies that there is a path from @xmath307 to @xmath308 in @xmath309 .
we take a shortest path @xmath216 among such paths .
if it has length @xmath72 , then @xmath310 induces a subgraph isomorphic to @xmath311 or @xmath312 depending on the adjacency between @xmath71 and the middle vertex of @xmath216 .
if it has length at least @xmath73 , then by lemma [ lem : pathtoobs ] , @xmath313 $ ] contains an induced subgraph in @xmath164 , which contradicts to our assumption .
thus , @xmath71 is also a cut vertex of @xmath2 .
since @xmath306 are cut vertices of @xmath2 , by ( 2 ) of lemma [ lem : structure ] , we have @xmath314 .
if @xmath315 is a directed path in @xmath304 , then for each @xmath316 , we take the subgraph @xmath317 of @xmath2 induced on the union of the set of all pendant vertices adjacent to @xmath129 and the vertex set of the block of @xmath2 containing @xmath130 . by ( 3 ) of lemma [ lem : structure ] , @xmath318 and @xmath319 are the canonical thread blocks listed in @xmath305 .
now , we assume that @xmath320 is a directed path in @xmath304 . in this case , for each @xmath321 , the subgraph @xmath322 of @xmath2 induced on the union of the set of all pendant vertices adjacent to @xmath141 and the vertex set of the block of @xmath2 containing @xmath130 is a canonical thread block listed in @xmath305 .
then we obtain @xmath317 from @xmath322 by removing all pendant vertices adjacent to @xmath138 and adding all pendant vertices adjacent to @xmath129 .
it is not hard to observe that @xmath317 is a canonical thread block whose first and last vertices are @xmath132 , respectively ; we can take a reverse ordering for vertices in the block containing @xmath132 , and replace each label @xmath323 with @xmath324 and each label @xmath324 with @xmath323 .
now , we set @xmath325 .
we claim that @xmath326 . as each thread block
@xmath131 or @xmath317 in @xmath301 satisfies that @xmath327 or @xmath328 , we have that * @xmath329 , and * @xmath330 . as @xmath151 is a canonical thread decomposition of @xmath139 and @xmath331 , by ( 3 ) of lemma [ lem : structure ] ,
every vertex of @xmath2 is either a pendant vertex adjacent to one of @xmath332 , or contained in a block containing two consecutive vertices @xmath132 .
it implies that @xmath333 , and we have * @xmath334 and @xmath335 .
therefore , we conclude that @xmath326 .
let @xmath2 be a connected @xmath269-free graph and suppose that @xmath2 is not a thread graph . by theorem [ thm : structurethread ]
, @xmath2 contains an induced cycle of length at least @xmath270 .
let @xmath266 be a shortest cycle among induced cycles of length at least @xmath270 in @xmath2 . for convenience ,
let @xmath336 .
we regard @xmath266 as a directed cycle on vertex set @xmath337 ( with @xmath277 ) , where for each @xmath338 , @xmath339 is an arc .
we prove by induction on @xmath340 that @xmath2 is a necklace graph whose underlying cycle is @xmath266 .
we may assume that @xmath341 .
clearly , @xmath139 is again @xmath269-free graph and @xmath266 is a shortest cycle among induced cycles of length at least @xmath270 in @xmath139 . by the induction hypothesis
, @xmath139 is a necklace graph whose underlying cycle is @xmath266 and thereby has a canonical thread decomposition @xmath342 where @xmath343 .
we claim that all vertices of @xmath344 have pairwise distance at most @xmath72 in @xmath139 .
suppose that there are two vertices @xmath345 that have distance at least @xmath73 in @xmath139 .
by lemma [ lem : excludecycle8 ] , the distance in @xmath139 between @xmath46 and @xmath110 is at most @xmath278 . we take a shortest path @xmath216 from @xmath46 to @xmath110 in @xmath139
. then by lemma [ lem : pathtoobs ] , @xmath346 $ ] contains an induced subgraph isomorphic to either @xmath161 , @xmath162 , @xmath163 , or an induced cycle @xmath347 of length @xmath348 with @xmath349 .
as @xmath266 was selected as a smallest induced cycle of length @xmath277 , if there is an induced cycle @xmath347 , then it has length at most @xmath35 .
this contradicts the assumption that @xmath2 is @xmath269-free .
thus , all vertices of @xmath344 have pairwise distance at most @xmath72 in @xmath139 .
it implies that @xmath350 for some @xmath351 .
let us consider @xmath352 $ ] .
observe that @xmath353 is a thread graph .
the facts that @xmath2 is @xmath354-free and that @xmath52 does not belong to a cycle of @xmath2 of length @xmath69 or more ( see paragraph above ) , imply that @xmath56 is @xmath355-free and thereby a thread graph ( theorem [ thm : structurethread ] ) . by construction @xmath353
is a thread graph with a canonical thread decomposition @xmath356 where @xmath357 and @xmath120 is the directed path @xmath358 .
so we apply lemma [ lem : extendthread ] on @xmath56 and @xmath353 .
let @xmath359 be the resulting canonical thread decomposition of @xmath56 .
then @xmath360 is a canonical thread decomposition of @xmath2 .
in this section , we give two single - exponential fpt algorithms for the lrw1-vertex deletion problem . both fpt algorithms
are based on branching algorithms that reduce an input graph to a @xmath269-free graph .
we first show that one can easily find a minimum lrw1-deletion set in a @xmath269-free graph .
it implies that there is an fpt algorithm with running time @xmath36 with a simple branching algorithm .
next , we provide a way to reduce the polynomial factor @xmath37 into @xmath361 using an algorithm to find a fixed subgraph on graphs of bounded cliquewidth .
we prove the following .
[ prop : nltothread2 ] given a @xmath269-free graph @xmath2 , we can compute a minimum lrw1-deletion set of @xmath2 in time @xmath362 .
we observe that every necklace graph can be turned into a thread graph by removing a vertex on the underlying cycle .
[ lem : nltothread ] let @xmath2 be a connected necklace graph with the underlying directed cycle @xmath266 . for each @xmath363 , @xmath139 is a thread graph .
let @xmath266 be the directed cycle on vertex set @xmath337 , where for each @xmath338 , @xmath339 is an arc . and let @xmath364 be the set of thread blocks mergeable with @xmath266 such that @xmath365 . because of the symmetry , it is sufficient to show that @xmath366 is a thread graph . as the pendent vertices of @xmath2 that are adjacent to @xmath70 become isolated vertices after removing @xmath70 , we may assume that @xmath2 has no pendent vertices adjacent to @xmath70 . for each @xmath351
, let @xmath248 and @xmath367 be the ordering and labeling of @xmath191 , respectively .
suppose that @xmath368 .
since @xmath369 has no pendant vertices adjacent to @xmath70 , the vertex @xmath370 is labelled either @xmath371 or @xmath323 .
let @xmath372 be the restriction of @xmath373 on @xmath374 , and for each @xmath375 , let @xmath376 it is easy to check that @xmath377 is a thread block with the ordering @xmath372 and labeling @xmath378 . if @xmath379 , then we can regard @xmath380 as the last thread block of @xmath366 .
similarly , if @xmath381 , then we can regard @xmath382 as the first thread block of @xmath366 . otherwise , we regard @xmath383 as the first thread block .
let @xmath384 and @xmath385 .
we conclude that @xmath366 is a thread graph with the underlying directed path @xmath120 where @xmath386 let @xmath2 be a @xmath269-free graph and let @xmath3 be the minimum size of a lrw1-deletion set of @xmath2 . by theorem [ thm : mainlrw ] , each connected component of @xmath2 is either a thread graph or a necklace graph . for each component @xmath56 of @xmath2 , we can test whether @xmath56 is a thread graph or not in time @xmath387 using theorem [ thm : charlrw1split ] . by lemma [ lem : nltothread ] ,
it is enough to remove exactly one vertex to make each necklace component a thread graph .
thus , @xmath3 is equal to the number of its necklace components .
moreover , in each necklace component @xmath56 , we can identify a vertex @xmath52 on the underlying cycle by testing whether @xmath353 is a thread graph for every vertex @xmath52 in @xmath56 .
it takes a @xmath388 time .
therefore , we can find a minimum lrw1-deletion set of @xmath2 in time @xmath389 .
we give an fpt algorithm using a simple branching algorithm .
[ thm : algorithm1 ] the lrw1-vertex deletion problem can be solved in time @xmath33 .
let @xmath390 be an instance of the lrw1-vertex deletion problem where @xmath2 is a graph on @xmath1 vertices .
first recursively find an induced subgraph of @xmath2 isomorphic to a graph in @xmath269 and branch by removing one of the vertices in the subgraph . because the maximum size of graphs in @xmath269 is @xmath35
, we can find such a vertex subset in time @xmath391 if exists . in the end
, we transform the instance @xmath390 into at most @xmath392 sub - instances @xmath393 such that each sub - instance consists of a @xmath269-free graph @xmath394 and a remaining budget @xmath395 .
thus , this branching step takes a time @xmath36 .
clearly , @xmath396 is a yes - instance if and only if one of sub - instances @xmath393 is a yes - instance .
let @xmath393 be a sub - instance obtained from the branching algorithm .
since @xmath394 is @xmath269-free , using the algorithm in proposition [ prop : nltothread2 ] , we can compute a minimum lrw1-deletion set of @xmath394 and decide whether @xmath393 is a yes - instance in time @xmath397 . by checking all sub - instances
, we can decide whether @xmath390 is a yes - instance in time @xmath36 .
theorem [ thm : algorithm1 ] already gives a single - exponential fpt algorithm for the lrw1-vertex deletion problem , but the polynomial factor @xmath37 makes it impractical . in the next subsection
, we give an algorithm with better polynomial factor , using the branching algorithm based on a cliquewidth expression .
the _ cliquewidth _ of a graph @xmath2 is the minimum number of labels needed to construct @xmath2 using the following four operations : 1 .
creation of a new vertex @xmath52 with label @xmath398 ( denoted by @xmath399 ) .
disjoint union of two labeled graphs @xmath2 and @xmath56 ( denoted by @xmath400 ) .
3 . joining by an edge each vertex with label @xmath398 to each vertex with label @xmath401 ( @xmath134 , denoted by @xmath402 ) .
4 . renaming label @xmath398 to @xmath401 ( denoted by @xmath403 ) .
every graph can be defined by an algebraic expression using these four operations .
such an expression is called a _ @xmath3-expression _ if it uses at most @xmath3 different labels .
thus , the cliquewidth of @xmath2 is the minimum @xmath3 for which there exists a @xmath3-expression defining @xmath2 .
= [ sibling distance=30 mm ] = [ sibling distance=15 mm ] child node @xmath404 child node @xmath405 child node @xmath406 child node @xmath404 child node @xmath407 child node @xmath404 child node @xmath408 child node @xmath409 child node @xmath410 child node @xmath404 child node @xmath411 child node @xmath412 child node @xmath413 ; for instance , the cycle @xmath414 of length @xmath69 admits the following @xmath73-expression : @xmath415 we can represent this expression as a tree - structure , depicted in figure [ fig : c5expression ] .
we call this tree the _ labelled tree induced by the @xmath3-expression of @xmath2_. as we need the relation between rankwidth and linear rankwidth , we define rankwidth .
a tree is _ subcubic _ if it has at least two vertices and every inner vertex has degree @xmath73 .
a pair @xmath416 is called a _
rank - decomposition _ of a graph @xmath2 if @xmath170 is a subcubic tree and @xmath211 is a bijection from the vertices of @xmath2 to the leaves of @xmath170 . for each edge @xmath206 in @xmath170 , @xmath417 induces a partition @xmath418 of the leaves of @xmath170 .
the _ width _ of an edge @xmath206 is defined as @xmath419)$ ] .
the _ width _ of a rank - decomposition @xmath420 is the maximum width over all edges of @xmath170 .
the _ rankwidth _ of @xmath2 is the minimum width over all rank - decompositions of @xmath2 . if @xmath421 , then @xmath2 admits no rank - decomposition and it has rankwidth @xmath422 .
the following approximation algorithm was given by oum @xcite .
[ thm : approxcw ] given a graph @xmath2 and positive integer @xmath3 , one can output a rank - decomposition of width at most @xmath423 of @xmath2 or confirm that the rankwidth of @xmath2 is larger than @xmath3 in time @xmath7 .
moreover , given a rank - decomposition of a graph @xmath2 of width @xmath3 , one can output a @xmath424-expression of @xmath2 in time @xmath425 .
remark that the rankwidth of a graph can be decreased by at most @xmath0 when removing a vertex , and the rankwidth of a graph is at most its linear rankwidth .
thus , if a given graph has rankwidth larger than @xmath38 , then we can not make it a graph of linear rankwidth at most @xmath0 by removing at most @xmath3 vertices .
therefore , using the algorithm in theorem [ thm : approxcw ] , we can assume that the instance is given with a @xmath40-expression , and we can use it to find an induced subgraph of constant size .
[ prop : inducedoncw ] let @xmath56 be a graph on @xmath426 vertices .
given a graph @xmath2 with its @xmath3-expression , one can test whether @xmath2 contains an induced subgraph isomorphic to @xmath56 in time @xmath427 , and output a vertex set inducing @xmath56 if exists , in the same time complexity .
let @xmath428 be the vertices of @xmath56 and @xmath429 be the edges of @xmath56 .
let @xmath430 be the given @xmath3-expression defining @xmath2 , and let @xmath170 be the labelled rooted tree induced by @xmath430 .
for every node @xmath205 of @xmath170 , let @xmath431 be the graph defined at node @xmath205 , and for each @xmath432 , let @xmath433 $ ] be the subgraph of @xmath431 induced on the vertices with label @xmath398 . for every node @xmath205 and every vector @xmath434 , we define the following value : * @xmath435 = 1 $ ] if there exists an injective mapping @xmath436 from @xmath437 to @xmath438 such that * * for each @xmath439 , @xmath440)$ ] , * * for each @xmath441 with @xmath442 , we have @xmath443 , @xmath444 , and @xmath445 , * * for each @xmath441 with @xmath446 , @xmath443 and @xmath444 , we have @xmath447 , * @xmath435 = 0 $ ] , otherwise
. the values @xmath448 $ ] will capture all possible inequivalent subgraphs of @xmath56 at @xmath431 , where two subgraphs are equivalent if their corresponding vertices are in the same labels .
a vector @xmath449 is _ complete _ if for each @xmath351 and @xmath450 , @xmath193 and @xmath451 are non - zero .
one can observe that @xmath2 contains an induced subgraph isomorphic to @xmath56 if and only if there is a complete vector @xmath52 such that @xmath452=1 $ ] .
a vector @xmath453 is called a _ sub - vector _ of @xmath454 if for each @xmath455 and @xmath456 , @xmath457 and @xmath458 .
now , we present how the values of @xmath459 $ ] are computed . at each node @xmath205 , we compute the value @xmath435 $ ] as follows . 1 . ( creation of a new vertex @xmath25 with label @xmath398 ) each vertex of @xmath56
can be mapped to @xmath25 .
thus , for each vector @xmath460 where exactly one of @xmath461 is @xmath398 and @xmath462 for all @xmath463 , we assign @xmath464:=1 $ ] and for all other vectors , assign @xmath464:=0 $ ] .
2 . ( disjoint union node with two children @xmath465 ) we take all possible two sub - vectors @xmath466 and @xmath467 of @xmath454 such that
* for each @xmath455 , if @xmath468 is non - zero , then exactly one of @xmath469 and @xmath470 is equal to @xmath468 , and the same thing holds for @xmath471 , @xmath472 , and @xmath473 .
+ we assign @xmath464:=1 $ ] if there exist such @xmath195 where @xmath474=c[t_2 , v_2]=1 $ ] , and @xmath464:=0 $ ] otherwise .
this formula holds because @xmath431 contains a subgraph @xmath475 if and only if there is a vertex partition @xmath476 and @xmath477 of @xmath475 with no edges between @xmath476 and @xmath477 such that @xmath478 $ ] is a subgraph of @xmath479 and @xmath480 $ ] is a subgraph of @xmath481 ( @xmath476 or @xmath477 may be an empty set ) .
this also implies that for a vector @xmath52 with @xmath482 and @xmath483 , it suffices to consider a pair @xmath484 which satisfies either of the following : @xmath485 , or @xmath486 .
+ note that there are at most @xmath487 possible pair of vectors .
thus , we need @xmath487 iterations to compute @xmath488 $ ] for fixed @xmath52 . in total , we need @xmath489 iterations to compute @xmath488 $ ] for all @xmath52 .
( join node with the child @xmath490 such that each vertex with label @xmath398 is joined to each vertex with label @xmath401 ) if there are @xmath491 and @xmath463 where @xmath483 with @xmath492 , @xmath493 and @xmath462 , then we assign @xmath488:=0 $ ] .
this correctly assigns the value as two vertices with labels @xmath398 and @xmath401 should be adjacent at this join node .
we can check it in time @xmath494 .
now , we assume that the given vector @xmath52 satisfies that for all @xmath491 and @xmath463 with @xmath483 with @xmath492 and @xmath493 , we have @xmath482 .
+ we take all sub - vectors @xmath495 of @xmath454 such that for @xmath463 , * @xmath496 if @xmath483 with @xmath492 , @xmath493 , and @xmath482 , and * @xmath497 otherwise .
+ then we assign @xmath464:=1 $ ] if there exists such @xmath498 where @xmath499=1 $ ] , and @xmath464:=0 $ ] otherwise .
this formula holds because @xmath431 contains a subgraph @xmath475 if and only if @xmath500 contains a subgraph @xmath501 where @xmath475 is obtained from @xmath501 by adding edges between @xmath502)$ ] and @xmath503)$ ] .
+ note that there are at most @xmath504 possible sub - vectors .
thus , we need @xmath505 iterations to compute @xmath488 $ ] for all @xmath52 .
4 . ( renaming label @xmath398 to @xmath401 )
if there is @xmath506 with @xmath507 , then we assign @xmath488:=0 $ ] .
this correctly assigns the value as there is no vertex with label @xmath398 at this node .
we can check it in time @xmath508 .
we assume that the given vector @xmath52 satisfies that for all @xmath506 , @xmath509 .
+ for each vector @xmath454 , we take all sub - vectors @xmath510 of @xmath454 such that for @xmath506 , * @xmath511 if @xmath512 , and * @xmath513 otherwise .
+ we assign @xmath464:=1 $ ] if there exists such @xmath498 where @xmath499=1 $ ] , and @xmath464:=0 $ ] otherwise .
this formula trivially holds as we just change the labels on vertices .
+ note that there are @xmath514 possible vectors .
thus , we need @xmath515 iterations to compute @xmath488 $ ] for all @xmath52 .
finally , we can test whether @xmath2 contains an induced subgraph isomorphic to @xmath56 by checking whether there is a complete vector @xmath52 such that @xmath452=1 $ ] .
thus , we can find an induced subgraph of @xmath2 isomorphic to @xmath56 in time @xmath427 , if exists .
also , if we find such a vector @xmath52 , then we can track which vertices are contributed to make the vector @xmath52 , and thus , we can compute a vertex set inducing @xmath56 in the same time complexity .
now , we give a second fpt algorithm for the lrw1-vertex deletion problem .
[ thm : algorithm2 ] the lrw1-vertex deletion problem can be solved in @xmath7 time .
all procedures in the algorithm will be same as in theorem [ thm : algorithm1 ] except the refined branching algorithm .
let @xmath390 be an instance where @xmath2 is a graph on @xmath1 vertices . using the algorithm in theorem [ thm : approxcw ]
, we can decide whether the rankwidth of @xmath2 is at most @xmath38 in time @xmath7 .
if the rankwidth of the input graph is more than @xmath38 , then we can not make it a graph of linear rankwidth at most @xmath0 by removing at most @xmath3 vertices .
thus , we say that it is a no - instance in this case .
otherwise , the algorithm outputs a @xmath40-expression @xmath430 of @xmath2 .
let @xmath170 be the labelled rooted tree induced by @xmath430 .
for convenience let @xmath516 .
let us fix a graph @xmath56 in @xmath269 .
note that @xmath517 . using the algorithm in proposition [ prop : inducedoncw ]
, we can test whether @xmath2 contains an induced subgraph isomorphic to @xmath56 and outputs a vertex set inducing @xmath56 if exists , in time @xmath518 .
if there is such a vertex set @xmath47 , then branch by removing one of the vertices @xmath52 in @xmath47 and decrease @xmath3 by @xmath0 . note that we can obtain the cliquewidth expression for @xmath519 from @xmath430 just by removing the node introducing @xmath52 .
we recurse this branching algorithm until there are no such vertex sets @xmath47 . combining the remaining steps described in the proof of theorem [ thm : algorithm1 ] ,
it is easy to verify that the lrw1-vertex deletion problem can be solved in time @xmath7 .
in this section , we show that the algorithm in theorem [ thm : main1 ] is best possible in some sense . our lower bound is based on a well - known complexity hypothesis formulated by impagliazzo , paturi , and zane @xcite . *
exponential time hypothesis ( eth ) .
* there is a constant @xmath520 such that 3-cnf - sat with @xmath1 variables and @xmath521 clauses can not be solved in time @xmath522 .
we use the known lower bound for the vertex cover problem .
[ thm : vclow ] there is no @xmath8-time algorithm for vertex cover , unless eth fails .
we show the following .
there is no @xmath8-time algorithm for lrw1 vertex deletion , unless eth fails .
for contradiction , suppose there exists an algorithm for solving the lrw1-vertex deletion problem in time @xmath8 .
let @xmath390 be an instance of the vertex cover problem .
we construct a graph @xmath394 from @xmath2 as follows : 1 . for every vertex @xmath281 ,
add a pendant vertex @xmath498 adjacent to @xmath52 , and 2 . for every edge @xmath198 in @xmath2 , we replace it with two vertex disjoint paths of length 2 from @xmath52 to @xmath25 .
let @xmath394 be the resulting graph .
note that for each edge @xmath198 in @xmath2 , @xmath523 and two disjoint paths of length @xmath72 from @xmath52 to @xmath25 in @xmath394 form an induced subgraph isomorphic to @xmath311 .
we have @xmath524 .
we claim that @xmath2 has a vertex set @xmath47 of size at most @xmath3 such that @xmath167 has no edges if and only if @xmath394 has a lrw1-deletion set of size at most @xmath3 .
suppose that @xmath2 has a vertex set @xmath47 of size at most @xmath3 such that @xmath167 has no edges .
it is easy to confirm that @xmath525 is a disjoint union of stars , which has linear rankwidth at most @xmath0 .
for the converse direction , suppose that @xmath394 has a lrw1-deletion set @xmath47 of size at most @xmath3 .
if @xmath47 contains a vertex of degree @xmath0 , then we can replace it with its neighbor .
we may assume that @xmath47 has no pendant vertices of @xmath394 .
let @xmath198 be an edge of @xmath2 and @xmath526 be the vertices of degree @xmath72 that are adjacent to @xmath52 and @xmath25 in @xmath394 .
as @xmath47 does not contain pendant vertices of @xmath394 , we have @xmath527 , otherwise , @xmath394 contains an induced subgraph isomorphic to @xmath311 . for each edge @xmath198 , if @xmath47 contains one of @xmath526 , then we replace it with one of @xmath52 and @xmath25
. let @xmath528 be the resulting set .
then @xmath529 and @xmath530 contains at least one of @xmath52 and @xmath25 for each edge @xmath198 of @xmath2 .
we conclude that @xmath2 has a vertex set @xmath47 of size at most @xmath3 such that @xmath167 has no edges .
therefore , using the algorithm for the lrw1-vertex deletion problem , we can decide whether @xmath390 is a yes - instance of the vertex cover problem , in time @xmath8 , which is not possible unless eth fails by theorem [ thm : vclow ] .
in this section , we show the following .
[ thm : main22 ] the lrw1-vertex deletion problem has a kernel with @xmath34 vertices .
we use the sunflower lemma to find a hitting set for obstructions of small size with a special property .
it consists in finding a subset @xmath170 with size bounded by @xmath531 for some function @xmath5 whose removal turns @xmath2 into a graphs of linear rankwidth at most @xmath0 with the property that for every set @xmath58 of size at most @xmath3 , the following are equivalent ( lemma [ lem : shrink ] ) : * @xmath47 is a minimal vertex set such that @xmath167 has no obstructions in @xmath269 .
* @xmath47 is a minimal vertex set such that @xmath532\setminus s$ ] has no obstructions in @xmath269 .
this property implies that if there is a minimal lrw1-deletion set @xmath47 , then each vertex of @xmath533 should be used to remove at least one long induced cycle .
it will be used to find an irrelevant vertex in a large thread block in @xmath534 .
moreover , we can preprocess the instance so that there is no small obstruction containing exactly one vertex of @xmath170 .
this will be used to bound the length of the sequence of thread blocks in each connected component .
let @xmath396 be an instance of lrw@xmath0-vertex deletion .
we start with an easy reduction rule .
[ rule : threadcomponent ] if @xmath2 has a connected component that has linear rankwidth at most @xmath0 , then we remove it from @xmath2 .
let @xmath171 be a family of subsets over a set @xmath535 .
a subset @xmath536 is called _ a hitting set _ of @xmath171 if for every set @xmath537 , @xmath538 . for a graph @xmath2 and a family of graphs @xmath171 , a set @xmath58
is also called _ a hitting set _ for @xmath171 if for every induced subgraph @xmath56 of @xmath2 that is isomorphic to a graph in @xmath171 , @xmath539 .
the following lemma can be obtained from the sunflower lemma .
[ lem : fomin ] let @xmath171 be a family of sets of size at most @xmath540 over a set @xmath535 , and let @xmath3 be a positive integer .
then in time @xmath541 , we can find a nonempty set @xmath542 such that 1 .
for every @xmath536 of size at most @xmath3 , @xmath543 is a minimal hitting set of @xmath171 if and only if @xmath543 is a minimal hitting set of @xmath544 , and 2 .
@xmath545 .
using proposition [ prop : nltothread2 ] and lemma [ lem : fomin ] , we identify a vertex set @xmath170 of @xmath2 with size polynomial in @xmath3 that allows us to forget about small obstructions .
[ lem : shrink ] let @xmath390 be an instance of lrw@xmath0-vertex deletion .
there is a polynomial time algorithm that either concludes that @xmath390 is a no - instance or finds a non - empty set @xmath546 such that 1 .
@xmath534 has linear rankwidth at most @xmath0 , 2 . for every set @xmath58 of size at most @xmath3 ,
@xmath47 is a minimal hitting set for @xmath269 in @xmath2 if and only if it is a minimal hitting set for @xmath269 contained in @xmath532 $ ] , and 3 .
@xmath547 .
let @xmath171 be the set of vertex sets @xmath47 of @xmath2 such that @xmath48 $ ] is isomorphic to a graph in @xmath269 .
since the maximum size of a set in @xmath59 is @xmath35 , using lemma [ lem : fomin ] , we can find a subset @xmath548 of @xmath59 such that 1 . for every vertex subset @xmath549 of size at most @xmath3
, @xmath550 is a minimal hitting set of @xmath171 if and only if @xmath550 is a minimal hitting set of @xmath544 , and 2 .
@xmath551 .
let @xmath552 . from the condition 1
, @xmath553 has no induced subgraph isomorphic to a graph in @xmath269 and by theorem [ thm : mainlrw ] , @xmath553 is a necklace graph . using the algorithm in proposition [ prop : nltothread2 ]
, we can find a minimum lrw1-deletion set @xmath554 of @xmath553 in polynomial time .
if @xmath555 , then we conclude that @xmath390 is a no - instance .
otherwise , we add @xmath554 to @xmath556 , increasing its size by at most @xmath3 .
we conclude that @xmath557 is a required set .
let us fix a subset @xmath170 of @xmath42 obtained by lemma [ lem : shrink ] .
we preprocess using the following reduction rule .
[ rule : onevertex ] let @xmath558 such that for every @xmath559 , there exists an induced subgraph @xmath56 of @xmath2 isomorphic to a graph in @xmath269 with @xmath560 .
if @xmath561 , then @xmath396 is a no - instance ; otherwise , remove @xmath535 from @xmath2 and reduce @xmath3 by @xmath562 , and use @xmath563 instead of @xmath170 .
it can be done in polynomial time because we only need to look at obstructions of @xmath269 in @xmath2 .
[ lem : oneintersect ] reduction rule [ rule : onevertex ] is safe .
we claim that every minimal lrw1-deletion set in @xmath2 contains @xmath535 .
let @xmath47 be a minimal lrw1-deletion set in @xmath2 .
then there exists a vertex subset @xmath564 such that @xmath528 is a minimal hitting set for graphs of @xmath269 in @xmath532 $ ] . from the property of @xmath170
, @xmath528 is also a minimal hitting set for graphs of @xmath269 in @xmath2 , and we must have @xmath565 as @xmath528 hits the sets @xmath566 for each @xmath559 .
it implies that if @xmath567 , then @xmath396 is a no - instance . otherwise , since @xmath535
is always contained in any minimal lrw1-deletion set of @xmath2 , we have that @xmath390 is a yes - instance if and only if @xmath568 is a yes - instance . from now on
, we assume that @xmath2 is reduced under reduction rules [ rule : threadcomponent ] and [ rule : onevertex ] .
a vertex @xmath52 of @xmath2 is called _ irrelevant _ if @xmath396 is a yes - instance if and only if @xmath569 is a yes - instance . for convenience ,
let @xmath570 .
we first show that if a thread block in @xmath534 is large , then we can always find an irrelevant vertex in there .
[ prop : irrelevant ] if @xmath534 contains a thread block of size at least @xmath571 , then we can find an irrelevant vertex in polynomial time .
we use the following lemma .
[ lem : diamond ] let @xmath2 be a graph and let @xmath572 be an induced path of length @xmath158 in @xmath2 .
if two distinct vertices @xmath573 in @xmath574 are adjacent to @xmath307 and @xmath308 , then @xmath309 contains an induced subgraph isomorphic to a graph in @xmath269 .
see figure [ fig : lemma65 ] for the following cases . if @xmath70 is adjacent to @xmath575 but not adjacent to @xmath576 , then @xmath577 is an induced path of length @xmath73 and @xmath575 is adjacent to its end vertices . by lemma [ lem : pathtoobs ] , @xmath578 $ ] has an induced subgraph isomorphic to a graph in @xmath269 .
considering all symmetric cases , we may assume that for each @xmath579 , @xmath52 is adjacent to both @xmath573 or neither of them .
depending on the adjacency between @xmath580 and @xmath581 , and the adjacency between @xmath575 and @xmath576 , we have one of the @xmath582 graphs in @xmath269 , which are @xmath583 . .
] suppose that @xmath534 contains a thread block of size at least @xmath571 .
we can find such a thread block @xmath584 and compute its ordering @xmath89 and labeling @xmath102 in polynomial time using the algorithm in lemma [ lem : splittreetothreadblock ] .
let @xmath585 be the first and last vertices of @xmath584 and let @xmath586 be the ordering obtained from @xmath89 by removing @xmath46 and @xmath110 .
we mark some vertices of @xmath584 as follows .
we set @xmath587 . 1 . for each vertex @xmath52 of @xmath170
, choose the first @xmath588 vertices @xmath589 of @xmath586 that are neighbors of @xmath52 with @xmath590 , and choose the last @xmath588 vertices @xmath589 of @xmath586 that are neighbors of @xmath52 with @xmath591 , and add them to @xmath592 .
2 . for each pair of two vertices @xmath52 , @xmath498 in @xmath170 , choose @xmath588 common neighbors of @xmath52 and @xmath498 in @xmath584 , and add them to @xmath592 .
3 . choose the first @xmath588 vertices @xmath589 of @xmath586 with @xmath590 , and choose the last @xmath588 vertices @xmath589 of @xmath586 with @xmath591 , and add them to @xmath592 . in each case ,
if there are at most @xmath38 such vertices , then we add all of them to @xmath592 .
then @xmath593 since @xmath594 , there exists a vertex @xmath25 in @xmath595 .
we claim that @xmath25 is an irrelevant vertex .
if @xmath390 is a yes - instance , then @xmath596 is clearly a yes - instance .
suppose that there is a vertex set @xmath597 with @xmath598 such that @xmath599 is a thread graph .
we may assume that @xmath600 is not a thread graph .
so , @xmath600 has an induced subgraph containing @xmath25 that is isomorphic to a graph in @xmath164 .
let @xmath601 be a minimal hitting set for @xmath269 in @xmath2 . from the property of the set @xmath170
, @xmath83 is a minimal hitting set for @xmath269 in @xmath532 $ ] , which implies that @xmath602 .
thus @xmath600 must have an induced cycle of length at least @xmath270 that contains @xmath25 .
let @xmath266 be an induced cycle of length at least @xmath270 containing @xmath25 in @xmath600 .
we will find an induced subgraph of @xmath599 that is isomorphic to a graph in @xmath269 , which leads to a contradiction .
let @xmath603 be the consecutive vertices on @xmath266 . to apply lemma [ lem : diamond ]
, we find two vertices that are adjacent to @xmath307 and @xmath71 .
these cases are depicted in figure [ fig : irrelevant ] . 1 .
@xmath604 . )
since @xmath307 and @xmath71 have a common neighbor @xmath25 in @xmath605 , @xmath592 contains at least @xmath588 common neighbors of @xmath307 and @xmath71 . since @xmath598 , there exist two vertices @xmath606 that are common neighbors of @xmath307 and @xmath71 .
2 . ( case 2 .
one of @xmath307 and @xmath71 is contained in @xmath170 . ) from the symmetry , we may assume that @xmath607 and @xmath608 .
since @xmath609 , @xmath71 is contained in @xmath584 .
if @xmath157 and @xmath610 , then @xmath592 contains the first @xmath588 vertices @xmath589 of @xmath586 that are neighbors of @xmath307 with @xmath590 .
we choose two vertices of them that are not in @xmath550 . in case
when @xmath98 and @xmath611 , we use the last @xmath588 vertices @xmath589 of @xmath586 that are neighbors of @xmath307 with @xmath591 to identify two vertices similarly .
3 . ( case 3 .
neither @xmath307 nor @xmath71 is contained in @xmath170 . ) since @xmath609 , @xmath307 and @xmath71 are contained in @xmath584 .
if @xmath612 , then @xmath613 , @xmath614 and it implies that @xmath615 , which is a contradiction . also , @xmath616 can not happen
thus , both of @xmath307 and @xmath71 appear either before @xmath25 in @xmath89 or after @xmath25 in @xmath89 . by the symmetry
, we may assume that @xmath307 and @xmath71 appear before @xmath25 in @xmath89 .
so , @xmath613 , @xmath617 , and @xmath98 .
since @xmath592 contains the last @xmath588 vertices @xmath589 of @xmath586 with @xmath591 , there exist two vertices @xmath573 from those @xmath588 vertices that are not in @xmath550 and @xmath266 . in all cases , @xmath599 has an induced subgraph isomorphic to a graph in @xmath269 by lemma [ lem : diamond ] .
it contradicts to the assumption that @xmath618 is a lrw1-deletion set of @xmath2 .
therefore , @xmath600 is a thread graph , and we conclude that @xmath396 is a yes - instance . in the next lemma , we show that if a vertex @xmath52 in @xmath170 has neighbors on @xmath619 distinct blocks in a connected component of @xmath534 , then we can find a subgraph @xmath56 isomorphic to one of @xmath620 such that @xmath621 , which is not possible by reduction rule [ rule : onevertex ] .
using it , whenever a connected component of @xmath534 has a long sequence of thread blocks , we can identify a sequence of consecutive thread blocks not touched by any obstruction in @xmath269 .
this allows us to contract one of these thread blocks to a vertex .
[ lem : reducelength ] if @xmath534 has a connected component with at least @xmath622 thread blocks , then we can in polynomial time transform @xmath2 into a graph @xmath394 with @xmath623 such that @xmath396 is a yes - instance if and only if @xmath624 is a yes - instance .
suppose that @xmath534 has a connected component @xmath56 such that @xmath56 consists of at least @xmath625 thread blocks .
we can find such a connected component in polynomial time using the algorithm in lemma [ lem : splittreetothreadblock ] , if exists .
let @xmath626 be the sequence of thread blocks of @xmath56 .
we claim that every vertex @xmath52 of @xmath170 has neighbors in at most @xmath582 thread blocks of @xmath56 . for contradiction ,
suppose that there is a vertex @xmath52 in @xmath170 having neighbors in at least @xmath619 thread blocks .
then there are three thread blocks @xmath627 having a neighbor of @xmath52 in @xmath2 such that @xmath628 and @xmath629 . for each @xmath630
, let @xmath631 be a neighbor of @xmath52 in @xmath632 . since each thread block consists of at least two vertices , we can choose a neighbor @xmath633 of @xmath631 in @xmath632 for each @xmath630 . depending on the adjacency between @xmath52 and the vertices @xmath634 , we have an induced subgraph isomorphic to a graph in @xmath635 such that it has exactly one vertex of @xmath170 .
this contradicts to the assumption that @xmath396 is an instance reduced by reduction rule [ rule : onevertex ] .
now , for each vertex @xmath52 of @xmath170 , we mark the thread blocks @xmath584 of @xmath56 containing a neighbor of @xmath52 .
since the number of thread blocks in @xmath56 is at least @xmath636 and @xmath637 , there exist @xmath638 and @xmath639 such that @xmath640 are non - marked thread blocks .
let @xmath641 be the two end vertices of @xmath642 .
we transform the graph @xmath2 into a graph @xmath394 by removing the thread block @xmath643 and adding a new vertex @xmath589 that are adjacent to @xmath644 .
let @xmath475 be the connected component of @xmath645 that is modified from the connected component @xmath56 of @xmath534 .
since we remove at least two vertices from @xmath2 and add one vertex , we have @xmath623 .
we show that @xmath396 is a yes - instance if and only if @xmath646 is a yes - instance .
suppose that @xmath2 has a minimal thread vertex set @xmath550 .
we first assume that @xmath550 contains a vertex @xmath647 in @xmath643 .
since @xmath550 is a minimal lrw1-deletion set and all small obstructions of @xmath269 are contained in @xmath648 , @xmath647 hits an induced cycle of length at least @xmath270 in @xmath2 , and the cycle must pass through the vertices @xmath46 and @xmath110 .
thus , @xmath649 is a lrw1-deletion set of @xmath394 with size at most @xmath650 .
assume that @xmath651 .
we may assume that @xmath652 is not a thread graph .
then @xmath652 must have an induced cycle @xmath266 of length at least @xmath270 intersecting the new vertex @xmath589 .
the cycle obtained from @xmath266 by replacing @xmath589 with the edge @xmath107 is also an induced cycle of length at least @xmath270 in @xmath600 .
it contradicts to the assumption that @xmath600 is a thread graph .
now suppose that @xmath394 has a minimal lrw1-deletion set @xmath550 .
if @xmath653 , then @xmath589 hits an induced cycle of length at least @xmath270 in @xmath394 because of the minimality of @xmath550 and the distance from @xmath46 to the end vertices of @xmath654 and @xmath655 . because @xmath46 hits all induced cycles of length at least @xmath270 in @xmath2 having a vertex of @xmath643 , @xmath656 is again a lrw1-deletion set of @xmath2 .
assume that @xmath657 .
we may assume that @xmath600 is not a thread graph .
so , @xmath600 has an induced cycle @xmath266 of length at least @xmath270 passing through @xmath46 and @xmath110 .
let @xmath658 be the cycle obtained from @xmath266 by replacing the edge @xmath107 with the vertex @xmath589 .
this cycle @xmath658 clearly exists in @xmath652 and it has length at least @xmath270 because it should contain at least one vertex from the thread blocks @xmath659 with @xmath660 .
this contradicts to the assumption that @xmath652 is a thread graph .
we conclude that @xmath390 is a yes - instance if and only if @xmath646 is a yes - instance .
we bound the number of connected components using the following lemma .
[ lem : reducecomponent1 ] 1 .
the graph @xmath534 has at most @xmath661 connected components containing at least two vertices .
2 . if @xmath534 has at least @xmath662 isolated vertices , then we can find an irrelevant vertex in polynomial time .
\(1 ) by reduction rule [ rule : threadcomponent ] , each connected component @xmath56 of @xmath534 contains a vertex that has a neighbor in @xmath170 .
let @xmath663 be the set of connected components of @xmath534 which consist of at least two vertices , and suppose that @xmath664 .
since every connected component of @xmath56 has a vertex having a neighbor in @xmath170 , there exists a vertex @xmath665 such that @xmath175 has neighbors in three distinct connected components of @xmath663 .
since each connected component of @xmath663 has at least two vertices , @xmath2 has a vertex set @xmath47 where @xmath48 $ ] is isomorphic to a graph in @xmath635 and @xmath666 .
it contradicts to the assumption that @xmath396 is reduced by reduction rule [ rule : onevertex ] .
\(2 ) suppose that @xmath534 has at least @xmath662 isolated vertices .
let @xmath47 be the union of isolated vertices in @xmath534 .
we may assume that every vertex in @xmath47 has a neighbor in @xmath170 .
we define a set @xmath592 to identify an irrelevant vertex . for each pair of two vertices in @xmath170 , choose @xmath588 common neighbors in @xmath47 , and add them to @xmath592 .
if there are at most @xmath38 common neighbors , then we add all of them into @xmath592 .
since @xmath667 , there is a vertex @xmath25 in @xmath668 .
we claim that @xmath25 is an irrelevant vertex of the problem .
if @xmath396 is a yes - instance , then there exists a vertex subset @xmath550 of size at most @xmath3 in @xmath2 such that @xmath600 is a thread graph . since @xmath599 is also a thread graph
, @xmath596 is a yes - instance .
suppose that @xmath596 is a yes - instance .
we choose a minimal vertex set @xmath550 in @xmath669 such that @xmath598 and @xmath599 is a thread graph .
we may assume that @xmath600 is not a thread graph .
let @xmath601 be a hitting set for @xmath269 in @xmath532 $ ] .
then by the property of @xmath170 , @xmath83 also hits all induced subgraphs in @xmath2 that are isomorphic to a graph of @xmath269 .
since @xmath550 already hits all small obstructions in @xmath2 , there exists an induced cycle @xmath266 of length at least @xmath270 in @xmath600 containing @xmath25 .
let @xmath573 be the neighbors of @xmath25 on the cycle @xmath266 . since @xmath573 have @xmath588 common neighbors in @xmath592 , we may choose two vertices @xmath670 that are common neighbors of @xmath575 and @xmath576 . by lemma [ lem : diamond ]
, we have that @xmath599 has an induced subgraph isomorphic to a graph in @xmath269 , which implies that @xmath599 is not a thread graph .
it is a contradiction , and we conclude that @xmath390 is a yes - instance .
let us now piece everything together and analyze the kernel size .
let @xmath396 be an instance of lrw1-vertex deletion . by reduction rule [ rule :
threadcomponent ] , we may safely assume that @xmath2 has no connected components that are thread graphs . then using the algorithm in lemma [ lem : shrink ] , in polynomial time ,
either we conclude that @xmath390 is a no - instance or find a non - empty set @xmath546 stated in lemma [ lem : shrink ] .
we apply reduction rule [ rule : onevertex ] .
[ lem : oneintersect ] guarantees that for every vertex set @xmath58 such that @xmath48 $ ] is isomorphic to a graph in @xmath635 , @xmath671 . combining proposition [ prop : irrelevant ] and lemma [ lem : reducelength ]
, we can assume that every connected component of @xmath534 has size at most @xmath672 ( otherwise the instance can be reduced in polynomial time ) .
note that for each connected component @xmath56 of @xmath534 , there exists a vertex in @xmath56 that has a neighbor in @xmath170 .
therefore , by lemma [ lem : reducecomponent1 ] , the number of non - trivial components of @xmath534 is at most @xmath661 and the number of isolated vertices in @xmath534 is at most @xmath673 .
it follows that @xmath674
we consider the problem linear rankwidth-@xmath25 vertex deletion when @xmath675 .
a next step is to investigate the problem for bigger @xmath25 , or for any fixed @xmath25 .
a closely related problem is rankwidth-@xmath25 vertex deletion , which asks whether @xmath2 has a vertex subset of size at most @xmath3 such that @xmath167 has rankwidth at most @xmath25 .
( linear ) rankwidth-@xmath25 vertex deletion is fixed - parameter tractable for the following reason .
note that any yes - instance has rankwidth at most @xmath676 .
having bounded ( linear ) rankwith can be characterized by a finite list of forbidden vertex - minors @xcite . from @xcite , having a vertex - minor can be expressed in @xmath677 , i.e. , monadic second order logic without edge set quantification where we can express the parity of @xmath678 for a vertex set @xmath550 .
fixed - parameter tractability follows as a consequence of courcelle , makowsky , rotics @xcite . as for rankwidth
, this result can be turned into a constructive algorithm as @xcite provides an explicit upper bound on the size of vertex - minor obstructions for rankwidth at most @xmath3 for fixed @xmath3 . however , the exponential blow - up in the running time is huge with respect to both @xmath25 and @xmath3 .
it is a challenging question whether a reasonable dependency on @xmath3 can be achieved .
a single - exponential time would be ideal , which was achievable for its treewidth counterpart .
a first realistic goal is to consider the case when @xmath675 , that is , the distance - hereditary vertex deletion .
we leave it as an open question whether this problem can be solved in time @xmath679 time for some constant @xmath13 . for linear rankwidth
, there is no known upper bound on the size of vertex - minor obstructions for linear rankwidth at most @xmath3 , and thus , obtaining such an upper bound is an interesting open question .
the third author would like to thank sang - il oum for suggesting the refined branching algorithm using cliquewidth .
i. adler , m. m. kant , and o. kwon .
linear rank - width of distance - hereditary graphs . in _
graph - theoretic concepts in computer science - 40th international workshop , wg 2014 , nouan - le - fuzelier , france , june 25 - 27 , 2014 . revised selected papers _ ,
pages 4255 , 2014 .
unit interval editing is fixed - parameter tractable . in m.
m. halldrsson , k. iwama , n. kobayashi , and b. speckmann , editors , _ automata , languages , and programming _ , volume 9134 of _ lecture notes in computer science _ , pages 306317 .
springer berlin heidelberg , 2015 .
v. chvtal and p. l. hammer .
aggregation of inequalities in integer programming . in _ studies in integer programming ( proc .
workshop , bonn , 1975 ) _ , pages 145162 . ann . of discrete math .
1 . north - holland , amsterdam , 1977 .
f. v. fomin , d. lokshtanov , n. misra , and s. saurabh .
planar f - deletion : approximation , kernelization and optimal fpt algorithms . in _
53rd annual ieee symposium on foundations of computer science , focs 2012 , new brunswick , nj , usa , october 20 - 23 , 2012 _ , pages 470479 , 2012 . s. gaspers and s. szeider .
backdoors to satisfaction . in _ the multivariate algorithmic revolution and beyond - essays dedicated to michael r. fellows on the occasion of his 60th birthday _ , pages 287317 , 2012 .
j. jeong , e. j. kim , and s. oum .
constructive algorithm for path - width of matroids . in r.
krauthgamer , editor , _ proceedings of the twenty - seventh annual acm - siam symposium on discrete algorithms , soda 2016 , arlington , va , usa , january 10 - 12 , 2016 _ , pages 16951704 .
siam , 2016 .
m. m. kant , e. j. kim , o. kwon , and c. paul . . in t.
husfeldt and i. kanj , editors , _ 10th international symposium on parameterized and exact computation ( ipec 2015 ) _ , volume 43 of _ leibniz international proceedings in informatics ( lipics ) _ , pages 138150 , dagstuhl , germany , 2015 .
leibniz - zentrum fuer informatik .
e. j. kim , a. langer , c. paul , f. reidl , p. rossmanith , i. sau , and s. sikdar .
linear kernels and single - exponential algorithms via protrusion decompositions . in _
automata , languages , and programming - 40th international colloquium , icalp 2013 , riga , latvia , july 8 - 12 , 2013 , proceedings , part i _ , pages 613624 , 2013 .
g. philip , v. raman , and y. villanger . a quartic kernel for pathwidth - one vertex deletion . in d.
thilikos , editor , _ graph theoretic concepts in computer science _ ,
volume 6410 of _ lecture notes in computer science _ ,
pages 196207 .
springer berlin heidelberg , 2010 .
r. van bevern , c. komusiewicz , h. moser , and r. niedermeier .
measuring indifference : unit interval vertex deletion . in
_ graph - theoretic concepts in computer science _ , volume 6410 of _ lecture notes in comput
_ , pages 232243 .
springer , berlin , 2010 . | _ linear rankwidth _ is a linearized variant of rankwidth , introduced by oum and seymour [ approximating clique - width and branch - width . , 96(4):514528 , 2006 ] .
motivated from recent development on graph modification problems regarding classes of graphs of bounded treewidth or pathwidth , we study the linear rankwidth-@xmath0 vertex deletion problem ( shortly , lrw1-vertex deletion ) . in the lrw1-vertex deletion problem , given an @xmath1-vertex graph @xmath2 and a positive integer @xmath3 , we want to decide whether there is a set of at most @xmath3 vertices whose removal turns @xmath2 into a graph of linear rankwidth at most @xmath0 and find such a vertex set if one exists . while the meta - theorem of courcelle , makowsky , and rotics implies that lrw1-vertex deletion can be solved in time @xmath4 for some function @xmath5 , it is not clear whether this problem allows a running time with a modest exponential function .
we first establish that lrw1-vertex deletion can be solved in time @xmath6 .
the major obstacle to this end is how to handle a long induced cycle as an obstruction . to fix this issue
, we define _ necklace graphs _ and investigate their structural properties .
later , we reduce the polynomial factor by refining the trivial branching step based on a cliquewidth expression of a graph , and obtain an algorithm that runs in time @xmath7 . we also prove that the running time can not be improved to @xmath8 under the exponential time hypothesis assumption .
lastly , we show that the lrw1-vertex deletion problem admits a polynomial kernel . |
shortly after the discovery that blazars can be strong sources of high energy @xmath0-rays , we suggested that very high energy @xmath0-ray beams from blazars can be used to measure the intergalactic infrared radiation field , since pair - production interactions of @xmath0-rays with intergalactic ir photons will attenuate the high - energy ends of blazar spectra ( stecker , de jager & salamon 1992 ) . in recent years , this concept has been used successfully to place upper limits on the the intergalactic ir field ( iirf ) ( stecker & de jager 1993 ; dwek & slavin 1994 ; stecker & de jager 1997 ; stanev & franceschini 1997 ; biller _ et al .
determining the iirf , in turn , allows us to model the evolution of the galaxies which produce it . as energy thresholds are lowered in both existing and planned ground - based air cherenkov light detectors , cutoffs in the @xmath0-ray spectra of more distant blazars are expected , owing to extinction by the iirf .
these can be used to explore the redshift dependence of the iirf . on the other hand , by modelling the intergalactic infrared and optical radiation fields as a function of redshift , one can calculate the expected @xmath0-ray opacity as a function of redshift and comparison of predicted absorption with the data can yield important new information about the evolution of the iirf ( stecker & de jager 1998 ; salamon & stecker 1998 ) .
exploring the iirf with high energy @xmath0-rays has both advantages and disadvantages over direct infrared observations .
an important advantage is that since the absorption process only acts effectively in intergalactic space over large distances , the @xmath0-ray approach does not suffer from problems of subtraction of foreground emission from both zodiacal light and interstellar dust in the galaxy .
as we have seen at this meeting , these can be formidable sources of uncertainty in determining the iirf .
also , high energy @xmath0-ray observations of blazar and @xmath0-ray burst spectra at high redshifts can be used to probe the past evolution of the iirf whereas direct infrared observations can only tell us about the present iirf . on the other hand ,
the disadvantage of the @xmath0-ray approach is the uncertainty in the unabsorbed spectrum of the source .
this uncertainty is somewhat ameliorated by the strong energy dependence of the absorption effect itself .
therefore , it is my contention that using both direct infrared and high energy @xmath0-ray observations in a `` synoptic '' approach can yield the most information about the iirf .
the formulae relevant to absorption calculations involving pair - production are given and discussed in stecker , de jager & salamon ( 1992 ) , where we derived the absorption formulae with cosmological and redshift effects included .
for @xmath0-rays in the tev energy range , the pair - production cross section is maximized when the soft photon energy is in the infrared range : @xmath1 where @xmath2 is the compton wavelength of the electron . for 15 tev @xmath0-rays
, absorption will occur primarily by interactions with mid - infrared photons having a wavelength @xmath3 20@xmath4 m .
( pair - production interactions actually take place with photons over a range of wavelengths around the optimal value as determined by the energy dependence of the cross section ; see eq . (
if the emission spectrum of an extragalactic source extends beyond 20 tev , then the extragalactic infrared field should cut off the _ observed _ spectrum between @xmath5 gev and @xmath6 tev , depending on the redshift of the source ( stecker & de jager 1998 ; salamon & stecker 1998 ) .
stecker & de jager ( 1998 ) ( hereafter sd98 ) have calculated the absorption coefficient of intergalactic space using a new , empirically based calculation of the spectral energy distribution ( sed ) of intergalactic low energy photons by malkan & stecker ( 1998 ) ( hereafter ms98 ) obtained by integrating luminosity dependent infrared spectra of galaxies over their luminosity and redshift distributions . after giving their results on the @xmath0-ray optical depth as a function of energy and
redshift out to a redshift of 0.3 , stecker & de jager ( 1998 ) ( sd98 ) applied their calculations by comparing their results with the spectral data on mrk 421 ( mcenery _ et al .
_ 1997 ) and spectral data on mrk 501 ( aharonian , _ et al .
_ 1997 ) .
sd98 made the reasonable simplifying assumption that the iirf is basically in place at a redshifts @xmath7 0.3 , having been produced primarily at higher redshifts ( madau 1995 ; salamon & stecker 1998 ) .
therefore sd98 limited their calculations to @xmath8 .
sd98 assumed for the iirf , two of the spectral energy distributions ( seds ) given in ms98 ( shown in figure 1 ) .
the lower curve in figure 1 ( adapted from ms98 ) assumes luminosity evolution proportional to @xmath9 out to @xmath10 , whereas the middle curve assumes such evolution out to @xmath11 .
evolution in stellar emissivity is expected to level off at redshifts greater than @xmath12 ( steidel 1999 ; hopkins , connolley & szalay 2000 ) using these two seds for the iirf , sd98 obtained parametric expressions for the optical depth @xmath13 for @xmath8 , taking a hubble constant of @xmath14 km s@xmath15mpc@xmath15 ( gratton 1997 ) .
the double - peaked form of the sed of the iirf requires a third order polynomial to approximate the opacity @xmath16 in a parametric form .
sd98 give the following approximation : @xmath17\simeq\sum_{i=0}^3a_i(z)(\log_{10}e_{\rm tev})^i \;\;{\rm for}\;\;1.0<e_{\rm tev}<50,\ ] ] where the z - dependent coefficients are given by @xmath18 table 1 gives the numerical values for @xmath19 , with @xmath20 , and @xmath21 .
the numbers before the brackets are obtained using the lower iirf sed shown in figure 1 ; the numbers in the brackets are obtained using the middle iirf sed .
equation ( 2 ) approximates @xmath22 to within 10% for all values of z and e considered .
figure 2 shows the results obtained by sd98 for @xmath13 .
@xmath23 & @xmath24 & @xmath25 & @xmath26 & @xmath27 + 0&1.11(1.46 ) & -0.26 ( 0.10 ) & 1.17(0.42 ) & -0.24 ( 0.07 ) + 1&1.15(1.46 ) & -1.24(-1.03 ) & 2.28(1.66 ) & -0.88(-0.56 ) + 2&0.00(0.15 ) & -0.41(-0.35 ) & 0.78(0.58 ) & -0.31(-0.20 ) + the advantage of using empirical data to construct the sed of the iirf , as done in ms98 , is particularly indicated in the mid - infrared range . in this region of the spectrum ,
galaxy observations indicate more flux from warm dust in galaxies than that taken account of in more theoretically oriented models ( _ e , g , _ macminn & primack ( 1996 ) ) . as a consequence , the mid - infrared `` valley '' between the cold dust peak in the far - infrared and cool star peak in the near ir
is filled in more in the ms98 results and is not as pronounced as in previously derived models of the ir background sed . as can be seen in figure 1 ,
the background sed predicted in ms98 is in excellent agreement with the results obtained from ultradeep _ isocam _ galaxy count observations at 15 @xmath4 m ( altieri _ et al .
_ 1999 ) .
2.0truecm the sd98 calculations predict that intergalactic absorption should only slightly steepen the spectra of mrk 421 and mrk 501 below @xmath3 7 tev , which is consistent with the data in the published literature the sd98 calculations further predict that intergalactic absorption should turn over the spectra of these sources at energies greater than @xmath3 15 tev .
the _ hegra _ group has observed the bl lac object mrk 501 in the flaring phase , obtaining an energy spectrum for this source up to an energy of 24 tev ( aharonian _ et al .
_ 1999 ) . the mrk 501 spectrum obtained by the _
group are well fitted by a source spectrum power - law of spectral index @xmath28 steepened at in the multi - tev energy range by intergalactic absorption , with the optical depth calculated by sd98 ( konopelko , _ et al .
figure 3 clearly shows this .
-3.0truecm = 8.0truein -5.0truecm 1.0truecm [ hegra ]
malkan & stecker ( 2000 ) have used their empirically based model ( ms98 ) to predict infrared luminosity functions and deep infrared galaxy counts at various wavelengths .
they have also examined their predictions for the iirf for comparison with the subsequent determinations from the _ cobe - dirbe _ data analysis . using the formalism of luminosity evolution proportional to @xmath29 out to a redshift of @xmath30 and constant ( no evolution ) for @xmath31
, they find that a comparison of their predictions with current _ iso _ galaxy counts at 15 and 175@xmath4 m favor their baseline model " with @xmath32 and @xmath33 ( the middle curve in figure 1 ) .
the @xmath0-ray limits ( sd97 ) also favor @xmath34 . on the other hand , the _ cobe - dirbe _
far infrared determinations seem to favor a stronger evolution with @xmath35 up to @xmath36 .
for example , the upper curve in figure 1 ( malkan & stecker 2000 ) assumes @xmath37 and @xmath38 .
this curve gives a flux in the mid - infrared which is a factor of @xmath31.8 above the `` baseline model '' and would imply a corresponding increase in the opacity of the universe to multi - tev @xmath0-rays .
( thus , one should multiply the appropriate bracketed numbers in table 1 by a factor of 1.8 ) therefore , @xmath16(15 tev ) would increase from @xmath28 to @xmath39 in the case of mrk 501 and mrk 421 ( @xmath40 ) .
this _ prima facie _ conflict can be resolved in two ways : either ( a ) the _ cobe - dirbe _ far - infrared estimates may suffer from undersubtraction of foreground emission and therefore are too high , or ( b ) the _ isophot _ galaxy counts may be missing a significant fraction of sources . in this later case ,
one may also have require that the @xmath0-ray results are wrong in that the _ hegra _ energy determinations have been overestimated , mimicing the effect of absorption which would be produced by a lower iirf .
another possibility is one involving new physics , _ viz . _
that lorentz invariance may be broken , allowing the universe to be transparent to multi - tev photons ( coleman & glashow 1999 ; kifune 1999 ) .
this `` new physics''scenario presents problems in that the mrk 501 spectrum does exhibit exactly the characteristics expected for high - energy @xmath0-ray absorption from pair - production ( konopelko _ et al .
_ 1999 ) . however
, we will discuss this possibility more quantitatively in the next section ( glashow & stecker 2000 ) .
possibility ( a ) finds support in the independent analysis of the _ cobe _ data by lagache et al .
( 1999 ) who obtain a flux at 140 @xmath4 m which is only 60% of the flux obtained by hauser _
( 1998 ) shown in figure 1 .
et al . _ ( 2000 ) also obtained a smaller flux at 240 @xmath4 m . in this
regard , one should also note that the results reported by hauser _ et al . _
( 1998 ) were at the 4@xmath41 level .
if the far - infrared galaxy counts are incomplete ( possibility ( b ) ) , this would imply stronger evolution in the far - infrared emission of galaxies than in the mid - infrared .
although the ms98 model already includes some differential evolution of this type , based on the data of spinoglio _ et al . _
( 1995 ) , it is conceivable that starburst galaxies at redshifts at @xmath42 might produce an even higher ratio of @xmath43 m to @xmath44 m rest - frame fluxes than their present - day counterparts . radiation at these widely separated wavelengths is known to be emitted by quite different dust grains which could have different evolutionary development , particularly for ulirgs ( ultraluminous infrared galaxies ) and agn ( active galactic nuclei ) .
one should note that _ this possibility can make the @xmath0-ray and infrared data compatible _
, since absorption of @xmath3 15 tev @xmath0-rays is caused by interactions with mid - infrared ( @xmath3 20@xmath4 m ) photons and not 140@xmath4 m far - infrared photons .
with the idea of spontaneous symmetry breaking in particle physics came the suggestion that lorentz invariance ( li ) might be weakly broken at high energies ( sato & tati 1972 ) .
although no real quantum theory of gravity exists , it was suggested that li might be broken as a consequence of such a theory ( amelino - camilia _ et al .
a simpler formulation for breaking li by a small first order perturbation in the electromagnetic lagrangian which leads to a renormalizable treatment has been given by coleman & glashow ( 1999 ) .
the breaking of li at high energies is one way of avoiding the predicted , but unseen , `` gzk '' cutoff in the ultrahigh energy cosmic - ray spectrum owing to photomeson interactions with 2.7k cosmic background photons ( greisen 1966 ; zatsepin & kuzmin 1966 ) , which produces an effective absorption mean - free - path for ultrahigh energy cosmic rays in intergalactic space of @xmath45 mpc ( stecker 1968 ) .
it has recently been suggested that li breaking could also make the universe transparent to high energy @xmath0-rays ( kifune 1999 ) .
such li breaking implies a perferred frame of reference in the universe which would naturally be associated with a rest frame for the 2.7k cosmic background radiation .
we follow here the formalism proposed by coleman & glashow ( 1999 ) for li breaking . within this scenario ,
the maximum attainable velocity of an electron , @xmath46 , the velocity of the photon .
let us consider the case where @xmath47 and we define @xmath48 in this case , electrons above an energy @xmath49 will be superluminal
. they will radiate cherenkov light if their velocity is large enough , _
i.e. _ , if @xmath50 which implies that @xmath51^{-1}$ ] .
this determines the maximum electron energy above which electrons would rapidly lose energy by cherenkov radiation : @xmath52 ( in this section , we adopt the standard paticle physics convention @xmath53 . ) since electrons are seen in the cosmic radiation up to an energy @xmath3 1 tev , this implies that @xmath54 tev which gives an upper limit on @xmath55 of @xmath56 . if li is broken so that @xmath57 , the threshold energy for the pair production process is altered because the square of the four - momentum becomes @xmath58 where @xmath59 is the energy of the low energy ( infrared ) photon and @xmath60 is the angle between the two photons . the second term on the left - hand - side comes from the fact that @xmath61 . for head - on collisions ( @xmath62 )
the minimum low energy photon energy for pair production becomes @xmath63 it follows that the condition for a significant increase in the energy threshold for pair production is @xmath64 or , equivalently , @xmath55 must be greater than @xmath65 .
thus , for a significant decrease in the optical depth to mrk 501 for @xmath66 tev , we must have @xmath67 .
the effect of breaking li is to exclude photons of energy below @xmath68 from pair producing , thus reducing the number of target photons . in this way
, one can exclude the high flux of far - infrared photons implied by the _ cobe - dirbe _ analysis from participating in the absorption process while still allowing mid - infrared photons to produce an absorption feature .
one can rule out this hypothesis is if one observes electrons with energies @xmath69 1 tev .
this would reduce the upper limit on @xmath55 to the point where there will be no significant reduction in absorption by pair - production interactions with infrared photons .
on the other hand , if one observes @xmath0-rays above 100 tev from an extragalactic source , this would be strong evidence for li breaking .
this is because the very large density ( @xmath3 400 @xmath70 ) of 3k cosmic microwave photons would otherwise absorb @xmath71 tev @xmath0-rays within a distance of @xmath3 10 kpc ( see eq .
. finally , we note that even if absorption is reduced in mrk 501 , one can still have absorption in sources at higher redshifts owing to interactions with higher energy photons ( see next section ) .
we now discuss the absorption of 10 to 500 gev @xmath0-rays at high redshifts . in order to calculate such high - redshift absorption properly ,
it is necessary to determine the spectral distribution of the intergalactic low energy photon background radiation as a function of redshift as realistically as possible out to frequncies beyond the lyman limit .
this calculation , in turn , requires observationally based information on the evolution of the spectral energy distributions ( seds ) of ir through uv starlight from galaxies , particularly at high redshifts .
conversely , observations of high - energy cutoffs in the @xmath0-ray spectra of blazars as a function of redshift , which may enable one to separate out intergalactic absorption from redshift - independent cutoff effects , could add to our knowledge of galaxy formation and early galaxy evolution . in this regard
, it should be noted that the study of blazar spectra in the 10 to 300 gev range is one of the primary goals of a next generation space - based @xmath0-ray telescope _ glast _
( gamma - ray large area space telescope ) ( bloom 1996 ) as well as _
veritas _ and other future ground based @xmath0-ray telescopes .
salamon and stecker ( 1998 ) ( hereafter ss98 ) have calculated the @xmath0-ray opacity as a function of both energy and redshift for redshifts as high as 3 by taking account of the evolution of both the sed and emissivity of galaxies with redshift . in order to accomplish this
, they adopted the recent analysis of fall _ et al . _
( 1996 ) and also included the effects of metallicity evolution on galactic seds .
their results indicate that the extragalactic @xmath0-ray background spectrum from blazars should steepen significantly above 20 gev , owing to extragalactic absorption .
future observations of such a steepening would thus provide a test of the blazar origin hypothesis for the @xmath0-ray background radiation .
pei and fall ( 1995 ) have devised a method for calculating stellar emissivity which bypasses the uncertainties associated with estimates of poorly defined luminosity distributions of evolving galaxies .
the core idea of their approach is to relate the star formation rate directly to the evolution of the neutral gas density in damped lyman @xmath72 systems , and then to use stellar population synthesis models to estimate the mean co - moving stellar emissivity @xmath73 of the universe as a function of frequency @xmath74 and redshift @xmath75 .
the ss98 calculation of stellar emissivity closely follows this elegant analysis , with minor modifications
. damped lyman @xmath72 systems are high - redshift clouds of gas whose neutral hydrogen surface density is large enough ( @xmath76 @xmath77 ) to generate saturated lyman @xmath72 absorption lines in the spectra of background quasars that happen to lie along and behind common lines of sight to these clouds .
these gas systems are believed to be either precursors to galaxies or young galaxies themselves , since their neutral hydrogen ( hi ) surface densities are comparable to those of spiral galaxies today , and their co - moving number densities are consistent with those of present - day galaxies ( worthy 1986 ; pei & fall 1995 ) .
it is in these systems that initial star formation presumably took place , so there is a relationship between the mass content of stars and of gas in these clouds ; if there is no infall or outflow of gas in these systems , the systems are `` closed '' , so that the formation of stars must be accompanied by a reduction in the neutral gas content . such a variation in the hi surface densities of lyman @xmath72 systems with redshift is seen , and is used by pei and fall ( 1995 ) to estimate the mean cosmological rate of star formation back to redshifts as large as @xmath78 .
following fall , charlot & pei ( 1996 ) , ss98 used the bruzual - charlot model ( charlot & bruzual 1991 ; bruzual & charlot 1993 ) corresponding to a salpeter stellar initial mass function , @xmath79 , where @xmath80 .
the mean co - moving emissivity @xmath81 was then obtained by convolving over time @xmath82 the specific luminosity with the mean co - moving mass rate of star formation .
ss98 also obtained metallicity correction factors for stellar radiation at various wavelengths .
increased metallicity gives a redder population spectrum ( worthy 1994 ; bertelli , _ et al .
_ 1994 ) .
ss98 calculated stellar emissivity as a function of redshift at 0.28 @xmath4 m , 0.44 @xmath4 m , and 1.00 @xmath4 m , both with and without a metallicity correction .
their results agree well with the emissivity obtained by the canada - french redshift survey ( lilly , _
et al _ 1996 ) over the redshift range of the observations ( @xmath83 ) .
the stellar emissivity in the universe is found to peak at @xmath84 , dropping off steeply at lower reshifts and more slowly at higher redshifts .
indeed , madau , pozzetti & dickinson ( 1998 ) have used observational data from the hubble deep field to show that metal production has a similar redshift distribution , such production being a direct measure of the star formation rate .
the co - moving radiation energy density @xmath85 is the time integral of the co - moving emissivity @xmath73 , @xmath86 where @xmath87 and @xmath88 is the redshift corresponding to initial galaxy formation .
the extinction term @xmath89 accounts for the absorption of ionizing photons by the clumpy intergalactic medium ( igm ) that lies between the source and observer .
although the igm is effectively transparent to non - ionizing photons , the absorption of photons by hi , hei and heii can be considerable ( madau 1995 ) .
ss98 evaluated the co - moving energy density @xmath85 using equation ( 9 ) above .
they then calculated the optical depth for @xmath0-rays owing to electron - positron pair production interactions with photons of the stellar radiation background .
this can be determined from the expression ( stecker , de jager & salamon 1992 ) @xmath90\sigma_{\gamma\gamma}(s)\ ] ] where @xmath91 , @xmath92 is the observed @xmath0-ray energy at redshift zero , @xmath74 is the frequency at redshift @xmath75 , @xmath93 is the redshift of the @xmath0-ray source , @xmath94 , @xmath60 being the angle between the @xmath0-ray and the soft background photon , @xmath95 is planck s constant ( @xmath96 ) , and the pair production cross section @xmath97 is zero for center - of - mass energy @xmath98 , @xmath99 being the electron mass . above this threshold , @xmath100,\ ] ] where @xmath101 .
figure 4 , based on the results of ss98 , shows the predicted critical energy for a 1/e absorption by pair production versus redshift obtained for the cases of correction for metallitcity evolution and no correction .
absorption of @xmath0-rays of energies below @xmath315 gev is negligible .
= 4.0truein the weak redshift dependence of the opacity at the higher redshifts , as shown in figure 4 , indicates that the opacity is not very sensitive to the initial epoch of galaxy formation , contrary to the speculation of macminn and primack ( 1996 ) .
in fact , the uncertainty in the metallicity correction ( see figure 4 ) would obscure any dependence on @xmath102 even further . with the @xmath0-ray opacity @xmath103 calculated out to @xmath104 , the cutoffs in blazar @xmath0-ray spectra caused by extragalactic pair production interactions with stellar photons can be predicted .
the left graph in figure 5 , from ss98 , shows the effect of the intergalactic radiation background on a few of the blazars observed by _
egret _ , _ viz .
_ , 1633 + 382 , 3c279 , 3c273 , and mrk 421 , assuming that the mean spectral indices obtained for these sources by _ egret _ extrapolate out to higher energies attenuated only by intergalactic absorption .
( observed cutoffs in blazar spectra may be intrinsic cutoffs in @xmath0-ray production in the source , or may be caused by intrinsic @xmath0-ray absorption within the source itself . )
the right hand graph in figure 5 shows the background spectrum predicted from unresolved blazars ( stecker & salamon 1996 ; salamon & stecker 1998 ) compared with the _ egret _ data ( sreekumar , _ et al .
note that the predicted spectrum steepens above 20 gev , owing to extragalactic absorption by pair - production interactions with intergalactic optical and infrared photons , particularly at high redshifts .
studies of the absorption of high energy @xmath0-rays in the spectra of extragalactic objects can be used to probe both the present intergalactic infrared background and the infrared and optical radiation that existed in intergalactic space at higher redshifts .
the results of such observations are free of the effects of solar system and galactic foreground contamination .
such studies are most effective when combined with direct infrared observations in a synoptic approach .
high energy @xmath0-ray observations have already provided the best constraints on the mid - infrared background and they support a galaxy evolution scenario where emissivity evolution in the mid - infrared evolved as @xmath105 with @xmath34 .
this is consistent with observations of galaxy counts using _
isocam_. on the other hand , the analysis of _ cobe - dirbe _
observations favors a stronger evolution in far - infrared emissivity with @xmath35 .
there are various possible interpretations of these results ; among them is the possibility that special relativity is modified at high energies .
future @xmath0-ray and infrared observations will be needed to resolve this situation .
stecker , f.w . &
de jager , o.c .
1997 , in _ towards a major atmospheric cerenkov detector v _ , proc .
kruger national park workshop on tev gamma ray astrophysics , ed .
. de jager ( potchefstroom : wesprint ) pg . | in this paper , i will take a synoptic approach to determining the intergalactic infrared radiation field ( iirf ) .
this approach draws on both the multi - tev @xmath0-ray observations and the infrared background observations and relates them _ via _ the semi - empirical modelling method of malkan & stecker .
i discuss the evidence for an intergalactic infrared background obtained by an analysis of the _ hegra _ observations of the high energy @xmath0-ray spectrum of mrk 501 and from constraints from mrk 421 deduced from the whipple air cherenkov telescope results .
i will show that this evidence is in accord with the predictions made by malkan & stecker ( 1998 ) for the intergalactic infrared spectral energy distribution produced by galaxies .
the malkan - stecker predictions are also in excellent agreement with mid- and far infrared galaxy counts .
however , there may be potential problems relating these predictions with the results of the analysis of _ cobe - dirbe _ far infrared data .
the @xmath0-ray and _ cobe - dirbe _ observations may also need to be reconciled .
i will discuss possible ways to resolve this situation including a partial nullification of the @xmath0-ray absorption process which can hypothetically occur if lorentz invariance is broken .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the first observation of emission lines from a resolved region around a qso was that of @xcite .
other examples followed @xcite .
the origin of these extended emission - line regions ( eelrs ) around low - redshift qsos is still uncertain .
the distribution of the emission - line gas generally appears to be uncorrelated with either the extended continuum emission or the radio structure , although both the incidence and the luminosity of eelrs is higher for steep - spectrum radio quasars than for flat - spectrum quasars or radio - quiet qsos @xcite .
two principal suggestions have been made : ( 1 ) that the gas is debris from strong interactions or mergers @xcite , and ( 2 ) that the gas is due to cooling flows from a hot surrounding medium @xcite .
we report here on an investigation of the eelr associated with the 4c37.43 .
this quasar has the most luminous eelr known among qsos at @xmath9 @xcite , making it an obvious candidate for detailed study . at the same time , the morphology and spectrum of this eelr are similar to those of many other low - redshift qsos , so the results of this investigation should have a wider application .
we use several types of ground - based data on 4c37.43 : ( 1 ) imaging in a 29 bandpass centered on [ ] @xmath10 , to determine the morphology and brightness distribution of the ionized gas ; ( 2 ) imaging in the essentially line - free continuum region between restframe 5050 and 6050 ; ( 3 ) imaging in a 290 bandpass including h@xmath11 , along with an adjacent continuum bandpass , to determine the h@xmath11 flux within an aperture equivalent to that used for our _ hubble space telescope _ ( _ hst _ ) fos spectroscopy ; ( 4 ) low - resolution spectroscopy to determine both the velocity structure of the ionized gas and line ratios to compare with photoionization models to estimate reddening ; and ( 5 ) higher - resolution spectroscopy to measure the [ ] @xmath12,3729 doublet ratio as a diagnostic of the electron density .
we obtained a total of 9000 s of narrowband [ ] imaging with the university of hawaii ( uh ) 2.2-m telescope and a tektronix @xmath13 ccd , through a filter centered at 6865 and having a fwhm of 29 .
in addition , we obtained 3000 s in an essentially line - free continuum bandpass filter centered at 7606 , with a fwhm of 1372 .
the images were reduced via standard techniques , using flat - field exposures of the illuminated interior of the dome , and the flux calibration was determined from images of the spectrophotometric standard star feige 67 @xcite .
stellar objects have a fwhm of 080 in the [ ] image and 076 in the line - free continuum image . for the brighter regions in both images , our s / n
ratio is high enough to allow useful deconvolution , which we have carried out using the stsdas contributed task acoadd ( * ? ? ?
* and references therein ) .
we also obtained 10,800 s of h@xmath11 imaging at the uh 2.2-m telescope , with a filter centered at 9131 and having a 290 fwhm bandpass .
although redshifted h@xmath11 lies at @xmath14 near the edge of the filter bandpass , the transmission at this position was greater than 70% .
line - free continuum exposures totaling 1500 s were obtained through a filter centered at 8338 , with a fwhm of 880 .
these fwhm for stellar objects in these images was @xmath15 .
some of the low - resolution spectroscopy was obtained with image slicers @xcite and the f/31 spectrograph on the uh 2.2-m telescope .
the image slicers provided 3 contiguous slits on the sky and produced 3 independent spectra on the tektronix @xmath13 ccd .
we tried two approaches : half of the data was obtained with a slicer having slit widths of 115 at a position angle ( pa ) of @xmath16 ; the remaining half with a slicer having slit widths of 21 at a pa of @xmath17 . in both cases ,
the center slit was centered on the bright emission peak about 4 east and slightly north of the quasar . by using these image slicers
, we were able to collect essentially all of the flux from the main emission region and eliminate any potential problem with atmospheric dispersion in comparing the flux from h@xmath11 and h@xmath18 .
we obtained a total of 16,200 s of useful integration .
we obtained additional low - resolution spectroscopy , with a total integration of 3600 s , with the low - resolution imaging spectrometer ( lris ; oke et al .
1995 ) on the keck ii telescope . with a 300 groove mm@xmath8 grating ,
the 1-wide slit gave a fwhm of @xmath19 .
the slit was placed through the qso and the bright emission region to the east at position angle 727 .
our high - resolution spectroscopy of the [ ] doublet was obtained with the uh faint - object spectrograph and a texas instruments @xmath20 ccd on the canada - france - hawaii telescope ( cfht ) . using a 1200 groove
mm@xmath8 grating and a 14 slit , we had a dispersion of 0.7 pixel@xmath8 and an instrumental fwhm of 3.3 pixels , or 2.3 .
we obtained 7200 s total integration centered on the peak of the emission - line region at slit pa @xmath21 .
we also obtained additional lower - resolution ( fwhm 5.0 ) spectroscopy during the same run , using a 500 groove mm@xmath8 grating and an image slicer with three 07 slits on 14 centers and at position angle 755 . by using 4 interleaved exposures ( 3 of 2700 s duration , 1 of 2100 s )
, we were able to completely cover a region of @xmath22 .
the main purpose of these observations was to determine the velocity field of the ionized gas over most of the emitting region through measurements of the [ ] @xmath10 line .
our hst observations of 4c37.43 consist of cycle 2 spectroscopy with the faint - object spectrograph ( fos ) and cycle 6 imaging with the wide - field planetary camera ( wfpc2 ) .
the original objective of the fos spectroscopy was to determine the ly-@xmath11 flux , for comparison with groundbased observations of balmer - line fluxes .
the entrance aperture was @xmath23 , but the height of the linear diode array limited the effective aperture to @xmath24 , where the 43 dimension was parallel to the dispersion axis . for an aperture of this size , the pre - refurbishment _ hst _ optical quality was sufficiently good that the residual spherical aberration had no significant impact on our results .
the center of this effective aperture was offset from the quasar by 4 in position angle @xmath17 , to the peak of the extended emission . in order to minimize the amount of scattered light from the wings of the aberrated quasar psf entering the spectrograph , we used an engineering mode that allowed us to deflect the electron beam within the digicon detector so that only the lower third of the square entrance aperture ( that closest to the quasar ) was projected onto the readout array . a total of 5550 s of exposure was obtained in two orbits .
the fos observations were reduced with the improved stsci `` calfos '' pipeline in 1994 .
this version included geomagnetic correction .
in addition , a manual correction was applied for internal scattered light , although this had only a minor impact on the resulting spectrum . from the absence of broad wings on ly-@xmath11 and @xmath25 and comparison with the line ratios observed in the fos nuclear spectrum @xcite
, we established that any contribution of light from the quasar itself was small , in agreement with our estimate of the expected light in the wings of the quasar psf .
the wfpc2 imaging comprised both wfc3 and pc1 observations in the f814w filter , which is dominated by continuum radiation ( although h@xmath11 falls on the longward wing of the filter profile , where the transmission is @xmath26% of the peak ) , and wfc2 observations with a linear - ramp filter ( lrf ) centered on the redshifted [ ] @xmath10 emission line .
the wfc3 f814w imaging consisted of two exposures at each of four dither positions , with a total of 6400 s ; the pc1 f814w imaging , of four exposures at each of four dither positions , with a total of 2560 s ; and the wfc2 lrf imaging , of two exposures at each of two dither positions , with a total of 9200 s. after standard pipeline reduction , the images were combined using the stsdas _ drizzle _ procedure @xcite .
cosmic rays were first identified using the stsdas _ crrej _ task , with careful adjustment of the parameters to avoid any significant compromising of real data , and masks for the cosmic rays were combined with the pipeline - generated data - quality files to produce the masks used to reject pixels in the final combination .
figure [ gbimages ] shows our ground - based images in essentially line - free continuum and in [ ] emission , obtained with the uh 2.2 m telescope .
the most striking impression is how _ unlike _ the continuum and line images are : the ionized gas seems to know virtually nothing of the distribution of the continuum ( presumably stellar ) material .
the continuum extension to the east , possibly forming a tidal bridge to the companion galaxy ( which is known to be at the same redshift as the quasar ; @xcite ) , has been described previously by @xcite , @xcite , @xcite , and @xcite .
the strong emission - line peak @xmath27 east - northeast of the quasar was first noted in @xmath28-band imaging ( which includes the strong [ ] lines ) by stockton ( 1976 ) ; a lower - resolution image in a narrow - band filter centered on [ ] @xmath10 has been given by @xcite , and integral - field spectroscopy has been carried out by @xcite and @xcite .
this emission structure , which we shall henceforth refer to as `` @xmath29 '' , is the subject of much of the remainder of this paper . particularly in the deconvolved , psf - subtracted [ ] @xmath10 image ( lower - right panel of fig .
[ gbimages ] ) , considerable new detail is seen in the structure of the ionized gas .
@xmath29 , previously seen simply as an elongated region , is now clearly double , with each of the peaks sporting suggestive tail - like appendages .
the long linear structure to the northwest of the quasar takes on an undulating appearance and breaks up into a series of knots .
finally , there appears to be a previously undetected emission peak even brighter than those to the east just 05 west and slightly south of the quasar .
the _ hst _ images generally confirm and extend these results .
the wfc3 f814w continuum image is shown in fig .
[ wfc_img ] , and the lrf [ ] @xmath10 image in fig .
[ oiii_img ] .
the relatively shallow pc f814w image shows no significant new structure .
the wfc3 f814w continuum images are saturated at the center of the quasar , so it is difficult to recover structure very close to the quasar by deconvolution or psf fitting and subtraction .
the pc1 f814w exposures were planned to avoid saturation of the quasar ; on the other hand , these images are strongly dominated by ccd readout noise , so low - surface - brightness features have poor s / n . the linear - ramp - filter [ ] @xmath10 images avoid saturation of the quasar and have similar surface - brightness sensitivity to our best ground - based images .
however , the shift of the central wavelength with position on the ccd that is a characteristic of these filters means that the [ ] emission more than about 8 from the quasar will be lost .
the deep continuum image ( fig . [ wfc_img ] ) shows that the bridge - like structure to the east has a quite narrow core .
the high - surface - brightness region northeast of the quasar , quite smooth at ground - based resolution , now shows some knots that may be star - forming regions , although some of this structure is due to regions of bright h@xmath11 emission leaking through the wings of the f814w filter profile .
the [ ] image ( fig . [ oiii_img ] ) shows an impressive amount of detail .
the two bright peaks comprising @xmath29 break up into a series of discrete knots , the long structure on the west side of the quasar now has a braided appearance , and the very bright region just west of the quasar is seen to have two discrete bright peaks ; these last features are confirmed in the pc image , where they are apparently due to the h@xmath11 leak .
the material to the southwest of the quasar forms an irregular ring , somewhat reminiscent of the one seen in the extended emission region around the quasar 4c25.40 @xcite .
however , probably the most important new features are the faint linear structures extending from the quasar to the northeast and southeast : these are almost certainly the edges of the ionization cone illuminating the bright emission regions east of the qso .
they are unlikely to be artifacts of the psf , since they persist undiminished after careful psf subtraction , using a star observed at the same detector position ( fig .
[ oiii_img ] , upper - left insert ) .
the presumed cone has an apparent opening angle of @xmath30 and its axis has a position angle of 77 , not particularly well aligned with the axis of the extended symmetric double radio source , which has a position angle of @xmath31 ( @xcite ; see fig .
[ gbimages ] ) .
in retrospect , these linear features are confirmed in our deconvolved ground - based [ ] image in fig .
[ gbimages ] , although the resolution there is certainly not high enough to allow them to have been recognized as an ionization cone . as mentioned above , at distances greater than a few arcsec , the lrf bandpass shifts off the [ ] @xmath10 line , so some of the [ ] emission seen at large distances east and west of the quasar in the our ground - based [ ] image is not present in the _ hst _ image .
it is perhaps significant that the axis defined by the most distant material seen in the ground - based [ ] image ( fig .
[ gbimages ] , bottom panels ) lies close to the axis of the ionization cone .
in fact , most of the extended emission lies within the projected edges of this cone or those of an inferred counter cone , although some strong emission does not .
it is possible that this emission lying outside the projected cone is evidence that the obscuring torus is leaky or irregular along some lines of sight . before we can discuss emission - line ratios
, we must estimate the reddening along the line - of - sight from both galactic dust and that associated with 4c37.43 itself .
for the galactic extinction , the dust column densities found by @xcite imply @xmath32 mag . after correcting the line fluxes by a standard galactic extinction curve normalized to this value
, we obtain an h@xmath11 to h@xmath18 flux ratio of @xmath33 and an h@xmath34 to h@xmath18 flux ratio of @xmath35 ( the former ratio was obtained from our spectrum from the uh 2.2 m telescope , the latter from our keck lris spectrum ( fig . [ lrisspec ] ) , which has a much better signal - to - noise ratio ) .
assuming that the reddening of the emission lines by intervening dust associated with 4c37.43 can be approximated by a standard galactic reddening law ( _ e.g. , _ * ? ? ?
* ) , we obtain intrinsic ratios @xmath36h@xmath11@xmath37h@xmath18@xmath38 and @xmath36h@xmath11@xmath37h@xmath18@xmath39 with @xmath40 .
these ratios are quite consistent with case b recombination , with little or no enhancement of h@xmath11 due to collisional excitation .
the uv spectrum of @xmath29 obtained with the _ hst _ fos is shown in fig . [ fosspec ] .
the observed flux in the ly-@xmath11 line is @xmath41 erg @xmath42 s@xmath8 . from our ground - based h@xmath11
imaging , using a synthetic aperture designed to match as closely as possible the effective fos aperture , we obtain an h@xmath11 flux of @xmath43 erg @xmath42 s@xmath8 , where most of the quoted error is due to estimated uncertainties in the aperture matching .
these values give a measured value of the ly-@xmath11/h@xmath11 flux ratio of @xmath44 .
we must now attempt to correct for reddening .
extrapolating the reddening found from the balmer lines into the uv is very uncertain .
if we simply assume that a standard galactic reddening law , with @xmath45 , holds over the whole range , then the ly-@xmath11 to h@xmath11flux ratio is @xmath46 ; if we consider values of @xmath28 ranging from 2.5 to 5 ( roughly corresponding to the extreme values seen along lines of sight in the galaxy ) , we obtain flux ratios ranging from @xmath47 to @xmath48 , respectively .
the expected ratio under case b conditions with no dust in the emission region , @xmath49 k , @xmath50 , and uncomplicated geometries would be @xmath51 ( our more detailed photoionization models given in [ photoion ] give a ratio of @xmath26 ) .
thus , the intrinsic ratio could be consistent with little or no dust in the emitting region , or it could be a factor of @xmath26 lower than this value , depending on the uv extinction properties of the _ intervening _ dust .
although it is often assumed that ly-@xmath11 emission is easily destroyed by small amounts of internal dust because of the increase of effective path before escape due to multiple scatterings , realistic simulations indicate that rather large concentrations of dust are required to produce significant deviations from the case b ratio @xcite .
this result is largely due to the small gas column density in typical clouds , thus requiring few average scatterings before escape from the cloud .
there can also be rather large variations in the ly-@xmath11/h@xmath11 ratio because of geometrical effects @xcite .
our bottom line is that , if something close to a standard galactic reddening law with @xmath45 applies along the line of sight , both in our galaxy and in the host galaxy of 4c37.43 , and if the individual emitting clouds are fairly isotropic in their emission of ly-@xmath11 , then we infer substantial quantities of internal dust .
however , the uncertainties in both the intervening reddening and the nature of the emitting clouds is such that we can not exclude the possibility of little or no dust . the luminosity - weighted average electron density in the [ ] -emitting region can be determined from the ratio of the [ ] @xmath12,3729 lines , and the electron temperature in the [ ] -emitting region similarly can be determined from the ratio of the [ ] @xmath52,5007 lines ( _ e.g. , _ * ? ? ?
figure [ oii ] shows our high - resolution profile of the [ ] @xmath12,3729 doublet .
while the two lines are not quite resolved , the profile allows a unique decomposition .
the fit shown is based on gaussian profiles placed at the laboratory wavelengths of the lines , fitted to the deredshifted spectrum , where the redshift has been determined from the [ ] @xmath53 line . the ratio @xmath54 , implying @xmath55 @xmath3 if @xmath56 k , or @xmath57 @xmath3 if @xmath58 k. earlier , @xcite had found @xmath59 by estimating the ionization parameter from the emission - line spectrum and the incident flux from the quasar luminosity and the projected distance to the emitting region .
the [ ] -line ratio @xmath60 is well determined from our keck lris spectrum ( fig .
[ lrisspec ] ) . the observed ratio is @xmath61 .
after correction for reddening , as determined according to [ reddening ] , we obtain @xmath62 ( we assume @xmath45 both for the correction for galactic reddening and for reddening at the redshift of 4c37.43 , but the difference over this wavelength range from using any other reasonable value is similar to the random uncertainty ) . for any likely value of @xmath63 ,
@xmath64 k. it is clear even without running the models that no single - parameter model can reproduce the observed spectrum : an ionization parameter consistent with the [ ] / [ ] and [ ] @xmath10/h@xmath18 intensity ratios produces relatively much weaker [ ] @xmath65 emission than is observed .
these and other inconsistencies between observed spectra of agn narrow - line and extended emission regions and computed photoionization models based on a single ionization parameter have led to several attempts to achieve better agreement by considering at least two kinds of clouds with significantly different physical properties .
one of the most successful of these approaches , pioneered by @xcite and explored in considerable detail by @xcite , involves using a combination of high - ionization - parameter matter - bounded clouds and low - ionization - parameter ionization - bounded clouds .
the usefulness of this approach was confirmed by @xcite , who found that such a mixed - medium model not only gave a better fit to the emission - line spectrum of extended emission in the radio galaxy 3c321 but also resulted in a more reasonable value for the photoionizing flux of the ( hidden ) uv source . achieving good agreement with the observed emission spectrum may also require the use of non - solar abundances ( _ e.g. , _ * ? ? ?
we have run grids of models using the mappings3 photoionization code @xcite .
guided by the investigations just mentioned , we have sought an acceptable model by ( 1 ) using two components , ( 2 ) allowing either or both components to be density bounded , and ( 3 ) allowing non - solar metallicities .
we estimate the flux density at the lyman limit by making a plausible interpolation between the uv and x - ray luminosities given by @xcite , converting the luminosity density to our assumed flat cosmology with @xmath66 km s@xmath8 mpc@xmath8 and @xmath67 , and taking the actual distance to @xmath29 to equal the projected distance of 4 ( @xmath68 kpc ) .
we obtain the flux density from the quasar continuum at the lyman limit incident on @xmath29 to be @xmath69 erg @xmath42 s@xmath8 hz@xmath8 .
we assume a two - component power - law photoionizing continuum , where a component with index @xmath70 dominates in the uv and a flatter component with index @xmath71 dominates in the x - ray region ( @xmath72 ) .
this continuum is normalized at the lyman limit to the value of @xmath73 found above .
we have also considered a normalization to a flux density reduced by a factor of 2 , corresponding to a projection factor of @xmath74 in the distance .
this reduced incident flux results in best - fit models only slightly different from those for unity projection factor ( _ i.e. , _ using our assumed @xmath73 ) , requiring minor changes to the densities and metallicity . to obtain an acceptable fit to the observations
, we find that we need to have a metallicity significantly below solar , one component with a density of several hundred @xmath3 , and the other with a density near 1 @xmath3 .
neither component can be extremely optically thin at the lyman limit ; on the other hand , we get slightly better agreement if the low - density component is somewhat density limited , with optical depths at the lyman limit of 10 or so .
our best agreement is for a model with about 1/3 solar metallicity , with @xmath75 of the flux of h@xmath18 coming from an essentially ionization - bounded component with average density @xmath4 @xmath3 , and @xmath76 from a density - bounded component with average density @xmath2 @xmath3 and an optical depth at the lyman limit of 16 .
the run of temperature in these regions is determined in a self - consistent manner from the models and averages close to @xmath77 k in the high - density component and @xmath78 k in the low - density component .
the average dimensionless ionization parameters are @xmath79 for the low - density region and @xmath80 for the high - density region .
the results from two models , combining these two components in different ratios , are compared with the observed line ratios in table [ lineratio ] .
models 1 and 2 have , respectively , 25% and 33% of the h@xmath18 flux coming from the high - density component .
model 1 fits most of the line ratios very well , but underestimates the [ ] emission by 40% .
model 2 fits all of the line ratios within @xmath81% .
lcccccccc observed & 0.34 & 1.42 & 1.46 & 0.96 & 0.19 & 0.20 & 10.55 & 0.07 + model 1 & 0.34 & 0.87 & 0.90 & 0.85 & 0.19 & 0.20 & 10.31 & 0.07 + model 2 & 0.30 & 1.15 & 1.18 & 0.81 & 0.17 & 0.20 & 9.35 & 0.07 + from our estimate of the dereddened h@xmath11 flux ( [ lyalpha ] ) , an estimate of the volume over which this flux is emitted , and the relative amounts of balmer emission from the two components of our photoionization model , we can estimate the filling factors of the two density regimes .
if @xmath82 is the observed h@xmath18 flux from a given region , the filling factor @xmath83 can be written as @xmath84 where @xmath85 is the emission coefficient for h@xmath18 , @xmath86 is the angular radius of a sphere having the same volume as the emitting region , and @xmath87 is the luminosity distance .
the quantity @xmath88 is nearly independent of density over the relevant density range and can be taken to be a function of temperature only ( _ e.g. , _ * ? ? ?
* ) . for the case of @xmath29 ,
if we assume that @xmath89 and that @xmath90 , with values of 2 @xmath3 and 500 @xmath3 for the low - density and high - density regions , respectively , we then obtain filling factors @xmath91 and @xmath92 . given the strong sensitivity of these calculations to the size of the emitting region , along with the simplicity of our model , these results likely point to a situation in which the low - density , [ ] -emitting gas fills the volume , in which are embedded tiny , dense regions of [ ] -emitting gas . normalizing to the h@xmath11 flux over a @xmath93 region centered on @xmath29 , we obtain @xmath94 of gas in the low - density component and @xmath95 of gas in the high - density component .
while our two - component model is clearly a great oversimplification , it can still give some qualitative insight into the properties of the emission - line region .
it is clear that most ( @xmath96% ) of the [ ] emission comes from regions of low density , while the [ ] emission comes almost entirely from regions with densities @xmath97 times higher .
this means that attempts to estimate the ionization parameter ( and thence the pressure ) from the [ ] / [ ] ratio ( _ e.g. , _ * ? ? ?
* ; * ? ? ?
* ) are not likely to be reliable .
as it happens , we do indeed find a pressure of @xmath98 @xmath3 k for the high - density regions , close to the average pressure found by @xcite for the eelr around 4c37.43 .
but the pressure in the low - density region , which is arguably the component that is likely to be in pressure equilibrium with a hot external medium , is @xmath99 @xmath3 k , over 2 orders of magnitude lower .
the high - density regions are thus clearly not in pressure equilibrium with their surroundings .
because they are located some 20 kpc ( in projection ) from the quasar and do not show a strong correlation with the continuum structure of the host galaxy ( and are hence unlikely to be gravitationally confined ) , they will dissipate on a timescale on the order of the sound - crossing time .
our photoionization calculations indicate that these dense clouds ( or at least the ionized part of them ) are typically only about 0.1 pc thick , so , with a typical sound speed of @xmath100 km s@xmath8 , they are likely to have lifetimes of @xmath6 years .
it seems most likely , then , that these clouds are continuously regenerated by shocks propagating through the surrounding medium . on the other hand , the relative weakness of uv lines such as @xmath101 and @xmath25 with respect to @xmath102 indicates that photoionization by the central qso strongly dominates over ionization by the shocks themselves in generating the observed emission lines . our moderate - resolution image - sliced spectroscopy of the [ ] h@xmath18 region covers most of the obvious emission seen in our deep narrow - band [ ] imaging .
velocities ( relative to the nuclear narrow - line region ) measured from the [ ] @xmath10 line are shown in fig .
[ oiiivel ] .
our velocity map is in general agreement with those of @xcite and @xcite .
the relative velocities range from @xmath103 to @xmath104 km s@xmath8 .
one s general impression is that the velocity field is locally ordered but globally disordered : individual filaments and components show continuous velocity gradients , but the velocities from region to region do not suggest any sort of simply organized kinematics .
note that the brightest regions on both the east and west sides of the qso have substantial negative velocities .
@xcite mention that `` ... it is notable that the region of largest blueshift and of higher ionization in both the eastern and western clouds , and the highest fwhm in the western cloud ( but not the eastern cloud ) are all where the optical gas crosses the radio source axis . ''
raising the possibility of a jet - cloud interaction , they however point out that `` ... there is no obvious distortion in the radio source to support this interpretation . '' as fig .
[ oiiivel ] shows , while some fairly high - velocity gas lies close to the radio axis on the western side , gas at similar velocities exists at many locations , and the highest velocities are found elsewhere . in any case , it would seem difficult to understand why an interaction between the radio jet and the gas should give large negative radial velocities for _ both _ the approaching and the receding jet
. we can not make detailed comments on the variations in ionization because in many areas the h@xmath18emission is too weak to measure .
what measurements we do have show that the peak [ ] /h@xmath18 ratio falls near the peaks of the [ ] intensity , significantly to the north of the radio jet on both the east and west sides .
in fact , our measurement of the [ ] /h@xmath18 ratio at the brightest region crossed by the jet on the east side is 5.3 , the lowest we are able to measure anywhere , compared with 11.0 about 35 north of this position .
we show our measurements of the [ ] line width in the lower left map in fig .
[ oiiivel ] .
these measurements must be taken with some caution , since multiple velocities are present along many lines of sight , and it may not always have been possible to deconvolve indivdual components . with this proviso , there seems to be some enhancement in the line width near where the radio jet crosses the linear structure on the west side , but similar line widths are seen at many other locations in the nebula .
the largest line widths seem to be associated with the high velocity gas about 2 north of the quasar .
overall , then , we find no compelling evidence for a significant effect on the gas kinematics or ionization from an interaction with the radio jet .
considering qso eelrs as a class , we can distinguish at least four different types .
the first are those in which the emission - line gas is closely associated with the morphology of the host galaxy ; apparently , cold disk gas in the host galaxy is photoionized by the qso nucleus .
such examples among qsos ( as contrasted with seyferts ) are rare , possibly because few qsos have host galaxies with intact cold disks , and the ionization cones are likely not to intercept the disks in any case .
the only likely example we are aware of is pg0052 + 251 @xcite .
the second type is that in which the emission - line gas appears to be connected with the radio morphology . at least among low - redshift qsos ,
strong emission associated with radio structure is not common , but we do have an example in pks2251 + 11 , where the two brightest peaks in the extended emission bracket the southeast radio lobe @xcite . in such cases , it is likely that the emitting gas is either ambient material that has been shocked by the radio jet , or that it is gas from the host galaxy that has become entrained in the radio jet .
a third type includes cases for which a neighboring gas cloud , confined either gravitationally ( as in a companion galaxy ) or by thermal pressure ( as in a cooling flow , as envisioned by @xcite ) , lies within the ionization cone of the qso and is close enough to be photoionized by the qso s uv continuum .
one trivial example of this type is a small galaxy @xmath105 west of mrk1014 .
@xcite noticed that this galaxy was enhanced in the narrow - band image centered on the wavelength of [ ] @xmath10 in mrk1014 , and recent spectroscopy confirms strong [ ] emission from this object at the redshift of mrk1014 @xcite .
an intriguing possibility is that ionization by nearby qso may make visible objects that would be very difficult to detect by any other means .
there is reasonable evidence that dwarf galaxies are heavily dark - matter dominated ( _ e.g. , _ @xcite , and references therein ) , so such galaxies can , in principle , trap gas in potential wells that are not well advertised in terms of their stellar luminosity .
the ubiquitous dwarf star - forming galaxies ( _ aka _ extragalactic regions ) are a case in point : they can not have sustained their present rate of star formation for more than a small fraction of a hubble time .
many of these must have been virtually invisible in their earlier quiescent state ; yet they have to have had substantial reservoirs of gas at moderately high densities in order to allow the current star formation to proceed .
objects such as these , with few stars but some gas in a largely dark matter potential well , would normally be undetectable at moderate redshifts , but they would become quite noticeable if the gas were to be ionized by a nearby qso
. this might be one way of detecting at least one class of the `` ghost galaxies '' proposed by kormendy & freeman ( 1998 ; see also @xcite ) .
a possible example of such a case is the discrete [ ] -emitting object near nab0205 + 05 @xcite , for which the equivalent width of the [ ] @xmath10 line is extremely high .
finally , there is the type of eelr which is closely associated with the qso , but for which the ionized gas , though often having a highly structured morphology , shows no obvious relation to either the radio or the optical continuum morphology .
most qso eelrs seem to be of this type , although few have the luminosity or show the richness of structure of that of 4c37.43 .
to recapitulate our main results from [ photoion ] , the region of the 4c37.43 eelr that we have investigated appears to comprise two main regimes , one with a density of @xmath2 @xmath3 and a temperature of @xmath106 k , the other with a density of @xmath4 @xmath3 and a temperature of @xmath107 k. the filling factor of the low - density gas is close to unity , while that of the higher - density gas is @xmath5 .
it is possible that the low - density gas is close to pressure equilibrium with a surrounding hot external medium , but the high - density gas can not be .
the relatively low pressure of @xmath99 @xmath3 k that we deduce for the low - density gas argues against a cooling flow origin for the emission - line region of 4c37.43 , as proposed by @xcite .
the referee has pointed out that photoionization of a typical giant molecular cloud ( gmc ) complex by a quasar could produce a similar two - phase density structure , as the photoionized region eats into the dense cloud and the ionized gas rapidly expands to reach pressure equilibrium with the surrounding medium .
this is an attractive suggestion , since gmcs are self - gravitating and could be stable over timescales of @xmath108 years .
however , as we have noted above , there is virtually no correlation between host galaxy morphology and the distribution of the gas in the eelr for 4c37.43 and similar objects ; in these cases , the quasar can not simply be `` lighting up '' _ in situ _ interstellar clouds . the brightest emission regions around 4c37.43 are mostly at projected distances of @xmath81 kpc , where one would not normally expect to find dense molecular clouds .
furthermore , strong extended emission is not the norm in qsos , even amongst the steep - radio - spectrum classical double sources for which it is statistically more likely
. why do a few objects like 4c37.43 have such prolific displays of extended emission , while the majority of apparently similar objects show little or none ?
as we have previously noted @xcite , there is an apparent correspondence between the incidence of strong extended emission among qsos and the presence of overt signs of strong interaction , such as close companion galaxies and tails or bridges seen in continuum emission .
this is certainly true in the case of 4c37.43 , where the continuum asymmetry and bridge- or tail - like structure to the east strongly suggest a recent merger .
one might therefore suppose that gmcs could have been tidally ejected to large distances and out of the plane of the original disk in which they were formed .
but gmcs are sufficiently dense that their trajectories should not be significantly affected by hydrodynamic interaction with the ambient gas , and they should follow essentially the same paths the stars do . once again , the general lack of correspondence between the continuum and emission - line morpholgies in 4c37.43 argues against this scenario .
because the high - density clouds can not be confined by external pressure , and gravitational confinement seems unlikely , they have probable lifetimes of @xmath6 years .
it therefore seems likely that they must be continuously regenerated by shocks .
if so , it remains a possibility that the shocks responsible for the dense regions are simply due to collisions of sheets of gas during the merger @xcite .
the velocities of up to 700 km s@xmath8 that we see are a significant , but perhaps not completely fatal , objection to this picture .
it is quite possible that much of the extended ionized gas we see was originally ejected from one or both of the galaxies by tidal forces during the merger .
however , it now seems likely to us that the origin of the shocks needed to produce the high - density clouds is an _
indirect _ result of the merger : a starburst - driven galactic superwind .
virtually all ultra - luminous infrared galaxies ( uligs ) are starbursts triggered by major mergers ( _ e.g. , _ * ? ? ?
at least some significant fraction of these also have qso nuclei ( _ e.g. , _ * ? ? ?
* ) , which must also have been triggered by the merger @xcite .
the ulig phase can dominate the ir emission for up to @xmath109 myr @xcite ; if the qso has a lifetime greater than this , some objects identified as `` ordinary '' qsos ( _ i.e. , _ those with no evidence of excessive fir emission ) will have previously been uligs as well .
one of the consequences of a vigorous starburst is a galactic superwind ( * ? ? ?
* and references therein ) , which will interact with the surrounding medium and particularly with any tidally ejected gas .
this superwind provides a mechanism for producing shocked gas , potentially at large distances from the qso .
it also may be responsible for entraining any ambient cool gas ( including that possibly ejected during the encounter ) and controlling both its location and its velocity .
this picture is consistent with both the distribution and the velocities of the ionized gas we see in 4c37.43 .
the main emission regions are found at roughly the same projected distances to the east and west of the quasar , and knots of emission are arranged along linear or arc - like structures roughly perpendicular to the direction to the quasar , which plausibly correspond to surfaces of the expanding bubbles . as can be seen from fig .
[ oiiivel ] , the bulk of the gas in the most luminous emission regions has projected velocities between about @xmath110 and @xmath111 km s@xmath8 .
there are similar positive velocities in regions with weaker emission ; also in weaker - emitting regions , there are negative velocities ranging up to @xmath103 km s@xmath8
. the predominance of negative velocities could simply be due to chance : the uneven distribution of the gas ( particularly if it is mostly tidal debris ) , coupled with the illumination pattern of the uv continuum from the quasar .
however , if the velocities have a significant outflow component , a bias towards negative velocities could also be a natural consequence of screening by dust blown out by a superwind . also ,
the tendancy for the highest velocities to be found amongst the weaker - emitting components is consistent with the expectation that initially less dense and less massive clouds would be accelerated to higher velocities .
furthermore , of the dozen most luminous eelrs in the survey of low - redshift qsos by @xcite ( including , of course , that of 4c37.43 ) , two are known to be associated with qsos having ultraluminous ir host galaxies having confirmed starburst or recent post - starburst stellar populations ( 3c48 , @xcite ; mrk1014 , @xcite ) .
this association lends support to a connection between the eelr and starburst activity .
@xcite have suggested that qsos that are also powerful radio sources , such as 3c48 , can break through a dust cocoon associated with the starburst more rapidly than can most radio - quiet qsos , allowing the central uv source to illuminate and photoionize the extended gas early on .
if this is true , it could explain why the most luminous eelrs are associated with powerful extended radio sources @xcite .
the observations presented here have given us the clearest view yet of an extended emission region around a quasar .
the _ hst _ wfpc2 imaging in the [ ] emission line shows the first clear evidence of an ionization cone on the east side of the quasar and confirms the presence of very high surface brightness emission @xmath112 west of the quasar .
much of the emission is arrayed in linear or arc - like structures , many of which appear to lie about 4 from the quasar and to be centered on it .
the spectrum of the bright emission complex @xmath29 is consistent with photoionization by the quasar , provided that the region comprises two main components , both having metallicities around 1/3 solar : ( 1 ) a low - density ( @xmath2 @xmath3 ) gas with essentially unity filling factor , and ( 2 ) a high - density ( @xmath4 @xmath3 ) gas with a very low ( @xmath5 ) filling factor and distributed in thin ( @xmath113 pc ) filaments or sheets .
the very short ( @xmath107 year ) lifetimes of these dense regions suggest their origin in shocks . from the spectroscopic and imaging evidence , together with our knowledge that at least some qsos have gone through a strong starburst phase as a result of major mergers , we believe that it is quite likely that such shocks are consequences of superwinds driven by a recent starburst in the quasar host galaxy .
we thank gabriela canalizo for assisting with the keck lris observations , geoff bicknell , mike dopita , and josh barnes for helpful discussions , and mike dopita , lisa kewley , and ralph sutherland for help with running the mappings photoionization code .
we also thank the anonymous referee for a careful reading of the paper and for comments that helped us improve the presentation .
support of the hst observations was provided by nasa through grant nos .
go-3538.01 - 91a and go-06490.01 - 95a from the space telescope science institute , which is operated by aura , inc .
, under nasa contract no .
nas5 - 26555 .
some of the ground - based observations were supported by nsf grant no .
ast95 - 29078 .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration , and which was made possible by the financial support of the w.m . keck foundation .
the authors recognize the very significant cultural role that the summit of mauna kea has within the indigenous hawaiian community and are grateful to have had the opportunity to conduct observations from it .
bahcall , j. , et al .
1993 , , 87 , 1 bergeron , j. , & durret , f. 1987 , , 184 , 93 binette , l. , wilson , a. s. , & storchi - bergmann , t. 1996 , , 312 , 365 binette , l. , wang , j. , villar - martin , m. , martin , p. g. , & magris c. , g. 1993 , , 414 , 535 block , d. l. , & stockton , a. 1991 , , 102 , 1928 boroson , t. a. , & oke , j. b. 1982 , , 296 , 397 boroson , t. a. , & oke , j. b. 1984 , , 281 , 535 boroson , t. a. , persson , s. e. & oke , j. b. 1985 , , 293 , 120 canalizo , g. , & stockton , a. 2000a , , 528 , 201 canalizo , g. , & stockton , a. 2000b , , 120 , 1750 canalizo , g. , & stockton , a. 2001 , , 555 , 719 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 crawford , c. s. , & vanderriest , c. 2000 , , 315 , 433 dopita , m. a. , & sutherland , r. s. 1995 , , 455 , 468 durret , f. , pcontal , e. , petitjean , p. , & bergeron , j. 1994 , , 291 , 392 fabian , a. c. , crawford , c. s. , johnstone , r. m. , & thomas , p. a. 1987 , , 228 , 963 fruchter , a. s. , & hook , r. n. 1997 , in applications of digital image processing xx , ed .
a. tescher , proc .
spie , 3164 , 120 heckman , t. m. 2001 , asp conf .
, 240 , 345 hickson , p. , & hutchings , j. b. 1987 , , 312 , 518 hook , r. n. , & lucy l. b. 1993 , st - ecf newsletter 19 , 6 hook , r. n. , lucy l. b. , stockton , a. , & ridgway , s. e. 1994 , st - ecf newsletter 21 , 16 kormendy , j. , & freeman , k. c. 1998 , , 30 , no . 4 , 21.05 laor , a. , fiore , f. , elvis , m. , wilkes , b. j. , & mcdowell , j. c. 1994 , , 435 , 611 lutz , d. , veilleux , s. , & genzel , r. 1999 , , 517 , l13 massey , p. , strobel , k. , barnes , j. v. , & anderson , e. 1988 , , 328 , 315 mateo , m. 1998 , , 36 , 435 miller , p. , rawlings , s. , & saunders , r. 1993 , , 263 , 425 oke , j. b. , et al .
1995 , , 107 , 375 osterbrock , d. e. 1989 , astrophysics of gaseous nebulae and active galactic nuclei , mill valley , california : university science books richstone , d. o. , & oke , j. b. 1977 , 213 , 8 robinson , t. g. , tadhunter , c. n. , axon , d. j. , & robinson , a. 2000 , , 317 , 922 sanders , d. b. & mirabel , i. f. 1996 , , 34 , 749 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 stockton , a. 1973 , , 246 , 25 stockton , a. 1976 , , 205 , l113 stockton , a. 1978 , , 223 , 747 stockton , a. 1986 , , 118 , 487 stockton , a. 1990 , in _ dynamics and interactions of galaxies _ , ed .
r. wielen , ( springer - verlag : berlin ) , 440 stockton , a. , & canalizo , g. 2000 , asp conf .
, 195 , 385 stockton , a. , & mackenty , j. w. 1987 , , 316 , 584 trentham , n. , m " oller , & ramirez - ruiz , e. 2001 , , 322 , 658 viegas , s. m. , & prieto , m. a. 1992 , , 258 , 483 wampler , e. j. , burbidge , e. m. , baldwin , j. a. , & robinson , l. b. 1975 , , 198 , l52 | we have explored the nature of the extended emission - line region around the @xmath0 quasar 4c37.43 , using extensive ground - based and _ hst _ imaging and spectroscopy .
the velocity field of the ionized gas shows gradual gradients within components but large jumps between components , with no obvious global organization .
the _ hst _ [ ] image shows radial linear features on the east side of the qso that appear to mark the edges of an ionization cone . concentrating on the bright emission peaks @xmath1 east of the quasar , we find through modeling that we require at least two density regimes contributing significantly to the observed emission - line spectrum : one with a density of @xmath2 @xmath3 , having essentially unity filling factor , and one with a density of @xmath4 @xmath3 , having a very small ( @xmath5 ) filling factor .
because the temperatures of these two components are similar , they can not be in pressure equilibrium , and there is no obvious source of confinement for the dense regions .
we estimate that the dense regions will dissipate on timescales @xmath6 years and therefore need to be continuously regenerated , most likely by shocks . because we know that some qsos , at least , begin their lives in conjunction with merger - driven massive starbursts in their host galaxies ,
an attractive interpretation is that the extended emission region comprises gas that has been expelled as a result of tidal forces during the merger and is now being shocked by the galactic superwind from the starburst .
this picture is supported by the observed distribution of the ionized gas , the presence of velocities ranging up to @xmath7 km s@xmath8 , and the existence of at least two qsos having similarly luminous and complex extended emission regions that are known to have ultra - luminous ir galaxy hosts with current or recent starbursts . |
crystallization , vitrification and gelation in systems with short - ranged attractive interactions like colloidal particles and proteins are widely studied phenomena @xcite .
crystallization kinetics is especially important for proteins because good crystals are necessary for the determination of their 3d structure .
furthermore , crystallization of biologically active proteins can lead to diseases like the human genetic cataract @xcite . while the crystal is an equilibrium phase @xcite , gels and glasses are non - equilibrium phases that occur due to dynamical arrest @xcite .
vitrification in a fluid of @xmath0 hard spheres with diameter @xmath1 in a volume @xmath2 , occurs for packing fractions @xmath3
@xcite , while the equilibrium phase diagram predicts a stable crystal phase for @xmath4 @xcite . in
the _ repulsive _ colloidal glass phase , the particles are arrested as they become localized in a `` cage '' formed by the surrounding spheres .
interestingly , the addition of non - adsorbing polymer , resulting in an effective attraction between the colloids , can enhance the relaxation process so that the system can crystallize .
nevertheless , for increasing attraction strengths the particles become more and more localized due to long - lived bonds between the neighboring particles until the system is arrested in an _ attractive _ glass at smaller packing fractions @xcite .
an arrested state at low @xmath5 is often called a gel .
however , the relation between gels and attractive glasses is still highly debated . in general ,
gels can be classified as chemical , if the particles are linked by irreversible bonds , or as physical , if the bonds are not permanent .
physical gels can form through a non - equilibrium mechanism due to the arrest of a ( metastable ) gas - liquid separation by the attractive glass transition @xcite . on the other hand
, equilibrium gelation can occur in systems where the gas - liquid phase separation is suppressed by e.g. long - ranged repulsions or directional interactions @xcite . for increasing contact energy
these two mechanisms for the formation of physical gels continuously evolve towards a diffusion limited cluster aggregation ( dlca ) that leads to chemical gels .
the metastable gas - liquid separation is therefore important for the formation of non - equilibrium gels @xcite , and it is widely recognized as an important factor in protein crystallization @xcite .
there has been much recent experimental , theoretical and simulation effort in understanding the equilibrium phase diagram or the kinetics and dynamics of these systems , but only a few studies explored the relationship between the equilibrium phase diagram and the underlying kinetic pathways for gelation and crystallization .
compared to previous works @xcite we systematically determine the crystallization behavior in relation to the equilibrium phase diagram in the region of the metastable gas - liquid phase separation and we investigate the relation between crystallization and gelation .
we study a model system of @xmath6 hard spheres of diameter @xmath7 and non - adsorbing polymers with diameter @xmath8 described by the asakura - oosawa - vrij ( aov ) model @xcite .
the polymers induce an effective short - ranged attractive depletion interaction between the colloids @xcite .
the total interaction potential is defined as @xmath9 with @xmath10 \
, \label{te : aopot1}\end{aligned}\ ] ] where @xmath11 is the distance between two colloidal particles with @xmath12 the center - of - mass of colloid @xmath13 , and @xmath14 with @xmath15 the boltzmann s constant and @xmath16 the temperature . thus the size ratio @xmath17 determines the range of the interaction .
the polymer reservoir packing fraction @xmath18 , is proportional to the polymer density @xmath19 in the reservoir with which our system is in osmotic contact .
this parameter controls the strength of attraction . as a measure of this interaction strength we also use the contact energy @xmath20 and the experimentally accessible reduced second virial coefficient @xmath21 , which is normalized by the second virial coefficient of hard spheres with a diameter @xmath7 .
although colloids often have a coulombic repulsion @xcite and the attractive potential in proteins is often patchy @xcite our simple model captures the basic phenomena that play a role in the competition between equilibrium and non - equilibrium phases of systems with short - ranged attractions .
we determine the equilibrium phase diagram by calculating the dimensionless free energy density @xmath22 as a function of the colloid packing fraction @xmath23 and contact energy @xmath24 ( polymer reservoir packing fraction @xmath25 ) with monte carlo simulations .
we use thermodynamic integration to relate the free energy of the effective system to that of a reference system at the same colloid volume fraction @xmath23 . to this end , we write the total free energy density as the sum of two contributions @xmath26 where @xmath27 is the free energy density of a system of @xmath0 hard spheres in a volume @xmath2 , for which we used the carnahan - starling expression @xcite for the fluid phase , and the hall expression @xcite for the face - centred - cubic ( f.c.c . )
solid phase . the free energy density @xmath28 is the contribution of the depletion potential ( [ te : aopot1 ] ) to the free energy density , and it is computed using the @xmath29-integration @xcite . in order to map out the phase diagram we determine the total free energy density @xmath30 for many state points @xmath31 in simulations , and we employ common tangent constructions at fixed @xmath32 to obtain the coexisting phases @xcite .
to trace the gas - solid binodal we use @xmath0=512 particles with @xmath33 monte carlo moves per particle for equilibration and @xmath33 monte carlo moves per particle for production . to check the accuracy of the results
few simulations were repeated using @xmath34 monte carlo moves per particle for equilibration and @xmath34 monte carlo moves per particle for production . to trace the metastable binodal while avoiding crystallization @xcite we use a smaller system with @xmath0=120 particles
to get good statistics 5 to 10 independent monte carlo runs are necessary for each value of the density . to study the dynamics of the system we perform standard brownian dynamics ( bd ) simulations @xcite by quenching homogeneous configurations to one of the state points indicated by the points in fig .
[ fig : phdfull2 ] .
the time step @xmath35 is chosen to be @xmath36 with @xmath37 , where @xmath38 is the stokes - einstein diffusion coefficient .
we neglect all hydrodynamic interactions between the particles .
the random forces mimic the interaction between particles and solvent , and are sampled from a gaussian distribution with variance @xmath39 . the random force and the dissipative term provide the system with a heat bath at constant temperature .
we carry out bd simulations using particle numbers @xmath6 ranging from 600 to 1200 and to avoid discontinuities in the potential we replace the hard - core potential with the equivalent repulsive soft potential @xmath40 .
we studied the evolution of a system for a total time @xmath41 . in order to study the dynamical arrest , we compute the bond correlation function @xmath42 where @xmath43=1 if the separation between particle @xmath13 and @xmath44 is smaller than a cutoff value of 1.25 @xmath7 , otherwise @xmath43=0 . the cutoff value corresponds to the minima between the first and second peak of the radial distribution function .
the function @xmath45 is similar in spirit to the one used in @xcite , and counts which fraction of the bonds at @xmath46=0 is still present at time @xmath46 .
we characterise the local structure around a particle @xmath13 by a set of numbers @xmath47 where @xmath48 are spherical harmonics , @xmath49 is a unit vector in the direction of the bond between particle @xmath13 and particle @xmath44 .
the sum runs over all @xmath50 neighbours of particle @xmath13 .
we then construct the dot product @xmath51 where @xmath13 and @xmath44 are neighbouring particles , and @xmath52 is the complex conjugate of @xmath53 .
we normalized the vector @xmath54 , such that @xmath55 . in our analysis particles @xmath13 and @xmath44 are defined as neighbors if @xmath56 , i.e. @xmath57 .
two particles are defined to be joined by a crystal bond if @xmath58 .
particle @xmath13 is crystal - like if at least 7 of its bonds with neighbouring particles are crystal - like @xcite . as in other studies
@xcite we use the @xmath59=6 order parameter .
we determine the phase diagram by calculating free energies in monte carlo simulations @xcite for the potential ( [ te : tot ] ) and for a size ratio @xmath60 , which corresponds approximately to the experimental value of recent work @xcite .
[ fig : phdfull2 ] shows the equilibrium phase diagram in the contact energy @xmath24 ( polymer reservoir packing fraction @xmath25 ) , colloid packing fraction @xmath23 representation .
for @xmath24=0 we recover the hard - sphere phase diagram with a fluid phase at @xmath61 in coexistence with a face - centered - cubic ( fcc ) crystal phase at @xmath62 . for increasing @xmath24
the region of fluid - solid coexistence opens up . at the critical energy @xmath633.48 ( ( @xmath64 ) we find a gas - liquid transition which is metastable with respect to a broad fluid - solid coexistence . in order to follow the crystallization
, we characterize the local structure around particle @xmath13 by the @xmath65 order parameter @xcite .
we define the crystal fraction @xmath66 as the number of crystal - like particles at time @xmath46 divided by the total number of particles .
the red ( dark grey ) region in fig . [ fig : phdfull2 ] indicates the statepoints for which the crystal fraction is larger than 0.4 at @xmath67 while the orange ( light grey ) region indicates a fraction of crystal smaller than 0.4 .
let us discuss first the low @xmath24 limit of the red ( dark grey ) region .
clearly , at low density the metastable gas - liquid binodal has a dramatic effect in inducing ( spontaneous ) crystallization of the colloidal particles , with the bottom boundary of the crystallization region as denoted by the red ( dark grey ) region in fig .
[ fig : phdfull2 ] coinciding with the metastable gas - liquid binodal . below the binodal line , and at low colloid density , we do not observe spontaneous nucleation , in agreement with the observations of @xcite .
neither do we observe , in the time scale of our simulations , crystallization just below to the critical point despite the presence of critical fluctuations that lower the nucleation barrier @xcite . on the other hand , at medium - high colloid density we observe crystallization also outside the binodal .
we now focus on the high @xmath24 limit for crystallization .
[ fig : fraccry ] shows that the crystal fraction slowly decays to zero for increasing attractive energy @xmath24 , and that the crystal fraction does not depend strongly on @xmath23 for fixed @xmath24 .
this indicates a crystallization process hindered by a slowing down of the dynamics .
the boundary between the orange ( light grey ) and red ( dark grey ) regions in fig .
[ fig : phdfull2 ] is a horizontal line at @xmath68 .
as it has been demonstrated for short - ranged attractive systems , isodiffusivity lines are also horizontal when approaching the gas - liquid binodal @xcite .
therefore , the slow decrease of crystallinity can be seen as an indicator of the proximity of the ( attractive ) glass transition line that intersects the gas - liquid binodal .
it is worthwile noting that the attractive glass transition was determined to be at @xmath69 using the long - time scaling of the incoherent correlation functions obtained from computer simulations for a system with a size ratio of q=0.1 , and using a pair potential that included a long - range repulsive barrier to prevent gas - liquid phase separation @xcite .
this value is remarkable close to our value of @xmath70 for which we find hardly any crystallization . as a function of the colloid packing fraction @xmath23 and interaction strength @xmath71 ( polymer reservoir packing fraction @xmath72 and the reduced second virial coefficient @xmath73 ) .
the circles and the triangles indicate the stable fluid - solid coexistence region .
( the circles are the results of simulations where a total of @xmath74 mc moves per particles were used .
the triangles are the results of more accurate simulations where a total of @xmath75 mc moves per particles were used ) .
the continuous lines are a guide to the eye .
the squares indicate the calculated coexistence points of the gas - liquid binodal . the thin dashed line is a fit to the binodal .
the points indicate the state points where we carried out bd simulations .
the red ( dark grey ) region indicates where the crystal fraction is larger than 0.4 at @xmath41 .
the orange ( light grey ) region indicates where the crystal fraction is smaller than 0.4 at @xmath41 .
, width=302 ] at @xmath41 as a function of the interaction energy @xmath71 for ( a ) @xmath76=0.10 .
( b ) @xmath76=0.25 .
( c ) @xmath76=0.40.,width=302 ] , but we indicate the state points discussed in the text . for clarity the binodal symbols are removed .
the thick ( blue ) dashed line indicates the transition between percolating and non - percolating states.,width=302 ] fig .
[ fig : phdfull3 ] shows the phase diagram as in fig .
[ fig : phdfull2 ] but we label state points that will be discussed below . to demonstrate the existence of a dynamical arrest we compute the bond correlation function @xmath77 ( fig .
[ fig : corr ] ) measured starting at a time t=600 @xmath78 after the initial quench in the state points a - d in fig .
[ fig : phdfull3 ] .
we observe a divergence in the characteristic decay time between @xmath79 ( point b in fig . [
fig : phdfull3 ] ) and @xmath80 ( point d in fig .
[ fig : phdfull3 ] ) , giving evidence of a fluid - solid ( either crystalline or amorphous ) transition .
the behavior of @xmath77 is very similar whether the solidification is due to crystallization ( @xmath80 ) or an attractive glass transition ( @xmath81 ) . as a function of time @xmath46 for @xmath76=0.25 .
circles indicate results for energy @xmath82 ( point a in fig . [
fig : phdfull3 ] ) , triangles - up indicate @xmath79 ( point b in fig .
[ fig : phdfull3 ] ) , triangles - down indicate @xmath80 ( point c in fig .
[ fig : phdfull3 ] ) , triangle - left indicate @xmath81 ( point d in fig .
[ fig : phdfull3 ] ) .
, width=302 ] we now focus our attention to the state points inside the gas - liquid region .
we find that in certain regions of the phase diagram the phase separation is arrested in a permanently percolated structure .
the thick dashed line in fig .
[ fig : phdfull3 ] corresponds to the percolation limit . for quenches inside the gas - liquid coexistence
we find , on the left side of the percolation line , clusters that are crystalline at low @xmath24 ( red ( dashed ) region in fig .
[ fig : phdfull3 ] ) and amorphous at high @xmath24 ( orange ( light grey ) region in fig .
[ fig : phdfull3 ] ) , while on the right - hand side , we find a ` long - lived ' percolating network ( gel ) with high - density branches that are crystalline at low @xmath24 ( red ( dashed ) region in fig .
[ fig : phdfull3 ] ) and amorphous at high @xmath24 ( orange ( light grey ) region in fig .
[ fig : phdfull3 ] ) .
we now analyze in more detail the kinetic pathways in different regions of the phase diagram .
the first state point we consider is point e in fig .
[ fig : phdfull3 ] ( @xmath76=0.05 , @xmath83 , and @xmath25=0.34 ) , just beyond the gas binodal . the direct path to the equilibrium phase would be nucleation of the crystal phase in the dilute fluid phase .
the actual kinetic pathway is shown in the attached movies @xcite .
[ fig : path1](a ) and ( b ) show typical configurations at @xmath84 and @xmath41 , respectively .
we observe first in fig .
[ fig : path1](a ) the formation of metastable liquid droplets ( yellow / small light spheres ) surrounded by gas particles ( large light spheres ) .
subsequently the nucleation of the crystal ( red / large dark spheres ) starts within the fluid droplets .
the crystalline clusters grow until a local equilibrium between the cluster and a surrounding gas phase is reached .
these metastable clusters survive for a long time ( fig .
[ fig : path1](b ) ) due to low diffusion , but irreversibly coalesce upon touching , in contrast with the findings of @xcite .
( @xmath76=0.05 and @xmath80 ) at t=50 @xmath78 ( a ) , and at t=600 @xmath78 ( b ) .
gas particles are indicated by large light grey spheres , fluid particles are indicated by yellow ( small light grey ) spheres and crystal particles are indicated by blue ( large dark grey ) spheres .
the complete kinetic pathway is shown in the attached movies @xcite .
, title="fig:",width=302 ] ( @xmath76=0.05 and @xmath80 ) at t=50 @xmath78 ( a ) , and at t=600 @xmath78 ( b ) .
gas particles are indicated by large light grey spheres , fluid particles are indicated by yellow ( small light grey ) spheres and crystal particles are indicated by blue ( large dark grey ) spheres .
the complete kinetic pathway is shown in the attached movies @xcite . ,
title="fig:",width=302 ] ( @xmath76=0.25 and @xmath80 ) at t=600 @xmath78 ) ( a ) and for state point d in fig .
[ fig : phdfull3 ] ( @xmath76=0.25 and @xmath81 ) at t=600 @xmath78 ( b ) .
gas particles are indicated by large light grey spheres , fluid particles are indicated by yellow ( small light grey ) spheres and crystal particles are indicated by blue ( large dark grey ) spheres .
the complete kinetic pathway is shown in the attached movies @xcite.,title="fig:",width=302 ] ( @xmath76=0.25 and @xmath80 ) at t=600 @xmath78 ) ( a ) and for state point d in fig .
[ fig : phdfull3 ] ( @xmath76=0.25 and @xmath81 ) at t=600 @xmath78 ( b ) .
gas particles are indicated by large light grey spheres , fluid particles are indicated by yellow ( small light grey ) spheres and crystal particles are indicated by blue ( large dark grey ) spheres . the complete kinetic pathway is shown in the attached movies @xcite.,title="fig:",width=302 ] at higher colloid density @xmath85 and @xmath24=3.74 ( point c in fig .
[ fig : phdfull3 ] ) , we observe spinodal decomposition , and subsequently nucleation of crystal nuclei within the fluid branches of the spinodal structure .
the crystalline clusters grow until the whole spinodal structure is crystalline .
[ fig : path2](a ) shows that the spinodal structure is arrested by crystallization . at higher interaction strength @xmath24=6.6 and @xmath76=0.25 ( point d in fig .
[ fig : phdfull3 ] ) , we observe spinodal decomposition , and only very small crystal nuclei appear within the fluid branches of the spinodal structure .
the spinodal decomposition is arrested into a branched gel by the formation of a glassy solid ( fig .
[ fig : path2](b ) ) .
this scenario is consistent with non - equilibrium gelation induced by spinodal decomposition arrested by an attractive glass transition .
the metastable gas - liquid phase separation is arrested due to decreased diffusion of single particles by crystallization at low @xmath24 . on the other hand , at high @xmath24 the phase separation
is arrested by a slow dynamics that is consistent with vitrification , but we do not explicitly calculate the exact location of the glass transition here . as a function of time @xmath46 for ( a ) @xmath24=3.3 ( line a in fig . [
fig : phdfull3 ] ) , ( b ) @xmath24=3.52 ( line in b fig .
[ fig : phdfull3 ] ) , ( c ) @xmath24=4.18 ( line in c in fig . [ fig : phdfull3 ] ) , ( d ) @xmath24=6.6 ( line d fig .
[ fig : phdfull3 ] ) . squares indicate a colloid packing fraction @xmath76=0.05 , circles indicate @xmath76=0.20 , triangles - up indicate @xmath76=0.40 , and triangles - down indicate @xmath76=0.45 ( same symbols as in fig .
3),title="fig:",width=264 ] as a function of time @xmath46 for ( a ) @xmath24=3.3 ( line a in fig .
[ fig : phdfull3 ] ) , ( b ) @xmath24=3.52 ( line in b fig .
[ fig : phdfull3 ] ) , ( c ) @xmath24=4.18 ( line in c in fig . [ fig : phdfull3 ] ) , ( d ) @xmath24=6.6 ( line d fig .
[ fig : phdfull3 ] ) .
squares indicate a colloid packing fraction @xmath76=0.05 , circles indicate @xmath76=0.20 , triangles - up indicate @xmath76=0.40 , and triangles - down indicate @xmath76=0.45 ( same symbols as in fig .
3),title="fig:",width=264 ] as a function of time @xmath46 for ( a ) @xmath24=3.3 ( line a in fig .
[ fig : phdfull3 ] ) , ( b ) @xmath24=3.52 ( line in b fig .
[ fig : phdfull3 ] ) , ( c ) @xmath24=4.18 ( line in c in fig . [ fig : phdfull3 ] ) , ( d ) @xmath24=6.6 ( line d fig .
[ fig : phdfull3 ] ) . squares indicate a colloid packing fraction @xmath76=0.05 , circles indicate @xmath76=0.20 , triangles - up indicate @xmath76=0.40 , and triangles - down indicate @xmath76=0.45 ( same symbols as in fig .
3),title="fig:",width=264 ] as a function of time @xmath46 for ( a ) @xmath24=3.3 ( line a in fig .
[ fig : phdfull3 ] ) , ( b ) @xmath24=3.52 ( line in b fig .
[ fig : phdfull3 ] ) , ( c ) @xmath24=4.18 ( line in c in fig . [ fig : phdfull3 ] ) , ( d ) @xmath24=6.6 ( line d fig .
[ fig : phdfull3 ] ) . squares indicate a colloid packing fraction @xmath76=0.05 , circles indicate @xmath76=0.20 , triangles - up indicate @xmath76=0.40 , and triangles - down indicate @xmath76=0.45 ( same symbols as in fig .
3),title="fig:",width=264 ] fig .
[ fig : growth ] shows the crystal fraction @xmath86 as a function of time @xmath46 for @xmath24=3.3 , 3.52 , 4.18 , and 6.6 and for varying @xmath76 ( filled symbols along the red lines a - d in fig .
[ fig : phdfull3 ] ) .
the evolution of the crystal fraction is characterized by three distinct regimes . in the first regime
the fraction of crystal is equal to zero .
this regime corresponds to the time elapsed between the quench and the formation of a critical nucleus ( induction time ) .
the second regime corresponds to the fast growth of the crystal .
[ fig : growth](a ) shows that for statepoints @xmath23=0.05 , 0.20 , and 0.40 of line a in fig .
[ fig : phdfull3 ] , the induction time is significantly different from zero .
we note here that the induction time is larger than 600 @xmath87 for @xmath88 and 0.20 .
all these statepoints are outside the metastable gas - liquid binodal region . in fig .
[ fig : growth](b ) , we find that statepoint @xmath88 of line b in fig .
[ fig : phdfull3 ] shows no crystallization at all , and hence the induction time is again larger than @xmath89 .
this statepoint also lies outside the gas - liquid coexistence region , i.e , @xmath90 , where @xmath91 denotes the coexisting packing fraction of the metastable gas phase . for quenches to @xmath92 ,
i.e. , @xmath23=0.20 , 0.40 and 0.45 , we find that the induction time is close to zero .
similarly , for quenches inside the gas - liquid spinodal as shown in fig .
[ fig : growth](c ) and [ fig : growth](d ) corresponding to line c and d in fig .
[ fig : phdfull3 ] , the induction time is also close to zero as the formation of the spinodal structure is followed immediately by nucleation of small crystallites and subsequent growth of the crystals . in this case , the induction time is related to the spinodal decomposition @xmath93 @xcite that is not visible on the timescale of figure [ fig : growth ] . to summarize , we find that the induction time is close to zero for all quenches beyond the gas binodal of the metastable gas - liquid coexistence region , i.e. , for @xmath92 .
the fraction of crystal grows rapidly after the induction time due to the huge difference in chemical potential between the liquid and the growing crystal phase . in the third regime
, the crystal fraction grows slowly by coarsening and defect removal processes .
the fraction of crystal reached by the system depends strongly on the contact energy . at low @xmath24 particle diffusion
is fast enough for crystallization . on the other hand , at high @xmath24
the particles diffuse slowly and despite the thermodynamic drive to form the crystal the system is trapped in an amorphous solid .
to summarize , for state points inside the binodal we always observe an induction time equal to zero , consistent with the nucleation of small crystallites in the initial stage , and subsequent growth of the crystals . outside the binodal ,
the induction time is close to zero only at high colloid packing fraction .
furthermore , we observe that the fraction of crystal is independent of the packing fraction for fixed @xmath24 , while it decreases for increasing contact energy @xmath24 .
the objective of this work was to systematically study the relationship between equilibrium and non - equilibrium phases in systems with short - ranged attractions . to study the dynamics of the system we carried out bd simulations by quenching a homogeneous fluid phase to certain state points in the phase diagram .
second , we observe that the crystallization is enhanced by the metastable gas - liquid phase separation , as demonstrated by the fact that the line of spontaneous crystallization ( line a in fig . [
fig : sum ] ) at low density follows closely the binodal of the gas - liquid separation ( dashed line in fig .
[ fig : sum ] ) .
the effect is caused by the two - step process crystallization process inside the gas - liquid binodal : the system first forms liquid domains , either through nucleation or spinodal decomposition ; subsequently , crystal nucleation occurs within the dense liquid domains .
this process follows the ostwald rule of stages that states that the transition from one phase to another may proceed via intermediate metastable states .
hence , the effective nucleation rate of the crystal depends on the nucleation rate of the liquid in the gas and the crystal in the liquid phase .
this two - stage crystallization process was also observed in proteins @xcite .
last , we find two competing mechanisms of arrest of the gas - liquid spinodal decomposition . at low @xmath24 , we observe that crystallization starts within the liquid branches before the spinodal structure can coarsen into macroscopic phase separation .
the crystallization slows down the coarsening process and arrests the system in structures typical of spinodal decomposition . for higher @xmath24 ,
the system becomes amorphous as crystallization is inhibited .
this result supports a gelation caused by a glass transition that meets the liquid binodal and thereby arrests the gas - liquid phase separation @xcite .
we find that crystallization and vitrification are both responsible for the formation of arrested spinodal patterns ( gels ) , albeit at different interaction energies .
these two mechanisms are also responsible for the formation of two different types of long - lived , but transient clusters at low colloid density .
the clusters are crystalline at low @xmath24 and glassy at high @xmath24 . at high colloid density , on the right hand side of the gas - liquid binodal ,
the amorphous arrested phase is structurally very similar to a ( porous ) attractive glass , while the crystalline arrested phase can be seen as a polycrystalline porous material .
[ fig : sum ] shows schematically the location of the different type of phases we found in relation to the equilibrium phase diagram .
finally , we point out that our phase diagram can provide a guide to experimentally obtain and control the morphology of crystals of colloids or proteins .
the crystallization window can be located experimentally since it is delimited at low temperature by an arrested phase with liquid - like local structure and at high temperature by a mobile fluid phase .
thus , by changing the temperature it is possible to control the degree of crystallinity .
the highest fraction of crystal is obtained for state points in the lower interaction strength ( higher temperature ) part of the crystallization window
. for increasing interaction strength ( decreasing temperature ) the crystal fraction decreases . for
the range of packing fractions studied in this work ( @xmath94 ) , the colloid packing fraction does not influence the fraction of crystal or the growth rate , but has a strong influence on the morphology of the final phase . while at very low @xmath76 clusters form , a branched structure is obtained at intermediate packing fractions , and porous materials
are obtained at high @xmath76 .
we thanks neer asherie for critically reading the manuscript .
this work is part of the research program of the _ stichting voor fundamenteel onderzoek der materie _
( fom ) , that is financially supported by the _ nederlandse organisatie voor wetenschappelijk onderzoek _ ( nwo ) .
nwo - cw is acknowledge for the top - cw funding . | we systematically study the relationship between equilibrium and non - equilibrium phase diagrams of a system of short - ranged attractive colloids .
using monte carlo and brownian dynamics simulations we find a window of enhanced crystallization that is limited at high interaction strength by a slowing down of the dynamics and at low interaction strength by the high nucleation barrier .
we find that the crystallization is enhanced by the metastable gas - liquid binodal by means of a two - stage crystallization process .
first , the formation of a dense liquid is observed and second the crystal nucleates within the dense fluid . in addition , we find at low colloid packing fractions a fluid of clusters , and at higher colloid packing fractions a percolating network due to an arrested gas - liquid phase separation that we identify with gelation . we find that this arrest is due to crystallization at low interaction energy and it is caused by a slowing down of the dynamics at high interaction strength .
likewise , we observe that the clusters which are formed at low colloid packing fractions are crystalline at low interaction energy , but glassy at high interaction energy .
the clusters coalesce upon encounter . |
the problem of calculating heavy ion induced continuum @xmath0 pair production to all orders in @xmath1 has received some renewed interest in the past several years .
realization that in an appropriate gauge@xcite , the electromagnetic field of a relativistic heavy ion is to a very good approximation a delta function in the direction of motion of the heavy ion times the two dimensional solution of maxwell s equations in the transverse direction@xcite , led to an exact solution of the appropriate dirac equation for excitation of bound - electron positron pairs@xcite . given this solution ,
it was perhaps not surprising that the solution of the dirac equation was obtained independently and practically simultaneously by two different collaborations@xcite for the analagous case of continuum @xmath0 pair production induced by the corresponding countermoving delta function potentials produced by ultrarelativistic heavy ions in a collider such as rhic .
an extended discussion and reanalyis of this solution , with comments on early parallel work in the literature , shortly followed@xcite .
one apparent physical consequence of this solution was that the rates for pair production in the exact solution agreed with the corresponding perturbation theory result@xcite .
several authors subsequently argued@xcite that a correct regularization of the exact dirac equation amplitude should lead to deviations from perturbation theory , the so called coulomb corrections .
although , as has been pointed out@xcite , the derived exact semiclassical dirac amplitude is not simply the exact amplitude for the excitation of a particular ( correlated ) electron - positron pair , there are observables , such as the total pair production cross section , that can be constructed from this derived amplitude
. the exact amplitude for a correlated electron - positron pair will not be treated here .
it is the coulomb corrections to the observables that _ can _ be constructed from this exact dirac equation amplitude that are the topic of this paper . in what follows it
will be shown from a somewhat different approach from what has been done before that coulomb corrections must exist , that they arise from the physical cutoff of the tranverse coulomb potential , and the accuracy of their evaluation has been up to now limited by an effective two - peak approximation to the exact retarded dirac amplitude .
one begins the semiclassical dirac solution by representing the electromagnetic effect of one heavy ion on the other as the linhard - wiechart potential produced by a point charge on a straight - line trajectory @xmath2 ^ 2+(z - v t)^2}}\ ] ] @xmath3 is the impact parameter , perpendicular to the @xmath4axis along which the ions travel , @xmath5 , @xmath4 , and @xmath6 are the coordinates of the potential relative to a fixed target ( or ion ) , @xmath7 is the dirac matrix , and @xmath8 and @xmath9 are the charge , velocity and relativistic @xmath9 factor of the moving ion . if one makes a gauge transformation on the wave function@xcite @xmath10 where @xmath11\ ] ] the interaction potential @xmath12 is gauge transformed to @xmath2 ^ 2+(z - v t)^2 } } -{\alpha z(1-(1/v)\alpha_z)\over\sqrt{b^2/\gamma^2+(z - v t)^2}}\ ] ] the second term is pure gauge and serves to reduce the range of the potential in @xmath13 to more closely map the @xmath13 range of the @xmath14 and @xmath15 fields , which have the denomenator to the @xmath16 power rather than the @xmath17 power of the untransformed lorentz gauge potential eq .
( 1 ) . in the ultrarelativistic limit (
ignoring correction terms in @xmath18 ^ 2$])@xcite @xmath19 this is the potential that allowed the closed form solution of the dirac equation for the bound - electron positron problem .
the full solution of the problem is in perturbation theory form , but with an eikonalized interaction in the transverse direction @xmath20 - 1 ) .\ ] ] in place of the pertubation interaction eq .
( 5 ) producing the higher order effect in @xmath1 .
recall that this exact semiclassical solution produced a reduction of a little less than 10% in the predicted cross section for au + au at rhic@xcite ; one can identify this reduction as a coulomb correction to bound - electron positron pair production
. in the bound - electron positron problem one conveniently takes the electromagnetic field of one moving heavy ion seen in the rest frame of the heavy ion that receives the created electron . for production of continuum pairs in an ultrarelativistic heavy ion reaction one may work in in the center of mass frame and the electromagnetic interaction goes to the limit of two countermoving @xmath21 function potentials @xmath22 where @xmath23 the semi - classical dirac equation with this potential has been solved in closed form@xcite .
baltz and mclerran@xcite noted the apparent agreement of the obtained amplitude with that of perturbation theory even for large @xmath24 .
segev and wells@xcite also noted the agreement with perturbation theory and noted the scaling with @xmath25 seen in cern sps data@xcite .
these data were obtained from reactions of 160 gev / c pb ions on c , al , pa , and au targets as well as 200 gev / c s ions on the same c , al , pa , and au targets . the group presenting the cern data , vane et al . , stated their findings in summary : `` cross sections scale as the product of the squares of the projectile and target nuclear charges . '' on the other hand , it is well known that photoproduction of @xmath0 pairs on a heavy target shows a negative ( coulomb ) correction proportional to @xmath26 that is well described by the bethe - maximon theory@xcite .
as noted in the introduction , several authors have argued that a correct regularization of the exact dirac equation amplitude must lead to coulomb corrections .
the first analysis was done in a weizsacker - williams approximation@xcite .
subsequently , lee and milstein argued@xcite the existence of coulomb corrections by an approximate analysis of the closed form solution of the dirac equation .
we will take as our starting point a somewhat extended consideration of the results of lee and milstein . to begin let us write the previously derived semiclassical amplitude for electron - positron pair production@xcite in the notation of lee and milstein@xcite @xmath27 { \cal m}({\bf k } ) f_b({\bf k } ) f_a({\bf q_{\perp } + p_{\perp } - k}).\ ] ]
@xmath28 and @xmath29 are the four - momenta of the produced electron and positron respectively , @xmath30 is an intermediate transverse photon momentum to be integrated over , @xmath31 and the effect of the potential eq .
( 7 - 8 ) is contained in integrals , @xmath32 and @xmath33 , over the transverse spatial coordinates taking the form @xmath34 \ { \exp [ -i 2 z \alpha \ln { \rho } ] - 1 \ } \nonumber \\ & = & 2 \pi \int_0^{\infty } \rho \ , d \rho j_0(k \rho ) \ { \exp [ -i 2 z \alpha \ln { \rho } ] - 1 \}.\end{aligned}\ ] ] @xmath35 has to be regularized or cut off at large @xmath36 . how it is regularized is the key to understanding coulomb corrections .
if one merely regularizes the integral itself at large @xmath36 one obtains@xcite apart from a trivial phase @xmath37 all the higher order @xmath1 effects in @xmath38 are contained only in the phase of the denominator of eq .
( 12 ) . as we will see
, it directly follows that calculable observables are equal to perturbative results . before considering the lee and milstein analysis
, we will discuss the observables that can be calculated@xcite from the solution of a dirac equation such as eq .
( 9 - 12 ) .
we have pointed out that the derived semiclassical dirac amplitude @xmath38 is not simply the exact amplitude for the excitation of an electron - positron pair@xcite .
the point is that exact solution of the semi - classical dirac equation may be used to compute the inclusive average number of pairs not an exclusive amplitude for a particular pair .
calculating the exact exclusive amplitude to all orders in @xmath1 is not easily tractable due to need for feynman propagators@xcite .
the possibility of solutions of the semi - classical dirac equation is connected to the retarded propagators involved . in this paper
we do not consider the exclusive ( feynman propagator ) amplitude at all .
we concentrate on observables that _ can _ be construted from the above amplitude and investigate the coulomb corrections contained in them .
the occupation number or inclusive number of electrons created in state @xmath28 ( at impact parameter @xmath39 ) is @xmath40 likewise the inclusive number of positrons created in state @xmath29 is @xmath41 these inclusive expressions say nothing about correlations between electrons in state @xmath28 and and positrons in state @xmath29 .
the mean number of electron - positron pairs is of course equal to either the mean number of positrons or the mean number of electrons and may be obtained by integrating over either of the previous expressions .
@xmath42 it is possible to calculate well - defined observables from the occupation numbers by integrating over the impact parameter @xmath39 @xmath43 @xmath44 and @xmath45 @xmath46 is the cross section for an electron of momentum @xmath47 where the state of the positron is unspecified .
likewise @xmath48 is the cross section for a positron of momentum @xmath49 with the state of the electron unspecified .
note that @xmath50 corresponds to a peculiar type of inclusive cross section which we should call the `` number weighted total cross section '' , @xmath51 in contrast to the usual definition of an inclusive total cross section @xmath52 for pair production , @xmath53 now we can write for the factor common to all the cross sections @xmath54 \nonumber \\ & \times & { \cal m}({\bf k } ) { \cal m}({\bf k'})^ * f_b({\bf k } ) f_b({\bf k'})^ * \nonumber \\ & \times & f_a({\bf q_{\perp } + p_{\perp } - k } ) f_a({\bf q_{\perp } + p_{\perp } - k'})^ * .\end{aligned}\ ] ] integrating @xmath55 $ ] over the impact parameter @xmath39 in the usual way gives @xmath56 and so @xmath57 one now obtains expressions for @xmath46 , @xmath48 , and @xmath50 that appear identical to the result of perturbation theory ( scaling as @xmath58 ) when our previous expression for @xmath35 eq .
( 12 ) is employed .
@xmath59 @xmath60 @xmath61 obviously @xmath32 and @xmath33 still have to be regularized or cut off at small @xmath62 and @xmath63 .
the strategy of the first paper of lee and milstein@xcite was to evaluate coulomb corrections by taylor expanding @xmath64 around @xmath65 , i.e. @xmath66 . the derivative @xmath67 is evaluated at @xmath65 , and also in the evaluation of e.g. eq .
( 26 ) @xmath30 is ignored in @xmath68 .
all the @xmath30 dependence of the integral is then contained in @xmath69 .
lee and milstein then invite us to consider the integral representing the difference between the exact solution and the perturbative solution @xmath70\ ] ] where @xmath71 \ { \exp [ -i \chi(\mbox{\boldmath $ \rho$ } ) ] - 1 \},\ ] ] with the transverse form of the potential not yet specified @xmath72 and @xmath73 \chi(\mbox{\boldmath $ \rho$})\ ] ] is the perturbative expression limit of @xmath35 .
lee and milstein keep the @xmath74 form for @xmath75 but switch the order of integration between @xmath36 and @xmath76 .
they integrate @xmath76 to some finite upper limit @xmath77 an then claim to set @xmath77 to infinity in the resulting expression .
actually @xmath77 simply falls out of the problem by a rescaling of @xmath36 to @xmath78 .
next , after integrating over the rescaled @xmath36 , the expression they obtain is a universal function of @xmath1 @xmath79,\ ] ] where @xmath80 is the digamma function and @xmath81 is euler s constant .
this expression may be alternatively expressed as @xmath82 where @xmath83 is the same function that was derived by bethe and maximon for coulomb corrections to @xmath0 photoproduction on heavy nuclei and takes the form @xmath84 the derivation and result may seem a little mystifying .
lee and milstein state , `` thus , we come to a remarkable conclusion : although the main contribution to the integral in eq .
( 4 ) comes from the region of small @xmath76 , where @xmath85 differs from @xmath86 and depends on the regularization parameters ( the radius of screening ) , nevertheless the integral @xmath87 itself is a universal function of @xmath1 . '' as we will see later the only part of this quoted statement that is completely true is that @xmath87 is a universal function of @xmath1 .
@xmath87 is then used by lee and milstein to calculate the coulomb correction arising from ion @xmath88 by taking ion @xmath89 is to lowest order in @xmath1 .
generalizing this approach , the corresponding coulomb correction arising from ion @xmath89 is also evaluated@xcite .
the sum of these two contributions then agree with the coulomb corrections as evaluated by ivanov , schiller , and serbo@xcite using the weizsacker - williams method .
let us try to understand lee and milstein s result by putting in a physical cutoff to the transverse potential @xmath75 ( which has been up to now set to @xmath90 ) . instead of regularizing the integral itself and letting the cutoff radius go to infinity
as was originally done@xcite , we will apply an appropriate physical cutoff to the interaction potential . in the weizsacker - williams or equivalent photon treatment of electromagnetic interactions the potential is cut off at impact parameter @xmath91 , where @xmath9 is the relativistic boost of the ion producing
the photon and @xmath92 is the energy of the photon . as lee and milstein
subsequently recall ( but do not utilize ) if @xmath93 and @xmath94 is cut off in a physically motivated way , such as an equivalent photon cutoff , then @xmath95 \over r } \ ] ] where @xmath96 with @xmath97 the energy of the photon from ion @xmath89 moving in the positive @xmath4 direction and @xmath98 the energy of the photon from ion @xmath88 moving in the negative @xmath4 direction . for simplicity we will suppress the subscripts on @xmath92 , remembering however for future possible use that @xmath99 are well defined in terms of @xmath100 and @xmath101 . the integral eq .
( 34 ) can be carried out to obtain @xmath102 and @xmath103 -1 \ } .\ ] ] the modified bessel function @xmath104 for small @xmath36 and cuts off exponentially at @xmath105 .
this is the physical cutoff to the transverse potential .
one may define @xmath106 and rewrite eq.(38 ) @xmath107 - 1 \}.\ ] ] it is now clear that @xmath108 is a function of @xmath109 times some function of @xmath110 .
the perturbative limit @xmath111 is analytically soluable and takes the form @xmath112 fig .
( 1 ) displays the results of numerical calculation of the scaled magnitude of @xmath35 as a function of @xmath113 for @xmath114 ( essentially the perturbative form eq .
( 40 ) ) and for @xmath115 .
note that the upper cutoff of @xmath36 at @xmath116 has the effect of regularizing @xmath35 at small @xmath76 .
@xmath35 goes to the constant @xmath117 as @xmath76 goes to zero in the @xmath118 perturbative case ; it goes to a reduced constant value as @xmath76 goes to zero for @xmath119 .
the form of the original solution eq .
( 11 ) @xmath37 is simply wrong because it is unphysical .
since it lacks a proper physical cutoff in @xmath36 , it not only blows up at @xmath120 , but it also fails to exhibit the correct reduction in magnitude that occurs when @xmath113 is not too large .
( 2 ) is an alternate display of results of the numerical calculations showing the fractional decrease in the ratio @xmath121 for various values of @xmath24 as a function of @xmath113 .
it is clear from the two figures that for increasing @xmath24 coulomb corrections reduce @xmath35 from the perturbative result for @xmath122 . only for @xmath123
does the magnitude of @xmath35 go over into the original form of eq .
( 41 ) .
now let us consider @xmath87 again with the specific forms of @xmath35 displayed in fig .
( 1 - 2 ) @xmath70\ ] ] note that , given the @xmath124 dependence of @xmath35 and of eq .
( 39 - 40 ) , this is a logarithmic integral of @xmath76 ( i.e @xmath125 ) times a function of @xmath113 .
therefore the integration is really over the combination variable @xmath113 .
thus @xmath126 falls out of the integral , and the coulomb correction function @xmath87 does not depend on @xmath9 or @xmath92 .
i have evaluated @xmath87 numerically and found it exactly converging to lee and milstein s result according to the expected improved precision with decreasing mesh size .
i attained agreement to one part in @xmath127 . conjecturing that the detail of the cutoff should not matter
, i replaced the function @xmath128 with a different function that also goes as @xmath129 ( plus an irrelevant constant , @xmath130 ) for small @xmath36 and also cuts off exponentially at @xmath131 : @xmath132.\ ] ] calculations of @xmath87 with @xmath133 in place of @xmath134 similarly converge numerically to the result of lee and milstein with agreement to one part in @xmath127 .
note however the non - identical shapes of the contribution to @xmath87 as a function of @xmath135 for the @xmath134 and @xmath133 transverse potential forms exhibited in fig .
( 3 ) , even though the area above the two curves ( the value of @xmath87 ) is identical .
now we can begin to understand the result of lee and milstein .
the reason that `` the integral @xmath87 itself is a universal function of @xmath1 '' is that the first order @xmath136 factor from the expansion makes the integral @xmath87 logarithmic and so , contrary to what lee and milstein state , @xmath87 does not `` depend ... on the regularization parameters ( the radius of screening ) '' . the radius of screening , i.e. @xmath126 , is finite , but it has fallen out of the problem .
furthermore `` the main contribution to the integral '' does not `` come from the region of small @xmath76 '' but , as is seen from the plot of the physically motivated @xmath134 curve in fig.(3 ) , the main contribution is peaked at @xmath137 and spreads out between half - maxima at @xmath138 and @xmath139 .
note that the decoupling of the coulomb corrections from @xmath126 seen in @xmath87 is only valid to first order in @xmath76 .
including higher order terms in @xmath76 or , alternatively carrying out a full numerical evaluation of e.g. eq .
( 26 ) , would necessarily restore some dependence on @xmath126 to the coulomb corrections .
a previous monte carlo perturbation theory calculation of bottcher and strayer@xcite displays the pair production cross section as a function of @xmath140 , and shows a significant deviation between an exact monte carlo evaluation of the cross section and evaluation using a two peak approximation ( in particular see fig.(9 ) of ref.@xcite ) .
since in carrying out their calculation , lee and milstein made a variety of a two peak approximation ( assuming @xmath140 small ) , one has to assume that the precision of their results is limited .
to lowest order in transverse momentum ( small @xmath76 and small @xmath140 ) , coulomb corrections do exist as a universal function of @xmath141 , where @xmath141 is the same function of bethe and maximon derived for coulomb corrections to electron - positron pair photoproduction .
these coulomb corrections reduce the uncorrelated electron or positron production cross sections and the number weighted total pair cross section .
in general and not limited to lowest order in transverse momentum , coulomb corrections are a function of only @xmath24 and the combination variable @xmath142 .
coulomb corrections arise from the finite cutoff of the transverse spatial integral at @xmath126 and vanish for large @xmath113 .
since the cern data cover a large part of the momentum range of produced positrons and scale perturbatively , they still seem to present a puzzle .
it would be useful to carry out full calculations of the total number weighted cross section @xmath50 as well as of the uncorrelated momentum dependent electron and positron cross sections @xmath143 and @xmath144 , utilizing the transverse integrals with a correct physical cutoff . since the cern data only detects positrons , comparison with a full calculation of @xmath144 is appropriate .
i would like to thank francois gelis and larry mclerran for useful discussions .
this manuscript has been authored under contract no .
de - ac02 - 98ch10886 with the u. s. department of energy .
a. j. baltz , m. j. rhoades - brown , and j. weneser , phys . rev . * a 44 * , 5568 ( 1991 ) .
a. j. baltz , phys . rev . * a 52 * , 4970 ( 1995 ) .
a. j. baltz , phys .
lett . * 78 * , 1231 ( 1997 ) .
b. segev and j. c. wells , phys . rev . * a 57 * , 1849 ( 1998 ) .
anthony j. baltz and larry mclerran , phys . rev .
* c 58 * , 1679 ( 1998 ) .
b. segev and j. c. wells , phys . rev .
* c 59 * , 2753 ( 1999 ) .
u. eichmann , j. reinhardt , s. schramm , and w. greiner , phys . rev . * a 59 * , 1223 ( 1999 ) .
ivanov , a. schiller , and v. g. serbo , phys .
b * 454 * , 155 ( 1999 ) .
r. n. lee and a. i. milstein , phys . rev . * a 61 * , 032103 ( 2000 ) .
r. n. lee and a. i. milstein , phys . rev . * a 64 * , 032106 ( 2001 ) .
a. j. baltz , f. gelis , l. mclerran , and a. peshier , nucl . phys . *
a695 * , 395 ( 2001 ) . c. r. vane , s. datz , e. f. deveney , p. f. dittner , h. f. krause , r. schuch , h. gao , and r. hutton , phys
. rev . * a 56 * , 3682 ( 1997 ) .
h. a. bethe and l. c. maximon , phys . rev . *
93 * , 768 ( 1954 ) ; handel davies , h. a. bethe and l. c. maximon , phys .
rev . * 93 * , 788 ( 1954 ) .
j. f. reading , phys . rev . * a 8 * , 3262 ( 1973 ) . j. reinhardt , b. mller , w. greiner , and g. soff , phys .
* 43 * , 1307 ( 1979 ) .
klaus rumrich , klaus momberger , gerhard soff , walter greiner , norbert grn , and werner scheid , phys .
lett . * 66 * , 2613 ( 1991 ) .
j. c. wells , v. e. oberacker , a. s. umar , c. bottcher , m. r. strayer , j. s. wu , g. plunien , phys . rev . * a 45 * , 6296 , ( 1992 ) . c. bottcher and m. r. strayer , phys .
* d 39 * , 1330 ( 1989 ) . | coulomb corrections to perturbation theory for producing electron - positron pairs in ultrarelativistic heavy ion collisions are considered in a part - analytical , part - numerical approach .
production probabilities are reduced from perturbation theory with increasing charge of the colliding heavy ions , as has been previously argued in the literature .
it is shown here that the reduction from perturbation theory comes from the appropriate physical spatial cutoff of the electromagnetic potentials arising from the colliding ultrarelativistic heavy ions . + * pacs : 25.75.-q , 34.90.+q * = 100000 specialpagefalse @page=1 oddheaddraftto be submitted to phys .
rev .
a oddhead evenheadoddhead oddfoot@font evenfootoddfoot specialpagefalse oddhead evenheadoddhead |
the term _ duality _ , in the meaning we use it here , was introduced in @xcite in the studies of the lam system . there , potentials with different values of the elliptic parameter @xmath0 were named dual if their spectra ( i.e. , the band - gap edges ) interchange under the switch @xmath1 . in the self - dual case ( the one corresponding to @xmath2 ) , the spectrum of the potential maps onto itself under the said switch .
one way to understand this phenomenon lies through algebraisation of the lam potential , i.e. , through writing the hamiltonian in a matrix form .
the property of the spectrum to transform into itself under the change @xmath3 was originally established in the studies of a sextic quasi - exactly solvable ( qes ) potential .
this property was termed as the @xmath4_energy reflection ( er ) symmetry_@xmath4 . @xmath4 in our article , we shall refer to it as the @xmath4_energy spectrum reflection ( esr ) property_@xmath4 .
@xmath4 in contrast to the lam model , wherein _ all _ the band - gap edges can be found by the matrix means , the qes problems normally permit only for partial algebraisation of the , generally infinite , spectrum . consequently , the esr symmetry refers only to its algebraic part . in this ,
restricted sense , the sextic potential can be regarded self - dual .
therefore , it is natural to hypothesise that there also exist _ dual _ sextic potentials whose algebraic parts of the spectra interchange under @xmath5 .
such potentials were found in @xcite , and their duality was proven both by matrix methods and by the quantum - mechanical means ( with the schrdinger equation employed ) . here
we study in a great detail the latter of these two approaches . in @xcite , it was spelled out for the first time , that the symmetry of the quantum energy levels of a qes system stems from a symmetry of the corresponding classical system . in _ ibid .
_ , the symmetry of the classical system was discussed in the canonical language of phase space . in the present work ,
we shall show that classical and quantum symmetries can be uniformly described in the language of the action , for which , as we shall demonstrate , the term _ duality _ can be applied as well .
this formalism allows one to explore at a deeper level the connection between the symmetries of the classical and quantum systems .
specifically , it turns out that the symmetries of both kinds of systems can be understood from studying the global properties of the riemannian surfaces of classical and quantum momenta , correspondingly .
this section addresses the esr symmetry of two classical systems governed by sextic potentials . to explore the behaviour of such systems
, we shall employ some tools from the complex analysis , with elements of topology involved . aside from theoretical yield for classical mechanics , this study will lay the basis for the subsequent research on symmetry between the quantised counterparts of the classical systems considered .
we begin with exploring one - dimensional motion in the potential @xmath6 two sextic potentials of the type ( [ vcl ] ) , which share the values of the parameters @xmath7 and @xmath8 , and have opposite values of the parameters @xmath9 and @xmath10 , possess the following duality property : @xmath11 in the special case of @xmath12 , this becomes self - duality : @xmath13 we now introduce the canonical momentum as @xmath14 with the mass set to be @xmath15 , and with the positive sign in front of the square root chosen . for the two said potentials , the duality property of the corresponding momenta reads as : @xmath16 for an arbitrary initial energy , the motion in each of these potentials is periodic , as both potentials become infinite at @xmath17 .
the abbreviated classical action is defined as an integral of the momentum over a period : , with @xmath18 being the lagrangian . ]
@xmath19 where the contour @xmath20 encloses the two ( real ) turning points @xmath21 and @xmath22 in the complex @xmath23-plane , between which oscillations are taking place.@xmath4 should be close enough to the real axis , in order to avoid enclosing the singularities other than those located on the branch cut of the momentum between the two real turning points.]@xmath4 these two turning points are such two real solutions of the equation @xmath24 that @xmath25 and no other real solutions of lies between them . in ,
the integral along the closed contour @xmath20 is nonzero due to the presence of a branch cut of the momentum inside it , between the turning points.@xmath4 this singularity makes the contour non - contractible or , in the parlance of topology , homotopically non - trivial .
the momentum is set positive along the bottom of the cut , in order for the action to stay positive when the direction of the contour in is chosen counter - clockwise . for various values of the parameters entering the potential , and for various values of energy , the equation may have up to six real roots , see figure [ f1 ] .
we shall be interested in the following special cases : + 1 . in the case of @xmath26
the potential has a double - well shape , with two absolute minima and one local maximum .
2 . in the case
@xmath27 the following two subcases will be considered : * if @xmath28 the potential is of a single - well type , i.e. , has a single absolute minimum and no maxima ; * if @xmath29 the potential has three wells , and has one of the local minima located at @xmath30 .
the case of @xmath31 is of no interest to us and will not be considered . in both cases , and ,
there exists a range of energies for which one of the equations @xmath32 has four real solutions , while another equation has only two .
below we assume the energy @xmath33 to stay in the range where has two solutions , while has four . for a detailed discussion in application to a particular form of the potential . ]
let us write down the action for ( a ) the potential @xmath34 and energy @xmath35 , and ( b ) the potential @xmath36 and energy @xmath33 .
we define the integration cycles @xmath37 and @xmath38 as @xmath39 with @xmath40 and @xmath41 being contours in the complex @xmath42-planes of @xmath43 and @xmath44
. represents only one of the two sheets of the riemannian surface of the momentum .
the second sheet corresponds to the negative sign in front of the square root in .
however , it turns out that all the calculations needed in this section can be performed on one sheet only . ]
now we shall express the action of the potential @xmath45 in terms of the integration cycles of the potential @xmath46 .
since figures [ roots6 ] and [ roots62 ] are identical up to a rotation by the angle @xmath47 , identical are the integrals over the contours @xmath40 and @xmath48 . indeed , @xmath49 an analogous derivation works for @xmath50 .
so one obtains : @xmath51 @xmath52 [ primeq ] these integration cycles are linked to the actions as @xmath53 hereafter primes will be omitted , because the equalities make them redundant .
.,scaledwidth=100.0% ] .,scaledwidth=100.0% ] it will be helpful to introduce the cycle @xmath54 this is but an integral over a contour @xmath55 enclosing all branching points .
so evaluation of this integral can be carried out with the aid of the residue at infinity : @xmath56 the key property of this result is that @xmath57 does @xmath4_not_@xmath4 depend on the energy . in other words , @xmath58 and @xmath59 , as functions of energy , turn out to be linearly dependent .
a similar result can be obtained for @xmath60 and @xmath61 .
thus we arrive at @xmath62 an equality to be referred to as _ duality of the action_. for the sum @xmath63 to come out energy - independent , not only the duality property was essential but also a simple analytic structure of the potential .
had we started with potentials of a higher order , we would have encountered more branching points in the complex plane .
so the sum of the cycles responsible for the classical actions would no longer be proportional to the residue at infinity , because other nontrivial cycles ( enclosing additional branching points ) would have emerged : @xmath64 the latter is equivalent to @xmath65 [ 26 ] with the right - hand side ( and , therefore , also the left - hand side ) being energy - dependent .
this section was largely motivated by @xcite .
it turns out that the method developed above for a polynomial potential can be easily applied to a periodic one . before demonstrating this ,
let us recall the major differences between polynomial potentials and periodic potentials ( the latter assumed continuous along the real direction ) .
first of all , if a potential ( as a function of a complex variable ) has a single period , then , instead of the entire complex plane , it can be defined on a strip . in this case
, one of the coordinates can be compactified , and the function can be treated as if it were defined on a cylinder . if the potential is doubly - periodic , the situation is even better , in that the function can be naturally defined on a parallelogram compactifiable into a torus deprived of , at least , one point .
this point corresponds to the pole which the periodic potential does necessarily have .
indeed , while the polynomial potential has its only pole at infinity , the doubly - periodic function , in order not to be constant , must have poles inside the fundamental parallelogram , according to the liouville theorem .
the riemannian surface of the corresponding momentum function can , generally , be of a different genus .
another fundamental difference between the polynomial and periodic potentials comes from the fact that , while the former only allow for the bounded motion , the latter permit for both bounded and unbounded motion , see table [ tab : motcla ] .
unbounded motion corresponds to the energy set above the absolute maximum of the potential . in this case
, there are no turning points , and the momentum behaves quite differently for the energies below and above the absolute maximum of the potential . for the bounded motion , the integration contour in the action integral is non - contractible due to the presence of the branch cut inside it . for the unbounded motion ,
it is homotopically non - trivial due to the global properties of the riemannian surface . below absolute maximum & .types of motion in classical mechanics [ cols="^ " , ] & + to solve the schrdinger equation , it is often convenient to change variables : @xmath66 then the function @xmath67 satisfies the ricatti equation : @xmath68 in the limit @xmath69 , the function @xmath70 becomes the classical momentum .
so we shall call it the quantum momentum function ( qmf ) and introduce the notation : @xmath71 so the equation takes the form : @xmath72 the quantisation condition , whence the @xmath73-th energy level is determined , is normally obtained from the requirement of single - valuedness of the function @xmath74
. however , in @xcite a more interesting option was proposed .
it was based on the fact that the wave function corresponding to the @xmath73-th energy level has @xmath73 zeros on the real axis , between the classical turning points ( the latter points being the zeros of the classical momentum ) @xcite . in these zeroes ,
the qmf has poles .
indeed , it trivially follows from ( [ ch ] ) and ( [ 290 ] ) that @xmath75 for analytic potentials , the pole of the function @xmath76 is of the first order , and the residue at this pole is @xmath77 .
therefore , the integral of the qmf along the contour @xmath20 enclosing classical turning points is : @xmath78 where the contour @xmath20 should be close enough to the real axis , in order to avoid containing the poles and branch cuts of @xmath76 , that are off the real axis . in the classical limit ( @xmath69 ) , the series of poles inside @xmath20 coalesces into a branch cut of the classical momentum @xcite .
the function @xmath79 is sometimes referred to as the quantum action function ( qaf ) @xcite , and the equality itself as the generalised bohr - sommereld quantisation condition ( gbs ) .
the gbs is often employed as a starting point in studies of the spectra of quantum systems . the two common approaches to extract information from the ricatti equation are : 1 . to employ the expansion of the qmf in the powers of @xmath80 ( the wkb method ) :
@xmath81 a strong side of this approach lies in its clear physical meaning , in that the first two terms give the ordinary bohr - sommerfeld quantisation condition .
a disadvantage of this method is that the form of implies the neglect of exponentially small non - perturbative terms .
hence all the results are valid only when tunneling is unimportant ( i.e. , at a distance from the relative maxima of the potential ) .
this drawback can , however , be overcome , if one begins looking for the quantum action ( and , thereby , the energy ) in a form of trans - series in @xmath80 rather then of a polynomial .
moreover , until recently , the usage of the gbs was the only systematic way of finding the multi - instanton corrections to the energy in quantum mechanics @xcite .
+ substituting into the ricatti equation gives a recursive relation @xmath82 which allows to express all the higher terms through the classical momentum : @xmath83 + next , we substitute the expansion into and get : . ]
@xmath84 + the zeroth and first terms in render : @xmath85 + the constant arising from the first term is often referred to as maslov index and can be calculated in various ways @xcite .
importantly , it does not depend on the form of the potential well . after moving it to the rhs of
, we arrive to the famous bohr - sommerfeld quantisation condition : @xmath86 + one may also proceed with calculating the higher - order terms in the lhs of .
this will , for example , provide a way to generate the perturbative expansion in the cases where it exists . to this end
, one will have to solve for the energy , inverting the series term by term .
when taking the integrals , one should take into account then all the odd terms in the expansion of the qmf starting from @xmath87 are total derivatives and , so , the corresponding integrals in vanish .
2 . to employ the laurent expansion of the qmf :
@xmath88 this approach has not been studied in depth so far . in @xcite and @xcite an approximate method of finding the energies was suggested based on this idea . in this paper , we use the laurent expansion to find the residue of the qmf at infinity , in order to prove the esr property in the language of the qaf .
these results rely on those obtained in @xcite .
we also correct some overlooks made in _
ibid_. in this section we introduce the sextic qes potentials and discuss how the energy spectrum reflection property follows from the matrix form of the hamiltonians
. then we learn how the esr can be formulated and proved on the language of the quantum action .
lastly , we discuss in greater detail connection between the duality of classical and quantum actions . through section [ sexticqes ]
, we shall employ the units in which @xmath89 .
the most well - known examples of qes problems are the even and odd sixth - order polynomial potentials : @xmath90\,x^2 - \mu\left(j+\dfrac{1}{4}\right ) \quad,\vspace{2 mm } \label{even}\\ v_{\text{odd}}\,(\nu,\mu , j , x ) = \dfrac{\nu^2}{8}\,x^6 + \dfrac{\mu\,\nu}{4 } \ , x^4 + \left [ \dfrac{\mu^2}{8 } - 2 \ , \nu\ , \left ( j + \dfrac{5}{8 } \right)\right]\,x^2 - \mu\left(j+\dfrac{3}{4}\right)\quad,\vspace{4 mm } \label{odd}\\ \nonumber \nu \in \mathbb{r}^+,\quad,\qquad \mu \in \mathbb{r } \quad,\qquad j = 1/2,\ , 1,\ , 3/2,\ , 2,\,\ldots\end{gathered}\ ] ] [ v ] these potentials are even and odd in the following sense . for them , the first @xmath91 lowest even ( odd ) energy levels can be obtained through diagonalisation of the following tridiagonal matrices : @xmath92 this is owing to the fact that , after an eigenvalues - preserving quasi - gauge transformation,@xmath4[multiblock footnote omitted ] the hamiltonians corresponding to potentials can be written as : @xmath93 where the operators @xmath94 obey the commutation relation : @xmath95 = 2t^0\quad,\qquad [ t^+,t^0 ] = -t^+\quad,\qquad[t^-,t^0 ] = t^-\quad\ ] ] and therefore can be interpreted as the generators of a representation of the @xmath96 group , with @xmath97 being the spin and @xmath98 the representation dimension .
since the subsequent discussion applies to both potentials shown in ( [ v ] ) , we shall concentrate on the even one , eqn .
( [ even ] ) .
the latter can be obtained from the classical potential by setting @xmath99 \quad,\qquad c = \dfrac{\mu\,\nu}{4 } \quad,\qquad d = \dfrac{\nu^2}{8 } \quad.\label{abc}\end{gathered}\ ] ] the change of sign @xmath100 leads to @xmath101 , whence the potentials with opposite values of @xmath102 are dual . in the notation @xmath103
the duality and self - duality conditions assume the form of @xmath104 in @xcite , a theorem was proven stating that the eigenvalues of a tridiagonal matrix change their signs , when one changes the signs of the diagonal elements .
the matrix is exactly the case .
thus , in accordance with the said theorem , the energies from the algebraic parts of the spectra of the even and odd potentials change their signs under the switch @xmath105 : @xmath106 [ symmy ] here @xmath107 and @xmath108 denote the highest energy levels in the algebraic sectors of the potentials : @xmath109 for different values of the parameters , both qes sextic potentials fall either into the category or into .
this means that the corresponding classical systems are dual to each other .
let us discuss these two cases using our new notations .
here we shall limit our discussion to the even potential . 1 . in the case @xmath110
both potentials , @xmath111 and @xmath112 are of the double - well shape , see figure [ f1 ] , ( a ) and ( b ) .
the duality takes place in the following range of energies : @xmath113 2 . in the case @xmath114
one of the potentials has a single minimum , while another one has three of them , see figure [ f1 ] , ( c ) and ( d ) . for definiteness ,
assume @xmath115 .
then the energy ranges wherein the duality occurs look as follows : @xmath116 in both cases , @xmath117 and @xmath118 denote the values of the corresponding potentials at the minima , and the following equality takes place : @xmath119 from the formul and it ensues that , without loss of generality , we can assume that the equation @xmath120 has four solutions .
consequently , the equation @xmath121 will then have two real solutions . for the odd potential
, one has to replace @xmath122 with @xmath123 in both and . a different way to prove the esr property of the energy levels of dual potentials follows from relating their quantum actions .
this implies derivation of a quantum analogue to .
despite the fact that we do not know the expression for the qmf ( obtaining whereof would be equivalent to solving the schrdinger equation ) , we can calculate its residue at infinity by substituting the laurent expansion into the ricatti equation , as was proposed in @xcite .
, is small .
here we discuss a general method applicable to arbitrary potentials . ] for the time being , we shall perform our calculations for the even potential .
thereafter , we shall provide the results for the odd potential , and shall briefly discuss the difference between these two situations .
our starting point will be the equation . as was agreed above
, we assume that has four solutions . then reads as : @xmath124 where , as before , the contours are defined as in figure [ roots6 ] . from the equation and the form of the potential
, it follows that the duality property for the qmf has exactly the same form as that for the classical momenta : @xmath125 it allows us to calculate the lhs of the for the positive energy levels of the potential @xmath112 , as an integral of @xmath126 along the imaginary axis : @xmath127 the sum of the integrals of @xmath128 over the contours @xmath129 , @xmath130 and @xmath131 is equal to an integral over the large contour @xmath132 enclosing _ all _ the poles of @xmath128 .
the latter integral can be evaluated with the aid of the residue of @xmath126 at infinity : @xmath133 here we made use of the fact that the function @xmath128 has no poles other than those inside the contours @xmath129 , @xmath130 or @xmath131 .
indeed , as was demonstrated in @xcite , the poles of the qmf emerge only on the branch lines of the classical momentum , which in our case are the segments of the real and imaginary axes , enclosed by @xmath129 , @xmath130 and @xmath131 .
this remarkable property allows us to employ the same integration contours for the qmf , as those employed for the classical momentum .
we also assume the point @xmath134 to be an isolated singularity , which is indeed the case for the considered qes potential , as was demonstrated in @xcite .
the equality ( [ 36 ] ) establishes a relation between the quantisation condition for a system described by the potential @xmath111 , with the energy @xmath135 , and the quantisation condition for a system described by @xmath112 , with the opposite energy @xmath33 . in other words , starting from , we can _ derive _ the quantisation condition in the form for @xmath112 . to calculate the residue at infinity
, we introduce the variable @xmath136 : @xmath137 in terms whereof the ricatti equation assumes the shape of . ]
@xmath138 the laurent expansion for @xmath139 reads as @xmath140 the lowest power of @xmath136 is determined by the highest power of @xmath23 in @xmath141 in .
insertion of the above expansion into the integral ( [ 36 ] ) gives : @xmath142 where the contour @xmath143 encloses the origin in the complex @xmath136-plane .
to find @xmath144 , we insert the laurent series in the ricatti equation and match the coefficients of the five lowest powers of @xmath136 : @xmath145 the two sets of solutions are : @xmath146 following @xcite , we employ the condition of square integrability of the wave function @xmath147\quad.\ ] ] for large @xmath23 ( and small @xmath136 ) one gets : @xmath148 \quad,\ ] ] and one has to choose the second set of solution with @xmath149 and @xmath150 .
this yields : @xmath151 or , equivalently : @xmath152 where @xmath153 is the number of the highest energy level in the algebraic sector . moving the first integral in , with an opposite sign , from the lhs to the rhs , and then using the equality , one arrives at @xmath154 while can be written down as @xmath155 from looking at the formul and , one may misconclude that the @xmath156-th energy level of the potential @xmath111 is symmetric to the @xmath157-th energy level of @xmath112 .
this , however , would be a mistake .
there is a fundamental difference between the cases of two and four classical turning points for a given energy . in the former case
, the equation has a single solution for the energy , while in the latter case it , usually , possesses two solutions ( an exception to this rule to be discussed shortly ) : @xmath158 we intend to comply with a standard convention wherein ( a ) the notation @xmath159 stands for the @xmath73-th energy level from the bottom , and ( b ) the bottom level corresponds to @xmath160 .
to that end , the energy levels below @xmath161 should be enumerated as : @xmath162 strictly speaking , does not cover all the possibilities .
a situation may emerge , wherein the highest level below @xmath161 is @xmath163 , and it lacks a counterpart @xmath164 .
nonetheless , in this case the proper numbering is @xmath165 , as is .
see also the discussion in section [ exampleee ] .
now it becomes evident that , on the right - hand side of the equation , the quantity @xmath166 is the number of the even energy level , @xmath167 .
in other words , for an even @xmath73 , the @xmath73-th energy level is symmetric to the @xmath168-th energy level : @xmath169 it is worth noting that the energies @xmath170 and @xmath171 share a perturbative part and differ only due to the non - perturbative splitting called into being by the tunneling effects .
these effects get smaller near the bottom of the wells , and become strong closer to the maximum . at the perturbative level ( in neglect of the splitting ) , one can say that the level @xmath172 is symmetric to two levels : @xmath159 and @xmath173 .
for the odd potential , integrating @xmath174 over the contour @xmath175 furnishes : @xmath176 where the number of the highest energy level in the algebraic sector is now given by : @xmath177 then , starting again with , we arrive at : @xmath178 be mindful that in the case of the odd potential , one chooses @xmath179 when solving .
therefrom , the energy spectrum reflection reads : @xmath180 it would be worth pointing out that the equation reveals a peculiarity inherent in all potentials with degenerate minima . for such potentials , the set of integer numbers in the rhs of , for which gbs has solutions , typically has gaps in it . in the case of the double - well potential
, the set looks as : @xmath181 here @xmath182 denotes the largest value of @xmath73 for which has solutions ( or a solution ) below its maximum , while @xmath183 is the smallest one for which the solution above the maximum exists .
this , however , should not surprise us .
indeed , since the lhs of has the meaning of action , it should experience a jump , as the energy crosses the level of the maximum of the potential . for symmetric double - well potentials
, we can expect @xmath184 in quantum mechanics , the above equality is not , generally , exact though it becomes asymptotically exact for highly excited states ( i.e. , for @xmath185 ) . in the next section , this will be illustrated with examples . in this section ,
we consider the even potential with @xmath186 ( i.e. , @xmath187 ) and @xmath188 , and with different values of @xmath102 .
let @xmath189 be such a value of the integer @xmath190 in @xmath191 that solution of this equation gives the energy level @xmath192 . as we discussed above , in the case of a single well it is but the number of the energy level , @xmath193 . in the case of two wells , one either has two energy levels with numbers @xmath194 and @xmath195 or , possibly , the level @xmath194 solely . due to the esr property
, these numbers for the dual potentials are linked via @xmath196 in other words , if one of two symmetric levels lies below the maximum of the potential , another one has to reside above the maximum . 1 .
@xmath197 + in this case , the potential @xmath111 is of a single - well shape , while @xmath112 is of a triple - well shape .
the numbering of the levels is @xmath198 in the right column , one can clearly observe the pattern usual for energy splitting .
+ @xmath199 + [ sums5 ] 2 .
@xmath200 + in this case , both potentials are of a double - well shape . however , the ground state energy in @xmath111 is above the maximum of the potential : @xmath201 and the numbering of the levels is the same as in .
3 . @xmath202 + in this case , the ground state energy in @xmath111 is still above the maximum of the potential . however , the numbering of the levels changes slightly ( though this change does not touch the levels possessing the esr property ) .
@xmath203 the fifth energy level in the potential @xmath112 is no longer below the maximum of the potential .
we can summarise the difference as follows : @xmath204 in other words , for the potential @xmath205 , the fifth energy level can be found from the equation @xmath206 while for the potential @xmath207 , the fifth energy level is obtained from the equation @xmath208 4 .
@xmath209 + in this case , the ground state energy in the potential @xmath111 is below its maximum , while the first excited state is above the maximum .
this is sufficient for us , to be able to state that the fourth state of the potential @xmath112 is above its maximum .
@xmath210 + @xmath211 + [ sums25 ] + mind the difference between and .
we had to add the factor of two to @xmath212 due to the fact that the ground state is below @xmath213 for @xmath209 .
+ one may have noticed from that , as we go from the third to the fourth energy level in @xmath112 , the quantum action becomes four times larger .
this is , of course , a purely quantum effect .
the general rule for our potential is : @xmath214 with @xmath183 and @xmath182 introduced in section [ quaction ] .
@xmath215 + in this case , for @xmath111 one has two levels below its maximum , and has four levels below the maximum for @xmath112 . @xmath216
the enumeration of the energy levels possessing the esr property has not changed , and stays same as in .
@xmath217 + in this case , the self - duality takes place , see eq . .
@xmath218 the energy of the second excited state is equal precisely to the value of the potential at the maximum , which is a general rule for the integer values of @xmath97 .
@xmath219 this energy level can be treated both as ( a ) a level below the maximum , with zero non - perturbative splitting ; or ( b ) a level above the maximum , with a zero non - perturbative shift .
the summation reads as : + @xmath220 as we have already mentioned , our studies of the qes systems in the language of the qaf formalism were largely inspired by the results obtained in @xcite by geojo , ranjani and kapoor .
we agree with _ ibid .
_ on the following items : * for the sextic potential , the assumption of the qmf to have an isolated pole at infinity leads to the condition of quasi - exact solvability for this potential .
* the residue of the qmf can be calculated explicitly by substituting the laurent expansion of the qmf into the ricatti equation .
however , in @xcite an oversight was made when the authors implicitly assumed the contour enclosing the real poles of the qfm to enclose _ all _ the poles of the qmf . in our notation
( see figure [ roots6 ] ) , this oversight reads as : @xmath221 in reality , however , the singularities which are off the real axis should also be taken into account , in order to obtain the correct expression .
due to the said omission , the authors got a wrong result for the residue of the qmf at infinity , which in their case turned out to be dependent on the number of the energy level.@xmath4 from the equation ( 6 ) in @xcite had to be different from the meaning of @xmath73 from the equation ( 20 ) in _ ibid_. in the former equation , @xmath222 is related to the number of the energy level , while in the latter it is the number of the highest energy level from the algebraic sector ( @xmath107 in our notation ) . ]
now we would like to trace the connection between the duality property of the classical action and its quantum analogue . to do so
, we shall employ the wkb method and show that the equations and can be obtained in the bohr - sommerfeld approximation ( i.e. , by taking into account the leading - order and next - to - leading - order terms in the expansion of the qmf in powers of @xmath223 ) .
this however is not a general rule . in the next section
, we shall perform an analogous calculation for a periodic potential .
as we shall see , in that case it will be necessary to take into account _ all _ the perturbative orders of the wkb expansion , in order to reconstruct the exact esr property .
we begin with writing down the classical duality condition in the notation introduced in . for the even potential ,
we arrive at : @xmath224 which virtually coincides with .
it is only the extra term @xmath225 in the brackets , that makes different from .
this term arises owing to the order-@xmath80 correction to the classical action , a sum of the maslov indices .
indeed , insertion of the series expansion into the rhs of furnishes us : , we mentioned that , beginning from @xmath226 , it is only the even terms that appear in the lhs of . ]
@xmath227 thus , due to the unique feature of the considered qes potential , @xmath228 the quantum esr symmetry can be explained in the bohr - sommerfeld approximation .
had we carried out our proof starting from the usual bohr - sommerfeld quantisation condition instead of , we would have been able to prove the esr symmetry at the perturbative level only . in other words
, we would have proved that the two levels below in the double - well energy region of one of two dual potentials are symmetric to some level in the single - well region of another one .
we now offer a couple of comments on non - perturbative corrections which never showed up in our developments . as one calculates the energy in potentials with degenerate minima
, such corrections typically arise : ( a ) below the absolute maximum of the potentials , due to the tunneling between the two wells ; ( b ) above the absolute maximum ( in the single - well region ) , owing to the over - the - barrier reflection . in the case of deep wells ,
these corrections become exponentially small for the energy levels far from the maximum of the potential . in the double - well ( single - well ) region of the sextic potential ,
the corrections are negative ( positive ) for even levels , and are positive ( negative ) for the odd ones
. it would be important to undescore the following two circumstances : 1 .
up to the equation , no non - perturbative corrections have been omitted , since we had made no assumption on the specific form of the qmf .
for this reason , the equations and are _ exact _ a fact already known from the algebraic considerations , see section [ algappr ] ) .
2 . as we already stated
, the form of the series in implies the neglect of the non - perturbative terms .
the more remarkable it is that we have nevertheless arrived at the exact result .
the reason for this is easy to guess .
the instanton action ( a cycle enclosing @xmath229 in figure [ roots6 ] ) and the action emerging due to the over - the - barrier reflection ( a cycle enclosing @xmath230 in figure [ roots62 ] ) have opposite signs and cancel in . accordingly , the corrections to the symmetric energy levels have opposite signs .
the exponentially small leading - order corrections are easy to obtain , and they indeed do match . from the duality property of the potential
, it ensues that @xmath231 in this section , we consider a qes periodic potential and perform calculations analogous to those done in section [ sexticqes ] for the sextic potentials .
first , we find an exact expression interrelating the qmf s of dual potentials .
then we study its wkb expansion . through section [ ellqes ]
, we shall be using the units in which @xmath232 .
the potential turns out to be quasi - exactly solvable if one sets @xmath233 for convenience , we also set @xmath234 thus we arrive at the potential of the lam model : @xmath235 the corresponding shrdinger equation @xmath236 \psi\,(x ) = e \psi\,(x)\ ] ] has a continuous spectrum consisting of @xmath237 finite bands and an upper band extending to infinity .
for the potential @xmath238 , we label the band edges as @xmath239 $ ] , with @xmath240 . in this notation ,
the band - gap structure of the potential reads as @xmath241 the band - edge energies for this potential can be found as eigenvalues of the operator @xmath242 = j^2_x + \nu\,j^2_y - \dfrac{1}{2}\,j\,(j+1)\,{\mathds{1}}\quad , \label{lamop}\ ] ] where @xmath243 is a unity matrix , while @xmath244 are generators of the @xmath245 group in the spin @xmath237 representation , obeying the usual commutation relation @xmath246 = { { i\mkern1mu}}\,j_z}\,$ ] .
the duality property @xmath247 = - e_{2j+2-l}\,[1-\nu ] \label{erlamee}\ ] ] immediately ensues from @xmath242 = j^2_x + \nu\,j^2_y - \dfrac{1}{2}\,j\,(j+1)\,{\mathds{1}}= - \left [ \left(j^2_z + ( 1-\nu)\,j^2_y\right ) - \dfrac{1}{2}\,j\,(j+1)\,{\mathds{1}}\right]\quad , \label{lamopdual}\ ] ] because the operator @xmath248\,$ ] has the same eigenvalues as @xmath248\,$ ] . in the next section ,
we see study the duality property in the language of the quantum action .
an equation defining the qmf for the lam potential comes up through insertion of into the ricatti equation .
the duality property has the same shape as : @xmath249 now , taking into account that @xmath250 and changing @xmath251 we finally arrive at : @xmath252 the latter equality looks very much like the equation for the sextic potential , and leads us to analogous results . in terms of the quantum actions , the esr property reads as : @xmath253 [ dudu ] starting from , we shall demonstrate that condition has to be fulfilled .
the integration contours for the qmf stay the same as in the classical case , see figure [ pxnuriem ] .
the lhs of can be expressed as : @xmath254 the residue of the qmf at its pole is calculated by expanding it into the laurent series and plugging the result into the ricatti equation : @xmath255 insertion of the latter equation into the former furnishes : @xmath256 to understand the physical meaning of the index @xmath222 , recall that the equation renders us the band - edges @xmath257 and @xmath258 . in the situations where the non - perturbative terms can be neglected , this equation defines the _ locations of bands and gaps_. now let us employ the expansion for the qmf , and derive the equation order by order in @xmath80 .
the classical part of is given by : @xmath259 which in the limit @xmath260 differs from by maslov index @xmath261 .
for convenience , we introduce the notation . @xmath262
taking into consideration the higher - order wkb terms in renders : @xmath263 \\
\\= 2\,\pi\,\hbar \left [ \kappa - \dfrac{1}{2 } + \dfrac{1}{8\,\kappa } - \dfrac{1}{128\,\kappa^3 } + \ldots \right ] = 2\,\pi\,\hbar\ , \dfrac{\sqrt{4\,\kappa^2 + 1}-1}{2 } = 2\,\pi\,j\,\hbar \quad .
\end{gathered } \label{sumcancel2}\ ] ] we see that summation of all ( perturbative ) orders of the qaf reproduces the exact equality .
the explicit form of the higer - order terms in the wkb expansion of the qmf can be used to perform the calculation of the residues , see the equation .
we remind that all the odd terms starting from @xmath87 vanish ( and the corresponding residues have to be equal to zero ) . in other words , in the bohr - sommerfeld approximation , having started from , we would end up with @xmath264 in the rhs of . in the latter calculation , we managed to obtain the correct result , having completely ignored the non - perturbative effects .
the reason for this is the same as was in the case of the double - well potential .
the integration cycles corresponding to the tunneling and over - the - barrier reflection have opposite signs and cancel each other in .
in our paper , we have introduced the notion of reflection symmetry of the action in classical systems , see the equations and .
generally , we expect equalities of this kind to hold for the classical counterparts of all quantum systems having the esr symmetry of the quantum energy levels . where @xmath107 and @xmath108 stand for the numbers of the highest energy levels in the algebraic sector for the even and odd potentials , correspondingly ; while @xmath237 stands for the number of bands for the elliptic potential . for a particular form of the sextic potential , we discussed in great detail how the numbers of the energy levels are related to the integer in the rhs of the generalised bohr - sommerfeld quantisation condition .
gerald v dunne and m shifman 2002 duality and self - duality ( energy reflection symmetry ) of quasi - exactly solvable periodic potentials , _ annals of physics _
* 299 * ( 2002 ) , pp .
143 - 173 , http://arxiv.org/abs/hep-th/0204224v1[[arxiv:hep-th/0204224 ] ] .
m shifman 1998 quasi - exactly solvable spectral problems , _ proceedings of the 29th cracow school of theoretical physics : structure of fundamental interactions . 2 - 12 june 1989 .
zakopane , poland . _
itep lectures in particle physics and field theory .
edited by m. shifman_. singapore , world scientific , vol * 2 * , pp .
775 - 875 .
k g geojo , s sree ranjani and a. k. kapoor 2003 a study of quasi - exactly solvable models within the quantum hamilton - jacobi formalism , http://arxiv.org/abs/quant-ph/0207036v2[[arxiv:quant-ph/0207036 ] ] .
t gulden , m janas , p koroteev and a kamenev 2013 statistical mechanics of coulomb gases as quantum theory on riemann surfaces , _ journal of experimental and theoretical physics _ , vol .
* 117 * ( 3 ) , pp .
517 - 537 . | the phenomenon of duality reflects a link between the behaviour of a system in different regimes .
the goal of this work is to expose the classical origins of such links , and to demonstrate how they come to life in some quasi - exactly solvable problems of quantum mechanics . by studying the global properties of the riemannian surface of the classical momentum
, we reveal that the abbreviated classical action possesses a symmetry which holds also at the quantum level and underlies the energy reflection symmetry of the quantum energy levels . |
the advanced camera for surveys ( acs ; ford et al . 2002 ) , by virtue of its high spatial resolution and sensitivity , allows us to study galaxy clusters in great detail up to redshifts of unity and beyond . at these redshifts ,
galaxy clusters are still assembling and galaxies are evolving towards the populations that we observe today .
recent results from our acs intermediate redshift cluster survey ( blakeslee et al . 2003a ; lidman et al .
2004 ; demarco et al . 2005 ; goto et al . 2005
; holden et al . 2005a ; holden et al .
2005b ; homeier et al .
2005 ; postman et al .
2005 ) have shown that galaxy clusters at redshift around unity show many similarities with local clusters , in terms of galaxy populations and their distribution , but also significant differences in galaxy morphology , ellipticity , and mass luminosity ratios .
the strongest evolution observed in the early type population is a deficit of a s0 population in this sample when compared to lower redshift samples ( postman et al . 2005 ) .
this would give evidence that the formation of the s0 population is still under way in clusters at redshift unity .
one of the most striking similarities is that the tight relation between early - type galaxy colors and luminosities that applies locally ( the color magnitude relation ; cmr ) is already in place at redshifts as high as @xmath14 ( e.g. stanford et al .
1997 ; mullis et al .
the cmr in local samples of galaxy clusters presents universal properties , in terms of scatter and zero point ( bower et al .
1992 ; van dokkum et al .
1998 , hogg et al . 2004 ; lpez cruz et al . 2004 ; bell et al . 2004
; bernardi et al . 2005 ; mcintosh et al . 2005 ) that evolve back in time in agreement with passively evolving models ( ellis et al . 1997 ; stanford , eisenhardt , & dickinson 1998 ; van dokkum et al .
2000 , 2001 ; blakeslee et al . 2003a ; holden et al .
2004 ; de lucia et al . 2004 ; blakeslee et al .
acs enables accurate measurement of the scatter around the cmr , with enough precision to seriously constrain galaxy formation age , which is impossible to obtain from ground based data ( see for example holden et al .
the measurement of the cmr scatter of the first cluster in our acs cluster survey , rxj1252.9 - 292 , permitted us to constrain the mean luminosity weighted age for the ellipticals to be @xmath15 gyr ( @xmath16 ) ( blakeslee et al .
2003a ) , based on simple modeling . in this paper , we extend the results obtained in blakeslee et al .
( 2003a ) to rxj 0910 + 5422 .
rxj 0910 + 5422 is part of the acs cluster survey ( guaranteed time observation , gto , program # 9919 ) , that includes eight clusters in the redshift range at @xmath17 , selected in the x ray , optical and near ir ( ford et al .
rxj 0910 + 5422 was selected from the rosat deep cluster survey ( rosati et al .
1998 ) and confirmed with near - ir and spectroscopic observations by stanford et al .
extensive followup spectroscopy at the keck observatory has been carried out in a magnitude limited sample reaching @xmath18 mag in the central @xmath19 arcmin ( stanford et al .
2005 ; in preparation ) .
the mean redshift of the cluster was measured to be @xmath20 ( stanford et al .
2002 ) . in this paper , we combine acs imaging with ground based spectroscopy and near ir imaging to constrain galaxy ages and formation histories from the study of their color magnitude relation . we discuss the properties of the elliptical ( e ) and lenticular ( s0 ) populations separately in the light of simple galaxy formation scenarios .
rxj 0910 + 5422 was observed in march 2004 with the acs wfc ( wide field camera ) in the f775w ( @xmath21 ) and f850lp ( @xmath1 ) bandpasses , with total exposure times of 6840 s and 11440 s , respectively .
the acs wfc scale is 0.05/pixel , and its field of view is @xmath22 . the apsis pipeline ( blakeslee et al .
2003b ) , with a _ lanczos3 _
interpolation kernel , was used for processing the images .
the acs photometric zero points ( ab system ) are 25.654 mag and 24.862 mag in @xmath21 and @xmath1 , respectively ( sirianni et al . 2005 ) . a galactic reddening of @xmath23 towards rxj 0910 + 5422 was adopted ( schlegel et al .
1998 ) , with @xmath24 and @xmath25 ( sirianni et al . 2005 ) .
the acs wfc field covers an area that at the redshift of this cluster , @xmath20 , corresponds to @xmath26 in the wmap cosmology ( spergel et al .
( 2003 ) : @xmath27 , @xmath28 , @xmath29 , adopted as our standard cosmology hereafter ) .
[ cluster ] shows the acs color image with x ray contours from chandra acis ( advanced ccd imaging spectrometer ) data that have been adaptively smoothed ( stanford et al . 2002 ) .
near - ir @xmath30 and optical @xmath31-band images were obtained at palomar observatory as described in detail by stanford et al .
( 2002 ) .
optical spectroscopy of galaxies in rxj 0910 + 5422 was obtained using the low resolution imaging spectrometer ( lris ; oke et al .
1995 ) on the keck 1 and 2 telescopes ( stanford et al .
2005 ; in preparation ) .
our typical errors in redshift correspond to errors in velocity between 100 and 300 km / s .
objects for spectroscopy were chosen initially from the catalog of objects with @xmath32 mag ( vega magnitudes ) within the ir imaging area ; outside of this area objects were chosen with @xmath33 mag from the @xmath31-band image to fill out masks .
our final sample included 66% of the objects with @xmath32 mag .
spectra were obtained using the 400 lines mm@xmath34 grating for all runs except for the initial two discovery masks as reported in stanford et al .
nine more masks were observed using lris during four runs between january 2001 and february 2003 . usually each mask was observed in a series of four 1800 s exposures , with small spatial offsets along the long axis of the slits . on average ,
the seeing was 0.9 @xmath35 .
the blue side data were generally not used since the rest frame wavelengths probed at @xmath36 fall far to the blue of the spectral features of interest for galaxies in the cluster . in total , 149 redshifts were obtained .
the slit mask data were separated into slitlet spectra and then reduced using standard long - slit techniques .
a fringe frame was constructed for each exposure from neighboring exposures , each offset from the previous by 3@xmath35 , in an observing sequence for each mask , and then subtracted from each exposure to greatly reduce fringing in the red .
the exposures for each slitlet were reduced separately and then co - added .
one - dimensional spectra were extracted for each targeted object , as well as the occasional serendipitous source .
wavelength calibration of the 1-d spectra was obtained from arc lamp exposures taken immediately after the object exposures .
a relative flux calibration was obtained from long - slit observations of the standard stars hz44 , g191b2b , and feige 67 ( massey & gronwall 1990 ) .
sextractor ( bertin & arnouts 1996 ) was used to find objects in the @xmath21 and @xmath1 images and measure their magnitudes .
threshold and deblending settings were used as in bentez et al .
although we have extensive spectroscopy , the acs imaging reaches considerably deeper along the cluster luminosity function .
thus , we have chosen to use colors ( @xmath21 - @xmath1 ) and @xmath37 to isolate a set of probable cluster members . in fig .
[ colage ] , the ( @xmath21 - @xmath1 ) and @xmath37 colors are shown as a function of galaxy age , using bc03 stellar population models , redshifted to @xmath38 .
early type cluster members would have ages of at least 0.5 gyr , corresponding to @xmath390.8 mag and @xmath40 mag . at first
, we give a larger color margin and select as potential cluster members all morphologically - classified early - type galaxies with @xmath41 mag and @xmath40 mag , down to @xmath42 mag ( the limiting magnitude of postman et al .
( 2005 ) morphological classification , that included all clusters in our sample at redshift unity ) .
our results in this paper are based on this morphological classification and a detailed discussion of the uncertainties in this classification can be found in that work .
this selected sample includes 38 galaxies within the acs field .
our final colors were measured within galaxy effective radii ( @xmath43 ) , to avoid biases due to galaxy internal gradients , following the approach in blakeslee et al .
( 2003a ) and van dokkum et al .
( 1998 , 2000 ) .
@xmath43 values were derived with the program galfit ( peng et al .
2002 ) , constraining the _ sersic _ index @xmath44 ( as in blakeslee et al .
2003a ) . to remove differential blurring effects ( the psf is @xmath45 broader in the @xmath1 band ) each galaxy image in both @xmath21 and @xmath1 was deconvolved using the clean algorithm ( hgbom et al . 1974 ) .
the @xmath46 colors were measured on the deconvolved images within a circular aperture of radius equal to @xmath43 , or 3 pixels , whichever is larger .
our median @xmath43 is @xmath11 5.5 pixels ( @xmath11 13kpc at z=1.106 ) .
our final results do not change ( within the uncertainties ) if the effective radii are calculated via a two component ( sersic bulge + exponential disk ) surface brightness decomposition technique using gim2d ( marleau & simard 1998 ; rettura et al .
, in preparation ) , that permits us to better fit the galaxy light profile .
the photometric uncertainties due to flat fielding , psf variations , and the pixel - to - pixel correlation for acs ( sirianni et al . 2005 ) were estimated by measuring the standard deviation of photometry in the background for circular apertures in the range of the measured effective radii .
these photometric errors were added in quadrature to the poisson uncertainties in the measured fluxes for each object .
the derived errors in the colors are between 0.01 and 0.03 mag down to @xmath47 mag .
sextractor mag_auto were used for the @xmath1 magnitude in the color
magnitude relation ; these are fairly robust , though may systematically miss a small fraction of the light ( bentez et al .
2004 ) .
we finally color selected 34 early type ( e , s0 and s0/a ) galaxies with @xmath48 mag within 2@xmath49 from the cluster center , taken as the center of the x ray emission ( stanford et al .
images of the color selected galaxies are shown in fig [ epostage ] , fig [ sopostage ] and fig [ sapostage ] . moreover , there are late type galaxies with luminosities that are similar to the red sequence bright early type galaxies ( fig .
[ spipostage ] ) . of the 34 color selected galaxies , 15 are spectroscopically confirmed cluster members , one ( s0/a , with magnitude @xmath50 mag ) is a confirmed non - member , and the others were not targeted for spectroscopy .
the selection in @xmath46 at @xmath51 therefore appears to be robust : only one of the 16 selected galaxies with measured redshifts is a non - member .
we expect few of the other 18 to be interloper field galaxies .
the color magnitude relation for the final color selected objects is shown in fig [ cmd ] .
red dots are ellipticals , orange squares and stars are s0 and s0/a galaxies , respectively .
smaller black symbols represent early type galaxies that do not lie on the red sequence .
small triangles are late - type galaxies .
boxes are plotted around confirmed cluster members .
confirmed interlopers are circled in the figure .
surprisingly , the two brightest cluster members are not ellipticals , but s0 .
the brightest of these two galaxies lies @xmath11 700 kpc ( @xmath111.2@xmath49 ) from the cluster center , and the other bright s0 at @xmath11 300 kpc ( @xmath11 0.6@xmath49 ) . moreover , there are late type galaxies with luminosities that are similar to the red sequence bright early type galaxies .
two of them lie on the red sequence and are confirmed cluster members , at @xmath11 80 kpc from the cluster center ( see also below in the discussion of the color and morphology distribution as a function of distance from the cluster center ) .
we fitted the following linear color magnitude relation to various subsamples of the galaxies : @xmath52 the solid line in fig [ cmd ] is the fit to the color
magnitude relation for the ellipticals , the black dotted line is the fit to the cmr for the s0s , and the dashed dotted line the fit to the full sample , within 2@xmath49 from the cluster center ( see discussion below ) .
the dashed line is the fit to the full sample of early type galaxies in rxj1252.9 - 292 from blakeslee et al .
( 2003a ) , scaled to this redshift with bc03 evolved stellar population models , with solar metallicity and a formation age of 2.6 gyr ( since blakeslee et al .
2003a obtains elliptical mean ages @xmath53 2.6 gyr ) .
the long dashed vertical line is the magnitude limit of the morphological classification @xmath42 mag .
the results for different morphological samples are given in table [ results ] . a robust linear fit based on bisquare weights ( tukey s biweight ; press et al .
( 1992 ) ) has been used to fit the color
magnitude relation .
uncertainties on the parameters were estimated by bootstrapping on 10,000 simulations .
the scatter around the fit was estimated from a biweight scale estimator ( beers , flynn & gebhardt 1990 ) , that is insensitive to outliers , in the same set of bootstrap simulations . the internal color scatter ( @xmath54 ) was measured in two ways : 1 ) to the scatter around the fit , we have subtracted in quadrature the average uncertainty due to the galaxy color error ; and 2 ) we have calculated the internal scatter for which the @xmath55 of the fit would be unity .
both methods give us internal scatters consistent to within a few 0.001 mag .
all galaxies in this sample lie within three sigma from the fit .
the x ray distribution appears to be very symmetric , and largely confined within 15 from the cluster center .
we calculated the cmr zero point and scatter within 1@xmath49 ( which corresponds to a scale of @xmath11 0.5 mpc at this redshift ) , and within 15 ( @xmath110.7 mpc ) , and within 2@xmath49 ( @xmath11 1 mpc , the scale used for the analysis of rxj1252.9 - 292 ) .
according to the results in table [ results ] , the internal color scatter increases when adding populations between 1@xmath49 and 2@xmath49 , especially for the s0 and s0/a populations , as also observed in local samples ( e.g. van dokkum et al . 1998 ) , with only a small increase in sample size .
we will therefore focus on the results obtained for color selected galaxies within 1@xmath49 from the cluster center ( where 90% of the color - selected galaxies lie ) .
the slope of the elliptical cmr ( @xmath56 ) is slightly steeper than the observed slope in rxj1252.9 - 292 ( @xmath57 ) , and in coma when the latter are shifted to the observed colors at @xmath58 , using non - evolving bc03 stellar population models , but still consistent within the uncertainties .
we do not find a flatter ( with respect to coma ) slope as in stanford et al .
however , the s0 sample shows a much shallower slope ( @xmath59 ) than the ellipticals , resulting in a much flatter slope for ellipticals and s0s together ( @xmath60 ) .
this can explain why a shallower slope was found in that work , in which elliptical and s0s were not separated . the spectroscopically - selected elliptical plus s0 slope ( @xmath61 ) is also flattened by the s0 population , while the spectroscopically - selected ellipticals have a slope ( @xmath62 ) similar to that of rxj1252.9 - 292 and coma . all the difference in slope are however within the uncertainties and are statistically insignificant .
using bruzual & charlot ( 2003 ; bc03 ) stellar population models , as in blakeslee et al .
( 2003a ) , we derive a constraint on the age of the stellar populations in the galaxies from galaxy colors and the scatter of the cmr ( van dokkum et al . 2001 ; blakeslee et al .
two simple models have been considered and our conclusions will depend on the chosen models .
the first model is a _ single burst _
model , in which galaxies form in single bursts at random times @xmath63 , between the age of the cluster and the recombination epoch .
the second is a model with _ constant star formation _ in a range of time between @xmath64 and @xmath65 , randomly chosen to be between the age of the cluster and the recombination epoch .
colors for 10,000 galaxies were simulated with their scatter around the cmr to be dependent on the burst age . in fig [ scatterage ]
, we show the simulated scatters as a function of burst age , with solar , half solar and twice solar metallicity model .
we will assume solar metallicity in what follows .
from the scatter ( @xmath66 ) in the colors of the galaxies classified as ellipticals , we obtain ages @xmath67 gyr ( @xmath68 ) , with a mean luminosity weighted age @xmath69 gyr ( @xmath70 3.1 ) , assuming the random single burst model . from the constant star formation model ,
we obtain ages @xmath71 gyr ( @xmath72 ) , with a mean luminosity weighted age @xmath73 gyr ( @xmath70 3 ) .
this agrees with the conclusion ( e.g. blakeslee et al .
2003a ; holden et al .
2004 ; lidman et al .
2004 ; de lucia et al . 2005 ) that the elliptical population in clusters of galaxies formed at @xmath74 3 , and has evolved mainly passively until @xmath36 . in the ( u - b)rest frame ( using bc03 stellar population models with solar metallicity and age equal to 4 gyr ) ,
a scatter in @xmath75 of @xmath76 corresponds to @xmath77 . as pointed out in van dokkum ( 2000 ) and blakeslee et al .
( 2003a ) , cmr scatters vary little with redshift .
the blakeslee et al .
( 2003a ) scatter for the elliptical cmr in rdcs 12522927 ( @xmath78 ) , correspond to a scatter of @xmath79 in the ( u - b)rest frame , indistinguishable within the uncertainties from our result .
the scatter in the cmr for galaxies classified as s0 ( @xmath80 ) is comparable to the one in the e cmr , but the galaxy colors are bluer and are not compatible with a population that is as old as the ellipticals .
all the s0s lie below the elliptical color magnitude relation .
in fact , between the elliptical and the s0 cmr fits there is a zero point difference of @xmath10 mag , with the s0s being bluer than the ellipticals .
one of the three s0/a galaxies has a color that is 0.07 mag redder than the elliptical cmr , and another has a color that is @xmath11 0.15 bluer than the cmr relation for es .
the inclusion of the s0/a galaxies does not significantly change the fitted cmr for the s0s .
when we consider all galaxies within 2@xmath49 from the cluster center , the s0 and total early type slopes are similar , while the color offset in the cmr is still present ( @xmath81 mag ) .
the s0 population of this cluster has a very peculiar cmr with respect to the average cluster of galaxies .
in fact , for several other studies the cmr of the s0 population has a similar zero point and on average a larger scatter with respect to the elliptical population ( van dokkum et al . 1998 ; blakeslee et al . 2003a
; holden et al . 2004 ; de lucia et al .
2004 ) , quite different from our results .
we will discuss this peculiar behavior in detail in the rest of the paper , including an examination of orientation effects on the classification in fig .
[ ircolors ] , the near - ir ( vega magnitudes ) and @xmath0 ( ab magnitudes ) colors are shown compared with single burst stellar population model predictions from bc03 .
the s0 colors are consistent with young ( @xmath82 2 gyr ) solar metallicity , or older ( @xmath82 3.5 gyr ) , half solar metallicity populations . if the difference in e and s0 mean colors is mainly due to metallicity , then even if the two populations were formed at the same epoch , ellipticals must have been able to retain more metals than the s0s , i.e. , they were more massive at a given luminosity ( given the observed mass metallicity relation for early - type galaxies , e.g. , tremonti et al .
2004 , bernardi et al .
2005 , and references therein ) .
this would imply higher mass to light ratios for the ellipticals with respect to the s0s .
however , the lack of strong evolution in the slope and scatter of the cmr from the present out to @xmath83 suggests that the cmr is mainly the result of a metallicity mass ( i.e. metallicity magnitude ) relation ( e.g. , kodama & arimoto 1997 ; kauffman & charlot 1998 vazdekis et al .
2001 ; bernardi et al . 2005 ) .
so , at a given magnitude we do not expect large metallicity variations .
if the offset is due to a different star formation history , a model with solar metallicity and with an exponential decay of the star formation will reproduce the offset at a galaxy mean age of @xmath11 3.5 gyr .
this age is consistent with the small scatter observed in the s0 cmr .
we would then be observing galaxies that followed different star formation : single burst and passive evolution for the ellipticals and exponentially decaying star formation for the s0s . an exponential decay in the star formation
is observed in field spiral samples ( rowan robinson 2001 ) .
if this is the case , our s0 population might be the evolved product of an old spiral population that was already in place in this cluster when the ellipticals formed and then gradually lost available gas for star formation .
if the e vs s0 color difference is mainly due to a difference in age , for a solar metallicity and a single burst bc03 template with age 4 gyr , the color difference corresponds to an age difference of @xmath131 gyr . for clusters of galaxies at z @xmath53 1 , the cluster members on the red sequence are only a part of all the progenitors of present day early type galaxies .
some of today s galaxy progenitors would have been bluer than the red sequence at these redshifts ( van dokkum & franx 2001 ) . in the s0 population of this cluster
, we may be seeing the transitional progenitor population that in @xmath131 gyr will evolve onto the same red sequence as now occupied by the ellipticals .
either of the latter two scenarios would be consistent with the postman et al .
( 2005 ) observed deficit of the s0 population of our acs cluster sample , when compared to lower redshift samples , implying that part of the s0 population is still forming in clusters at redshifts around unity .
since the galaxies classified as s0 in rxj 0910 + 5422 are found to be systematically bluer ( with respect to the red sequence ) than the s0 populations observed in previous studies , we wish to examine further the properties of these galaxies in terms of their shapes and light distributions , and how they compare to the elliptical and spiral samples in this and other clusters . the shape parameters that we will consider are concentration and asymmetry ( abraham et al .
1996 ; conselice et al .
2004 ) , sersic index @xmath84 , and galaxy axial ratios . in fig [ cas ] ( left )
, we compare the asymmetry a and the concentration c for ellipticals , s0s , and spirals with @xmath46 colors between 0.5 and 1.2 mag .
the asymmetry parameter is obtained by subtracting a 180degree rotated image from each original galaxy image , summing the residuals and including a correction for the background .
the concentration parameter is defined as in abraham et al .
( 1996 ) as the sum of the galaxy flux within an aperture @xmath85 divided the total flux .
@xmath85 is calculated using the sextractor fit to the galaxies at 1.5 @xmath86 above the background .
the obtained semi major and semi minor axes from this fit were multiplied by 0.3 to derive the @xmath85 aperture ( see also homeier et al .
early type and late type galaxies lie on different regions in this a vs c plane .
all our red sequence s0s have @xmath87 and @xmath88 .
all but one have @xmath89 and @xmath88 .
this is the same locus in the a
c plane that is occupied by most early type galaxies in abraham et al .
( 1996 ) and in our postman et al .
( 2005 ) low redshift sample ( fig [ cas ] , right ) .
this last sample includes 5 strong lensing clusters observed as part of our acs gto program [ zw1455 + 2232 ( @xmath90 ) , ms1008 - 1224 ( @xmath91 ) , ms1358 + 6245 , cl0016 + 1654 ( @xmath92 ) , and ms j0454 - 0300 ( @xmath93 ) ] .
visual and automated classification for this cluster in the @xmath21band for all galaxies with @xmath94 was performed by postman et al .
we conclude that the s0 population presents statistical parameters typical of an early type population ( low asymmetry and high compactness ) .
fig [ justsersic ] plots sersic index @xmath84 as a function of galaxy effective radius @xmath43 , both from our galfit modeling , for ellipticals , s0s and spirals with @xmath46 colors between 0.5 and 1.3 mag .
red sequence ( @xmath46 color between 0.8 and 1.1 mag ) galaxies are shown by large diamonds .
most spirals have @xmath95 and most early types have @xmath96 , as expected .
however , the @xmath84 values do not permit us to discriminate between s0s and ellipticals in a unique way , unless they are combined with goodness of fit information for the sersic model ( e.g. the _ bumpiness _ parameter introduced by blakeslee et al . 2005 )
in fig [ ab ] we compare the apparent axial ratio from sextractor versus effective radii for elliptical and s0s with @xmath46 colors between 0.5 and 1.3 mag .
the axial ratios have been verified by using elliprof ( the isophotal fitting software that is used for surface brightness fluctuations analysis in tonry et al .
1997 and mei et al .
2005 ) on each galaxy image , after the cleaning procedure .
as above , red sequence galaxies are shown by large diamonds .
the red sequence ellipticals and the bluer s0s have different axial ratio distributions , with all red sequence s0s showing axial ratios @xmath97 , and nearly all red sequence ellipticals with @xmath98 .
assuming axisymmetric disks ( oblate ellipsoids ) viewed with random orientation , and with a gaussian distributed intrinsic axial ratio ( with mean equal to @xmath99 ( extreme thin disk ) , @xmath100 ( early type galaxy ) and @xmath101 ( elliptical ) ( jorgensen & franx 1994 ) ) , one would expect @xmath10240% , @xmath10260% , and @xmath10290% , respectively , of the s0s to have axial ratios above 0.7 .
just 9% ( 1 out of 11 ) of the red sequence s0s are observed to have an axial ratio this large ( or 22% for the full s0 sample in this cluster field ) ( fig [ ab2 ] ; top ) .
the random probability that the s0 axial ratios would show such a low fraction with @xmath98 is less than 1% .
this is a very simple model , but it points out a lack of round s0s , indicating either that there is some orientation bias in the classification , or that this class of objects is intrinsically prolate in shape .
jorgensen & franx ( 1994 ) found a similar deficit of round s0s in the center of the coma cluster .
they concluded that part of the face on s0s were classified as elliptical galaxies .
fabricant et al . (
2000 ) also found a deficit of round s0s in the cluster cl 1358 + 62 , at z=0.33 .
their analysis shows that ellipticities and bulge to total light
ratio do not allow us to distinguish elliptical from s0 galaxies .
the other two acs gto clusters at @xmath103 ( rxj1252.9 - 292 and rx j0848 + 4452 ) do not show a similar lack of round s0s , as 90% of the red sequence ellipticals ( out of @xmath10470 ) and 47% of the s0s ( out of @xmath10435 ) have @xmath98 ( fig [ ab2 ] ; bottom ) .
this bias is also not observed in other clusters of our acs intermediate redshift cluster survey ( see fig 5 from postman et al .
2005 ) , for which more than 40% of the s0 galaxies have axial ratios above 0.7 .
the observed peculiarity of the rxj 0910 + 5422 s0 axial ratio distribution might call into question our result above that the s0s have a significant color offset with respect to the ellipticals . for instance , if there is a bias in our color measurement procedure which causes elongated objects to have colors that are too blue , then the color offset found above may be artificial . such a color bias might occur if the high inclination angles bias our @xmath43 measurements to higher values , and if the s0s become progressively bluer at larger radii .
we first examine this possibility , then proceed to discuss resolutions to the peculiarity of the s0 axial ratio distribution , with the aim to establish if a misclassification of face on s0s as ellipticals would bias the measurement of the offset between the elliptical and s0 cmr zero points .
it is conceivable that our @xmath46 colors could be biased by aperture effects in the nearly edge on s0s , for which the ( possibly ) bluer outer disks might contribute more to the galaxy colors than in the rounder ellipticals population .
if this effect were severe enough , it might mimic the offset in color of the s0s and ellipticals found above .
we test this possibility here .
s0 and elliptical internal color profiles are shown in fig [ gradso ] and fig [ grade ] .
the gradients have been calculated with aperture photometry on the same images used to calculate our @xmath46 colors .
circles are @xmath46 colors at different radii , the cross the @xmath46 color at the effective radius , used for the cmr .
the s0 colors profiles do not show strong gradients .
in particular , the @xmath46 colors calculated at the effective radii are not systematically bluer than colors determined at smaller radii .
most of s0 galaxies have flat profiles ; one has a blue inward gradient ( acs i d 1621 ; @xmath105 mag , @xmath106 mag ) . in two galaxies ( acs i d 1393 and 3177 ) the colors in the central 0.15 @xmath107 are redder than the color at the effective radius .
when compared with elliptical gradients , on average s0 colors do not appear biased towards higher effective radii and bluer colors than the ellipticals .
we note that three elliptical galaxies show blue inward gradients ( acs i d 1753 , 1519 , 3323 ) .
orientation biases are known to occur in the classification of ellipticals and s0s in local galaxy samples .
for instance , rix & white ( 1990 , 1992 ) showed , based on both isophotal and dynamical modeling , that a large fraction of ellipticals contain a disk component with at least @xmath1220% of the light , but which is hidden due to projection effects .
jorgensen & franx ( 1994 ) found a strong deficit of round s0s in a sample of 171 galaxies in the central square degree of the nearby coma cluster , and concluded that inclination angle played a large part in the classification of es and s0s .
michard ( 1994 ) proposed that , except for the bright boxy ellipticals without rotational support , early - type galaxies comprise a single class of oblate rotators with orientation being the main criterion for classification as either e or s0 . on the other hand ,
van den bergh ( 1994 ) explained the predominance of flattened s0s by invoking two distinct subpopulations : bright disky objects intermediate between ellipticals and spirals , and a fainter population of prolate objects .
we now address the question of whether the galaxies classified as s0 in rxj 0910 + 5422 are preferentially flattened in shape because of an orientation bias in the classifications or intrinsically prolate shapes .
if it is an orientation bias , then this could mean either that ( 1 ) face - on s0s have been misclassified as ellipticals because their disks are not apparent , or ( 2 ) that edge - on spirals tend to be called s0s because the spiral structure is obscured .
either would result in a predominance of flattened s0s .
however , in the former case , the misclassification of face - on s0s as ellipticals would tend to blur any color separation between the two classes , while in the latter case , a color offset might be introduced between the two classes because of contamination by bluer spirals . because we do observe a color offset between galaxies classified as e and s0 , with the s0s being bluer , it is possible to look for the `` missing '' population of face - on blue galaxies by examining ellipticity versus color .
if a population of round blue objects is found , we can then determine the nature of the classification bias , and whether it biases our color offset measurement .
histograms of the axial ratio distributions for ellipticals and s0s are shown in the top panel of fig .
we find that 60% of the early - type red sequence galaxies have @xmath98 , but 95% of these low - ellipticity galaxies are classified as es .
however , if we split galaxies instead by color , using @xmath108 mag as the separation point , then we find that 54% of all galaxies bluer than this separation have @xmath98 , while 43% of the red sequence galaxies bluer than this [ @xmath109 mag ] have @xmath98 ( fig .
[ ab3 ] ) .
thus , the deficit of round s0 galaxies ( which are also significantly bluer than the mean of the e class ) is not found when the early - type galaxies are split based purely on color .
this suggests that some of the bluer round galaxies are the face - on counterparts of those classified as s0 .
fig [ blue ] shows axial ratios vs @xmath110 color residuals with respect to the total early - type galaxy cmr relation , for all galaxies in the rxj 0910 + 5422 red sequence .
galaxy types are coded with different symbols ; using the color residuals in this way takes out the effect of the magnitude dependence of the colors .
there are five round blue ellipticals , i.e. , with @xmath98 and on the blue side of the early - type cmr .
if these are the face - on counterparts of the blue s0s , then the s0 axial ratio distribution becomes much more in line with expectations for a randomly oriented disk population ( 35% are rounder than @xmath111 ) .
further , we note that four other ellipticals were classified e / s0 in postman et al .
( 2005 ) ; if these are also taken as face - on s0s , then the axial ratio distribution comes in very close agreement with expectations .
we conclude that the face - on s0s in rxj 0910 + 5422 are classified as ellipticals , just as in local early - type galaxy samples . from our axial ratio simulations and statistics from our z@xmath111 acs gt0 sample
, the number of blue s0 galaxies might be between 25 and 40% higher than estimated in section 5.3.1 .
moreover , the inclusion of a few blue s0s in the elliptical red sequence will have the effect of increasing very slightly the observed cmr scatter for the ellipticals .
the uncertainty in the postman et al .
( 2005 ) s0 fraction in rxj 0910 + 5422 is significantly larger than 40% and hence a systematic change by this amount for one cluster does not alter any of the results or conclusions in that paper . throughout the paper
, we continue to use elliptical and s0 classifications from postman et al .
( 2005 ) , but keeping in mind the presence of this possible projection effect .
stanford et al . ( 2002 ) have analyzed the x ray and near - ir properties of this cluster .
this system appears to be fairly relaxed based on its regular x
ray profile ; however , they find indications that the cluster is in an early phase of formation .
in fact , the chandra acis data show evidence for temperature structure , possibly due to an infalling group or mass streaming along a filament .
the soft component of the x ray emission ( 0.52 kev ) dominates the x ray center of the cluster , while to the south there is a harder component ( 26 kev ) ( see fig . 6 from stanford et al .
the cluster does not have a central bcg or cd galaxy , and the x ray emission center does not correspond to an optical grouping of galaxies ; rather a number of luminous confirmed cluster members are linearly distributed , at least as projected on the sky ( as shown in fig [ cluster ] ) .
we do not observe any strong trend of the galaxy color @xmath46 with distance from the cluster x ray center ( fig . [ colors ] ) .
surprisingly , though not statistically significant , the bluer galaxies are concentrated toward the cluster center , instead of the outskirts , as in the other acs intermediate redshift cluster sample ( demarco et al . 2005
; goto et al . 2005 ; homeier et al . 2005 ; postman et al . 2005 ) .
this tendency might support the hypothesis that we are observing sub - groups of galaxies in an edge on sheet , e.g a group of bluer disk galaxies on a redder , older population of ellipticals .
there are 10 confirmed elliptical and 5 confirmed s0 members in the center of the cluster ( r @xmath82 500 kpc ) .
the average s0 redshift is 1.102 @xmath112 0.002 and the average elliptical redshift is 1.105 @xmath112 0.007 ( the given uncertainties are standard deviations around the mean ) .
this indicates small relative velocities ( the two redshifts are indistinguishable given the errors ) between the classes and is true regardless of the possible classification bias discussed above .
unfortunately , our spectroscopic sample does not permit us to track the cluster central structure in detail .
the average relative velocity between the confirmed e and s0s of @xmath11 500 km / s ( that also corresponds to the cluster velocity dispersion ; see below ) is fairly small . if merging of two distinct groups of galaxies is happening along the line of sight , we expect much higher velocity dispersions and/or relative velocities between the infalling s0s and the ellipticals .
stanford et al . ( 2002 ) also suggest that active galaxy - galaxy merging should be observed , based on the x ray temperature structure . to investigate any on going dynamical activity , we calculated the cluster velocity dispersion , and the merger rate . from the 25 spectroscopically confirmed members ( all galaxy types included ) the line of sight rest
frame velocity dispersion is @xmath113 km / s , using the software rostat from beers , flynn , & gebhardt ( 1990 ) .
the available chandra data give x - ray temperatures ranging from @xmath114 kev ( stanford et al .
2002 ) to @xmath115 kev ( ettori et al .
wu et al . ( 1999 ; see also rosati et al .
2002 ) gives the relationship between @xmath116 and @xmath86 for relaxed clusters , which predicts that the velocity dispersion corresponding to the measured x - ray temperature should be @xmath11 1000 km / s , considerably higher than we have found .
again , in the case of merging groups ( along the line of sight ) we would also have expected a higher velocity dispersion .
the quality of the acs data allows us to discern merging activity among the cluster galaxies . if we assume that galaxy pairs with projected separations less than @xmath117 kpc are physically associated , we observe 13 associated early type galaxies , nine galaxies of which lie on a filamentary structure about @xmath11 100 kpc from the cluster center ( fig .
[ cluster ] ) .
as noted above , rxj 0910 + 5422 lacks any cd galaxy near the center of the x ray emission ( see also fig . [ cluster ] ) , but rather has a filamentary group of galaxies around the x ray center .
the nine early - type interacting galaxies within this filamentary structure ( at radius of @xmath118 kpc from the x ray center ) include 3 unique pairs ( yellow arrows in fig .
[ cluster ] ) , plus a galaxy triplet ( the three components marked with red arrows in fig . [ cluster ] ) .
each of the 3 pairs consists of a bright elliptical with a smaller companion ( all closer than @xmath119 kpc ) , while the triplet is a large e with two smaller s0s ( also closer than @xmath119 kpc ; one of these two s0s is acs i d 1621 , the s0 with an inward blue gradient ) .
one of the pairs ( the middle pair in the figure ) and the two nearest galaxies in the triplet , have essentially zero relative velocity and thus are likely merger candidates .
also two of the ellipticals with blue inward gradients lie on the central filaments , and they are both small satellites of a larger galaxy .
the other two pairs , which do not lie on the filamentary structure , are at @xmath120 kpc ( two e with similar size , both with weak o ii emission ( stanford et al . 2002 ) and @xmath121 kpc ( one s0/a and one e of similar size ) from the x ray center , and have relative velocities of 10,000 km / s and 2000 km / s , respectively .
the presence of the low - velocity pairs is consistent with the low velocity dispersion , and provides evidence for the on going hierarchical growth of the cluster ( e.g. , van dokkum et al .
the pairing of the bright ellipticals with smaller s0s also argues against the view that we are observing a blue s0-dominated group infalling into a red cluster elliptical population , but rather complex stellar population evolution within a filamentary structure .
the red sequence s0/a confirmed members also lie within this filamentary structure .
the observations of a significant number of red galaxy pairs in a cluster at z @xmath13 1 is interesting in the context of the recent findings by van dokkum ( 2005 ) of red galaxy interactions in @xmath11 70% of 86 early type galaxies in a selected sample of nearby red galaxies from the musyc ( multiwavelength survey by yale chile ; gawiser et al .
2005 ) and the noao deep wide field survey .
this work concluded that most of the ellipticals in local samples were assembled by red galaxy
galaxy mergers , denominated _ dry _ mergers because they would involve gas poor early type galaxies . at higher redshift , tran et al .
( 2005b ) confirmed red galaxy mergers first observed by van dokkum ( 1999 ) in ms 1054 - 03 at z=0.83 .
tran et al . selected mergers as associated pairs with projected separation less than @xmath119 kpc and relative line of sight velocities less than 165 km / s . as in rxj 0910 + 5422
, the red early type galaxies involved in these mergers are among the brightest cluster members .
their results suggest that most early type galaxies grew from passive red galaxy
galaxy mergers . in our sample
we observe a triplet and three red galaxy pairs with projected distances less than @xmath119 kpc . of those ,
the triplet and one galaxy pair show zero relative velocity . of the two other pairs , composed of a bright and a fainter companion , redshifts are not available for the faint companions .
to obtain a deeper understanding of the rxj 0910 + 5422 galaxy population , we constructed galaxy luminosity functions in the following way .
we start with the original sextractor catalog ( described in section 3 ) .
all objects with magnitudes brighter than @xmath122 mag were considered as foreground objects .
nine of these bright objects are confirmed non - members .
the remaining seven are objects that do not belong to the red sequence and whose sizes and luminosities are much larger than those of the confirmed members , in particular those of the bright red sequence galaxies ; therefore they are very unlikely to be at the cluster redshift .
the contribution to the luminosity function from both foreground and background field galaxies ( hereinafter , the field ) has been estimated from the galaxy counts in a reference field .
the control region is taken from the goods - s ( great observatories origins deep survey
south ; giavalisco et al .
2004 ) acs field , observed in the same filter as the cluster field .
point like objects were eliminated in a consistent way in the cluster and in the control field , by identification of the stellar locus in the diagnostic plot of the sextractor parameters
mag_auto vs flux_radius ( the selected objects have fwhm equal to the psf in the image ) .
cluster and control field luminosity functions were normalized to the cluster area .
both cluster and field counts were binned with a bin size of 0.5 mag . for each bin ,
the field counts are subtracted from the cluster counts , taking into account the extensive spectroscopic sample ( more than 60% of the objects used for the lf determination brighter than m@xmath123 have measured redshifts ) .
known interlopers were excluded from the analysis .
the uncertainties in the cluster counts after subtraction of the field contribution are calculated by adding in quadrature poissonian uncertainties .
the luminosity functions are shown in fig [ lumfun ] .
the filled circles with errors are the total background
corrected cluster luminosity function .
we do not include errors from cosmic variance due to the choice of the background control region . the red sequence elliptical and spheroidal ( s0 and s0/a )
luminosity functions are shown respectively in blue and green .
the red histogram is the luminosity function of all early type galaxies with color @xmath124 mag , excluding confirmed non members .
the histogram of red sequence galaxies in rxj1252.9 - 292 is shown as the dashed red line .
the background contribution is very small for the early - type sample .
the red arrows show the histogram values after background subtraction .
the rest - frame @xmath125 magnitudes are shown on the top of the plot , calculated from colors obtained from the bc03 stellar population model and templates ( sbc , scd ) from coleman , wu & weedman ( 1980 ) .
the solid black line is a schechter function fit to the total cluster luminosity function .
it is obtained by calculating the c ( cash 1979 ) statistic ( a maximum likelihood statistic to fit data with poissonian errors ) on a grid in the m@xmath123
@xmath126 plane for each combination of m@xmath123 and @xmath126 : first the normalization ( @xmath127 ) is calculated in order to reproduce the observed number of galaxies in the observed magnitude range , then the c statistic is computed as @xmath128 , where @xmath129 is the observed number of galaxies in the @xmath130th bin and @xmath131 is the number of galaxies predicted in that bin by the schechter function with parameters m@xmath123 , @xmath126 , and @xmath127 .
the combination m@xmath123,@xmath126 which minimizes the c statistic is taken as the best - fit .
if the c statistic is defined as above , the 1- 2- and 3-@xmath86 confidence levels for m@xmath123 and @xmath126 can be estimated from @xmath132 .
we obtain m@xmath133 mag and @xmath134 .
most of the faint
end population is composed of s0 and s0/a galaxies .
the two brightest galaxies in the red sequence are s0 . with respect to rxj1252.9 - 292 , a bright population of red sequence ellipticals is missing in rxj 0910 + 5422 .
however , the large poissonian errors on the bright end of the cluster population prevent us from definitively excluding the hypothesis that the two clusters could be drawn from the same parent population .
similarly , small number statistics do not permit us to study the luminosity functions of the different red and blue faint populations in this cluster .
in this paper we have studied the color magnitude relations of galaxies in rxj0910 + 5422 to constrain their ages and formation histories .
our results show that the color
magnitude relation for the elliptical galaxies is consistent both in slope and scatter with that of rxj1252.9 - 292 ( blakeslee et al
. 2003a , lidman et al . 2004 ) and recent results from holden et al . (
2004 ) and de lucia et al .
( 2004 ) , confirming that elliptical galaxies in galaxy clusters show a universal color
magnitude relation consistent with an old passively evolving population even at @xmath135 . from the color
magnitude relation of the ellipticals , we derive a mean luminosity weighted age @xmath6 gyr ( @xmath7 3 ) .
we find that the s0s in rxj 0910 + 5422 define a color magnitude sequence with a scatter similar to that found for the ellipticals , but shifted bluer by @xmath10 mag .
this is peculiar with respect to previous cluster studies , which more typically found that the s0s followed the same cmr as the ellipticals , but with somewhat larger scatter ( bower et al .
1992 ; ellis et al .
1997 ; stanford et al . 1997 ; lpez cruz et al . 2004 ; stanford , eisenhardt , & dickinson 1998 ; van dokkum et al . 2000 , 2001 ; blakeslee et al . 2003a ; holden et al .
2004 ) . only one earlier study , van dokkum et al .
( 1998 ) found a significantly bluer s0 population .
we examine this population of blue s0s in some detail , noting that there is a strong predominance of flattened systems with axial ratios @xmath98 , and conclude that the face - on members of the population have likely been classified as ellipticals . if so , the color offset between the two classes would become even more significant , and the true cmr scatter for the ellipticals would be slightly lower than we have estimated .
this peculiarity is not observed in other clusters of our acs intermediate redshift cluster survey , and its amplitude is smaller than the uncertainties adopted in postman et al .
( 2005 ) .
if the observed color difference between the ellipticals and s0s is mainly due to metallicity at the same age , this would imply that the redder ellipticals were able to retain more metals than the s0s , i.e. , they are more massive . however , current data suggest that the cmr is mainly the result of a metallicity mass ( i.e. metallicity magnitude ) relation ( e.g. , kodama & arimoto 1997 ; kauffman & charlot 1998 vazdekis et al . 2001 ; bernardi et al .
this implies that we do not expect large metallicity variations at a given magnitude .
if , instead , the offset is mainly due to age , then the implied age difference would be @xmath131 gyr for single - burst solar metallicity bc03 models .
it could also result from different star formation histories , with the s0s experiencing a more extended period of star formation .
a model with solar metallicity and with an exponential decay of the star formation reproduces the offset at a galaxy mean age of @xmath11 3.5 gyr .
the blue s0s may comprise a group infalling from the field onto a more evolved red cluster population , or they may be a transitional cluster population not yet evolved all the way onto the elliptical red sequence ( van dokkum & franx 2001 ) .
assuming passive evolution , they will reach this red sequence after about 1 gyr .
high fractions of faint blue late type galaxies were observed in substructures infalling in a main cluster ( e.g. abraham et al .
1996 ; tran et al .
2005a ) , and proposed as the progenitors of faint s0s in clusters .
the view of this cluster as a structure still in formation is supported by x
ray observations of the cluster temperature structure ( stanford et al .
2002 ) , the lack of a cd galaxy , and its filamentary structure that suggests merging of substructures .
however , we derive a small cluster velocity dispersion , unusual for merging substructures . moreover , the blue s0s in this sample span the same luminosity range of the bright ellipticals , are distributed towards the center of the cluster , and some of the faintest ones are physically associated with brighter ellipticals belonging to the central filamentary structure .
these elements would argue against the bluer s0s being a young group merging with an existing red cluster population , and support the hypothesis that we are observing a transitional blue s0 population in a cluster core that is still evolving onto the elliptical red sequence .
this result is also consistent with the deficit of s0s observed in our acs cluster sample , when compared to lower redshift samples , that implies that the s0 population is not yet in place but still forming in clusters at redshifts around unity ( postman et al .
interestingly , we also observe in this cluster potential progenitors for bright s0 galaxies : four bright spirals ( spectroscopically confirmed cluster members ) with @xmath1 brighter than 22.5 mag and @xmath75 between 0.5 and 1.3 mag .
red galaxy pairs are also observed .
a triplet and three red galaxy pairs have projected distances less than @xmath119 kpc , and of those , the triplet and one pair show zero relative velocity .
this would be the evidence of red galaxy mergers at z@xmath13 1 .
van dokkum ( 2005 ) and tran et al . (
2005b ) have observed mergers of red galaxies in a nearby elliptical sample and in ms 1054 - 03 at z=0.83 , respectively . they suggested a scenario in which most of the early type galaxies were formed from passive red galaxy
galaxy mergers , called _ dry _ mergers , because they involve gas poor early type galaxies .
future papers will analyze the ages and masses of the cluster members using our optical spectroscopy along with newly obtained spitzer irac imaging .
a larger sample would be needed to draw firmer conclusions about the formation of s0s .
acs was developed under nasa contract nas 5 - 32865 , and this research has been supported by nasa grant nag5 - 7697 and by an equipment grant from sun microsystems , inc .
the space telescope science institute is operated by aura inc .
, under nasa contract nas5 - 26555 .
we are grateful to k. anderson , j. mccann , s. busching , a. framarini , s. barkhouser , and t. allen for their invaluable contributions to the acs project at jhu .
we thank w. j. mccann for the use of the fitscut routine for our color images .
sm thanks tadayuki kodama for useful discussions . cccccccccccc + sample & @xmath136&@xmath137&@xmath138 & @xmath54 + & & ( mag)&&(mag ) & + + e+s0+s0/a@xmath139 & 31 & 0.99 @xmath112 0.01 & -0.030 @xmath112 0.020 & 0.060 @xmath112 0.008 + e+s0@xmath140 & 14 & 1.00 @xmath112 0.02 & -0.010 @xmath112 0.033 & 0.054 @xmath112
0.009 + e@xmath139 & 19 & 1.02 @xmath112 0.01 & -0.033 @xmath112 0.015 & 0.042 @xmath112
0.011 + e@xmath140 & 10 & 1.02 @xmath112 0.04 & -0.020 @xmath112 0.044 & 0.047 @xmath112
0.022 + s0@xmath139 & 9 & 0.95 @xmath112 0.02 & 0.005 @xmath112 0.023 & 0.044 @xmath112 0.02 + s0+s0/a@xmath141 & 12 & 0.95 @xmath112 0.02 & -0.007 @xmath112 0.027 & 0.057 @xmath112 0.015 + + e+s0+s0/a@xmath142 & 32 & 0.99 @xmath112 0.01 & -0.032 @xmath112 0.019 & 0.060 @xmath112 0.008 + e+s0@xmath143 & 15 & 0.99 @xmath112 0.02 & -0.021 @xmath112 0.034 & 0.054 @xmath112
0.009 + e@xmath142 & 19 & 1.02 @xmath112 0.01 & -0.033 @xmath112 0.015 & 0.042 @xmath112 0.011 + e@xmath143 & 10 & 1.02 @xmath112 0.04 & -0.020 @xmath112 0.044 & 0.047 @xmath112
0.022 + s0@xmath142 & 10 & 0.95 @xmath112 0.02 & -0.012 @xmath112 0.036 & 0.051 @xmath112
0.018 + s0+s0/a@xmath144 & 13 & 0.96 @xmath112 0.02 & -0.015 @xmath112 0.033 & 0.065 @xmath112 0.015 + + e+s0+s0/a@xmath145 & 34 & 0.99 @xmath112 0.01 & -0.036 @xmath112 0.018 & 0.059 @xmath112 0.008 + e+s0@xmath146 & 15 & 0.99 @xmath112 0.02 & -0.022 @xmath112 0.035 & 0.054 @xmath112
0.009 + e@xmath145 & 20 & 1.01 @xmath112 0.01 & -0.032 @xmath112 0.015 & 0.044 @xmath112
0.010 + e@xmath146 & 10 & 1.02 @xmath112 0.04 & -0.021 @xmath112 0.046 & 0.047 @xmath112 0.022 + s0@xmath145 & 11 & 0.96 @xmath112 0.02 & -0.022 @xmath112 0.038 & 0.053 @xmath112 0.015 + s0+s0/a@xmath147 & 14 & 0.96 @xmath112 0.02 & -0.024 @xmath112 0.034 & 0.065 @xmath112 0.013 + | the color - magnitude relation has been determined for the rdcs j0910 + 5422 cluster of galaxies at redshift z = 1.106 .
cluster members were selected from hst advanced camera for surveys ( acs ) images , combined with ground based near ir imaging and optical spectroscopy .
postman et al .
( 2005 ) morphological classifications were used to identify the early - type galaxies . the observed early type color magnitude relation ( cmr ) in @xmath0 versus @xmath1 shows an intrinsic scatter in color of @xmath2 mag , within @xmath3 from the cluster x ray emission center .
both the ellipticals and the s0s show small scatter about the cmr of @xmath4 mag and @xmath5 mag , respectively . from the scatter about the cmr , a mean luminosity weighted age @xmath6 gyr ( @xmath7 3 )
is derived for the elliptical galaxies , assuming a simple stellar population modeling ( single burst solar metallicity ) .
this is consistent with a previous study of the cluster rdcs1252.9 - 292 at @xmath8 ( blakeslee et al . ) .
strikingly , the s0 galaxies in rdcs j0910 + 5422 are systematically bluer in @xmath9 by @xmath10 mag , with respect to the ellipticals .
the blue s0s are predominantly elongated in shape ; the distribution of their ellipticities is inconsistent with a population of axisymmetric disk galaxies viewed at random orientations , suggesting either that they are intrinsically prolate or there is some orientation bias in the s0 classification .
the ellipticity distribution as a function of color indicates that the face - on s0s in this particular cluster have likely been classified as elliptical .
thus , if anything , the offset in color between the elliptical and s0 populations may be even more significant .
the color offset between s0 and e corresponds to an age difference of @xmath11 1 gyr , for a single - burst solar metallicity model .
alternatively , it could be the result of a different star formation history ; a solar metallicity model with an exponential decay in star formation will reproduce the offset for an age of 3.5 gyr , i.e. the s0s have evolved gradually from star forming progenitors .
the color offset could also be reproduced by a factor of @xmath122 difference in metallicity , but the two populations would each need to have very small scatter in metallicity to reproduce the small scatter in color .
the early type population in this cluster appears to be still forming . the blue early - type disk galaxies in rdcs j0910
+ 5422 likely represent the direct progenitors of the more evolved s0s that follow the same red sequence as ellipticals in other clusters .
thirteen red galaxy pairs are observed and the galaxies associated in pairs constitute @xmath1340% of the cmr galaxies in this cluster .
this finding is consistent with the conclusions of van dokkum and tran et al . that most of the early type galaxies grew from passive red mergers . |
both grand unified theories ( gut ) @xcite and brane world scenario with extra dimensions , though neither are observed in nature yet , have fascinated us very much @xcite .
so far , they have been developed mostly as independent paradigms . in this work ,
we provide a framework in which gut and the brane world scenario inseparably cooperate in extending standard model ( sm ) .
a rich vacuum structure of gut , especially in supersymmetric theories , often gives a number of discrete vacua which can form domain walls . in @xmath4 dimensional spacetime , domain walls can provide spontaneous generation of our @xmath1 dimensional world @xcite .
there have been a few works @xcite combining gut and extra dimensions by considering a domain wall in which all gut particles are trapped .
in contrast , we aim at _ unifying _ them in such a way that the breaking pattern of gut gauge symmetry and profile of massless sm particles ( one - generation of quarks and leptons ) are determined as a consequence of wall dynamics .
while domain wall solutions often have unwanted flat directions associated to moduli , our model stabilizes them with mass @xmath3 .
we overcome the long - standing problem of localizing gauge fields on the domain wall @xcite by adopting field - dependent gauge kinetic term @xcite .
this localization mechanism assures charge universality of matter fields by preserving the @xmath1-dimensional gauge invariance .
the proton decay can be highly suppressed .
we extend the minimal @xmath0 gut in @xmath1 dimensions to @xmath4 dimensions .
in addition to fermionic matters @xmath5 in an anti - fundamental representation and @xmath6 in an anti - symmetric representation , we have two scalar fields @xmath7 and @xmath8 which are 5 by 5 matrices . @xmath9 and @xmath10 are in the adjoint representation while @xmath11 and @xmath12 are singlets . the minimal gut in @xmath1 dimensions
needs only @xmath9 to break @xmath0 gauge symmetry . in our model
, @xmath9 plays the same role , but together with @xmath11 , it provides domain walls and traps the chiral fermions on them . on the other hand ,
the role of @xmath13 is to localize the massless gauge fields .
our lagrangian consists of three parts @xmath14 .
we have @xmath15 $ ] with @xmath16}^2\,,\\ v_1 & = -\mu^2 \hat t^2 + \alpha \hat t^4\ , , \label{eq : v1}\end{aligned}\ ] ] where metric is @xmath17 , and @xmath18 , @xmath19 and @xmath20 are mass and coupling constants , while the covariant derivatives are defined by @xmath21}$ ] and similarly for @xmath13 .
we choose the potential ( [ eq : volpot ] ) to be simple to ensure analytic solutions if @xmath22 is absent @xcite .
the role of @xmath22 in ( [ eq : v1 ] ) with @xmath23 is to stabilize undesirable moduli of the domain wall solutions .
@xmath24 includes the field - dependent gauge kinetic function @xcite @xmath25\,,\ ] ] with @xmath26}$ ] .
@xmath27 is given by _
f & = & i |_|5^m d_m_|5 + i + & & + h|_|5t^t_|5 + h , [ eq : l_f ] where @xmath28 and @xmath29 are yukawa couplings , while the covariant derivatives are @xmath30 , and @xmath31 .
let us first consider the case of @xmath32 .
@xmath33 gives the multiple degenerate vacua @xmath34 , @xmath35 with @xmath36 .
up to symmetry transformation , we can label the vacua by the number of @xmath37 eigenvalues of @xmath38 .
for example , we refer to the vacuum with @xmath39 as @xmath40 .
clearly , @xmath38 determines breaking pattern of @xmath0 . in the vacua , the components in @xmath41 except for those eaten by the gauge fields have masses @xmath42 , and all components in @xmath13 have @xmath43 .
the fermion masses are @xmath44 and @xmath45 .
we assume all these masses to be the same order or larger than @xmath3 .
since all the vacua are degenerate , static and stable domain walls exist .
we choose the @xmath0 unbroken vacua as the boundary condition @xmath46 , @xmath47 at @xmath48 , and assume @xmath49 .
then we find exact domain walls solutions @xmath50 with a @xmath51 hermitian matrix @xmath52 containing all the parameters i.e. moduli of the solution . without loss of generality , we can diagonalize @xmath52 . depending on the number of coincident walls and the ordering of their positions on @xmath53 , we find ten qualitatively different patterns .
phenomenologically most interesting one is the 3 - 2 splitting configuration with @xmath54 , where three walls are located at @xmath55 and the remaining two are at @xmath56 , interpolating the left - most vacuum @xmath57 , the middle @xmath40 , and the right - most vacuum @xmath58 .
the @xmath0 is broken to @xmath59 in the middle @xmath40 vacuum . to identify the four dimensional effective fields and their spectra around this configuration ,
let us consider small fluctuations @xcite t & = & t_0(y ) + ( t_3(x^,y ) & t(x^,y ) + t(x^,y)^ & t_2(x^,y ) ) , [ eq : fluctuation ] where @xmath11 stands for the background solution , @xmath60 ( @xmath61 ) is @xmath62 ( @xmath63 ) hermitian matrix , and @xmath64 is 3 by 2 complex matrix .
we consider fluctuations for @xmath13 similarly .
the gauge fields are given by @xmath65 where @xmath66 corresponds to @xmath67 gauge field , and @xmath68 is the @xmath69 complex matrix .
it was found @xcite that the lightest modes in @xmath70 are massless , and those with the index 3 ( 2 ) are localized around the 3 ( 2 ) coincident walls at @xmath71 ( @xmath72 )
. localization of the gauge fields is achieved by the field dependent gauge kinetic term ( [ eq : ohta ] ) .
let @xmath73 and @xmath74 .
then the fluctuations @xmath75 ( @xmath76 ) in the axial gauge @xmath77 is expanded a _
, = _ n _ , ^(n)(x ) v_^(n)(y)/_(y ) , [ eq : wf_gauge ] with the four - dimensional vector fields @xmath78 . here
we are interested in divergence free part @xmath79 .
the lowest mode was found @xcite to be massless and its mode function is flat @xmath80 .
nevertheless , the gauge field zero modes are localized on domain walls thanks to the field dependent gauge kinetic function @xmath81 in ( [ eq : ohta ] ) which provides an effective mode function @xmath82 hence , the massless @xmath83 ( @xmath84 ) gauge fields are localized around @xmath71 ( @xmath72 ) , and those of @xmath85 are around both @xmath72 and @xmath86 .
the dimensionless effective gauge couplings for @xmath87 gauge group are @xmath88 .
the zero modes @xmath89 in @xmath64 in eq.([eq : fluctuation ] ) and @xmath90 in @xmath91 ( the off - diagonal component of @xmath13 ) are the would - be nambu - goldstone ( ng ) modes for the spontaneous breaking of @xmath92 , and are extended between the 3 coincident walls and the 2 coincident walls ( t^(0 ) + s^(0 ) ) = ( _ 3 - _ 2 + _ 3 - _ 2 ) ^(0)(x ) ^(0)^(0)(x ) , [ eq : zero_t ] where @xmath93 is a 3 by 2 complex matrix and @xmath94 .
they are absorbed by the off - diagonal gauge fields @xmath68 in eq.([eq : vectorfluctuation ] ) due to the geometric higgs mechanism @xcite .
let us expand the divergence free part of @xmath68 ( @xmath95 ) as b_= _
n ^(n)_(x ) ^(n)(y)/_+(y ) . with @xmath96 .
although it is difficult to derive the spectrum analytically we can obtain qualitative features @xcite . when @xmath97 vanishes gauge field has zero mode ( no symmetry breaking ) .
hence the mass of the lightest mode starts linearly for small @xmath98 as @xmath99 for @xmath100 . on the other hand , when @xmath101 , @xmath68 is localized around the well - separated walls with the width @xmath102
therefore , the mass becomes independent of @xmath98 as @xmath103 .
it was found that all the other excited modes in @xmath41 , @xmath13 and @xmath104 have masses of the order or heavier than @xmath3 , so that they are decoupled in the low energy physics on the domain walls @xcite .
so far , we have considered @xmath32 where we have degeneracy of symmetric and symmetry - breaking vacua .
it yields undesirable massless scalar fields of @xmath105 ( @xmath106 ) and light vector fields @xmath68 for @xmath100 . from now on we incorporate the moduli - stabilization potential @xmath22 to make all the undesirable massless or light fields sufficiently massive by lifting the vacuum degeneracy as @xmath107 . by excluding configurations with the highest vacuum @xmath108 out of ten possible patterns of wall configurations ,
we are left with only three possibilities : 3 - 2 splitting , 2 - 1 - 2 splitting or no splitting .
( solid lines ) for @xmath109 , @xmath110 , @xmath111 and @xmath112 .
the broken curves stand for the analytic solution ( [ eq : gensol3 ] ) for @xmath113 with the same separation @xmath114 .
( top ) the mode functions of the fermions and massive gauge boson with @xmath115 .
the shared horizontal axis is @xmath53.,width=321 ] to figure out which is the ground state configuration , we survey the parameter space @xmath116-@xmath117 by using the gradient flow equation @xmath118 @xcite . here
, @xmath119 stands for the scalar fields and we introduce a fictitious flow - time @xmath120 and let the field configuration relax toward the local minimum of the euclidean action @xmath121 as @xmath122 . to find global minima , we use initial configurations in ( [ eq : gensol3 ] ) and solve the gradient flow equation with randomly chosen @xmath52 for each point in the @xmath116-@xmath117 space repeatedly .
we confirm that there exists a large region in the @xmath117-@xmath116 parameter space where the 3 - 2 splitting is the global minimum , see fig .
[ fig : survey ] . -@xmath117 space .
the 3 - 2 ( 2 - 1 - 2 ) splitting is the global minimum for the blue circle ( green diamond ) . both of them exist as global and local minima for the orange square ( @xmath109 , @xmath110).,width=321 ] taking all the parameters with mass dimension in @xmath123 to be the same order and identify them with @xmath3 , the 3 - 2 walls are stabilized at the separation @xmath124 where the inter - wall repulsive force is balanced against the confining force due to the excess of the @xmath40 energy , as illustrated in fig .
[ fig : wall_sol ] .
thus , @xmath22 lifts the moduli in @xmath125 ( @xmath126 ) and gives @xmath127 to @xmath68 .
the yukawa terms for @xmath128 and @xmath41 provides massless chiral fermions around the domain walls .
the fermion localization depends only on the solution @xmath41 which we can obtain by numerically solving gradient flow when @xmath129 . as shown in fig .
[ fig : wall_sol ] , difference between the numerical solution ( solid curves ) for @xmath129 and the analytic solution ( dashed curves ) given in eq .
( [ eq : gensol3 ] ) is tiny .
thus , we can make use of ( [ eq : gensol3 ] ) with @xmath54 in the following .
let us decompose the fermions as @xmath130 and @xmath131 [ eq : fdecom ] _
|5 = ( _ |3 + _ 2 ) , _ 10 = ( _ |3 & _ 3,2 + -_3,2^t & _ 1 ) .
we expand @xmath132 $ ] where effective fields @xmath133 are chiral fermions in @xmath1 dimensions defined by @xmath134 , @xmath135 ( @xmath136 ) . furthermore , four - dimensional gamma matrices @xmath137 are defined as @xmath138 and @xmath139 where @xmath140 .
the other components can be expanded similarly . plugging these into linearized field equation for @xmath141
, we find equations for mode functions @xmath142 where the subscript @xmath143 distinguishes the components in eq .
( [ eq : fdecom ] ) .
the hamiltonian for the left ( right ) component is @xmath144 ( @xmath145 ) with @xmath146 and @xmath147 .
the effective yukawa couplings are approximated by @xmath148 for @xmath149 , @xmath150 for @xmath151 , @xmath152 for @xmath153 , @xmath154 for @xmath155 , and @xmath156 for @xmath157 .
the mode functions @xmath158 for zero modes are the kernels of @xmath159 and @xmath160 .
when @xmath161 and @xmath162 , the normalizable zero modes appear only in the left - handed components as f_|3,l^(0 ) = n_|3 ^ - , and similarly for @xmath163 , @xmath164 and @xmath165 by replacing appropriately @xmath28 with @xmath29 , and @xmath86 with @xmath166 .
the mode function for @xmath167 is given by f_3,2,l^(0 ) = n_3,2 ^- , where @xmath168 and @xmath169 are normalization constants .
we identify one generation of quarks and leptons as usual @xmath170 , @xmath171 , @xmath172 , @xmath173 , and @xmath174 .
we show typical mode functions in fig .
[ fig : wall_sol ] , where we observe that quarks and leptons are localized around the walls associated to the gauge fields with which they interact .
maintaining charge universality has been difficult for localized gauge fields @xcite . in our model , it holds since the sm gauge symmetry is fully maintained . in other words ,
the overlap integrals with the sm gauge fields and quarks ( leptons ) are independent of wall positions , since mode functions of massless gauge fields are flat @xmath80 .
the massive gauge boson @xmath175 with mass @xmath176 plays the same role as @xmath177 gauge bosons in the usual @xmath1 dimensional gut , where its exchange is suppressed by the large mass @xmath3 .
when the symmetry breaking occurs by the wall splitting , however , their coupling to quarks and leptons can have significant additional suppression .
this suppression results from the small overlap of mode functions caused by two factors : the quark and lepton mode functions can be displaced from each other @xcite , and their mode functions are sharply peaked when their yukawa couplings are large .
the relevant triple couplings for the proton decay can be found in the kinetic terms of @xmath128 as _ pd & = & g h_lq1 |d^c^^(0)*_l + g h_dq _ 3 + & + & g h_lq2 _ 2 + . here , we use a canonical notation by redefining @xmath178 , @xmath179 , and @xmath180 .
the @xmath98-dependent triple couplings are given by the overlap integrals @xmath181 @xmath182 with @xmath183 , @xmath184 , and @xmath185 . ( blue circle ) and @xmath186 ( orange square ) as a function of @xmath187 .
the parameters are @xmath109 and @xmath188 and we set @xmath114.,width=321 ] the effective four - fermi coupling for the process @xmath189 is given by _
zz = h_z h_z g^2 / _ 0 ^ 2 .
the suppression factor @xmath190 compared to the standard gut is calculable in our model , since the mode function @xmath191 , shown in fig .
[ fig : wall_sol ] , is reliably obtained unlike in the previous model @xcite based on an ambitious assumption of nonperturbative confinement in @xmath4 dimensions . on top of being a function of separation of walls @xmath98 , @xmath190 also depends strongly on the magnitude of yukawa couplings @xmath28 and @xmath29 .
the width of the mode function is inversely proportional to yukawa couplings @xmath28 and @xmath29 .
hence the large yukwa coupling suppresses the overlap exponentially @xmath192 .
we plot @xmath193 and @xmath186 in fig .
[ fig : proton_decay ] , where we see that they are suppressed by more than @xmath194 and @xmath195 , respectively , with the yukawa coupling of order @xmath196 .
we note that there is no conflict with experimental data for these large yukawa couplings .
finally let us consider the baryon number violating process mediated by the off - diagonal component @xmath64 of scalar field @xmath41 through the yukawa couplings .
although it is difficult to solve the schrdinger problem for @xmath64 @xcite , it is still possible to get qualitative features .
since the zero modes of @xmath64 are absorbed by @xmath68 , the lightest mode of @xmath64 should have mass of order @xmath3 .
in addition , since the potential of the schrdinger problem should make the mode functions localized around the walls , similarly to zero mode case , the overlaps of mode functions are suppressed similarly to the vector case .
hence we expect that @xmath41-mediated proton decay enjoys a suppression factor of the same order as that of the massive vector bosons .
we propose a @xmath4 dimensional model which unifies @xmath0 gut and the brane world scenario .
our @xmath1 dimensional spacetime dynamically emerges with the symmetry breaking @xmath197 together with one generation of the sm matter fields .
we solve the gradient flow equation and confirm the 3 - 2 splitting configuration is the global minimum in a large parameter region . by applying the idea of the field - dependent gauge kinetic function @xcite to our model , we solve the long - standing difficulties of the localization of massless gauge fields and charge universality .
all the undesirable moduli are stabilized .
furthermore , the proton decay can be exponentially suppressed .
we have not yet included the sm higgs field and the second and higher generations , but our framework can easily incorporate the former similarly to ref.@xcite and the latter with the mass hierarchy in the spirit of ref.@xcite . furthermore , our model can be extended to other gut gauge group like @xmath198 .
supersymmetry and/or warped spacetime with gravity can also be included without serious difficulties .
since our model has strong resemblance to d - branes in superstring theory , we hope that our field theoretical model can give some hints for simple constructions of sm by d - branes .
f. b. was an international research fellow of the japan society for the promotion of science , and was supported by grant - in - aid for jsps fellows , grant number 26004750 .
this work is also supported in part by the ministry of education , culture , sports , science ( mext)-supported program for the strategic research foundation at private universities `` topological science '' ( grant no .
s1511006 ) , by the japan society for the promotion of science ( jsps ) grant - in - aid for scientific research ( kakenhi ) grant numbers 25400280 ( m.a . ) , 26800119 and 16h03984 ( m. e. ) , and 25400241 ( n. s. ) and by the albert einstein centre for gravitation and astrophysics financed by the czech science agency grant no . 14 - 37086 g ( f. b. ) .
s. dimopoulos , s. raby and f. wilczek , `` supersymmetry and the scale of unification , '' phys .
d * 24 * ( 1981 ) 1681 .
n. arkani - hamed , s. dimopoulos and g. r. dvali , `` the hierarchy problem and new dimensions at a millimeter , '' phys .
b * 429 * ( 1998 ) 263 [ hep - ph/9803315 ] .
a. davidson , d. p. george , a. kobakhidze , r. r. volkas and k. c. wali , `` su(5 ) grand unification on a domain - wall brane from an e(6)-invariant action , '' phys .
d * 77 * ( 2008 ) 085031 [ arxiv:0710.3432 [ hep - ph ] ] .
g. r. dvali and m. a. shifman , `` domain walls in strongly coupled theories , '' phys .
b * 396 * ( 1997 ) 64 [ erratum - ibid .
b * 407 * ( 1997 ) 452 ] [ arxiv : hep - th/9612128 ] .
g. r. dvali , g. gabadadze and m. a. shifman , `` ( quasi)localized gauge field on a brane : dissipating cosmic radiation to extra dimensions ?
, '' phys .
b * 497 * ( 2001 ) 271 [ hep - th/0010071 ] . k. ohta and n. sakai , `` non - abelian gauge field localized on walls with four - dimensional world volume , '' prog .
* 124 * ( 2010 ) 71 erratum : [ prog .
* 127 * ( 2012 ) 1133 ] [ arxiv:1004.4078 [ hep - th ] ] . | we present a field theoretical model unifying grand unified theory ( gut ) and brane world scenario . as a concrete example , we consider @xmath0 gut in 4 + 1 dimensions where our @xmath1 dimensional spacetime spontaneously arises on five domain walls .
a field - dependent gauge kinetic term is used to localize massless non - abelian gauge fields on the domain walls and to assure the charge universality of matter fields .
we find the domain walls with the symmetry breaking @xmath2 as a global minimum and all the undesirable moduli are stabilized with the mass scale of @xmath3 .
profiles of massless standard model particles are determined as a consequence of wall dynamics .
the proton decay can be exponentially suppressed .
yghp-17 - 06 |
variability in non - magnetic , cataclysmic variable ( cv ) close binaries has been largely attributed to instabilities .
outbursts in recurrent and classical novae can be caused by thermonuclear runaway on the surface of a white dwarf or by a thermal and viscous instability in the disk .
the latter outburst involves cycling from low - to - high thermal and viscous states ( see e.g. , meyer & meyer - hofmeister 1981 and references within ) , and thus cycling from low - to - high mass transfer rates ( see e.g. , lasota 2001 ) , that results in a release of accretion energy as mass is drained onto the white dwarf .
this type of outburst is similar to that experienced by cv dwarf novae ( dn ) , but larger in scale , and by transient x - ray binaries that have neutron stars or black holes as the primary star in the binary .
another type of instability , a tidal instability , results in cv dn su uma and nova - like ( nl ) outbursts , of which the larger amplitude and longer duration superoutbursts involve both a thermal - viscous instability and a tidal instability . both cv dn su uma and nl outbursts and superoutbursts show characterisitc hump - shaped modulations , known as positive superhumps , in their light curves .
positive superhumps have periods that are a few percent longer than the orbital period .
in addition to positive superhumps , some non - magnetic cv dn systems ( e.g. , tt ari ) show negative superhumps in their light curves .
these negative superhumps have a period that is a few percent shorter than the orbital period .
negative superhumps are suggested to be a consequence of a partially tilted disk ( patterson et al .
1993 ) or are related to a warped accretion disk ( petterson 1977 ; murray & armitage 1998 ; terquem & papaloizou 2000 ; murray et al .
2002 ; foulkes , haswell , & murray 2006 ) .
our previous numerical simulations ( wood , montgomery , & simpson 2000 ; montgomery 2004 , montgomery 2009 ) show that negative superhumps can be produced by a disk fully tilted out of the orbital plane . in montgomery ( 2009 )
, we suggest that the source to the negative superhump is extra light from inner annuli of a tilted accretion disk , and this extra light waxes and wanes with the amount of gas stream overflow and particle migration to inner annuli as the secondary orbits .
accretion disks have been postulated to warp or tilt via a potpourri of sources .
for example , disk tilt in x - ray binaries can be from gas streaming at an upward angle from the inner lagrange point for one half of the orbit and at a downward angle for the second half of the orbit ( boynton et al . 1980 ) .
in some cvs , a disk tilt can be held constant by a gas stream that is fed via the magnetic field of the secondary ( barrett , odonoghue , & warner 1988 ) . also for cv systems , a disk tilt instability can result from a coupling of an eccentric instability to lindblad resonances ( lubow 1992 ) .
a vertical resonant oscillation of the disk midplane can be caused by tidal interactions between a massive secondary and a coplanar primary ( lubow & pringle 1993 ) . a warping instability can be caused by irradiation from the primary ( pringle 1996 , 1997 ) .
a warping can be caused by direct tidal forces from a secondary orbiting on an inclined orbit ( e.g. , papaloizou & terquem 1995 , larwood et al .
1996 , larwood 1997 , larwood & papaloizou 1997 ) . a disk tilt can be induced if the secondary has a body axis that is tilted relative to the orbital plane ( roberts 1974 ) .
a disk warp can be caused by misalignments of the spin axis of a compact and/or magnetized primary and the disk axis ( see e.g. , kumar 1986 , 1989 ) . to date , no consensus has been found regarding the source to a disk tilt or warp .
we do acknowledge that murray & armitage ( 1998 ) find that cv dn accretion disks do not tilt significantly out of the orbital plane by instabilities .
regardless of how an accretion disk tilts or warps , non - coplaner accretion disks should precess in the retrograde direction , that is , the direction opposite to spin and orbital motion .
precession is the motion of a body axis around a fixed axis in space .
retrograde precessional motion traces out a cone around the cone s symmetric axis , a fixed axis in space .
two sources are known to generate retrograde precession .
one is torques acting on the body .
a second kind is fast precession which applies to bodies that are free .
an example of the latter is a coin thrown into the air in such a way that the coin spins fast , but the normal to the coin wobbles .
of the two , many typically cite torques as the source to retrograde precession .
for example , retrograde precessional activity has been suggested for x - ray binaries ( e.g. , katz 1973 , larwood 1998 , wijers & pringle 1999 ) , for common systems that produce jets ( e.g. , livio 1999 ) , for protostars ( e.g. , papaloizou & terquem 1995 , larwood et . al
. 1996 , larwood 1997 ) , for quasars and supermassive black holes ( e.g. , romero et al .
2000 , caproni & abraham 2002 ) , for cvs ( barrett et al . 1988 , patterson et al .
1993 , harvey et al .
1995 ) , to name a few .
katz ( 1973 ) suggests that the permanently tilted edge of a disk precesses due to the tidal field of the secondary .
roberts ( 1974 ) suggests that the secondary precesses due to the tidal field of the primary and the tilted disk retrogradely precesses as a consequence .
barrett et al .
( 1988 ) and kondo et al .
( 1983 ) suggest that particles may be orbiting a tilted disk and as a result , the disk retrograde precesses .
patterson et al . ( 1993 ) notice a near 2:1 ratio of the negative - to - positive precessional periods of several cv systems that exhibit both modulations in their light curves , and they note that this ratio is curiously similar to that of the earth which has a retrograde nodding precessional period @xmath0=18.61 yr and a prograde precessional period @xmath1=8.85 yr ( see e.g. , stacy 1977 ) .
patterson et al . ( 1993 ) suggest that the source to retrograde precession in these cv dn systems may be the same as that which causes the earth to retrogradely precess .
petterson ( 1977 ) suggests that the earth s retrograde precession is caused by outside forces , namely tidal torques on the earth s bulge by the moon and sun with the moon having the most effect due to its closer proximity to earth .
hence , patterson et al . (
1993 ) suggests that a tidal torque by the secondary on a tilted accretion disk could be the source to retrograde precession in cv dn non - magnetic , accreting disks .
others have found that only portions of their disk retrogradely precesses . for semidetached binaries , bisikalo et al .
( 2004 ) show that a one - armed spiral shock wave can be generated in inner annuli of cooler disks ( i.e. , @xmath2 ) .
they find that their inner annuli spiral wave precesses retrogradely , but the disk does not .
based on their numerical simulations , they show that the wave is generated by sheared elliptical orbits .
they suggest that the sheared elliptical orbits are due to retrograde precession within the disk , and the retrograde precession is due to an influence by the mass - losing star by the processes discussed in kumar ( 1986 ) and warner ( 1995 , 2003 )
. however , this suggestion can not be correct because kumar ( 1986 ) finds that retrograde precession is due to the tidal influence of the secondary on a tilted disk s outer annulus , not inner annulus , and because the simulations by bisikalo et al .
( 2004 ) do not have warped or tilted disks ( private communication ) . in this work , we test the idea of patterson et al .
( 1993 ) on the tilted accretion .
we assume the disk is fully tilted out of the orbital plane because our previous works ( wood , montgomery , & simpson 2000 ; montgomery 2004 , montgomery 2009 ) are successful in showing that fully tilted disks retrogradely precess . in addition , our works also generate negative superhump modulations in our artificial light curves that have the correct shape and negative superhump periods of the right percentage as compared with observations . in this work
we do not explain how accretion disks tilt but instead check to see if the source to retrograde precession is like that which causes the spinning , tilted earth to retrogradely precess . in 2 of this paper ,
we develop a generic theoretical expression to describe retrograde precession using the earth - moon - sun system as a model . in 3 , we reduce the generic theoretical expression by considering the geometry of the system and by making appropriate assumptions , thereby obtaining expressions generated or used by others to describe retrograde precession in protoplanetary , protostellar , x - ray , black hole , and non - magnetic accreting cv binary systems . in 4 ,
we summarize our numerical simulations . in 5 , we list observational data .
is 6 , we compare retrograde precessional values obtained from our theoretical expressions with those obtained from our numerical simulations and with those obtained from observational data . in 7 , we provide a discussion and in 8 , we summarize , conclude , and list future work .
we consider two systems , one being the earth - moon - sun system and one being a non - magnetic , accreting cv dn system where the accretion disk is misaligned with the orbital plane . in the earth - moon - sun system ,
the spinning earth is known to have an oblate spheroidal shape .
the resultant equatorial bulge is not symmetric about earth s rotation axis . as such , a net torque on this equatorial bulge results in the rotational axis of the earth precessing in the opposite sense to the earth s spin and earth s orbital directions . to simplify the many - body problem of the earth - moon - sun system and to generate like coordinates among the two systems , we only consider net torque effects by the sun on the earth .
figure 1 , top panel , exaggerates the earth and sun size and distance scales to show details . in this panel ,
the sun is the small sphere on the @xmath3 axis located at a distance @xmath4 from the origin .
the earth is the large oblate spheroid that is centered at the origin in the ( @xmath5 ) coordinate system of the ecliptic .
the @xmath6 axis defines the normal to the ecliptic and orbital motion @xmath7 is prograde .
the ( @xmath8 ) coordinate system ( @xmath9 not shown ) is the equatorial system of the earth .
earth rotates around the @xmath10 axis with angular velocity @xmath11 .
as shown , the angle from the @xmath6 to the @xmath10 axes is the obliquity @xmath12 .
note that @xmath13 and @xmath14 share the same axes and thus they coincide with the line of equinoxes at this instant in time in the earth - sun system ( or with the line of nodes in the non - magnetic , accreting cv dn system ) .
also note that @xmath10 , @xmath6 , and @xmath3 are in the same plane and thus eulerean rotation is about the @xmath15 axis .
because the gravitational attraction forces differ from a point on earth nearer to the sun , denoted by the force vector @xmath16 , than a point on earth further from the sun , denoted by the force vector @xmath17 , torques around the origin result .
figure 1 , bottom panel , represents the cv dn system and is similar to figure 1a except that the oblate spheroidal earth is replaced with a massive primary star @xmath18 that has angular velocity @xmath19 and that is surrounded by a disk of mass @xmath20 where @xmath21 .
also , the sun is replaced with a secondary star of mass @xmath22 . in this non - magnetic cv dn system ,
the massive primary can not be seen as it is smaller in size compared to the disk size .
we assume that the separation @xmath4 of the system is much larger than the thickness of the disk , and we assume that the disk has negligible mass compared to that of the primary . like the top panel ,
this bottom panel is not to scale .
ultimately we seek the retrograde precession of the line of nodes in the spinning , tilted cv dn system .
to find this precession , we use the precession of the equinoxes in the earth - sun system as a model . for continuous bodies ,
its angular momentum vector @xmath23 is related to its angular velocity vector @xmath24 in any coordinate system through a linear transformation matrix known as the moment of inertia tensor @xmath25 , @xmath26 using a right - hand coordinate system whose general coordinates are ( x,y,z ) , we can identify the diagonal elements of the matrix as the moment of inertia coefficients and the off - diagonal elements as the products of inertia .
if we assume , for example , that @xmath27 is about the @xmath14 axis and therefore @xmath23 is along the @xmath14 axis , then @xmath28 if the density @xmath29 of a solid is known , then the moment of inertia can be found from @xmath30 where @xmath31 is the perpendicular radial distance from any rotational axis , @xmath32 is small volume element , and density is a function of radius . in cartesian coordinates , the moment of inertia tensor is @xmath33 dv'\ ] ] where dv=dxdydz. if cylindrical coordinates are desired , then we can integrate from @xmath34 to @xmath35 along the @xmath36 axis where @xmath37 is the height of the cylinder , from @xmath38 to @xmath39 along the @xmath40 axis where @xmath39 is the radius of the cylinder , and from @xmath41 to @xmath42 along the @xmath43 axis .
however , the solution to this integral is most easily found using curvilinear coordinates @xmath44 , @xmath45 , @xmath46 with scale factors @xmath47 , @xmath48 , and @xmath49 for a circular cylinder . in this solution , @xmath50 , @xmath51 , @xmath52 , and @xmath53 .
the limits of integration are @xmath54 on @xmath55 , @xmath56 on @xmath57 , and @xmath58 on @xmath59 . in the earth - sun system
, the potential we seek is that of the earth and thus we center this attracting body at the origin of our coordinate system . as shown in figure 2 ,
the attracting body has mass @xmath60 ; an elemental mass @xmath61 is located at a distance @xmath62 from the center of mass of the attracting body and at an angle @xmath63 to the line of centers connecting the center of mass of the attracting body to the center of mass of a far - away point mass object @xmath64 .
we assume in the earth - sun system that the sun is located at a distance @xmath4 where @xmath65 .
the potential @xmath66 at the far away point mass object is @xmath67 where we have assumed that earth has uniform density @xmath29 and rotation is possible about the line of centers . in this equation , @xmath68 is the universal gravitational constant .
this potential is that of maccullagh , a result that is applicable for slightly non - spherical celestial bodies and for @xmath65 .
this potential is valid for the oblate spheroidal earth in the earth - moon - sun system .
we assume that this potential is also valid for the spinning , tilted cv dn system even though the primary is surrounded by a disk ( and hence the geometry is by far not spherical ) and the radius of the white dwarf primary plus accretion disk is not significantly less than the distance to the point mass secondary star . as the first term in the potential dominates over the other terms ( of which only the second term is shown as all others are assumed negligible ) and as @xmath21 , the potential is most affected by @xmath64 and @xmath69 . as @xmath18 is assumed spherical and of uniform density and as the secondary is significantly further away than compared with the radius of the primary , we accept this potential as valid for the spinning , tilted cv dn system .
we note that the second term in equation ( 4 ) is not negligible as the disk s geometric shape affects the potential via the moment of inertia .
we also note that as the radius of the disk is dependent on the mass ratio of the system , with smaller mass ratios yielding larger disk radii , accretion disks in small mass ratio systems may be subject to additional perturbations .
these additional correction terms are not considered in this work . in figure 2 ,
the earth in the earth - sun system and the primary in the cv dn system are the attracting mass bodies that are centered at the origin .
the sun and the secondary star are considered to be the far away point mass objects . as the potential is evaluated at the far away point mass object ,
the force , which is proportional to the gradient of the potential , is on the far away point mass object . by
newton s third law , the far away point mass object exerts an equal and opposite force on the attracting body .
as the forces are not co - linear , a torque about the origin arises .
figure 1 is exaggerated to show non - colinear forces . because the spinning earth has rotational flattening and because the earth s spin axis is tilted , the force pulling on the near - side of earth ( i.e. , the side closer to the sun ) is greater than the force pulling on the far - side of the earth , and a net non - zero torque results about the @xmath13 axis ( see figure 1 , top panel ) . in the spinning ,
tilted cv dn system , a torque about the origin arises for the same reasons - because the disk is tilted and spinning and because this combination results in non - colinear near and far side forces on the disk ( see figure 1 , bottom panel ) . as shown in figure 1 ,
the body is rotating counter - clockwise about the @xmath10 body axis , and the body moment of inertia coefficients are along the body axes ( @xmath70 ) . by definition ,
the net torque around the origin of the body is @xmath71 x @xmath72 where any force @xmath73@xmath66 ) .
as @xmath74 x @xmath74 = 0 , then the @xmath75 in equation ( 4 ) is the only term that is not just a function of radius but also a function of ( x,y,z ) .
therefore , we reduce equation ( 4 ) to @xmath76 where we have evaluated the potential at a distance @xmath77 from the origin .
the net torque in the ( @xmath8 ) coordinate system is @xmath78 by employing transformation equations ( see e.g. , goldstein 1980 ) , we can rotate the net torque from the ( @xmath8 ) coordinate system to that of the ( @xmath5 ) coordinate system ( see figure 1 ) . in this transformation , we assume that the obliquity angle @xmath12 remains constant , the orbit is circular , and the mean orbital motion is @xmath79 as shown in figure 1 .
the resultant net torque in the ( @xmath5 ) coordinate system becomes @xmath80 with component magnitudes @xmath81 \\ \gamma_{y } & = & \frac{3gm}{d^{3 } } ( i_{x'x'}-i_{z'z'})[\cos(\dot{\omega } t + \psi)\sin(\dot{\omega } t + \psi ) \sin\theta ] \\ \gamma_{z } & = & \frac{-3gm}{d^{3 } } ( i_{y'y'}-i_{x'x'})[\cos(\dot{\omega } t + \psi)\sin(\dot{\omega } t + \psi ) \cos\theta ] .\end{aligned}\ ] ] where @xmath82 is some phase angle , and @xmath83 is time , and @xmath84 is an angle that increases in the orbital direction from perihelion . in these equations , we have neglected any variations over one orbit because the precessions we seek are over long periods of time , much longer than one orbital period . to simplify the equations further , we assume symmetry about the @xmath10 rotation axis , and we assume a circular disk .
therefore , @xmath85 and @xmath86 is zero .
we can not simplify @xmath87 any further but we we can substitute the identity @xmath88 into @xmath89 . because the magnitude of this torque is not constant , we need to take averages . for example , the magnitude is a minimum at either of the equinoxes and a maximum at either of the solstices in the earth - sun system .
likewise , the magnitude is a minimum when the primary , secondary , and the disk s line of nodes align and a maximum when the primary , secondary , and disk s line of antinodes align .
therefore , we can find the average of @xmath89 and @xmath87 over a time period of one orbit using an integral average of some continuous function f(t ) over an interval @xmath90 , @xmath91 . if we assume in @xmath89 that only the @xmath92 is not constant and we assume the phase constant is zero
, then we find that @xmath93 .
for the same time interval , we find that @xmath94 . of equations ( 9)-(11 ) ,
only @xmath89 is left , @xmath95 we can use unit vector notation and employ the cross product definition , @xmath96 to obtain the component of the net torque along @xmath97 , @xmath98.\ ] ] for the earth - sun system , @xmath99 and is not in solar masses .
for the earth - sun system , the component of the average torque along the line of equinoxes is @xmath100.\ ] ] for the non - magnetic cv dn system , @xmath101 and is not in solar masses . for the misaligned cv dn system ,
the component of average torque along the line of nodes is @xmath102.\ ] ] note that the direction is negative as indicated by @xmath10 @xmath103 @xmath6 .
a rotating body responds to a torque that has been applied over a small period of time by gaining a small amount of angular momentum , @xmath104 .
the direction of the small gain in angular momentum is parallel to the the applied torque and perpendicular to the spin axis @xmath10 of the body .
the rotating body responds over this same small time interval by moving its spin angular momentum vector to a new position . by repeating this process many times over a significantly longer period time
, the spin axis of the rotating body @xmath10 traces out a surface of a cone whose axis coincides with the normal to @xmath6 .
the direction of the trace is retrograde to the direction of the spin of the body and orbit as shown in figure 1 .
the angular speed of the trace is the magnitude of the precessional angular velocity .
thus , to find the retrograde precession we need to find the time rate of change of the angular momentum and equate that to the torque found in equation ( 14 ) . to establish the total angular momentum ,
we need to find the total angular velocity . by definition , the total angular velocity of a body rotating about its axis @xmath27 is the sum of the spin rate @xmath105 , which is along the @xmath106 axis , and the motion of its spin axis @xmath10 , which is along the @xmath107 axis , @xmath108 in this equation , we assume that no net torque is applied and @xmath85 . from euler s equations of motion
, we find @xmath109 upon substitution into equation ( 1 ) , the total angular momentum becomes @xmath110 of which the time rate of change is @xmath111 in obtaining this last equation , we assume a constant spin rate and constant moment of inertia coefficients .
the time rate of change of the angular momentum is just the torque we found in the previous subsection .
even though the magnitude of this torque is not constant over one orbit and therefore the motion of the angular momentum vector moving to its new position pulses with this driving torque , we can assume that over long periods of time this net retrograde precessional angular velocity @xmath112 is steady .
other than what is mentioned here , we do not consider nodding motions any further in this work .
after setting the time rate of change of the angular momentum equal to the torque found in the previous subsection and after simplifying , we obtain the quadratic @xmath113 \dot{\gamma } + \frac{3gm}{2d^{3}}\frac{(i_{z'z'}-i_{x'x'})}{i_{x'x ' } } = 0\ ] ] whose roots are real and can be summed , the smaller root being @xmath114 like for equations ( 15 ) and ( 16 ) , we can substitute @xmath99 in equation ( 22 ) for the sun that is located at a distance @xmath4 from the origin and @xmath115 for the moon that is located at a distance @xmath116 from the origin @xmath117 this equation is the retrograde precession of the zero - point of right ascension ( a.k.a .
, the first point of aries ) in the earth - moon - sun system ( see any fundamentals of geophysics or celestial mechanics book such as danby 1962 and stacey 1977 ) .
likewise , we could substitute @xmath101 for a secondary that is located at a distance @xmath4 from the origin @xmath118 this equation is the retrograde precession of a point on the rim of the spinning , mis - aligned disk .
this point can be considered equivalent to the zero - point of right ascension ( a.k.a .
, the first point of aries ) in the earth - moon - sun system .
because the disk is spinning and because it is tilted , forces that arise from the gradient of the potential are not colinear , and these non - colinear forces cause a net torque on the disk , one component of which is along the line of nodes .
this component of net torque causes the spin axis of the disk to precess in a direction that is retrograde to the orbital direction of the secondary .
the net motion traces out a cone whose axis coincides with the normal to the orbital plane .
we note that this retrograde precessional equation for a spinning tilted accretion disk in a non - magnetic system is generic in that we have yet to take into account the moment of inertia coefficients , a differentially rotating disk , and any effects due to the gas stream on the accretion disk .
retrograde precession expressions that have been found by others for various systems can also be found by the technique developed in 2 , and we show this in this section . insight is gained by comparing derivation results . in all examples , we assume equation ( 24 ) is valid . in addition , we assume the disk is of constant density , orbits are circular , and equation ( 3 ) is valid for finding the moment of inertia .
we solve equation ( 3 ) in cylindrical coordinates to find the moment of inertia of a rigid , continuous disk that has constant density .
we also assume the radius of the disk is much larger than the height of the disk , @xmath119 . for a thin circular disk , @xmath120.\ ] ]
we assign @xmath121 , @xmath122 , and @xmath123 to the diagonal elements in the matrix , noting that @xmath121=@xmath122 and thus @xmath124 .
substituting these into equations ( 24 ) , the retrograde precession becomes @xmath125 equation ( 26 ) is the same result as that obtained by katz et al .
( 1982 ) who approximated a disk as a thin ring to find the nodding motion of accreting rings and disks in x - ray binaries . in their work
, they consider the ring / disk to be tilted and acted upon by a secondary mass . the keplerian orbital angular velocity , @xmath126 is easily identified in equation ( 26 ) .
we assume the angular velocity of the disk around the @xmath10 axis is @xmath127 .
thus , the retrograde precession of a thin , circular , continuous , constant density , rigid , tilted disk is @xmath128 at the edge of the continuous rigid disk , the centripetal force equals the gravitational force .
therefore , we can also represent the angular velocity of the disk at the disk edge by @xmath129 we can substitute this last equation and equation ( 27 ) into equation ( 28 ) to obtain @xmath130 where mass ratio @xmath131 .
if we assume @xmath132 in the polytropic equation of state used by larwood ( 1997 ) , then their expression for retrograde precession for protoplanetary accretion disks reduces to our equation ( 30 ) . likewise , if we assume @xmath133 in the polytropic equation of state used by romero et al .
( 2000 ) , then the expression they assume for retrograde precession in quasars and black hole accretion disks reduces to our equation .
this equation is proportional to the result obtained by kondo , wolff , & van flandern ( 1983 ) , and subsequently by barrett , odonoghue , & warner ( 1988 ) , who establish the precession rate of an orbiting particle in a tilted disk .
their proportionality constant is 1/2 , however . substituting newton s version of kepler s third law for the binary orbit into equation ( 30 )
, we obtain yet another version of retrograde precession @xmath134 if we assume a small angle approximation , then equation ( 31 ) reduces to a later derivation by warner ( 1995 , 2003 ) that is based on a retrograde precession of a tilted disk s outer annulus ( kumar 1986 , 1989 ) .
the absolute magnitude of equation ( 31 ) is the same result as that obtained by osaki ( 1985 ) for prograde precession in non - tilted or non - warped disks , a result that is too high for prograde precession as pressure effects within the disk have been neglected ( lubow 1992 ) .
the expressions for retrograde precession developed by others , shown above , are the equivalent to our derivations if we assume that the disk rigidly rotates at the same rate as the primary , i.e. , @xmath135 . in other words ,
differential rotation has been neglected . in obtaining @xmath75 in equation ( 25 ) ,
we assume the disk to be circular , thin , of constant density , rigid , and continuous .
however , papaloizou & terquem ( 1995 ) note that accretion disks are not solid objects , and thus angular frequency is not constant with radius in fluid disks .
that is , disks are fluid and thus individual rings can rotate differentially and this should be considered in evaluating equation ( 25 ) . from equation ( 25 ) , we note that @xmath136 .
we assume this same equation holds in a differentially rotating disk .
equation ( 24 ) simplifies to @xmath137 angular frequency is not expected to be constant with disk radius and thus equation ( 29 ) is revised to its most general form , @xmath138 where @xmath139 is the radial distance from the origin to any particle within the disk . as we still assume the disk is circular , we still expect @xmath121=@xmath122 even though the disk is differentially rotating .
we do expect that the differential angular velocity to affect the @xmath123 coefficient . with these assumptions and allowing @xmath140 , the retrograde precession of a thin , circular , constant density , differentially rotating , tilted disk is @xmath141 where @xmath142 note that because the units of @xmath123 change when we include the @xmath143 term inside the integral , the integral is redefined to @xmath144 .
the solution to this integral is , assuming a constant density disk , @xmath145 .
similar calculations of @xmath146 and @xmath147 show that our assumption of @xmath148 is correct in equation ( 32 ) . as @xmath149 for a cylinder , then @xmath150 . substituting this result into equation ( 34 ) and substituting in our prevously found solution for @xmath151 , the retrograde precession of a thin , circular , constant density , differentially rotating , tilted accretion disk is @xmath152 to obtain equation ( 37 ) ,
we have substituted the angular velocity at the disk edge , equation ( 29 ) , into equation ( 36 ) . to obtain equation
( 38 ) , we have substituted in newton s version of kepler s third law . as we noted earlier , this derivation assumes constant surface density and constant disk radius @xmath39 .
equation ( 37 ) is the retrograde precession found by larwood et al .
( 1996 ) for pre - main sequence binary systems where they assume a polytropic gamma of 5/3 ( i.e. , isentropic ) in their equation of state .
if viscous dissipation occurs , then an efficient cooling mechanism is assumed if fluids are to remain isentropic throughout .
equation ( 37 ) is the same result as that found by terquem , papaloizou , & nelson ( 1999 ) for protostellar disks that have tidally induced warps .
equation ( 37 ) is proportional to that found by larwood ( 1998 ) for forced precession in x - ray binaries .
larwood ( 1998 ) found a 3/7 proportionality to our 15/32 as they averaged the tidal torque over the disk .
equation ( 38 ) is from montgomery ( 2004 ) .
we note that equation ( 35 ) is not entirely correct . in this equation , we integrate from the origin to the outer radius of the disk .
the inner rim of the disk does not extend to the origin .
however , the outer radius of the disk is so large compared to the radial extent of the white dwarf or to the radial extent to the inner rim of the disk that the effect on precession is minimal .
hence , equation ( 35 ) is valid for our study of precession in cv dn systems . in the derivation of equations ( 36)-(38 ) , we assume that the disk differentially rotates . implied in this derivation is that the disk precesses as a unit .
support for this assumption is given by papaloizou & terquem ( 1995 ) who analytically show , and larwood et al .
( 1996 ) numerically verify , that rigid body precession of a keplerian disk is possible so long as the sound crossing time scale in the disk is small compared to the precession time scale of the disk .
further , numerical simulations by murray ( 1998 ) and observations by patterson et al .
( 1998 ) do not indicate that disks differentially precess . in 2.4
, we assume that the major components in accreting cv dn systems are the primary , secondary , and the accretion disk .
we assume that the orbit is circular , symmetry is about the @xmath10 rotation axis , and the disk is circular .
a circular disk is mirror symmetric and rotationally symmetric around its @xmath36 axis and , as a result , we could find equal body moment of inertia coefficients @xmath153 and a constant rotation around the normal axis to the plane of the disk , equation ( 24 ) . however , accretion disks in cv dn systems have long been thought to be non - circular ( see e.g. , warner 1995 , 2003 ) .
an elliptical disk does not have equal radii .
thus , the moment of inertia coefficients about the principle axes are not equal , @xmath121 @xmath154 @xmath122 , and @xmath155 . yet , over a time period of one orbit , @xmath156 still dominates .
equations ( 31 ) and ( 38 ) are still valid in non - circular disks so long as the disk precesses as a unit , the distance @xmath4 is much greater than the radius of the primary as discussed in 2.2 , and precession time is much longer than the orbital time . in the derivation of equations ( 36)-(38 ) , we assume that the disk differentially rotates yet the disk precesses as a unit .
we have justified the geometry of the disk in neglecting a small , centrally located inner hole where the primary is located and in neglecting the outer shape of the disk being more elliptical than circular in cv dn non - magnetic systems .
we have taken into account major components of the system - the primary , secondary , and the accretion disk .
however , we have neglected the effect of the gas stream on the accretion disk .
therefore , we introduce additional geometrical shapes that may be used to help describe the overall geometrical shape of the disk as the overall geometrical disk shape may have changed due to the gas stream striking the disk . in this subsection , we consider the geometrical shape of a ring that is centered around the primary and that is tilted at the same angle as the disk .
the inner rim radius of the ring @xmath157 is located near the innermost rim of the disk and the outer rim radius of the ring is located at @xmath158 . in this derivation
, we also assume rotation and precession as a unit . in modeling black holes , bardeen & petterson ( 1975 ) state that the combined effects due to precession and viscosity should be to align the rotation of the inner disk annuli with the spin axis of the hole .
kumar and pringle ( 1985 ) show that the alignment of the disk and the hole does occur in slightly tilted , steady state disks .
borrowing these ideas , we return to equation ( 32 ) and assume that @xmath159 @xmath160 where @xmath161 is the angular velocity at @xmath157 . note that we assume that the inner edge of the ring rotates at the same rate as the primary .
using kepler s third law , we can relate the radius and angular velocity of the inner ring edge to that of the outer ring edge @xmath162 to obtain @xmath163 substituting in kepler s third law for the orbital period of the binary orbit , equation ( 41 ) , and the moment of inertia coefficients from equation ( 25 ) , equation ( 42 ) becomes @xmath164 or equivalently @xmath165 or equivalently @xmath166 notice the similarities between equations ( 30 ) and ( 44 ) and between equations ( 31 ) and ( 45 ) . to reduce equation ( 45 )
further , we need to know the radius to the inner rim of the circular ring .
the inner rim radius of the ring is unlikely to be the radius of the white dwarf as the inner rim of the ring is unlikely to extend down to the surface of the primary .
also , the inner rim radius of the ring is not likely to be the distance of closest approach of a free - falling gas stream to the primary ( lubow & shu , 1975 ) @xmath167 which is good for @xmath168 .
although this distance of closest approach yields a radius that is larger than the radius of the white dwarf , this radius is not large enough to explain a circularized ring : the initial path of the free - falling gas stream forms a parabolic shape around the primary , and after interaction with the primary s roche lobe , the path then forms a rosette pattern . upon collisions of the gas stream with itself ,
the rosette pattern of the gas stream settles into a circularized ring that has a radius that is larger than the distance of closest approach of the initial free - falling gas stream . yet the inner rim radius of the ring can not be the circularization radius ( hessman & hopp , 1990 ) @xmath169 which is good for @xmath170 , as this equation describes the @xmath171 radius of the ring s @xmath172 radius . as the ring is thin
, we shall assume that @xmath157 is half way between the minimum radius and the circularization radius @xmath173 as warner ( 2003 ) finds that @xmath174 , @xmath175 and equation ( 45 ) becomes @xmath176 in summary , we assume a thin ring that is centered about the origin and is located near the inner edge of the disk .
this ring retrogradely precesses as a unit .
we note that we could have repeated the process of differential rotation that we introduced in 3.2 .
however , because the ring is thin , we find that equation ( 49 ) is minimally affected .
as the orbital period is slightly longer than the period of the secondary encountering the same disk node per orbit , we can relate the negative superhump period @xmath177 and the orbital period to the retrograde precession period by @xmath178 another relation is the negative superhump period excess @xmath179^{-1 } \end{aligned}\ ] ] where we substitute equation ( 50 ) into equation ( 51 ) to obtain equation ( 52 ) .
for example , we can find the negative superhump period excess for the rigid , continuous , circular , tilted disk from 3.1 .
if we assume that the radius of the disk is that given by paczynski ( 1977 ) @xmath180 which is valid for mass ratios in the range @xmath181 , then we can combine equations ( 31 ) and ( 52 ) to obtain @xmath182^{-1}.\end{aligned}\ ] ] likewise , we can find the negative superhump period excess for the differentially rotating , circular , tilted disk from 3.2 by substituting equation ( 53 ) into equation ( 38 ) and then into equation ( 52 ) to obtain @xmath183^{-1}.\end{aligned}\ ] ] likewise , we can find the negative superhump period excess for the rotating , circular , tilted ring from 3.3 by substituting equation ( 49 ) into equation ( 53 ) and then into equation ( 52 ) to obtain @xmath184^{-1}.\ ] ] note that we have dropped the negative sign that redundantly indicates retrograde precession .
equation ( 55 ) appears in montgomery ( 2004 ) . if we assume a primary mass @xmath18=0.8@xmath185 and mass ratio @xmath186 , then the secondary mass @xmath22=0.32@xmath185 .
the radius of the disk is @xmath187 , where @xmath188 ( montgomery 2009 ) . the rotation period at the disk edge , determined from equation ( 29 ) , is around one hundred seconds .
this rate is fast compared with the orbital period @xmath189 hours found using newton s version of kepler s third law . from equation ( 55 ) , we find @xmath19080 hours , assuming small angle approximation for disk tilt . instead of mass ratio , we can express equations ( 54 ) , ( 55 ) , and ( 56 ) in terms of the orbital period .
for this substitution , we shall assume non - magnetic , accreting binaries that follow the smith & dhillon ( 1998 ) secondary mass - period relation , @xmath191 where @xmath192 is in hours .
we do not show these lengthy negative superhump period excess equations in this work , but they can easily be found by variable substitution where @xmath193 and @xmath194 .
in montgomery ( 2009 ) , we simulate non - magnetic cv dn systems that have mass ratios within the range @xmath195 and low mass transfer rates . by limiting our mass ratio range to longer orbital periods ,
we eliminate systems that apsidally precess and thus show positive superhumps in their light curves , thereby minimizing variables .
we adopt a smith & dhillon ( 1998 ) secondary mass - period relation which is for main - sequence dwarfs as the secondaries . because we simulate accreting , non - magnetic cv dn systems in or above the period gap , we choose an average white dwarf primary mass @xmath194 , the average value found by smith & dhillon ( 1998 ) for this mass ratio range .
we refer the reader to montgomery ( 2009 ) for details regarding the code yet we provide a summary here . for our numerical simulations , we use smoothed particle hydrodynamics ( sph ) to simulate accretion disks that have constant and uniform mass particles . in the code , we assume assume a constant and uniform smoothing length @xmath196 and an ideal gamma - law equation of state @xmath197 where @xmath198 is pressure , @xmath199 is the adiabatic index , @xmath29 is density , and @xmath200 is specific internal energy .
the sound speed is @xmath201 .
the code utilizes a lattanzio et al .
( 1986 ) artificial viscosity , and we choose @xmath202 viscosity coefficients .
our viscosity is approximately equivalent to a shakura & sunyaev ( 1973 ) viscosity parametrization ( @xmath203 where @xmath204 is the disk scaleheight ) @xmath63=0.05 . as smak ( 1999 )
estimates @xmath205 0.1 - 0.2 for dn systems in high viscosity states , our simulations are more for quiescent systems . as only approaching particles
feel the viscous force and since neither radiative transfer nor magnetic fields are included in this code , all energy dissipated by the artificial viscosity is transferred into changing the internal energies . as radiative cooling
is not included in this code , an adiabatic index @xmath199=1.01 is incorporated to prevent internal energies from becoming too large . by integrating the changes in the internal energies of all the particles over a specific time interval @xmath206 , variations in the bolometric luminosity yield an approximate and artificially generated light curve @xmath207 in all simulations , we build a disk to 100,000 particles . at orbit 200 ,
we artificially rotate the disk out of the orbital plane 5@xmath208 to induce negative superhumps in the light curve and to force the disk to precess in the retrograde direction .
for example , if @xmath186 then @xmath22=0.32 @xmath185 .
our simulation unit length scales to @xmath209 if we assume the secondary mass - radius relationship @xmath210 ( warner 2003 ) that applies for @xmath211 and the eggleton ( 1983 ) volume radius of the roche lobe secondary @xmath212 in this equation , @xmath213 and @xmath22 are in solar radii and solar mass , respectively .
the eggleton ( 1983 ) relation is good for all mass ratios , accurate to better than 1% .
taking the radius of the primary to be 6.9x10@xmath214 cm , the scaled radius of the white dwarf is @xmath215 .
after building the disk , we allow the disk to evolve in the short term to a quasi - equilibrium state where particles are injected at the rate they are removed from the system by either being accreted onto the primary or the secondary or lost from the system .
the average net rate of accretion for a quasi - static disk is around 500 particles per orbit as shown in montgomery ( 2009 ) .
as this net rate is approximately half that injected to build the disk , the steady state mass transfer rate reduces to @xmath216 1 x @xmath217/yr or a rate similar to an su uma in quiescence .
we estimate the density of the gas near @xmath218 to be @xmath219 g @xmath220 using @xmath221 . as we maintain 100,000 particles in the steady state disk , then the total mass of the disk is @xmath222= 100,000 x @xmath223 = 3.5x10@xmath224 g or @xmath225 2x10@xmath226 , a negligible value compared to the mass of either star .
if we assume the outer radius of the disk is @xmath227 , or twice the circularization radius ( warner 2003 ) @xmath228 where @xmath229 , then for @xmath186 we find @xmath230 and @xmath231 . as @xmath232 , self gravity is neglected in these simulations . in montgomery ( 2009 ) , we find the calculated negative superhump period , its propagated error , and the nodal superhump period excess as output to the code .
we list these values in table 1 .
the units for period are hours .
we also find the source that powers the negative superhump .
our results suggest that negative superhumps are due to an increase in the number of particles that reach innermost annuli of a tilted disk ( montgomery , 2009 ) .
as inner annuli have higher energies , the combination of increased number of particles and higher energy inner annuli yields more light being emitted .
this additional light waxes and wanes with the amount of gas stream overflow and the amount of particle migration into innermost annuli . as the secondary orbits the tilted disk
, the gas stream mostly oveflows one face of the tilted disk for approximately the first half of an orbit and then over the other face of the tilted disk for approximately the second half of an orbit . as a result of the disk tilt and thus gas stream overflow and additional particle migration to innermost annuli , a yellow colored ring that is denser that its surroundings is generated near the innermost annuli ( see figure 3 ) .
this denser ring is important to the generation of the negative superhump .
we do note that the relative surface densities across the face of the disk vary by only @xmath233 from annuli to annuli , regardless if the disk is tilted or not .
exceptions are the disk rim and a denser ring , colored green , located near @xmath234d in figure 3 .
the latter is caused by gas stream overflowing the tilted disk and striking the disk face .
of the two accretion disks shown in figure 3 , the tilted disk is more blended in density .
hence , assuming a constant density disk in our derivations of retrograde precession in @xmath235 is justified .
note in figure 3 that the ring is tilted at the same angle as the disk and that the ring is rotating .
therefore , in addition to the disk , the ring is also subject to precession .
also note that numerically simulated cv dn disks ( figure 3 ) are elliptical and do not have excessively large , centrally located inner holes . as the disk radius is much larger than the radius of this small hole ,
our previous assumption of integration from the origin of the system to the outer edge of the disk is justified .
table 2 lists observational data including orbital periods , negative superhump periods , and negative superhump period excesses for several non - magnetic , accreting cv dn systems .
if errors are known , they are listed in parentheses . in the table , nl is novalike , ps is the permanent superhumper category of nl , ip is intermediate polar , sw sex and vy scl are members of nls , er uma is a member of the su uma subclass . in table 2 , we also list positive superhump periods and precessional data for those systems that precess both in the prograde and in the retrograde direction .
prograde precession is due to density waves forming radially through the middle - to - outer annuli of the disk ( smith et al .
2007 ) and is described by a dynamical prograde term plus a retrograde pressure term ( lubow 1991a , b ) .
the retrograde precession term is a correction to the prograde precession term due to these spiral density waves . in this work
, we do not establish the geometrical shape of an accretion disk with these radial density waves as these radial density waves are not generated in long orbital period cv dn systems .
we leave this as future work .
figure 4 shows a comparison of the observational nodal superhump period excess ( @xmath236 for ips , @xmath237 for am cvn , * for cv dn that progradely and retrogradely precess , and x for those systems that only retrogradely precess ) to the numerical negative superhump period excess ( @xmath238 ) values listed in tables 1 and 2 , respectively , as a function of orbital period .
the width of the symbol can be taken as the error in the orbital period .
in addition , we overplot a short dashed line that represents a retrogradely precessing rigid disk - equation ( 54 ) , a dash - dot line that represents a retrogradely precessing but differentially rotating disk - equation ( 55 ) , a long dash line that represents a retrogradely precessing ring - equation ( 56 ) , and a solid line that represents a disk that is differentially rotating yet retrogradely precessing as a unit and has a precessing ring . as shown in figure 4 , the retrograde precession of an inner annuli ring ( long dashed line ) does not match any of the observations , a result in agreement with the idea that the disk can not be represented by a ring .
the retrograde precession of solid , rigid disk also does not match any of the observations , a result in agreement with the idea that the disk rotates differentially as the disk is a fluid .
also shown in figure 4 , the retrograde precession of a differentially rotating disk does not match the numerical simulations .
of the four lines shown in figure 4 , the numerical simulations are best described by the solid line . to generate the solid line in figure 4
, we assume that the net retrograde precession is due to a differentially rotating disk that has superimposed on it a rotating ring due to the gas stream overflowing the tilted disk and particle migration as the secondary orbits .
the net precession is @xmath239 likewise , the net negative superhump period excess is @xmath240 as shown in figure 4 .
smith & dhillon ( 1998 ) find that cv secondary stars with orbital periods shorter than 7 - 8 hours are indistinguishable from main - squence stars in detached binaries in terms of spectral type , mass , and radius .
they find that cvs with periods up to 7 - 8 hours almost always have main - sequence secondaries , although they do point out that systems with peculiar secondaries do exist .
we assume their secondary mass - period relation in both our theoretical expressions and in our numerical simulations .
if smith & dhillon ( 1998 ) and this work are correct , then the solid line in figure 4 should represent those systems that have main sequence secondaries .
the one observational system that is on the solid line is v1193 ori .
hs 1813 + 6122 is nearby to one of our numerical simulation points . as smith & dhillon ( 1998 )
find that the average white dwarf mass above the period gap is @xmath241 , the same average mass we chose for the numerical simulations and theoretical expression for retrograde precession , we find @xmath242 and @xmath243 for v1153 ori .
we find @xmath244 and @xmath245 for hs 1813 + 6122 . in agreement with the ideas of smith & dhillon ( 1998 )
, these above - the - period - gap systems have stellar masses .
we do note the large scatter in the observational data . as shown in figure 4 by the various lines ,
the geometric shape of the disk significantly affects the precession rate and could be a major source to the scatter . another smaller source to the scatter
not considered in this work is density waves that may be generated in the disk .
therefore , we have identified the observational data in figure 4 by various different symbols to indicate systems with different geometries or otherwise unusual disks .
for example , those systems in table 2 labeled ips are expected to have a large absence of inner annuli due to the primary s large magnetic field that disrupts the disk . in this work
, we did not find the retrograde precession of a differentially rotating , retrogradely precessing , hollow cylinder and thus we flagged these observational systems in figure 4 as squares .
a hole as large as that experienced by an ip s disk would not be described by the solid line or any other line in figure 4 as the geometrical shape of this system is different than that described by these lines .
likewise , the geometrical shape of an accretion disk that both progradely and retrogradely precesses would not be described by any of the lines in figure 4 .
these systems have radial density waves in their accretion disks and may also have an innermost ring due to gas stream overflow and particle migration to innermost annuli because the disk is tilted .
we did not consider these geometrical shapes or disks with density waves in this work .
we leave these different geometrical shape disks for future work .
warps in disks can be localized , say in inner regions or in outer regions ( e.g. , boffin et al .
2003 ) , or across the entire disk .
an example of localized warps are those generated in inner disk annuli .
nelson & papaloizou ( 1999 ) find inner disk warps that are induced by misaligned angular momentum vectors of the disk and the rotating black hole .
petterson ( 1977 ) and iping and petterson ( 1990 ) study radiation pressure as a source to disk warp .
if a primary is a source of strong radiation , then the surface of the accretion disk can absorb this radiation and re - emit it normal to the disk surface , inducing a torque on the disk that may warp the disk if the central source luminosity exceeds a critical value ( see e.g. , petterson 1977 ; pringle 1996 , 1997 ; wijers & pringle 1999 ) .
the strong radiation source twists the disk so that each annuli affected precesses at the same rate . in these studies ,
the warp precesses relative to the inertial frame as a rigid body .
bisikalo et al .
( 2004 ) study cv accretion disks and they too find a localized inner disk warp that precesses yet the disk as a whole does not precess .
an example of entire disk warp that precesses as a unit are the studies of larwood et al .
( 1996 ) , larwood & papaloizou ( 1997 ) , and larwood ( 1997 ) .
these authors find precession rates due to a tidal interaction between a circumprimary disk or a circumbinary disk and a secondary point mass on a parabolic trajectory in protostellar systems .
their results indicate that disks become modestly warped ( @xmath246 ) during non - coplaner encounters and that disks precess approximately as a unit .
murray et al .
( 2002 ) find that a warp can be induced in cv accretion disks by an inclined magnetic dipole field that is centered on the secondary .
they also find that their warp precesses retrogradely and as a unit .
an example of a localized warp that may cause the entire disk to warp is the studies by foukes , haswell , & murray ( 2006 ) who generate warps in x - ray binary disks .
their source to warp is also radiation driven by the primary .
they find that inner disk annuli can warp and the warp precesses nearly as a rigid body in the retrograde direction . for less extreme mass ratio systems
, they find that inner and outer regions of the disk warp to the point that they can say that the entire disk is tilted to the binary plane . as a result of the entire disk tilt , the whole disk precesses in the retrograde direction
, a result also found in montgomery ( 2004 , 2009 ) .
for cv dn systems , warps seem unlikely and are usually not considered ( e.g. , boffin et al .
2003 ) as cv dn primary are not known to be strong radiation sources .
in addition , this study considers only cv dn that are considered non - magnetic .
the geometry suggested in larwood et al .
( 1996 ) , larwood & papaloizou ( 1997 ) , and larwood ( 1997 ) is unlikely for cv dn systems .
also unlikely is a misaligned disk and rotating primary angular momentum vectors as this source seems to only warp inner annuli without the generation of negative superhumps ( foulkes , haswell , & murray 2006 ) .
disk tilts have been postulated to involve the entire disk or only a portion such as the outer edge ( bonnet - bidaud , motch , & mouchet 1985 ; kumar 1986 , 1989 ) . as foulkes , haswell , & murray ( 2006 ) and montgomery ( 2009 )
find that whole disk needs to be tilted in order for the gas stream to flow over and under the disk edge , entire disk tilts are the likely geometry for non - magnetic , accreting , tilted , spinning systems if negative superhumps of the correct shape and frequency per orbit are expected to be seen in the light curves .
the source to disk tilt in non - magnetic systems remains unknown , and we leave this as future work . as figure 5 shows , a fully tilted disk may look like a disk warp as the system retrogradely precesses . in figure 5 , we only show one half of the total disk precessional period . in montgomery ( 2009 ) , we find that if the entire disk is tilted , then the gas stream can overflow a disk face and reach inner annuli . as the secondary orbits , the gas stream flows over one face of the disk for @xmath2471/2 an orbit and the flows over the other face of the disk for @xmath2471/2 an orbit .
the location where the gas stream transitions on the disk rim from flowing over to under and vice versa ( i.e. , the nodes ) precesses in the retrograde direction . like the slow westward movement of the first point of aries along the equator of the earth
, the line of nodes also has a slow westward movement along the rim of the disk .
if the disk precesses as a unit and the line of nodes shifts westward by a small amount each orbit , then the location where the gas stream strikes the innermost annuli must also shift westward by a small amount each orbit .
that is , the spinning , tilted ring formed near the innermost annuli of the disk and formed by the gas stream overflowing the tilted disk also precesses in the retrograde direction .
therefore , to explain precession in our numerical simulations , we net the retrograde precession of the tilted , rotating ring and the tilted , rotating disk .
bisikalo et al .
( 2004 ) finds that only their spiral density wave retrogradely precesses , not the entire numerical simulated cv accretion disk .
observations by patterson et al .
( 1998 ) do not indicate that disks precess differentially .
numerical simulations by murray ( 1998 ) do not indicate that disks precess differentially .
our results do not indicate that disks precess differentially .
papaloizou & terquem ( 1995 ) analytically show , and larwood et al .
( 1996 ) numerically verify , that rigid body precession of a keplerian disk is possible so long as the sound crossing time scale in the disk is small compared to the precession time scale of the disk .
we find in montgomery ( 2009 ) that the entire disk needs to tilt for additional particle migration to reach inner annuli and produce negative superhumps .
we find in this work that both outer and inner annuli precess in fully tilted disks .
these results suggest that partially retrogradely precessing non - magnetic cv dn tilted accretion disks are unlikely .
driven precession involves torques on a body .
one example is a wobbling coin on a table .
the coin spin rate is slow but its wobble rate is fast due to a torque on the body .
another example is a spinning top inclined at an angle from the vertical .
the spin rate is fast but the precession rate is slow due to torques on the body .
the precession of the earth due to tidal torques by the moon and the sun is another example of a slow precession rate compared to earth s ( relatively ) fast spin rate . for a cv dn system with @xmath186 , we previously calculated the retrograde precessional period to be approximately 80 hours . this period is slow compared to the primary s rotation period of around one hundred seconds , if we assume that the primary s rotation period is similar to that at the disk edge .
therefore , driven precession could be a viable source of retrograde precession in cv dn systems that show negative superhumps in their light curves .
katz ( 1973 ) , roberts ( 1974 ) , katz et al .
( 1982 ) , kondo et al . ( 1983 ) , kumar ( 1986 , 1989 ) , barrett et al . ( 1988 ) , warner ( 1995 , 2003 ) are all examples of those employing or assuming torques to induce retrograde precession .
as we show in figure 4 , retrograde precession in cv dn disks is likely due to the same source that causes the earth to retrogradely precess , that is , by tidal torques .
the difference in precessional values of the earths equinoxes and a cv dn s line of nodes is due to the geometric shape of the precessing object ( e.g. , oblate spheroid , disk , etc . ) , the state of matter ( i.e. , fluid , solid , etc . ) , and any effects due to the accretion stream .
bisikalo et al .
( 2004 ) generate a spiral density wave in innermost annuli of their numerical simulated cv disks .
their spiral density wave is caused by sheared elliptical inner disk annuli .
they find that only the spiral density wave retrogradely precesses , and they cite kumar ( 1986 ) as the source to retrograde precession .
however kumar ( 1986 ) finds the retrograde precession due to a tilted disk s outer annuli and thus can not be the source to retrograde precession of the spiral density wave .
however , we may be able to help bisikalo et al .
lubow & shu ( 1975 ) find that pressure can be generated in the disk due to shocks that are induced by gas stream overflow , and lubow ( 1991a , b ) finds that additional pressure due to spiral density waves could cause the disk to retrogradely precess .
if the inner disk has a spiral density wave instead of a dense ring , then the net retrograde precession can still be explained by the solid line in figure 4 .
that is , the disk would retrogradely precess due to combined effects due to tidal torques and pressure from a spiral density wave located near the innermost disk annuli ( instead of a ring ) .
we plan to show this in a future work .
negative superhumps are found in observational systems with high mass ratios , long orbital periods , and high mass transfer rates . in montgomery ( 2009 ) , we find that the negative superhump can be due to extra light being emitted from inner disk annuli .
the extra light is modulated by gas stream overflow that strikes inner disk annuli and by particle migration as the secondary orbits . for this result to be true
, we find in montgomery ( 2009 ) that the disk must be fully tilted .
the source to disk tilt has not yet been resolved and we leave this as future work . fully tilted disks could be subject to driven torques that cause the disk to retrogradely precess .
patterson et al . ( 1993 ) suggest that the retrograde precesison in cv dn accretion disks may be by the same source that causes the earth to retrogradely precess .
using the earth - moon - sun system as a model , we generate a generic theoretical expression for a spinning disk that is misaligned with the orbital plane to test this idea .
we assume the potential of maccullogh is valid for our spining , tilted disk that contains a massive primary that is smaller than the thickness of the disk and the distance to the secondary .
our expression matches that which describes the precession of earths equinoxes . by making appropriate assumptions ,
we reduce our generic expression for retrograde precession to those generated by others , or to those used by others , to describe retrograde precession in protostellar , protoplanetary , x - ray binary , non - magnetic accreting cv dn , quasar and black hole systems . because our generic expression reduces to those generated by others for a variety of systems , retrograde precession by tidal torques on a spinning , misaligned
primary may be ubiquitous .
previous expressions treated the disk or ring as rigid and/or did nt include the effects on the disk by the accretion stream .
because the disk is fully tilted , the gas stream overflows one face of the disk and particles migrate to inner annuli , creating a spinning , tilted ring that is denser than its surroundings as shown in our numerical simulations . in this work
, we generate a theoretical expression for the retrograde precession of a spinning , tilted ring .
we assume that the rotation rate of the primary matches that of the inner rim of the spinning , tilted ring .
this implies that the axes of the ring and the primary align .
we show that the net retrograde precession in cv dn non - magnetic accretion disks needs to include the effects due to both the precessing disk and the precessing ring .
if we include the net moment of inertia due to the disk and the ring , then our theoretical expression agrees well with our numerical results .
our findings support the suggestion by patterson et al .
( 1993 ) that tidal torques like those by the moon and the sun on the spinning , tilted earth are the source to retrograde precession in spinning , tilted cv dn accretion disks . in our theoretical expression and our numerical simulations ,
we assume an average primary mass @xmath248 that matches the average primary mass @xmath241 found by smith & dhillon ( 1998 ) .
if the results in smith & dhillon ( 1998 ) and this work are correct , then from figure 4 we find that v1193 ori and hs 1813 + 6122 secondaries may have stellar masses .
the scatter in observational precessional values seems to be mostly from differing geometrical disk shapes .
the scatter could also be from any density waves that may be present in the disk .
bisikalo et al .
( 2004 ) finds that a spiral density wave is generated in innermost annuli of their numerically simulated cv accretion disks .
they find that their spiral density wave retrogradely precesses .
pressure effects generated within the disk from a spiral density wave could add to retrograde precession .
the disk could retrogradely precess due to combined effects due to tidal torques and pressure from a spiral density wave .
we plan to show this in a future work , and we anticipate that the result is the same solid line shown in figure 4 .
however , unlike that found by bisikalo et al .
( 2004 ) , we expect the disk to still precess as a unit .
we do note that we did not consider significant mass loss of the secondary over time and this may also affect precessional values
. if the mass lost by the secondary is not significantly acquired by the primary and is instead mostly lost via e.g. , winds , jets , or circumbinary disks , then the geometric shape of the disk , moment of inertia , mass ratio , and/or orbital period changes and thus so changes the precessional values .
whether the retrograde precession and orbital period values change one - to - one is not fully understood and we leave this as future work as well .
the author would like to acknowledge an aas small research grant and a ucf - uf space research initiative that enabled us to generate this work on our eight - node server .
we would like to thank undergraduate mark guasch for colorizing the disks for this work and undergraduate jon edmiston for finding the elements of the moment of inertia tensor for a rigid elliptical disk .
we thank graham wynn and luigi stella , the reviewer and scientific editor for apj , respectively , as their comments enabled a manuscript that flowed better than the original version .
we especially thank graham wynn for his suggestion of differentiating potentially different geometrical disk shapes with different symbols and addressing other geometrical shape disks ( e.g. , ip disks ) that are not considered in this work .
99 ak t. , retter a. , liu a. , & esenoglu h.h .
, 2005 , pasa , 22 , 105 ak t. , retter a. , liu a. , & esenoglu h.h . , 2005b , newa , 11 , 147 araugo - betancor a. , knigge c. , long k.s . ,
hoard d.w .
, szkody p. , rodgers b. , krisciunas k. , dhillon v.s .
, hynes r.i .
, patterson j. , & kemp j. , 2003 , apj , 583 , 437 arenas j. , catalan m.s .
, augusteijn t. , & retter a. , 2000 , mnras , 311 , 135 bardeen j.m . & petterson j.a . , 1975 , apj , 195 , l65 barrett p. , odonoghue d. , & warner b. , 1988 , mnras , 233 , 759 barrera l.h . & vogt n. , 1989 , a&a , 220 , 99 biro i.b . , 2000 ,
a&a , 364 , 573 bisikalo d.v . ,
boyarchuk a.a . , kaygorodov p.v . ,
kuznetsov o.a . ,
& matsuda t. , 2004 , arep , 48 , 449 boffin h.m.j . , stanishev v. , kraicheva z. , & genkov v. , 2003 , aspc , 292 , 297 bonnet - bidaud j.m . , motch c. , & mouchet m. , 1985 , 143 , 313 borczyk w. , schwarzenberg - czerny a. , szkody p. , 2003 ,
a&a , 405 , 663 boyton p.e . ,
crosa l.m .
, deeter j.e . , 1980 , apj , 237 , 169 caproni a. & abraham z. , 2002 , rmxac , 14 , 74 danby j.m.a . , 1962 , in fundamentals of celestial mechanics ( new york : macmillan company )
dhillon v.s .
, jones d.h.p .
, marsh t. r. , & smith r.c .
, 1992 , mnras , 258 , 225 eggleton p.p . , 1983 ,
apj , 268 , 368 foulkes s.b .
, haswell c.a . , & murray j.r . , 2006 , mnras , 366 , 1399 fu h. , li z .- y .
, leung k .- c .
, zhang z .- s .
, & gaskell c.m . , 2004 ,
chjaa , 4 , 88 gao w. , li z. , wu x. , zhang z. , & li y. , 1999 , apj , 527 , 55 gulsecen h. , retter a. , liu a. , esenoglu h. , 2009 , newa , 14 , 330 harvey d. , skillman d.r .
, patterson j. , & ringwald f.a .
, 1995 , pasp , 107 , 551 hellier & buckley d.a.h . , 1993 , mnras , 265 , 766 hessman f.v .
& hopp u. , 1990 , a&a , 228 , 387 hoard d.w . &
szkody p. , 1997
, apj , 481 , 433 huber m.e .
howell s.b . , ciardi d.r . , & fried r. , 1998 , pasp , 110 , 784 iping r.c .
& petterson j.a . , 1990 , a&a , 239,221 kang t.w . , retter a. , liu a. , & richards m. , 2006 , aj , 131 , 1687 kato t. , ishioka r. , & uemura m. , 2002 , pasj , 54 , 1033 katz j.i . , 1973 , nature phys .
, 246 , 87 katz j.k .
, anderson s.f . , margoon b. , & grandi s.a . , 1982 , apj , 260 , 780 kondo y. , wolff c.l . , & van flandern t.c . , 1983 , apj , 273 , 716 kozhevnikov v.p . , 2004 ,
a&a , 419 , 1035 kozhevnikov v.p . , 2007 , mnras , 378 , 955 kraicheva z. , stanishev v. , genkov v. , & iliev l. , 1999 a&a , 351 , 607 kumar s. , 1986 , mnras , 223 , 225 kumar s. , 1989 , in theory of accretion disks , eds .
f. meyer , w.j .
duschl , j. frank , & e. meyer - hofmeister , kluwer , dordrecht , p. 297
kumar s. & pringle j.e .
, 1985 , mnras , 213 , 435 larwood j. , 1997 , mnras , 290 , 490 larwood j. , 1998 , mnras , 299 , l32 larwood j. , nelson r.p . , papaloizou j.c.b . , & terquem c. , 1996 , mnras , 282 , 597 larwood j. & papaloizou j.c.b .
, 1997 , mnras , 285 , 288 lasota j .-
, 2001 , new astronomy review , 45 , 449 lattanzio j.c .
, monaghan j.j . , pongracic h. , & schwarz m.p . , 1986 , j. sci .
comput . 7 , 591 li z. , leung k .- c .
, & gaskell c.m . , 2004 , rmxac , 21 , 259 livio m. , 1999 , phr , 311 , 225 lubow s.h . &
shu f.h . , 1975 , apj , 198 , 383 lubow s.h . , 1991a , apj , 381 , 259 lubow s.h . ,
1991b , apj , 381 , 268 lubow s.h .
, 1992 , apj , 398 , 525 lubow s.h . &
pringle j.e . , 1993 , apj , 409 , 360 meyer f. & meyer - hofmeister e. , 1981 , a&a , 104 , l10 montgomery m.m . , 2004 , ph.d thesis , florida institute of technology montgomery m.m .
, 2009 , mnras , 394 , 1897 murray j.r .
, 1998 , mnras , 300 , 561 murray j.r . & armitage p.j . , 1998 , mnras , 300 , 561 murray j.r . , truss m.r . ,
& wynn g.a . , 2002 , aspc , 261 , 416 nelemans g. , steeghs d. , & groot p.j .
, 2001 , mnras , 326 , 621 nelson r.p . &
papaloizou j.c.b . , 1999 ,
mnras , 309 , 929 olech a. , rutkowski a. , & schwarzenberg - czerny a. , 2007 , aca , 57 , 331 osaki y. , 1985 , a&a , 144 , 369 paczynski b. , 1977 , apj , 216 , 822 papaloizou j.c.b .
& terquem c. , 1995 , mnras , 274 , 987 patterson j. , 1998 , pasp , 110 , 1132 patterson j. , thomas g. , skillman d.r .
, & diaz m. , 1993 , apjs , 86 , 235 patterson j. , jablonski f. , koen c. , odonoghue d. , & skillman d.r . , 1995 , pasp , 107 , 1183 patterson j. , kemp j. , saad j. , skillman d. , harvey d. , fried r. , thorstensen .r . , ashley r. , 1997 , pasp , 109 , 468 patterson j. , kemp j. , richman h.r .
, skillman d.r . ,
vanmunster t. , jensen l. , buckley d.a.h .
, odonoghue d. , & kramer r. , 1998 , pasp , 110 , 415 patterson j. , thorstensen j.r .
, fried r. , skillman d.r . , cook l.m .
, & jensen l. , 2001 , pasp , 113 , 72 patterson j. , fenton w.h .
, thorstensen j.r .
, harvey d.a . ,
skillman d.r .
, fried r.e . , monard b. , odonoghue d. , beshore e. , martin b. , niarchos p. , vanmunster t. , foote j. , bolt g. , rea r. , cook l. , butterworth n. , & wood m. , 2002 , pasp , 114 , 1364 patterson j. , kemp j. , harvey d.a . , fried r.e .
, rea r. , monard b. , cook l. , skillman d.r .
, vanmunster t. , bolt g. , armstrong e. , mccormick j. , krajci t. , jensen l. , gunn j. , butterworth n. , foote j. , bos m. , masi g. , & warhurst p. , 2005 ,
pasp , 117 , 1204 petterson j. , 1977 , apj , 216 , 827 pringle j.e .
, 1996 , mnras , 281 , 357 pringle j.e . , 1997 , mnras , 292 , 136 retter a. , leibowitz e.m . ,
& ofek e.o . , 1997 , mnras , 286 , 745 retter a. , hellier c. , augusteijn t. , naylor t. , bedding t.r . , bembrick c. , mccormick j. , & velthuis f. , 2003 , mnras , 340 , 679 retter a. , liu a. , & bos m. , 2005 , assl , 332 , 251 roberts w.j . , 1974 , apj , 187 , 575 rodriguez - gil p. ( and 17 others ) , 2007 , mnras , 377 , 1747 rodriguez - gil p. , schmidtobreick l. , & gansicke b.t . ,
2007b , mnras , 374 , 1359 roelofs g.h.a .
, groot p.j . , nelemans g. , marsh t.r . , &
steeghs d. , 2006 , mnras , 371 , 1231 romero g.e . , chajet l. , abraham z. , & fan j.h . , 2000 ,
a&a , 360 , 57 schoembs r. , 1982 , a&a , 115 , 190 shafter a.w . &
szkody p. , 1983 ,
pasp , 95 , 509 shakura n.i . &
sunyaev r.a . , 1973 , a&a , 24 , 337 skillman d.r . , patterson j. , kemp j. , harvey d.a . , fried r.e . , retter a. , lipkin y. , vanmunster t. , 1999 , pasp , 111 , 1281 smak j. , 1999 , acta astronomica , 49 , 391 smith a.j .
, haswell c.a . , murray j.r . ,
truss m.r . ,
& foulkes s.b . , 2007 , mnras , 378 , 785 smith d.a .
& dhillon v.s .
, 1998 , mnras , 301 , 767 stacey f.d . , 1977 , physics of the earth , second edition ( new york : john wiley & sons ) szkody p. & howell s.b . , 1993 ,
apj , 403 , 743 taylor c.j . ,
thorstensen j.r . , patterson j. , fried r.e . , vanmunster t. , harvey d.a .
, skillman d.r .
, jensen l. , & shugarov s. , 1998 , pasp , 110 , 1148 thorstensen j.r . , taylor c.j . , becker c.m . , & remillard r.a . , 1997 ,
pasp , 109 , 477 terquem c. , papaloizou j.c.b . , & nelson r.p . , 1999 , astrophysical disks
asp conference series , vol .
160 , 71 terquem c. & papaloizou j.c.b . , 2000 , a&a , 360 , 1031 tramposch j. , homer l. , szkody p. , henden a. , silvestri n.m . , yirak k. , fraser o.j . , & brinkmann j. , 2005 , pasp , 117 , 262 warner b. , 1995 , cataclysmic variable stars ( new york : cambridge u. press ) warner b. , 2003 , cataclysmic variable stars ( new york : cambridge u. press ) wijers r.a.m.j . & pringle j.e . , 1999 , mnras , 308 , 207 wood m.a . , montgomery m.m .
, & simpson j.c . ,
2000 , apjl , 535 , l39 woudt p.a . & warner b. , 2002 , mnras , 335 , 44 woudt p.a . , warner b. , & spark m. , 2005 , 364 , 107 wu x. , li z. , ding y. , zhang z. , & li z. , 2002 , apj , 569 , 418 @xmath249nl , @xmath250vy scl / nl , @xmath251ps / nl , @xmath252su uma / dn , @xmath253er uma / dn , @xmath254ip / nl , @xmath255z cam , @xmath214sw sex / nl , @xmath256nova , @xmath257classification uncertain + @xmath258taylor et al . ( 1998 ) , @xmath259huber et al .
( 1998 ) , @xmath260patterson et al .
( 2005 ) , @xmath261kraicheva et al .
( 1999 ) , @xmath262wu et al .
( 2002 ) , @xmath263warner ( 1995 ) , @xmath264borczyk et al .
( 2003 ) , @xmath265patterson et al .
( 1997 ) , @xmath266arenas ( 2000 ) , @xmath267harvey et al .
( 1995 ) , @xmath268patterson et al .
( 1995 ) , @xmath269thorstensen et al .
( 1997 ) , @xmath270skillman et al .
( 1999 ) , @xmath271nelemans et al .
( 2001 ) , @xmath208roelofs et al .
( 2006 ) @xmath272retter et al .
( 2003 ) , @xmath273hellier & buckley ( 1993 ) , @xmath274schoembs ( 1982 ) , @xmath275barrera & vogt ( 1989 ) , @xmath276gao et al .
( 1999 ) , @xmath277patterson et al .
( 2001 ) , @xmath278rx j1643.7 + 3402 - patterson et al .
( 2002 ) , @xmath279shafter & szkody ( 1983 ) , @xmath280retter et al . ( 1997 ) , @xmath281patterson ( 1998 ) , @xmath282rodriguez - gil et al .
( 2007b ) , @xmath283biro ( 2000 ) , @xmath284araujo - betancor et al .
( 2003 ) , @xmath285boffin et al .
( 2003 ) , @xmath286dhillon et al .
( 1992 ) , @xmath287hoard & szkody ( 1997 ) , @xmath288 woudt & warner ( 2002 ) , @xmath289kozhevnikov ( 2004 ) , @xmath290li et al .
( 2004 ) , @xmath291fu et al .
( 2004 ) , @xmath292retter et al .
( 2005 ) , @xmath293sdss j040714.78 - 064425.1 - ak et al .
( 2005 ) , @xmath294sdss j210014.12 + 004446.0 - tramposch et al .
( 2005 ) , @xmath295woudt et al .
( 2005 ) , @xmath296kozhevnikov ( 2007 ) , @xmath297kato et al . ( 2002 ) , @xmath298rodriguez - gil et al .
( 2007 ) , @xmath299kang et al .
( 2006 ) , @xmath300gulsecen et al .
( 2009 ) , @xmath301szkody & howell ( 1993 ) , @xmath302olech et al .
( 2007 ) , @xmath303ak et al .
( 2005b ) + | net tidal torque by the secondary on a misaligned accretion disk , like the net tidal torque by the moon and the sun on the equatorial bulge of the spinning and tilted earth , is suggested by others to be a source to retrograde precession in non - magnetic , accreting cataclysmic variable ( cv ) dwarf novae systems that show negative superhumps in their light curves .
we investigate this idea in this work .
we generate a generic theoretical expression for retrograde precession in spinning disks that are misaligned with the orbital plane .
our generic theoretical expression matches that which describes the retrograde precession of earths equinoxes . by making appropriate assumptions ,
we reduce our generic theoretical expression to those generated by others , or to those used by others , to describe retrograde precession in protostellar , protoplanetary , x - ray binary , non - magnetic cv dn , quasar and black hole systems .
we find that spinning , tilted cv dn systems can not be described by a precessing ring or by a precessing rigid disk .
we find that differential rotation and effects on the disk by the accretion stream must be addressed .
our analysis indicates that the best description of a retrogradely precessing spinning , tilted , cv dn accretion disk is a differentially rotating , tilted disk with an attached rotating , tilted ring located near the innermost disk annuli . in agreement with observations and numerical simulations by others
, we find that our numerically simulated cv dn accretion disks retrogradely precess as a unit .
our final , reduced expression for retrograde precession agrees well with our numerical simulation results and with selective observational systems that seem to have main sequence secondaries .
our results suggest that a major source to retrograde precession is tidal torques like that by the moon and the sun on the earth .
in addition , these tidal torques should be common to a variety of systems where one member is spinning and tilted , regardless if accretion disks are present or not .
our results suggest that the accretion disk s geometric shape directly affects the disk s precession rate . |
and @xmath0 are the color indices of quarks and gluons , respectively_,width=529 ] in standard model the properties of quarks and leptons are remarkably similar , as far as the electroweak interactions are concerned .
the quarks with six different flavours are grouped into the three doublets @xmath1,@xmath2,@xmath3 , in one - to - one correspondence to the three lepton doublets @xmath4,@xmath5 , @xmath6 , as shown in the schematic chart in fig .
[ fig : chart ] .
quarks and leptons interact in a similar way with the electroweak vector bosons @xmath7,@xmath8 and @xmath9 .
furthermore , it is anticipated that the universal higgs mechanism is responsible for the generation of the quark and lepton masses ( for more details see @xcite ) .
in addition , quarks interact `` on their own '' , revealing their specific property , the colour charge ( _ colour _ ) , a conserved quantum number which is absent for leptons .
a quark of a given flavour has three different colour states with equal masses and electroweak charges .
colour - induced interactions between quarks are mediated by _ gluons _
, the massless and electroweakly - neutral spin-1 particles , in these lectures i will discuss the quark - gluon `` corner '' of the standard model , introducing the basics of quantum chromodynamics ( qcd ) , the theory of quark - gluon interactions . throughout this survey , the main emphasis will be put on the relation between qcd and _ hadrons _ , the observed bound states of quarks .
this first lecture is devoted to the basic properties of the quark - gluon interactions .
i will frequently refer to quantum electrodynamics ( qed ) , the more familiar theory of electromagnetic ( e.m . )
interactions , which is a useful prototype of qcd . in fig .
[ fig : diags ] the feynman diagram of the electron - muon e.m .
scattering is drawn together with the analogous diagram of the quark - quark interaction .
for definiteness , i specify the quarks as having @xmath10 and @xmath11 flavours , the counterparts of @xmath12 and @xmath13 in the standard model .
the two interactions have many important similarities : * the colour - charged quarks emit and absorb gluons in the same way as the electrically charged leptons emit and absorb photons ; * gluon and photon are massless ; * both gluon and photon have spin 1 . being guided by this analogy , one would expect that gluon exchanges generate a coulomb - type interquark force , similar to the usual attraction / repulsion between the electrically charged particles . in reality ,
quark - gluon interactions are far more complicated .
in particular , since the colour charge has three components , quarks can change their color states after emitting / absorbing gluons , as indicated on the diagram in fig .
[ fig : diags ] .
hence , due to the colour conservation , gluons also carry colour quantum numbers , and , as a result , interact with each other .
in fact , it is the gluon self - interaction that makes qcd - dynamics so peculiar .
yet qcd has one basic property in common with qed .
both theories have specific `` internal '' symmetries , named _ local gauge symmetries _ , to be discussed below .
the lepton - photon dynamics is described by one compact formula of the qed lagrangian , @xmath14 where the physical degrees of freedom of the photon field ( the electric and magnetic fields ) are combined in the field strength tensor @xmath15 , with @xmath16 . in the above
, @xmath17 is the lepton dirac field with spin 1/2 and mass @xmath18 , and a compact notation for the covariant derivative @xmath19 is introduced . starting from @xmath20 it is possible to derive dirac equations for the leptons and maxwell equations for the photon .
furthermore , eq . ( [ lqed ] ) yields the basics elements of the qed feynman diagrams : the photon and lepton propagators and the lepton - photon interaction vertex . employing these elements ,
one obtains the amplitude of a given e.m .
process in a form of a perturbative expansion in the numerically small coupling @xmath21 .
for us the most interesting property of @xmath20 is the gauge symmetry which reveals itself if one locally changes the phase of the lepton fields : @xmath22\psi_l(x)\ , , \nonumber \\
\bar{\psi}_l(x ) & \to & \bar{\psi}'_l(x)=\bar{\psi}_l(x ) exp\left[i\chi(x)\right]\ , , \label{u1}\end{aligned}\ ] ] @xmath23 being an arbitrary function of the 4-coordinate @xmath24 .
if one simultaneously adds the derivative of the same function to the photon field : @xmath25 the lagrangian ( [ lqed ] ) remains invariant .
it is easy to check that the transformations ( [ u1 ] ) and ( [ u1a ] ) leave intact the physical observables , such as the e.m .
current @xmath26 or @xmath27 .
importantly , the local gauge symmetry implies that the photon mass vanishes , @xmath28 .
indeed , adding to @xmath20 a mass term @xmath29 for the photon field would evidently violate the symmetry because the latter term changes under ( [ u1a ] ) .
the particular case @xmath30 ( when only the lepton fields are transformed ) corresponds to the _ global _ gauge symmetry which is responsible for the e.m .
current conservation in qed . from the mathematical point of view
the qed gauge transformations form a _
group_. let me remind that a given set of elements @xmath31 can be qualified as a group if three conditions are simultaneously fulfilled : 1 ) a multiplication rule can be defined @xmath32 , that is , a correspondence is established between a given pair of elements @xmath33 and @xmath34 and a certain third element @xmath35 belonging to the same set ; 2 ) the unit element @xmath36 exists , so that @xmath37 for each @xmath33 and 3 ) the inverse element @xmath38 can be specified for each @xmath34 , so that @xmath39 .
all three conditions are valid for the ( infinite and continuous ) set of transformations ( [ u1 ] ) and ( [ u1a ] ) generated by the set of the arbitrary functions @xmath23 . indeed , performing two gauge transformations one after the other , with @xmath40 and @xmath41 ,
is equivalent to the gauge transformation with @xmath42 .
the unit element of this `` multiplication '' rule is simply the transformation with @xmath43 and the inverse element for each @xmath23 is @xmath44 . importantly , the group multiplication is commutative ( the group is abelian ) because the result of the overlap of two phase transformations is independent of their order .
the group we are discussing is called @xmath45 .
mathematically , it is equivalent to the group of rotations of the cartesian coordinate system around one of its axes .
the rotation angle plays the same role as the phase @xmath46 .
gauge transformations in qcd have a more rich geometry .
they are somewhat similar to the general rotations in the 3-dimensional space , involving @xmath47 matrices , which do not commute .
more specifically , a colour gauge transformation of the quark field @xmath48 with a given flavour @xmath49 , involves transitions between different color components , a sort of `` rotations of colour coordinates '' : @xmath50 the elements of the @xmath51 matrix @xmath52 depend arbitrarily on the 4-point @xmath24 .
furthermore , this matrix is unitary : @xmath53 or , in a symbolic form , @xmath54 . to explain why unitarity is necessary ,
let me invoke the following physical argument .
since photon does not distinguish quark colours , the only admittable form for the e.m .
interaction of quarks is @xmath55 where @xmath56 for @xmath57 and the summation goes over the quark colour indices .
the gauge transformation ( [ su3transf ] ) applied to the quark fields yields @xmath58 clearly , only if ( [ unit ] ) is valid , @xmath59 remains invariant .
the usual exponential representation of the gauge transformation matrix is : @xmath60\ , .
\label{matrix}\ ] ] it contains eight independent and arbitrary functions @xmath61 multiplied by eight reference matrices @xmath62 ( @xmath0 ) .
the latter have the form chosen by gell - mann : @xmath63 the @xmath64-matrices have the following properties : @xmath65 ( hermiticity ) , @xmath66 , and @xmath67= f^{abc}\lambda^c$ ] ( noncommutativity ) , where @xmath68 are totally antisymmetric constants ( @xmath69 , etc . ) .
it is easy to check that @xmath70 defined in ( [ matrix ] ) obeys unitarity and has a unit determinant @xmath71 .
the ( infinite and continuous ) set of noncommutative @xmath72 matrices forms a group , which is called @xmath73 .
one may ask : why there are eight independent functions @xmath74 determining the rotations of the three color states ? is nt eight too many ?
the point is that the quark fields @xmath75 are complex functions , hence the matrix @xmath70 is also a complex function of @xmath24 determined by @xmath76 real functions .
only 8 of them are independent because there are altogether 10 constraints : nine provided by the unitarity relation ( [ unit ] ) and one by the unit determinant . with eight `` rotation angles '' the group @xmath73 is of course quite different from the group of rotations in three dimensions which has only 3 parameters .
the qcd lagrangian has to be exactly symmetric with respect to the local gauge transformations ( [ su3transf ] ) with matrices ( [ matrix ] ) .
this property serves as a guiding principle for constructing @xmath77 .
we start from the part of the lagrangian which describes the propagation of free quarks : @xmath78 and yields the usual dirac equation for spin 1/2 particle for each quark with a given flavour and colour .
similar to the case of qed , the expression ( [ quarkl ] ) is not invariant with respect to the local gauge transformations ( [ su3transf ] ) .
an additional term with derivatives of @xmath61 remains : @xmath79\gamma^\mu\psi_{q}^{j}(x)\ , .
\label{quarkp}\ ] ] to restore gauge invariance one follows the same scenario as in qed or in the electroweak theory @xcite .
the idea ( put forward long ago by yang and mills ) is to introduce `` compensating '' spin-1 fields interacting with quark fields .
there should be one separate spin-1 field for each of the eight degrees of freedom determining the gauge transformations ( [ su3transf ] ) . in this way
eight gluons enter the game , with the following quark - gluon interaction term added to the lagrangian : @xmath80 here @xmath81 is the dimensionless coupling analogous to @xmath12 in @xmath20 .
in contrast to e.m .
interaction , where the photon field is electrically neutral , the gluon fields also carry colour charge , so that the colour state of a quark changes after emitting / absorbing a gluon .
the colour of the gluon distinguished by the index @xmath82 can be identified with a superposition of quark and antiquark colours .
for example the gluon field @xmath83 is in the same colour state as the quark - antiquark pair @xmath84 .
the @xmath85-fields in ( [ lintqcd ] ) have to be gauge - transformed in the following way : @xmath86 so that the overall change of @xmath87 cancels the symmetry breaking term on r.h.s . of ( [ quarkp ] ) .
it is a simple exercise to check that the combination of transformations ( [ su3transf ] ) and ( [ atransf ] ) leaves the sum @xmath88 invariant .
we see that in qcd the `` compensating '' transformation of gluon fields ( [ atransf ] ) is more complicated than its analog ( [ u1a ] ) in qed : the addition of the derivative over @xmath46-functions is accompanied by a `` color rotation '' . to complete
the lagrangian one has to add a gauge - invariant term describing the propagation of gluon fields : @xmath89 where @xmath90 is the gluon field - strength tensor .
the local gauge invariance of @xmath91 implies that gluons are massless . at the same time
, @xmath92 is a more complicated object than its qed analog @xmath27 . indeed , substituting ( [ gmunu ] ) to ( [ gluel ] ) , we notice that not only the terms quadratic in @xmath85 emerge ( propagation of gluons ) but also the three- and four - gluon vertices ( gluon self - interactions ) . note that formally , both properties of gluons : colour quantum number and self - interactions are due to the noncommutativity of the gauge - transformation group ( the fact that @xmath93 ) . the final form of qcd lagrangian
is obtained by adding together the three pieces introduced above : @xmath94 where @xmath95 .
to summarize , @xmath77 describes not only quark - gluon interactions but also _ gluodynamics _ , the specific gluon self - interactions which have no analog in qed .
quantum correction ( when photons exchange virtual electrons via box diagrams ) . for a classical maxwellian electrodynamics , self - interacting e.m .
fields would mean `` new physics beyond standard theory '' .
i am not aware of any discussions of photon self - interactions in the times before quantum field theory .
interestingly , the light emitting light was mentioned in poetry .
i found the following sentence written in xiii century by armenian poet kostandin erznkazi @xcite : `` and so the light was born from the light , the great light of sun ... '' ( in translation from armenian ) ] the quark and gluon propagators and vertices derived from @xmath77 are collected in table 1 .
the formula of the gluon propagator has a certain degree of freedom related to the fact that the physical massless gluon has only two polarization / spin states whereas the field @xmath85 has four components . to make things working ,
one follows the same recipe as in qed .
an additional constraint on the gluon field is introduced , the so called gauge - fixing condition .
the gluon propagator given in table [ tab : fr ] corresponds to the usual feynman gauge .
note that in qcd purely gluonic loop diagrams are possible , in which case one has to take care of subtracting the contributions of unphysical components of @xmath85 also in these loops .
it is usually done by adding specially designed fictitious particles ( the so called fadeev - popov ghosts ) which only appear in the loops , and are not shown in table 1 .
._propagators and vertices in qcd . _ [ cols="^,^ " , ] quark - gluon interaction is flavour - independent . therefore , given that a @xmath96 meson with a certain @xmath97 exists , the mesons with the same @xmath97 containing all possible quark - antiquark flavour combinations should also be observed .
the flavour partners of the pion ( @xmath98-meson ) with @xmath99 ( @xmath100 ) are listed in table [ tab : mesons ] .
almost all of them have been observed ; the masses and other characteristics are given in @xcite .
the only temporary exceptions are the pseudoscalar @xmath101 meson ( @xmath102 ) and the vector @xmath103 meson ( @xmath104 ) .
these two states are not yet in @xcite , due mainly to experimental reasons . the _ heavy quarkonia _ , i.e. , the mesons consisting of heavy quark and antiquark ( @xmath105 or @xmath106 ) are of a special interest . here
the masses of interacting quarks are large enough compared to their characteristic energies within hadrons : @xmath107 . in other words ,
heavy quarks are nonrelativistic objects with respect to qcd interactions .
it is therefore possible to approximate the quark - gluon interactions with a nonrelativistic potential , putting the hadronic calculus on the safe ground of quantum mechanics .
the coulomb quark - antiquark potential @xmath108 , derived from the one - gluon exchange , is valid at small distances . to provide quark confinement ,
a certain long - distance part of the potential , infinitely growing at @xmath109 should also exist ( e.g. , oscillator or linear potential ) .
this part of the potential can not be directly calculated from @xmath77 and is usually modeled and fitted to the observed quarkonium spectra .
importantly , numerical studies of qcd on the lattice confirm the existence of the confining potential force between heavy quark and antiquark .
in addition to the exact colour- and space - time symmetries , qcd lagrangian possesses approximate flavour symmetries originating from the pattern of quark masses . since the latter
are generated by some external mechanism , the flavour symmetries do not have fundamental roots in qcd .
nevertheless , they provide very important relations for hadron masses and hadronic amplitudes .
let us start with the @xmath110 and @xmath10 quarks and rewrite the qcd lagrangian , isolating these two flavours : @xmath111 the smallness of the @xmath110 and @xmath10 masses , @xmath112 , implies that also their difference is small : @xmath113 neglecting this difference we use a new notation for the common @xmath114 quark mass : @xmath115 in this approximation , @xmath116 where a new , two - component spinor field ( doublet ) is introduced : @xmath117 the theory described by the r.h.s . of ( [ qcdisosp ] )
is not exactly qcd , but is very close to it .
the new lagrangian @xmath118 contains two degenerate quark flavours and has a symmetry with respect to the general phase rotations in the `` two - flavour space '' : @xmath119 where @xmath120 are arbitrary ( @xmath24-independent ) parameters and @xmath121 are the @xmath122 pauli matrices .
the symmetry transformations ( [ su2transf ] ) form a group su(2 ) . 2 matrix with the unit determinant . ]
one has to emphasize that the approximate su(2)-flavour symmetry emerges `` by chance '' , simply because the @xmath110 and @xmath10 quark masses turn out to be almost degenerate . note also that e.m .
interaction violates this symmetry , due to different electric charges of @xmath110 and @xmath10 quarks .
but this @xmath123 effect is again small , at the same level of @xmath124 , as the @xmath125 violation due to the quark mass difference .
the approximate degeneracy of @xmath110 and @xmath10 flavours manifests itself in hadrons . replacing @xmath110 quarks by @xmath10 quarks or vice versa in a given hadron , yields a different hadron which has a very close mass and other properties .
this qualitative prediction is nicely confirmed by the measured mass differences between proton ( @xmath126 ) and neutron ( @xmath127 ) , @xmath128 and @xmath129/\sqrt{2})$ ] , @xmath130 and @xmath131 , etc .. the typical mass splittings for the @xmath132 hadronic partners are at the level of few mev .
thus , qcd nicely explains the origin of _ isospin _ symmetry introduced by heisenberg in 30 s in nuclear physics , to describe the similarities between the `` mirror '' isotopes , obtained from each other by interchanging protons and neutrons .
the second part in the word isospin reflects the analogy with the electron spin symmetry , the degeneracy of the spin - up and spin - down electron states in quantum mechanics .
in fact , one introduces a similar formalism in qcd , attributing isospin @xmath133 to the doublet of @xmath110 and @xmath10 quarks and treating these two flavours as `` up '' and `` down '' components with @xmath134 and @xmath135 , respectively .
the hadrons containing @xmath110 and @xmath10 quarks in different combinations form isomultiplets , with the isospin counting similar to the spin algebra in quantum mechanics . in the case of proton and neutron ,
the diquark @xmath136 has isospin 0 , therefore , adding @xmath110 or @xmath10 quark to the diquark , we get the nucleon isodoublet ( @xmath133 ) consisting of proton with @xmath134 and neutron with @xmath135 .
another isodoublet is formed by @xmath137 and @xmath138 , where @xmath139 quark , which has no isospin , is combined with @xmath110 and @xmath10 , respectively . in the same way ,
@xmath140 and @xmath141 , or @xmath142 and @xmath143 build isodoublets . note that antiquarks have the opposite signs of @xmath144 : @xmath145 ( @xmath146 ) has @xmath135 ( @xmath134 ) . combining @xmath110 and @xmath10 quark with their antiquarks ,
one gets four states .
three of them belong to isotriplet ( @xmath147 ) : @xmath148 for example , pions ( @xmath149 , @xmath150 and @xmath151 ) , as well as @xmath98 mesons ( @xmath152 , @xmath153 and @xmath154 ) form isotriplets .
the fourth quark - antiquark state is an isosinglet ( @xmath155 ) : @xmath156 which deserves a separate discussion . in general , the @xmath157 and @xmath158 states transform into each other via intermediate gluons . in mesons
this transition takes place at long distances , due to some nonperturbative mechanism , not necessarily described by diagrams with a fixed number of gluons . in any case
, the transition amplitude has a characteristic scale of @xmath159 , much larger than the mass difference @xmath160 between the @xmath157 and @xmath158 states .
the @xmath161 degeneracy yields two orthogonal physical states , the @xmath162 component of the isotriplet ( [ isotriplet ] ) and the isosinglet ( [ iso0 ] ) . turning to strange quarks
one encounters the second @xmath155 state @xmath163 .
the same gluonic transition mechanism provides a mixing between ( [ iso0 ] ) and the @xmath163-state , now , the difference between the masses is not small , being of @xmath164 .
hence , the amount of mixing depends on the magnitude of the @xmath165 transition amplitude .
the latter is quite sensitive to the spin - parity of the meson state .
for example , the two isosinglet mesons with @xmath99 shown in table [ tab : mesons ] have ( up to small deviations ) the following quark content : @xmath166 indicating that mixing in the @xmath167 channel is large .
for the @xmath100 mesons the situation is completely different . from the quark content of the isosinglet
mesons : @xmath168 one concludes that the transition from strange to nonstrange quark pairs in the @xmath169 state is suppressed .
returning to the isospin symmetry , it is worth to mention that it yields useful relations between hadronic amplitudes . to give just one simple example ,
let us consider the four observable hadronic decays of @xmath170 -meson .
the amplitudes of these decays are related via isospin symmetry , so that only one amplitude is independent : @xmath171 to obtain these relations , one does not necessarily need to apply formulae for su(2 ) group . in the isospin limit , all four decays are described by a single quark diagram , shown in fig . [
fig : kstkpi ] , where the initial and final mesons are taken in the valence - quark state .
hadronic decays ; the four possible combinations of @xmath172 and @xmath173 correspond to the four decay modes related by isospin symmetry._,width=302 ] each individual decay mode has its own combination of @xmath110 and @xmath10 quarks to be substituted in this diagram .
furthermore , the coefficients @xmath174 originate from the quark content of @xmath150 . naturally , the hadronic amplitude attributed to the quark diagram in fig . [
fig : kstkpi ] is a nonperturbative , long - distance object and can not be directly calculated in qcd .
however all we need are the amplitude relations between individual decay modes and not the value of the amplitude itself . squaring the relations ( [ su2rel ] ) and taking into account the phase space ,
one predicts the ratios of branching fractions confirmed by the experimental values given in @xcite .
another flavour symmetry , widely used in hadron phenomenology , is @xmath175 corresponding to the limit of qcd with all three quarks , @xmath110,@xmath10 and @xmath11 , having equal masses . since in reality
@xmath176 is only inessentially smaller than @xmath177 , the magnitude of @xmath175-violation in hadrons is not universal , depending on their quark content and quantum numbers .
the @xmath175 symmetry is very helpful in `` organizing '' the spectra of strange and nonstrange hadrons in multiplets .
the light - quark meson multiplets are obtained by combining quark flavour - triplets and antiquark flavour - antitriplets . without using the specific rules of @xmath73-algebra , which can be found in many textbooks , it is easy to figure out that the nine quark - antiquark states split into a singlet @xmath184 and octet @xmath185 .
the octet has its own isospin substructure .
is a subgroup of @xmath175 . ]
the singlet - octet pattern provides a reasonable description for pseudoscalar mesons , in particular , @xmath186 meson is close to the @xmath175 singlet state ( [ etaprime ] ) .
the isotriplet of pions ( @xmath187 ) , two isodoublets of kaons ( @xmath188 and @xmath189 ) and the isosinglet @xmath190 , given by ( [ eta ] ) , together form an octet .
however , this pattern is not universal .
e.g. , in the case of vector mesons , @xmath191 and @xmath192 states in ( [ omega ] ) are neither octets , nor singlets . to complete the counting of meson @xmath175 multiplets
, one has to mention also triplets and antitriplets of heavy - light mesons .
for example , in the case of @xmath193 quark , @xmath194 ( @xmath195 ) form a triplet ( antitriplet ) . generally , @xmath175 works quite well for baryons , because their characteristic mass scale is a few times larger than @xmath177 .
importantly , the spin - parity of the lowest baryon @xmath175 multiplets is fixed , due to the total antisymmetry of the baryon `` wave - function '' in the @xmath175/spin / colour coordinates required by fermi - statistics .
there is an octet with @xmath196 ( including proton and neutron ) and decouplet with @xmath197 .
let us , for instance , have a look at the latter .
it contains the isoquadruplet ( @xmath198 ) of @xmath199-resonances ( @xmath200 , @xmath201 , @xmath202 , @xmath203 ) , the isotriplet of @xmath204 resonances ( @xmath205 , @xmath206 , @xmath207 ) , the isodoublet of @xmath208-resonances ( @xmath209 , @xmath210 ) and the isosinglet @xmath211 .
consulting @xcite for the masses of these baryons , one notices a distinct hierarchy : each constituent @xmath11 quark adds an amount of @xmath212 to the baryon mass .
returning to the quark diagram in fig .
[ fig : kstkpi ] , we may now replace the @xmath11 quark by @xmath110 or @xmath10 quark . in @xmath175 limit the new diagram obtained after this replacement is equal to the one with @xmath11-quark , yielding relations between the @xmath213 and @xmath214 hadronic amplitudes , e.g. @xmath215 such relations are typically violated at the level of 20 - 30 % , but are still useful from the phenomenological point of view . to complete our survey of flavour symmeties , we now turn to the @xmath216 quark sector of qcd lagrangian .
the fact that @xmath217 , together with the flavour - independence of quark - gluon interactions , allows one to consider an interesting limit of @xmath77 where both @xmath193 and @xmath218 quarks have infinitely heavy mass : @xmath219 at first sight , the limit is not justified , because in reality @xmath220 is substantially larger than @xmath221 . as we shall see , the fact that both masses are large turns out to be more important for qcd dynamics .
formally , in the limit ( [ hlimit ] ) one can introduce a doublet of heavy - flavour fields @xmath222 and rewrite the lagrangian in a form invariant with respect to su(2)-rotations in the @xmath223 flavour space : @xmath224 this particular form of the heavy - quark limit for @xmath77 is however not convenient , because the heavy mass scale @xmath225 is still present explicitly . to understand why it is desirable to effectively remove that scale , let us consider the heavy - quark limit ( [ hlimit ] ) for @xmath226 or @xmath227 meson . since @xmath221 and @xmath220 have no direct relation to qcd , it makes sense to discuss a hypothetical heavy - light meson @xmath228 with a mass @xmath229 containing a heavy quark with an arbitrarily large mass @xmath225 . in the rest frame of @xmath228
the constituent heavy quark stays almost at rest , providing a static source of color charge which emits and absorbs gluon fields .
the meson mass , to a good approximation is @xmath230 where @xmath231 is the energy of the light quark - gluon `` cloud '' surrounding the heavy quark .
the situation very much resembles the hydrogen atom where the total mass of the atom is a sum of an extremely large ( @xmath232 gev ) proton mass and a small energy of the electron cloud ( @xmath233 mev ) .
the essential point is that the electron itself is nonrelativistic .
one can isolate the electron mass from the rest of the energy , introduce the coulomb potential and kinetic energy , and eventually solve the equations of motion , determining the electron energy levels . in heavy hadrons
the light - quark cloud is purely relativistic ( @xmath234 ) and has a complicated long - distance nature .
nevertheless , one essential feature is common for both bound states . in the atom the energy of the electron cloud
does not depend on the proton mass .
likewise , in the heavy - light meson @xmath231 in ( [ massh ] ) is ( up to @xmath235 corrections ) independent of @xmath225 . from atomic physics we know that the electron energy levels in hydrogen and deuterium coincide to a great precision .
the fact that deuteron is twice more massive than proton , does not play a role for the energy levels , because in both cases the atomic nuclei are static .
important is that the electric charge does not change by switching from proton to deuteron .
similarly , in the @xmath228 meson @xmath231 changes very little if one replaces @xmath225 by @xmath220 or by @xmath221 , because the colour charge of the heavy quark does not change .
thus , the heavy - flavour symmetry is in reality the symmetry between the light - quark remnants of the heavy hadrons , so that the heavy - quark mass scale indeed plays a secondary role . to achieve a quantitative level , a special formalism of _ heavy quark effective theory _ ( hqet ) was developed for applications of qcd to heavy - light hadrons .
one starts from @xmath77 and introduces transformations which decouple the @xmath236 part of the heavy - quark field from the part which has the remnant momentum @xmath237 .
only the latter part strongly interacting with the light quark - gluon cloud is relevant for qcd dynamics . in hqet one
integrates out the heavy degrees of freedom and works with the lagrangian containing a new effective quark field carrying the flavour of @xmath238 but no mass .
not only the static limit ( [ hlimit ] ) but also an expansion in powers of @xmath235 can be systematically treated .
the field - theoretical aspects of heavy - mass expansion are nicely explained in the literature ( see e.g. , @xcite,@xcite ) ; i will only focus on some important phenomenological consequences of heavy - flavour symmetry .
one famous example is the @xmath239 decay involving weak @xmath240 transition in standard model .
the unknown part of the decay amplitude is the hadronic matrix element @xmath241 determined by the long - distance interactions involving the initial and final heavy quarks as well as the surrounding light quark - gluon `` cloud '' .
one chooses a special kinematical configuration , the `` zero recoil point '' where the momentum transfer to the lepton pair is equal to @xmath242 in the @xmath227 meson rest system @xmath243 this point correspond to the final @xmath226 meson at rest . in the heavy - quark limit
the replacement of @xmath218 quark by @xmath193-quark does not change the hadronic state : @xmath244 and the matrix element ( [ zerorec ] ) reduces to a trivial normalization factor .
one can therefore predict the decay amplitude in the zero recoil point up to @xmath235 corrections .
in fact there is a theorem stating that these corrections are even smaller and start from @xmath245 , but to derive this and other important details one needs a full - scale hqet framework . without resorting to the effective theory ,
it is possible to understand the origin of another important symmetry emerging in the heavy - quark limit . in the hydrogen atom ,
the electron and proton have magnetic moments related to their spins and yielding interactions with the external magnetic fields or with each other ( spin - spin interactions ) .
the magnetic moments are inversely proportional to the masses , so that the proton magnetic moment plays no role for the electron energy levels . each level is degenerate with respect to the proton spin direction .
since qed and qcd have very similar vector boson interactions with spin 1/2 particles , the spin @xmath246 quarks also have _ chromomagnetic _ moments . and interact with the `` magnetic '' parts of gluonic fields and with other quarks . for the heavy nonrelativistic quark the chromomagnetic moment is inversely proportional to the heavy quark mass @xmath225 . in the infinite mass limit the interaction vanishes and ,
hence the light - cloud energy @xmath231 is independent of the spin orientation of the heavy quark .
one arrives at a new classification of heavy - light states based on this _ heavy - quark spin symmetry_. instead of adding together the orbital momentum and the total spin of quarks as we did in ( [ spin ] ) it is more appropriate to introduce , for a @xmath247 meson ( @xmath248 ; @xmath249 ) , the total angular momentum of light degrees of freedom : @xmath250 adding the heavy quark spin @xmath251 to @xmath252 one gets degenerate doublets of heavy - light mesons with total angular momentum @xmath253 . at @xmath254
one simply has @xmath255 and therefore a doublet of mesons with @xmath100 and @xmath99 consisting of @xmath227 and @xmath256 ( @xmath226 and @xmath257 ) in the @xmath218 quark ( charm ) sector .
the mass differences within doublets are indeed very small @xcite : @xmath258 indicating that the heavy - quark spin symmetry works quite well , especially for the heavier @xmath218 quark .
taking into account that the mass differences are @xmath259 effects , one expects that @xmath260 which is also in accordance with ( [ diffm ] ) and ( [ qmasses ] ) .
i leave as an exercise to show that at @xmath261 there are two degenerate doublets : one with @xmath262 and another one with @xmath263
. the colour confinement principle does not exclude hadronic states with an `` exotic '' valence quark content , different from @xmath264 or @xmath265 .
quarks , antiquarks and gluons can be added together in any combination , e.g. , @xmath266 , @xmath267 , @xmath268 or @xmath269 , provided they are in a colour - neutral state . since one can not calculate the spectrum of hadrons in qcd with a good precision , predictions of exotic states are generally model - dependent .
it is always problematic to distinguish an exotic hadron from the excitation of an ordinary hadron with the same @xmath97 and flavour quantum numbers .
moreover , in this case one expects mixing between ordinary and exotic hadrons .
for example , if there is a @xmath99 state composed of two gluons @xmath267 ( _ glueball _ ) , it should be mixed with @xmath186 to a certain degree , so that @xmath186 acquires a glueball component . therefore , the most interesting , `` smoking gun '' signatures are the hadrons with exotic quantum numbers ( flavour content and/or @xmath270 ) , forbidden for quark - antiquark mesons or three - quark baryons . for example
it is impossible to arrange a flavour - neutral quark - antiquark state with @xmath271 . the @xmath272 and @xmath273 ( charge - conjugation ) parities of a fermion - antifermion state are determined by the rules : @xmath274 and @xmath275 , so that @xmath276 means @xmath277 .
hence , the only possibility to have @xmath278 is @xmath279 .
however , adding together even values of @xmath280 and @xmath281 one can not get @xmath282 . on the other hand ,
adding one constituent gluon to a @xmath264 pair , one easily makes a `` hybrid '' meson with @xmath283 content and @xmath271 quantum numbers .
searches for hybrid mesons are currently being carried out , but the experimental situation is not yet settled .
the recently observed narrow baryon resonance @xmath284 decaying to @xmath285@xcite is another promising candidate for hadron exotics , a state @xmath286 with five valence constituents ( _ pentaquark _ ) .
flavour symmetries are important model - independent tools to confirm / reject the experimental candidates for exotic resonances . in particular , an important task is to find the symmetry partners of these hadrons and to fill the relevant isospin and @xmath175 multiplets ( in case of the pentaquark it is actually the @xmath175 antidecouplet ) .
in this lecture we return to the quark - gluon interactions at large momentum transfers ( short distances ) . in this region , practically at @xmath287 gev , @xmath288 is small , allowing one to apply the perturbative expansion .
it is then possible to test qcd quantitatively , calculating various quark - gluon interaction processes at large @xmath238 and comparing the results with the available experimental data .
note however , that the traditional way to study interactions by scattering one object on the other is not applicable to quarks and gluons .
they simply are not available in free - particle states .
one needs to trace quarks inside hadrons , where the long - distance forces are important .
take as an example the elastic pion - proton scattering at large momentum transfers ( see fig . [ fig : pin ] ) .
here one has to combine the perturbative quark - quark scattering amplitudes ( two - gluon exchange ) with the `` wave functions '' of quarks inside the initial and final hadrons . to obtain these functions one needs to go beyond perturbative qcd
therefore , an unambiguous extraction of the perturbative amplitude from the data on the scattering cross section is not a realistic task .
the situation is not so hopeless , actually , since we have at our disposal electroweak bosons ( @xmath289 ) interacting with quarks in a pointlike way .
electroweak interactions at large momentum transfer serve as external probes of short - distance dynamics . in lecture 1 , we already discussed one example : the quark - antiquark pair production in @xmath9 decay . to list all possible electroweak sources of quarks in a more systematic way , i start with the photon .
its interaction with the quark e.m .
current was already given in ( [ lqed ] ) , let me write it down again : @xmath290 introducing a compact notation for the quark e.m .
current : @xmath291 where the summation over colour indices is not shown for brevity .
the quark weak current entering the quark-@xmath292 flavour - changing interaction : @xmath293 is more complicated and includes the ckm mixing matrix : @xmath294 finally , the quark - z interaction is @xmath295 where the quark electroweak neutral current is a mixture of vector and axial - vector parts . _ , width=302
] in @xmath297 annihilation at high energies the virtual photon provides a short - distance source of quark - antiquark pairs .
this process is hadron - free in the initial state .
the photon - exchange diagram is depicted in fig .
[ fig : eehad ] ( for simplicity , i ignore the additional @xmath9-exchange diagram ) .
the experimentally measured total cross section @xmath298 depends on one kinematical variable @xmath299 .
the virtual timelike photon transfers its energy @xmath300 to the hadronic state . at very large @xmath301 and @xmath302 ,
the initial pair of quarks is produced at an average distance of @xmath303 , much smaller than the typical hadronic distance scale @xmath304 .
due to asymptotic freedom of qcd , gluonic interactions of the produced quark pair are suppressed by small @xmath288 .
hence , @xmath297 annihilation at high energies provides an almost pointlike source of quasi - free quark pairs . at long distances
the created quarks and antiquarks are inevitably converted into some hadronic state . since in the total cross section the summation
is done over all hadronic states produced at a given energy , the total probability of hadronization sums up to a unit .
hence , at @xmath305 the hadronic cross section is well approximated by the cross section of the free quark - antiquark pair production : @xmath306 summed over all quark flavours with @xmath307 .
this , so called _ parton model _
approximation for @xmath308 is confirmed by experimental data .
moreover , the majority of events saturating the cross section at high @xmath300 consists of two distinct hadronic jets originating from the initial quark pair . the way we obtained ( [ sigmatot ] ) may seem too qualitative and a bit `` hand - waving '' . in the following
, we shall derive the asymptotic cross section ( [ sigmatot ] ) in a more rigorous way . in this derivation
several important concepts will be introduced , to be used in discussing further topics covered by these lectures .
we start with the formal definition of the total cross section : @xmath309 where the sum over the final hadronic states @xmath310 includes phase - space integration and implies summation over spins ( polarizations ) .
the matrix elements @xmath311 ( in ( [ sum ] ) @xmath312 and @xmath313 ) determine the general @xmath281-matrix of the theory : @xmath314 the usual representation of @xmath281-matrix in terms of lagrangian has the time - ordered exponential form : @xmath315 \right\}\ , , \label{smatr}\ ] ] where @xmath20 includes e.m .
interactions of quarks and leptons .
furthermore , the unitarity of @xmath281 matrix is used : @xmath316 or @xmath317 from now on we consider the forward scattering @xmath318 . replacing @xmath319 and substituting ( [ s ] ) in ( [ unit1 ] ) ,
one obtains the unitarity relation for @xmath320 ( the optical theorem ) : @xmath321 to apply this universal relation to @xmath297- scattering , we take @xmath322 with four - momentum @xmath323 , so that @xmath324 .
furthermore , we choose @xmath325 restricting the set of intermediate states by hadronic states . as a result we obtain a rigorous relation between the amplitude of the forward @xmath326 scattering via hadronic intermediate states , @xmath327 , and the sum over the squared @xmath328 amplitudes . the latter sum , according to ( [ sum ] ) is proportional to the total hadronic cross section . the optical theorem ( [ tunit ] ) takes the form : @xmath329 diagrammatically , this relation is represented in fig .
[ fig : unit ] . the amplitude @xmath330 contains the photon propagators and the products of electron and positron spinors in both initial and final states , written according to the qed feynman rules .
the nontrivial part of this amplitude , denoted as @xmath331 , describes the @xmath332 transition , and is called the _ correlation function _ ( or correlator ) of quark currents .
the formal expression for this object reads : @xmath333 _ , width=453 ] due to the conservation of e.m .
current ( @xmath334 ) , the correlation function depends on one invariant amplitude : @xmath335 substituting ( [ pimunu2 ] ) in ( [ aee ] ) and taking imaginary part from both sides , we obtain @xmath336 it is convenient to normalize the hadronic cross section to the @xmath337 cross section known from qed .
one can literally repeat the derivation done above , taking instead of hadronic states the @xmath338 states : @xmath339 .
the resulting relations are quite similar to ( [ unit2 ] ) , ( [ aee ] ) and ( [ impart ] ) : @xmath340 @xmath341 and @xmath342 where the muonic correlation function @xmath343 is nothing but a 2-point muon - loop diagram .
furthermore , the cross section in ( [ unit_mu ] ) taken from qed textbooks , reads : @xmath344 dividing the hadronic unitarity relation ( [ unit2 ] ) by the muonic one ( [ unit_mu ] ) and using ( [ impart ] ) and ( [ impartmu ] ) , we obtain a useful ratio : @xmath345 corrections .
wavy lines denote external currents with 4-momentum @xmath346 , solid lines quarks and dashed lines gluons . _ , width=453 ] the next key point in our derivation is the analysis of the correlation function @xmath347 at spacelike @xmath348 . at large @xmath349 ,
the long - distance domain in the space - time integral in ( [ pimunu ] ) is suppressed by the strongly oscillating exponent and the short distances / times @xmath350 dominate .
this justifies using qcd perturbation theory with @xmath351 at @xmath352 . the leading - order asymptotically free result is given by the 2-point quark - loop diagram ( fig .
[ fig:2loop]a ) , and the next - to - leading corrections are determined by @xmath353 two - loop diagrams ( fig .
[ fig:2loop]b , c , d ) . calculation of these diagrams yields ( at @xmath354 ) : @xmath355 where each flavour @xmath346 contributes with the same expression and a coefficient @xmath356 .
the natural scale for @xmath288 is @xmath357 . in recent years , due to tremendous calculational efforts , the @xmath358 and even @xmath359 corrections to @xmath360
have been calculated ; also the loop diagrams for the massive quark are known with a high accuracy . in the final stage of our derivation the calculated function @xmath347 at @xmath361
is related to @xmath362 at positive @xmath11 .
one employs cauchy theorem for the function @xmath363 obtained from @xmath347 by analytically continuing the real variable @xmath357 to the complex values , @xmath364 : @xmath365 the integration contour c is shown in fig .
[ fig : cont ] .
it circumvents the singularities of the function @xmath363 , i.e. the points / regions where @xmath366 . according to ( [ unit2 ] ) and ( [ impart ] ) the location of singularities at real @xmath367
is determined by the masses of resonances and/or the thresholds of multiparticle hadronic states produced in @xmath308 ; the lowest one is at @xmath368 corresponding to the threshold of the lightest two - pion state .
subdividing the contour @xmath273 into : 1 ) a large circle with the radius @xmath369 , 2 ) an infinitely small semicircle @xmath370 surrounding @xmath371 and 3 ) two straight lines from @xmath371 to @xmath369 , we can rewrite the integral in terms of three separate contributions : @xmath372 suppose the function decreases at @xmath373 : @xmath374 , where @xmath375 .
then the first integral vanishes at @xmath376 .
taking infinitely small semicircle , one makes the third integral also vanishing .
furthermore , since there are no singularities of @xmath363 at @xmath377 , the integrand in the second integral reduces to the imaginary part : @xmath378 ( due to schwartz reflection principle ) . finally , we obtain the desired _
dispersion relation _ @xmath379 with l.h.s . calculated in qcd in a form of perturbative expansion ( [ pimunuloop ] ) and r.h.s . related to the cross section via ( [ r ] ) : @xmath380 ) .
the crosses on the positive real axis indicate singularities of @xmath347 _ , width=302 ] the quantity @xmath381 is directly measurable in @xmath297 experiments .
it remains to determine @xmath382 .
knowing the answer for the quark loop diagram in fig .
[ fig:2loop]a it is very easy to write down the expression for the muon loop .
since we neglect masses in both diagrams , the only difference is the factor 3 , from summing up the colour states in the quark loop .
this factor is naturally absent for the muon loop . from ( [ pimunuloop ] )
, taking imaginary part , one obtains @xmath383 finally , to guarantee the convergence of the dispersion integral ( [ disprel ] ) , let us differentiate both parts in @xmath357 : @xmath384 where ( [ pimunuloop ] ) gives for l.h.s .
@xmath385 note that divergent and constant terms disappeared after differentiation and play no role in our derivation .
the final form of the dispersion relation is @xmath386 where ellipses denote higher - order in @xmath288 corrections .
the fact that ( [ rel2 ] ) is valid at @xmath387 unambiguously fixes the constant limit of @xmath381 at large @xmath11 : @xmath388 where @xmath389 indicates that @xmath369 includes all `` active '' quark flavours , for which the condition @xmath302 is fulfilled .
finally , we notice that ( [ rasympt ] ) coincides with the parton model prediction ( [ sigmatot ] ) , taking into account that the free - quark- and muon - pair cross sections differ only by the colour factor times the quark charge squared : @xmath390 importantly , qcd not only reproduces the parton model prediction for @xmath381 but also provides perturbative corrections , as well as predicts the integral ( [ rel2 ] ) over @xmath381 . from @xcite_,width=529 ]
the experimental data collected in various regions of @xmath300 nicely confirm ( [ rasympt ] ) .
according to fig.[fig : ree ] taken from @xcite , the ratio @xmath381 approaches first the constant value @xmath391 at energies @xmath392 gev , above the region of vector meson resonances @xmath98,@xmath191,@xmath192 ( and below the charm threshold ) .
that is exactly the value anticipated from ( [ rasympt ] ) for @xmath393 .
well above charmonium resonances , a new constant level is achieved : @xmath394 . and
finally , @xmath395 is settled at energies above @xmath396 resonances , where all five quark flavours are in their asymptotic regime . actually , the current data on @xmath381 are so precise that one should also include small @xmath288-corrections to @xmath397
. there are other similar inclusive observables calculable in qcd , among them the total widths @xmath398 and @xmath399 .
they have the same status as @xmath381 , but a fixed scale @xmath400 or @xmath401 instead of @xmath300 .
one has also to mention an interesting and well developed sub - field of perturbative qcd related to the jet and/or heavy - quark production in @xmath297 and hadron collisions at high energies .
the underlying short - distance quark - gluon processes are successfully traced in the experimentally observed multijet structure of the final state .
naturally , hadrons can not be completely avoided , because , after all , quarks and gluons hadronize . in fact , hadronization in jet physics is nowadays a somewhat routine procedure described by qcd - oriented models ( e.g. the lund model integrated within pythia @xcite ) . at lower scales , @xmath402 gev ,
inclusive decays of @xmath403-lepton are among useful tools to study perturbative qcd ( see , e.g. @xcite ) . the processes of lepton - nucleon _ deep inelastic scattering _ ( dis ) , @xmath404 and @xmath405 ( @xmath406 ) , are among the best testing grounds of qcd .
the long - distance structure of the initial nucleon makes these processes more complicated than @xmath296 . for definiteness ,
let us consider the electron - nucleon scattering mediated by the virtual photon @xmath407 with the 4-momentum @xmath346 , whereas @xmath408 is the nucleon 4-momentum .
to measure the dis cross section , one only has to detect the final electron . in the nucleon rest frame
the invariant variables @xmath409 and @xmath410 are related to the initial and final energies and the scattering angle of the electron : @xmath411 , @xmath412 .
the specific kinematical region @xmath413 has to be chosen to reveal the spectacular effect of asymptotic freedom . in this region
the experimentally measured dis cross section , normalized to the cross section of the electron scattering on a free pointlike quark @xmath414 , depends essentially on the ratio @xmath415 .
this effect ( _ bjorken scaling _ ) , was first interpreted in terms of a beautiful phenomenological model suggested by feynman .
one considers the reference frame with large nucleon momentum @xmath408 , so that @xmath416 gives the longitudinal direction . neglecting long - distance binding forces between quarks
, the initial nucleon is represented as a bunch of free constituents ( _ partons _ ) : quarks , antiquarks and gluons moving in the longitudinal direction . ) for the pion . ] for simplicity the quark masses and transverse momenta as well as the nucleon mass are neglected , in comparison with @xmath417 and @xmath418 .
the electron scatters on one of the nucleon constituents ( excluding gluons , of course , because they are electrically neutral ) which has the momentum fraction @xmath419 , so that after the pointlike collision the quark 4-momentum is @xmath420 .
the massless quark has to remain on - shell , @xmath421 , therefore @xmath422 and @xmath423 the cross section is then represented as a sum of all possible `` elementary '' processes integrated over @xmath419 @xmath424 where @xmath425-function takes into account ( [ condi ] ) .
the sum goes over all quark and antiquark species inside proton , and @xmath426 are the _ parton distributions _ defined as the probabilities to find the @xmath427-th constituent with the momentum fraction @xmath419 . in qcd ,
the scaling behavoiur ( [ sigmadis ] ) corresponds to the asymptotic - freedom approximation ( @xmath428 ) .
switching on the perturbative quark - gluon interactions one predicts logarithmic corrections @xmath429 to this formula which are nicely reproduced by experiment .
there are many excellent reviews and lectures where dis in perturbative qcd are discussed ( see e.g.@xcite ) . here
we shall concentrate on one essential aspect : separation of long- and short - distance effects .
let us approach dis from the quantum - field theory side , as we did above for @xmath430 .
omitting for simplicity the initial and final electrons , one can represent the hadronic part of the dis cross section in a form of @xmath431 cross section .
employing unitarity relation ( [ tunit ] ) with @xmath432 one is able to relate the dis cross section to the @xmath433 forward - scattering amplitude ( see fig . [
fig : dis ] ) : @xmath434 the amplitude @xmath435 contains the photon polarization vectors multiplied by a new purely hadronic object @xmath436 which resembles the correlation function of two currents we introduced above , but , instead of vacuum , has nucleons in the initial and final states . one can prove that at large @xmath417 and @xmath418 the dominant contribution to the above integral stems from small space - time intervals @xmath437 .
is a more general condition than the small - distance / time dominance @xmath438 which takes place in @xmath430 . ] in other words , the points of absorption and emission of the virtual photon are located close to the light - cone @xmath439 .
the process takes place in the asymptotic freedom regime , that is , a single quark absorbs the photon , and penetrates quasi - freely at small @xmath440 before emitting the photon .
all other contributions , for example , with different quarks emitting and absorbing initial and final photons are suppressed by powers of @xmath441 . to describe the free - quark propagation from @xmath24 to
0 we use the quark propagator @xmath442 neglecting the quark mass . substituting the e.m .
currents in ( [ amplit ] ) in terms of quark fields , contracting the fields of the propagating quark and using ( [ prop1 ] ) we obtain @xmath443 in this expression , where only the leading term is shown , the calculable short - distance part ( the quark propagator ) is separated from the long - distance part which is represented by the quark - antiquark matrix element taken between nucleon states .
this hadronic matrix element is a complicated object which has to be either determined from experiment or calculated using methods beyond perturbative qcd which will be discussed in the next two lectures .
i skip the derivation of the cross section formula from the imaginary part of @xmath444 which allows to relate the matrix element introduced in ( [ tmunu1 ] ) with the parton distributions .
also @xmath288 corrections can be systematically calculated ; they contain important and observable @xmath445 effects .
important is that the long - distance matrix element ( or parton distributions ) are universal characteristics of the nucleon and they do not change if the short - distance part of the process changes ( e.g. , by switching to neutrino - nucleon scattering @xmath446 where the charmed quark is produced ) . to summarize , in dis it is possible to separate the short - distance domain by choosing appropriate kinematical region and defining convenient physical observables .
the short - distance quark - gluon interactions are calculable in a form of perturbation theory in @xmath288 , whereas the long - distance part is parameterized in terms of universal hadronic matrix elements .
the procedure of approximating the product of currents in ( [ tmunu1 ] ) by a quark - antiquark operator and separating short and long distances is called _ operator - product expansion _ ( ope ) . implicitly or explicitly
, ope is used in almost any perturbative qcd treatment of hadronic processes .
qcd at short distances does not essentially help in understanding the long - distance dynamics of quarks and gluons . from the short - distance side we only know that the running coupling @xmath447 increases at low - momentum scales and
eventually diverges at @xmath448 ( see fig . [
fig : alphaslow ] ) . is the growth of @xmath288 the only dynamical reason for confinement ?
it is not possible to answer this question remaining within the perturbative qcd framework , because the language of feynman diagrams with propagators and vertices is not applicable already at @xmath449 .
qcd in nonperturbative regime is currently being studied using other methods , first of all , the lattice simulations . from these studies , there is a growing confidence that long - distance dynamics is closely related to the nontrivial properties of the physical vacuum in qcd . for a given dynamical system ,
vacuum is a state with the minimal possible energy .
evidently , in qcd the vacuum state contains no hadrons , because creating any hadron always costs a certain amount of energy . in that sense ,
qcd vacuum has to be identified with the @xmath450 state in the correlation function ( [ pimunu ] ) .
given that the vacuum state contains no hadrons does not yet mean that it is completely empty
. there could be quantum fluctuations of quark and gluon fields with nonvanishing densities .
the existence of vacuum fields is manifested , e.g. by _ instantons _ , special solutions of qcd equations of motion having a form of localized dense gluonic fields ( for an introductionary review on instantons see , e.g. , @xcite ) .
lattice qcd provides an independent evidence for quark / gluon fields in the vacuum . without going into further theoretical details ,
i will rather concentrate on the phenomenology of vacuum fields in qcd .
we shall see that properties of hadrons are influenced by the existence of quark and gluon vacuum fluctuations with nonvanishing average densities , the so called _ vacuum condensates_. formally , in the presence of vacuum fields , the matrix elements of quark and gluon field operators between the initial @xmath451 and final @xmath452 states are different from zero . the combinations of fields have to obey lorentz - invariance , colour gauge symmetry and flavour conservation , so that the simplest allowed composite operators are @xmath453 \left[\bar{\psi}_{q'\,j}(\gamma^a)^j_l \psi^l_{q'}\right]\ , , \label{vacuum}\end{aligned}\ ] ] where @xmath454 $ ] and @xmath455 are various combinations of lorentz- and colour matrices . the indices at @xmath456 reflect their dimension @xmath10 in gev units .
furthermore , if the operators are taken at different 4-points , care should be taken of the local gauge invariance .
for instance , the quark - antiquark nonlocal matrix element has the following form : @xmath457\psi_q(0)\mid 0\rangle\ , , \label{qqbarvac}\ ] ] where @xmath458=\exp\left[ig_s\int_0 ^ 1 dv x^\mu a^a_\mu(vx)(\lambda^a/2)\right ] $ ] is the so - called gauge factor .
only the matrix elements with the light quarks @xmath249 are relevant for the nonperturbative long - distance dynamics .
a pair of heavy @xmath193 ( @xmath218 ) quarks can be created in vacuum only at short distances / times of @xmath459 ( @xmath460 ) , that is , perturbatively . without fully solving qcd ,
very little could be said about vacuum fields , in particular about their fluctuations at long distances which have typical `` wavelengths '' of @xmath461 .
thus , we are not able to calculate the matrix element ( [ qqbarvac ] ) explicitly , as a function of @xmath24 .
it is still possible to investigate the vacuum phenomena in qcd applying certain approximations .
one possibility is to study the average local densities .
the vacuum average of the product of quark and antiquark fields , @xmath462 corresponds to the @xmath463 limit of the matrix element ( [ qqbarvac ] ) .
the simplest vacuum density of gluon fields is @xmath464 due to translational invariance , both @xmath465 and @xmath466 are independent of the 4-coordinate .
these universal parameters are usually called the densities of quark and gluon condensates , respectively . as we shall see in the following subsection
, the nonvanishing quark condensate drastically influences the symmetry properties of qcd .
let us return to the isospin symmetry limit ( [ qcdisosp ] ) of the qcd lagrangian : @xmath467 since @xmath468 , a reasonable approximation is to put @xmath469 , so that @xmath110- and @xmath10-quark components of the @xmath470-doublet become massless .
each dirac spinor can be decomposed into the left- and right - handed components , @xmath471 where , by definition , the left - handed ( right - handed ) quark has an antiparallel ( parallel ) spin projection on its @xmath472-momentum .
similarly , for the conjugated fields one has : @xmath473 rewriting @xmath470 in terms of the left- and right - handed components we obtain the following decomposition of the lagrangian ( [ qcdchiral ] ) in the massless limit : @xmath474 the quark - gluon interaction term in @xmath118 is now split into two parts , @xmath475 and @xmath476 , so that quarks conserve their _ chirality _ ( left- or right - handedness ) after emitting /absorbing an arbitrary number of gluons .
importantly , also the interaction of quarks with photons has the same property ( take the e.m .
interaction ( [ em1 ] ) and decompose it into @xmath280 and @xmath369 parts in the same way as above ) . in the massless
( _ chiral _ ) limit , quarks of left- and right chiralities propagate and interact independently from each other .
in fact , it is possible to introduce two independent isospin @xmath477 transformations , separately for @xmath280 and @xmath369 fields : @xmath478 restoring the mass in @xmath118 leads to the violation of chiral symmetry .
the lagrangian mass term can be represented as an effective transition between left- and right - handed quarks : @xmath479 having in mind the smallness of the @xmath114-quark masses , one naturally expects that an approximate chiral symmetry manifests itself in the observable properties of hadrons . in reality , the symmetry is violated quite substantially , as the two following examples demonstrate . the first one is a real hadronic process : the photon scattering on a polarized ( e.g. left - handed ) proton .
one of the diagrams is shown in fig .
( [ fig : compt ] ) .
this process is a very complicated mixture of quark - photon and quark - gluon interactions at long distances , determined by the quark structure of the proton .
importantly , all these interactions obey an approximate chiral symmetry at the lagrangian level .
hence , the amplitude of the proton - chirality flip is expected to be very small , @xmath480 of the total scattering amplitude . to check this conjecture ,
let me consider the case when the initial photon energy is much smaller than the proton mass . in this case ,
it is possible to approximate the proton with a point - like particle .
the @xmath481 amplitude is then simply obtained from the @xmath482 amplitude in qed ( compton effect ) , replacing the electron spinors and propagators by the proton ones : @xmath483 u_l(p_1)\epsilon_2^\alpha\epsilon^\beta_1 , \label{compt}\end{aligned}\ ] ] where @xmath484 are the polarization vectors of the initial and final photons with the 4-momenta @xmath485 and @xmath486 , respectively . without even completing the calculation ,
we notice that the @xmath487 terms in the amplitude flip the initial proton chirality @xmath488 , whereas the @xmath489 terms preserve chirality .
importantly , the contributions of both types are of the same order , determined by the scale @xmath490 , indicating that chiral symmetry for the photon - proton scattering is broken at 100% level . to present the second example of the chiral symmetry violation , i start from the correlation function @xmath491 very similar to the one introduced in lecture 3 , but containing a slightly different quark current : @xmath492 which produces the @xmath147 and @xmath100 quark - antiquark states .
note that this current is conserved , @xmath493 , even if @xmath494 .
we then follow the same derivation as in lecture 3 and obtain the dispersion relation ( [ disprel ] ) for @xmath347 .
the only change is in the imaginary part ( [ impi ] ) , where now only the states with @xmath147 contribute to the total cross section ( @xmath98 meson and its radial excitations , the two - pion state and other states with an even number of pions ) , so that the ratio @xmath381 has to be replaced by @xmath495 in parallel , we consider the correlation function @xmath496 of two axial - vector currents with @xmath147 : @xmath497 this current is conserved only in the chiral symmetry ( @xmath498 ) limit : @xmath499 decomposing the correlation function ( [ corra ] ) in two tensor structures : @xmath500 we notice that in the chiral limit there is only one independent invariant amplitude : @xmath501 the dispersion relation for @xmath502 has the same form as ( [ disprel ] ) : @xmath503 where @xmath504 is the corresponding threshold . in order to determine the imaginary part via relation similar to ( [ impi ] ) , we introduce a slightly artificial cross section @xmath505 of hadron production mediated by @xmath9 boson coupled to the axial - vector current
. then , @xmath506 where @xmath507 both @xmath347 and @xmath508 can be calculated at large @xmath509 from the same 2-point quark - loop diagrams shown in fig .
[ fig:2loop ] . in the chiral limit ,
the only difference is in two extra @xmath510 matrices present in @xmath511 .
hence , in perturbative qcd @xmath512 at any order in @xmath288 .
this equation is trivial for the leading - order loop diagrams ( fig .
[ fig:2loop]a ) . using dirac algebra
, it is easy to check that the extra @xmath510 matrices cancel each other ( @xmath513 ) in the absence of masses in the propagators .
furthermore , each gluon line inserted in the loop brings two more @xmath7-matrices which do not influence that cancellation . from ( [ eqp ] ) follows the equation of two dispersion relations ( [ disprel ] ) and ( [ dispa ] ) . to keep the dispersion integrals convergent ,
we differentiate them @xmath514 times at a certain @xmath348 .
the result is : @xmath515 note that even in the presence of quark masses the corrections to ( [ eqp ] ) and to ( [ eqdisp ] ) are very small , @xmath516 .
if @xmath357 is not very large , @xmath517 gev@xmath518 , the integrals in ( [ eqdisp ] ) are dominated by the contributions of the low - mass hadronic states to the corresponding @xmath308 cross sections . in the case of vector current ,
the states are @xmath98-meson and its radial excitations .
hence , @xmath381 represents a resonance curve with the peaks located at @xmath519 , @xmath520 etc .
( see fig .
[ fig : spectra ] ) .
the validity of ( [ eqdisp ] ) at arbitrary ( but large ) @xmath348 implies that also @xmath521 .
we then expect the spectrum of resonances generated by the axial - vector current to resemble , in gross features , the @xmath98-spectrum .
however , experimental data @xcite reveal a completely different picture .
the lowest resonances in the axial - vector channel are the pion with @xmath99 , the axial meson @xmath522 and the radial excitation @xmath523 .
._,width=302 ] summarizing , there is a clear evidence , based on the observed properties of hadrons , that chiral symmetry in qcd is violated much stronger than it is expected from @xmath77 .
an additional source of the chiral symmetry violation is provided by the quark condensate .
decomposing the quark and antiquark fields in ( [ qqbar ] ) in the left - handed and right - handed components , @xmath524 we realize that condensate causes vacuum transitions between quarks of different chiralities . hence , in qcd one encounters a _
spontaneous broken _ chiral symmetry , a specific situation when the interaction ( in this case @xmath77 ) obeys the symmetry ( up to the small @xmath525 corrections ) , whereas the lowest - energy state ( qcd vacuum ) violates it .
we conclude that in order to correctly reproduce the properties of hadrons and hadronic amplitudes ( e.g. , correlation functions ) , one has to take into account the vacuum fields , in particular , the quark condensate .
let us return to the correlation functions ( [ corrv ] ) and ( [ corra ] ) , restoring nonzero @xmath110 and @xmath10 quark masses , that is , working with full qcd .
the vacuum quark - gluon fields generate new contributions to @xmath360 or @xmath502 .
in addition to the perturbative loop diagrams in fig .
[ fig:2loop ] , there are diagrams shown in fig .
[ fig : cond ] , where gluons , quarks and antiquarks penetrate to long distances , being absorbed and emitted by vacuum fluctuations .
the vacuum fields have characteristic momenta of @xmath159 . therefore ,
if the momentum scale in the correlator is large , @xmath526 , one can approximate the vacuum state by a set of constant fields . in other words , the virtual quark propagating at short distances / times between the points @xmath24 and @xmath527 ,
can not `` resolve '' the long - distance fluctuations of vacuum fields and percepts them in a form of averaged static fields .
this approximation makes the calculation of diagrams in fig .
[ fig : cond ] straightforward .
in addition to feynman rules of perturbative qcd for virtual quarks and gluons , one has to form all possible combinations ( [ vacuum ] ) of vacuum fields and replace them by the corresponding condensate densities , so that the 4-momenta of the crossed lines on these diagrams are neglected . for example , the product of quark and antiquark vacuum fields in fig .
[ fig : cond ] d has to be replaced by @xmath465 . more details on these calculations can be found , e.g. , in the review @xcite .
the result for the vector - current correlator has the following schematic form : @xmath528 where the first term on r.h.s .
corresponds to the perturbative diagrams in fig .
[ fig:2loop ] , whereas the sum contains the contributions of vacuum condensates with dimensions @xmath10 obtained from the diagrams in fig .
[ fig : cond ] . to compensate the growing dimension of the operators
@xmath529 , the coefficients @xmath530 contain increasing powers of @xmath531 .
thus , the condensate contributions die away at @xmath532 and do not alter the perturbative asymptotics of the correlator given by @xmath533 .
furthermore , as soon as we work at relatively large @xmath534 , it is possible to retain only a few first terms in the sum , that is , neglect diagrams with more than 3 - 4 vacuum fields emitted from the virtual quarks in the correlation function . for the axial - vector correlation function
@xmath508 one obtains an expression similar to ( [ ope ] ) with the same perturbative part ( up to very small corrections of @xmath535 ) , but with different coefficients @xmath530 at certain condensate terms . the most important deviation from the vector - current case is in the value and sign of @xmath536 , i.e. , in the 4-quark condensate terms .
thus , addition of condensate effects leads to an explicit violation of ( [ eqp ] ) .
it is then not surprising that hadron resonances contributing to @xmath381 and @xmath537 in ( [ eqdisp ] ) are different .
furthermore , the method of correlation functions allows to reproduce an important relation for the pion mass , explaining the smallness of @xmath538 .
note that from the point of view of naive quark model @xmath539 and @xmath98 mesons differ only by orientations of quark spins .
why is then @xmath540 and , moreover , @xmath541 ?
we consider the correlation function similar to ( [ corra ] ) , but for simplicity , containing charged axial currents : @xmath542 where @xmath543 .
this current is the part of the standard model weak current ( [ jweak ] ) , responsible for @xmath544 transition , e.g. the @xmath545 decay . the hadronic matrix element which determines this decay , @xmath546 is parameterized via the pion decay constant @xmath547 which plays an essential role in our analysis .
it is convenient to multiply ( [ corra1 ] ) by @xmath548 , forming a combination of invariant amplitudes : @xmath549 note that @xmath550 at @xmath498 and , in particular , the perturbative part of @xmath551 vanishes as @xmath535 .
concerning nonperturbative part , the only first - order in @xmath552 contribution is given by the quark condensate diagram in fig .
[ fig : cond]d @xmath553 to proceed , we use for @xmath551 the dispersion relation of the type ( [ dispa ] ) . to obtain the imaginary part
, one has to return to the unitarity relation ( [ tunit ] ) , identifying @xmath554 with @xmath555 and @xmath320 with the correlation function .
the result is : @xmath556 importantly , only pseudoscalar states contribute to the above sum , because the matrix elements for the axial - vector mesons @xmath557 and its excitations vanish , being proportional to the transverse polarization vectors of these mesons : @xmath558 using ( [ unitpi ] ) and the definition ( [ fpi ] ) we obtain the following expression for the dispersion relation : @xmath559 where the one - pion state contribution is shown explicitly ( with @xmath560 ) including the phase space proportional to @xmath425-function .
ellipses denote excited pions and multiparticle states with the same quantum numbers . integrating out @xmath425-function and substituting ( [ pi5tilde ] ) in l.h.s .
we obtain @xmath561 where the sum over higher states is also shown schematically .
to fulfil this equation at large @xmath357 one has to demand that @xmath562 .
simultaneously , the decay constants of excited states have to obey : @xmath563 , otherwise the @xmath357 asymptotics of l.h.s and r.h.s . in ( [ finalpi ] )
is different at @xmath525 .
we then reproduce the well known gell - mann - oakes - renner relation @xmath564 this relation reflects the special nature of pion in qcd .
the anomalously small pion mass is not accidental and is closely related to the spontaneous chiral symmetry breaking via condensate . if a symmetry in quantum field theory is broken spontaneously , there should be massless states ( nambu - goldstone particles ) , one per each degree of freedom of broken symmetry .
the three pions , @xmath565 play a role of massless nambu - goldstone particles in qcd . in other words , due to the specific structure of qcd vacuum fields , the amount of energy needed to produce a pion state tends to zero .
the fact that pions still have small nonvanishing masses is due to the explicit violation of chiral symmetry via @xmath114 quark masses .
how large is the quark condensate density ?
using @xmath566 mev , @xmath567 mev @xcite and taking the @xmath114 quark mass values from ( [ qmasses ] ) , renormalizing them at @xmath568 gev , one obtains from ( [ gor ] ) , typically , @xmath569 where we assume isospin symmetry for the condensates @xmath570 .
not surprisingly , the estimated value is in the ballpark of @xmath177 !
being not a measurable physical quantity , the condensate density is a scale - dependent parameter . since r.h.s . of ( [ gor ] )
is determined by hadronic parameters @xmath571 , which are both scale - independent , the running of the quark condensate should compensate the running of the quark mass given in ( [ runmass ] ) , that is : @xmath572 the gluon condensate density is another important characteristics of nonperturbative qcd .
this parameter can not be easily estimated from correlation functions with light quarks , because the latter are dominated by quark condensates .
a very useful object , sensitive to the gluon condensate is the correlation function of @xmath193-quark currents : @xmath573 where @xmath574 is the @xmath193-quark part of the quark e.m .
current @xmath575 .
following the same derivation as in lecture 3 , we write down the dispersion relation for @xmath576 relating im@xmath577 with the ratio @xmath578 defined as : @xmath579 where the cross section @xmath580 includes hadronic states with @xmath105 content produced in @xmath297 : charmonium resonances ( @xmath581 ) , pairs of charmed hadrons etc .
in addition to the perturbative @xmath193-quark loop diagrams in fig [ fig:2loop ] , one has to include also the contribution of the diagrams shown in fig [ fig : cond]a , b , c , with @xmath193 quarks emitting vacuum gluons .
( remember that nonperturbative quark condensate for the heavy @xmath193 quark is absent . )
the resulting relation has the following form : @xmath582 where @xmath583 is the mass of the lowest @xmath584 state in this channel .
the function @xmath585 evaluated from the massive @xmath193 -quark loop diagrams has a more complicated form than in the massless case .
the function @xmath586 is the calculable short - distance part of the diagrams in fig .
[ fig : cond]a , b , c . to achieve a better convergence at @xmath587 , the dispersion relation
is usually differentiated @xmath514 times .
note that in this case the point @xmath588 is also accessible : @xmath193 quarks are still highly virtual at @xmath588 because the long - distance region starts at @xmath589 .
the set of power moments obtained from ( [ rcdisp ] ) with @xmath358 accuracy has a form @xmath590\ , , \label{moments}\ ] ] with calculable coefficients @xmath591 .
the natural scale for @xmath288 in the perturbative loops is in this case the virtuality @xmath592 .
the moments ( [ moments ] ) are used to extract the gluon condensate density and , simultaneously the @xmath193 quark mass , employing the experimental data on @xmath578 on l.h.s .
the estimate of the condensate density obtained first in @xcite is : @xmath593 again within the range of @xmath177 .
the value of the gluon condensate density is usually given multiplied by @xmath288 for convenience , because this product is scale - independent .
the relation ( [ moments ] ) obtained first in @xcite is a well known example of a _ qcd sum rule_. the method developed by shifman , vainshtein and zakharov @xcite employs quark - current correlation functions calculating them in the spacelike region , including perturbative and condensate contributions .
consider for example the correlation function ( [ corrv ] ) , with the result of qcd calculation having the form ( [ ope ] ) .
note that in the latter expression short- and long - distance effects are separated .
the perturbative part @xmath533 and the coefficients @xmath594 take into account short - distance quark - gluon interactions with characteristic momenta larger than the scale @xmath238 . both @xmath585 and @xmath530 are _ process - dependent _ , i.e. , depend on the choice of the currents . on the other hand , the condensate densities absorb , in an averaged way , the long - distance interactions with momenta less than @xmath238 and are process - independent .
the universality of condensates allows to calculate correlation functions in different channels without introducing new inputs , in an almost model - independent way . )
represent another example of ope in qcd ( see lecture 3.3 ) , when a product of two currents is expanded in a set of local operators @xmath529 . the perturbative part in this case is interpreted as a unit operator with zero dimension . ] to obtain the sum rule , the qcd result for the correlation function is matched , via dispersion relation , to the sum over hadronic contributions ( the integral over hadronic cross section ) .
a twofold use of the sum rule relations is possible .
firstly , using experimental data one saturates the hadronic sum in the dispersion integral and extracts the universal qcd parameters : quark masses , @xmath288 , condensate densities .
one example is the @xmath595 sum rule ( [ moments ] ) discussed above .
secondly , having fixed qcd parameters , one calculates , with a certain accuracy , the masses and decay amplitudes of the lowest hadrons entering the dispersion integral .
the relation ( [ gor ] ) obtained from the sum rule ( [ finalpi ] ) can serve as an example .
note that ( [ finalpi ] ) is unique , because all states in the hadronic sum except the pion and all qcd terms except the quark condensate are absent at @xmath525 . to demonstrate in more detail how the method works ,
let me outline the calculation of the @xmath98-meson decay constant from the correlation function ( [ corrv ] ) .
the dispersion relation for @xmath331 can be written in the following form : @xmath596(-g_{\mu\nu}q^2+q_\mu q_\nu)\right\}_{q^2=s}\ , , \label{disprhoo}\end{aligned}\ ] ] where the @xmath98-meson contribution ( @xmath597 ) is isolated from the sum and the hadronic matrix element is substituted : @xmath598 determined by the @xmath98 decay constant @xmath599 . the integrand @xmath600 ( _ spectral density _ ) includes the sum over excited and multihadron states .
we take into account the experimental fact that the @xmath98 resonance strongly dominates in the low - energy region @xmath601 gev , so that @xmath600 practically starts from some threshold value @xmath602 gev . switching to the invariant amplitude @xmath347 and using the result ( [ ope ] ) of qcd calculation ,
one obtains from ( [ disprhoo ] ) the desired relation , a prototype of qcd sum rule : @xmath603 which is valid at sufficiently large @xmath509 . to proceed
, one applies to both sides of this equation the _ borel transformation _ defined as : @xmath604 this transformation deserves a clarifying comment . differentiating @xmath347 many times in @xmath357 means , one is effectively approaching the long - distance region . indeed , with an infinite amount of derivatives the function @xmath347 is defined at any @xmath357 , including @xmath367 .
the @xmath605 limit works in the opposite direction : one penetrates into the deep spacelike asymptotics .
combining the two transformations in ( [ borel2 ] ) at @xmath606 , one fixes the virtuality scale at @xmath607 . using the school - textbook definition of the exponent : @xmath608
, it is an easy exercise to prove that @xmath609 as a result , borel transformation exponentially suppresses the integral over @xmath600 in ( [ sumrule1 ] ) with respect to the @xmath98-meson term .
furthermore , after applying @xmath610 to r.h.s . of ( [ sumrule1 ] )
the coefficients @xmath611 contain powers of @xmath612 .
hence , at large @xmath613 it is possible to retain only a few low - dimension condensates in the sum , e.g. , at @xmath614 gev@xmath518 a reasonable approximation is to neglect all operators with @xmath615 . the explicit form of the qcd sum rule ( [ sumrule1 ] ) after borel transformation is @xcite : @xmath616\ , .
\label{svzrho}\end{aligned}\ ] ] in obtaining the above relation , the four - quark vacuum densities are factorized into a product of quark condensates .
the quark - gluon condensate has very small coefficient and is neglected .
the running coupling @xmath288 is taken at the scale @xmath617 i.e. , at the characteristic virtuality of the loop diagrams after borel transformation .
importantly , there exists a svz region @xcite of intermediate @xmath613 where the @xmath98 meson contribution alone saturates the l.h.s . of the sum rule ( [ svzrho ] ) .
meson decay constant calculated from the sum rule ( [ svzrho ] ) neglecting all excited and continuum states ( solid ) , as a function of the borel parameter , in comparison with the experimental value ( boxes ) .
the dashed curve corresponds to an improved calculation , where the sum over excited and continuum states is estimated using quark - hadron duality with a threshold @xmath618 gev@xmath518 .
_ , width=491 ] to illustrate this statement numerically , in fig .
[ fig : frho ] the experimentally measured @xmath599 ( obtained from the @xmath619 width ) is compared with the same hadronic parameter calculated from the sum rule ( [ svzrho ] ) where all contributions of excited and continuum states are neglected .
one indeed observes a good agreement in the region @xmath620 1 gev@xmath518 .
an important step to improve the sum rule ( [ svzrho ] ) is to use the _ quark - hadron duality _ approximation .
the perturbative contribution to the correlation function ( the sum of fig .
[ fig:2loop ] diagrams ) is represented in the form of a dispersion integral splitted into two parts : @xmath621 the integral over the spectral function @xmath600 in ( [ sumrule1 ] ) is approximated by the second integral over the perturbative spectral density @xmath622 in ( [ qcddual ] ) .
the latter integral is then subtracted from both parts of ( [ sumrule1 ] ) .
correspondingly ( [ svzrho ] ) is modified : the l.h.s .
contains only the @xmath98 term , and , on the r.h.s .
, the perturbative contribution has to be multiplied by a factor @xmath623 . the numerical result obtained from the duality - improved sum rule ( [ svzrho ] )
is also shown in fig .
[ fig : frho ] .
the agreement between the sum rule prediction and experiment is impressive : @xmath624 whereas @xmath625 @xcite .
the estimated theoretical uncertainty quoted in ( [ frho1 ] ) is typical for qcd sum rules , reflecting the approximate nature of this method .
meson decay constant calculated from qcd sum rule @xcite for the correlation function of axial - vector currents ( solid line ) , in comparison with the experimental value ( boxes ) .
the quark - hadron duality threshold is fitted simultaneously as @xmath626 gev@xmath518 .
the uncertainty of about 10 - 15% has to be added to the theoretical curve._,width=453 ] so far we only reproduced @xmath599 , using the experimental value of @xmath627 . in principle , it is possible to go one step further and estimate also the @xmath98-mass from the sum rule .
furthermore , a very similar sum - rule analysis for the correlation function of axial currents ( [ corra1 ] ) ( the invariant amplitude @xmath502 ) successfully reproduces the value of @xmath547 ( see fig . [
fig : fpi ] ) .
it is possible to extend the method of qcd sum rules to baryons @xcite .
the idea is to construct special quark currents with baryon quantum numbers which can serve as a source of baryon production / annihilation from / to qcd vacuum . in reality ,
such currents hardly exist , but they are allowed in qcd if colour - neutrality is obeyed .
a well known example is the ioffe current with the nucleon quantum numbers ( i.e. , spin 1/2 ) : @xmath628 where @xmath629 are color indices , @xmath630 is the charge conjugation matrix and , for definiteness , the proton flavour content is chosen . as a next step ,
one constructs a correlation function @xmath631 correction , ( c),(d ) -quark condensate diagrams._,width=453 ] the corresponding diagrams are shown in fig .
[ fig : baryons ] , including perturbative loops and vacuum condensates .
they look quite different from the quark - antiquark loops , but differences concern the short - distance parts of the diagrams .
the universality of condensates allows one to calculate @xmath632 without introducing new input parameters .
the hadronic contribution contains a total sum over states produced and annihilated by the current @xmath633 , starting from the lowest possible state , the nucleon : @xmath634 the derivation of the qcd sum rule is done along the same lines as in the previous subsection . omitting the details , let me only mention that from this sum rule an approximate formula for the nucleon mass is obtained , @xmath635^{1/3}\,,\ ] ] relating it to the quark - condensate density .
in fact , the quark masses @xmath552 themselves generate very small corrections to this relation and are neglected .
thus , qcd sum rules provide an answer to the question that was raised in lecture 1 : almost @xmath636 of the baryonic mass in the universe are due to vacuum condensates .
as already mentioned , the sum rule ( [ moments ] ) can be used to extract the @xmath193-quark mass . here , the role of borel transformation is played by a simple differentiation at @xmath588 , which turns out to be more useful . at low @xmath514 ,
the moments ( [ moments ] ) are especially convenient for @xmath221 determination because the gluon condensate effects are small . replacing @xmath637 , @xmath638 ,
, one obtains analogous sum rule relations for @xmath218 quark , where gluon condensate is much less important , being suppressed by @xmath639 .
recent analysis @xcite of these sum rules yields @xmath640 gev , @xmath641 gev .
another subset of charmonium sum rules ( higher moments at fixed large @xmath348 ) was recently employed in @xcite , with a prediction for @xmath221 in agreement with the above .
the heavy - quark mass determination using sum rules is also done in a different way , employing the large @xmath514 moments ( [ moments ] ) which are less sensitive to the cross - section above the open flavour threshold .
these moments , however , demand a careful treatment of coulomb interactions between heavy quark and antiquark in the perturbative diagrams .
remember that one - gluon exchange yields coulomb potential .
close to the threshold of heavy quark - pair production , @xmath642 ( @xmath643 ) this part of quark - antiquark interaction becomes important , and at large @xmath514 the near - threshold region dominates in the perturbative coefficients @xmath644 in ( [ moments ] ) .
a systematic treatment of this problem is possible in _ nonrelativistic qcd ( nrqcd ) _ , a specially designed effective theory obtained from qcd in the infinite heavy quark mass limit ( for a review see , e.g. @xcite ) .
equally well , qcd sum rules allow to estimate the masses of light @xmath645 quarks . to give only one typical example from the vast literature
let me refer to the recent analysis @xcite , based on the correlation function for derivatives of the @xmath11-flavoured vector current @xmath646 , ( @xmath172 ) : @xmath647 the qcd answer for @xmath648 is proportional to @xmath649 ( due to @xmath650 ) , making this correlator very sensitive to @xmath176 .
one calculates the usual set of diagrams shown in figs .
[ fig:2loop ] and [ fig : cond ] , where the quark lines are now @xmath11 and @xmath346 .
furthermore , the recent progress in the multiloop qcd calculations allows to reach the @xmath359 accuracy in the perturbative part of @xmath648 and to include also @xmath353 corrections to the condensate contributions .
the hadronic spectral density @xmath651 for this correlator is saturated by @xmath652 states with @xmath11 -flavour , e.g. @xmath653 states with @xmath254 .
the sum rule has the form @xmath654\ , , \label{strange}\ ] ] data of kaon @xmath281-wave scattering on @xmath655 were used to reproduce @xmath656 .
the resulting prediction @xcite for the mass is @xmath657 the ratios of the light ( @xmath645 ) quark masses can be predicted in qcd from the relations for pions and kaons , similar to ( [ gor ] ) .
a systematic derivation is done employing _
chiral perturbation theory _ , an effective theory obtained from qcd in the a low - energy limit , using instead of quarks and gluons , the pion and kaon degrees of freedom ( for a review see @xcite ) .
the result is @xcite : @xmath658 from the above ratios and the value ( [ msvalue ] ) one obtains @xmath659 mev and @xmath660 mev .
we see that qcd sum rules are extremely useful for the quark mass determination .
the @xmath661 values extracted from sum rules are included , together with the lattice determinations and results of other methods , in the world - average intervals in @xcite presented in ( [ qmasses ] ) . in @xmath227-meson decays the ckm parameters of the standard model are inseparable from hadronic matrix elements .
hence , without qcd calculation of these matrix elements with an estimated accuracy , it is impossible to use experimental data on @xmath227 decays for extracting the standard model parameters and for detecting / constraining new physics effects .
currently , lattice qcd provides many hadronic parameters for @xmath227 physics , with a continuously improving accuracy .
qcd sum rules represent another actively used working tool . with condensates and
quark masses determined from a set of experimentally proven sum rules for light - quark and heavy quarkonium systems , one has a real possibility to assess the theoretical accuracy of the sum rule predictions by varying the input within allowed intervals .
one of the most important applications of qcd sum rules is the determination of the @xmath227-meson decay constant @xmath662 defined via the matrix element @xmath663 ( @xmath664 ) .
note that @xmath662 multiplied by @xmath665 determines the width of leptonic @xmath227 decays , such as @xmath666 . to calculate @xmath662 from qcd sum rules ,
one usually employs the correlation function : @xmath667 so that the lowest @xmath227 meson term in the hadronic sum for this correlation function contains @xmath662 : @xmath668 to obtain the sum rule , one needs to calculate @xmath669 from the perturbative and condensate diagrams in figs .
[ fig:2loop ] and [ fig : cond ] , with @xmath510 vertices emitting and absorbing @xmath218 and @xmath346 lines .
one , therefore needs an input value for @xmath220 and @xmath176 ( @xmath552 can safely be neglected ) .
these values are taken from the analyses overviewed in the previous subsection .
the sum rule has a form similar to ( [ svzrho ] ) for @xmath599 , naturally , the expressions for the perturbative part and for the coefficients @xmath530 are completely different .
also the hierarchy of contributions in the heavy - light correlation function differs from the light - quark case .
now the quark condensate becomes very important being proportional to @xmath670 .
the recent updates of the sum rule for @xmath662 obtained in @xcite take into account the @xmath358 corrections to the heavy - light loop calculated in @xcite .
the numerical prediction of the sum rule , taking @xmath671 gev , is : @xmath672 mev and @xmath673 mev @xcite , in a good agreement with the most recent lattice qcd determinations . to complete our brief survey of qcd sum rules ,
let me introduce one important version of this method , the light - cone sum rules ( lcsr ) @xcite used to calculate various hadronic amplitudes relevant for exclusive processes . in the following , we consider the application of lcsr to the @xmath674 transition amplitude ( see fig .
[ fig : bpi ] ) .
the latter is determined by the hadronic matrix element @xmath675q_\mu , \label{bpiformf}\ ] ] generated by the @xmath676 weak current ( [ jweak ] ) . due to spin - parity conservation
only the vector part of the current contributes .
there are two independent 4-momenta @xmath408 and @xmath346 , and one independent invariant @xmath357 , the momentum transfer squared .
the initial and final mesons are on shell , @xmath677 and @xmath678 .
it is quite obvious that one needs two invariant functions of @xmath357 , the _ form factors _ @xmath679 and @xmath680 , to parameterize this matrix element
. only one form factor @xmath681 is interesting , the other one is kinematically suppressed in the measurable @xmath682 semileptonic decay rate ( @xmath406 ) .
to derive lcsr for @xmath683 one uses a specific correlation function , which itself represents a hadronic matrix element .
it is constructed from the product of the weak @xmath684 current and the current @xmath685 used to generate @xmath227 in ( [ fb ] ) .
the currents are taken at two different 4-points and sandwiched between vacuum and the one - pion state .
the formal definition of this correlation function reads : @xmath686 where it is sufficient to consider only one invariant amplitude @xmath687 .
note that we now have two kinematical invariants , @xmath357 and @xmath688 , still @xmath677 . ) , from left to right : the leading - order , @xmath353 correction , soft gluon .
the blob with @xmath539 denotes the pion distribution amplitude .
_ , width=453 ] diagrammatically , the correlation function ( [ bpicorr ] ) is represented in fig .
[ fig : bpidiag ] . at spacelike @xmath689 and @xmath348
, very similar diagrams describe the process @xmath690 , the one - pion production by two virtual photons via e.m .
currents . two - photon ( double - tagged ) collisions .
] one only has to replace all quarks in the diagrams in fig .
[ fig : bpidiag ] by either @xmath110 or @xmath10 quarks .
both objects : the heavy - light correlation function ( [ bpicorr ] ) and the @xmath690 amplitude contain one virtual quark propagating between vertices and a quark - antiquark pair which is emitted at points @xmath24 and @xmath527 and converted into a real pion state . at large spacelike external momenta
@xmath691 the space - time interval @xmath692 approaches the light - cone .
hence , the virtual quark in both @xmath693two - currents @xmath694 pion@xmath695 amplitudes propagates at short distances allowing a perturbative qcd description .
the calculable short - distance parts are process - dependent . in case of the correlation function ( [ bpicorr ] )
we have a virtual @xmath218 quark propagating between vertices of flavour - changing currents , whereas in the @xmath690 amplitude the light quark propagates between e.m
. vertices .
the long - distance part in both cases is , however , the same vacuum - to - pion matrix element of light quark and antiquark emitted at points @xmath24 and @xmath527 : @xmath696 where in one case @xmath697 , @xmath698 , @xmath699 and in the other case @xmath700 , @xmath701 , @xmath702 . due to isospin symmetry
, the difference between these two configurations is indeed very small . in order to treat the diagram with one extra gluon entering the pion
, one has to introduce an additional quark - antiquark - gluon matrix elements of the type @xmath703 where @xmath704 . in the above @xmath705
denote certain combinations of dirac matrices .
having separated short- and long - distances , one is able to calculate the correlation function ( [ bpicorr ] ) in a form of _ light - cone ope _
, where the short - distance part ( the virtual @xmath218 quark propagator plus gluon corrections ) is multiplied by a long - distance part , the universal matrix elements such as ( [ matrpi ] ) , ( [ matrixglue ] ) .
the latter can be parameterized in terms of _ light - cone distribution amplitudes _ of the pion @xcite , the most important of them is defined by @xmath706 where @xmath707 is a characteristic momentum scale , determined by the average @xmath440 in the correlation function . in the above definition , @xmath110 and @xmath708 are the fractions of the pion momentum @xmath408 carried by the quark and antiquark , in the approximation where one neglects the transverse momenta of the constituents with respect to the longitudinal momenta .
the leading - order answer for @xmath687 obtained from the first diagram in fig .
[ fig : bpidiag ] using ( [ pionwf ] ) is quite simple : @xmath709 the pion da @xmath710 plays here the same role of nonperturbative input as the condensates in the conventional qcd sum rules considered above .
asymptotically , that is at @xmath711 , qcd perturbation theory implies @xmath712 .
however , at the physical scale @xmath713 , at which the ope is applied to the correlation function ( [ bpicorr ] ) , nonasymptotic effects also contribute , which we will not discuss for brevity .
importantly , there is a power hierarchy of different contributions stemming from the diagrams in fig .
[ fig : bpidiag ] , determined by the large scale in the correlation function .
this scale is given by the virtuality of @xmath218 quark , i.e. by the quantity which stands in the denominator of the @xmath218 -quark propagator : @xmath714 .
importantly , this quantity remains large when @xmath357 is positive ( timelike ) but not very large , @xmath715 , allowing one to penetrate into the lower part of the kinematical region @xmath716 of @xmath682 decay .
an example of a subleading contribution is the diagram with an additional gluon entering pion da ( the third one in fig .
[ fig : bpidiag ] ) .
it has two powers of inverse scale and is suppressed with respect to the leading - order diagram . over recent years
@xmath353 corrections and quark - antiquark - gluon contributions have been calculated to improve the result ( [ qcd ] ) .
having evaluated @xmath687 as a function of @xmath357 and @xmath688 , we are still half - way from the final sum rule .
the next important step is writing down a dispersion relation in the variable @xmath688 , the external momentum of the current with @xmath227 meson quantum numbers .
the diagrammatical representation of the dispersion relation is shown in fig.[fig : bpidisp ] .
it contains the same set of hadronic states as in the sum rule for @xmath662 , starting from @xmath227 meson ground state . using ( [ fb ] ) and ( [ bpiformf ] ) one obtains the hadronic representation we need : @xmath717 the ellipses in the above denote the contributions from the excited @xmath227 and from continuum states . equating the qcd result ( [ qcd ] ) in the region of validity @xmath689 with the dispersion relation ( [ disp ] ) one obtains a raw sum rule relation for @xmath718 .
.)_,width=491 ] the rest of the calculation follows the usual qcd sum rule procedure : borel transformation in @xmath688 and subtraction of the contribution from higher states invoking quark - hadron duality .
one finally arrives at an expression for the desired form factor , the dominant term of which obtained directly from ( [ qcd ] ) is given by @xmath719 here , @xmath617 is the borel mass parameter , and the scale @xmath720 reflects the characteristic virtuality of the correlation function , @xmath721 .
the integration limit @xmath722 depends on the effective threshold @xmath723 above which the contribution from higher states to the dispersion relation ( [ disp ] ) is canceled against the corresponding piece in the qcd representation ( [ qcd ] ) .
the parameters @xmath724 are usually taken from the sum rule for @xmath662 considered in the previous subsection .
the most recent predictions for @xmath725 obtained from lcsr @xcite were used to extract @xmath726 from the measurements of the exclusive semileptonic width at @xmath227 factories @xcite , using the formula for the decay rate : @xmath727 ^ 2\ , .
\label{bpiwidth}\ ] ] where @xmath728 is the pion energy in the @xmath227 meson rest frame .
qcd has a thirty - years history and embraces many approaches , some of them developed quite independently from the others .
due to self - interactions of gluons this theory has an extremely rich dynamics , combining asymptotic freedom at short distances with the self - emerging energy scale @xmath177 and confinement at long distances .
accordingly , qcd has two different phases : the perturbative one responsible for quark - gluon processes at large momentum transfers and the nonperturbative one where the only observable states are hadrons formed by confined quarks and gluons .
yet there is no complete analytical solution for the hadronic phase of qcd .
one has to rely on approximations : either numerical ( qcd on the lattice ) or analytical ( qcd sum rules ) .
in addition , several effective theories corresponding to different limits of qcd and exploiting the rich symmetry pattern of the theory are successfully used . in the overview of qcd given in these lectures
i tried to emphasize the importance of the hadronic aspects of qcd , where the most nontrivial phenomena and challenging problems are accumulated .
i would like to thank the organizers of the school for inviting me to give these lectures and for an enjoyable meeting at tsakhkadzor .
i am grateful to the discussion leaders for their help , especially to a. pivovarov for useful remarks on the manuscript .
my special thanks are to the students of the school for their stimulating interest and for many questions and comments .
k. hagiwara _ et al .
_ [ particle data group collaboration ] , phys . rev .
d * 66 * ( 2002 ) 010001 .
s. r. sharpe , lectures given at tasi 94 : cp violation and the limits of the standard model , boulder , co , hep - ph/9412243 ; + a. s. kronfeld , in shifman , m. ( ed . ) : at the frontier of particle physics .
vol . 4 , 2411 - 2477,hep - lat/0205021 .
r. fleischer , lectures at this school .
t. nakano _ et al .
_ [ leps collaboration ] , phys . rev .
* 91 * ( 2003 ) 012002 , hep - ex/0301020 ; + s. stepanyan _ et al . _
[ clas collaboration ] , phys .
lett . * 91 * ( 2003 ) 252001 , hep - ex/0307018 + v. v. barmin _ et al . _
[ diana collaboration ] , phys . atom . nucl .
* 66 * ( 2003 ) 1715 , hep - ex/0304040 .
b. l. ioffe , nucl .
b * 188 * ( 1981 ) 317 [ erratum - ibid .
b * 191 * ( 1981 ) 591 ] ; z. phys .
c * 18 * ( 1983 ) 67 .
y. chung , h. g. dosch , m. kremer and d. schall , phys .
b * 102 * ( 1981 ) 175 ; nucl .
b * 197 * ( 1982 ) 55 .
i. i. balitsky , v. m. braun and a. v. kolesnichenko , nucl .
* b312 * ( 1989 ) 509 ; + v. m. braun and i. e. filyanov , z. phys .
* c44 * ( 1989 ) 157 .
v. l. chernyak and i. r. zhitnitsky , nucl .
* b345 * ( 1990 ) 137 .
g. p. lepage and s. j. brodsky , phys .
lett . * 87b * ( 1979 ) 359 ; phys .
* d22 * ( 1980 ) 2157 ; + a. v. efremov and a. v. radyushkin , phys .
* b94 * ( 1980 ) 245 ; theor .
* 42 * ( 1980 ) 97 ; + v. l. chernyak and a. r. zhitnitsky , jetp lett . * 25 * ( 1977 ) 510 ; sov .
j. nucl .
* 31 * ( 1980 ) 544 ; + a. khodjamirian , r. ruckl , s. weinzierl , c. w. winhart and o. yakovlev , phys . rev .
* d62 * ( 2000 ) 114002 ; + p.
ball and r. zwicky , jhep * 0110 * ( 2001 ) 019 .
s. b. athar _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 68 * ( 2003 ) 072003 . | notes of five lectures given at the 2003 european school of high - energy physics , tsakhkadzor , armenia , september 2003 |
the manipulation of the magnetization by electric fields , which is referred to as converse magnetoelectric effect , is one of the most challenging tasks in today s spintronic devices.@xcite the presence of such a cross - coupling between ordered magnetic and dielectric states in so - called multiferroic systems allows for the development of a new family of devices , where the spin degree of freedom is controlled by electric instead of magnetic fields.@xcite it has been shown that a robust electric field control of magnetism can be realized by using extrinsic multiferroic hybrid structures consisting of ferroelectric and ferromagnetic materials.@xcite in these systems , the extrinsic magnetoelectric effects rely on either electric field effects@xcite using carrier mediated ferromagnets,@xcite or employing exchange coupling effects between antiferromagnetic , ferroelectric and ferromagnetic compounds.@xcite moreover , from the beginning the elastic coupling at the interface between ferromagnetic and ferroelectric materials has been exploited to realize a robust electric field control of the magnetization.@xcite in this case , the electric and magnetic degrees of freedom are elastically coupled at the interface between the two ferroic constituents .
more precisely , strain mediated converse magnetoelectric effects rely on an electric field induced strain in the ferroelectric via the converse piezoelectric effect , or via ferroelastic reorientations .
this strain is then transferred into the ferromagnetic thin film clamped onto the ferroelectric , where converse magnetoelastic effects modify the magnetization .
indirect strain - mediated magnetoelectric effects can be realized using particulate composites,@xcite laminated composites,@xcite nanostructured composites,@xcite and horizontal multiferroic hybrid structures . in the latter case ,
the discovery of giant sharp and persistent converse magnetoelectric effects in ferromagnetic thin films epitaxially grown on batio@xmath0 ( bto ) crystals@xcite triggered an increasing research effort in this field , using different ferromagnetic material systems , such as magnetic ferrites ( fe@xmath0o@xmath1,@xcite cofe@xmath2o@xmath1,@xcite and nife@xmath2o@xmath1@xcite ) , magnetic perovskites and double - perovskites ( la@xmath3sr@xmath4mno@xmath0,@xcite la@xmath3ca@xmath4mno@xmath0,@xcite pr@xmath3ca@xmath4mno@xmath0,@xcite and sr@xmath2crreo@xmath5@xcite ) , and @xmath6-magnets ( fe,@xcite ni,@xcite co,@xcite and cofe@xcite ) .
in particular , strong electric field - induced changes of the magnetic state were reported in fe@xmath0o@xmath1-based extrinsic multiferroic hybrids.@xcite since fe@xmath0o@xmath1 has a ferrimagnetic ground state with a high curie temperature of about 850k,@xcite a pronounced magnetostrictive effect,@xcite and a high spin polarization at the fermi level,@xcite fe@xmath0o@xmath1 is a promising candidate for possible future magnetoelectric devices . moreover ,
a sizable spontaneous ferroelectric polarization was reported in fe@xmath0o@xmath1 films@xcite which is caused by a non - centrosymmetric charge order.@xcite therefore , fe@xmath0o@xmath1 seems to be not only the oldest magnetic material but also the first magnetoelectric multiferroic known to mankind.@xcite to quantitatively understand the strain - mediated magnetoelectric effects in bto - based multiferroic hybrid structures , as a first step , we recently investigated the magnetization changes as a function of temperature in fe@xmath7co@xmath7/bto and ni / bto hybrids caused by magnetoelastic effects.@xcite we demonstrated that the manipulation of the magnetization in bto - based multiferroic hybrid structures can be theoretically well described by controlling the ferromagnetic and ferroelastic domain configuration in the multiferroic hybrid . here , as a second step
, we show that by knowing the ferroelastic domain configuration in the bto substrate as a function of the applied electric field , the strain - mediated converse magnetoelectric effect in epitaxially grown fe@xmath0o@xmath1/bto hybrid structures can be theoretically simulated using a simple model , consisting of two magnetic regions in the fe@xmath0o@xmath1 thin film .
this demonstrates that by controlling the ferroelastic domain state in the bto substrate large and robust manipulations of the magnetization as a function of the applied electric field are feasible at room temperature in bto - based multiferroic hybrid structures .
the article is organized as follows : in sec .
[ sec : exp ] , we discuss the growth process and the structural as well as magnetic properties of fe@xmath0o@xmath1 thin films epitaxially grown on bto substrates . since x - ray diffraction measurements using synchrotron radiation carried out on these fe@xmath0o@xmath1/bto hybrid structures reveal two differently strained regions in the fe@xmath0o@xmath1 thin film at low electric fields , a two - region model is introduced in sec . [
sec : model ] . in this model ,
the room - temperature magnetization behavior of fe@xmath0o@xmath1/bto hybrids as a function of the applied electric field can be explained by the magnetic behavior of parts of the fe@xmath0o@xmath1 thin film elastically clamped onto two different ferroelastic domains of the bto substrate .
thus , to simulate the strain - mediated converse magnetoelectric effect , at first the volume fraction of the ferroelastic domains in the bto substrate is determined ( cf .
[ sec : xc ] ) , and then the magnetic state of the fe@xmath0o@xmath1 thin film on top of these domains is calculated using magnetoelastic theory ( cf .
[ sec : mamc ] ) . in sec .
[ sec : expsimu ] , we show that there is an excellent agreement between the experimental results and the theoretical simulations based on the two - region model .
this demonstrates that the experimentally obtained room - temperature converse magnetoelectric effects in bto - based multiferroic hybrids are well described within this approach for magnetic fields larger than the magnetic saturation field .
the fe@xmath0o@xmath1 thin films with thicknesses between 40 nm and 70 nm were epitaxially grown on 0.5 mm thick ( 001)-oriented bto substrates by laser molecular - beam epitaxy ( laser - mbe)@xcite monitored by an _ in - situ _ high pressure rheed system.@xcite the energy density of the krf excimer laser ( @xmath8 nm ) at the target was set to 3.1j/@xmath9 and the laser repetition rate was 2hz .
the deposition was carried out in an atmosphere of pure argon with a pressure of @xmath10 mbar at a temperature of 593k , i.e. , in the cubic phase of the bto substrate .
the growth process of the fe@xmath0o@xmath1 thin films is illustrated in fig .
[ fig : fig1 ] on the basis of the intensity evolution of the rheed @xmath11 spot , which was recorded during the growth of a 64 nm thick fe@xmath0o@xmath1 film . before starting the growth process
, the rheed pattern of the bto substrate reveals three spots , which belong to the first laue circle and can be indexed with @xmath12 , @xmath11 , and @xmath13 [ cf .
inset ( i ) of fig .
[ fig : fig1 ] ] . by starting the deposition at @xmath14s ,
the high intensity of the @xmath11 spot drastically decreases [ cf .
[ fig : fig1 ] ( a ) ] .
after 50pulses ( @xmath15s ) , the deposition was stopped [ cf .
[ fig : fig1 ] ( ii ) ] .
the rheed pattern recorded at this moment does not show any reflections [ cf .
inset ( ii ) of fig .
[ fig : fig1 ] ] .
this indicates a highly disordered and nearly amorphous growth within the first monolayers of fe@xmath0o@xmath1 .
further deposition leads to a three dimensional crystalline growth as indicated by a checkerboard like rheed pattern [ cf .
inset ( iii ) of fig .
[ fig : fig1 ] ] . upon further continuing the deposition , the three - dimensional growth mode turns over into a two - dimensional one with increasing rheed intensity and emerging intensity oscillations [ cf .
[ fig : fig1 ] ( b ) ] . in the last stage of the growth process ( @xmath16s ) ,
a real two - dimensional layer - by - layer growth mode can be obtained [ cf .
[ fig : fig1 ] ( c ) ] . at this stage ,
a single unit cell of fe@xmath0o@xmath1 , which is manifested by four rheed oscillations,@xcite was deposited followed by a growth interruption of 30s , allowing for the relaxation of the film surface . the rheed pattern recorded at the end of the growth process [ cf .
inset ( iv ) of fig .
[ fig : fig1 ] ] does not show any indication of transmission like spots , suggesting a smooth surface of the fe@xmath0o@xmath1 thin film .
using atomic force microscopy as well as x - ray reflectometry , a surface roughness ( rms value ) ranging between 0.8 nm and 2.5 nm was found .
therefore , as obvious from fig .
[ fig : fig1 ] , fe@xmath0o@xmath1/bto hybrid structures can be fabricated in a true layer - by - layer growth mode .
however , the highly disordered growth at the beginning of the deposition of fe@xmath0o@xmath1 , which might be explained by the lattice mismatch of fe@xmath0o@xmath1 and bto of 4.7% at the growth temperature , results in a large mosaic spread .
x - ray diffraction measurements ( not shown here ) reveal no secondary phases as well as a full width at half maximum of the rocking curves around the fe@xmath0o@xmath1(004 ) reflection of about @xmath17 .
this is more than one order of magnitude larger than observed for fe@xmath0o@xmath1 thin films grown on lattice - matched mgo substrates.@xcite o@xmath1 film grown on a ( 001)-oriented bto substrate measured at 300k with the magnetic field applied in the film plane .
the magnetic field dependence of the magnetization can be fitted by a @xmath18 behavior using @xmath19 ( red solid line ) .
( a ) the @xmath20 loop reveals a coercive field of 35mt and a saturation magnetization of @xmath21ka / m at @xmath22 t .
( b ) the normalized remanent magnetization as a function of temperature measured after field cooling the sample with @xmath23 t discloses a verwey transition at 121k . ] the magnetic properties of the hybrid structures were studied via superconducting quantum interference device ( squid ) magnetometry in the temperature range between 2k and 300k at magnetic fields of up to 7 t applied in the film plane .
figure [ fig : fig2 ] shows the magnetic behavior of a 45 nm thick fe@xmath0o@xmath1 film grown on a bto substrate . the magnetic hysteresis loop @xmath20 measured at 300k reveals a coercive field of 35mt [ cf .
inset ( a ) of fig .
[ fig : fig2 ] ] , which is close to the value reported for fe@xmath0o@xmath1 thin films on lattice - matched mgo substrates.@xcite furthermore , a high saturation magnetization of @xmath21ka / m at @xmath24 t is obtained , which corresponds to [email protected] .. the difference @xmath26 between @xmath27 and the bulk value @xmath28f.u .
is generally explained by the presence of anti - phase boundaries ( apb ) .
apbs exhibit a strong antiferromagnetic coupling , which results in a @xmath18 behavior of the magnetization , where the parameter @xmath29 depends on the density of apbs.@xcite using @xmath19 , the magnetic hysteresis can be nicely fitted [ cf .
red line in fig .
[ fig : fig2 ] ] .
this value is much lower than values reported for fe@xmath0o@xmath1 thin films grown on ( 001)-oriented mgo or al@xmath2o@xmath0
substrates.@xcite this might be explained by the disordered growth within the initial phase of the deposition process , leading to a disconnection of the fe@xmath0o@xmath1 thin film with the surface structure of the bto substrate .
thus , the deviation in saturation magnetization @xmath26 might not be explained by apbs alone .
oxygen non - stoichiometry fe@xmath30o@xmath1 with @xmath31 is another possibility , which affects @xmath27.@xcite the degree of oxygen non - stoichiometry @xmath32 can be investigated by measuring the verwey transition @xmath33 , which is a hallmark of the quality of fe@xmath0o@xmath1 thin films .
the inset ( b ) of fig .
[ fig : fig2 ] shows the remanent magnetization as a function of temperature measured after field cooling the sample in an external magnetic field of @xmath23 t .
the verwey transition is clearly observable at 121k , which is identical to the bulk value .
thus , our fe@xmath0o@xmath1 thin films grown on bto substrates have a low amount of oxygen vacancies and the correct iron oxide phase . other iron oxides such as maghemite
can be safely excluded . note that a small reduction of the phase transition temperature was observed in other samples , which can be translated into an oxygen non - stoichiometry of @xmath34 assuming a linear dependence of @xmath33 on @xmath32.@xcite altogether , the measured reduction of the saturation magnetization with respect to the bulk value might be explained by a combination of effects caused by apbs and a slight oxygen non - stoichiometry in the fe@xmath0o@xmath1 thin films .
one of the intriguing properties of multiferroic hybrid structures consisting of ferromagnetic and ferroelectric materials is the possibility to manipulate the magnetization @xmath35 by an electric field @xmath36 . to investigate this converse magnetoelectric effect in the fe@xmath0o@xmath1/bto hybrid structures ,
an au bottom electrode was sputtered on the backside of each bto substrate , which enables us to apply an electric field @xmath37 across the bto substrate along the @xmath38 direction using fe@xmath0o@xmath1 as top electrode .
the fe@xmath0o@xmath1/bto hybrids are further heated to 450k well above the curie temperature of bto and slowly cooled down to room temperature while applying an electric field of 400kv / m .
o@xmath1(38nm)/bto multiferroic hybrid .
the magnetic field was applied in the film plane along the @xmath39 direction .
a miscut bto substrate was used for the experiment with a misalignment of @xmath40 ( @xmath41 ) of the [ 001 ] ( [ 100 ] ) direction of the ferroelastic unit cell with respect to the @xmath38 ( @xmath42 ) direction describing the surface of the bto substrate .
the variation of the ferroelastic domain configuration as a function of the electric field is illustrated in the bottom part of the figure . at high positive and negative fields ,
a ferroelastic single @xmath43 domain state is expected , while at low electric fields a @xmath43/@xmath44 multi - domain state is present in the bto crystal .
the blue arrows indicate the orientation of the ferroelectric polarization and the blue regions denote ferroelastic domain walls .
polarization rotation induced by the electric field is neglected for simplicity . ] as an example , fig .
[ fig : fig3 ] shows the modification of the magnetization as a function of the applied electric field , @xmath45 , for different values of the applied magnetic field .
the curves were measured at room temperature , i.e. , in the tetragonal phase of bto , using a fe@xmath0o@xmath1(38nm)/bto multiferroic hybrid . here , the projection of the magnetization @xmath46 on the magnetic field direction , which was oriented along the @xmath39 direction , is plotted as a function of the electric field strength @xmath37 applied across the bto substrate . in analogy to ni / bto hybrid structures discussed in ref .
, a butterfly shape of the @xmath45-loops is observable . on the basis of the measured magnetization changes @xmath47
, a pseudo - linear magnetoelectric coupling coefficient can be estimated using @xmath48 , which results in @xmath49s / m .
this value is lower than the magnetoelectric constant of the prototype intrinsic magnetoelectric material cr@xmath2o@xmath0 ( @xmath50s / m),@xcite demonstrating that the magnitude of the converse linear magnetoelectric effect in bto - based hybrid structures is comparable or even less than in intrinsic magnetoelectric materials due to the unfavorable ratio between the thickness of the ferromagnetic thin film ( @xmath51 ) and the thickness of the bto substrate ( @xmath52 ) .
however , the observed magnetoelectric effect is robust at room temperature . only very recently ,
large room - temperature magnetoelectric effects were reported in magnetoelectric hexaferrites.@xcite this was the first observation of significant magnetoelectric effects in intrinsic magnetoelectric materials at room temperature .
the variation of the magnetization as a function of an applied electric field in bto based multiferroic hybrids at 300k can be explained by taking into account the ferroelastic domains of bto.@xcite in the tetragonal phase , bto exhibits three types of ferroelastic domains with the polarization vector pointing either along the out - of - plane direction ( @xmath43 domains ) or one of the two in - plane directions ( @xmath53 domains and @xmath44 domains ) . the tetragonal unit cell of the @xmath54 domains is oriented with its longer axis parallel to the bto surface .
it induces tensile strain in the overlying ferromagnetic thin film , which modifies the magnetic anisotropy due to converse magnetoelastic effects .
thus , to quantitatively describe converse magnetoelectric effects in bto - based multiferroic hybrids , the volume fraction of the different ferroelastic domains has to be known .
the requirements are relaxed by using miscut bto crystals . in this case only a single type of @xmath54 domains is formed .
o@xmath1(38nm)/bto hybrid recorded in the ( a ) @xmath55-plane and ( b ) @xmath56-plane at 0kv / m using a miscut bto crystal ( yellow / red : high intensity , light blue / dark blue : low intensity ) .
only reflections caused by ferroelastic @xmath43 domains and tilted @xmath44 domains are visible .
the expected positions of the reflections caused by diffraction from tilted ferroelastic @xmath53 domains are marked by dashed circles . ]
figure [ fig : fig4 ] shows x - ray diffraction measurements around the asymmetric bto ( 204 ) reflection . in the experiment a fe@xmath0o@xmath1(38nm)/bto hybrid with a bto substrate is used , which has a misalignment of about @xmath40 ( @xmath41 ) of the [ 001 ] ( [ 100 ] ) direction of the ferroelastic unit cells with respect to the @xmath38 ( @xmath42 ) direction describing the surface of the bto crystal ( cf .
inset of fig .
[ fig : fig3 ] ) .
the mesh scan around the bto ( 204 ) reflection reveals only two reflections .
one is caused by scattering from ferroelastic @xmath43 domains at @xmath57rlu and the other by tilted ferroelastic @xmath44 domains .
the second type of @xmath54 domains ( @xmath53 domains ) could not be detected , neither in the @xmath55-plane [ cf .
[ fig : fig4](a ) ] nor in the @xmath56-plane [ cf . fig .
[ fig : fig4](b ) ] . therefore , only a single type of @xmath54 domains is present in the used miscut bto crystals at zero electric field . in this case
the evolution of the ferroelastic domains as a function of the applied electric field @xmath37 can be described as follows : at high positive and negative electric fields , a ferroelastic single @xmath43 domain state is present , which changes into a ferroelastic multi - domain state around @xmath58kv / m .
this ferroelastic multi - domain state consists of ferroelastic @xmath43 and tilted @xmath44 domains as well as @xmath43@xmath43 and @xmath54@xmath43 domain walls ( cf .
bottom part of fig .
[ fig : fig3 ] ) .
o@xmath1/bto hybrid structure .
the measurements were performed at the bm28 beamline at the european synchrotron facility ( esrf ) . upon increasing the electric field from -28kv / m to 400kv / m again ( black open symbols ) , the initial measurement at 400kv / m is reproduced .
the effect of the strain imposed on the fe@xmath0o@xmath1 film is shown in more detail in ( b ) and ( c ) . in the case of 400kv / m
, the ( 004 ) reflection of the fe@xmath0o@xmath1 thin film ( black symbols ) can be well fitted using a single peak profile ( black solid line ) . at 0kv / m , two peak profiles ( dashed lines )
are needed to fit the measured reflection ( red solid line ) .
the gray lines display the difference between the experimental data ( symbols ) and the fits ( solid lines ) , and the blue arrows mark the @xmath59 positions of the reflections . ]
the impact of the ferroelastic domain state on the overlying ferromagnetic thin film due to the imposed strain can be investigated by performing x - ray diffraction measurements on crystalline fe@xmath0o@xmath1/bto hybrid structures . as an example , fig .
[ fig : fig5 ] shows @xmath60-scans around the bto ( 002 ) reflection recorded at different electric field strengths @xmath37 applied across the fe@xmath0o@xmath1/bto hybrid structure .
the measurements shown in fig . [ fig : fig5 ] were carried out at the bm28 beamline at the european synchrotron radiation facility ( esrf ) using synchrotron radiation with an energy of 7.1255kev . at @xmath61kv / m ,
the bto substrate yields only one reflection at @xmath62rlu [ cf .
full black symbols in fig .
[ fig : fig5](a ) ] .
this confirms that a single ferroelastic @xmath43 domain state exists at this electric field strength . by reducing the electric field to values close to the coercive field of bto ( @xmath63kv / m ) , a second reflection around @xmath64rlu emerges , which is caused by ferroelastic @xmath44 domains formed in the bto substrate [ cf
full red symbols in fig .
[ fig : fig5](a ) ] . by increasing the electric field to the starting value of @xmath61kv / m , the first measurement is reproduced , i.e. , a single ferroelastic @xmath43 domain state is restored [ cf .
open black symbols in fig .
[ fig : fig5](a ) ] .
the impact of the strain imposed on the fe@xmath0o@xmath1 thin film was investigated in detail around the fe@xmath0o@xmath1 ( 004 ) reflection [ figs .
[ fig : fig5](b ) and ( c ) ] . the peak profile of the fe@xmath0o@xmath1 ( 004 ) reflection measured while applying an electric field of @xmath61kv
/ m can be well fitted using a single pearson vii peak function@xcite [ cf .
solid line in fig .
[ fig : fig5](b ) ] .
this demonstrates that the fe@xmath0o@xmath1 thin film is homogeneously strained at this electric field strength .
an out - of - plane lattice constant of @xmath65 nm can be derived from the peak position of the fe@xmath0o@xmath1 ( 004 ) reflection .
furthermore , x - ray diffraction measurements carried out around the fe@xmath0o@xmath1 ( 404 ) reflection ( not shown here ) reveal an in - plane lattice constant of @xmath66 nm . comparing these values to the bulk lattice parameters indicates that our fe@xmath0o@xmath1 thin films exhibit a tensile out - of - plane strain of @xmath67% and a compressive in - plane strain of @xmath68% .
this is in contrast to x - ray diffraction measurements published recently.@xcite however , those measurements were performed without any electric field applied to the samples , i.e. , in a ferroelastic multi - domain state of the bto substrate .
the derived values of the in - plane and out - of - plane strain yield a poisson ratio of 0.33 , which is in good agreement with ref . .
in contrast to the x - ray measurements at @xmath61kv / m , the intensity of the fe@xmath0o@xmath1 ( 004 ) reflection at @xmath58kv / m is strongly reduced and a second peak occurs close to the first one [ fig .
[ fig : fig5](c ) ] .
two different peak functions are needed to reproduce this situation [ cf .
dashed lines in fig .
[ fig : fig5](c ) ] .
while the position of the first one is almost identical to the peak position measured at @xmath61kv / m , the second peak is located at a higher @xmath59-value [ cf .
blue arrows in figs .
[ fig : fig5](b ) and ( c ) ] .
this clearly demonstrates that two differently strained regions are present in the fe@xmath0o@xmath1 thin film at @xmath58kv / m .
the one region can be attributed to those parts of the fe@xmath0o@xmath1 thin film clamped to ferroelastic @xmath43 domains of the bto crystal , while the other to those regions of the fe@xmath0o@xmath1 thin film positioned on top of ferroelastic @xmath44 domains .
the @xmath44 domains induce a tensile strain along the @xmath39 direction in the overlaying fe@xmath0o@xmath1 thin film , thereby causing a compressive strain along the out - of - plane direction due to elasticity .
this shifts the fe@xmath0o@xmath1 ( 004 ) reflection to higher @xmath59 values . from the difference in position of the two peaks visible in fig .
[ fig : fig5](c ) , the induced compressive out - of - plane strain of the fe@xmath0o@xmath1 thin film due to the transformation of ferroelastic @xmath43 domains into @xmath44 domains in the bto crystal can be calculated to @xmath69% . by employing the poisson ratio of 0.33 ,
the induced in - plane strains @xmath70% and @xmath71% are derived .
these values are in excellent agreement with the theoretical values @xmath72% and @xmath73% caused by domain reorientations from ferroelastic @xmath43 to @xmath44 domains in the bto crystal .
this provides clear evidence for a perfect strain transmission from the bto substrate into the fe@xmath0o@xmath1 thin film .
figure [ fig : fig5 ] indicates that the large tensile strain is imposed on the fe@xmath0o@xmath1 thin film by the formation of ferroelastic @xmath44 domains in the bto substrate . due to magnetoelastic effects
this strain should strongly affect the magnetization @xmath74 of those parts of the fe@xmath0o@xmath1 film elastically clamped onto these domains .
in contrast , since the ( 004 ) reflection of the fe@xmath0o@xmath1 thin film positioned on top of ferroelastic @xmath43 domains stays nearly unaffected as a function of the applied electric field , the magnetization @xmath75 of these regions is expected to depend only very weakly on the applied electric field .
thus , the converse magnetoelectric effects in fe@xmath0o@xmath1/bto hybrid structures at room temperature can be described in first order by a two - region model @xmath76 here , @xmath77 denotes the volume fraction of the ferroelastic @xmath43 domains in the bto substrate as a function of the applied electric field , and @xmath78 the remaining volume fraction of the emerging ferroelastic @xmath44 domains . in this model , the magnetizations @xmath74 and @xmath75
are considered independent of the electric field .
thus , eq . only considers ferroelastic domain reorientations and does not take into account linear converse piezoelectric effects as well as effects caused by polarization rotations.@xcite this is reasonable , since the inverse linear piezoelectric effect of bto leads to a maximum in - plane strain , which is three orders of magnitude smaller than the strain caused by ferroelastic domain reorientation .
@xcite furthermore , effects due to ferroelastic domain walls are also not included in the model . to calculate the converse magnetoelectric effects using eq .
, the volume fraction @xmath77 as well as the magnetization values @xmath75 and @xmath74 have to be determined .
domains @xmath77 using x - ray diffraction .
( a ) , ( b ) mesh scans around the bto ( 002 ) reflection recorded while applying an electric field of 400kv / m and 0kv / m across the fe@xmath0o@xmath1(38nm)/bto hybrid . to obtain @xmath79 ,
the ratio of the integrated intensities for @xmath80rlu and @xmath81rlu is calculated .
( c ) @xmath79 as a function of the applied electric field . ] the volume fraction of the ferroelastic @xmath43 domains @xmath77 is determined by means of x - ray diffraction measurements.@xcite since the ferroelastic @xmath43 and @xmath54 domains exhibit different unit cell dimensions as well as a change of the polarization direction , both the bragg angles and the intensities of the respective reflections are affected.@xcite for the experimental determination of @xmath82 , x - ray diffraction measurements around the bto ( 002 ) reflection were carried out for different values of the electric fields applied across the fe@xmath0o@xmath1/bto hybrid structure.@xcite figures [ fig : fig6](a ) and ( b ) show two reciprocal space maps recorded with @xmath61kv / m and @xmath58kv / m . for @xmath61kv / m , only a single reflection around @xmath62rlu is visible , which is a marker for ferroelastic @xmath43 domains .
this demonstrates that at this electric field strength a purely ferroelastic @xmath43 domain state is present in the bto substrate . by reducing the electric field strength ,
tilted @xmath44 domains are formed continuously , until the electric coercive field is reached .
scattering from these @xmath44 domains causes a finite intensity around @xmath83rlu [ cf .
[ fig : fig5](b ) ] . to determine @xmath77 ,
the detected intensities were integrated for @xmath80rlu ( @xmath84 ) and @xmath85rlu ( @xmath86 ) . using these intensities ,
@xmath77 can be estimated as @xmath87 .
the resulting volume fraction as a function of the applied electric field is shown in fig .
[ fig : fig6](c ) . by comparing @xmath77 with the @xmath45-loops depicted in fig .
[ fig : fig3 ] , the pronounced interdependence of the magnetization in the ferromagnetic thin film and the volume fraction of the ferroelastic @xmath43 domains in the bto substrate becomes obvious .
this provides further evidence for the validity of the two - region model described by eq . .
we note , however , that for the calculation of @xmath45 , the ferroelastic domain configuration at the surface of the bto is essential , since these domains are elastically coupled to the overlaying ferromagnetic thin film . as x - rays penetrate deep into the bto substrate , the obtained volume fraction @xmath77 is an averaged value . by using different angles of the incoming x - ray beam , i.e. , varying the penetration depth of the x - rays , an uncertainty of less than 3% for @xmath77 can be estimated . as obvious from fig .
[ fig : fig6](c ) , this uncertainty is not larger than the already displayed error bars , describing the experimental uncertainty at a constant angle of the incoming x - rays .
having determined the ferroelastic domain state in the bto substrate as a function of the applied electric field strength , in the two - region model approach [ cf .
eq . ] the strain - mediated converse magnetoelectric effects @xmath88 in the fe@xmath0o@xmath1/bto hybrid depend only on the magnetization values @xmath75 and @xmath74 of the regions elastically clamped to ferroelastic @xmath43 and @xmath44 domains , respectively .
therefore , the remaining step is to determine the magnetization values @xmath75 and @xmath74 . in general , elastic strain can modify the magnetization by two main effects .
first , the direction of the magnetization can be altered by a strain - induced change of the magnetic anisotropy , since the magnetoelastic energy can be regarded as an additional uniaxial magnetic anisotropy in a phenomenological free energy approach.@xcite second , due to strain - induced changes of the bond length and bond angle , the exchange interaction can be modified .
this may cause variations of the magnetic ordering temperature and the magnetization .
even a strain - induced transition from a magnetically ordered / electrically conducting to a nonmagnetic / insulating state is possible by changing the spin state of the magnetic ions.@xcite since previous experiments showed that changes of the magnetic anisotropy are the driving force behind the converse magnetoelectric effects in fe@xmath0o@xmath1/bto hybrids,@xcite we neglect any strain - induced modification of the saturation magnetization @xmath27 and assume @xmath27 to be homogeneous throughout the fe@xmath0o@xmath1 thin film . in this approximation , the magnetization values @xmath75 and
@xmath74 are described by @xmath89 and @xmath90 , respectively .
using magnetoelastic theory , the unit vectors @xmath91 and @xmath92 can be derived by minimizing the free enthalpy density @xmath93 of the fe@xmath0o@xmath1 thin film , which is based on the magnetic energy density @xmath94 given by ( see ref . for more details ) @xmath95 here , @xmath96 is the magnetic anisotropy contribution , @xmath97 the purely elastic energy density , @xmath98 the magnetoelastic energy density , and @xmath99 with @xmath100 are the components of the unit vectors @xmath101 , i.e. , the directional cosines . the strain components @xmath102 ( @xmath103 ) are given in matrix notation with @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , and @xmath109.@xcite to account for the shape anisotropy in ferromagnetic thin films with finite dimensions , an additional contribution @xmath110 is added to eq . . in the framework of a magnetic single domain model ,
@xmath94 is only a function of the magnetization direction @xmath99 and the strain components @xmath102 .
the components @xmath102 with @xmath111 are zero , since no shear strains are expected in the tetragonal phase of bto at room temperature . as discussed in section [ sec : model ] , the strain state of those parts of the fe@xmath0o@xmath1 thin film elastically coupled to ferroelastic @xmath43 domains ( @xmath112 ) and @xmath54 domains ( @xmath113 ) of the bto substrate is @xmath114 assuming a cubic symmetry of the fe@xmath0o@xmath1 thin film , the magnetic anisotropy @xmath96 can be expressed as @xmath115 , with the first order cubic anisotropy constant @xmath116 . the second term in eq
. represents the purely elastic energy @xmath117 . here , @xmath118 are the elastic stiffness constants of the cubic fe@xmath0o@xmath1 thin film , which are regarded as material constants ( @xmath119n / m@xmath120 , @xmath121n / m@xmath120 , and @xmath122n / cm@xmath120).@xcite the interaction between the elastic and the magnetic anisotropy energies is described by the first order magnetoelastic energy density @xmath98 : @xmath124\ , , \label{eq : magel}\end{aligned}\ ] ] with the magnetoelastic coupling coefficients @xmath125 and @xmath126 . for bulk
magnetostrictive samples , @xmath125 and @xmath126 can be expressed as a function of the magnetostrictive strains @xmath127 and @xmath128 : @xmath129 and @xmath130 .
we note that the magnetoelastic coupling coefficients @xmath131 in ferromagnetic thin films may deviate from the bulk values due to surface effects and/or the influence of strain.@xcite in our analysis , we neglect these effects and use the bulk values @xmath132 and @xmath133 for the calculations.@xcite a proportionality factor @xmath134 is introduced to account for any deviation in the magnetoelastic coupling from bulk - like behavior . plotted versus the angle @xmath135 defining the in - plane orientation of the applied magnetic field .
@xmath136 corresponds to the @xmath42 direction .
the data are taken at an electric field of @xmath61kv / m and an external magnetic field of 25mt ( black symbols ) and 20mt ( red symbols ) .
the @xmath137-curves can be fitted using a cubic anisotropy field of @xmath138mt ( solid lines ) , assuming a misalignment of @xmath139 between the sample normal and the rotation axis . ] to determine the cubic anisotropy field @xmath140 , squid magnetometry measurements were carried out for different in - plane orientations of the applied magnetic field . to ensure a well defined ferroelastic single @xmath43 domain state in the bto substrate , an electric field of @xmath61kv /
m was applied across the hybrid structure .
the angular dependence of the magnetization projection @xmath141 along the external magnetic field @xmath142 is shown in fig .
[ fig : fig7 ] for @xmath143mt and @xmath144mt .
obviously , a four - fold symmetry is visible , which is a clear sign of a cubic magnetic anisotropy in the fe@xmath0o@xmath1 thin film .
the angular dependence @xmath141 measured at constant strain state @xmath112 can be simulated by means of eq . .
the best fit between experiment and simulation was obtained using a cubic anisotropy field of @xmath138mt ( cf .
solid lines in fig .
[ fig : fig7 ] ) . a misalignment of @xmath139 between the sample normal and the rotation axis
was assumed to account for the different magnetization values measured at @xmath145 , @xmath146 , and @xmath147 ( cf .
[ fig : fig7 ] ) . using the cubic anisotropy field and the elastic parameters determined above , the fe@xmath0o@xmath1/bto hybrid structure is well described except for the remaining unknown parameter @xmath134 , which characterizes the magnetoelastic coupling strength .
this parameter is the only fitting parameter for the simulation of the converse magnetoelectric effect in the following .
o@xmath1(38nm)/bto multiferroic hybrid for different magnetic field strengths .
the magnetic field was aligned in the film plane along the @xmath39 direction of the bto crystal and the temperature was set to 300k . for better visibility , the curves are vertically shifted with respect to each other .
the best fit between experiment and simulation was obtained by assuming a magnetoelastic coupling efficiency of 55% ( @xmath148 ) .
a misalignment of the sample surface with respect to the magnetic field of @xmath149 was assumed for the calculation . ] by calculating the magnetization values @xmath75 and @xmath74 according to eq . the converse magnetoelectric effects in the fe@xmath0o@xmath1(38nm)/bto hybrid ( cf .
[ fig : fig3 ] ) can be simulated for different magnetic field strengths assuming that ferroelastic domain reorientation is the dominant contribution [ cf .
the normalized magnetization projection along the @xmath39 direction as a function of the applied electric field obtained in this way is shown in fig . [
fig : fig8 ] , together with the experimental results taken from fig .
[ fig : fig3 ] . an excellent agreement between experiment and simulation could be obtained by assuming a magnetoelastic coupling efficiency of 55% ( @xmath148 ) .
the fact that this value is well below 100% indicates that the magnetoelastic behavior of the fe@xmath0o@xmath1 thin film is quite different from the bulk . nevertheless , as obvious from fig .
[ fig : fig8 ] , the variation of the magnetization with the applied electric field can be well simulated for magnetic fields @xmath150mt using the two - region model described in eq . .
at magnetic field values close to the coercive field , converse piezoelectric effects and magnetic domain effects come into play , leading to large changes of the magnetic coercivity as a function of the electric field.@xcite these changes can not be explained within the simple two - region model alone . in this case , the two - region model must be further extended to take into account piezoelectric effects as well as strain effects due to polarization rotations to simulate the electrical manipulation of the magnetic coercive fields .
we have investigated strain - mediated converse magnetoelectric effects in extrinsic multiferroic hybrid structures consisting of epitaxially grown ferrimagnetic fe@xmath0o@xmath1 thin films on ferroelectric bto substrates at room temperature . as evident from the time evolution of the intensity recorded by in - situ rheed ,
the growth of fe@xmath0o@xmath1 thin films onto bto substrates is highly disordered for the first few monolayers and then continuously evolves into a true layer - by - layer growth mode .
squid magnetometry measurements on these fe@xmath0o@xmath1 thin films reveal a saturation magnetization close to the bulk value as well as a pronounced verwey transition , which is a hallmark of the quality of fe@xmath0o@xmath1 .
furthermore , butterfly - shaped changes of the magnetization as a function of the applied electric field were observed at room temperature .
the use of miscut bto substrates , where only a single type of ferroelastic @xmath54 domains exist , allows us to quantitatively simulate the observed converse magnetoelectric effect by using a simple two - region model .
this model considers only effects caused by reorientations of ferroelastic @xmath43 domains into ferroelastic @xmath54 domains in the bto substrate . in the framework of this model ,
the magnetization values of those parts of the fe@xmath0o@xmath1 thin film elastically clamped onto ferroelastic @xmath43 domains and @xmath54 domains are considered as electric field independent quantities , which can be calculated using magnetoelastic theory . by comparing the results of the simulations to the experimental data ,
we demonstrate that the room - temperature converse magnetoelectric effects of fe@xmath0o@xmath1/bto - based multiferroic hybrid structures can be well understood on the basis of the two - region model for applied magnetic fields larger than the saturation field of fe@xmath0o@xmath1
. the detailed understanding of the converse magnetoelectric effect at room temperature in bto - based multiferroic hybrid structures opens the way to develop novel extrinsic multiferroic hybrids exhibiting giant electric - field induced changes of the magnetization .
we thank thomas brenninger for continuous technical support . financial support from the german research foundation ( projects gr 1132/13 and 14 ) and the german excellence initiative via the nanosystems initiative munich ( nim ) is gratefully acknowledged .
87ifxundefined [ 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 10.1088/0953 - 8984/17/2/l06 [ * * , ( ) ] link:\doibase 10.1038/nature05023 [ * * , ( ) ] link:\doibase 10.1038/nmat1805 [ * * , ( ) ] link:\doibase 10.1002/pssr.200802011 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/11/1/013021 [ * * , ( ) ] link:\doibase 10.1063/1.3624663 [ * * , ( ) ] link:\doibase 10.1038/ncomms1564 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.217202 [ * * , ( ) ] link:\doibase 10.1002/adma.201201004 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/24/33/333201 [ * * , ( ) ] link:\doibase 10.1080/09500830701194165 [ * * , ( ) ] link:\doibase 10.1063/1.2836410 [ * * , ( ) ] link:\doibase 10.1002/adma.200904326 [ * * , ( ) ] link:\doibase 10.1002/adma.201190024 [ * * , ( ) ] link:\doibase 10.1140/epjb / e2012 - 20675 - 4 [ * * ( ) , 10.1140/epjb / e2012 - 20675 - 4 ] link:\doibase 10.1038/nmat3130 [ * * , ( ) ] link:\doibase 10.1002/adma.200900278 [ * * , ( ) ] link:\doibase 10.1016/j.jmmm.2003.12.344 [ * * , ( ) ] link:\doibase 10.1038/nmat2184 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.75.054408 [ * * , ( ) ] link:\doibase 10.1038/nmat1886 [ * * , ( ) ] link:\doibase 10.1007/bf00540771 [ * * , ( ) ] link:\doibase 10.1143/jjap.40.4948 [ * * , ( ) ] link:\doibase 10.1126/science.1094207 [ * * , ( ) ] link:\doibase 10.1063/1.3069280 [ * * , ( ) ] link:\doibase 10.1063/1.2206121 [ * * , ( ) ] link:\doibase 10.1063/1.2844858 [ * * , ( ) ] link:\doibase 10.1063/1.3330890 [ * * , ( ) ] link:\doibase 10.1063/1.3073751 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.104405 [ * * , ( ) ] link:\doibase 10.1063/1.2370881 [ * * , ( ) ] link:\doibase 10.1088/0022 - 3727/46/5/055001 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.014408 [ * * , ( ) ] link:\doibase 10.1063/1.1578186 [ * * , ( ) ] link:\doibase 10.1063/1.2161424 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.144416 [ * * , ( ) ] link:\doibase 10.1063/1.1825075 [ * * , ( ) ] link:\doibase 10.1063/1.3200236 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.012410 [ * * , ( ) ] link:\doibase 10.1002/pssa.201026403 [ * * , ( ) ] link:\doibase 10.1063/1.3609237 [ * * , ( ) ] link:\doibase 10.1063/1.3684695 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.092108 [ * * , ( ) ] link:\doibase 10.1063/1.3628464 [ * * , ( ) ] link:\doibase 10.1063/1.3377923 [ * * , ( ) ] link:\doibase 10.1063/1.3675868 [ * * , ( ) ] link:\doibase 10.1038/ncomms2398 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.054410 [ * * , ( ) ] link:\doibase 10.1021/nl300849u [ * * , ( ) ] link:\doibase 10.1038/srep00258 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.104445 [ * * , ( ) ] link:\doibase 10.1002/adfm.200801907 [ * * , ( ) ] link:\doibase 10.1063/1.3354104 [ * * , ( ) ] @noop _ _ , ed .
( , ) link:\doibase 10.1002/pssa.2210340129 [ * * , ( ) ] link:\doibase 10.1103/physrevb.44.13319 [ * * , ( ) ] link:\doibase 10.1002/adma.200901381 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.144105 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.056403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.212404 [ * * , ( ) ] link:\doibase 10.1103/physrevb.12.5166 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.134432 [ * * , ( ) ] link:\doibase 10.1117/12.397845 [ * * , ( ) ] link:\doibase 10.1088/0953 - 2048/12/11/398 [ * * , ( ) ] link:\doibase 10.1016/s0304 - 8853(99)00706 - 4 [ * * , ( ) ] link:\doibase 10.1007/s00339 - 003 - 2105 - 9 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.134405 [ * * , ( ) ] link:\doibase 10.1063/1.1808497 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.155438 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.134436 [ * * , ( ) ] link:\doibase 10.1103/physrevb.40.9090 [ * * , ( ) ] link:\doibase 10.1103/physrevb.43.8461 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1038/nmat2826 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.177201 [ * * , ( ) ] link:\doibase 10.1107/s0021889877012849 [ * * , ( ) ] link:\doibase 10.1007/s100510050060 [ * * , ( ) ] link:\doibase 10.1038/35002022 [ * * , ( ) ] link:\doibase 10.1146/annurev.matsci.36.090804.100242 [ * * , ( ) ] link:\doibase 10.1107/s0021889803028395 [ *
* , ( ) ] link:\doibase 10.1063/1.1722606 [ * * , ( ) ] link:\doibase 10.1103/physrevb.69.140403 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevb.79.024432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.66.2798 [ * * , ( ) ] link:\doibase 10.1103/physrev.99.1210 [ * * , ( ) ] | the quantitative understanding of converse magnetoelectric effects , i.e. , the variation of the magnetization as a function of an applied electric field , in extrinsic multiferroic hybrids is a key prerequisite for the development of future spintronic devices .
we present a detailed study of the strain - mediated converse magnetoelectric effect in ferrimagnetic fe@xmath0o@xmath1 thin films on ferroelectric batio@xmath0 substrates at room temperature .
the experimental results are in excellent agreement with numerical simulation based on a two - region model .
this demonstrates that the electric field induced changes of the magnetic state in the fe@xmath0o@xmath1 thin film can be well described by the presence of two different ferroelastic domains in the batio@xmath0 substrate , resulting in two differently strained regions in the fe@xmath0o@xmath1 film with different magnetic properties .
the two - region model allows to predict the converse magnetoelectric effects in multiferroic hybrid structures consisting of ferromagnetic thin films on ferroelastic substrates . |
the topological properties of @xmath0-knots , also called genus one 1-bridge knots , have recently been investigated in several papers ( see @xcite ) .
these knots are very important in the light of some results and conjectures involving dehn surgery on knots ( see in particular @xcite and @xcite ) .
moreover , the strict connection between cyclic branched coverings of @xmath0-knots and cyclic presentations of groups have been pointed out in @xcite , @xcite and @xcite . roughly speaking ,
a @xmath0-knot is a knot which can be obtained by gluing along the boundary two solid tori with a trivial arc properly embedded .
a more formal definition follows .
a set of mutually disjoint arcs @xmath4 properly embedded in a handlebody @xmath5 is _ trivial _ if there exist @xmath6 mutually disjoint discs @xmath7 such that @xmath8 , @xmath9 and @xmath10 for all @xmath11 and @xmath12 .
let @xmath13 be a genus @xmath14 heegaard splitting of a closed orientable 3-manifold @xmath15 and let @xmath16 ; a link @xmath17 is said to be in _ @xmath6-bridge position _ with respect to @xmath18 if : ( i ) @xmath19 intersects @xmath18 transversally and ( ii ) @xmath20 and @xmath21 are both the union of @xmath6 mutually disjoint properly embedded trivial arcs .
the splitting is called a _
@xmath22-decomposition of @xmath19_. a link @xmath19 is called a _ @xmath22-link _ if it admits a @xmath22-decomposition .
note that a @xmath23-link is a link in @xmath24 which admits a @xmath6-bridge presentation in the usual sense .
so the notion of @xmath22-decomposition of links in 3-manifolds generalizes the classical bridge ( or plat ) decomposition of links in @xmath24 ( see @xcite ) . obviously , a @xmath25-link is a knot , for every @xmath26 .
therefore , a _ @xmath0-knot _
@xmath27 is a knot in a lens space @xmath2 ( possibly in @xmath24 ) which admits a @xmath0-decomposition @xmath28 where @xmath29 is an ( attaching ) homeomorphism which reverses the standard orientation on the tori ( see figure [ fig .
it is well known that the family of @xmath0-knots contains all torus knots ( trivially ) and all two - bridge knots ( see @xcite ) in @xmath24 . in this paper
we develop an algebraic representation of @xmath0-knots through elements of @xmath1 , the mapping class group of the twice punctured torus . in section [ ( 1,1 )
e mcg ] we establish the connection between the two objects . in section [ standard ] we prove that every @xmath0-knot in a lens space @xmath2 can be represented by an element of @xmath1 which is the composition of an element of a certain rank two free subgroup and of a standard element only depending on the ambient space @xmath2 .
this representation will be called `` standard '' . as a notable application , in sections [ torus ] and [ due ponti ]
we obtain standard representations for the two most important classes of @xmath0-knots in @xmath24 : the torus knots and the two - bridge knots .
moreover , applying certain results obtained in @xcite , we give explicit cyclic presentations for the fundamental groups of all cyclic branched coverings of torus knots of type @xmath3 , with @xmath30 and @xmath31 odd .
-decomposition . ] in what follows , the symbol @xmath2 will denote any lens space , including @xmath32 and @xmath33 .
moreover , homotopy and homology classes will be denoted with the same symbol of the representing loops .
let @xmath36 be a closed orientable surface of genus @xmath14 and let @xmath37 be a finite set of distinguished points of @xmath36 , called _
punctures_. we denote by @xmath38 the group of orientation - preserving homeomorphisms @xmath39 such that @xmath40 .
the _ punctured mapping class group _ of @xmath36 relative to @xmath41 is the group of the isotopy classes of elements of @xmath38 . up to isomorphism
, the punctured mapping class group of a fixed surface @xmath36 relative to @xmath41 only depends on the cardinality @xmath42 of @xmath41 .
therefore , we can simply speak of the _ @xmath42-punctured mapping class group _ of @xmath36 , denoting it by @xmath43 . moreover , for isotopy classes we will use the same symbol of the representing homeomorphisms .
the _ @xmath42-punctured pure mapping class group _ of @xmath36 is the subgroup @xmath44 of @xmath43 consisting of the elements pointwise fixing the punctures .
there is a standard exact sequence @xmath45 where @xmath46 is the symmetric group on @xmath42 elements .
a presentation of all punctured mapping class groups can be found in @xcite and in @xcite . in this paper
we are interested in the two - punctured mapping class group of the torus @xmath1 . according to previously cited papers , a set of generators for @xmath1
is given by a rotation @xmath47 of @xmath48 radians which exchanges the punctures and the right - handed dehn twists @xmath49 around the curves @xmath50 respectively , as depicted in figure [ fig .
since @xmath47 commutes with the other generators , we have @xmath51 . ]
the following presentation for @xmath52 has been obtained in @xcite : @xmath53 the group @xmath52 ( as well as @xmath1 ) naturally maps by an epimorphism to the mapping class group of the torus @xmath54 , which is generated by @xmath55 and @xmath56 .
so we have an epimorphism @xmath57 defined by @xmath58 and @xmath59 .
the group @xmath60 will play a fundamental role in our discussion . in order to investigate its structure ,
let us consider the two elements @xmath61 and @xmath62 , where @xmath63 is the right - handed dehn twist around the curve @xmath64 depicted in figure [ fig . 3 ] .
the effect of @xmath65 and @xmath66 is to slide one puncture ( say @xmath67 ) respectively along a meridian and along a longitude of the torus , as shown in figure [ fig .
observe that , since @xmath68 , we have @xmath69 . and @xmath66 . ]
the following result can be obtained from ( * ? ? ?
* th . 1 ) and ( * ? ? ?
5 ) by classical techniques .
[ nucleo ] the group @xmath60 is freely generated by @xmath61 and @xmath62 , where @xmath70 .
now , let @xmath71 be a @xmath0-knot with @xmath0-decomposition @xmath72 and let @xmath73 be a fixed orientation - reversing homeomorphism , then @xmath74 is an orientation - preserving homeomorphism of @xmath75 .
moreover , since two isotopic attaching homeomorphisms produce equivalent @xmath0-knots , we have a natural surjective map from the twice punctured mapping class group of the torus @xmath1 to the class @xmath76 of all @xmath0-knots @xmath77 if @xmath78 , then @xmath79 is a @xmath0-knot in the lens space @xmath80 @xcite , and therefore it is a knot in @xmath24 if and only if @xmath81 . as will be proved in section 3 , we have the following `` trivial '' examples : * if either @xmath82 or @xmath83 or @xmath84 , then @xmath79 is the trivial knot in @xmath85 ; * if @xmath86 , then @xmath79 is the trivial knot in @xmath24 .
moreover , it is possible to prove that if @xmath87 , then @xmath79 is the knot @xmath88 , where @xmath89 is any point of @xmath90 .
so , in this case , @xmath79 is a standard generator for the first homology group of @xmath85 .
every element @xmath91 of @xmath1 can be written as @xmath92 , @xmath93 , where @xmath94 .
since @xmath47 can be extended to a homeomorphism of the pair @xmath95 , the @xmath0-knots @xmath79 and @xmath96 are equivalent .
so , for our discussion it is enough to consider the restriction @xmath97
in this section we show that every @xmath0-knot @xmath71 admits a representation by the composition of an element in @xmath60 and an element which only depends on @xmath2 . a representation of this type
will be called `` standard '' .
note that a similar result , using a rank three free subgroup of @xmath1 , has been obtained in ( * ? ? ?
* theorem 3 ) .
first of all , we deal with trivial knots in lens space .
let @xmath98 be the subgroup of @xmath52 generated by @xmath55 and @xmath99 .
there exists a disk @xmath100 , with @xmath101 and @xmath102 , such that @xmath103 .
so any element of @xmath98 produces a trivial knot in a certain lens space . on the other hand ,
any trivial knot in a lens space admits a representation through an element of @xmath98 , as will be proved in proposition [ trivialknots ] .
we need a preparatory result .
[ matriciequivalenti ] let @xmath27 be a @xmath0-knot in @xmath2 .
then , for each @xmath104 such that @xmath105 there exists @xmath106 , with @xmath78 , such that @xmath107 .
let @xmath108 , with @xmath109 .
since @xmath110 , there exist @xmath111 such that @xmath112 and @xmath113 . if @xmath114 , we have @xmath115 , since @xmath116 can be extended to a homeomorphism of the pair @xmath95 .
moreover @xmath117 for integers @xmath118 such that @xmath119 and @xmath120 consider the sequence of equations of the euclidean algorithm ( with @xmath121 , @xmath122 ) : @xmath123 with @xmath124 .
the @xmath125 s are the coefficients of the continued fraction @xmath126 in the following we will use the notation @xmath127 $ ] .
[ trivialknots ] * the trivial knot in @xmath32 is represented by @xmath128 . * the trivial knot in @xmath33
is represented by @xmath129 .
* let @xmath118 be integers such that @xmath119 and @xmath120 .
if @xmath127 $ ] , then the trivial knot in the lens space @xmath2 is represented by @xmath130 since all the involved homeomorphisms belong to @xmath98 , all the knots are trivial .
it is easy to check ( see also @xcite ) that , for suitable @xmath104 , we have : @xmath131 since @xmath132 , @xmath133 , and @xmath134 , the statement is obtained .
now we can prove the result announced at the beginning of this section .
[ decomposition ] let @xmath27 be a @xmath0-knot in @xmath2
. then there exist @xmath135 such that @xmath107 , with @xmath136 . by lemma [ matriciequivalenti ]
, there exists @xmath91 , with @xmath137 , such that @xmath107 .
it suffices to define @xmath138 and @xmath139 .
a representation @xmath106 of a @xmath0-knot will be called _ standard _ if @xmath91 is of the type described in the previous theorem .
we point out that @xmath0-knots admit different ( usually infinitely many ) standard representations .
for example @xmath140 represents the trivial knot in , for all @xmath111 .
in this section we give a standard representation for all torus knots in @xmath24 .
let @xmath141 be a torus knot of type @xmath142 .
then @xmath143 , and we can assume that @xmath27 lies on the boundary @xmath144 of a genus one handlebody @xmath5 canonically embedded in @xmath24 .
the homology class of @xmath27 is @xmath145 , where @xmath146 and @xmath147 respectively denote a longitude and a meridian of @xmath148 . by slightly pushing ( the interior of ) an arc @xmath149 outside @xmath5 and @xmath150 inside @xmath5
, we obtain an obvious @xmath0-decomposition of @xmath27 .
observe that @xmath151 can be assumed without loss of generality ( see @xcite ) . in the next statement
@xmath152 denotes the integral part of @xmath153 .
[ torusknots ] the torus knot @xmath154 is the @xmath0-knot @xmath79 with : @xmath155 where @xmath61 and @xmath156 .
up to isotopy , we can suppose that the arc @xmath157 lies on @xmath158 , as in figure [ fig .
the arc @xmath159 can be transformed into an arc @xmath160 in such a way that @xmath161 is a trivial knot in @xmath24 , represented by the standard homeomorphism @xmath128 , via a suitable sequence of homeomorphisms @xmath66 and @xmath65 , according to the following algorithm .
consider the sequence of equations : @xmath162 . ]
where @xmath163 , for @xmath164 .
moreover , define @xmath165 .
so @xmath166 , for @xmath167 .
now define the homeomorphisms @xmath168 , for @xmath169 .
figure [ fig .
6 ] depicts the effect of @xmath66 and @xmath170 on @xmath159 . as a consequence
, the homeomorphism @xmath171 transforms the arc @xmath159 into the arc @xmath160 ( figure 7 shows the case @xmath172 ) , and therefore we have @xmath173 .
so @xmath174 represents the torus knot @xmath175 . for example , @xmath176 , with @xmath177 ( see figure [ fig . 7 ] ) . as a consequence ,
we obtain a cyclic presentation for the fundamental group for all cyclic branched coverings of a particular class of torus knots .
[ torus2 ] the fundamental group of the @xmath42-fold cyclic branched covering of the torus knot @xmath178 , with @xmath179 odd and @xmath180 , admits the cyclic presentation @xmath181 , where @xmath182 is equal to @xmath183 ( subscripts are taken modulo @xmath42 ) .
let @xmath184 . from theorem [ torusknots ]
we have @xmath185 with @xmath186 . applying ( * ? ? ?
* proposition 1 ) , we obtain @xmath187 with @xmath188
. then @xmath189 . since , up to equivalence , @xmath190 , we have @xmath191 .
we set @xmath192 , therefore @xmath193 with @xmath194 .
the statement derives from a straightforward application of ( * ? ? ?
* theorem 7 ) , for example , the fundamental group of the @xmath42-fold cyclic branched covering of @xmath172 admits the cyclic presentation @xmath181 , where @xmath195
in this section we give a standard representation for all two - bridge knots in @xmath24 .
let @xmath196 be a non - trivial two - bridge knot in @xmath24 of type @xmath197 .
then we can assume @xmath198 , @xmath199 odd , @xmath6 even and @xmath200 , without loss of generality ( see ( * ? ? ?
it is known that @xmath196 admits a conway presentation with an even number of even parameters @xmath201 $ ] ( see figure [ figconway ] ) , satisfying the following relation : @xmath202 [ twobridge ] the two - bridge knot @xmath203 having conway parameters @xmath201 $ ] is the @xmath0-knot @xmath79 with : @xmath204 where @xmath205 is the right - handed dehn twist around the curve @xmath206 depicted in figure [ figtwobridge ] .
now we show that @xmath205 ( note that no disk bounded by @xmath206 and properly embedded in @xmath5 is disjoint from @xmath159 ) . referring to figure [ fig .
17 ] , the following `` lantern '' relation @xmath208 holds ( see @xcite ) .
so we obtain @xmath209 and therefore @xmath210 .
since @xmath211 we have @xmath212 now , using the relations of ( [ presentation ] ) we get @xmath213 _ acknowledgements . _ the authors would like to thank sylvain gervais and andrei vesnin for their helpful suggestions .
we also would like to thank the referee for his valuable comments and remarks .
work performed under the auspices of the g.n.s.a.g.a . of i.n.d.a.m .
( italy ) and the university of bologna , funds for selected research topics .
j. r. parker and c. series , the mapping class group of the twice punctured torus . to appear in _ proceedings of the conference `` groups : combinatorial and geometric aspects '' _ ( bielefeld , 15 - 23 august 1999 ) , london mathematical society lecture note series . | we develop an algebraic representation for @xmath0-knots using the mapping class group of the twice punctured torus @xmath1 . we prove that every @xmath0-knot in a lens space @xmath2 can be represented by the composition of an element of a certain rank two free subgroup of @xmath1 with a standard element only depending on the ambient space . as notable examples ,
we obtain a representation of this type for all torus knots and for all two - bridge knots .
moreover , we give explicit cyclic presentations for the fundamental groups of the cyclic branched coverings of torus knots of type @xmath3 .
+ + _ mathematics subject classification 2000 : _ primary 57m05 , 20f38 ; secondary 57m12 , 57m25 .
+ _ keywords : _
@xmath0-knots , heegaard splittings , mapping class groups , two - bridge knots , torus knots . |
in this paper , we study the cauchy problem for the hartree equation @xmath3 here @xmath4 is a nonlinear function of hartree type for @xmath5 , where @xmath6 denotes the convolution in @xmath7 . in practice , we use the integral formula of @xmath8 @xmath9 where @xmath10 .
if the solution @xmath11 of @xmath8 has sufficient smoothness and decay at infinity , it satisfies two conservation laws : @xmath12 as explained in @xcite , the energy is also conserved for the energy solutions @xmath13 . from the viewpoint of the fractional integral ,
we rewrite the equation @xmath8 as @xmath14 for dimension @xmath15 , the exponent @xmath16 is the unique exponent which is energy critical in the sense that the natural scale transformation @xmath17 leaves the energy invariant , in other words , the energy @xmath18 is a dimensionless quantity .
the cauchy problem of the hartree equation has been intensively studied ( [ 4 - 10 ] , [ 15 , 16 , 18 , 19 ] . with regard to the global well - posedness and scattering results , they all dealt with the @xmath0-subcritical case @xmath19 in the energy space or some weighted spaces . in @xcite , we obtained the small data scattering result for the @xmath0-critical case in the energy space .
for the large initial data for the @xmath0-critical case @xmath20 in the energy space , the argument in @xcite can not yield the global well - posedness , even with the conservation of the energy @xmath21 , because the time of existence given by the local theory depends on the profile of the data as well as on the energy . concerning the @xmath0-subcritical case @xmath22 , using the method of morawetz and strauss @xcite , j. ginibre and g. velo @xcite developed the scattering theory in the energy space , where they exploited the properties of @xmath23 and
obtain the usual morawetz estimate @xmath24 later , k. nakanishi @xcite exploited the properties of @xmath25 and used a certain related sobolev - type inequality to obtain a new morawetz estimate @xmath26 which was independent of the nonlinearity . in this paper , we deal with the cauchy problem of the hartree equation with the large data for the @xmath0-critical case @xmath27 .
inspired by the approach of bourgain @xcite and tao @xcite in the case of the @xmath0-critical schrdinger equation with the local nonlinear term , we obtain the global well - posedness and scattering results for the hartree equation for the large radial data in @xmath0 .
the new ingredient is that we take advantage of the following localized estimate for the first time @xmath28 to rule out the possibility of energy concentration , instead of the classical morawetz estimate @xmath29 due to the nonlinear term .
our main result is the following global well - posedness result in the energy space . [
main ] let @xmath15 , and @xmath30 be radial .
then there exists a unique global solution @xmath31 to @xmath32 where @xmath33 and on each compact time interval @xmath34 $ ] , we have @xmath35\times \mathbb{r}^n ) } \leq c(\big\| \varphi \big\|_{\dot{h}^1}).\ ] ] as the right hand side of @xmath36 is independent of @xmath37 , we can obtain the global spacetime estimate . as a direct consequence of the global @xmath38 estimate , we have scattering , asymptotic completeness , and uniform regularity .
let @xmath39 be radial and have finite energy .
then there exists finite energy solutions @xmath40 to the free schrdinger equation @xmath41 such that @xmath42 furthermore , the maps @xmath43 are homeomorphisms from @xmath44 to @xmath44 . finally , if @xmath45 for some @xmath46 , then @xmath47 for all time @xmath48 , and one has the uniform bounds @xmath49 the paper is organized as follows . in section @xmath50
, we introduce notations and the basic estimates ; in section @xmath51 , we derive the local mass conservation and morawetz inequality ; in section @xmath52 , we discuss the local theory for @xmath53 ; in section @xmath54 , we obtain the perturbation theory ; finally , we prove the main theorem in section @xmath55 .
we will often use the notations @xmath56 and @xmath57 to denote the estimate @xmath58 for some @xmath59 .
the derivative operator @xmath60 refers to the space variable only .
we also occasionally use subscripts to denote the spatial derivatives and use the summation convention over repeated indices .
we define @xmath61 , @xmath62 , @xmath63 ; for @xmath64 we denote by @xmath65 the dual exponent , that is , @xmath66 . for any time interval @xmath67
, we use @xmath68 to denote the mixed spacetime lebesgue norm @xmath69 with the usual modifications when @xmath70 .
when @xmath71 , we abbreviate @xmath72 by @xmath73
. we use @xmath10 to denote the free group generated by the free schrdinger equation @xmath74 .
it can commute with derivatives , and obeys the inequality @xmath75 , @xmath76 .
we say that a pair @xmath77 is admissible if @xmath78 and @xmath79 for a spacetime slab @xmath80 , we define the _ strichartz _ norm @xmath81 by @xmath82 and define @xmath83 by @xmath84 when @xmath85 , the spaces @xmath86 and @xmath87 are banach spaces , respectively . based on the above notations
, we have the following _ strichartz _ inequalities @xcite , @xcite[lemstri1 ] let @xmath11 be an @xmath88 solution to the schrdinger equation @xmath8 .
then @xmath89 for any @xmath90 and any admissible pairs @xmath77 .
the implicit constant is independent of the choice of interval @xmath67 . from sobolev embedding
, we have [ lemstri2 ] for any function @xmath11 on @xmath80 , we have @xmath91 where all spacetime norms are on @xmath80 . for convenience ,
we introduce two abbreviated notations . for a time interval @xmath67 , we denote @xmath92 [ nle ] let @xmath93 , where @xmath33 . for any time interval @xmath67 and @xmath90 , we have @xmath94 * proof : * by strichartz estimates , hardy - littlewood - sobolev inequality and hlder inequality , we have @xmath95
in this section , we will prove two useful estimates .
one is a local mass conservation estimate and the other is a morawetz inequality , which appears in morawetz identity . the local mass conservation estimate is used to control the flow of mass through a region of space , and the morawetz inequality is used to prevent concentration .
we recall a local mass conservation law that has appeared in @xcite , @xcite and @xcite . for completeness
, we give the sketch of the proof .
let @xmath96 be a bump function supported on the ball @xmath97 that equals @xmath98 on the ball @xmath99 .
observe that if @xmath11 is a finite energy solution of @xmath53 , then @xmath100 we define @xmath101 differentiating the above quantity with respect to time , we obtain by the integration by parts @xmath102 hence , we have @xmath103 this implies that if the local mass @xmath104 is large for some time @xmath48 , then it can also be shown to be similarly large for nearly time @xmath48 , by increasing the radius @xmath105 if necessary to reduce the rate of change of the mass .
on the other hand , from sobolev and hlder inequalities , we have @xmath106 this gives the control of mass in small volumes . to prevent the concentration of the energy , we need a morawetz estimate .
the morawetz estimate is based on some integral identity derived by variation of the lagrangian .
we define @xmath107 by latexmath:[\[2\ell(u)=\langle iu_t , u\rangle + |\nabla u|^2 + \frac12 associated to the equation @xmath8 . from the definition of the variation of the functional @xmath109
, we have @xmath110 using this identity together with @xmath111 and @xmath112 , we obtain the following formula : @xmath113 as a consequence of the above dilation identity , we have the following morawetz estimate , which plays an important role in our proof . [ mor ]
let u be a solution to @xmath53 on a spacetime slab @xmath114 .
then for any @xmath115 , we have @xmath116 where @xmath117
we have @xmath118 * proof : * we define @xmath119 , then @xmath120 and @xmath121 where we use the symmetry of @xmath122 and @xmath123 .
let @xmath124 and let @xmath125 be a bump function adapted to the ball @xmath126 which equals 1 on the ball @xmath127 .
we set @xmath128 . for @xmath127
, we have @xmath129 and for @xmath130 , we have bounds @xmath131 thus we have @xmath132 where @xmath133 , @xmath134 meanwhile @xmath135 @xmath136 moreover , from sobolev and hlder inequalities , we have @xmath137 so if we integrate by parts on a time interval i and take @xmath138 , we obtain @xmath139 for @xmath140 . the proof is completed .
in this section , we develop a local well - posedness and blow - up criterion for the @xmath0-critical hartree equation .
first , we have [ lwp ] let @xmath141 , and @xmath67 be a compact time interval that contains @xmath142 such that @xmath143 for a sufficiently small absolute constant @xmath144 .
then there exists a unique strong solution to @xmath53 on @xmath114 such that @xmath145 * proof : * the proof of this proposition is standard and based on the contraction mapping arguments .
we define the solution map to be@xmath146 then @xmath147 is a map from @xmath148 with the metric @xmath149onto itself because @xmath150 @xmath151 it suffices to prove @xmath147 is a contraction map .
let @xmath11 , @xmath152 , then @xmath153 by lemma [ nle ] , we have @xmath154 in the same way , we have @xmath155 as long as @xmath125 is chosen sufficiently small .
then the contraction mapping theorem implies the existence of the unique solution to ( [ equm ] ) on i. next , we give the blow - up criterion of the solutions for @xmath53 . the usual form is similar to those in @xcite , @xcite , which is in the form of a maximal interval of existence . for convenience ,
we obtain [ blowup ] let @xmath30 , and let @xmath11 be a strong solution to @xmath53 on the slab @xmath156 such that @xmath157 then there exists @xmath158 such that the solution @xmath11 extends to a strong solution to @xmath53 on the slab @xmath159\times \mathbb{r}^n$ ] . * proof : * by the absolute continuity of integrals , there exists a @xmath160 , such that @xmath161 then by lemma [ nle ] , we have @xmath162therefore @xmath163 now we write @xmath164 then @xmath165 by the absolute continuity of integrals again , there exists a @xmath166 , such that @xmath167 thus we may apply proposition [ lwp ] on the interval @xmath168 $ ] to complete the proof . in other words , this lemma asserts that if @xmath169 is the maximal interval of existence and @xmath170 , then @xmath171
in this section , we obtain the perturbation for hartree equation , which shows that the solution can not be large if the linear part of the solution is not large .
this is an analogue of lemma @xmath172 in @xcite , and later , killip , visan and zhang @xcite gave the similar perturbation result for the schrdinger equation with the quadric potentials .
[ pertl ] let @xmath11 be a solution to @xmath53 on @xmath173 $ ] such that @xmath174 where @xmath125 is sufficiently small constant depending on the norm of the initial data , then @xmath175 where @xmath176 for @xmath177 .
* proof : * from strichartz estimate and lemma @xmath178 , we obtain @xmath179 if @xmath125 is sufficiently small , we have the first claim @xmath180 as for the second claim , we give the proof for @xmath181 , the case @xmath182 is similar . using strichartz estimate and lemma @xmath178 again
, we have @xmath183 therefore , the second claim follows by the triangle inequality and choosing @xmath125 sufficiently small .
in this section , we give the proof of theorem @xmath184 .
the new ingredient is that we first take advantage of the the estimate of the term @xmath185 in the localized morawetz identity to rule out the possibility of energy concentration , which is independent of the nonlinear term . for the schrdinger equation ,
tao @xcite used the classical morawetz estimate , which depends on the nonlinearity , to prevent the concentration . for readability , we first take some constants @xmath186 which come from several constraints in the rest of this section .
all implicit constants in this section are permitted to depend on the dimension @xmath187 and the energy .
fix @xmath188 , @xmath189 $ ] , @xmath11 .
we may assume that the energy is large , @xmath190 , otherwise the claim follows from the small energy theory @xcite . from the boundedness of energy and sobolev embedding
, we can obtain @xmath191 for all @xmath192 $ ] .
assume that the solution @xmath11 already exists on @xmath189 $ ] .
by lemma @xmath193 , it suffices to obtain a priori estimate @xmath194 ) } \leq o(1),\ ] ] where @xmath195 is independent of @xmath196 , @xmath197 .
we may assume that @xmath198 ) } \geq 2\eta,\ ] ] otherwise it is trivial .
we divide @xmath199 $ ] into @xmath200 subintervals @xmath201 $ ] for some @xmath202 such that @xmath203 where @xmath125 is a small constant depending on the dimension @xmath187 and the energy . as a consequence , it suffices to estimate the number @xmath200 . from @xmath209
, we obtain the upper bound on the number of exceptional intervals , @xmath210 .
we may assume that there exist unexceptional intervals , otherwise the claim would follow from this bound and @xmath211 .
therefore , it suffices to compute the number of unexceptional intervals .
* proof : * by time translation invariance and scale invariance , we may assume that @xmath215 $ ] .
we subdivide @xmath206 further into @xmath216 $ ] and @xmath217 $ ] .
by @xmath211 and the pigeonhole principle and time reflection symmetry if necessary , we may assume that @xmath218 ) } \geq \frac{\eta}{4}. \endaligned\ ] ] thus by lemma @xmath219 , we have @xmath220)}\geq \frac{\eta}{8}.\ ] ] on the other hand , by @xmath211 , lemma @xmath224 , lemma @xmath178 and lemma @xmath219 , we have @xmath225 ) } & \lesssim \big\|u\big\|^2_{x([\frac12 , 1 ] ) } \big\|u\big\|_{w([\frac12 , 1])}\\ & \lesssim \eta^2\big\|u\big\|_{\dot{s}^1([\frac12 , 1 ] ) } \lesssim \eta^2 .
\endaligned\ ] ] thus the triangle inequality implies that @xmath226 ) } & \geq \frac{1}{100 } \eta , \endaligned\ ] ] provided @xmath125 is chosen sufficiently small .
hence , if we define @xmath227 then we have @xmath228 ) } \geq \frac{1}{100 } \eta.\ ] ] next , we estimate the upper bound on @xmath229 .
we have by @xmath230 and the triangle inequality @xmath231 ) } & \leq \big\|u\big(t-\frac12\big)u(\frac12 ) \big\|_{\dot{s}^1([\frac12 , 1 ] ) } + \big\|u(t - t_-)u(t_- ) \big\|_{\dot{s}^1([\frac12 , 1 ] ) } \\ & \qquad + \big\| \int^\frac12_{0}u(t - s)f(u(s))ds \big\|_{\dot{s}^1([\frac12 , 1 ] ) } \\ & \lesssim \big\|u\big\|_{\dot{h}^1 } + \big\|u\big\|^2_{x([0 , \frac12 ] ) } \big\|u\big\|_{w([0 , \frac12 ] ) } \\ & \lesssim \big\|u\big\|_{\dot{h}^1 } + \big\|u\big\|^2_{x([0 , \frac12 ] ) } \big\|u\big\|_{\dot{s}^1([0 , \frac12 ] ) } \\ & \lesssim 1 .
\endaligned\ ] ] where we use strichartz estimate , @xmath211 and lemma @xmath219 .
* proof : * by the chain rule , hlder inequality and sobolev embedding , we have @xmath238 it follows by @xmath239 @xmath240\times \mathbb{r}^n ) } \leq \sup_{t\in [ \frac12 , 1 ] } \int^0_{t_- } \frac{1}{|t - s|^2 } \big\| \nabla f(u)(s)\big\|_{l^{\frac{2n}{n+4}}_x } ds \lesssim 1 .
\endaligned\ ] ] from @xmath241 and interpolation , we have @xmath242\times \mathbb{r}^n ) } & \leq \big\| \nabla v\big\|^{2/3}_{l^{\infty}_tl^{\frac{2n}{n-4}}_x([\frac12 , 1]\times \mathbb{r}^n ) } \big\| \nabla v\big\|^{1/3}_{l^{\infty}_tl^2_x([\frac12 , 1]\times \mathbb{r}^n ) } \\ & \lesssim 1 .
\endaligned\ ] ] from the fundamental theorem of calculus , we have @xmath243\times \mathbb{r}^n ) } & \lesssim |h| . \endaligned\ ] ] this implies @xmath244\times \mathbb{r}^n ) } & \leq \int \chi(y)\big\| v(t , x+\eta^{c_2}y ) -v(x ) \big\|_{l^{\infty}_tl^{\frac{6n}{3n-8}}_x([\frac12 , 1]\times \mathbb{r}^n ) } dy \\ & \lesssim \int \chi(y ) |\eta^{c_2}y|dy\\ & \lesssim \eta^{c_2}. \endaligned\ ] ] hence from hlder inequality , we obtain @xmath245 ) } \lesssim \big\|v - v_{av}\big\|_{l^{\infty}_tl^{\frac{6n}{3n-8}}_x([\frac12 , 1]\times \mathbb{r}^n ) } \lesssim \eta^{c_2}. \endaligned\ ] ] on the other hand , by hlder inequality , young inequalities and @xmath241 , we have @xmath250\times \mathbb{r}^n ) } & \lesssim \big\|v_{av}\big\|_{l^{\infty}_tl^{\frac{2n}{n-2}}_x([\frac12 , 1]\times \mathbb{r}^n ) } \\ & \lesssim \big\|v\big\|_{l^{\infty}_tl^{\frac{2n}{n-2}}_x([\frac12 , 1]\times \mathbb{r}^n ) } \\ & \lesssim 1 . \endaligned\ ] ] interpolating with @xmath251 gives @xmath252\times \mathbb{r}^n ) ) } & \gtrsim \big\|v_{av}\big\|^{\frac{3n-6}{2}}_{x([\frac12 , 1 ] ) } \big\|v_{av}\big\|^{-\frac{3n-8}{2}}_{l^{\frac{2(3n-8)}{n-2}}_tl^{\frac{2n}{n-2}}_x([\frac12 , 1]\times \mathbb{r}^n)}\\ & \gtrsim \eta^{\frac{3n-6}{2}}. \endaligned\ ] ] thus there exists @xmath253\times \mathbb{r}^n$ ] such that @xmath254 hence , by cauchy - schwarz inequality , we have @xmath255 next ,
we use the radial assumption to show that the bubble of mass concentration must occur at the spatial origin . in the forthcoming paper
, we shall use the interaction morawetz estimate with the frequency localized @xmath270 almost - conservation law to rule out the possibility of the energy concentration at any place and deal with the non - radial data .
the corresponding results for the schrdinger equation with local nonlinearity , please see @xcite , @xcite and @xcite .
* proof : * if @xmath272 in proposition @xmath246 is within @xmath273 of the origin , then the result follows immediately . otherwise by the radial assumption , there would be at least @xmath274 many distinct balls each containing at least @xmath275 amount of mass . by hlder inequality , this implies @xmath276 that is @xmath277 [ control ] assume that the solution @xmath11 is spherically symmetric .
for any interval @xmath282 $ ] and @xmath67 be a union of consecutive unexceptional intervals @xmath206 .
then @xmath283 and moreover , there exists a @xmath284 such that @xmath285 * proof : * for any unexceptional interval @xmath206 , from hlder inequality and corollary @xmath286 , we have @xmath287 therefore @xmath288 we integrate this over each unexceptional interval @xmath206 and sum over @xmath284 , @xmath289 the second claim follows from the first and the fact that @xmath290 this completes the proof . [ cascade ] let @xmath67 be an interval tiled by finitely many intervals @xmath291 .
suppose that for any continuous family @xmath292 of the unexceptional intervals , there exists @xmath293 such that @xmath294 for some small @xmath295 .
then there exist @xmath296 distinct indices @xmath297 such that @xmath298 and for any @xmath299 , @xmath300 hold for @xmath301 . by hypothesis
, @xmath67 contains at least one interval of length @xmath302 .
all intervals with length larger than @xmath303 belong to the first generation . by the total measure
, we see that there are at most @xmath304 intervals in the first generation .
removing there intervals from @xmath67 leaves at most @xmath305 gaps , which are tiled by intervals @xmath206 .
each iteration of the algorithm removes at most @xmath304 many intervals and produces at most @xmath305 gaps .
suppose that there are @xmath308 consecutive unexceptional intervals initially , and we perform at most @xmath309 times iterations
. then the number @xmath309 obeys @xmath310 which leads to the claim @xmath296 .
let @xmath311 be the interval obtained after @xmath312 iterations and @xmath313 be any interval in @xmath311 .
for @xmath314 , let @xmath315 be the @xmath316-generation gap which contains the @xmath313 , and assign the @xmath317 be any ith - generation interval which is contained in @xmath315 ( see figure @xmath98 ) .
by the construction , for any @xmath299 , we have @xmath318 for all @xmath319 .
let us denote the union of these consecutive unexceptional intervals by @xmath67 . by corollary @xmath340 , the hypotheses of proposition @xmath341 are satisfied with @xmath342 and so we can find a cascade of @xmath309 intervals and they satisfied the hypotheses of proposition @xmath343 .
the bound on @xmath309 implies the bound on @xmath308 , namely , @xmath344 j. ginibre and g. velo , scattering theory in the energy space for a class of hartree equations , nonlinear wave equations ( providence , ri , 1998 ) , 29 - 60 , contemp . math .
, 263 , amer .
soc . , providence , ri , 2000 . | we consider the defocusing , @xmath0-critical hartree equation for the radial data in all dimensions @xmath1 .
we show the global well - posedness and scattering results in the energy space .
the new ingredient in this paper is that we first take advantage of the term @xmath2 in the localized morawetz identity to rule out the possibility of energy concentration , instead of the classical morawetz estimate dependent of the nonlinearity . |
according to the two - fluid model , the flow in superfluids at finite temperatures involves separate motions of the normal and superfluid components .
these components have densities @xmath1 and @xmath2 and velocities @xmath3 and @xmath4 , respectively . in this paper
we are concerned only with processes which happen at nearly constant temperature . in such simplified cases
the equations of the two - fluid hydrodynamics can be understood to originate from the euler equation for the superfluid component and from the navier - stokes equation for the normal component , with the mutual friction force providing the interaction between the two components @xcite .
the mutual friction interaction is mediated by quantized vortex lines . when a straight vortex moves with velocity @xmath5 in a superfluid it experiences two forces ( fig .
[ friction]a ) : the first one is the magnus lift force from the superfluid component @xmath6 . here
@xmath7 is the quantum of circulation and @xmath8 is a unit vector along the vortex core .
the second force arises from the scattering of the quasiparticles , which form the normal component , from the vortex cores .
it has the components @xmath9 , where @xmath10 and @xmath11 ( the subscript @xmath12 denotes projections to the plane perpendicular to the vortex core ) .
vortex mass can usually be ignored and the equation of motion reduces simply to the force balance , @xmath13 . solving this for @xmath5 yields @xmath14 here
@xmath15 and @xmath16 are the mutual friction parameters , characteristics of the superfluid , which describe the interaction of thermal quasiparticles with the vortex cores . inserting @xmath5 from eq . to the expression for @xmath17 we find that in the mutual friction force @xmath18 the term with @xmath19 is always directed opposite to the so - called counterflow velocity @xmath20 and thus leads to dissipation , while the term with @xmath21 is non - dissipative .
the mutual friction force can be averaged over vortex lines if they are locally parallel to each other . in this case
we get a force per unit mass of the superfluid component @xmath22 -\alpha ' ( \vels - \veln ) \times \oms~. \label{eq : fns}\ ] ] here the vorticity @xmath23 and @xmath24 is a unit vector in the direction of @xmath25 .
note that the friction coefficients in eq .
are the same as in eq . only because complete local polarization of vortex lines is assumed .
later we will introduce effective mutual friction parameters to describe phenomenologically situations where this is not the case .
if the vortex line is curved , then its tension @xmath26 , originating from the quantized superflow around the core , leads to an additional force acting perpendicular to the core .
balancing it against the magnus force and averaging over the vortex lines we get the contribution @xmath27
where @xmath28 , @xmath29 is the intervortex distance and @xmath30 is the core size . including the tension force to the balance of forces acting on a vortex modifies also the expression for the mutual friction force , but we will not consider this here . since @xmath31 has the small prefactor @xmath32 , in many cases it can be neglected . however , there are situations where it is essential , like in the kelvin - wave instability @xcite or in the angular momentum balance of the propagating vortex front @xcite , considered below .
inserting the @xmath33 and @xmath31 forces to the euler equation , we get the coarse - grained hydrodynamic equation for the superfluid component @xmath34 where @xmath35 is the chemical potential . to account for the normal component , one inserts the @xmath33 force with a negative sign to the navier - stokes equation , where the viscous term then damps the drive from mutual friction . by comparing the magnitudes of these two terms ,
it is possible to work out a criterion , when the normal component can be considered not to respond to the motion of the superfluid component in the hydrodynamic regime @xcite : @xmath36 .
here @xmath37 is the kinematic viscosity of the normal component .
the fermi superfluid @xmath0he has a large value of @xmath38 , which satisfies this criterion and allows us to consider the normal component as stationary .
this is quite unlike bose superfluids where usually @xmath39 @xcite . at low temperatures in the ballistic regime of quasiparticle
motion one might expect that the drag from the quasiparticle gas on the vortex lines depends strongly on the ratio of the intervortex distance to the scattering cross - section in the quasiparticle - vortex scattering process .
it turns out that at the typical experimental values of @xmath40 mm this consideration is not important .
numerical calculations at higher vortex densities show , however , some non - trivial effects in quasiparticle motion @xcite .
( _ a _ ) forces acting on a vortex moving with respect to the superfluid and normal components : magnus force @xmath17 and reaction of the normal component @xmath41 , with @xmath42 .
( _ b _ ) measured values of the mutual friction coefficients as a function of temperature .
the main panel shows @xmath19 and @xmath21 at @xmath43 as determined from the damping of an oscillating diaphragm @xcite .
insert _ shows values of @xmath19 at @xmath44 as derived from the decay of laminar spin - down from rotation @xcite .
the line is a fit to the theoretical dependence @xmath45 with @xmath46 as fitting parameter and the bulk energy gap @xmath47 at 29bar pressure .
[ friction ] ] values of the mutual friction coefficients @xmath19 and @xmath21 are thus important for understanding the dynamic properties of the superfluid . in the isotropic
b phase of superfluid @xmath0he they have been measured in wide temperature and pressure ranges by the manchester group @xcite ( fig .
[ friction]b ) .
the results are in fair agreement with the theory of mutual friction which includes the iordanskii force from the scattering of bulk thermal quasiparticles by the flow field around the vortex @xcite and the kopnin force originating from the spectral flow of the vortex - core - bound fermions and their scattering from the bulk quasiparticles @xcite .
the energy gap @xmath48 in the spectrum of bulk quasiparticles is the main source for the fast variation of mutual friction with temperature .
the original measurements left an uncertainty @xcite whether the accepted value of the gap should be renormalized to fit the experimental data .
new measurements of friction to lower temperatures @xcite leave no doubt that the bulk @xmath48 properly describes the temperature dependence , in accordance with theory ( insert in fig .
[ friction]b ) .
an unsolved problem , both experimentally and theoretically , is the behavior of mutual friction at very low temperatures , below about @xmath49 . there is a prediction of a new friction mechanism @xcite , which depends on the acceleration of vortex lines , but not on temperature and thus allows for a finite dissipation even at zero temperature through the emission of non - thermal quasiparticles from the vortex cores
there are observations of temperature - independent contributions to friction @xcite , but it is not yet clear , whether they can be attributed to a mechanism of this kind .
to establish the ranges of stability for laminar and turbulent flows in superfluids with a stationary normal component , one can apply the same dimensional arguments as used in classical hydrodynamics , by introducing the reynolds number re .
the classical reynolds number @xmath50 has the physical meaning of being the ratio of the magnitudes of the inertial and dissipative terms in the navier - stokes equation . here
@xmath51 is the typical velocity of the flow and @xmath52 is the spatial scale .
turbulent motion is observed for @xmath53 . in superfluids , starting from eq . for the superfluid velocity , we put @xmath54 and omit the tension term so that we get the equation @xmath55 here the inertial term is the same as in the navier - stokes equation , but renormalized with the prefactor @xmath56 .
the dissipative term , proportional to @xmath19 , has a completely different structure .
in fact , both the inertial and dissipative terms have the same scaling @xmath57 and thus their ratio becomes an internal parameter of the superfluid , independent of the velocity and the geometry of the flow @xcite @xmath58 this parameter has the same physical meaning as the reynolds number in classical hydrodynamics and thus we call it the superfluid reynolds number
. in analogy to classical turbulence we can expect superfluid flow to be turbulent when @xmath59 and laminar in the opposite limit .
measurements show , however , that the value of @xmath60 at the transition from laminar to turbulent dynamics , which we call the onset value @xmath61 , is actually for many types of flow much closer to unity than in classical systems .
thus we write the criterion for superfluid turbulence as @xmath62 it is possible to approach the limits for quantum turbulence also from a microscopic point of view .
the essential processes are vortex reconnections , growth of vortex rings , and kelvin waves on vortex lines .
one can estimate within some model when these processes proliferate or when they are suppressed by mutual friction .
different models have been considered : the pseudo - vinen equation @xcite , damping of kelvin waves on vortex bundles @xcite , and the reorientation of a vortex ring in circulating flow @xcite . in all cases
the same criterion eq . is obtained .
he can be stabilized in the central section of the sample with a magnetic field .
the upper b - phase section is about 1 cm shorter than the lower . in rotation
the ab interface becomes unstable at a well - defined velocity and a bundle of @xmath63 closely packed vortex loops is injected from the a into the b phase , as shown for the upper ab interface . in the lower b - phase section an alternative way is illustrated for putting vortices in applied flow .
it uses slowly evolving remanent vortices which terminate at the cylinder wall .
( _ b _ ) turbulent evolution of the injected vortex loop(s ) .
( _ c _ ) the number of vortices within the nmr pick - up coil is determined from the height of the so - called counterflow peak in the nmr spectrum of @xmath0he - b @xcite .
[ expsetup ] ] when the vortex tension is included in eq .
, two additional dimensionless parameters can be defined .
the ratio of the inertial term to the tension term is @xmath64 an alternative , but similar definition of @xmath65 is often used .
owing to the functional similarity of this expression to the classical reynolds number , this combination is also called a superfluid reynolds number .
( to add to the confusion , in superfluids with a mobile normal component even more superfluid reynolds numbers can be defined . )
the physical meaning of @xmath66 is different , however .
the condition @xmath67 coincides with the feynman criterion which specifies when the first vortex line becomes stable in the flow @xcite .
when @xmath68 the flow can support many vortex lines .
this is a necessary condition for the applicability of the coarse - grained hydrodynamic equations .
here we consider only cases where this condition is fulfilled .
rad / s , where some vortices connect to the cylindrical wall , owing to the residual inclination ( of @xmath69 ) of the long cylinder with respect to the rotation axis .
when @xmath70 is rapidly ramped to @xmath71 , the curved vortex ends become unstable in the applied flow and generate new vortices , first in slow single - vortex processes , and finally in a turbulent burst , when the newly created vortices interact turbulently .
the number of vortices within the top and bottom nmr coils ( top panel ) are obtained from the height of the counterflow ( cf ) peak , which is continuously scanned , as shown in the bottom panel . in laminar
flow the height of the cf peak does not decrease with time , while here it collapses suddenly . in this measurement
there is not a - phase barrier layer ; thus from the arrival times of the vortex fronts to the nmr coils one obtains the moment when the turbulent burst happened ( marked with an arrow in the upper panel ) and the location of the burst , which in this measurement happens to be 4 cm above the bottom nmr coil .
[ slowformation ] ] finally , the ratio of the tension term to the dissipation term @xmath72 is the parameter which controls the superfluid decoupling phenomenon @xcite , as discussed later in this paper .
many conventional methods for generating turbulence in classical liquids or in superfluid @xmath73he are not applicable in @xmath0he - b owing to the special requirements at ultra - low temperatures and the large viscosity of the normal component .
the most detailed measurements of the transition to turbulence have been performed in rotating flow with a cylindrical sample container of radius @xmath52 and oriented along the rotation axis , fig .
[ expsetup]a .
the walls of the container are sufficiently smooth to avoid vortex pinning and surface friction .
this is possible with @xmath0he - b since the vortex core radius is relatively large , @xmath74 m , depending on pressure . and
@xmath75 are determined .
[ probabs ] ] in uniform rotation the normal component of @xmath0he - b is in solid - body rotation : @xmath76 .
the most obvious laminar solution of eq .
is then @xmath77 .
the formation of quantized vortices out of nothing at moderate flow velocities @xmath78cm / s is prohibited by a large energy barrier and thus this metastable vortex - free state can persist for the duration of the experiment in the absence of seed vortices @xcite .
another state of laminar flow is solid - body rotation of the superfluid component at an angular velocity @xmath79 : @xmath80 . in this state
the sample is filled with rectilinear vortices at a density @xmath81 , aligned along the rotation axis .
if the rotation drive @xmath70 is increased from @xmath79 , then the vortices move towards the axis or if @xmath70 is reduced they move away from it . in both cases the velocity of the laminar expansion or contraction of such a cluster of parallel vortex lines is determined by @xmath19 @xcite
. one more example of laminar flow becomes important in the laminar regime of vortex motion @xmath82 , when short seed vortices initially occupy only part of the height of the cylinder , while the rest remains vortex - free .
these vortices bend to the cylindrical wall , as seen in fig .
[ expsetup]a .
they continuously expand , until they become fully rectilinear line vortices , as their ends move with an axial velocity @xmath83 ( or somewhat smaller , depending on the curvature of the vortex ) @xcite . among the different types of turbulence in rotating flows here we discuss only two , fig .
[ expsetup]b .
the first is the turbulent burst @xcite , when a few closely packed vortex loops , attached to the cylindrical wall interact via reconnections and expanding kelvin waves and quickly fill the cross - section of the cylinder within a short vertical section .
the second process is the expansion of this turbulence towards the vortex - free region(s ) as propagating turbulent vortex front(s ) @xcite .
the axial propagation velocity of the turbulent front @xmath84 can significantly exceed the value which the laminar expansion velocity @xmath85 would have in the same conditions . in the experiment of fig .
[ expsetup]a we first create counterflow by rotating the sample in the vortex - free state ( or a state with a few rectilinear vortices in a central cluster ) and then inject seed vortices .
one injection technique uses the kelvin - helmholtz instability of the interface between the a and b phases of superfluid @xmath0he @xcite .
another method is to start at a low rotation velocity @xmath86 with a few remanent vortices connected to the cylindrical wall and to create the flow by rapidly increasing rotation to @xmath71 @xcite .
a third useful injection technique is triggered by an absorption event of a thermal neutron which deposits , depending on the counterflow velocity , one or a few vortex loops of size @xmath87 mm close to the cylindrical wall @xcite .
after injection the seed vortices interact and evolve , expanding along the cylinder . using nmr techniques ( fig .
[ expsetup]c ) we measure the number of vortex lines close to both ends of the sample tube . if the vortex expansion follows the laminar scenario , then they do not reconnect , their number is not changed , and the height of the counterflow peak in the nmr spectrum decreases only a little .
if a turbulent front develops , then the counterflow peak disappears completely when the front passes through a pick - up coil .
there are practically no intermediate cases , which can not be classified as laminar or turbulent .
an example of a measurement of the turbulent dynamics with boundary - attached seed vortices is shown in fig .
[ slowformation ] . if the vortex - injection measurement is repeated a number of times at the same conditions , we obtain the probability for the turbulent response .
the results for different injection methods and different flow velocities are shown in fig .
[ probabs ] .
it reveals a sharp transition from laminar to turbulent dynamics with decreasing temperature from which we determine the onset values @xmath88 and @xmath61 when the probability is 1/2 . in agreement with the criterion
, we find that @xmath89 .
note that in the case of the ab interface instability , when a small bundle of vortex loops is injected in bulk b - phase , @xmath61 is velocity - independent , as discussed above for bulk superflow .
when the cylinder wall is closely involved in the turbulent processes , then the transition acquires some velocity dependence .
however overall , the transition region is narrow , with the ratio of @xmath60 being around 2 at its boundaries , when measured with a given injection technique .
although the cylinder wall approaches an ideal solid surface , its influence on the dynamics can not be neglected at low mutual friction @xmath90 : the drive from the counterflow reaches its maximum close to the side wall and in addition the self - induced velocity has to be taken into account which arises from the curvature of the vortex end at the attachment point on the wall . thus reconnections with the wall , loop formation , and annihilation become important processes since their influence extends ever deeper into the bulk volume when @xmath91 . in the next two sections we discuss the influence of the boundary on the transition to turbulence and the dependence of @xmath61 on the strength of the flow perturbation .
the remarkable observation from the measurements in fig . [ probabs ] is that the injection of even a single vortex in the applied flow in a neutron absorption event results in a substantial probability to start turbulence @xcite already at a relatively small @xmath92 .
what is the process responsible for the increase from one single vortex to a number of loops which start interacting and produce turbulence ?
insight in this question is shed by the experiment outlined in fig .
[ multseq ] . here
we start with the equilibrium vortex state at @xmath93 where some vortices are curved and connect to the side wall owing to the residual tilt of the cylinder . at the initial high temperature @xmath94
, rotation is increased to a sufficiently large value @xmath71 so that all vortices are collected in a central cluster of rectilinear lines . the sample
is then cooled in this state to a temperature @xmath95 . at this temperature @xmath70 can be increased or decreased , so that the vortex cluster contracts or expands , and no turbulence is observed as long as no vortices connect to the side wall of the cylinder . however , as soon as @xmath70 is reduced below the velocity @xmath96 where the first vortex reconnects to the cylindrical wall , then on a subsequent increase of @xmath70 to @xmath71 turbulence evolves and ultimately brings @xmath97 . , the temperature @xmath98 , and the number of vortices @xmath99 as a function of time , while the sketches in the top row illustrate the vortex configurations .
see the text for details .
( _ b _ ) dependence of the rotation velocity @xmath96 at the instability on the preparation velocity @xmath100 ( symbols ) for the two b - phase sections , separated by the a phase barrier layer .
the lower section is longer and thus the first vortex reconnects to the side wall at higher @xmath96 . both dependencies can be fit to a geometrical model of the vortex cluster @xcite using the residual tilt angle of the cylinder with respect to the rotation axis @xmath101 as a fitting parameter ( lines ) .
the measurements here are performed with @xmath102 , @xmath103 and @xmath104rad / s .
[ multseq ] ] the turbulent process consists of two stages
. first during the precursor , the number of vortices @xmath99 increases linearly with time at a rate @xmath105s@xmath106 @xcite .
this slow increase is seen in fig .
[ slowformation ] .
later in the turbulent burst , @xmath99 suddenly increases close to the equilibrium number @xmath107 . from these observations
one concludes that for starting the initial increase in @xmath99 , it is necessary to have a curved vortex , attached to the boundary and moving in the applied flow .
( in contrast , vortex ends on the flat top and bottom walls of the cylinder see approximately zero counterflow and are stable . ) a sketch of this flow - induced single - vortex instability at a solid surface @xcite is presented in fig . [ wallinstab]a .
if a vortex close to the wall develops a kelvin - wave loop of proper orientation ( i ) , the loop starts to grow in the applied flow ( ii ) .
when the loop reconnects with the boundary , the reconnection event creates a new vortex and induces kelvin waves on the newly formed wall - attached vortex segments ( iii ) . in one of the segments
the kelvin waves will be properly oriented with respect to the counterflow and the loop will start to grow ( iv ) , repeating the process .
thus new vortices are created repeatedly without the need for a specially arranged vortex mill @xcite .
eventually the local density of vortices becomes sufficient to start the turbulent burst .
the main features of the above scenario are supported by numerical simulations , although to reproduce the whole sequence of events in fig .
[ multseq ] turned out to be difficult .
in particular , in fig . [ wallinstab]b one can see kelvin waves on the side - wall - attached vortex segments after the reconnection of a vortex loop with the boundary . in fig .
[ wallinstab]c a single vortex loop is allowed to expand and reconnect with a flat wall .
the probability to create a new vortex after the reconnection is calculated , averaged over all possible orientations of the original loop . remarkably , this model demonstrates both central messages of this paper .
first , the probability to create a new vortex rapidly grows with decreasing temperature in the range where @xmath108 , demonstrating the transition to turbulence .
second , at still lower temperatures , where @xmath60 approaches @xmath66 , the probability starts to decrease as a result of the decoupling of the superfluid from the reference frame of the wall . .
by repeating the calculation with arbitrary combinations of @xmath19 and @xmath21 , it is proven that the result depends only on @xmath60 , not on @xmath19 or @xmath21 separately @xcite .
[ wallinstab ] ] above the transition to turbulence was studied by introducing a perturbation in metastable laminar flow and then observing whether turbulence emerged or not .
similar investigations have been performed in viscous turbulence for the flow in a circular pipe @xcite .
laminar pipe flow is linearly stable which means that a finite - size perturbation is needed to turn it turbulent .
measurements of the amplitudes @xmath109 needed for such a perturbation lead to a scaling law @xmath110 . owing to the quantized nature of vortices in superfluids one might expect that a perturbation of finite amplitude is always required to turn the flow turbulent and thus the interesting question arises whether a scaling law , similar to that of viscous turbulence , might apply . qualitatively a dependence on the amplitude of the perturbation can be checked by comparing different injection methods @xcite and the respective values of @xmath61 .
for example , as seen in fig .
[ probabs ] , injection of @xmath111 - 20 closely packed vortex loops using the ab interface instability results in a higher probability to start turbulence at a given temperature and flow velocity compared to the injection of @xmath112 - 3 vortex rings using neutron absorption .
next we describe systematic measurements of @xmath61 using a method , where the strength of the flow perturbation , expressed as the number of injected vortices @xmath113 , can be continuously adjusted .
this is achieved by varying the initial velocity @xmath86 in the measurement with curved wall - attached vortices in fig .
[ slowformation ] .
a simple geometric model of vortex lines in the tilted cylinder @xcite gives the number of vortex ends attached to the cylindrical wall as @xmath114 . here
@xmath115 is the height of the sample , @xmath116 is the tilt angle with respect to the rotation axis and the term with @xmath117 takes into account the width of the equilibrium vortex - free annulus at the outer cylindrical boundary .
numerical calculations of the equilibrium vortex state in the tilted cylinder give a smaller number for @xmath113 owing to the effects from vortex curvature .
however , in our search for scaling laws we are not concerned with absolute numbers , but only use the dependencies @xmath118 and @xmath119 . as a function of @xmath86 , which controls the number of curved seed vortices at different values of applied flow @xmath120 .
the filled symbols represent measurements on the shorter top b - phase section , while open symbols refer to the bottom section , see fig .
[ expsetup]a . for the data on the bottom section ,
@xmath86 has been scaled with @xmath121 , where @xmath122 and @xmath123 are the heights of the two sections .
the solid lines represent power - law fits to all data with a common exponent ( excepting the bottom section at the lowest @xmath86 ) , but with a prefactor which depends on @xmath124 .
the _ insert _ shows @xmath61 at @xmath125rad / s as determined from the fits in the main panel .
the line is the power - law fit for the data points . [ scaling ] ]
thus @xmath113 can be controlled by changing @xmath86 and the applied flow velocity with respect to the normal component , @xmath126 , by varying @xmath127 . for each pair of @xmath128
we measure the probability curve @xcite , as in fig .
[ probabs ] , independently for the two b - phase sections and determine their respective onset values @xmath61 .
the results are presented in fig .
[ scaling ] .
the transition is seen to move to higher temperatures and smaller @xmath60 when either @xmath86 or @xmath124 is increased . in the same conditions
@xmath61 in the longer bottom section of the sample container is smaller that in the top section . the results from the two sections can be put on the same line assuming proportionality of @xmath113 to @xmath86 and @xmath115 .
this supports the relevance of @xmath113 as a measure of the strength of the perturbation .
it is also understandable why the value of @xmath61 in the bottom section deviates downwards from the common dependence at the lowest @xmath86 .
this is caused by the rim of the orifice in the bottom plate , where vortices become pinned and thus increase @xmath113 . remarkably , the @xmath129 dependence at all @xmath124 can be fit with the same power law @xmath130 while the prefactor in this fit also scales with @xmath124 ( insert in fig .
[ scaling ] ) .
when @xmath61 is expressed in terms of @xmath113 using the conversion from @xmath86 to @xmath113 given above , the overall scaling becomes @xmath131 .
note that in the previous section we considered the instability of a curved wall - attached vortex in the applied flow to develop independently of other vortices .
the existence of the scaling proves that this view is oversimplified and the transition to turbulence is a collective phenomenon . beyond that , however , the understanding of the measured scaling is missing and its explanation remains a task for future research .
the transition to turbulence in the flow through a circular pipe is one of the classic problems of hydrodynamics . in his pioneering work osborne
reynolds @xcite demonstrated the influence of the entrance to the pipe as well as the quality of his long straight flow channel on the downstream vorticity and turbulence . for superfluid pipe flow technical problems of this kind have not yet been solved . however
, numerical calculations can be performed to study the dynamics in superfluid pipe flow @xcite and to check whether the principles emerging from rotating measurements also apply for linear flow in the pipe .
the first message is that turbulence is suppressed at high temperatures also in pipe flow .
even when a number of vortex loops are introduced in the flow , so that they start interacting and reconnecting , eventually they decay away by annihilating , fig .
[ pipexmpl]a .
the reason is that under the influence of the mutual friction force vortex lines move also transverse to the stream lines across the entire cross section of the pipe and annihilate on the wall , leaving no remnants .
the second message is that at low temperatures the single - vortex instability switches on and this radically changes the dynamics . now
the vortex , when it is driven to the wall across the pipe , does not annihilate totally , but leaves two remnants , which start traversing the pipe cross - section in the opposite direction .
such multiplying processes make it possible even for a single vortex to grow to a turbulent plug of tangled vortices , fig .
[ pipexmpl]b , analogous to the turbulent plugs of varying length observed in viscous pipe flow @xcite . and a drive with a flat radial distribution of @xmath132 in the absence of vortices .
the flow perturbation is introduced in the form of a few vortex lines stretched across the flow channel .
the radius of the pipe is 3 mm .
( _ a _ ) at small @xmath60 , 15 vortex lines are unable to start turbulence and all vortices are eventually annihilated .
( _ b _ ) at large @xmath60 , one single vortex is sufficient to create a turbulent tangled vortex plug which moves downstream with the flow along the pipe .
[ pipexmpl ] ] the transition separating the two regimes of superfluid pipe flow is sharp and occurs at @xmath133 , as demonstrated in fig .
[ pipescaling ] .
the onset value @xmath61 scales with the magnitude of the flow perturbation , expressed as the number of injected vortices , in a manner similar to the measured dependence in rotating superfluid flow or in the pipe flow of classical liquids .
thus the calculations on superfluid pipe flow duplicate the essential features of the transition to turbulence , which were outlined in the earlier sections .
is shown as a function of the flow velocity @xmath132 ( left panel ) and as a function of the number of seed vortex lines @xmath113 ( right panel ) .
the value of @xmath61 is determined as the mid point between two calculations : one proving complete annihilation of all vortex lines at a smaller @xmath60 and a second demonstrating the formation of a turbulent vortex plug at a somewhat higher @xmath60 .
the separation of these two @xmath60 values is depicted with the uncertainty bars . [ pipescaling ] ]
when the superfluid dynamics becomes turbulent at @xmath134 , the assumption about the local polarization of vortex lines , used in deriving equation , starts to be violated .
one way to provide a phenomenological description of turbulent flows on the level of the coarse - grained equations is to supplement with effective friction parameters .
effective friction models , though , do not possess universality : they depend on the type of flow and on the particular physical process considered .
this is similar to how the scattering time of quasiparticles in @xmath0he is often treated : it is not a universal quantity , but depends eg . on whether one considers viscosity , heat conduction , or mutual friction .
so far the most detailed experimental information on effective friction ( or viscosity ) was collected for turbulent energy dissipation processes .
here we describe measurements which allow us to determine simultaneously the effective frictions for energy dissipation and for momentum transfer between the superfluid and its hydrodynamic drive @xcite .
we consider the lowest temperatures where @xmath21 can be ignored and discuss only effective values of @xmath19 .
[ expsetup]b illustrates the emergence of an axially expanding process which ultimately fills the rotating cylinder with rectilinear vortex lines in the equilibrium state of rotation . the most exciting part of this process is a turbulent vortex front @xcite propagating along the entire length of the long cylinder .
the front separates the vortex - free non - rotating superfluid from the rotating superfluid with a bundle of twisted vortex lines @xcite , which we assume to be in approximate solid - body rotation .
thus @xmath4 is forced in a configuration of shear flow axially across the front .
as usual for shear flow , such a configuration is unstable with respect to turbulence . at closer look it becomes obvious that vortices in the front , which terminate on the cylindrical side wall , precess around the cylinder axis at a smaller angular velocity than the twisted vortex bundle behind the front .
this differential motion leads to reconnections , which sustain the turbulence in stationary state while the front is propagating at constant velocity @xmath84 .
the axial velocity @xmath84 of the front is measured using nmr - based time - of - flight techniques and is shown in fig .
[ frontvel ] as a function of temperature and rotation velocity .
one of the characteristics is that @xmath84 saturates in the @xmath135 limit at a @xmath98-independent , but @xmath70-dependent value . since the motion of the front decreases the free energy of the superfluid
, a non - zero @xmath84 means that the turbulence in the front leads to finite energy dissipation even in the zero - temperature limit , in agreement with observations on the free decay of vortex tangles in @xmath0he - b @xcite .
another non - trivial feature of the front velocity is the temperature dependence with a faster variation of @xmath84 in the range @xmath136 , than at higher or lower temperatures .
this feature we call below the shoulder. the shoulder in fig.[frontvel ] falls in the temperature regime where @xmath137 and the quasiparticle density approaches the collisionless limit . here
momentum exchange of the superfluid component with the normal excitations becomes increasingly weaker and the effects from vortex tension and curvature gain importance in eq . .
this leads to partial decoupling of the superfluid from the reference frame of the container @xcite .
it means that behind the front the superfuid component rotates at an angular velocity @xmath138 , that is at a rate slower than what solid - body rotation in equilibrium with the cylinder wall requires . in fig .
[ decouple ] measurements of the precession velocity @xmath79 , as derived from the heat released in front propagation at @xmath139 , give a result which is less than half of that of the rotation drive @xmath70 .
this is consistent with @xmath140 obtained from direct observations of vortex precession with nmr and from numerical simulations at this temperature . .
in the region of the front the vortices terminate at the cylindrical side wall .
turbulence is sustained by vortex reconnections which are highlighted with yellow dots .
the front moves upward into the vortex - free region .
the plot on the left shows measurements of the axial front velocity ( circles ) as a function of temperature and angular velocity @xmath70 of the cylinder .
the lines are fit to the model eq . with parameters @xmath141 , @xmath142 , @xmath143 , and @xmath144 .
[ frontvel ] ] to account for the decoupling , one considers the angular momentum balance for the propagating front in eq . , where @xmath33 appears since @xmath145 and @xmath31 owing to the vortex twist .
in addition @xmath19 has to be replaced by an effective value for angular momentum @xmath146 . for the energy balance , which in the laminar regime was expressed as @xmath147 ,
one then recovers a generalized expression with @xmath70 replaced by @xmath79 and @xmath19 by a distinct effective value for the energy dissipation @xmath148 .
the combined model is @xcite @xmath149 we assume a simple linear relation of the effective friction parameters with @xmath150 : @xmath151 and @xmath152 , with @xmath153-independent phenomenological parameters @xmath154 , @xmath155 , @xmath156 and @xmath157 .
this model provides a good fit to the experiment in fig .
[ frontvel ] and reproduces the main qualitative features of @xmath84 including the shoulder .
the same parameter values also describe measurements of @xmath84 at different liquid pressures @xcite and the direct measurements of @xmath79 in fig .
[ decouple ] .
the systematic deviation of the model from the experimental data in the range ( 0.25 0.35)@xmath158 in fig .
[ frontvel ] may be caused by an oversimplified treatment of the angular momentum balance in the turbulent front which does not yet include all contributions considered for the laminar front @xcite .
an important message is conveyed by the residual values for @xmath143 and @xmath159 , which describe the @xmath160 contribution to effective friction from turbulence and possibly from surface friction .
when energy dissipation is considered , turbulence dramatically enhances effective friction , owing to the turbulent energy cascade , which leads to substantial dissipation even in the zero - temperature limit .
the minute value of @xmath157 , which is two orders of magnitude smaller than @xmath155 , attests that quantum turbulence does not efficiently increase momentum exchange with the normal component and the container . as a result ,
a new class of hydrodynamic phenomena becomes prominent owing to superfluid decoupling at temperatures where @xmath161 and @xmath162 . and reaches the end of the sample cylinder at @xmath163s , when the measured power is at the maximum .
the square - wave model ( blue curve ) assumes the superfluid component to reach the equilibrium vortex state immediately behind the front .
in fact , owing to its partial decoupling from the container rotation , the superfluid component comes to equilibrium much later , as the fitted analysis of the measured record shows ( green curve ) .
it proves that the vortex density behind the front corresponds to only @xmath164 , while the rest of the vortices are supplied mostly by laminar motion well after the front propagation has already stopped .
the insert shows the rotation velocity @xmath79 of the superfluid behind the front as a function of temperature ( normalized to the velocity @xmath70 of the rotation drive ) .
@xmath79 is determined from an analysis of the thermal signal in the main plot , from oscillating nmr signals @xcite , and from direct measurements of the number of vortices behind the front at higher temperatures @xcite ( filled symbols ) .
results from numerical simulations on front motion are shown with empty circles .
the line represents the model in eq . with the same parameter values as in fig .
[ frontvel ] at @xmath165rad / s .
[ decouple ] ] the difference in the contribution from quantum turbulence to energy dissipation and to momentum exchange with the normal component may originate in the damping properties of the kelvin waves , emitted from vortex reconnections .
this is illustrated numerically in a simple model system of two vortex rings , reconnecting in the presence of mutual friction damping @xcite , fig .
[ recrings ] . the damping causes a continuous decrease of the total energy and momentum of the system . after
the reconnection kelvin waves are excited and the oscillating motion causes substantial extra energy dissipation , while the rate of momentum change is essentially not affected by the reconnection event .
the explanation is that the direction of the mutual friction force alternates in the two half - periods of the oscillation .
thus momentum transfer to the normal component cancels out .
simultaneously the direction of the vortex velocity is also alternating , but it remains opposite to the mutual friction force .
thus the work done by mutual friction in the two half - periods of the oscillation adds up . since reconnections and the fluctuating vortex motion are the characteristic features of quantum - turbulent flows , this conclusion is likely to apply to quantum turbulence in general .
the normal component can be considered not to participate in the motion if @xmath166 in the hydrodynamic regime while in the ballistic regime the diffusive scattering of thermal quasiparticles from the walls should dominate the scattering from vortices .
these conditions are generally realized in superfluid @xmath0he but may also be applicable to superfluid @xmath73he in the low - temperature range .
when the normal component is stationary and surface pinning or friction can be ignored , the motion of the superfluid component is characterized by three dimensionless parameters : @xmath66 , @xmath60 and their ratio .
the parameter @xmath66 describes the transition from vortex - free or single - vortex dynamics at @xmath167 to collective dynamics with many vortices at @xmath168 .
the latter condition is required for quantum turbulence .
the parameter @xmath60 has the same physical meaning as the reynolds number in classical turbulence .
when mutual friction is large and @xmath169 , turbulent motion is damped .
one might think that by placing a sufficiently dense vortex tangle in high - velocity flow , one could observe turbulent motion even at high temperatures . in reality when @xmath19 is large , vortex lines move across the flow , i.e. towards the wall . if they annihilate at the wall without leaving remnants behind , which is the case at an ideal wall and small @xmath60 , then all vortices eventually disappear . with decreasing temperature , i.e. increasing @xmath60
, vortices start to move with the flow , which allows more time for interactions and reconnections . in this regime also the single - vortex instability at the wall @xcite becomes an effective mechanism for the formation of new loops .
experimentally one can distinguish between the formation of new vortices in the bulk - volume or in surface - mediated processes since the latter switch on at slightly lower temperatures .
but overall both types of processes contribute to the transition from laminar to turbulent superfluid dynamics at @xmath108 .
the value of @xmath60 at the transition depends on the type of flow and scales with the magnitude of the flow perturbation . in some cases laminar flow
can be exceptionally stable , like in the spin - down of a cylindrical container to rest . here a large perturbation , like tilting the sample cylinder by @xmath170 from the rotation axis , is needed to turn the flow turbulent even at @xmath171 @xcite .
this is in contrast to classical fluids , where the spin - down to rest is one of the most unstable flows @xcite .
the difference is caused by the different mechanisms of coupling of the fluid to the container : in classical fluids it is a surface effect , mediated by boundary layers , while in superfluids it is a volume effect ( in the absence of surface pinning or friction ) , mediated by the normal component . , momentum @xmath172 , and angular momentum @xmath173 , normalized to their respective values at @xmath174 .
the calculation demonstrates quite different behavior in energy and momentum loss after the reconnection event .
no externally applied flow is present .
[ recrings ] ] the implications from this coupling in low - temperature superfluid dynamics have been understood only recently .
it is by now well established that quantum turbulence turns a superfluid to a dissipative liquid , by providing intrinsic mechanisms for energy loss via the turbulent energy cascade .
in contrast , no such intrinsic mechanism exists for momentum exchange with the normal component / container / drive , if extrinsic mechanisms such as surface friction and pinning are excluded . with respect to momentum coupling and drag , a superfluid filled with a vortex tangle continues to behave like an almost ideal liquid and no boundary layer , similar to that in classical turbulence , is formed .
this duality in coupling mechanisms has profound implications on the low - temperature dynamics , when @xmath175 .
in particular , the rapidly decreasing momentum coupling leaves a characteristic anomaly in the overall energy dissipation rate , as demonstrated by the shoulder in the measured front velocity @xmath176 in fig .
[ frontvel ] .
the anomaly falls in the temperature range where another interesting phenomenon is expected to become important , the crossover in the turbulent energy cascade from the quasi - classical cascade at length scales exceeding the inter - vortex distance to the kelvin - wave cascade at smaller scales .
the problem of joining the two energy cascades of different nature across the crossover region @xcite , dubbed the bottleneck problem , has still no universally accepted theoretical resolution .
the mutual - friction - determined termination point of the energy cascade traverses the crossover region in the temperature range where @xmath177 @xcite or @xmath178 @xcite .
this is not very different from the range where decoupling becomes important owing to the rapid variation of @xmath179 .
it is thus not trivial to distinguish the two effects experimentally .
in fact , the shoulder in the front velocity was originally interpereted as an indication of the bottleneck @xcite , and only newer direct measurements of @xmath79 allowed to understand that decoupling is the main source of this anomaly .
an open question is whether the systematic deviation of the front velocity from the decoupling model in the range ( 0.25 0.35)@xmath158 in fig . [ frontvel ] may provide new information on the bottleneck .
a further consequence from the decreasing momentum coupling at low temperatures is the increased importance of laminar flow in the low - friction region , where all dynamics was earlier expected to be turbulent .
for example in the thermal measurement of the front propagation in fig .
[ decouple ] only about 40% of the energy is released by the turbulent process and the rest comes as slow laminar tail .
the reason is that in laminar flow the decoupling problem does not appear , as both energy dissipation and momentum transfer are determined by the same mutual friction parameter @xmath19 .
an interesting question is to what extent the decoupling phenomena are applicable to quantum turbulence in superfluid @xmath73he , where usually complicated experimental geometries and the pinning of vortices at the sample boundaries facilitate the coupling of the superfluid to the container .
simultaneously , however , much bigger values of @xmath60 can be achieved in @xmath73he compared to @xmath0he which in principle increases the role of decoupling .
suppression of the decoupling by pinning might be not complete for the following reasons : first , the pinning force is limited by vortex tension @xmath180 .
this limit has been experimentally observed in spin - up of @xmath73he in rough - wall containers @xcite .
second , it has been demonstrated that a vortex in @xmath73he can slide relatively easily along the wall by continuously reconnecting from one pinned vortex to another @xcite .
thus it may be worth to consider whether the decoupling affects the interpretation of experiments in superfluid @xmath73he .
one example is the measurements of spin - down of a cubic container filled with superfluid @xmath73he @xcite . here
the anomaly in the temperature dependence of the effective turbulent viscosity @xmath181 was observed , in a way similar to the shoulder in the front velocity @xmath84 in our experiments .
the current explanation of this effect is linked to the physics of the crossover region in the turbulent energy cascades @xcite .
we note , however , that the decrease of @xmath181 is observed in the temperature region where @xmath182 and thus the influence of the decoupling can not be a priori excluded .
the superfluid decoupling is most pronounced in flows where the momentum transfer from the drive to the superfluid is unidirectional .
the opposite extreme is quantum turbulence generated with various oscillating objects @xcite , currently the most popular technique for studying the free decay of turbulent vortex tangles @xcite . here the direction of momentum transfer is continuously alternating , with the average being close to zero . in this case
the decoupling phenomena are masked .
the recently developed floppy vibrating wire @xcite , which can be used in an oscillating mode or in a steady - motion mode , might provide a new tool to study the decoupling phenomenon , provided that surface - friction effects can be kept under control .
we thank g.e .
volovik , v.s . lvov and j.j .
hosio for useful discussions .
this work was partially supported by the academy of finland ( centers of excellence programme 2012 - 2017 and grant 218211 ) and the eu 7th framework programme ( fp7/2007 - 2013 , grant 228464 microkelvin ) .
hosio jj , et al .
( 2013 ) energy and angular momentum balance in wall - bounded quantum turbulence at very low temperatures .
_ nature comm _ 4:1614 .
finne ap , et al . (
2006 ) dynamics of vortices and interfaces in superfluid @xmath0he .
_ rep prog phys _ 69:31573230 .
reynolds o ( 1883 ) an experimental investigation of the circumstances which determine whether the motion of water shall be direct or sinuous and of the law of resistance in parallel channels .
_ phil trans roy soc _ 174:935982 .
skrbek l , vinen wf ( 2009 ) the use of vibrating structures in the study of quantum turbulence . in _ progress of low temperature physics vol .
xvi _ , eds tsubota m , halperin wp ( elsevier , the netherlands ) , pp . | in fermi superfluids , like superfluid @xmath0he , the viscous normal component can be considered to be stationary with respect to the container .
the normal component interacts with the superfluid component via mutual friction which damps the motion of quantized vortex lines and eventually couples the superfluid component to the container . with decreasing temperature and mutual friction the internal dynamics of the superfluid component becomes more important compared to the damping and coupling effects from the normal component .
this causes profound changes in superfluid dynamics : the temperature - dependent transition from laminar to turbulent vortex motion and the decoupling from the reference frame of the container at even lower temperatures .
n this paper we consider the motion of quantized vortices in superfluids , where the normal component is clamped to the walls of the container , that is , it is stationary in a reference frame moving with the wall .
this situation can be experimentally realized in superfluid @xmath0he . in such systems
turbulent motion can occur only in the superfluid component and thus , in principle , is easier to analyze . here
the role of the normal fluid is twofold : first , it provides friction in the superfluid motion , which is mediated by quantized vortices and works over a wide range of length scales .
second , it provides a coupling to the container walls , which acts uniformly over the whole volume of the superfluid .
this is quite unlike classical turbulence , where viscous dissipation operates only at the small kolmogorov scale and coupling to the walls is provided by thin boundary layers . as a result
, a variety of new phenomena is observed in experiments and numerical simulations as a function of temperature .
these are controlled by the normal - fluid density . at the highest temperatures turbulent motion
is suppressed completely .
when the temperature decreases , a sharp transition to turbulence is seen .
the transition can be characterized by a superfluid reynolds number , which is composed of the internal friction parameters of the superfluid and is independent of velocity .
when turbulence is triggered by a localized perturbation of the laminar flow , the critical value of the reynolds number is found to scale with the strength of the perturbation . when the temperature decreases further , friction from the normal component rapidly vanishes .
the overall dissipation rate in quantum turbulence remains nevertheless finite , owing to the contribution from the turbulent energy cascade , and reaches a temperature - independent value in the zero - temperature limit .
this zero - temperature dissipation can be characterized by an effective viscosity or friction .
it is found , however , that coupling to the walls is not essentially improved by the turbulence and can potentially become very small . at the lowest temperatures
the concept of a single effective friction breaks down ; for the proper description of quantum - turbulent flows one then has to introduce a separate effective friction for momentum exchange between the superfluid and the boundaries . |
the rest - frame uv spectra of qsos and active galactic nuclei ( agns ) typically contain strong and broad emission lines ( bels ) from the resonance doublets 1549 , 1240 and 1034 in the lithium - like isoelectronic sequence .
@xmath5774 is another resonant doublet in that sequence .
its short wavelength makes 774 difficult to measure , but this line deserves special attention because requires ionization energies ( 207 ev ) nearly twice as large as ( 114 ev ) and more than four times larger than ( 48 ev ) .
broad @xmath5774 lines can therefore probe the extreme high - ionization gas in the broad emission line regions ( belrs ) and test the possible relationship of this gas to other high - ionization phenomena in qso / agn environments .
for example , the -emitting belr has an ionization consistent with the x - ray `` warm '' absorbers ( halpern 1984 ) observed in some quasars and agns .
warm absorbers can be identified by bound - free edges of and near 0.8 kev ( for example , @xcite ) .
photoionization calculations indicate that the regions producing line emission would also produce these edges in soft x - rays if the gas lies along our line - of - sight to the x - ray continuum source ( @xcite ) .
there are now several reports of broad 774 emission from qsos ( @xcite , hamann , zuo & tytler 1995b ) , but the data are often of poor quality and the number of secure detections is small .
hamann ( 1995b ) made the first attempt to study the general strength and character of 774 emission in a sample of qsos .
that sample consists of 5 radio - loud and radio - quiet sources measured in the _ hubble space telescope _ ( _ hst _ ) absorption line snapshot survey ( absnap ) .
they found that 3 of the 5 sources , and possibly all 5 , have a bel at @xmath0774 .
they argued that the measured wavelengths and fluxes ( compared to photoionization models ) both point to 774 as the most likely identification .
more recently , zheng ( 1997 ) measured the @xmath0774 emission line in a composite spectrum derived from a larger number of qsos . here
we supplement that work with new photoionization calculations and new _ hst _ and ground - based spectroscopy of four qsos .
we also extract an unbiased sample of 11 qsos from the _ hst _ archives and combine them with our previous absnap sample to construct mean spectra for a total of 16 sources .
the new individual spectra provide the highest signal - to - noise measurements of the feature , while the means allow us to study the 774 emission properties in representative samples .
section 2 below describes the observations and measurements of the emission lines .
section 3 reexamines the 774 identification and provides an analysis of the kinematics and physical conditions based on photoionization simulations and fits to the line profiles .
section 4 discusses the implications of the 774 emission and , in particular , the possible relationship to warm absorbers .
section 5 provides a summary .
we obtained uv spectra of four bright qsos with the faint object spectrograph ( fos ) on board _
we observed pks 0355@xmath6483 and q 1435@xmath6015 because they were strong candidates for broad 774 emission in our previous study of the _ hst_-absnap sample ( @xcite ) .
we chose pg 1148 + 549 as another candidate for strong emission based on earlier observations with the _ international ultraviolet explorer _ ( _ iue _ ; @xcite ) .
we observed the qso pg 1522 + 101 as part of an unrelated absorption - line study and include it here because the _ hst _ spectra contain a strong emission line near 774 rest ( @xcite ) .
the _ hst _ spectra encompass 774 and other prominent lines in the rest frame uv but , for the two highest - redshift sources , q 1435@xmath6015 and pg 1522 + 101 , they do not include the important 1549 line .
we therefore obtained ground - based optical spectra of these qsos with the shane 3.0 m telescope at lick observatory .
table 1 provides a log of the observations , including the dates , the telescopes used ( tel . ) , the spectrographs and their setups / gratings ( instr./setup ) , the wavelengths covered ( @xmath7 ) , the approximate full - width - at - half - maximum resolutions @xmath8 , and the total exposure times ( exp . ) .
the emission - line redshifts listed with the qso names are from hewitt & burbidge ( 1993 ) , except for pg 1148 + 549 , which is from schmidt & green ( 1983 ) .
hst_-fos spectra were obtained through the 1 science aperture and calibrated at the _ space telescope science institute _ using their standard `` pipeline '' procedures .
we recalibrated the absolute wavelengths in the g130h and g270h spectra by requiring that galactic absorption lines appear at their laboratory wavelengths .
we then shifted the g190h or g160l spectra by matching features in spectral regions that overlap with g130h or g270h .
we expect that the final @xmath9 wavelength uncertainties are less than 0.5 diodes , that is @xmath104 with the g160l grating and @xmath100.51 with the higher resolution gratings .
the lick observations employed a 2 slit and a @xmath11 reticon ccd .
we reduced those data in the usual way using vista software ( modified slightly by dr .
t. a. barlow ) .
the wavelength uncertainties in the lick spectra should be less than @xmath00.2 .
we flux - calibrated those spectra by observing both the qsos and standard stars through a wide slit on the same night .
none of the data are corrected for galactic extinction .
we simply note that for a typical high - latitude galactic column density of @xmath12 ( murphy 1995 and references therein ) , a conversion factor of @xmath13 ) ( @xcite ) and a typical extinction curve with @xmath14 ( cardelli , clayton & mathis 1989 ) , the extinction peak at the @xmath02175 silicate feature is roughly 0.5@xmath15 .
this peak coincides with the 1034 emission line for @xmath16 .
figures 14 plot the new and old spectra together .
the new _ hst _ observations of pks 0355@xmath6483 and q 1435@xmath6015 ( figures 1 and 3 ) have higher signal - to - noise ratios and slightly higher resolutions than the absnap data because of the longer exposure times and improved line spread function of the post - costar fos . figure 2 includes two versions of the _ iue _ spectrum of pg 1148 + 549 for comparison to the _ hst _ data . the bottom spectrum in that figure , labeled `` old _ iue _ '' , is the original data as shown in hamann ( 1995a ) . the middle spectrum , labeled `` new _ iue _ '' , is a new reduction of the same data .
the old _ iue _ spectrum was reduced using an optimal extraction technique ( @xcite ) .
the new _ iue _ spectrum derives from the improved image processing system ( newsips ) used for the _ iue _ final archives ( @xcite ) and is the average from three separate images that have been cleaned of cosmic rays .
the new _ iue _ reduction of pg 1148 + 549 has a slightly higher signal - to - noise ratio , but there is a camera artifact at @xmath01663 ( crenshaw 1990 ) that is enhanced by newsips .
the emission line discussed by hamann ( 1995a ) is weaker in both the newsips reduction and the _ hst _ spectrum ( upper curve in figure 3 ) .
figure 4 plots the low ( g160l ) and high ( g190h+g270h ) resolution _ hst _ spectra of pg 1522 + 101 , with the longer wavelength lick observatory data appended to the latter . both _
_ observations of this source show strong @xmath0774 emission , but the higher resolution spectrum shows more clearly the `` forest '' of absorption lines shortward of the qso s emission .
these absorption features can significantly distort the continuum and bels in low resolution data .
there is also weak lyman limit absorption in this spectrum at @xmath01980 ( observed ) , which is confirmed by detection of the corresponding lyman series lines .
we constructed mean spectra of two samples of qsos measured with _ hst_-fos and the low - resolution grating ( g160l ) .
the first sample consists of the 5 qsos selected by hamann ( 1995b ) from the _ hst_-absnap database ( @xcite ) .
that sample includes qsos with ( 1 ) emission redshifts ( @xmath17 to 1.4 ) that allow wavelength coverage across both 774 and 1034 with the g160l grating , ( 2 ) no lyman - limit absorption across @xmath0774 in the qso rest frame , and ( 3 ) subjectively `` measurable '' emission .
the last criterion serves as a signal - to - noise discriminator and insures that we can establish an emission redshift for coadding the spectra .
this selection based on could introduce a bias toward sources with more high - ionization gas and thus stronger and emission lines ( see 2.4 below ) .
however , the qsos were _ not _ chosen for their characteristics .
the second sample consists of 11 qsos from the first _
absorption line key project database ( @xcite ) selected by criteria ( 1 ) and ( 2 ) above6 at 3@xmath18 ) and a poorly defined redshift ( see @xcite ) . ] .
seven of these qsos have measurable emission and 4 do not
. we will refer to the subset of 7 key project spectra with significant as kp - sub , and the full sample of 11 key project spectra as kp - full .
note that none of the spectra that lack emission , including pg 1407 + 265 ( footnote 2 ) , have a detectable line .
figure 5 plots the mean spectra derived from the various sub - samples and the total sample of 16 sources .
we calculated the means using the new _
hst_-fos spectra of the two absnap sources pks 0355@xmath6483 and q 1435@xmath6015 ( 2.1 ) .
the other absnap spectra are from hamann ( 1995b ) .
the key project spectra are from the _ hst _ archives with reductions by schneider ( 1993 ) .
pg 1148 + 549 and pg 1522 + 101 ( 2.1 ) are not in the absnap or key project samples and so are not included in the means .
for the 12 sources with well - defined lines ( in the absnap and kp - sub samples ) we shifted the individual spectra to the rest frame by forcing the centroid of the upper @xmath050% of the profile to be at 1033.8 .
for the 4 sources without , we shifted the spectra using published redshifts in hewitt & burbidge ( 1993 ) . applying the published redshifts , which derive from low - ionization lines such as ] 1909 or 2799 ,
leads to larger uncertainties ( by at least several ) in the rest wavelengths of and because ( 1 ) these redshifts do not compensate for the calibration uncertainties in the _
hst_-fos wavelengths and ( 2 ) there are known intrinsic redshift differences between high- and low - ionization lines ( @xcite ) . in all cases , we normalized the continua to unity by fitting with low - order polynomials and then averaged the spectra with equal weights .
wavelength regions at the ends of individual spectra were excluded from the means if they contain obvious lyman - limit absorption , severe noise spikes , or geocoronal emission lines such as in first or second order .
the histogram at the bottom of figure 5 indicates the number of spectra contributing to the total mean at each wavelength .
median spectra drawn from these samples are qualitatively the same as the means .
table 2 lists the centroid rest wavelengths ( @xmath19 ) , rest equivalent widths ( rews ) , observed fluxes ( flux ) , and full widths at half maximum ( fwhm ) for emission lines detected in the new _ hst _ and lick spectra ( figures 14 ) and in several of the mean spectra ( figure 5 ) .
not all of the lines labeled in figures 15 are present in the data ; for example , 615 and ] 895 are not detected in any source with approximate 3@xmath18 upper limits of rew @xmath20 in all cases .
the measurements for pg 1522 + 101 are from the higher resolution data after interpolating across the strongest of the overlying forest lines .
we made no attempt to correct for absorption lines in the other sources .
the fluxes given for the mean spectra are relative to 1034 , while those given for the individual sources are as observed with units 10@xmath21 ergs s@xmath22 @xmath4 . recall that the data are not corrected for galactic extinction ( 2.1 ) .
the equivalent widths are immune to extinction effects , but the fluxes are not . in a worst - case scenario , with the @xmath02175 extinction peak overlying the line ( 2.1 ) ,
the differential extinction between and for a typical galactic column density would be @xmath00.1@xmath15 and would lead us to overestimate the true / flux ratio by 10% .
the emission lines were measured by integrating the flux above the fits to the continua ( dotted curves in figures 15 ) between the wavelengths indicated by @xmath23 in table 2 .
we did not attempt to separate severely blended lines or line components .
in particular , the feature we attribute to 1034 could have a contribution from 1026 .
laor ( 1995 ) and wills ( 1995 ) estimate that the mean contributions to this blend are @xmath023% and @xmath1010% , respectively , in samples of low - redshift qsos measured with _
we will ignore this small contribution in our discussions below .
we will also ignore the weak 1527 line that can add to the measured 1549 flux .
the feature we identify as 774 is clearly present in the mean spectra , and it is stronger than the means ( both in rew and relative to other lines ) in the individual qsos pks 0355@xmath6483 ( radio - loud ) and pg 1522 + 101 ( radio - quiet ) .
the observations of q 1435@xmath6015 are less conclusive , but the new data are consistent with the previous report of strong @xmath0774 emission from this source ( @xcite ) .
hst_-fos spectrum of pg 1148 + 549 does not exhibit the strong @xmath0774 line noted in the original _ iue _ data ( @xcite ) .
hst _ spectrum of this source resembles the reprocessed ( `` new '' ) _ iue _ spectrum ( fig .
3 ) , indicating that the line is weaker than our previous claim . there is evidence in the means and in some of the individual spectra for emission from the excited - state line @xmath24 1176 and from a broad unidentified feature at @xmath01070 .
the unidentified feature was noted previously by laor ( 1994 and 1995 ) in their sample of low - redshift qsos . the @xmath24 line could prove to be a valuable diagnostic of belr physical conditions because it involves a transition between unusually high energy states ,
17.1 to 6.5 ev . the lower state is metastable and gives rise to the well - known ] 1909 emission line ( see @xcite for discussion ) .
unbiased measurements of the average @xmath0774 line strength require spectra of many qsos selected by only their redshifts and their lack of overlying lyman - limit absorption ( criteria 1 and 2 in 2.2 ) . among the data we collected , the kp - full sample of 11 spectra comes closest to this ideal
. the average @xmath0774 line in that sample has rew @xmath0 5.9 and a flux of @xmath042% of 1034 ( table 2 ) .
these results can be compared to rew @xmath0 3.2 and a flux ratio of 774/1034 @xmath0 21% reported by zheng ( 1997 ) from their composite _ hst _ spectrum .
the zheng composite includes @xmath020 spectra across @xmath0774 and 8090 across the line .
our kp - full sample is undoubtedly a subset of the spectra they used .
the different line strengths in these two samples could be caused by random fluctuations in the small numbers of spectra involved and/or by the large measurement uncertainties , which are dominated by the subjective continuum placement .
however , the reported differences might also be caused by selection effects .
in particular , there is an inherent `` baldwin effect '' in the zheng composite .
several studies have shown that this effect is pervasive in qso samples ; more luminous qsos have typically lower emission - line rews ( @xcite ) .
the spectra in the _ hst _ database are primarily of the brightest qsos at each redshift .
our kp - full sample includes a relatively narrow redshift range so that @xmath0774 through @xmath01034 are measured in the same data , but the zheng composite has many more low redshift and low luminousity qsos contributing at long wavelengths ( e.g. across ) than at short wavelengths ( at ) .
therefore their composite should have a larger rew in 1034 , a smaller rew in 774 and a lower 774/1034 strength ratio compared to our kp - full sample , which agrees with the measured results .
therefore , our kp - full sample , although still limited by the small number of spectra and significant measurement uncertainties , should provide the most representative measure of the 774 line strength and the 774/1034 flux ratio in ( bright ) qsos at redshift @xmath01 .
the absnap and kp - sub samples both have stronger 774 lines than kp - full , probably because of a selection bias introduced by the requirement for `` measurable '' emission ( 2.2 ) .
note , however , that while 774 is stronger in the absnap sample , 1034 and the blend are not ; therefore , the nature and extent of the selection bias is ambiguous .
the total mean mixes the absnap and kp - full samples and therefore has slightly stronger @xmath0774 line emission than kp - full alone .
theoretical models of the belr are needed to understand the conditions under which 774 forms and the likelihood that other lines near this wavelength might contribute to the measured emission .
some recent photoionization calculations have made specific predictions for the line strength ( @xcite ) .
here we update the calculations in hamann ( 1995a ) using a newer version of cloudy ( 90.02 ; @xcite ) to examine other strong lines at wavelengths @xmath251000 and test whether recent improvements in the atomic data affect the results ( see also @xcite ) .
the most likely contributors to broad line emission near 774 are 774 , 765 , 789 , 786 and 763 ( verner , verner & ferland 1996 ) .
the calculations by hamann ( 1995a ) indicate that the alternatives to should be relatively weak , and therefore is the most likely identification for the measured line .
figure 6 shows the line equivalent widths predicted by cloudy for different values of the ionization parameter @xmath26 ( defined as the dimensionless ratio of hydrogen - ionizing photon to hydrogen particle densities at the illuminated face of the clouds ) .
this plot is analogous to figure 2 in hamann ( 1995a ) .
the model clouds have a space density of @xmath27 , a column density of @xmath28 and solar element abundances .
they are illuminated on one side by a standard qso spectrum that is believed to be typical of low - redshift qsos and seyfert 1 nuclei ( @xcite ) .
we modified this spectrum to have a sharp decline at wavelengths longer than 2 @xmath29 m to avoid significant free - free heating ( @xcite ) .
the internal cloud velocities are assumed to be thermal .
the equivalent widths in figure 6 apply for emitting regions that completely cover the central qso ( over 4@xmath30 ster ) but are transparent to the radiation from other clouds ( e.g. and photons emitted on the far side of the qso are not absorbed or reflected by clouds on the near side ) .
the effects of different space densities and spectral shapes on some of the line strengths are shown by korista ( 1997 ) and netzer ( 1996 ) .
in general , the relative strengths of the permitted lines in figure 6 are only weakly dependent on the space density for a given @xmath26 .
the intercombination lines ] and ] are strongly suppressed above their common critical density of @xmath0@xmath31 .
smaller column densities would significantly reduce the emission from low- to intermediate - ionization lines ( e.g. , and ) at the right in figure 6 , because these lines form strictly at large depths in the high-@xmath26 clouds ( see fig . 3 in hamann 1995a ) .
examination of the heating and cooling processes indicates that , for the continuum shape used here , clouds with @xmath32 are thermally unstable ( @xcite ) .
the only important difference between figure 6 here and figure 2 in hamann ( 1995a ) is that the line is now several times stronger , particularly at low and intermediate @xmath26 ; but the main results are the same .
786 and 763 are too weak under any circumstances to produce the measured @xmath0774 line .
( the predicted 763 line , not shown in fig .
6 , is at least 5 times weaker than 786 . )
the 765 and 789 lines are the strongest alternatives to 774 .
they form in roughly the same gas as 1549 , so their strengths relative to are not sensitive to the uncertain geometry or physical conditions . scaling the predicted / and / ratios in figure 6 by the measurements in table 2
shows that the observed 774 lines in pks 0355@xmath6483 and pg 1522 + 101 are much too strong to have significant contributions from or .
in particular , for @xmath33 ( see 4 ) , + should have rew @xmath10 0.6 in pks 0355@xmath6483 and rew @xmath10 1.9 in pg 1522 + 101 , compared to the measured value of rew @xmath0 13.5 in both qsos .
the ratio of the fluxes predicted in + to that measured in @xmath0774 is therefore @xmath100.07 in pks 0355@xmath6483 and @xmath100.3 in pg 1522 + 101 .
the mean spectra ( fig .
5 ) do not include 1549 for comparison , but in large ground - based samples has rew @xmath0 20 to 32 , or about @xmath050% of ( see the compilation by @xcite ) .
if we adopt a conservatively large value of rew(1549 ) = 40 , then figure 6 implies that the average + emission should have rew @xmath34 for @xmath35 or rew @xmath36 for @xmath33 .
these estimates are no more than about half the measured mean of rew @xmath0 5.9 in the @xmath0774 line ( 2.4 ) .
since values of @xmath35 are generally favored for the emitting region ( 4 ) , the average contributions of and to the @xmath0774 feature should be @xmath1017% .
these results indicate that 774 dominates the average @xmath0774 emission and is the only significant contributor whenever this line is strong compared to .
experiments with cloudy show that this conclusion holds even if the metallicities are above solar and nitrogen is several times overabundant relative to the other metals ( @xcite and 1993 ) .
figure 6 also supports the conclusions in hamann ( 1995a ) regarding the ionization , column densities and covering factors of the emitting regions .
significant emission in 774 , relative to both the continuum and other lines , requires ionization parameters @xmath37 .
thermal stability in the clouds independently requires @xmath38 , for an overall range of @xmath39 in the line - forming gas .
the ratio of the observed mean ( 5.9 ; 2.4 ) to largest - predicted ( @xmath020 ; fig .
6 ) 774 equivalent widths implies an average minimum covering factor of @xmath030% .
the emitting regions must subtend at least this fraction of the sky as seen from the central qso to produce the average equivalent width .
( the same analysis shows that the strongest emitters , pks 0355@xmath6483 and pg 1522 + 101 , require covering factors @xmath165% . )
the observed mean 774 equivalent width also requires a minimum hydrogen column density of @xmath40 for solar abundances , even if the emitting region completely covers the central qso ( see figure 3 in @xcite ) .
( the strongest emitters require column densities at least twice as high . ) for the range in @xmath26 values given above , column densities beyond @xmath41 do not add to the flux , so we have an effective upper limit of @xmath42 for emission .
these results are uncertain by factors of a few because they depend on the poorly known shape of the ionizing spectrum .
the ionization parameter reflects the integrated continuum flux at energies @xmath113.6 ev , but the ionization depends only on the flux above 207 ev .
therefore , different continuum shapes will yield different estimates of the minimum @xmath26 value .
netzer ( 1996 ) argued that highly ionized species like are best described by an x - ray ionization parameter , @xmath43 , that includes energies between only 0.1 and 10 kev .
a similar problem is that the equivalent widths measure the line strengths relative to the _ local _ continuum ( e.g. at 774 ) , even though that continuum has little to do with the production of 774 .
recent calculations by netzer ( 1996 ) predict @xmath03 times lower equivalent widths for 774 compared to figure 6 because of a softer assumed continuum ; the flux ratio @xmath44 is @xmath03 times smaller in netzer s ( 1996 ) continuum than in the mathews & ferland ( 1987 ) continuum used in figure 6 .
conversely , experiments with cloudy show that harder ( flatter ) continua , such as the single power law @xmath45 or the `` baseline '' continuum used by korista ( 1997 ) , roughly double the peak equivalent width compared to figure 6 and shift the rew curve to more than 5 times lower @xmath26 .
even harder power - law continua , such as @xmath46 , still only double the peak equivalent width .
the ratio of observed ( @xmath05.9 ) to largest - predicted ( @xmath040 ) equivalent widths in these extreme cases therefore implies a lower limit on the average covering factor of @xmath015% .
another uncertainty is the importance of continuum pumping to the line emission . if there are large velocity dispersions along radial lines of sight through the belr , the resonant absorption of continuum photons can `` pump '' electrons into upper energy states and significantly enhance the emission line fluxes .
this pumping can take two forms , ( 1 ) resonant line scattering ( absorption and re - emission in the same transition ) , and ( 2 ) continuum fluorescence ( absorption into high energy states followed by cascades through intermediate energy levels ) .
these processes are negligible for metallic lines in `` standard '' belr models ( e.g. fig .
6 above ) because the continuum photons encounter only thermal line widths in the belr . however , if there is substantial turbulence or a large range of radial velocities ( e.g. in an outflow ) , the absorption of continuum photons can compete with other excitation mechanisms particularly in weak resonance lines .
since the velocity field in the belr is essentially unknown , we consider the possible effects of line scattering and continuum fluorescence here .
the amount of scattering in a given line depends on its velocity width and optical depth as encountered by the continuum radiation . for unblended resonance lines that ( 1 ) have no alternate decay routes , ( 2 )
are not collisionally deexcited ( e.g. at high densities ) , and ( 3 ) are on the flat part of the curve - of - growth ( optically thick but no damping wings ) , the scattering equivalent width resulting from thermal or micro - turbulent gas velocities can be derived by integrating over the saturated absorption line profile .
this integration yields @xmath47 where @xmath48 is the covering factor of the emission region ( @xmath49 ) , @xmath50 is the line wavelength , @xmath51 is the line - center optical depth , and @xmath52 is the doppler velocity ( thermal or turbulent ) .
similarly , radial outflows spanning a velocity range @xmath53 in optically thick lines will produce scattering equivalent widths of @xmath54 keeping in mind that @xmath55 in eqn . 1 and that well - separated doublets like 770,780 scatter twice as much flux as single or blended lines , one can use the values of @xmath51 plotted by hamann ( 1995a ) for strictly thermal velocities to estimate @xmath56 for different values of @xmath52 . here
we use cloudy to derive the @xmath51 directly for models identical to figure 6 but with larger @xmath52 .
we find that for clouds with @xmath57 and @xmath58 , the doublet has @xmath59 .
similarly , for @xmath60 and @xmath58 , 1549 has @xmath61 and the 765 and 789 lines both have @xmath62 .
continuum fluorescence will enhance the line emission further via cascades from upper energy states .
the importance of fluorescence versus scattering emission for a given line depends on the number of pumping transitions available to higher states , the optical depths in those transitions , the shape of the incident continuum , and the branching ratios ( including escape probabilities ) out of the upper states ( see also ferguson , ferland & pradhan 1995 ) . experiments with multi - level atoms in cloudy indicate that the equivalent widths due to scattering _ and _ fluorescence are factors of @xmath01.5 to @xmath02 larger than the estimates above for scattering alone .
if we add these illustrative scattering+fluorescence equivalent widths to the rews in figure 6 , we would derive lower minimum covering factors and minimum column densities for the 774 emitting region by factors of a few compared to 3.1.2 .
more importantly , the predicted contributions of 765 and 789 to the measured @xmath0774 lines would be several times larger than our estimates in 3.1.1 , because the theoretical ratio of + to emission becomes larger .
therefore , 774 might not dominate the average @xmath0774 emission if there are large radial velocity dispersions that induce substantial continuum pumping .
a thorough analysis of this possibility is beyond the scope of the present paper ; we note simply that continuum pumping should also enhance the emission in other weak lines .
future studies might test the continuum pumping hypothesis by examining these weak lines and looking for correlations with the measured @xmath0774 feature .
for example , strong continuum pumping in 765 and 789 at @xmath63 should be accompanied by strong pumped emission from 977 , 991 and 834 .
similarly , strong pumping contributions to 774 at @xmath64 should come with highly pumped 615 emission and an / flux ratio that is closer to unity than in figure 6 .
our limited cloudy experiments with @xmath58 did not reveal any significant discrepancies with the data that would immediately rule out or require large pumping contributions .
the _ hst _ spectra of pks 0355@xmath6483 and pg 1522 + 101 ( figs .
1 and 4 ) have the strongest and most reliably measured emission lines near 774 .
figures 7 and 8 show that their , , and emission lines all have similar profiles .
the only significant difference is that and are broader than in pks 0355@xmath6483 .
this small difference might be due to either the larger doublet separations in and ( see fig .
7 ) or small differences in the velocities in their emitting regions . in contrast , the @xmath0774 features in both qsos are dramatically broader than the other bels ( see also table 2 ) .
figure 9 shows similar results for q 1435@xmath6015 , although in this case comparisons with the @xmath0774 line are more uncertain .
overall , we measure fwhms in the @xmath0774 feature that are roughly 2 to 5 times larger than and the other bels in the same spectra ( table 2 ) . figures 10 and 11 show our attempts at fitting the @xmath0774 features in pks 0355@xmath6483 and pg 1522 + 101 using the measured profiles as templates .
we constructed the templates by smoothing the profiles without correcting for the doublet separation ( because it is relatively small , @xmath0500 ) .
the top panels in figures 10 and 11 show the best fits assuming that is the only contributor to the measured feature and both of the doublet members ( 770 and 780 ) have the same redshift and profile as .
the only freely varied parameters were the line strengths .
the best fits were determined by @xmath65 minimization ( with the narrow forest lines excluded from the minimization in pg 1522 + 101 ) .
the -only fits show that the @xmath0774 features are consistent with emission at the redshifts ( see also table 2 and figs .
79 above ) . in pks 0355@xmath6483 the redshifts of these lines are indistinguishable . in pg 1522 + 101
the purported line is redshifted by up to @xmath02000 with respect to the peak , but this shift might not be significant given the profile and centroid uncertainties in the broad @xmath0774 feature .
the much more significant result is that the profiles are far too narrow to explain the @xmath0774 features in terms of .
the lower two panels in figures 10 and 11 plot the best fits with large contributions from 765 and 789 .
these fits assume that and also have the same redshift and profile as , with the fluxes freely varied .
the results using 765 and 789 alone ( middle panels ) account better for the overall width of the @xmath0774 features but they do not match the profile shapes . fits using and together with the two doublet members ( bottom panels ) are much better , but we reiterate the conclusion from 3.1 ; 765 and 789 should be much weaker than they are portrayed in figures 10 and 11 , unless continuum pumping controls the line excitation .
if the observed @xmath0774 features are dominated by , the velocity profiles must be substantially broader than and other bels in the same spectra .
we can estimate the widths of the individual doublet components by fitting the @xmath0774 emisison in pks 0355@xmath6483 and pg 1522 + 101 with one gaussian for each doublet member .
for these fits we require that the two gaussian components have the same redshift and velocity width , but the values of those parameters and the strengths of the lines are all freely varied . the results ( shown in figure 12 ) match the data reasonably well ; the fit profiles are only slightly more centrally peaked than the observed lines . in both qsos , the observed lines are fit slightly better by a 1:1 doublet ratio , but 2:1 ratios are not ruled out .
the centroids of the fits are within 0.3 of the values listed in table 2 and the individual components have fwhm @xmath0 14,000 in both qsos .
this velocity width is similar to the fwhms measured directly from the profiles ( table 2 ) .
these results imply that the velocities in the region are 25 times larger than in the -emitting gas . to understand how much of the flux might come from the lower - velocity gas
, we performed additional fits to the pks 0355@xmath6483 and pg 1522 + 101 profiles using ( 1 ) the template at the redshift ( as in the top panels of figs .
10 and 11 ) plus ( 2 ) a gaussian doublet with freely - varied width and redshift ( as in fig .
12 ) . for both qsos
the @xmath65 minimization forced the template contributions to zero , yielding fits identical to fig .
12 . therefore , to first order , the measured profiles suggest emission from a high - velocity region with little or no contribution from the -emitting gas .
we did not fit the 774 profiles in the mean spectra because they might be broadened by uncertainties in the measured redshifts and by real redshift differences between and the other lines .
also , our selection requirement for measurable emission might bias the absnap and kp - sub samples toward more `` peaky '' line profiles .
nonetheless , we note that the mean profiles are comparable to those in pks 0355@xmath6483 and pg 1522 + 101 and more than twice as broad as both and ( table 2 ) . although possibly spurious , this result is consistent with the individual qsos discussed above .
we also note that the @xmath0774 centroid is slightly redshifted in the mean spectra ( by @xmath01000 ) with respect to the peak ( table 2 ) ; however , this velocity shift is probably smaller than the uncertainties .
the ionization parameters ( @xmath39 ) and covering factors ( @xmath130% ) derived for the emitting region ( 3.1.2 ) are both considerably larger than estimates based on low- to intermediate - ionization lines such as , 2799 , and 1549 .
in particular , the latter lines indicate @xmath66 ( @xcite ) , while values as high as @xmath67 have been inferred when 1034 is also considered ( @xcite ) .
the covering factor of the low - ionization region is expected to be @xmath1010% based on both the ratio of observed to predicted equivalent widths and the infrequency of lyman limit absorption edges at the emission redshift ( see @xcite ) .
the much larger covering factor of the region implies that the high - ionization gas is optically thin in the lyman continuum ( otherwise it would overproduce the emission ) .
optically thin , high - ionization clouds have been proposed before to explain the strength of the and lines ( @xcite ) , and there is evidence from agn variability studies that part of the 1549 emission comes from optically thin gas ( see @xcite and references therein ) .
our results suggest that the highest ionization lines have the largest contributions from optically thin gas .
george , turner & netzer ( 1995 ) reached a similar conclusion based on the tentative detection of a large line equivalent width in x - ray spectra of the seyfert 1 galaxy ngc 3783 . in spite of the low optical depths near the lyman limit
, the -emitting gas has a substantial opacity to soft x - rays .
our calculations show that and are the dominant species of oxygen in the -emitting region ( see , for example , fig . 3 in hamann 1995c ) .
the minimum column density of @xmath68 ( 3.1.2 ) implies a minimum column in + of @xmath69 .
the resulting optical depth through the gas is @xmath70 at the or edges ( based on photoionization cross - sections in @xcite ) .
therefore , the component of the belr would be a strong warm absorber , characterized by + edges at @xmath00.8 kev , if it lies along our line of sight to the x - ray continuum source ( see also @xcite ) .
moreover , the range of physical conditions needed for strong emission , @xmath39 and @xmath71 ( 3.1.2 ) , encompasses the specific conditions estimated for several strong warm absorbers ( e.g. mathur , wilkes & aldcroft 1997 , @xcite ) .
the broad 774 lines reported here clearly identify a belr component with warm absorber properties . these results support models in which the warm absorbers are part of , or closely related to , the belr ( e.g. @xcite and references therein ) .
other studies suggest that warm absorbers are also closely related to the blueshifted absorption lines observed in some qsos and seyfert galaxies ( @xcite ) .
these absorption lines form in outflows with line - of - sight velocities ranging from near zero to @xmath120,000 ( e.g. in the broad absorption line [ bal ] systems ; @xcite ) .
there is no direct evidence for outflow in the 774 emission - line data , but the ultra - broad line widths reported here , with fwhm @xmath0 14,500 in two well - measured cases , are consistent with high velocities in the warm absorber gas .
the large fwhms are also consistent with emission from the `` very broad line region '' ( vblr ) , which was identified by ferland , korista & peterson ( 1990 ) from the non - varying line wings in a seyfert 1 galaxy .
ferland ( 1990 ) present independent arguments that the vblr is highly ionized , optically thin at the lyman limit and the likely source of x - ray warm absorption ( see also @xcite ) .
the large covering factors and warm absorber - like properties of the emitting regions suggest that warm absorbers should be common in qsos .
the most recent x - ray surveys indicate that warm absorbers are common in seyfert 1 nuclei ( with a @xmath150% detection rate ; @xcite ) but _ rare _ in qsos ( @xmath05% detection rate ; @xcite ) .
the rarity of warm absorbers in qsos might be reconciled with the emission - line data if ( 1 ) qso ionizing spectra are harder than the mathews & ferland ( 1987 ) continuum , requiring smaller covering factors ( 3.1.2 ) , ( 2 ) large radial velocity dispersions ( @xmath11000 ) in the belr enhance the emission in 774 and other weak lines via continuum pumping ( 3.1.3 ) , or ( 3 ) qsos are observed over limited viewing angles that usually avoid gas along our line - of - sight to the x - ray continuum source .
viewing angles that avoid the belr might occur naturally in the recent disk - wind models ( @xcite ) . in these models ,
all of the bels form close to the accretion disk and the highest ionization lines form at the smallest disk radii .
murray ( 1996 ) pointed out that the exceptionally broad 774 profiles might be due to enhanced rotational broadening associated with the inner disk .
however , that interpretation might conflict with our estimates of large covering factors , which require substantial emission from regions away from the disk plane .
better data and more specific comparisons with the models are needed before drawing firm conclusions .
the large 774 line widths compared to 1549 and especially 1034 might provide strong constraints on models of the emitting regions in general .
the calculations in figure 6 indicate that the gas producing the ultra - broad line should also produce considerable emission .
this ultra - broad component to the lines might be hard to hide in the measured profiles , particularly for sources like pks 0355@xmath6483 where any high - velocity emission must be weak compared to 774 ( figs . 1 and 7 ) .
however , we can lower the predicted / flux ratio by adopting large values of @xmath26 and/or lower @xmath72 compared to that used in figure 6 .
( adding continuum pumping via large turbulent velocities [ 3.1.3 ] does not significantly alter the / flux ratio . ) limited experiments with cloudy indicate that 1034 can be @xmath02 or more times weaker than 774 if @xmath73 and @xmath74 .
larger gas densities and/or different ionizing continuum shapes can further reduce the theoretical / flux ratio ( see also @xcite ) . alternatively , or perhaps in addition , the measured @xmath0774 lines might be broadened by blending with 765 and 789 if their emission is enhanced by continuum pumping ( 3.1.3 and 3.2.1 ) . in that case , the true line widths could be comparable to the other bels and there would be no need for an ultra - high velocity emitting region .
the mean spectrum of an unbiased sample ( kp - full ) of 11 radio - loud and radio - quiet qsos clearly shows broad - line emission at @xmath0774 with an average rew of @xmath05.9 and an average flux equal to @xmath042% of 1034 .
this result supports our previous conclusion ( based on a smaller sample ) that @xmath0774 emission is common in qsos .
hst_-fos spectra of individual qsos reveal a strong and broad @xmath0774 line in pg 1522 + 101 ( also @xcite ) , and they support the prior claims of strong @xmath0774 lines in pks 0355@xmath6483 and q 1435@xmath6015 ( @xcite ) .
however , the new spectra contradict the report ( based on old _ iue _ data ) that the @xmath0774 line is strong in pg 1148 + 549 ( @xcite ) .
the centroids of the observed @xmath0774 features support as the single most likely identification , but the velocity widths are significantly larger than 1034 and all other bels in two well - measured qsos ( pks 0355@xmath6483 and pg 1522 + 101 ) and probably in the mean spectra .
the @xmath0774 line widths , ranging from fwhm @xmath75 11,000 to 15,000 , are @xmath02 to 5 times larger than the lower - ionization bels .
these ultra - broad profiles are consistent with significant contributions from 765 and 789 emission . however , photoionization calculations coupled with measurements of 1549 constrain the and line strengths to be at least several times weaker than the average @xmath0774 emission ( 3.1.1 ) .
we conclude that exceptionally broad 774 dominates the average feature and is the only significant contributor when this line is strong compared to 1549 . the only way to avoid this conclusion is to invoke large line - of - sight velocity dispersions ( @xmath11000 ) in the belr , so that the resonant absorption of continuum photons greatly enhances the and line strengths relative to both the continuum and 1549 ( 3.1.3 ) .
if dominates the @xmath0774 emission , fits to the line profiles indicate that most of the flux comes from a distinct high - velocity region , with little or no contribution from the low - velocity gas responsible for and the other lower - ionization lines ( 3.2.1 ) .
the photoionization calculations also indicate that the emitting regions have ionization parameters @xmath39 , total column densities @xmath71 , and covering factors @xmath130% of 4@xmath30 steradians ( for solar abundances and a standard ionizing spectrum ; 3.1.2 ) .
these results are uncertain by factors of a few because of the poorly known shape of the far - uv spectrum and the possible large contributions to the line fluxes from continuum pumping .
nonetheless , the component of the belr must be considerably more ionized and probably more extensive than the regions producing lines like , 2799 , and 1549 .
the physical conditions in the emitting clouds are characteristic of the highly - ionized `` warm '' absorbers observed in soft x - rays , suggesting a close relationship between the belr and the warm absorber gas .
the emitting regions would themselves produce warm absorption ( e.g. bound - free edges of + at @xmath00.8 kev ) if they lie along our line of sight to the x - ray continuum source .
we are grateful to g. j. ferland for providing the cloudy software and r. weymann for the reduced _ hst _ key project spectra .
we also thank l. zuo and d. tytler for help with the absnap data , g. j. ferland , k. korista and a. laor for useful discussions , and r. lyons for comments on this manuscript and help with the data processing .
finally , we thank an anonymous referee for helpful suggestions . this work was supported by nasa grants nag 5 - 1630 and nag 5 - 3234 , and by the space telescope science institute through the archival grant ar-5292 - 93a and the guest observer grant go-6103 - 94a .
bahcall , j. n. , 1993 , , 87 , 1 baldwin , j. 1977 , , 214 , 679 baldwin , j. a. , ferland , g. j. , hamann , f. , carswell , r. , phillips , m. , wilkes , b. , & williams , r. e. 1996 , , 461 , 664 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 cohen , r. , beaver , e. a. , burbidge , e. m. , hamann , f. , junkkarinen , v. t. , & lyons , r. w. 1994 , in poster papers accompanying _ the analysis of emission lines _ , ed .
m. livio , stsci symposium , ( cambridge : new york ) 10 crenshaw , d. m. , bruegman , o. w. , & norman , d. j. 1990 , , 102 , 463 davidson , k. , & netzer , h. 1979 , rev .
phys . , 51 , 715 diplas , a. , & savage , b. d. 1994 , , 427 , 274 espey , b. r. , carswell , r. f. , bailey , j. a. , smith , m. g. , & ward , m. j. 1989 , , 342 , 666 fabian , a. c. 1994 , , 46 , l59 ferguson , j. w. , ferland , g. j. , & pradhan , a. k. 1995 , , 438 , l55 ferland , g. j. 1996 _ hazy , a brief introduction to cloudy _
, university of kentucky , department of physics and astronomy , internal report ferland , g. j. , korista , k. t. , & peterson , b. m. 1990 , , l21 ferland , g. j. , peterson , b. m. , horne , k. , welsch , w. f. , & nahar , s. n. 1992 , , 387 , 95 fiore , f. , elvis , m. , mathur , s. , wilkes , b. j. , & mcdowell , j , 1993 , , 415 , 129 foltz , c. b. , chaffee jr .
, weymann , r. j. , & anderson , s. f. 1988 , in _ qso absorption lines : probing the universe _ , eds . j. c. blades , d. a. turnshek , & c. a. norman ( cambridge : cambridge univ .
press ) , p. 53
george , i. m. , turner , t. j. , netzer , h. , nandra , k. , mushotzky , r. f. , & yaqoob , t. 1997 , apjs , in press george , i. m. , turner , t. j. , & netzer , h , 1995 , , 438 , l67 halpern , j. p. 1984
, , 281 , 90 hamann , f. , barlow , t. a. , beaver , e. a. , burbidge , e. m. , cohen , r. d. , junkkarinen , v. , & lyons , r. 1995c , , 443 , 606 hamann , f. , & ferland , g. j. 1992 , , 391 , l53 hamann , f. , & ferland , g. j. 1993 , , 418 , 11 hamann , f. , & korista , k. t. 1996 , , 464 , 158 hamann , f. , shields , j. c. , cohen , r. d. , junkkarinen , v. t. , & burbidge , e. m. 1996 , in `` emission lines in active galaxies , '' ( i.a.u colloquium 159 , shanghai , china ) , eds .
b. peterson , f .- z .
cheng , and a. s. wilson , ( asp conf .
series ) , 96 hamann , f. , shields , j. c. , ferland , g. j. , & korista , k. 1995a , , 454 , 688 hamann , f. , zuo , l. , & tytler , d. 1995b , , 444 , l69 hewitt , d. & burbidge , g. 1993 , , 87 , 451 kinney , a. l. , rivolo , a. r. , & koratkar , a. r. 1990 , , 75 , 645 kinney , a. l. , bohlin , r. c. , blades , j. c. , & york , d. g. 1991 , , 75 , 645 korista , k. , baldwin , j. , ferland , g. , & verner , d. 1997 , , 108 , 401 kwan , j. , & krolik , j. 1981 , , 250 , 478 lanzetta , k. m. , turnshek , d. a. , & sandoval j. 1993 , , 84 , 109 laor , a. , jannuzi , b. t. , green , r. f. , & boroson , t. a. 1997a , , in press laor , a. , bahcall , j. n. , jannuzi , b. t. , schneider , d. p. , green , r. f. , & hartig , g. f. 1994 , , 420 , 110 laor , a. , bahcall , j. n. , jannuzi , b. t. , schneider , d. p. , green , r. f. , & hartig , g. f. 1995 , , 99 , 1 laor , a. , fiore , f. , elvis , m. , wilkes , b. j. , & mcdowell , j. c. 1997b , , 477 , 93 marziani , p. , & sultenic , j. w. 1993 , , 409 , 612 mathews , w. g. , & ferland , g. j. 1987 , , 323,456 mathur , s. 1994 , , 431 , l75 mathur , s. , elvis , m. , & singh , k. p. 1995
, , 455 , l9 mathur , s. , wilkes , b. , aldcroft , t. 1997 , , in press mathur , s. , wilkes , b. , elvis , m. , & fiore , f. 1994 , , 434 , 493 mcdowell , j. c. , canizares , c. , elvis , m. , lawrence , a. , markoff , s. , mathir , s. , & wilkes , b. j. 1995 , , 450 , 585 mihara , t. , matsuoka , m. , mushotzky , r. f. , kunieda , h. , otani , c. , miyamoto , s. , & yamauchi , m. 1994 , , 46 , l59 murphy , e. m. , lockman , f. j. , laor , a. , & elvis , m. 1996 , , 105 , 369 murray , n. 1996 , private comm .
murray , n. , & chiang , j. 1995 , , 454 , l105 murray , n. , chiang , j. , grossman , s. a. , & voit , g. m. 1995 , 451 , 498 nandra , k. , & pounds , k. a. 1994 , , 268 , 405 netzer , h. 1993 , , 411 , 594 netzer , h. 1996 , , 473 , 781 nichols , j. s. , & linsky , j. l. 1996 , , 111 , 517 osterbrock , d. e. 1986 , , 24 , 171 osterbrock , d. e. 1989 , _ astrophysics of gaseous nebulae and active galactic nuclei _ , ( mill valley : university science books ) otani , c. , 1996 , , 48 , 211 reynolds , c. s. 1997 , , 286 , 513 reynolds , c. s. , & fabian , a. c. 1995 , , 273 , 1167 schmidt , m. , & green , r. f. 1983 , , 269 , 352 schneider , d. p. , 1993 ,
apjs , 87 , 45 shields , j. c. , & ferland , g. j. 1993 , , 402 , 425 shields , j. c. , ferland , g. j. , & peterson , b. m. 1995 , , 441 , 507 shields , j. c. , & hamann , f. 1997 , , 481 , 752 turner , t. j. , & pounds , k. a. 1989 , , 240 , 833 turner , t. j. , george , i. m. , & mushotzky , r. f. 1993b , , 412 , 72 turner , t. j. , nandra , k. , george , i. m. , fabian , a. c. , & pounds , k. a. 1993a , , 419 , 127 tytler , d. , & fan , x .- m . 1992 , , 79 , 1 tytler , d. , & zuo , l. 1994 , private comm .
ulrich , m. h. 1988 , mnras , 230 , 121 verner , d. , verner , e. m. , & ferland , g. j. 1996 , atomic data and nuclear data tables , 64 , 1 weymann , r. j. , morris , s. l. , foltz , c. b. , & hewett , p. c. 1991,apj , 373 , 23 wiese , w. l. , smith , m. w. , & glennon , b. m. 1966 , atomic transition probabilities ( nbs - nsrds 4 , vol
. 1 ) ( washington : gpo ) wills , b. j. , 1995 , , 447 , 139 zheng , w. , kriss , g. a. , telfer , r.c . , grimes , j. p. , & davidson , a. f. 1997 , , 475 , 469 | 774 is an important tracer of the high - ionization gas in qsos .
we examine the emission - line properties using new _
hst_-fos spectra of four sources , mean spectra derived from two qso samples in the _ hst _ archives , and new photoionization calculations .
the results support our previous claim that broad lines are common in qsos , with an average flux of @xmath042% of 1034 and velocity widths that are @xmath02 to 5 times larger than , 1549 and other broad lines in the same spectra . the strongest and most reliably measured 774 lines ( in two sources ) have fwhm @xmath0 14,500 .
line profile fits in these cases show that the unusually large widths might be caused by blending with emission from 765 and 789 .
however , standard photoionization calculations indicate that , and all other lines near this wavelength should be too weak , leaving ( very broad ) as the only viable identification for the @xmath0774 feature .
( this conclusion might be avoided if there are large radial velocity dispersions [ @xmath11000 ] in the emitting region and the resonant absorption of continuum photons enhances the flux in weaker lines . )
the calculations also indicate that the emitting regions have ionization parameters in the range @xmath2 , total hydrogen column densities of @xmath3 @xmath4 , and an average covering factor of @xmath130% ( for solar abundances and a nominal qso continuum shape ) .
the emitting region is therefore more extensive , more highly ionized , and has much higher velocities than the rest of the broad emission line region ( belr ) .
this highly - ionized belr component would be a strong x - ray `` warm '' absorber if it lies along our line - of - sight to the x - ray continuum source . |
let @xmath0 be a field and @xmath1 $ ] be the polynomial ring in @xmath15 variables with coefficients from @xmath0 . the ring @xmath2 is naturally equipped with a @xmath16-grading .
stanley decomposition _ of a finitely generated @xmath16-graded @xmath2-module @xmath3 is a representation as a finite direct sum of @xmath0-vector spaces @xmath17,\ ] ] where each @xmath18 is homogeneous , and @xmath19 is a subset of @xmath20 .
stanley depth _ of this decomposition is @xmath21 , where @xmath22 is the cardinality of @xmath19 .
stanley depth of @xmath3 _ is defined to be @xmath23 in @xcite , stanley conjectured that @xmath24 for all finitely generated @xmath16-graded @xmath2-modules @xmath3 .
this conjecture has been confirmed in several special cases , for instance , when the module @xmath3 is clean in the sense of dress @xcite , that is , it allows a homogeneous prime filtration @xmath25 with @xmath26 .
the stanley - reisner ring @xmath27 $ ] of a simplicial complex @xmath28 is clean if and only if @xmath28 is shellable .
it was proved in @xcite that stanley s conjecture also holds for @xmath29 , where @xmath30 is a monomial ideal and @xmath31 . for other recent developments , see for example @xcite .
however , stanley s conjecture still remains open , partly because of the difficulty of computing the stanley depth of @xmath3 .
herzog , vladoiu and zheng proposed a method of attack in the special case where @xmath32 with @xmath33 being monomial @xmath2-ideals in @xcite .
they associate @xmath34 with a poset @xmath35 , and thus the computation of @xmath36 is reduced to a corresponding computation on the poset @xmath35 .
we briefly review their method for squarefree monomial ideals in section [ sec : prelim ] . though
a direct implementation of this method is computationally expensive , it has yielded a wealth of information ; see @xcite for its applications . in this paper , we investigate the _ squarefree veronese ideal _
@xmath8 , which is generated by all squarefree monomials of degree @xmath9 in @xmath1 $ ] .
the ideal @xmath8 is the stanley - reisner ideal of the @xmath37-skeleton of the @xmath15-simplex , hence the stanley - reisner ring @xmath38 is cohen - macaulay of dimension @xmath39 .
the ideal @xmath8 is a polymatroidal ideal @xcite , and has linear quotients .
the rees algebras of squarefree veronese ideals and their @xmath40-invariants were studied in @xcite .
herzog et al . showed in @xcite that stanley s conjecture holds for both @xmath8 and @xmath38 .
however , it is still difficult to determine @xmath41 .
the ideal @xmath42 is the homogeneous maximal ideal @xmath43 of @xmath2 .
herzog et al .
conjectured in @xcite that @xmath44 .
this formula was later confirmed by bir , howard , keller , trotter and young in @xcite using combinatorial techniques upon which this paper builds .
the stanley depth of @xmath8 in the other situation remains unexplored in general , although cimpoea considered a similar problem in @xcite . in this paper , we prove an exact formula for the stanley depth of @xmath8 for certain values of @xmath15 and @xmath9 and a bound for others .
our principal result is the following theorem : [ thm : main ] let @xmath0 be a field , @xmath1 $ ] be a polynomial ring in @xmath15 variables over @xmath0 .
furthemore , let @xmath8 be the squarefree veronese ideal in @xmath2 , generated by all squarefree monomials of degree @xmath9 . 1 . if @xmath45 , then the stanley depth @xmath46 .
2 . if @xmath12 and @xmath13 , then @xmath47 we begin the paper by reviewing the method of herzog et al . for associating a poset with a monomial ideal .
we also introduce our notation and prove some preliminary results in section [ sec : prelim ] .
section [ sec : proof ] is devoted to proving theorem [ thm : main ] in three stages .
by definition , the ideal @xmath8 is squarefree . the method of herzog et al . for determining the stanley depth of a squarefree monomial ideal @xmath48
using posets can be summarized in the following way .
let @xmath49 be the unique set of minimal monomial generators of @xmath48 .
for each @xmath50 , denote @xmath51 .
the monomial @xmath52 is squarefree precisely when @xmath53 for @xmath54 . by definition ,
all the @xmath55 are squarefree .
if @xmath52 is squarefree , let @xmath56:=\{1,\dots , n\}$ ] be the support of @xmath52 .
for the squarefree ideal @xmath57 , there is a poset with ground set @xmath58 \mid c \text { contains $ \supp(v_i)$ for some $ i$}}\ ] ] partially ordered by inclusion .
the poset @xmath59 is a subposet of the boolean algebra of subsets of @xmath60 $ ] ; it consists of all supersets of the support of the generators of @xmath48 .
for every @xmath61 with @xmath62 , define the interval @xmath63 $ ] to be @xmath64 .
let @xmath65 $ ] be a partition of @xmath59 , and for each @xmath66 , let @xmath67 be the tuples such that @xmath68 .
there is then a stanley decomposition @xmath69 of @xmath48 : @xmath70.\ ] ] it is clear that @xmath71 .
most importantly , herzog et al . showed in @xcite that if @xmath48 is a squarefree monomial ideal , then @xmath72 before introducing our combinatorial approach to this problem , we establish some bounds on the stanley depth of squarefree veronese ideals .
we also prove other results which show that our approach to proving theorem [ thm : main ] can focus on values of @xmath15 having a particular form .
[ lem : min - sdepth ] if @xmath48 is a squarefree monomial ideal and @xmath73 is the minimal monomial generating set of @xmath48 , then @xmath74 .
this follows directly from the method of herzog et al .
[ lem2 ] for integers @xmath75 , @xmath76 .
since @xmath8 is generated by monomials of degree @xmath9 , by lemma [ lem : min - sdepth ] we have @xmath77 .
the poset @xmath78 has a partition @xmath79 $ ] , satisfying @xmath80 . for each interval
@xmath81 $ ] in @xmath82 with @xmath83 , we have @xmath84 . furthermore , there are @xmath85 subsets of cardinality @xmath86 in this interval .
since these intervals are disjoint , counting the number of subsets of cardinality @xmath86 , we have the inequality : @xmath87 after simplification , we get the desired estimate .
the following corollary is now immediate . if @xmath88 , then @xmath89 .
we expect the following conjecture , which generalizes our theorem [ thm : main ] , to be true .
[ conj - ind ] for positive integers @xmath75 , @xmath90 .
this formula is already known to be valid for @xmath91 , see @xcite . a further implementation by computer shows that it is also valid for @xmath92 .
the structure of the boolean algebra of subsets of @xmath60 $ ] provides the following lemma which allows us to use known bounds on the stanley depth of a squarefree veronese ideal to establish other bounds .
[ plus1 ] suppose @xmath93 are positive integers such that @xmath94 and @xmath95
. then @xmath96 .
the poset @xmath97 has a partition @xmath98 with @xmath99 , while @xmath78 has a partition @xmath100 with @xmath101 .
now @xmath102\in \calp_1 } \left[c\cup\set{n+1},d \cup\set{n+1}\right]\right ) \cup \calp_2\ ] ] is a partition of @xmath103 with @xmath104 .
notice that when @xmath105 , we have @xmath106 .
in fact , the following lemma establishes that if we can prove conjecture [ conj - ind ] for such values of @xmath15 , it is true for all @xmath15 and @xmath9 , @xmath75 .
[ reduction ] suppose @xmath107 for all positive integers @xmath108 and @xmath9 .
then @xmath109 for all @xmath108 such that @xmath110 .
the proof proceeds by double induction , first on @xmath9 and then on @xmath15 . for the induction on @xmath9 ,
notice that in @xcite , biro et al . proved that when @xmath91 , @xmath111 .
so if @xmath112 then @xmath113 so suppose that for all @xmath114 and for all @xmath115 , we have for all @xmath116 that @xmath117 . now consider @xmath118 .
for this , we use induction on @xmath15 . for the base case ,
suppose @xmath119 , which is the smallest @xmath15 can be since @xmath120 and @xmath121 .
this case follows from the assumption of the lemma with @xmath122 . by assumption
we have that if @xmath123 for some @xmath108 , then @xmath124 .
for all @xmath125 for which @xmath126 , we have @xmath127 .
thus suppose there exists some @xmath128 such that @xmath129 as @xmath130 , our inductive hypothesis on @xmath15 yields @xmath131 .
also , @xmath132 , and therefore we have @xmath133 thus the inductive hypothesis for @xmath9 yields @xmath134 .
thus , by lemma [ plus1 ] , @xmath135 , as desired .
this concludes both inductions and finishes the proof of the lemma .
observe that if @xmath136 then @xmath137 .
thus if @xmath138 for all positive integers @xmath108 and @xmath9 , then by lemma [ lem2 ] and lemma [ reduction ] we have that for all @xmath139 , @xmath140 thus , in order to prove conjecture [ conj - ind ] it suffices to show that @xmath141 for all positive integers @xmath108 and @xmath9 .
in other words , we may restrict our attention to @xmath8 where @xmath142 and @xmath143 for some @xmath121 . in this paper
, we will frequently refer to the boolean algebra of subsets of @xmath60 $ ] partially ordered by inclusion as the _ @xmath15-cube_.
a @xmath144-set is a subset of @xmath60 $ ] of size @xmath144 .
recall that for @xmath145 in @xmath60 $ ] , we have defined the _
@xmath63 $ ] as the set of all @xmath146 $ ] such that @xmath147 .
notice that @xmath63 $ ] is isomorphic to the @xmath148-cube .
we say that an interval @xmath63 $ ] _ covers _ a set @xmath149 if @xmath150 $ ] . for our purposes , it is helpful to think of a subset @xmath2 of @xmath60 $ ] by evenly distributing the points @xmath151 around a circle in the plane , with the elements of @xmath2 depicted as solid points and the elements of @xmath60\backslash s$ ] depicted as open points . as such , the circle will have inherent clockwise and counterclockwise directions .
we will refer to this as the _ circular representation of @xmath60$]_. in order to facilitate our proofs in this paper , we first establish some definitions . given the circular representation of @xmath60 $ ] , a _
block _ is a subset of consecutive points on the circle . for @xmath152
$ ] we denote by @xmath153 $ ] the block starting at @xmath66 and ending at @xmath154 when traversing the circular representation of @xmath60 $ ] clockwise .
let @xmath155 $ ] and let @xmath156 with @xmath157 be a _
density_. the _ block structure of @xmath158 with respect to @xmath159 _ is a partition of the elements of the circular representation of @xmath60 $ ] into clockwise - consecutive blocks @xmath160 such that * the first ( going clockwise ) element @xmath161 of @xmath162 is in @xmath158 , * for all @xmath163 $ ] , @xmath164 , * for all @xmath163 $ ] , @xmath165 , * for all @xmath166 such that @xmath167 \subsetneq b_i$ ] , @xmath168|+1 \leq \delta \cdot |[b_i , y ] \cap a|.$ ] the purpose of ( iii ) and ( iv ) in the definition of block structure above is to ensure that the density of elements of @xmath158 in a block is close to @xmath169 . for two examples of a block structure , see fig .
[ fig : block - structure ] . block - structure ( 31,37)2 ( 38.3,28.3)4 ( 8.5,21.3)12 ( 23.3,39)@xmath170 ( 36,20)@xmath171 ( 11.5,13)@xmath172 ( 11,31)@xmath173 ( 89,29)4 ( 90,22)5 ( 78.5,9)8 ( 59,21)12 ( 67,37)15 ( 85,14)@xmath171 ( 62,13)@xmath172 ( 59,25)@xmath173 ( 62,33)@xmath170 ( 71,38)@xmath174 ( 84,37)@xmath175 for @xmath176 the block structure for a set @xmath158 on @xmath60 $ ] exists and is unique . it is clear for @xmath159 in this range that a block structure can be constructed by iteratively expanding blocks .
that is , for each element of @xmath158 assign the next @xmath177 elements cyclically to the block containing @xmath158 . if any of these blocks contain another element of @xmath158 , merge the two blocks and then add points from @xmath60\backslash a$ ] if the number of elements of @xmath158 in the block allows .
this process will eventually terminate , yielding a block structure , since @xmath178 .
now suppose that @xmath179 and @xmath180 are two block structures for the set @xmath158 with different blocks
. then there are two blocks , say @xmath162 and @xmath181 , which are not the same and overlap . from ( iii ) and ( iv ) it is clear that @xmath182 and so without loss of generality we may assume that @xmath161 precedes @xmath183 in the clockwise ordering .
we then note that if @xmath184 is nonempty ; that is , there is an element of @xmath181 following @xmath162 , then @xmath185 where the inequality comes from ( iv ) , applied to the initial segments of @xmath162 and @xmath181 , respectively .
note that this contradicts ( iii ) , as @xmath186 is not less than @xmath187 .
furthermore , this implies that @xmath188 .
but since @xmath189 , there must be some block @xmath190 ( @xmath191 ) that contains @xmath161 .
clearly , @xmath192 is nonempty .
specifically , it contains @xmath181 , and so by the above , @xmath193 .
but then @xmath194 , contradicting that the block structure is a partition .
we denote the set @xmath195 by @xmath159-@xmath196 and the union @xmath197 by @xmath198 .
each @xmath199 for @xmath163 $ ] is called a _ gap .
_ we denote the set @xmath200 by @xmath159-@xmath201 and the union @xmath202 by @xmath203 .
given density @xmath159 , let @xmath204 be the function that maps each @xmath155 $ ] with @xmath205 to @xmath206 $ ] . throughout this paper
we will be concerned with intervals of the form @xmath207 $ ] .
we will also require the use of some graph theoretic notions .
the _ neighborhood _ of a vertex @xmath208 is the set of all vertices adjacent to @xmath208 and is denoted @xmath209 . if @xmath2 is a set of vertices , then @xmath210 denotes all vertices not in @xmath2 that are adjacent to a vertex in @xmath2 .
bipartite graph _ is a graph whose vertex set can be partitioned into sets @xmath211 and @xmath212 such that all edges of @xmath213 have one end in @xmath211 and the other in @xmath212 .
a _ matching _ in a graph is a set of edges in which no two edges in the set have a common end .
a _ complete matching from @xmath211 to @xmath212 _ in a bipartite graph @xmath213 with bipartition @xmath214 is a matching containing precisely one edge incident with each vertex in @xmath211 .
a famous result concerning complete matchings is the following theorem due to hall .
[ thm : hall ] let @xmath213 be a bipartite graph with bipartition @xmath215 .
then @xmath213 has a complete matching from @xmath211 to @xmath212 if and only if @xmath216 for all @xmath217 . in this paper
, we will find the following consequence of hall s theorem useful .
[ cor : spec - hall ] let @xmath213 be a bipartite graph with bipartition @xmath218 .
if each vertex in @xmath211 has degree @xmath144 and @xmath144 is the maximum degree of @xmath213 , then @xmath213 has a complete matching from @xmath211 to @xmath212 .
the following lemma provides the first step toward proving theorem [ thm : main ] .
[ lem : c2 ] let @xmath12 be an integer and @xmath219
. then @xmath220 .
it suffices to show that the @xmath9-sets in @xmath78 can be covered by intervals @xmath63 $ ] with @xmath221 .
consider the bipartite graph in which @xmath211 consists of the @xmath9-sets of @xmath60 $ ] , @xmath212 consists of the @xmath222-sets , and a @xmath9-set is adjacent to precisely the @xmath222-sets of which it is a subset .
then every vertex has degree @xmath86 , so by applying corollary [ cor : spec - hall ] to @xmath211 , there is a complete matching from @xmath211 to @xmath212
. the edges of this matching then give rise to intervals of the desired form .
lemma [ lem : c2 ] is the case of @xmath106 with @xmath223 .
the next section tackles the more challenging cases of @xmath224 and @xmath225 .
to prove theorem [ thm : main ] , it suffices by lemma [ lem2 ] , lemma [ plus1 ] , lemma [ lem : c2 ] and the proof of lemma [ reduction ] to show that if @xmath106 and @xmath226 , then @xmath227 . to do this ,
we first prove some lemmas about intervals of the form @xmath228 $ ] . our first lemma deals with the disjointness of these intervals in certain cases .
[ no collisions ] given a positive integer @xmath15 , let @xmath229 $ ] with @xmath230 and @xmath231 , and let @xmath156 with @xmath157 . if @xmath232 , then @xmath233 $ ] does not intersect @xmath234 $ ] .
suppose not .
then @xmath233 $ ] and @xmath234 $ ] have nontrivial intersection .
in particular , @xmath235 \cap [ a',f_{\delta}(a')]$ ] .
this implies that @xmath236 for all @xmath237 , and @xmath238 for all @xmath239 . as @xmath230
, there exists an @xmath240 satisfying @xmath241 .
let @xmath242 be as in the definition of the block structure of @xmath158 and , without loss of generality , suppose @xmath243 .
note that @xmath244 therefore , the block in @xmath159-@xmath245 that contains @xmath246 , say @xmath247 , must also contain @xmath248 , the first element of @xmath173 .
as @xmath249 and @xmath250 it must be that @xmath251 .
let @xmath252 be the next element of @xmath253 found when proceeding clockwise around the circular representation of @xmath60 $ ] .
clearly @xmath254 , so it must be the case that @xmath255 as otherwise @xmath256 . proceeding clockwise in this manner , we find @xmath257 and @xmath258 . again , using the fact that @xmath259 and the fact that @xmath260 , we find that @xmath261 , the first element of @xmath174 , is also in @xmath247 .
applying the same argument as was applied to @xmath173 we find that @xmath257 and @xmath262 .
finally , proceeding clockwise and using analogous arguments , we conclude that @xmath263 .
however , @xmath264 , contradicting the fact that @xmath265 . we also require the following lemma about the size of @xmath266 .
[ right size ] let @xmath267 be positive integers and let @xmath268 .
let @xmath155 $ ] be a @xmath9-set
. then @xmath269 .
since @xmath270 , it follows from the definition of the block structure of @xmath158 with respect to the density @xmath108 that @xmath271 . since @xmath158 is a @xmath9-set , we find that @xmath272 .
thus , @xmath273 .
we now show that the intervals of the form @xmath274 $ ] with @xmath275 cover all the @xmath222-subsets of @xmath60 $ ] .
[ d+1 ] let @xmath267 be positive integers and let @xmath276 .
let @xmath277\mid a \subseteq [ n],~|a|=d \}.\ ] ] then each @xmath222-subset of @xmath60 $ ] is covered by a unique element of @xmath278 .
lemma [ right size ] implies that @xmath279 .
thus , we may apply lemma [ no collisions ] , which implies that each @xmath222-set is covered by at most one element of @xmath278 .
so it suffices to show that the number of @xmath222-sets covered by the elements of @xmath278 is precisely @xmath280 . by lemma [ right size ] , each element of @xmath278 is a @xmath281-cube .
thus , each element of @xmath278 covers exactly @xmath282 sets of size @xmath222 . since @xmath283 , we find that the elements of @xmath278 cover @xmath284 sets . moreover , since @xmath106 , it is easy to verify that @xmath285 as desired .
the following theorem resolves the first of our two remaining cases , that of @xmath224 .
[ c=3 ] let @xmath9 be a positive integer and let @xmath286 .
then there exists a partition of the subsets of @xmath60 $ ] of size at least @xmath9 into intervals @xmath63 $ ] such that @xmath287 .
let @xmath288\mid a \subseteq [ n ] , |a| = d\}$ ] . by lemma
[ right size ] , @xmath289 for all @xmath9-sets @xmath158 . by lemma [ d+1 ] ,
all @xmath222-sets are covered by the elements of @xmath278 .
since lemma [ right size ] implies that @xmath290 , we can apply lemma [ no collisions ] to the @xmath9-sets to show that @xmath291 for all @xmath292 .
moreover , the @xmath293-sets that are not covered by any element of @xmath278 and the subsets of @xmath60 $ ] of size at least @xmath294 can be covered with trivial intervals .
therefore we have found a partition of the desired type .
the next lemma provides the final tool required for our proof .
it shows that after selecting the intervals of the form @xmath274 $ ] with @xmath275 for our partition , any uncovered set has all of its supersets uncovered as well .
[ uncovered top ] let @xmath267 be positive integers with @xmath295 .
let @xmath268 and let @xmath296 be a positive integer .
let @xmath297\mid a \subseteq [ n ] , |a|=d\}$ ] .
suppose @xmath298 is a @xmath299-set that is not covered by any element of @xmath278 .
then there is no superset of @xmath298 that is covered by an element of @xmath278 .
suppose not .
let @xmath300 be a superset of @xmath298 that is covered by some @xmath301 .
by lemma [ right size ] , @xmath48 covers some @xmath302-set , say @xmath303 , that is also a superset of @xmath298 .
thus , it suffices to prove that there is no superset of @xmath298 of size @xmath302 that is covered by an element of @xmath278 . to this end , let @xmath304 \in \mathcal{i}$ ] with @xmath305 .
let @xmath306 .
we call such a combination of sets @xmath307 a _ pair_. we call the pair @xmath307 _ optimal _ if , among all pairs , @xmath308 is minimized .
let @xmath309 be an optimal pair .
notice that if @xmath310 , then since we know @xmath311 , we have that @xmath312 . in this case ,
@xmath313 $ ] covers @xmath298 , a contradiction .
thus @xmath314 .
consider @xmath315 .
let @xmath316-@xmath317 such that @xmath318 .
let @xmath319 be the first element in @xmath320 counterclockwise from @xmath321 in the circular representation of @xmath60 $ ] , and let @xmath322 be the last element of @xmath321 ( that is , the most clockwise element of @xmath321 ) .
the point @xmath322 exists and is distinct from @xmath246 since for @xmath323 , a block can not end in an element of @xmath158 .
let @xmath324 be the successive elements of the gaps of @xmath325 , indexed _
counterclockwise _ from @xmath319 . fix @xmath326 as small as possible so that @xmath327 .
such an @xmath326 must exist , as otherwise all the gap points belong to @xmath328 , so we would have @xmath329 , but @xmath330 .
we now define a sequence of pairs @xmath331 .
let @xmath332 and then define @xmath333 .
notice that @xmath334 , where @xmath335 is the last element of the block of @xmath108-@xmath336 that contains @xmath337 .
we then define @xmath338 .
see , for example , fig .
[ fig : changing - block - structure ] . shift - block ( 5,8)@xmath339 ( 21,24)@xmath321 ( 25,15)@xmath337 ( 33,7)@xmath335 ( 65,7)@xmath339 ( 78,24)@xmath321 ( 84,15)@xmath337 ( 92,8)@xmath335 now note that we have the following property from our definition .
@xmath340 for @xmath341 , we know that @xmath342 , and therefore @xmath343 . from the computation above , we see that this implies @xmath344 .
therefore , the pair @xmath345 is optimal for @xmath346 . on the other hand , @xmath327 , and therefore @xmath347
, so we have @xmath348 , contrary to the optimality of @xmath349 .
therefore , a set @xmath298 as described above can not exist . with this result in hand , we are now prepared to resolve the case @xmath225 .
[ c=4 ] let @xmath9 be a positive integer , and let @xmath350 .
then there exists a partition of the subsets of @xmath60 $ ] of size at least @xmath9 into intervals @xmath63 $ ] such that @xmath351 . let @xmath352\mid a \subseteq [ n ] , |a| = d\}$ ] . by lemma [ right size ] ,
@xmath353 for all @xmath9-sets @xmath158 . by lemma [ d+1 ] ,
all @xmath222-sets are covered by the elements of @xmath278 .
since lemma [ right size ] implies that @xmath354 , we can apply lemma [ no collisions ] to the @xmath9-sets to show that @xmath291 for all @xmath355 .
therefore we have partitioned all of the @xmath9-sets and @xmath222-sets into the appropriate intervals .
what remains to show is that the @xmath293-sets and @xmath356-sets that are not covered by any element of @xmath278 can be partitioned according to the theorem . for @xmath357 let @xmath358\mid |s| = d+i+1 , ~s \notin \cup_{[a , b ] \in \mathcal{i}}[a , b ] \}$ ] .
let @xmath213 be the bipartite graph with vertex set @xmath359 and @xmath360 if and only if @xmath361 is a subset of @xmath362 .
by lemma [ uncovered top ] with @xmath363 we have that @xmath364 for each @xmath365 .
furthermore , @xmath366 for all @xmath367 .
hence , by corollary [ cor : spec - hall ] , there exists a complete matching from @xmath368 to @xmath369 .
therefore we can use the 1-cubes corresponding to the matching to cover the elements of @xmath368 , and these intervals are disjoint .
the remaining elements of @xmath369 can then be covered by trivial intervals .
furthermore , the sets of size at least @xmath294 can be covered by trivial intervals as well .
thus , we have found a partition of the desired type .
this completes the proof of theorem [ thm : main ] .
although this paper only resolves a fraction of the possible cases of conjecture [ conj - ind ] , we believe that it is true in full generality , in part because many of the lemmas in this paper are stated and proved in greater generality than is required for their applications here . however , it is not clear how to apply them to advance this line of research .
it seems that after applying lemma [ uncovered top ] it may be beneficial to pass from our combinatorial setting back to the algebraic setting to acquire some insight .
although it is not immediately obvious , our approach here generalizes the constructive proof for @xmath91 given by bir et al . in @xcite .
the generalization comes through the notion of balanced sets as originally defined there and an equivalent method of finding the intervals .
unfortunately , we do not see how to generalize the method of finding intervals that appears in @xcite to higher values of @xmath9 .
we also note that the inductive proof for @xmath91 given by bir et al . seems to face significant challenges in generalizing , and while we would not rule out the use of such a method to prove conjecture [ conj - ind ] , it seems unlikely to be successful .
the authors are grateful to jrgen herzog , david m. howard and william t. trotter for useful conversations while pursuing this research . | let @xmath0 be a field and @xmath1 $ ] . in 1982 , stanley defined what is now called the stanley depth of an @xmath2-module @xmath3 , denoted @xmath4 , and conjectured that @xmath5 for all finitely generated @xmath2-modules @xmath3 .
this conjecture remains open for most cases . however , herzog , vladoiu and zheng recently proposed a method of attack in the case when @xmath6 with @xmath7 being monomial @xmath2-ideals .
specifically , their method associates @xmath3 with a partially ordered set
. in this paper we take advantage of this association by using combinatorial tools to analyze squarefree veronese ideals in @xmath2 . in particular , if @xmath8 is the squarefree veronese ideal generated by all squarefree monomials of degree @xmath9 , we show that if @xmath10 , then @xmath11 , and if @xmath12 and @xmath13 , then @xmath14 .
= 1 |
nova outbursts are associated with close binary systems in which the secondary star fills its roche lobe and transfers gas onto the companion white dwarf via an accretion disk ( gehrz , 1988 ) .
this hydrogen rich gas accumulates onto the surface of the white dwarf until temperatures and densities are high enough to ignite thermonuclear reactions .
a thermonuclear runaway process takes place leading to the ejection of gas at hundreds or thousands of kilometers per second ( saizar et al .
this expanding gas has a chemical composition which is more metal rich than solar , and later on is photoionized by the hot underlying white dwarf .
the thermonuclear runaway model can account for some cno enhancement abundances observed in novae .
some other heavy element abundances , however , as the ones measured in nova cygni 1992 ( austin et al .
1996 ; hereafter a96 ) , can apparently only be explained by a mixing of the accreted envelope with white dwarf s core material ( starrfield et al .
1986 and truran and livio 1986 ) .
two types of nova are commonly recognized : co novae , in which the progenitor is a carbon - oxygen white dwarf , and onemg novae , involving an oxygen - neon- magnesium white dwarf .
novae cygni 1992 belongs to this second type since the most prominent emission lines are [ ne iii ] 3869/3967 , [ ne v ] 3346/3426 and [ o iii ] 4959/5007 ( matheson et al .
1993 ) . here
we use spectroscopic observations taken over a 4 year period , coupled with photoionization models , in order to study the time evolution of the nebular line spectra of nova cygni 1992 from day 85 to day 1581 after the outburst .
the models provide the white dwarf s temperature and the hydrogen density and chemical composition of the emitting shell under the simple approximations of uniform and spherically symmetric shell .
the spectral line structure and hst images show a more complex morphology ( paresce et al .
1995 ) , which suggests one must interpret these results with caution .
the data were obtained primarily with the dominion astrophysical observatory 1.8-m plaskett telescope , but as the nova faded the kitt peak national observatory 2.1-m and the multi - mirror telescope were employed .
the log of the observations is shown in table 1 .
the nebular observations began on may 15 , 1992 ( 85 days after the outburst ) and continued until june 1996 ( 1581 days after maximum ) .
the data are all long - slit spectra using a ccd as a detector .
the spectral images were reduced using standard iraf packages to remove amplifer bias and flat field variations as well as to extract and calibrate the spectra .
moderate resolution combined with good wavelength coverage was achieved by taking spectra of the nova and standard stars at a variety of grating tilts , taking care that consecutive spectra overlapped in wavelength .
the spectra were then reduced individually and combined by scaling to match the overlap sections . for the dao observations , the slit width was 1.5@xmath2 to 2@xmath2 and standard stars
were observed near the same airmass as the nova , which reduces the effects of slit losses caused by differential refraction .
the kpno and mmt spectra were taken at the parallactic angle .
the major source of error in the observed line ratios comes from combining spectra taken with different grating tilts as this error can accumulate over a large span of wavelengths .
we estimate that ratios of lines with small wavelength separations ( @xmath3100 nm ) are good to 10% while lines ratios spanning @xmath4300 nm are accurate to about 30% .
the nebular emission lines of v1974 cyg were broad ( @xmath5 km / s fwhm ) with a symetric shape that persisted for the first two years after discovery .
the evolution of the spectral line shape is best exemplified by [ o iii ] 4959/5007 emission .
figure 1 displays the [ o iii ] line in 1992 .
early on the lines were extremly broad , suggesting a fast wind dominated emission and smeared out the kinematic features .
through out 1992 , the [ o iii ] line narrows and the individual features become more clearly defined .
figure 2 shows that between 1992 and 1995 the [ o iii ] lines become even more narrow , and the symmetry dissolves into many knots as the emission becomes dominated by low density gas that is slow to recombine .
the striking symmetry in the early spectral features suggests a simple geometrical structure of the ejecta and implies that the line profile can be modeled in a straightforward way by following the prescription developed by hutchings ( 1972 ) .
the ejecta from novae in hutchings model consists of an expanding ring and polar caps , which inherently reflect the binary symmetry of the system .
our basic kinematical model consists of a shell of emission expanding at a velocity of 1100@xmath6 .
the emission is enhanced in the equatorial region and polar regions and the viewing angle is 35@xmath7 from the pole .
figure 3 shows a comparison between a simple kinematical model and the observations of the [ o iii ] 5007 line profile for three epochs in 1992 .
this single - velocity equatorial ring / polar blob model reproduces the basic spectral features , except for the lowest velocity emission peak and the broad wings of the earliest data .
figure 4 displays a projection onto the sky at the [ o iii ] 5007 emission compared with an archival 1994 hst foc observation of cygni 1992 at the same wavelength range .
the model image perhaps exaggerates the polar caps , but overall it resembles the observations .
the parameters of the model , the same for the spectrum and image , are summarized in table 2 .
the line profiles have also been modeled by austin ( 1995 ) and by hayward et al .
( 1996 ) ( hereafter h96 ) , based on a more complete physical model .
note that hydrodynamical simulations of nova shells by gill & obrien ( 1999 ) fail to reproduce the equatorial and polar features that appear to be common in nova shells .
the emission line fluxes were modeled using the photoionization code cloudy , version 84.12a ( ferland 1993 ) .
the code simultaneously solves the equations of thermal and statistical equilibrium for a spherical shell of a dilute gas which is being photoionized by an underlying thermal source .
the input parameters for the code are the temperature and luminosity of the source ( assumed to be a blackbody ) and those describing the characteristics of the shell , i.e. the hydrogen density , radius , thickness and chemical composition .
there is also the possibility to use other parameters to describe the covering and filling factors to account for the inhomogeneities of the ejecta , but since we were interested only in the fluxes relative to , these parameters were not changed in the models .
the code then integrates the emission over this shell giving the emergent emission - line spectrum .
the best fit parameters for cloudy are shown in table 3 ( temperature of the ionizing source and hydrogen density of the shell ) and table 4 ( chemical abundance of the shell ) .
the intrinsic errors in the temperature and density determinations are estimated to be 15 and 20% respectively , but systematic errors are likely to dominate , due for example to the fact that we are assuming spherical symmetry , eventhough observations show this is not the case .
figures 5 through 10 we present the comparison between the modeled synthetic spectra and the dereddened observations for days 85 , 140 , 172 , 202 , 250 , 401 , 506 , 565 , 639 , 1381 and 1581 after the outburst .
the observations were dereddened with e(b - v)=0.32 ( chochol , d. et al .
table 5 presents the dereddened observed fluxes of some selected lines compared to the model predictions . in order to directly compare the model and the data ,
the continuum was subtracted in all the observed spectra ( a technique particularly important for the 1381 and 1581 days ) , and an arbitrary constant was added to set a constant continuum levels .
we created synthetic spectra from the model output using a gaussian function with a fixed fwhm of 30 to approximate the observed line widths . in the figures comparing the models to the data ,
the emission calculated from the photoionization models are identified in the upper row . only those lines with integrated flux @xmath8 relative to
are shown . in those cases where the lines are too close together only the brightest line from the `` blend '' is labeled ( although all the lines are marked ) .
the identification of the rest of the lines in a blend can be found in the figure captions . from the figures it is clear that the brightest lines change from one day to another as the physical conditions of the nebula evolve .
the lower row labels mark the lines which are observed but not predicted by the models , this is either because they are not included in the code or because their integrated fluxes are lower than @xmath9 .
this is the case , for instance , of the [ o i ] 6300/63 lines , which are observed but for which the models predict even stronger lines of [ s iii ] 6312 and fe x 6374 .
the abundances of the elements were initially taken from a96 , and since no dust formation was reported in nova cygni 92 ( paresce et al . 1995 ; h96 ) , we do not expect a significant depletion of heavy elements ; we therefore assume that the mean chemical composition of the gas remains constant in time .
we found that the initial values given by a96 overestimated the strength of the c ii 4267 , n ii 5680 , [ s iii ] 6312 , ar iii 7135 , plus some of the fe and o lines .
the gas abundances ( relative to hydrogen , with h = 1 for the solar value ) were then modified to improve the comparison with the observations as follows : s = 122.6 @xmath10 s = 1 , n = 282 @xmath10 n = 50 , o = 110 @xmath10 o = 80 , ar = 58.2 @xmath10 ar = 5.0 , and fe = 16 @xmath10 fe = 8 . the changes in the sulphur , nitrogen and argon abundances are relatively large , but this is part due to the fact that except for h , he , o , ne , n and fe , the initial abundances from a96 were not properly constrained ( starrfield 1997 ) .
the sulfur abundance was reduced to 1 ( to be considered as an upper limit ) , in agreement with matheson et al .
1993 , since the [ s iii ] 6312 line should be negligible compared to [ o i ] 6300 .
the nitrogen abundance was reduced to 50 to fit the [ n ii ] 5680/5755 lines , which provides a good match for the early days , but for day 250 and following days the lines are slightly underestimated . the abundance value of n is high when compared to that found by matheson et al .
( 1993 ) , but matches the value derived by h96 based on the mid - infrared spectra of nova cygni 92 .
the oxygen abundance was reduced from 110 to 80 to get a better fit for [ o iii ] 4959/5007 .
we therefore increase the ne / o ratio from 2.3 ( a96 ) to 3.1 , in closer agreement with the value of ne / o=4 found by salama et al .
the n / o ratio gives a value of 0.6 , which is consistent with a white dwarf s mass between 1.25 and 1.35 ( starrfield et al .
1992 ) . from the ar iii
7135/7007 lines we set an upper limit to the ar abundance of 5 .
finally , the fe abundance was set to fit the fe iii 5270 and fe vi 5177 lines .
it is important to remember that the only abundances our models constrain are those for which we have emission line observations .
the abundances we modeled and used are , he=4.5 , n=50.0 , o=80.0 , ne=250.0 , s=1.0 , ar=5.0 , fe=8.0 ( relative to their solar values in which hydrogen is unity ) .
they are within the @xmath11 confidence limits of those listed by a96 , except for n which is difficult to constrain since we do not have the strong nitrogen uv lines to compare ( see shore et al . 1996 ) . the abundances in table 4 include the values for other elements ( following a96 ) , that may not correspond to the best values .
we do not expect , however , that changes in them will modify the spectra .
as politano et al . ( 1995 ) and a96 point out , the high abundances provide further support for the mixing scenario in which core material from the onemg white dwarf is dredged up during the accretion process .
the time evolution of the temperature of the ionizing source undergoes three different phases ( see table 3 ) : a rapid rise during days 140 through 250 ( at least ) , reaching a plateau ( in our data ) at around day 400 , and beginning to decline after day 506 .
these phases are necessary to explain the different observed ionization stages as a function of time and they are consistent with the behavior of the x - ray flux observed with rosat by krautter et al .
( 1996 ) . as it is explained by krautter et al .
( 1996 ) thermonuclear runaway models of the nova outburst ( starrfield , 1989 ) , predict that not all of the accreted material is ejected during the outburst .
the remaining material ( ranging from 10% to 90% ) would radiate at a constant luminosity .
as mass is lost from this system and the white dwarf shrinks back to its equilibrium radius , the effective temperature of the remnant increases .
krautter et al .
( 1996 ) related the decline in the x - ray flux with a decrease of the white dwarf s temperature , due to the turnoff of thermonuclear burning at its surface .
the temperature for the central source that we derived for the different models is consistent with the @xmath12 confidence temperature values obtained by a96 .
our central source temperatures are lower in the rise phase than those obtained by h96 based on mid - infrared data , but they are similar during the plateau and decline phases . in the models we used a simple approximation of a constant luminosity of l@xmath13 = @xmath14 erg s@xmath15 for the central source to minimize the number of initial parameters to study .
the constant bolometric luminosity phase depends on the mass of the white dwarf , and according to krautter et al .
( 1996 ) , lasts at least 511 days for nova cygni 1992 ( see also shore et al .
a value of @xmath14 erg s@xmath15 was also used by h96 up to day 616 , but it was reduced to @xmath16 erg s@xmath15 for day 849 .
a luminosity of @xmath14 erg s@xmath15 is then clearly too high for our models corresponding to the 1381 and 1581 days .
( see note on days 1381 and 1581 on next section for further discussion ) .
in all our models the physical radius of the shell ( see table 3 ) was derived from hst images ( paresce et al .
1995 ) taking the mean value between the major and minor axes of the observed elliptical ring and adopting a distance of 2.8 kpc ( a96 ) . for days 1381 and 1581 we assumed a mean expansion velocity of 620 km s@xmath15 .
this is smaller than the 1100 km s@xmath15 derived above from the line shape model , but it is not critical for the line ratios .
as it is expected from a constant expansion velocity shell , the density decreases approximately as @xmath17 .
our results agree well with those found by a96 , hayward et al .
1996 and paresce et al . 1995 .
the thickness ( @xmath18r ) of the shell became an important parameter for days 1381 and 1581 , when the shell changed from being radiation - bounded to matter - bounded .
a similar behavior was found by morisset et al .
( 1996 ) , where the shell became matter - bounded by day 800 .
we chose a constant thickness of @xmath19 cm , which is the average between the lower and upper limit given by a96 . for days 1381 and 1581
this value is also consistent with the criteria followed by h96 who adopted a thickness equal to a value of a @xmath20 the radius of the shell .
saizar et al .
( 1996 ) also suggest that since the [ o i ] 6300/63 lines trace the presence of neutral gas , the lines can be used to determine when the shell changes from ionization - bounded to matter - bounded .
the [ o i ] lines are present in the spectra up to day 639 , but they disappear by day 1381 , indicating when the shell was matter - bounded . the comparison between between observations and models for the first seven epochs ( i. e. 85 through 565 ) is quite reasonable , and this can be seen for instance in the [ o iii ] 5007 + 4959/4363 line ratio ( see figure 11 ) . the ratio can be used to determine the electron temperature ( @xmath21 ) in the ejecta , for a range in the electron density of @xmath22 , and for values of @xmath21 of @xmath23 k ( see e. g. osterbrock 1989 ) .
departures from these ranges indicates in general a more complex ionization structure . by day 565 , one begins to notice some difference between our simple models and the observations .
these differences are more noticeable for the last two epochs of our observations , 1381 and 1581 . for days 1381 and 1581
, we encountered several problems .
first , the lack of consecutive observations during the 1994@xmath2495 period makes it very difficult to follow the time evolution of the white dwarf temperature and the hydrogen density ( the main parameters we try to determine ) .
second , as it was mentioned before , the luminosity of white dwarf is expected to decrease considerably after day 600 .
h96 suggest it goes down to @xmath16 erg s@xmath15 by day 849 .
as we already mentioned our approximation of a constant luminosity of @xmath14 erg s@xmath15 is clearly too high for those days .
we ran several models with a lower value of @xmath16 erg s@xmath15 and the general tendency was that the [ ne v]3346 and [ ne v]3426 lines did not appear in the models , while they are clearly present in the observations .
third , the ionizing spectrum at late times may depart significantly from a blackbody .
fourth , our models for days 1381 and 1581 were very sensitive to the thickness of the shell chosen , since the ionization structure of h , he , c , n and o depends strongly on the radius .
our simple approach of a single uniform density shell for these latter days , however , is not appropriate .
h96 in their modeling of the mid - infrared spectra , have used `` multiple components '' to reproduce the observations , i.e. three zones with different physical characteristics ( dense , intermediate and diffuse ) .
the contrast among these three components becomes more drastic after approximately day 600 .
we have used a `` single '' component model up to day 639 , with relatively good results .
nevertheless for days 1381 and 1581 this approximation seems to break down .
for example , our synthetic spectra can not reproduce the [ oiii](5007 + 4959)/4363 ratio , which is overestimated by a large factor .
since high ratios are indicative of low density material , the fact that the ratio is greatly overestimated indicate that a single low - density component model is not valid at late times , when the emission is probably dominated by denser clumps .
the spectra of nova cygni 1992 display some very high ionization lines of [ fe vi ] , [ fe vii ] and [ fe x ] .
the highest ionization forbidden fine - structure emission lines or coronal lines are expected to arise in a high temperature low density gas ; whether this gas is heated by photoionization alone or by the interaction through shocks of the ejecta with the circumstellar material , is still not known .
a96 and h96 found in their studies that photoionization alone fully accounts for the coronal lines seen in the optical and in the mid - infrared .
a similar conclusion was reached by wagner and depoy ( 1996 ) who concludes that the observed ir spectrum of nova cygni 1992 and the inferred electron temperature are inconsistent with an origin in a high - velocity shock - heated gas . from hst observations
paresce et al .
( 1995 ) also found that shocks are not responsible for a substantial part of the emission . on the other hand , greenhouse et al .
( 1990 ) , determined from studies on quvul that photoionization is insufficient to account for the spatial extent of the coronal - emitting region .
figure 12 shows the [ fe vi ] 5146/77 , [ fe vii ] 6087 + ca v 6087 and [ fe x ] 6374 coronal emission lines as a function of time and they are compared with the behavior of the `` super - soft''x - rays [ 0.1 @xmath24 2.4 kev ] observed by rosat ( krautter et al .
1996 ) and the values predicted by the models . on the one hand the similar behavior of the three quantities strongly suggests that an important fraction of the iron coronal lines could be due to photoionization . on the other hand all the lines are found to be systematically underestimated which may indicate some contribution from shocks .
it is also possible that the strength of the fe lines reflects our chosen fe abundance , which was reduced by a factor of two to match the spectra .
furthermore , the fe x 6374 line is blended with the [ o i ] 6363 line , so to obtain its flux we subtracted one third of the [ o i ] 6300 line from the blend ( assuming too that the contribution from the [ s iii ] 6312 line was negligible ) .
it could be also the case that the coronal lines are produced as a `` skin effect '' on the lower - density or rarified surfaces of denser clumps , which could explain the difference between the observed and modeled fe lines .
multiple component models would be required to explore this in more detail .
the spectroscopic time evolution of nova cygni 1992 has been studied using 4 years of observations coupled with the synthetic spectral models , in order to learn more about the physical conditions of its emitting gaseous shell .
the line fluxes for the models were obtained using a photoionization code ( cloudy ) , where the chemical gas abundances were kept _ fixed _ as a function of time .
the models are consistent with the results from similar studies ( austin et al . 1996 ) , although they require less drastic enhancements in the gas abundances of the heavy metals , in particular for sulphur , argon ( which were not modeled by a96 ) and nitrogen .
the time evolution of the surface temperature of the ionizing source ( which it is assumed to be a black body ) follows three phases ( a rise , a plateau and a decline ) and it ranges from log(t ) = 4.9 - 5.63 , similar to what it is observed in the soft x - rays ( krautter et al .
1996 ) .
the line profiles can be reproduced using a simple kinematical model , consisting of an equatorial ring and polar caps with slightly different densities ( 6.7@xmath2510@xmath26 - 9.5@xmath2510@xmath26 @xmath27 respectively ) , expanding at an average velocity of 1100 km / s .
the fe coronal emission lines follow also the time behavior observed in the soft x - rays , which suggests a common energy source for their photoionization . that the fe line fluxes are slightly underestimated ,
may indicate nevertheless that the contributions from shocks may also be important .
amm thanks mmo and ipac for providing access to their facilities during the completion of this work .
we thank gary ferland for generously making available his photoionization code cloudy .
we thank a. alpert for helping on the data reduction in the initial phase of the project and e. friel for her useful comments .
austin , s.j .
, 1995 , phd thesis , arizona state university austin , s.j . , wagn , r.m . ,
starrfield , s. , shore , s.n . ,
sonnerborn , g. & bertram , r. , 1996 , aj , 111 , 869 chochol , d. , hric , l. , urban , z. , komzik , r.,grygar , j. and papousek , j. , 1993 , a&a , 277 , 103c ferland , g.j . , 1993 ,
department of physics and astronomy internal report ( university of kentucky ) greenhouse , m.a . ,
grasdalen , g.l . ,
woodward , c.e . ,
benson , j. , gehrz , r.d . ,
rosenthal , e. & skrutskie , m.f .
, 1990 , apj , 352 , 307 gehrz , r.d . , 1988 ,
araa , 26 , 377 gill , c.d & obrien , t.j . , 1999 , mnras , 307 , 677 hayward , t.l . ,
saizar , p. , gehrz , r.d . , benjamin , r.a . , mason , c.g . ,
houck , j.r .
, miles , j.w . , gull , g.e . & schoenwald , j. , 1996 , apj , 469 , 854 hutchings , j.b . , 1972 , mon . not .
soc . , 158 , 177 krautter , j. , ogelman , h. , starrfield , s. , wichmann , r. & pfeffermann , e. , 1996 , apj , 456 , 788 matheson , t. , filippenko , a.v .
& ho , l.c . , 1993 , apj , 418 , l29 morisset , c. & pequignot , d. , 1996 , a&a , 312 , 135 paresce , f. , livio , m. , hack , w. & korista , k. , 1995 , a&a , 299 , 823 politano , m. , starrfield , s. , truran , j.w . , weiss , a. & sparks , w.m .
, 1995 , apj , 448 , 807 saizar , p. , pachoulakis , i. , shore , s.n . ,
starrfield , s. , williams , r.e . ,
rothschild , e. & sonnerborn , g. , 1996 , mon . not .
, 279 , 280 salama , a. , evans , a. , eyres , s.p.s . , leech , k. , barr , p. & kesster , m.f . , 1996 ,
a&a , 315 , l209 shore , s.n . ,
starrfield , s. & sonnerborn , g. , 1996 , apj , 463 , l21 starrfield , s. , 1989 , in classical novae , edited by m.f .
bode & a. evans ( wiley , new york ) , p39 starrfield , s. , sparks , w.m . and truran , j.w . , 1986 , apj , 303 , l5 starrfield , s. , shore , s.n . , sparks , w.m . , sonnerborn , g. , truran , j.w . & politano , m. , 1992 , apj , 391 , l71 starrfield , s. et al . , 1992 , baas , 24 , 1190 starrfield , personal communication , 1997 truran , j.w . and livio , m. , 1986 , apj , 308 , 721 wagner , r.m . , & depoy , d.l . , 1996 ,
apj , 467 , 860 lcccc may 15 , 92 & 85 & dao 1.8 m & 3500 - 8700 & 3.1 + jul 09 , 92 & 140 & dao 1.8 m & 3700 - 7500 & 2.2 + jul 12 , 92 & 143 & dao 1.8 m & 6500 - 6700 & 1.2 + aug 10 , 92 & 172 & dao 1.8 m & 3900 - 7300 & 3.1 + sep 09 , 92 & 202 & dao 1.8 m & 3300 - 6700 & 3.5 + sep 17 , 92 & 210 & dao 1.8 m & 4400 - 8000 & 6.5 + oct 27 , 92 & 250 & dao 1.8 m & 3300 - 6900 & 2.8 + mar 27 , 93 & 401 & dao 1.8 m & 4300 - 6900 & 2.2 + jul 10 , 93 & 506 & dao 1.8 m & 3300 - 6400 & 6.1 + sep 07 , 93 & 565 & dao 1.8 m & 3300 - 7300 & 5.3 + nov 20 , 93 & 639 & kpno 2.1 m & 4400 - 8000 & 5.8 + dec 02 , 95 & 1381 & mmt & 3200 - 8900 & 5.8 + dec 03 , 95 & 1382 & mmt & 4000
- 5500 & 2.0 + jun 19 , 96 & 1581 & mmt & 3300 - 8900 & 5.8 + lrrrr may 15 , 92 & 85 & 4.95 & 7.90 & 14.65 + jul 09 , 92 & 140 & 5.02 & 7.80 & 14.87 + aug 10 , 92 & 172 & 5.10 & 7.75 & 14.95 + sep 09 , 92 & 202 & 5.36 & 7.64 & 15.02 + oct 27 , 92 & 250 & 5.52 & 7.46 & 15.13 + mar 27 , 93 & 401 & 5.62 & 7.20 & 15.34 + jul 10 , 93 & 506 & 5.63 & 6.80 & 15.43 + sep 07 , 93 & 565 & 5.54 & 6.60 & 15.48 + nov 20 , 93 & 639 & 5.48 & 6.40 & 15.53 + dec 02 , 95 & 1381 & 4.88 & 4.46 & 15.86 + jun 19 , 96 & 1581 & 4.90 & 4.40 & 15.92 + lllrrrrrr & 3345.9 & f@xmath30 & & & & 3.3 & 9.5 & & & f@xmath31 & 0.005 & 0.02 & 0.1 & 3.0 & 9.5 & 20.6 & 3425.8 & f@xmath30 & & & & 7.7 & 23.9 & & & f@xmath31 & 0.01 & 0.06 & 0.3 & 8.0 & 25.7 & 55.7 & 3868.7 & f@xmath30 & 3.4 & 4.4 & & 5.8 & 1 0.1 & & & f@xmath31 & 3.5 & 4.7 & 6.2 & 9.2 & 10.4 & 17.0 & 3967.5 & f@xmath30 & 1.2 & 1.5 & 1.9 & 1.9 & 3.1 & & & f@xmath31 & 1.1 & 1.4 & 1.9 & 2.8 & 3.2 & 5.2 h@xmath32 & 4861.3 & f@xmath30 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & & f@xmath31 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 4958.9 & f@xmath30 & 0.6 & 0.7 & 0.8 & 1.0 & 1.6 & 3.0 & & f@xmath31 & 0.6 & 0.7 & 0.9 & 1.3 & 1.6 & 2.8 & 5006.9 & f@xmath30 & 2.1 & 2.2 & 2.6 & 3.2 & 5.0 & 8.9 & & f@xmath31 & 1.7 & 2.2 & 2.8 & 4.1 & 4.7 & 8.3 & 5680 & f@xmath30 & 0.08 & 0.06 & 0.05 & 0.07 & 0.1 & 0.1 & & f@xmath31 & 0.06 & 0.05 & 0.05 & 0.04 & 0.03 & 0.03 & 5755 & f@xmath30 & 0.1 & & 0.1 & 0.1 & 0.2 & & & f@xmath31 & 0.1 & 0.1 & 0.08 & 0.05 & 0.06 & 0.09 h@xmath33 & 6562.8 & f@xmath30 & 4.5 & 4.2 & 3.8 & 3.3 & 2.7 & 2.6 & & f@xmath31 & 4.8 & 3.7 & 3.3 & 2.9 & 2.8 & 2.7 lllrrrrr & 3345.9 & f@xmath30 & 31.4 & 21.9 & & 0.5 & 0 .4 & & f@xmath31 & 32.3 & 24.3 & 18.2 & 1.3 & 0.9 & 3425.8 & f@xmath30 & 87.5 & 67.1 & & 1.7 & 1.4 & & f@xmath31 & 87.5 & 65.8 & 49.4 & 3.5 & 2.6 & 3868.7 & f@xmath30 & 12.0 & 9.9 & & 1.0 & 0.9 & & f@xmath31 & 17.4 & 15.5 & 14.2 & 3.6 & 1.8 & 3967.5 & f@xmath30 & 3.8 & 3.1 & & 0.5 & 0 .6 & & f@xmath31 & 5.3 & 4.8 & 4.4 & 1.1 & 0.5 h@xmath32 & 4861.3 & f@xmath30 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & & f@xmath31 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 4958.9 & f@xmath30 & 4.2 & 4.9 & 8.1 & 1.2 & 0.9 & & f@xmath31 & 3.7 & 4.1 & 4.8 & 1.3 & 0.7 & 5006.9 & f@xmath30 & 12.1 & 13.8 & 24.8 & 3.7 & 2.3 & & f@xmath31 & 11.2 & 12.4 & 14.4 & 4.0 & 2.0 & 5680 & f@xmath30 & 0.1 & 0.1 & 0.07 & & & & f@xmath31 & 0.03 & 0.04 & 0.04 & & & 5755 & f@xmath30 & 0.3 & 0.3 & 0.2 & & & & f@xmath31 & 0.10 & 0.09 & 0.08 & & h@xmath33 & 6562.8 & f@xmath30 & & 2.8 & 3.8 & 2.6 & 3.3 & & f@xmath31 & 2.7 & 2.7 & 2.7 & 2.6 & 2.6 | optical observations of nova v1974 cygni ( cygni 1992 ) , spanning a 4 year period , have been used to study its spectroscopic evolution .
the data cover a wavelength range from @xmath0 and follows the nebular evolutionary phase of the ejecta .
we have modeled the integrated fluxes by means of the photoionization code cloudy .
the models were run at a _ fixed _ abundance value for the most prominent elements ( i.e. h , he , c , o , n , ne , fe , etc ) over the entire time sequence .
it is possible to constrain from this simple model some of the physical conditions of the gaseous emitting region , like temperature and density .
compared with previous studies of the gas abundances of the heavy elements , we found that smaller enhancements of s , n and ar , and comparable values for o and fe , are able to reproduce the observations .
the time evolution of the surface temperature of the ionizing source and the high - ionization iron lines [ fe vii ] 6087 and [ fe x ] 6374 , is similar to what it is observed in the soft x - rays .
the early line profiles can be reproduced using a simple kinematical model consisting of an equatorial ring and polar caps , expanding at a velocity of @xmath1 km / s .
this simple model also approximates the structure of the resolved shell observed by hst . considering the complicated structure of the shell , the lack of well defined values of its gas density and our limited knowledge of the time evolution of the surface temperature of the photoionization source
, the comparison between models and observations agrees remarkably well .
21.75 cm -0.5 cm # 1#2#3#4 # 1 , _ # 2 _ , * # 3 * , # 4 |
newtonian mechanics is our first comprehensive model of a fundamental particle dynamics .
most of our theoretical conceptual notions in physics are tied to its formulation . as such , they are classical concepts .
the physical world is quantum .
its description should be based on quantum concepts each of which may be very different from its classical counterpart , if there is at all a counterpart .
most if not all of our fundamental conceptual notions in physics as an intuitive , common sense , origin .
however , such intuitive concepts do not fix their possible mathematical formulation which is an abstraction necessary for the analysis in a physical theory with precision .
the formulation of such intuitive concepts into the concepts of classical physics is what physicists are familiar with , but not at all _ a priori _ more intuitive than their quantum counterparts . the unfortunate common miss - statement of quantum physics being less than intuitive has however hindered our appreciation of quantum physics with the more appropriate quantum formulation of the fundamental conceptual notions .
our model of space and time is the most central problem at hand .
a common theme in the classical concepts is that the corresponding physical quantities , such as the position of a particle in space , have real number values .
quantum physics indicates otherwise , at least not necessarily by a finite number of real numbers .
more generally , one should consider such quantities as having values modeled by some appropriate types of abstract algebraic entities .
the system of real numbers as one of the latter has no preferred role apart from being our first order approximation to the physical nature as in the classical theory .
we can summarize it all by saying that the main culprit in the limitation of the classical concept is the real number , which the philosopher w. quine once called a ` convenient fiction ' ; with quantum physics , the old fiction is not so convenient any more .
let us focus on the notion of the physical space as a collection of all possible positions , either of a free particle or equivalently as the center of mass for a system of particles . the first mathematical model is given by the three dimensional euclidean geometry , which had been the only available model for centuries till beyond newton s time . and
it was of course adopted by newton in his mechanics .
again , any mathematical model is beyond naive intuition .
when , as it stand today , we have more than one such model , which one is the correct , or simply better , model for the physical space is a physics question .
the answer is to be seek from matching the corresponding theories to experimental result .
the limited success of newtonian mechanics says that the newtonian model may be not good enough .
we are now familiar with the einstein models , for his theories of special and general relativity .
however , the einstein models are still limited to real number geometry . like the newtonian model , they are classical ( physical ) models .
we want to look at quantum models . a simple natural guess based on modern mathematics would be a kind of topological space which likely has some noncommutative geometry @xcite
the idea of quantum geometry is not new in the subject matter related to deep microscopic structure of spacetime and quantum gravity .
however , we present here a model of the physical space as behind simple quantum mechanics and illustrate its having the correct classical limit @xcite . at this point , some readers may feel uncomfortable about our suggestion of going beyond the notion of a three dimensional manifold calling it nothing counter - intuitive .
the following comment is in order .
the notion of a three ` dimensional ' space , or four ` dimensional ' spacetime , is no doubt quite intuitive .
the chinese term for the universe as a totality of all spacetime , for example , has two characters the first of which is the space part which literally means extending along the three principle axes .
however , it is more tri - axial than three dimensional .
nothing in our intuition points towards identifying a definite position along any axis , or equivalently a distance with the abstract mathematical entity called a real number .
one can see that the notion of being sort of tri - axial stays in our model .
the ` particle ' is the key concept for an ideal physical object in the newtonian theory , one that still play a key role in simple quantum mechanics as a quantized version of the latter . to analyze the logic behind the newtonian formulation ,
the first question is ` why particle ? ' .
the particle is matter ( with a single characteristic called mass ) which has an unambiguous position in the euclidean space .
the newton s laws are really definitions of notions like states ( of motion ) , valid frames of reference , force , and mass ( as inertia ) . note that time
is also modeled on an ( one dimensional ) euclidean geometry .
all the notions put together gives a model of physical phenomena the success of which obviously breaks down at the atomic scale and beyond .
we argue that the modified theory of quantum mechanics has all the notions modified including those of position in space , though time is not touched so long as the ` non - relativistic ' theory is concerned .
we stick to the setting here . in ref.@xcite
, we presented the perspective of taking the relativity symmetry as the starting point to look at the newtonian model as well as formulating the quantum model and its classical limit .
it is a perspective that is behind a big program aiming at eventually constructing the quantum spacetime model and its dynamics at the deep microscopic level @xcite .
symmetry is the single most important theoretical structure in modern physics .
a relativity symmetry is the key symmetry behind a theory of mechanics .
it is the group of reference frame transformation , and the symmetry of the spacetime model in the theory .
the configuration space and phase space of a free particle are all natural homogeneous spaces as representations of the symmetry .
the algebra of observables also carries a representation of it . for the newtonian theory ,
the relativity symmetry is given by the galilei group .
newtonian space - time , as well as the single particle configuration and phase spaces can all be identified as coset spaces of the lie group .
the coset for the space - time is given by the quotient of the galilei group @xmath1 factored by the @xmath2 subgroup of rotations and boosts .
the corresponding case for the einstein theory is the minkowski spacetime @xmath3 , the coset space of the poincar group factored by the lorentz group .
in fact , the newtonian limit is exactly to be recovered by a contraction of @xmath4 to @xmath1 as @xmath5 @xcite . what we have ,
as shown below , is much of a parallel picture for the quantum versus classical case .
the action of @xmath1 on the coset is given by the familiar expressions ( c t + x^i + 1 ) =( ccc 1 & 0 & b + v^i & r^i_j & a^i + 0 & 0 & 1 ) ( c t + x^j + 1 ) = ( c t+b + v^i t + r^i_j x^j + a^i + 1 ) , and the more convenient infinitesimal form ( c dt + dx^i + 0 ) =( ccc 0 & 0 & b + v^i & ^i_j & a^i + 0 & 0 & 0 ) ( c t + x^j + 1 ) = ( c b + v^i t + ^i_j x^j + a^i + 0 ) .
kinematical structure alone is enough to look at the configuration and phase spaces on which generic dynamics can be described with a time translation generated by a hamiltonian with a nontrivial potential .
hence , we first restrict our analysis the subgroup , @xmath1 with the time translations @xmath6 taken out ; denoted by @xmath7 .
the newtonian space can be identified with @xmath8 which is the same as @xmath9 . for convenience in matching with the familiar quantum picture
, we use @xmath10 generators in place of the boost generators @xmath11 .
the newtonian space then has @xmath12 , and for the phase space @xmath13 , we have also @xmath14 .
the phase space is given by @xmath15 .
the position and momentum coordinates @xmath16 and @xmath17 both transform as a three vector under rotations and have their own independent translations generated by @xmath18 and @xmath19 , respectively .
the quantum symmetry differs from the classical one .
we have the heisenberg commutation relation which corresponds to having the lie algebra to the galilean symmetry modified by = i_ij i , where @xmath20 is an extra central generator that commutes with all others .
it is a bigger group @xmath21 which is a @xmath22 central extension of @xmath1 @xcite . without the time translations ,
it is a heisenberg - weyl group supplemented with the rotations denoted by @xmath23 . in the setting , one has naively the space coset @xmath24 with [ qsc ] ( c dx^i + d + 0 ) = ( ccc ^i_j & 0 & |x^i + 0 & 0 & 0 ) ( c x^j + + 1 ) =( c ^i_j x^j + |x^i + 0 ) , and the phase space coset @xmath25 with [ qpsc ] ( c dp^i + dx^i + d + 0 ) = ( cccc ^i_j & 0 & 0 & |p^i + 0 & ^i_j & 0 & |x^i + -|x_j & |p_j & 0 & | + 0 & 0 & 0 & 0 ) ( c p^j + x^j + + 1 ) =( c ^i_j p^j + |p^i + ^i_j x^j + |x^i + ( |p_j x^j -|x_j p^j ) + | + 0 ) . note that @xmath23 is the central extension of @xmath7 above .
the funny cosets with a @xmath26 coordinate apparently do not serve our purpose . however , we know how to construct the quantum phase space from the second coset . that is the formulation of the hilbert space of canonical coherent states @xcite .
the phase space coset is isomorphic to the heisenberg - weyl subgroup @xmath27 .
a unitary representation of the latter is given by taking the linear span of the canonical coherent states each of which can be identified as a point in the coset .
they are given by e^i|p^i , x^i= u(p^i , x^i , ) |0where u(p^i , x^i , ) e^i e^i e^-ix^i_i e^ip^i_i = e^i(p^i_i- x^i_i + ) , and @xmath28 is a fiducial normalized vector , @xmath29 and @xmath30 are representations of the generators @xmath19 and @xmath18 as hermitian operators , and @xmath31 is the identity operator representing the central generator @xmath20 . as a basis , the set of coherent states is overcomplete .
wavefunction @xmath32 for each of the state is a symmetry minimal uncertainty gaussian centered at the state label , which correspond to expectation values rather than eigenvalues of the @xmath30 and @xmath29 observables .
the operators generates translations of @xmath16 and @xmath33 exactly as described by the coset representation , with the nontrivial phase transformation .
the @xmath34 is the same as the ground state for a harmonic oscillator .
the coherent states are like the classical state described in quantum mechanics .
what is particular interesting to note is that one can have the parallel construction starting from the ( configuration ) space coset of eq.([qsc]).the relevant subgroup is the one generated by @xmath18 and @xmath20 .
we have a unitary representation as the span e^i|x^i= u(x^i , ) |0where u(x^i , ) e^i e^-ix^i_i , @xmath35 is the fiducial normalized vector , and @xmath30 and @xmath31 the hermitian operators generating translations in @xmath16 and the phase rotation . checking from the coset space action , one can see that the @xmath36 states are position eigenstates .
we have hence a hilbert space as the one usually presented in basic quantum mechanics textbook , which is of course unitary equivalent to the previous one from the coherent states .
it looks very much like it is the quantum space we have , and the quantum configuration space for a particle is the same as its phase space .
the real quantum phase space is not the hilbert space but its projective counterpart .
each one dimensional subspace as a ray corresponds to a distinct physical state .
the projective hilbert space is known to have the structure of an infinite dimensional symplectic manifold and kahlr manifold @xcite .
we can look it that through an expansion of each state in terms of the fock state basis as | = ( q_n + ip_n)| n for real homogeneous coordinates @xmath37 and @xmath38 of the projective space .
the schrdinger equation i | = | for the state is actually equivalent to the set of hamilton equations & & q_n = h(p_n , q_n ) , + & & p_n = - h(p_n , q_n ) , with the hamiltonian function @xmath39 .
the basic fact is unfortunately much less appreciated than it should be .
the projective hilbert space also has a metric .
it has no problem serving as a model of the physical space either .
the countable infinite set of complex coordinates @xmath40 are just @xmath41 .
we can also think of @xmath42 as a set of complex coordinates .
a functional analog of the hamilton equations can be obtained for the real and imaginary parts of the function .
the real and imaginary parts of each @xmath43 then serves as a set of real coordinates .
it is obviously essentially the same for the coherent state wavefunction @xmath44 .
there are uncountable infinite number of them , hence a set with much redundancy .
the picture says a quantum wavefunction is a full and definite description of the position ( and momentum ) of the particle .
the space of all such position a free particle can possibly have , _
i.e. _ the full ( projective hilbert space is a model of the quantum space . our familiar position operator @xmath45 describe a notion of position which is obviously the one in the classical model .
the quantum particle does not have a unique position in the classical model of space for the main reason that the latter fails as a good model of the space .
we can also see that it is the phase transformations generated by the central charge that makes it no longer possible to have a reduced notion of the ( configuration ) space from the phase space .
the lagrangian submanifold of a symplectic manifold of a classical system , as the configuration space , can be taken as the real part of the symplectic manifold with a specific complex structure .
a phase transformation that mixes the real and imaginary parts is a canonical transformation which mixes the configuration and momentum variable and not a symmetry transformation of the ( configuration ) space itself . with the quantum symmetry
, we have phase transformations within the relativity symmetry .
they are hence symmetry transformation of ( configuration ) space .
the feature is essentially dictated by the heisenberg commutation relation .
we have our quantum model of the physical space . a nontrivial check
that the idea is admissible is that it can explain the success of the corresponding classical picture in the proper limit .
the latter is given by a symmetry contraction .
more specifically , one should take the original representation which describes the quantum physics to the required limit rather than directly building the classical physics description from the contracted symmetry .
we will see that the contraction of the representation does indeed give a representation of the contracted symmetry @xcite .
classical mechanics should be seen as such an approximation to quantum mechanics @xcite .
we require the quantum space model to reduce to the classical one as sort of the @xmath46 limit through tracing the impact of the contraction on the hilbert space representation(s ) described .
we are talking about the infinite dimensional curved space as the projective hilbert space approximated by a finite dimensional euclidean space .
it works !
the symmetry contraction is given on the lie algebra as the @xmath47 with @xmath48 and @xmath49 .
naively , one can take @xmath50 as @xmath51 .
the nonzero commutator within @xmath27 , now in terms of @xmath52 , @xmath53 and @xmath20 , goes as = _ ij i 0 .
hence , symmetry ( sub)group , as well as the full @xmath21 reduces to a trivial central extension of the classical symmetry .
the @xmath20 generator completely decoupled from the rest and becomes quite irrelevant .
one can see that the transformation properties of the coset space representations in eqs.([qsc ] ) and ( [ qpsc ] ) reduce to the classical analog as @xmath54 and @xmath55 explicitly , we have at the @xmath47 limit & & d=| , + & & dx^i_c = ^i_j x^j_c + |x^i_c , + & & dp^i_c = ^i_j p^j_c + |p^i_c . for the coherent state picture
, we first have the @xmath56 states relabeled as @xmath57 where @xmath58 and @xmath59 ( @xmath60 , @xmath61 ) .
the new label correspond to the expectation values of @xmath62 and @xmath63 .
we have then & & ^c_i,^c_i | _ i^c |_i^c,_i^c= , + & & ^c_i,^c_i| _
i^c |_i^c,_i^c= , with & & ^c_i,^c_i |_i^c,_i^c = , [ e ] + & & ^c_i,^c_i |_i^c,_i^c = 1 .
the contraction result correspond to the @xmath47 limit , hence @xmath46 .
one can see that the last exponential factor hence the full result in eq.([e ] ) goes to 0 .
the basis states becomes mutually orthogonal .
the equation above then say that the @xmath62 and @xmath63 operators become diagonal on the basis set .
the conclusion there is that the original hilbert space representation becomes a reducible representation .
it reduces to a simple sun of the one dimensional spaces for the @xmath57 states . but
such rays of the original hilbert space each correspond to a single state .
they are the only states remain in the classical limit .
all the nontrivial linear superpositions are eliminated .
that is to say , the original projective hilbert space reduces to the classical six dimensional phase space . what about the physical space or the configuration space ?
we can actually do better than simply saying the latter is naturally a lagrangian subspace of the ( classical ) phase space .
we can start from looking at the @xmath36 basis states we have in constructing the quantum hilbert space as the model of the physical space and trace its contraction limit .
the basis states are to be relabeled as @xmath64 , _
i.e. _ by eigenvalues of the @xmath62 .
now , @xmath63 and hence all operators , say , as polynomials of @xmath62 and @xmath63 commute .
we have the same conclusion for the reduction of the hilbert space now seeing as going to the simple sum of the rays of @xmath64 .
the projective hilbert space is of course the three dimensional euclidean space .
in physics , we are supposed to learn from experiments what constitutes a good / correct theoretical / mathematical model of any physical concept , and physical space should not be an exception . as physics advances , we see our understanding of all the intuitive concepts keeps deepening .
the abstract mathematical contents to be given to such concepts changes as we replace a fundamental theory with a better one .
spacetime , or space and time , is the most basic concepts in our all discourse on physics .
einstein s relativity theories advanced our understanding of spacetime .
looking at it from the proper point of view , quantum mechanics does the same . in a way
, it would be quite surprising if it does not .
our talk here offers exactly such a perspective .
a notion about spacetime being somehow emergent rather than fundamental has been getting popular lately . in many cases ,
it is not clear what the author(s ) exactly meant about that .
the way we see it : the familiar classical picture and all the related features are almost definitely not at all fundamental .
they are like our classical approximate description of spacetime
. it should be quite clear nowadays that the deep microscopic picture of spacetime has to be very different .
we sure expect any good model of quantum spacetime model to give back the classical model as an approximation under proper limits . as such
, one can certainly call the classical model a emergent one . however , to have a fundamental theory of physics built completely without some notion of spacetime in it , we think , would be quite impossible .
there has to be some dynamical variables , whatever unfamiliar mathematical objects needed to described them , in the theory .
for the part of such variables necessary to give a picture of the spacetime as we have in the known classical theories at the proper limit , it is only fair to call the collection of all their admissible values in the most general physical setting the spacetime model of the theory .
we would also like to point out that quantum field theory understood correctly gives another very interesting perspective about spacetime , a feature we expect to stay as a part of any model of the deep microscopic ( quantum ) spacetime .
quantum field theory says that spacetime is the only physical entity .
the various quantum fields are not independent dynamical ` objects ' but only dynamic degrees of freedom the spacetime has .
any state , says the state which corresponds to a single muon , involves all the degrees of freedom . with enough precision
, we can see the role of any degree of freedom in any other basic properties of the state .
the magnetic dipole moment of the muon has been used to reveal information about possible superparticles postulated new quantum fields the discovery of which is one of the key target of the large hadron collider experiments .
we know already from quantum entanglement that the full system is more than the sum of its parts .
there are no truly independent subsystems .
the above leads naturally to the question of gravitation or quantum gravity .
we believe _ `
noncommutative geometry is to quantum gravity as noneuclidean geometry is to gravity'_. one should build quantum gravity as a _ geometrodynamics of quantum spacetime_. our quantum space model presented here is of course very far from the final quantum spacetime model .
we need to first go from the @xmath21 setting to one that incorporates poincar symmetry .
actually , we have a basic perspective which goes beyond that to a full stable symmetry that can not be the contraction limit of another symmetry @xcite . our basic stable symmetry have completely noncommutative @xmath65 and @xmath66 .
we need also to see how the above perspective of quantum field theory .
we need full dynamical pictures which includes analysis of the corresponding algebras of observables @xcite .
we have an exciting long way to go .
let us have a last word on basic quantum mechanics .
starting with our quantum space model , quantum mechanics can be seen from a very different light .
for instance , it begs the question of why not sees as physical observables all functions of the ( quantum ) position and momentum , as coordinates of the quantum space / phase space .
the usual notion of physical observables as essentially functions as operator @xmath29 and @xmath30 may be too limited .
they are only observables which have a clear classical analog . and as such , there is no reason to restrict possible knowledge about the operator / observables on a state to their expectation values either .
all the other moments of the distributions are conceptually as measurable as the expectation values . on a complementary point of view ,
the noncommutative algebra of quantum observables ( operators ) can be thought of as all continuous function @xmath67 of a topological space @xmath68 is the general setting of noncommutative geometry @xcite .
thinking about the quantum space as having noncommutative coordinates , essentially @xmath69 and @xmath70 ( kind of @xmath29 and @xmath30 ) @xcite is an alternative picture the infinite number of real number coordinates here .
we are familiar with thinking about physical quantities as real number valued .
that is no more than an assumption .
the notion of physical quantities as having values in some other algebraic entity may be the way to go .
we are familiar with the bohr argument about all measurement being classical .
that has to be the case only if we restrict results of measurements to be real numbers .
a generic measurement should only to a process to extract information from a physical system .
the system being quantum says that we may want to extract from it quantum information rather than the familiar classical one .
we are still learning to manipulate quantum information well .
but we can do that , we can deal with true quantum measurements .
dealing with physical quantities as beyond real number values and quantum space(time ) as beyond real number geometries is like a new copernican revolution for our generation to push on ! the author is partially supported by research grants most
105 - 2112-m-008 - 017- from the most of taiwan .
99 connes a 1994 _ noncommutative geometry _
( academic press ) + van suijlekom w d 2015 _ noncommutative geometry and particle physics _ ( springer ) the results discussed here are based on chew c s , kong o c w and payne j 2016 _ preprint ncu - hep - k066 , arxiv:1603.05371 [ quant - ph ] _ a conceptual discussion in a boarder context has been presented in the earlier article in chinese kong o c w 2015 _ general relativity : a centennial perspective _ ,
( taiwan : locus publishing ) p 185 kong o c w 2008 _ phys .
b _ * 665 * 58 + see also kong o c w 2008 _ preprint arxiv:0906.3581 [ gr - qc ] _ cho d n and kong o c w 2014 _ ann .
phys . _ * 351 * 275 gilmore r 2005 _ lie groups , lie algebras , and some of their applications _
( dover ) + tung w k 1985 _ group theory in physics _
( world scientific ) for a detailed pedagogical presentation , see kong o c w and payne j _ preprint ncu - hep - k067 _ de azcrraga j a and izquierdo j m 1995 _ lie groups , lie algebras , cohomology and some applications in physics _ ( cambridge university press ) .
peremolov a m 1977 _ sov .
usp . _ * 20 * 103 + zhang w - m , feng d h andgilmore r 1990 _ rev .
phys . _ * 62 * 867 + klauder j r and e.c .
sudarshan e c 1968 _ fundamentals of quantum optics _
( benjamin ) + klauder j r 2010 _ a modern approach to functional integration _ ( birkhuser ) bengtsson i and yczkowski k 2006 _ geometry of quantum states _ ( cambridge university press )
chew c s , kong o c w and payne j 2017 _ preprint ncu - hep - k070 _ , in preparation , gives a full presentation of the dynamical picture . | the physical world is quantum
. however , our description of the quantum physics still relies much on concepts in classical physics and in some cases with ` quantized ' interpretations .
the most important case example is that of spacetime .
we examine the picture of the physical space as described by simple , so - called non - relativisitic , quantum mechanics instead of assuming the newtonian model .
the key perspective is that of ( relativity ) symmetry representation , and the idea that the physical space is to be identified as the configuration space for a free particle .
parallel to the case of the phase space , we have a model of the quantum physical space which reduces to the newtonian classical model under the classical limit .
the latter is to be obtained as a contraction limit of the relativity symmetry .
ncu - hep - k069 + feb 2017 * model of the physical space from quantum mechanics * + + _ department of physics and center for mathematics and theoretical physics , + national central university , chung - li , taiwan 32054 + e - mail : [email protected]_ * * a**bstract : the physical world is quantum .
however , our description of the quantum physics still relies much on concepts in classical physics and in some cases with ` quantized ' interpretations .
the most important case example is that of spacetime .
we examine the picture of the physical space as described by simple , so - called non - relativisitic , quantum mechanics instead of assuming the newtonian model .
the key perspective is that of ( relativity ) symmetry representation , and the idea that the physical space is to be identified as the configuration space for a free particle .
parallel to the case of the phase space , we have a model of the quantum physical space which reduces to the newtonian classical model under the classical limit .
the latter is to be obtained as a contraction limit of the relativity symmetry .
+ @xmath0 talk presented at the eighth international workshop dice 2016 spacetime , matter , quantum mechanics ( sep 12 - 16 ) , castiglioncello , italy . |
a riemannian metric @xmath4 is called _ einstein _ if the ricci tensor @xmath5 satisfies the equation @xmath6 , for some @xmath7 .
when @xmath8 is compact , einstein metrics of volume 1 can be characterized variationally as the critical points of the scalar curvature functional @xmath9 on the space @xmath10 of riemannian metrics of volume 1 .
if @xmath11 is a compact homogeneous space , a @xmath12-invariant einstein metric is precisely a critical point of @xmath13 restricted to the set of @xmath12-invariant metrics of volume 1 . as a consequence ,
the einstein equation reduces to a system of non - linear algebraic equations , which is still very complicated but more manageable , and in some times can be solved explicity .
thus most known examples of einstein manifolds are homogeneous . in a recent work @xcite
the first two authors classified all flag manifolds for which the isotropy representation decomposes into four pairwise inequivalent irreducible submodules , and found new invariant einstein metrics on these spaces .
recall that a generalized flag manifold is an adjoint orbit of a compact semisimple lie group @xmath12 , or equivalently a compact homogeneous space of the form @xmath14 , where @xmath15 is the centralizer of a torus @xmath16 in @xmath12 .
eventhough the problem of finding all invariant einstein metrics on @xmath8 can be facilitated by use of certain theoretical results ( e.g. the work @xcite on the total number of @xmath12-invariant complex einstein metrics ) , it still remains a difficult one , especially when the number of isotropy summands increases .
this difficulty also increases when we pass from exceptional flag manifolds to classical flag manifolds , because in the later case the einstein equation reduces to a parametric system .
in particular , eventhough all invariant einstein metrics were found for every generalized flag manifold with four isotropy summands , a partial answer was given for the spaces @xmath17 and @xmath18 .
we summarize the results obtained in @xcite about these spaces .
[ the1](@xcite ) the flag manifold @xmath17 ( @xmath2 and @xmath3 ) admits at least six @xmath0-invariant einstein metrics .
there are two non - khler einstein metrics and two pairs of isometric khler - einstein metrics .
[ the2](@xcite ) the flag manifold @xmath18 ( @xmath19 and @xmath20 ) admits at least four @xmath21-invariant einstein metrics , which are khler . for the special case
@xmath22 the following results have been obtained : [ the3](@xcite ) the flag manifold @xmath23 ( @xmath24 ) admits at least six @xmath25-invariant einstein metrics .
there are two non - isometric non - khler einstein metrics , and four isometric khler - einstein metrics . in the special case where @xmath26 there are two more non - khler einstein metrics , and
the total number of @xmath25-invariant einstein metrics is exactly eight .
[ the4](@xcite ) the flag manifold @xmath27 ( @xmath28 ) admits precisely six @xmath21-invariant einstein metrics
. there are four isometric khler - einstein metrics , and two non - khler einstein metrics . in the present paper
we find all @xmath0-invariant einstein metrics on the flag manifold @xmath17 , by using a new approach into manipulating the algebraic systems of equations obtained from the einstein equation .
the coefficients of the polynomials in such systems involve parameters , so a major difficulty appears when we try to show existence and uniqueness of solutions .
therefore , the contribution of the present work is , besides answering the original problem on einstein metrics , to give some insight towards the study of parametric systems of algebraic equations .
our main result is the following : the flag manifold @xmath18 will be treated in a forthcoming paper .
let @xmath29 be a generalized flag manifold of a compact simple lie group @xmath12 , where @xmath30 is the centralizer of a torus @xmath16 in @xmath12 .
let @xmath31 be the identity coset of @xmath32 .
we denote by @xmath33 and @xmath34 the corresponding lie algrebras of @xmath12 and @xmath35 .
let @xmath36 denote the killing form of @xmath33 . since @xmath12 is compact and simple
, @xmath37 is a positive definite inner product on @xmath33 .
with repsect to @xmath37 we consider the orthogonal decomposition @xmath38 .
this is a reductive decomposition of @xmath33 , that is @xmath39 , and as usual we identify the tangent space @xmath40 with @xmath41 . since @xmath30 ,
the isotropy group @xmath35 is connected and the relation @xmath39 is equivalent with @xmath42\subset{\mathfrak}{m}$ ] .
thus , for a flag manifold @xmath11 the notion of @xmath43-invariant and @xmath44-invariant is equivalent .
let @xmath45 be the isotropy representation of @xmath35 on @xmath40 .
since @xmath46 is equivalent to the adjoint representation of @xmath35 restricted on @xmath41 , the set of all @xmath12-invariant symmetric covariant 2-tensors on @xmath32 can be identified with the set of all @xmath43-invariant symmetric bilinear forms on @xmath41 . in particular ,
the set of @xmath12-invariant metrics on @xmath32 is identified with the set of @xmath43-invariant inner products on @xmath41 .
let @xmath47 be a @xmath48-orthogonal @xmath43-invariant decomposition of @xmath41 into pairwise inequivalent irreducible @xmath43-modules @xmath49 @xmath50 .
such a decomposition always exists and can be expressed in terms of @xmath51-roots ( cf .
@xcite , @xcite ) .
then , a @xmath12-invariant riemannian metric on @xmath8 ( or equivalently , an @xmath43-invariant inner product @xmath52 on @xmath53 ) is given by @xmath54 where @xmath55 . since @xmath56 as @xmath43-representation , any @xmath12-invariant metric on @xmath8
has the above form .
similarly , the ricci tensor @xmath5 of a @xmath12-invariant metric @xmath4 on @xmath8 , as a symmetric covariant 2-tensor on @xmath32 is given by @xmath57 where @xmath58 are the components of the ricci tensor on each @xmath49 , that is @xmath59 .
these components obtain o useful description in terms of the structure constants @xmath60 $ ] first introduced in @xcite .
let @xmath61 be a @xmath48-orthonormal basis adapted to the decomposition of @xmath41 , that is @xmath62 for some @xmath63 , and @xmath64 if @xmath65 ( with @xmath62 and @xmath66 ) .
set @xmath67 , x_{\gamma})$ ] so that @xmath68_{{\mathfrak}{m}}=\sum_{\gamma}a_{{\alpha}{\beta}}^{\gamma}x_{\gamma}$ ] , and @xmath60=\sum(a_{{\alpha}{\beta}}^{\gamma})^{2}$ ] , where the sum is taken over all indices @xmath69 with @xmath70 ( where @xmath71_{{\mathfrak}{m}}$ ] denotes the @xmath41-component ) .
then @xmath60 $ ] is nonnegative , symmetric in all three entries , and independent of the @xmath48-orthonormal bases choosen for @xmath72 and @xmath73 ( but it depends on the choise of the decomposition of @xmath41 ) .
[ ricc](@xcite ) let @xmath11 be a generalized flag manifold of a compact simple lie group @xmath12 and let @xmath74 be a decomposition of @xmath41 into pairwise inequivalent irreducible @xmath43-submodules .
then the components @xmath58 of the ricci tensor of a @xmath12-invariant metric ( [ inva ] ) on @xmath8 are given by @xmath75-\frac{1}{2d_{k}}\sum_{i , j}\frac{x_{j}}{x_{k}x_{i}}[kij ] , \qquad ( k=1 , \ldots , s).\ ] ] in wiew of proposition [ ricc ] , a @xmath12-invariant metric @xmath76 on @xmath8 , is an einstein metric with einstein constant @xmath77 , if and only if it is a positive real solution of the system @xmath78-\frac{1}{2d_{k}}\sum_{i , j}\frac{x_{j}}{x_{k}x_{i}}[kij]=e , \quad 1\leq k\leq s.\ ] ]
we review some results related to the generalized flag manifold @xmath79 ( @xmath80 ) obtained in @xcite .
its corresponding painted dynkin diagram is given by the isotropy representation of @xmath8 decomposes into a direct sum @xmath81 , which gives rise to a decomposition @xmath82 of @xmath53 into four irreducible inequivalent @xmath44-submodules .
the dimensions @xmath83 of these submodules can be obtained by use of weyl s formula ( * ? ? ?
* , p. 210 ) and are given by @xmath84 according to ( [ inva ] ) , a @xmath12-invariant metric on @xmath11 is given by @xmath85 for positive real numbers @xmath86 .
we will denote such metrics by @xmath87 .
it is known ( @xcite ) that if @xmath88 then @xmath8 admits two non - equivalent @xmath12-invariant complex structures @xmath89 , and thus two non - isometric khler - einstein metrics which are given ( up to scale ) by ( see also ( * ? ? ?
* theorem 3 ) ) @xmath90 if @xmath22 then @xmath8 admits a unique @xmath12-invariant complex structure with corresponding khler - einstein metric ( up to scale ) given by @xmath91 ( cf . also ( * ? ? ?
* theorem 8) where all isometric khler - einstein metrics are listed ) .
the ricci tensor of @xmath8 has been computed in @xcite and is given as follows : [ componentsii ] the components @xmath92 of the ricci tensor for a @xmath12-invariant riemannian metric on @xmath8 determined by ( [ metri ] ) are given as follows : @xmath93 where @xmath94 $ ] and @xmath95 $ ] . by taking into account the explicit form of the khler - einstein metrics above , and substituting these in ( [ compiii ] ) , we can find that the values of the unknown triples @xmath60 $ ] are given by @xmath96 and @xmath97 .
a @xmath12-invariant metric @xmath87 on @xmath11 is einstein if and only if , there is a positive constant @xmath77 such that @xmath98 , or equivalently @xmath99 by substituting the values of @xmath100 and @xmath101 into the components of the ricci tensor , system ( [ systemi ] ) is equivalent to the following equations : @xmath102
we consider the equation @xmath103 of system ( [ systemi ] ) .
this is equivalent to @xmath104 let @xmath105 .
then the system of equations @xmath106 becomes @xmath107 from ( [ 1 ] ) we get that @xmath108 note that @xmath109 if and only if @xmath110 . by substituting equation ( [ 3 ] ) into equation ( [ 2 ] ) , we obtain the following equation : @xmath111 from ( [ 2 ] ) we get that @xmath112 note that @xmath113 if and only if @xmath114 . by substituting equation ( [ 5 ] ) into equation ( [ 1 ] ) , we obtain the following equation : @xmath115 note that the relation between @xmath116 and @xmath117 is given by @xmath118 [ existence ] the equation @xmath119 has at least two solutions between @xmath120 and @xmath121 .
we consider the value @xmath116 at @xmath120 and @xmath121 .
we see that @xmath122 now , the value @xmath116 at @xmath123 is given by @xmath124 thus the equation @xmath119 has at least two solutions between @xmath120 and @xmath121 .
we need to show that the polynomial @xmath125 has only one local minimum ( i.e. the two solutions obtained in proposition [ existence ] are unique ) , with some exceptions which will also be studied .
[ 2p+5 ] for @xmath126 and @xmath127 the equation @xmath128 has exactly two positive solutions .
we have that @xmath129 @xmath130 and @xmath131 note that the quadratic polynomial @xmath132 attains its minimum at @xmath133 and we see that @xmath134 we set @xmath135 and we investigate the conditions for @xmath136 such that @xmath137 for @xmath138 .
we consider the coefficients of @xmath139 as a polynomial of @xmath140 .
we can write @xmath141 as @xmath142 we put @xmath143 note that @xmath144 thus we see that @xmath145 , @xmath146 for @xmath147
. therefore we see that @xmath148 for @xmath149 and @xmath147 and hence , @xmath150 is monotone increasing and the polynomial @xmath125 has only one local minimum for @xmath151 and @xmath152 .
thus the equation @xmath119 has exactly two positive solutions .
now we examine the values @xmath153 and @xmath154 .
[ p=2_3 ] ( 1 ) let @xmath153 .
then for @xmath155 the equation @xmath156 has exactly two positive solutions , and for @xmath157 it has exactly four positive solutions .
\(2 ) let @xmath154 .
then for @xmath158 the equation @xmath159 has exactly two positive solutions , and for @xmath160 it has exactly four positive solutions .
( 1 ) for @xmath161 we have that @xmath162 thus we see that @xmath163 for @xmath164 , and hence , @xmath165 is monotone increasing and the polynomial @xmath166 has only one local minimum for @xmath164 .
thus the equation @xmath156 has exactly two positive solutions for @xmath164 .
for @xmath167 , we consider polynomials @xmath166 one by one and we see that , for @xmath168 the equation @xmath156 has two positive solutions , and for @xmath169 the equation @xmath156 has four positive solutions .
@xmath170 @xmath171 @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 @xmath177 ( 2 ) for @xmath178 we have that @xmath179 thus we see that @xmath180 for @xmath164 , and hence , @xmath181 is monotone increasing and the polynomial @xmath182 has only one local minimum for @xmath164 .
thus the equation @xmath159 has exactly two positive solutions for @xmath164 . for @xmath183 , we consider polynomials @xmath182 one by one and we see that , for @xmath184 the equation @xmath159 has two positive solutions , and for @xmath185 the equation @xmath159 has four positive solutions .
@xmath186 @xmath187 @xmath188 @xmath189 @xmath190 @xmath191 @xmath192 next , we consider the case when @xmath193 .
we may assume that @xmath127 .
[ n=2p ] let @xmath22
. then the equation @xmath194 has exactly two positive solutions for @xmath195 and four positive solutions for @xmath196 .
we see that @xmath197 thus , the four solutions of the equation @xmath194 are given by @xmath198 since @xmath199 is negative for @xmath195 , we see that the equation @xmath194 has exactly two positive solutions @xmath195 and four positive solutions for @xmath196 .
@xmath200 @xmath201 @xmath202 [ n=2p+1 ] let @xmath203 .
then for @xmath127 the equation @xmath204 has exactly two positive solutions .
we see that @xmath205 and @xmath206 thus we see that @xmath207 for @xmath208 and hence , @xmath209 is monotone increasing and the polynomial @xmath210 has only one local minimum for @xmath208 .
thus the equation @xmath211 has exactly two positive solutions .
for @xmath212 , we see that @xmath213 is negative and two real solutions @xmath214 of the quadratic equation @xmath215 are given by @xmath216 since the polynomial @xmath217 of degree 3 takes a local minimum at @xmath218 , we consider the value @xmath219 .
we see that @xmath220 by evaluating the above expression for the integers @xmath212 , we see that @xmath221 for @xmath222 and @xmath223 for @xmath224 .
thus the polynomial @xmath210 has only one local minimum for @xmath225 , and @xmath226 has two local minima and one local maximum for @xmath227 .
however , we see that for @xmath228 the equation @xmath204 has exactly two roots between @xmath229 and @xmath230 , and this completes the proof
. @xmath231 @xmath232 [ n=2p+2 ] let @xmath233 .
then for @xmath127 the equation @xmath234 has exactly two positive solutions .
we see that @xmath235 and @xmath236 thus we see that @xmath237 for @xmath238 and hence , @xmath239 is monotone increasing and the polynomial @xmath240 has only one local minimum for @xmath238 .
thus the equation @xmath234 has exactly two positive solutions . for @xmath241 , we see that @xmath242 is negative and the two real solutions @xmath214 of the quadratic equation @xmath243 are given by @xmath244 since the polynomial @xmath245 of degree 3 takes local minimum at @xmath218 , we consider the value @xmath246 .
we see that @xmath247 by substituting integer @xmath241 , we see that @xmath248 for @xmath249 and @xmath250 for @xmath251 .
thus the polynomial @xmath240 has only one local minimum for @xmath252 , and @xmath253 has two local minima and one local maximum for @xmath254 .
however , we see that for @xmath254 the equation @xmath234 has exactly two roots between @xmath255 and @xmath230 .
@xmath256 by a similar method we can prove the next two lemmas . [ n=2p+3 ] let @xmath257 .
then for @xmath127 the equation @xmath258 has exactly two positive solutions .
[ n=2p+4 ] let @xmath259 .
then for @xmath127 the equation @xmath260 has exactly two positive solutions .
therefore we have obtained the following : [ solutions ] ( 1 ) if @xmath261 and @xmath262 , then @xmath8 admits exactly four @xmath263-invariant einstein metrics for the pairs @xmath264 , @xmath265 , @xmath266 , @xmath267 , @xmath268 , @xmath269 , @xmath270 , @xmath271 and two @xmath263-invariant einstein metrics for all other cases .
( 2 ) if @xmath261 and @xmath272 , then @xmath8 admits exactly four @xmath263-invariant einstein metrics for the pairs @xmath264 , @xmath265 , @xmath266 , @xmath273 , @xmath274 , @xmath268 , @xmath275 , @xmath271 and two @xmath263-invariant einstein metrics for all other cases .
part ( 1 ) is a consequence of proposition [ existence ] and lemmas [ 2p+5 ] [ n=2p+4 ] . for ( 2 ) , we consider the equation @xmath276 , and the result follows from the relation ( [ 7 ] ) .
let @xmath277 and set @xmath278 . from equation ( [ 31 ] ) we obtain that @xmath279 we need to show the following : [ kahler ] the system of equations @xmath280 and @xmath281 has no positive solutions , except khler - einstein metrics .
we substitute equation ( [ 41 ] ) and @xmath278 into the equations @xmath280 and @xmath282 , and we obtain the following equations : we consider the resultant of the polynomials @xmath283 and @xmath284 with respect to @xmath285 , which is a polynomial of @xmath286 , say @xmath287
. we factor @xmath287 as @xmath288 we first consider the cases when @xmath289 and we claim that we only get khler - einstein metrics on @xmath290 .
\1 ) let @xmath291
. then equations ( [ 51 ] ) and ( [ 61 ] ) reduce to @xmath292 if @xmath293 , we have @xmath294 by taking the resultant of these polynomials with respect to @xmath285 , we get @xmath295 and we see that the resultant is non - zero for @xmath296 . thus we get only @xmath297 for a solution of equations ( [ 51 ] ) and ( [ 61 ] ) . from ( [ 41 ] ) , we see @xmath298 .
thus we obtain a khler - einstein metric in this case .
notice that this metric corresponds ( up to scale ) to the khler - einstein metric @xmath299 of ( [ ke ] ) \2 ) let @xmath300
. then equations ( [ 51 ] ) and ( [ 61 ] ) reduce to @xmath301 if @xmath302 , we have @xmath303 by taking the resultant of these polynomials with respect to @xmath285 , we get @xmath304 now we have @xmath305 which is positive for @xmath296 .
thus we see that the resultant is non - zero and we only get @xmath306 for a solution of equations ( [ 51 ] ) and ( [ 61 ] ) . from ( [ 41 ] ) , we see @xmath307 .
thus we obtain a khler - einstein metric in this case . notice that this metric corresponds ( up to scale ) to the khler - einstein metric @xmath308 of ( [ ke ] ) \3 )
let @xmath309 .
by a similar method we obtain that for @xmath296 , @xmath310 is the only solution of equations ( [ 51 ] ) and ( [ 61 ] ) , and from ( [ 41 ] ) we see that @xmath311 .
thus we obtain a khler - einstein metric in this case .
\4 ) let @xmath312 . by a similar method we obtain that for @xmath296 , @xmath313 is the only positive solution of the equations ( [ 51 ] ) and ( [ 61 ] ) for @xmath314 , and from ( [ 41 ] ) we see that @xmath315 .
therefore , we obtain a khler - einstein metric in all four cases .
we now denote by @xmath316 the factor of degree @xmath317 in the factorization of @xmath287
. then we can write @xmath318 the case @xmath319 has been studied in @xcite .
we now proceed in two steps .
note that @xmath320 for @xmath321 .
we have that @xmath322 note that the coefficient of @xmath323 is @xmath324 .
the polynomial @xmath316 of degree @xmath317 attains a local minimum at @xmath325 , a local maximum at @xmath326 , and a local minimum at @xmath327 . by evaluating @xmath328 at the point @xmath329
, we have that @xmath330 since we can write @xmath331 we see that @xmath332 , thus @xmath333 also , by evaluating @xmath328 at the point @xmath334 , we have that @xmath335 since we can write @xmath336 we see that @xmath337 , thus @xmath338 therefore , the three real solutions @xmath339 , @xmath340 , @xmath341 of the polynomial @xmath342 of degree 3 satisfy @xmath343 @xmath344 @xmath328 @xmath328 @xmath344 since @xmath345 , in order to show that @xmath346 for @xmath347 , we need to prove the following : * the local minimum is positive . *
we show our claim by dividing into two cases , namely @xmath348 and @xmath349 . * case 1 .
* the polynomial @xmath316 is given by @xmath350 then the local minimum of @xmath316 at @xmath351 satisfies @xmath352 .
indeed , it is @xmath353 then @xmath354 and @xmath355 also , we have that @xmath356 note that @xmath357 and @xmath358 hence , the function @xmath316 is concave up for @xmath359 , so the local minimum @xmath327 satisfies @xmath360 .
we consider the tangent lines of the curve @xmath316 at @xmath361 and @xmath362 , given by the equations @xmath363 and @xmath364 respectively .
these are shown in the figure .
@xmath365 @xmath362
let @xmath366 be their point of intersection given by @xmath367 note that @xmath368 for @xmath369 .
therefore , the local minimal @xmath370 is greater than @xmath371 , and the claim has been proved .
* case 2 . .
* note that @xmath372 and @xmath373 now we have @xmath374 and @xmath375 note that @xmath376 thus we see that @xmath377 .
let @xmath378 be the tangent line of the curve @xmath316 at @xmath379 .
this is given by @xmath380 we consider the point @xmath381 such that @xmath382 .
then we see that @xmath383 we will show that @xmath384 for @xmath385 . indeed , we have @xmath386 where @xmath387 we shall show that @xmath388 for @xmath385
. we can write @xmath389 as a polynomial of @xmath390 of the form @xmath391 where @xmath392 can be written as follows : @xmath393 we see that the coefficients @xmath394 ( @xmath395 ) are positive for @xmath349 , which means that @xmath388 for @xmath396 . therefore , @xmath384 for @xmath385 .
now we compute @xmath397 .
we see that @xmath398 note that @xmath399 and @xmath400 therefore , the function @xmath316 is concave up for @xmath401 .
@xmath402 @xmath403 consider the tangent line @xmath404 of the curve @xmath316 at @xmath402 , which intersects @xmath405-axis at a point @xmath406 , and the tangent line @xmath407 of the curve @xmath316 at @xmath403 .
since @xmath408 and @xmath409 , the tangent lines @xmath410 intersect at a point @xmath366 with @xmath411 . since @xmath316 is concave up , we see that the curve @xmath412 @xmath413 lies inside the triangle given by the three points @xmath414 , @xmath366 and @xmath415 .
since the point @xmath416 is inside of this triangle , it follows that the local minimum @xmath417 is greater than @xmath418 , and the claim has also been shown in this case .
this reduces to case step 1 as follows .
we consider the resultant of @xmath283 and @xmath284 with respect to @xmath286 , which is a polynomial of @xmath285 ( instead of @xmath286 ) , and we denote this resultant by @xmath419 . by factorizing @xmath419
we have that @xmath420 we denote by @xmath421 the factor of degree @xmath317 in the above factorization
. then we can write @xmath422 if we replace @xmath423 with @xmath424 in the polynomial @xmath421 , we get exactly the same polynomial as @xmath425 , and thus we see that the equation @xmath426 has no positive solutions for @xmath427 .
the main theorem now follows from propositions [ solutions ] and [ kahler ] .
in this section we study the isometry problem for the new homogeneous einstein metrics of @xmath1 , corresponding to the pairs @xmath428 which are presented in the main theorem . recall that when @xmath22 , it was proved in @xcite that the non - khler homogeneous einstein metrics of the form @xmath429 , where @xmath285 is given by part @xmath430 of ( [ 400 ] ) , are not isometric .
however for the special case of @xmath431 , the isometry problem for the remaining two new einstein metrics @xmath432 , where @xmath285 and @xmath286 are determined by part @xmath433 of ( [ 400 ] ) , and @xmath434 respectively , has not been studied yet .
, were obtained in ( * ? ? ?
* theorem .
8) with respect to the normalization @xmath435 . for the special case
@xmath431 the new einstein metrics are given with respect to the normalization @xmath436 . ]
let us recall the method used in @xcite .
for any @xmath12-invariant einstein metric @xmath87 on @xmath1 ( with @xmath3 ) we determine a ( normalized ) scale invariant given by @xmath437 , where @xmath438 is the scalar curvature of the given metric @xmath4 , @xmath439 is the volume of @xmath4 and @xmath440 . in particular ,
the scalar curvature of @xmath4 is given by @xmath441}{2}(\frac{x_1}{x_2x_3 } + \frac{x_2}{x_1x_3 } + \frac{x_3}{x_1x_2 } ) - \frac{[134]}{2}(\frac{x_1}{x_3x_4 } + \frac{x_3}{x_1x_4}+\frac{x_4}{x_1x_3})\ ] ] where @xmath442 and @xmath443 $ ] , @xmath444 $ ] are given in section 2 .
note that @xmath445 is a homogeneous polynomial of degree @xmath446 on the variables @xmath447 , and the volume @xmath448 is a monomial of degree @xmath449 .
thus @xmath437 is a homogeneous polynomial of degree 0 , and it is invariant under a common scaling of the variables @xmath450 .
if two metrics are isometric then they have the same scale invariant , so if the scale invariants @xmath451 and @xmath452 are different , then the metrics @xmath4 and @xmath453 can not be isometric .
if @xmath454 we can not draw an immediate decision and conclude if the metrics @xmath4 and @xmath453 are isometric or not . finally , khler - einstein metrics which correspond to equivalent invariant complex structures on @xmath8 are isometric ( cf .
@xcite ) . in order to detect which pairs of einstein metrics in the main theorem
are isometric or not , first we need to give their approximate values . note that the non - khler einstein metrics are of the form @xmath432 , where @xmath285 is obtained by solving equation @xmath455 ( see @xmath456 ) , if we first substitute the corresponding values of @xmath457 and @xmath423 .
next , @xmath286 is easily obtained from ( [ 3 ] ) . in the following table
we present the case of @xmath458 . respectively .
in particular , as we can see from table 1 , the einstein metrics on @xmath461 are obtained from the einstein metrics on @xmath462 , by a permutation of the components @xmath463 , for any @xmath464 , and conversely . and
@xmath17 are isometric via an element of the weyl group of @xmath12 . ]
thus we obtain the isometries @xmath465 , @xmath466 and @xmath467 .
this result is also obtained from table 2 , where we give the values of the corresponding scale invariants for the einstein metrics @xmath468 and @xmath469 .
also , from table 2 we easily conclude that all non khler invariant einstein metrics on @xmath465 , @xmath470 and @xmath471 are not isometric , since for any case it is @xmath472 .
this completes the examination of the case @xmath458 . for the special case @xmath22 with @xmath431 , the scale invariants corresponding to the new non - khler einstein metrics on @xmath473 given by @xmath432 , where @xmath285 and @xmath286 are determined by part @xmath433 of ( [ 400 ] ) , and @xmath434
, respectively , are equal . however , for @xmath474 @xmath286 is given by @xmath475 and for @xmath476 @xmath286 is given by @xmath477 thus these two einstein metrics on @xmath8 are isometric . | we find the precise number of non - khler @xmath0-invariant einstein metrics on the generalized flag manifold @xmath1 with @xmath2 and @xmath3 .
we use an analysis on parametric systems of polynomial equations and we give some insight towards the study of such systems . we also examine the isometric problem for these einstein metrics . 2000 _ mathematics subject classification . _
primary 53c25 ; secondary 53c30 , 12d05 , 65h10 _ keywords _ : homogeneous manifold , einstein metric , generalized flag manifold , algebraic systems of equations , resultant . |
the study of systems out of thermal equilibrium has attracted great attention in recent years . as their equilibrium counterparts , these systems may display continuous phase transitions characterized by a set of critical exponents @xcite .
in particular , much interest exists in transitions from a fluctuating active state towards an absorbing state , i.e. a configuration where the dynamics is frozen .
a prototype of a 1d lattice model with a transition into an absorbing state is the contact process @xcite , in which each lattice site can be either empty ( @xmath11 ) or occupied by a particle ( @xmath12 ) with the reactions @xmath13 , @xmath14 .
depending on the relative rates of these processes , the stationary state is empty ( when @xmath14 dominates ) , or is occupied by a finite density of particles ( if the reaction @xmath15 dominates ) .
the contact process therefore displays a non - equilibrium phase transition which is known to belong to the directed percolation ( dp ) universality class . the empty lattice ( @xmath16 ) is the absorbing state .
a wide range of models with transitions into absorbing states was found to belong to the dp universality class . as typical examples we quote the branching - annihilating random walks ( barw ) with an odd number of offsprings @xcite , the domany - kinzel model @xcite and the pair contact process @xcite .
the dp class therefore appears to be extremely robust and quite common , but it is certainly not the only possible one .
another universality class that has by now been firmly established is the so - called parity conserving ( pc ) class @xcite . a prototype model in this class is the barw with an even number of offsprings @xcite in which particles can diffuse and undergo the reactions @xmath17 , @xmath18 , with @xmath19 an even integer . in that model ,
the particle conservation modulo two is believed to be the reason for the system to show non - dp critical behavior .
more recently it became clear that the parity conservation , at least at the microscopic level , is not a necessary condition for a pc transition to occur @xcite .
hinrichsen @xcite provided an example of this by introducing a one dimensional model where each lattice site can be occupied by at most one particle or can be in any of @xmath0 inactive states ( @xmath20 , @xmath21 @xmath22 ) .
the reactions are : @xmath23 we refer to the model defined by the reactions ( [ reac_1]-[reac_4 ] ) as to the generalized contact process ( gcp ) .
the original contact process @xcite , corresponds to the case @xmath2 , in which the reaction ( [ reac_4 ] ) is obviously absent .
notice that the reaction ( [ reac_4 ] ) in the case @xmath24 ensures that configurations as @xmath25 , with @xmath26 are not absorbing .
such configurations do evolve in time until the different domains coarsen and one of the @xmath0 absorbing states @xmath27 , @xmath28 ,
@xmath29 is reached . for the gcp with @xmath3
, it was found from simulations @xcite that the transition falls in the pc class if @xmath30 , while if the symmetry between the two absorbing states was broken ( @xmath31 ) a dp transition was recovered @xcite .
one aim of this paper is to investigate the case @xmath32 , which has not been studied so far .
our results are obtained by a numerical investigation based on density matrix renormalization group ( dmrg ) techniques @xcite for the case @xmath4 and @xmath5 and an exact solution in the limit @xmath33 . on the basis of these results
we are led to conjecture that for @xmath6 the model is always in the same universality class , which coincides with that of multispecies branching and annihilating random walks @xcite .
there are several reasons which make the gcp interesting .
firstly , it contains the two mayor universality classes for transitions into absorbing states , namely dp ( @xmath2 ) and pc ( @xmath3 ) .
secondly , for @xmath34 is it interesting to study the effect of breaking the permutational symmetry of the model and for @xmath35 the symmetry can be broken in different ways ( see below ) . finally , it is also interesting to investigate the performance of the dmrg algorithm for a system with several absorbing configurations and with several states per site , a situation which is definitely more complicated than what was considered so far @xcite .
we recall that the application of the dmrg to non - equilibrium processes is rather recent and its power / limitations have not been fully investigated yet , therefore the gcp represents another important testing ground for this purpose .
this paper is organized as follows : in sec .
[ sec : pd ] we present the numerical results for the cases @xmath36 and @xmath37 . in sec .
[ sec : exact ] we show that in the limit @xmath38 the dynamics of the model can be mapped onto that of the zero temperature @xmath0-state potts model , which allows the determination of one critical exponent .
next , we present in sec . [
sec : conjecture ] a conjecture for the critical behavior of the model for @xmath6 based on the numerical and exact results . finally , in sec .
[ sec : breaking ] we analyse the effect of breaking the symmetry in the @xmath4 case , while sec .
[ sec : concl ] concludes the paper .
as a starting point for our analysis we use the quantum form of the master equation @xcite to describe the evolution of the stochastic processes in continuous time : @xmath39 where @xmath40 is a state vector whose elements are the probabilities of finding the system in a certain configuration , and the entries of the matrix @xmath41 are the transition probabilities per unit of time between different configurations . as in quantum mechanics we will call @xmath41 the hamiltonian of the system
. however , in the most general case , as in the gcp , the matrix @xmath41 is non - hermitian , therefore one should distinguish between right and left eigenvectors , which are now not related by transposition . as a second consequence
, the eigenvalues could be complex , but @xmath41 always has at least one eigenvalue that is zero , and the real part of the non vanishing eigenvalues is strictly positive . since we are interested in the stationary behavior of the system and the relaxation towards it , we will determine the low lying spectrum of the hamiltonian , i.e. the part of the spectrum with the smallest real part . in sections
ii - iv we will consider the system to be symmetric in the ground states : @xmath42 .
first of all , the conservation of probability always ensures the existence of a trivial left eigenvector with zero eigenvalue @xmath43 , where the sum is extended over all possible configurations @xmath44 .
besides this left ground state , the gcp has @xmath0 trivial right ground states : the @xmath0 absorbing configurations @xmath45 , with @xmath46 . to study the rest of the low lying spectrum
we apply the dmrg in combination with finite size scaling .
the dmrg is an iterative algorithm through which one constructs approximate eigenvalues and eigenvectors of @xmath41 for long chains . at each iteration
the lattice size is increased and the configurational space is truncated efficiently , so that one considers , instead of the exact operator @xmath41 , effective matrices of reduced dimensions , which can be handled numerically .
the accuracy obtained for the dominant eigenvalues and eigenvectors is often very good .
although originally invented for hermitian problems , the dmrg also works in non - hermitian cases , as it has been shown in several examples @xcite . for @xmath47
the gcp has more than one absorbing state and therefore more than two states per site .
this , together with the fact that the hamiltonian is non - hermitian , makes a dmrg study of the model technically difficult .
consequently it is of great importance to have a way to check the results of our method . until now
the gcp has only been studied by means of monte carlo simulations , for @xmath48 @xcite .
therefore , this subsection is restricted to these two cases .
we will explain how we applied the dmrg , give some details on the finite size scaling for @xmath3 and compare our results with @xcite .
a first quantity which can be calculated by the dmrg is the gap @xmath49 between ( the real part of ) the eigenvalue of the first excited state of the hamiltonian @xmath41 and the absorbing ground states @xmath50 .
@xmath49 is the inverse relaxation time and enables us to determine the critical region and the dynamical critical exponent @xmath51 .
a direct implementation of the dmrg is however very unpractical since for @xmath47 we have a degenerate ground state , and the dmrg is known to work best for gapped systems .
we alleviated this problem by adding at the two boundary sites the following reactions @xmath52 ( recall that it is customary to use open boundary conditions in dmrg @xcite ) . with ( [ reac_5 ] ) , only @xmath53 is left as a ground state .
the bulk critical behavior is however expected to be unchanged .
next we performed the transformation @xcite : @xmath54 with @xmath55 and where @xmath41 is constructed from the reactions ( [ reac_1]-[reac_4 ] ) and ( [ reac_5 ] ) .
@xmath56 is no longer a stochastic hamiltonian , the zero eigenvalue is shifted to @xmath57 , but the rest of its spectrum is exactly the same as that of @xmath41 .
this can easily be checked by looking at the eigenvalues of the left eigenvectors which are the same for both matrices .
therefore the calculation of the gap of the original hamiltonian with an @xmath0-times degenerate ground state is reduced to the calculation of the lowest eigenvalue of @xmath56 ( obviously provided @xmath57 is bigger than the gap of @xmath41 ) .
this strategy could be generalized to other systems with several absorbing states , provided one finds an appropriate boundary reaction which eliminates " zero eigenstates , as done by ( [ reac_5 ] ) for the gcp .
0.2truecm before turning to the finite size analysis of the calculated gap , let us first present the phase diagram for @xmath2 and @xmath3 .
figure [ fig01 ] shows these diagrams , obtained by the dmrg method and standard finite size scaling techniques .
the region above the lines denotes the active state , where the system has a finite stationary density of particles , while below these lines the stationary density is zero and the system is in one of the absorbing states .
we find ( see below ) that the critical exponents are the same all along the critical line and are those of the dp and pc universality class for @xmath2 and @xmath3 respectively .
we notice that the active region increases from @xmath2 to @xmath3 .
the location of the critical lines agrees well with monte carlo simulations by hinrichsen @xcite .
it is worth while at this point to present in some more detail the finite size scaling analysis employed for @xmath3 , in order to clarify the differences with the higher @xmath0 case . for each choice of the rates
@xmath58 we can calculate the gap @xmath59 of the matrix @xmath41 for a system of length @xmath60 .
this gap equals the lowest eigenvalue of @xmath56 defined in eq .
( [ shift ] ) . as a function of l
, @xmath59 has a different scaling behavior in the three regions of the phase diagram . in the active phase
the gap should scale as @xmath61 , since asymptotically in l the absorbing states are degenerate with a state having a finite density of particles . at the critical line we have @xmath62 , with @xmath63 the dynamical exponent . for @xmath3
the system is in the pc universality class and the whole inactive phase is known to be algebraic : the gap decays as @xmath64 .
notice that the latter behavior is different in dp models where the gap is finite , i.e. @xmath65 , in the inactive phase . the finite size scaling behavior of @xmath59 can best be monitored by plotting the discrete logarithmic derivative of the gap : @xmath66 $ ] . in the active phase
the scaling form @xmath61 implies @xmath67 ( with @xmath68 a positive constant ) , while @xmath69 at the critical line .
0.2truecm figure [ fig02 ] shows a plot of @xmath70 for @xmath3 , @xmath71 as a function of the parameter @xmath72 for @xmath73 ( in the present case the calculation of the gap was extended up to @xmath74 ) . in the right part of the figure one distinguishes the scaling for the active phase , while in the left part ( large @xmath72 ) one sees that @xmath70 flattens out and approaches the value of @xmath75 , as expected for the algebraic inactive phase of pc models .
the maxima of @xmath70 , marked with a filled circle in fig .
[ fig02 ] , identify the boundary between the active and inactive phase and can be used as critical point estimates . from an @xmath76 extrapolation we find @xmath77 for our choice of @xmath71 , which determines point b of the critical line in fig .
[ fig01 ] . in the inset of fig .
[ fig02 ] we plot @xmath78 , as function of @xmath79 . extrapolating @xmath80 in the limit @xmath76 we find @xmath81 , in accurate agreement with the corresponding exponent known for the pc class @xmath82 @xcite . in this way
, the dmrg enables us to determine the critical region and the exponent @xmath83 . as we will show in the following subsection , using different boundary conditions we can also find estimates for other exponents . in table
[ table01 ] we show our estimates of the critical exponents @xmath83 and @xmath84 , in the points a and b of fig .
[ fig01 ] .
we recover the exponents of dp and pc as expected for @xmath2 and @xmath3 respectively , indicating that the dmrg is performing well
. since the results of the dmrg are consistent with the monte carlo results for @xmath2 and @xmath3 , we can confidently use it to study also the case @xmath4 .
again we start by an analysis of the gap : in fig .
[ fig03 ] we plot the quantity @xmath70 along the line @xmath85 for @xmath86 . as for @xmath3
we find a clear evidence of an active phase for small @xmath72 where @xmath87 .
however , when we want to determine the critical point by localising the maxima , we find that upon increasing @xmath60 the maxima shift towards larger values of @xmath72 . in the limit @xmath76
we find @xmath88 , indicating that the system is always active and that the critical point is shifted to @xmath89 .
this numerical evidence together with the arguments presented in the next section lead us to expect that the model is critical only if @xmath89 , as indicated in fig .
[ fig01 ] .
the estimate of the exponent @xmath83 at @xmath90 yields @xmath91 . in the next section
we will give an analytical treatment of the case @xmath89 confirming this value of @xmath83 .
to get access to more critical exponents , we introduce different boundary conditions .
we replace ( [ reac_5 ] ) with : @xmath92 through which particles are continuously injected at the boundary sites .
now the states @xmath93 are no longer absorbing configurations , and the spectrum of the hamiltonian is non - degenerate .
there is a unique right eigenvector @xmath94 , which for finite system lengths always has a finite density of particles . in the active phase
the gap @xmath95 is no longer asymptotically degenerate , as in the previous case , but it remains finite , indicating that the system relaxes exponentially fast towards @xmath94 . summarizing , if we consider the model defined by the reactions ( [ reac_1]-[reac_4 ] ) and the boundary term ( [ inject ] ) we have a gap behaving as @xmath96 in the active phase , while as before @xmath97 at the critical point .
this means we can use @xmath98 itself to discriminate between the active and critical domains .
0.2truecm 0.2truecm figure [ fig04 ] shows a plot of the gap @xmath98 along the line @xmath85 .
when we extrapolate @xmath76 , the gap remains finite for every value of @xmath72 , i.e. we find again that the system is active throughout the whole parameter space .
these data also give the possibility to estimate the correlation length exponent along the time direction : @xmath99 .
finite size scaling around the critical point @xmath89 gives us the following scaling relation : @xmath100 for @xmath72 fixed and @xmath76 , @xmath98 remains finite , so the scaling function should behave like @xmath101 . as a consequence in the thermodynamic limit @xmath76
the gap should vanish at the critical point as @xmath102 .
the linear behavior of the gap in fig .
[ fig04 ] , when plotted as a function of @xmath103 , already indicates that @xmath104 and this results is also consistent with the scaling collapse reported in the inset , where we took @xmath104 and @xmath105 .
finally using the same boundary term ( [ inject ] ) it is also possible to compute the critical exponent @xmath106 which describes the behavior of the particle density near the critical point . with the boundary condition ( [ inject ] ) the stationary state of any finite system has a non - zero density of particles .
we calculated in particular the average density of particles in the central site of the chain which we will denote as @xmath107 .
this quantity , calculated along the line @xmath85 , is shown in fig .
[ fig03 ] for chains of various lengths ( up to @xmath74 ) and plotted as function of @xmath108 . for large @xmath72 and @xmath109 ( @xmath110 ) the dmrg does not perform so well and we had to restrict the calculation to @xmath111 . 0.2truecm
as seen in the figure in the limit @xmath76 the stationary particle density vanishes only for @xmath112 , again indicating that the system is always active for any finite @xmath72 . for large @xmath72 the order parameter
is expected to decay as @xmath113 therefore the quantity : @xmath114 converges to @xmath106 for @xmath115 and @xmath38 .
the inset of fig .
[ fig03 ] shows a plot of @xmath116 as function of @xmath108 for chains of various system lengths .
the extrapolated result for @xmath117 and @xmath111 is @xmath118 . in summary
, we found that for @xmath4 , the system is always active , except for @xmath89 where it is critical and we have exponents consistent with @xmath105 , @xmath119 , @xmath120 . to conclude our numerical calculations for the case of symmetrical ground states , we studied the system with four inactive states per site .
here we restricted ourselves to a calculation of the density @xmath107 with boundary condition ( [ inject ] ) .
our results are shown in figure [ fig06 ] .
0.2truecm from these data we find , as was the case for @xmath4 that the system is active for any finite value of @xmath72 , and that the critical exponent @xmath121 .
in this section we show how the dynamics of the model can be simplified in the limit where one of the rates of the decay processes ( [ reac_1 ] ) or ( [ reac_2 ] ) goes to infinity .
for @xmath38 the resulting effective dynamics coincides with that of a zero temperature @xmath0-state potts model .
this leads to a determination of the exponent @xmath83 .
intuitively , the argument goes as follows . for @xmath38 , all particles in the system will disappear very quickly , and in any finite system one
will soon be in a configuration containing only the @xmath0 empty sites @xmath123 , @xmath124 and @xmath125 .
particles will then be created at the boundaries separating inactive domains by ( [ reac_4 ] ) , but again they will disappear immediately by reaction ( [ reac_2 ] ) . hence , when looked at the time scale of the slow processes ( [ reac_3 ] ) and ( [ reac_4 ] ) the dynamics can be limited to the set of configurations that contain only empty sites .
( in this limit the creation of two particles on nearest neighbour sites occurs with even smaller probability and the effect of ( [ reac_1 ] ) can therefore be neglected ) . as an example
consider : @xmath126 the whole process consists of moving the domain wall one unit to the left with rate ( on the time scale of the slow process ) equal to @xmath127 .
going in this way through all possibilities one can derive an effective dynamics on the slow time scale , which turns out to involve diffusion , annihilation and coagulation of domain walls .
the above heuristic reasoning can be made mathematically rigorous using a fast rate expansion introduced in @xcite .
we therefore write @xmath41 as @xmath128 where @xmath129 contains the reactions ( [ reac_1]),([reac_3 ] ) and ( [ reac_4 ] ) while @xmath130 contains only the reactions ( [ reac_2 ] ) . in the limit @xmath38 ,
it is appropriate to apply the schwinger - dyson formula to the operator @xmath131 which appears in the formal solution of the master equation ( [ f1 ] ) .
one has @xmath132 \label{f2}\end{aligned}\ ] ] where @xmath133 using this expansion it can be shown @xcite that for @xmath38 the time evolution operator @xmath131 becomes @xmath134 where @xmath135 and @xmath136 is the projector on the ground states of @xmath137 , i.e. @xmath138 equation ( [ f4 ] ) shows that for @xmath38 the generator of the effective dynamics is @xmath139 which is nothing but @xmath129 projected onto the ground states of @xmath137 . in the gcp , the ground states of @xmath137 for a chain of length @xmath60 are the @xmath140 configurations without particles and @xmath141 is then the effective hamiltonian of the slow processes projected in this reduced space .
this is the mathematical description of the physical arguments given in the beginning of this section .
if one works out the matrix elements of @xmath141 ( see the appendix ) one obtains that the following processes can occur with the indicated rates @xmath142 it is now useful to interpret the @xmath0 empty states as the possible spin values of an @xmath0-state potts model . in this language , the processes ( [ 18]-[20 ] ) can be summarized as follows : the central spin assumes the value of one of its neighbour spins with equal probability .
the dynamics of our model in the limit @xmath38 is therefore consistent with the requirements of detailed balance for an @xmath0-state potts model at zero temperature .
it is generally expected that if such a dynamics includes a domain wall diffusion , that then it is critical with a dynamic exponent @xmath105 @xcite , independently of @xmath0 .
in fact , the exponent @xmath105 can be derived exactly for the case that the rate of the process ( [ 18 ] ) equals @xmath143 , and those of the processes in ( [ 20 ] ) equal 1/2 @xcite .
hence , it is quite possible that also for our model , @xmath105 exactly . for @xmath3
we thus recover the known dynamical exponent @xmath105 in the inactive phase of a model with a pc transition . for @xmath6
our numerical data strongly suggest that for @xmath38 the gcp is critical , and hence the exponent @xmath105 must correspond to the dynamical exponent at criticality for these models .
this estimate is indeed consistent with the value that was determined numerically for @xmath4 in the previous section .
the result that numerically the exponent @xmath106 is the same for @xmath4 and @xmath5 , combined with the fact that @xmath105 if @xmath6 leads us to conjecture that for all @xmath6 the critical exponents are the same , i.e. @xmath120 , @xmath105 and @xmath119 .
in the next section we will give further arguments that support this conjecture .
it is also possible to study the effective dynamics of the gcp for @xmath122 . in this limit
, the dynamics of the model when considered on the time scale of the slow processes ( [ reac_2]-[reac_4 ] ) , will be restricted to the space of configurations without particle pairs .
each particle present in the system then separates two domains of empty sites .
therefore particles can be labelled by two indices and in this way they can be divided into classes .
we will denote by @xmath144 a particle @xmath12 that separates an @xmath145-domain on its left side from an @xmath146-domain on its right side . in the limit @xmath122 , it is useful to look at the dynamics of these particles .
when @xmath2 there is only one type of particle , and from a determination of the effective hamiltonian @xmath141 ( see the appendix ) we find that this particle can diffuse , and undergo the reactions @xmath147 and @xmath148 .
since there are no processes that create particles , we arrive at the conclusion that independently of @xmath72 , the gcp with @xmath2 must always be in the inactive state when @xmath122 .
this is in agreement with our numerical results ( see fig .
[ fig01 ] ) . for @xmath47 ,
the situation is less clear .
now there are processes that both destroy and create particles present in the effective dynamics ( see the appendix ) .
it is therefore in principle possible to have both an active and an inactive phase , depending on the value of @xmath72 . at this moment
, we can draw no firm conclusions for the form of the phase diagram when @xmath122 . on the basis of our numerical work and on the basis of our results for @xmath38
, we believe that the model is probably always active along that line , at least when @xmath35 . 0.2truecm
in his paper , hinrichsen @xcite gave an heuristic argument which relates the gcp for @xmath3 to the barw with two offsprings , thus explaining the pc universality found numerically .
this argument works as follows : indicating with @xmath149 the domain wall between two configurations @xmath150 and @xmath151 he considered an effective dynamics for the variables @xmath149 . as seen in the previous section such representation of the original model becomes exact either in the limit @xmath122 and then @xmath149 coincides with the particle @xmath144 , or in the limit @xmath38 where @xmath149 coincides with the bond variable @xmath152 . for finite values of these parameters
one can still apply this reasoning at a coarse - grained level . in this case
@xmath149 is not a sharp domain wall , but an object with a fluctuating thickness .
hinrichsen argued that in this coarse - grained representation the most likely reactions for @xmath3 are : @xmath153 ( for @xmath3 , one has obviously @xmath154 , @xmath155 or viceversa ) .
an example of such reactions is shown in fig .
[ fig07 ] ( a ) and ( b ) .
notice that the reactions ( a ) and ( b ) given in ( [ barwe ] ) are those for a branching - annihilating random walk ( barw ) with two offsprings , which suggests indeed , as found numerically that the universality class is pc .
these arguments can be extended to the case @xmath35 , where there is still the possibility of having reactions of type ( a ) and ( b ) , but also reactions involving three different domains ( @xmath156 , @xmath157 and @xmath158 ) .
: @xmath159 when @xmath35 there are actually @xmath160 domain walls , and the model described by reactions ( [ barwe ] ) and ( [ barwngt2 ] ) is now a barw with more than one type of particles . to our knowledge
this type of model has not been studied yet , but we expect it to be in the same universality class as the gcp with @xmath35 , i.e. always in an active state except when the rates for the processes ( a ) and ( c ) are zero , and with exponents @xmath105 , @xmath120 and @xmath161 . there is a barw with more than one type of particles that has attracted some attention recently .
the model was introduced by cardy and tuber @xcite , who considered a system with @xmath162 different particles @xmath163 , where @xmath164 , @xmath165 , @xmath162 which diffuse and undergo the reactions : @xmath166 where in the last reaction it is understood that @xmath167 .
the _ bosonic _
version of this model has precisely the same exponents we determined for the gcp with @xmath168 @xcite .
notice that the coarse - grained representation of the gcp as defined by the reactions in fig .
[ fig07 ] and the model defined by the reactions ( [ cardy_1],[cardy_2],[cardy_3 ] ) do not actually coincide . while there is an obvious correspondence between ( [ cardy_1 ] ) , ( [ cardy_2 ] ) with ( b ) , ( a ) of fig .
[ fig07 ] , the reaction ( [ cardy_3 ] ) does not have any obvious counterpart .
it is not a priori clear therefore that the two models are in the same universality class .
the coincidence of the critical exponents therefore suggests that one could replace ( [ cardy_3 ] ) with other reactions , as for instance @xmath169 , without changing the universality class . to our knowlegde ,
barw models with this kind of reactions have not been studied yet .
they form an interesting subject for further investigation .
the model of cardy and tuber has raised some interest recently since it has been found that fermionic and bosonic versions of the model are in different universality classes @xcite . in the fermionic version only one particle per lattice site is allowed , which implies that particles of different species block each other . in the fermionic version of the model it makes for instance a difference wether
the two offsprings produced by the reaction ( [ cardy_3 ] ) are placed to the same side or at opposite sides of the parent particle @xcite .
if , for instance , they are placed at opposite sides the offsprings can not annihilate through the reaction ( [ cardy_1 ] ) because of the presence of the parent particle that blocks them . in the bosonic model the two offspring can instead always recombine .
it is important to stress here that in the multispecies barw model we constructed ( ( [ barwe])-([barwngt2 ] ) ) from a coarse - grained representation of the gcp there are no blocking effects . by the very construction of the model two domain walls approaching each other
can always annihilate .
therefore even if the model is clearly of fermionic nature its universality class , as found numerically , is that of the bosonic multispecies barw .
as a final point we consider the effect of breaking the permutation symmetry of the inactive states of the model . for @xmath3 hinrichsen @xcite explicitely broke the @xmath170 symmetry by choosing @xmath171 in the reaction ( [ reac_2 ] ) . as a result he found the system to switch from pc to dp behavior , which was understood as pc being related to the presence of an exact @xmath170 symmetry .
0.2truecm for @xmath4 , we now perform a similar symmetry - breaking by considering the following two cases : ( a ) @xmath172 and ( b ) @xmath173 . in both cases the system has a @xmath170 symmetry due to the equivalence of the states @xmath21 and @xmath174 .
the difference is that starting from a random configuration in the first case the system is more likely to reach the absorbing states @xmath175 and @xmath176 compared with @xmath177 , while in the second case the reverse is true .
we calculated the particle density @xmath178 as before , using the boundary - term ( [ inject ] ) .
depending on the phase in which the model is , the density will behave as @xmath179 ( in the active region ) , @xmath180 ( at criticality ) , or @xmath181 ( in the inactive phase ) . if we define @xmath182 /\ln\left [ ( l+1)/\ln ( l-1)\right ] $ ] , we expect @xmath183 to be zero in the active phase , to be @xmath84 at critical points , and @xmath184 in the inactive phase . in fig .
[ fig08 ] we plotted @xmath185 as a function of @xmath79 for cases ( a ) and ( b ) with the choice @xmath71 and injection rate @xmath186 . in both cases for small @xmath72 one finds the typical scaling behavior of the active phase with @xmath187 just as for the case @xmath188 . for large @xmath189
however the situation differs from the symmetric model . for ( a )
, we find @xmath190 for large @xmath72 , i.e. one has a standard inactive phase with a particle density exponentially small in @xmath60 . in case ( b ) , @xmath185 becomes equal to @xmath143 , implying that the inactive phase is itself critical with @xmath191 . in between the active and the inactive phase
we have a critical point where @xmath185 is going to a distinct finite value .
the critical point estimates of @xmath185 are marked by filled circles in fig .
[ fig08 ] . for case ( a ) the critical point is at @xmath192 , while for ( b ) it is at @xmath193
. the value of @xmath84 agrees with that of dp and pc respectively , as can be seen in fig .
[ fig08 ] where the critical values of these universality classes are indicated with a dotted line .
this indicates that on breaking the symmetry of the inactive states the remaining symmetry of the _ dominant _ rates @xmath194 determines the critical behavior . in case ( b ) @xmath195 the dominant rates still have a @xmath170 symmetry leading to the pc universality class , while in case ( a ) @xmath196 , dp critical behavior is recovered .
in this paper , we studied a generalised contact process first introduced by hinrichsen .
the major part of our results were obtained by applying the dmrg to the model . with this technique
we verified that for @xmath3 , the critical line of the model is in the pc universality class , consistent with earlier results coming from simulations .
a first set of new results was obtained for the case @xmath4 , for which we found that the model is always active , except when @xmath38 , which corresponds to the critical line of the model . from our numerical work we determined the critical exponents to be equal to @xmath197 and @xmath120 . using well established scaling laws @xcite , other exponents can be determined from these three . for @xmath5 , we found evidence that the phase diagram is the same , and that the critical exponent @xmath106 also equals 1 .
secondly , using a fast rate expansion that becomes exact for @xmath198 , we were able to argue that in that limit @xmath105 .
it can be hoped that by examining the model for @xmath199 small using perturbation techniques , it may be possible to determine also the exponents @xmath106 and @xmath99 exactly . on the basis of these numerical and exact results we conjectured that the universality class of the model is the same for all @xmath6 .
the exponent values that we found for @xmath6 coincide with those of the barw model with more than one type of particles introduced by cardy and tuber @xcite .
we were able to give an heuristic argument that explains why the two models could be in the same universality class .
it is interesting to remark that despite many attempts the number of universality classes found for phase transitions out of an adsorbing phase , remains very limited .
it could have been hoped _
priori _ that for the generalised gcp studied here , new universality classes could appear for @xmath168 . in a sense
our results show that this is true , but only in the least exciting way possible : the universality class does not depend on @xmath0 , and moreover the exponents take on rather trivial values .
one could hope that by lowering the permutation symmetry to a @xmath200-symmetry , other universality classes could appear for @xmath201 .
this could be done , e.g. by having the rates of the process ( [ reac_4 ] ) depend on @xmath202 .
but since this would only make a difference for @xmath201 , it will probably be difficult to investigate such a model with the numerical techniques currently available .
we also verified that if one breaks the permutation symmetry of the model with @xmath4 , one recovers a dp or pc universality , suggesting that it is the symmetry of the largest rates that determines the universality class .
finally , we remark that the consistency of the dmrg results with those coming from simulation for @xmath3 , or with the exactly determined value of @xmath83 for @xmath203 , shows convincingly that the dmrg can be trusted as a powerful method in the study of ( criticality in ) non equilibrium systems .
jh and cv thank the iuap , belgium for financial support .
ec is grateful to infm for financial support through pais 1999 .
in this appendix we will calculate explicitly @xmath204 where @xmath205 is one of the rates of the gcp .
in section iii it was shown that this reduces to the calculation of @xmath206 , the hamiltion that describes the effective dynamics . since @xmath207 this is a projection of @xmath129 on the ground states of @xmath137 .
let us denote the right ground states of @xmath137 as @xmath208 and the left ones as @xmath209 . @xmath210
the physical meaning of @xmath211 is evident : they are the stationary states of @xmath137 , any state @xmath212 will under the dynamics of @xmath137 relax into one of these @xmath213 .
the left ground states @xmath209 can be interpreted as linear functionals giving the corresponding probabilities : any state @xmath212 will under the dynamics of @xmath137 relax into the ground state @xmath211 with probability @xmath214 ( where we assumed that @xmath209 are normalized : @xmath215 as @xmath216 since clearly @xmath217 \left| \psi\right\rangle = 1 $ ] for any ( normalized ) state @xmath212 , this projection conserves probability , meaning that when @xmath129 is a stochastic operator , so is @xmath218 .
furthermore we can write the transition rates of the effective dynamics between state @xmath213 and @xmath219 as @xmath220 these matrix elements determine the effective dynamics , and we will now calculate them explicitly .
we first study the limit where the rate @xmath72 of the processes @xmath221 goes to infinity .
the hamiltonian is of the form : @xmath222 where @xmath129 is the generator of the processes ( [ reac_1 ] ) , ( [ reac_3 ] ) , ( [ reac_4 ] ) and @xmath137 is the generator of process ( [ reac_2 ] ) with the factor @xmath72 brought out . in this case
the ground states @xmath211 of @xmath137 are all configurations containing no particles @xmath12 .
firstly , we note that the processes ( [ reac_1 ] ) and ( [ reac_3 ] ) can only act on configurations containing particles , so they can not contribute to the rates ( [ transitionrates ] ) of the effective dynamics and we redefine @xmath129 without them . when we now project @xmath129 which contains only 2-site interactions , onto the @xmath211 , the resulting operator will contain 3-site interactions .
it is therefore convenient to first rewrite @xmath129 as a 3-site operator .
since we only have reaction ( [ reac_4 ] ) left in @xmath129 , this becomes ( @xmath223 ) @xmath224{ll } $ \emptyset_{k}\emptyset_{l}\emptyset_{k}\longrightarrow\emptyset_{k } a\emptyset_{k}$ & rate 2\\ $ \emptyset_{k}\emptyset_{l}\emptyset_{m}\longrightarrow\emptyset_{k } a\emptyset_{m}$ & rate 2\\ $ \emptyset_{k}\emptyset_{l}\emptyset_{l}\longrightarrow\emptyset_{k } a\emptyset_{l}$ & rate 1\\ $ \emptyset_{k}\emptyset_{l}a\longrightarrow\emptyset_{k}aa$ & rate 1 \end{tabular } \label{three - site}\ ] ] ( together with some reactions that are obtained by reflection ) .
we finally notice that the last process of ( [ three - site ] ) is again not relevant for the projection on @xmath211 , and determine the effective dynamics in the following diagram @xmath225{ccc } \begin{tabular } [ c]{c } reaction\\ \multicolumn{1}{l}{with rate ... } \end{tabular } & \begin{tabular } [ c]{c } projection\\ \multicolumn{1}{l}{with probability ... } \end{tabular } & netto rate\\ & & \\ $ \emptyset_{k}\emptyset_{l}\emptyset_{k}\stackrel{\text{rate 2 } } { \longrightarrow}\emptyset_{k}a\emptyset_{k}$ & $ \stackrel{\text{1 } } { \longrightarrow}\emptyset_{k}\emptyset_{k}\emptyset_{k}$ & rate $ 2.1=2$\\ $ \emptyset_{k}\emptyset_{l}\emptyset_{m}\stackrel{\text{rate 2 } } { \longrightarrow}\emptyset_{k}a\emptyset_{m}$ & $ \stackrel{\text{1/2 } } { \longrightarrow}\emptyset_{k}\emptyset_{k}\emptyset_{m}$ & rate $ 2.\frac { 1}{2}=1$\\ & $ \stackrel{\text{1/2}}{\longrightarrow}\emptyset_{k}\emptyset_{m } \emptyset_{m}$ & rate $ 2.\frac{1}{2}=1$\\ $ \emptyset_{k}\emptyset_{l}\emptyset_{l}\stackrel{\text{rate 1 } } { \longrightarrow}\emptyset_{k}a\emptyset_{l}$ & $ \stackrel{\text{1/2 } } { \longrightarrow}\emptyset_{k}\emptyset_{k}\emptyset_{l}$ & rate $ 1.\frac { 1}{2}=\frac{1}{2}$\\ & $ \stackrel{\text{1/2}}{\longrightarrow}\emptyset_{k}\emptyset_{l } \emptyset_{l}$ & rate $ 1.\frac{1}{2}=\frac{1}{2}$ \end{tabular } % \end{equation}\end{aligned}\ ] ] which are the processes and their corresponding rates already given in ( [ 18]-[20 ] ) .
note that this calculation is exactly the same for @xmath226 .
next , we consider now the limit @xmath122 . in this case
we have @xmath227 where @xmath129 is the generator of the processes ( [ reac_2]-[reac_4 ] ) , and @xmath137 is the generator of process ( [ reac_1 ] ) with the factor @xmath109 brought out .
the ground states of @xmath137 are now all configurations containing no particle pairs .
in contrast with the previous case , all processes of @xmath129 are now relevant for the projection on the ground states of @xmath137
. we will start with the case @xmath2 , where there is only one inactive state @xmath11 , and process ( [ reac_4 ] ) can a priori not take place . for process ( [ reac_3 ] )
we again use the 3-site representation , while process ( [ reac_2 ] ) is so simple that we keep the 2-site representation .
we then get the following effective dynamics @xmath224{ccc } \begin{tabular } [ c]{c } reaction\\ \multicolumn{1}{l}{with rate ... } \end{tabular } & \begin{tabular } [ c]{c } projection\\ \multicolumn{1}{l}{with probability ... } \end{tabular } & netto rate\\ & & \\ $ a\emptyset\stackrel{\text{rate } \mu}{\longrightarrow}\emptyset\emptyset$ & $ \stackrel{\text{1}}{\longrightarrow}\emptyset\emptyset$ & rate $ \mu.1=\mu$\\ $ a\emptyset\emptyset\stackrel{\text{rate 1}}{\longrightarrow}aa\emptyset$ & $ \stackrel{\text{1/2}}{\longrightarrow}a\emptyset\emptyset$ & rate $ 1.\frac { 1}{2}=\frac{1}{2}$\\ & $ \stackrel{\text{1/2}}{\longrightarrow}\emptyset a\emptyset$ & rate $ 1.\frac{1}{2}=\frac{1}{2}$\\ $ a\emptyset a\stackrel{\text{rate 2}}{\longrightarrow}aaa$ & $ \stackrel { \text{1/2}}{\longrightarrow}a\emptyset a$ & rate $ 2.\frac{1}{2}=1$\\ & $ \stackrel{\text{1/4}}{\longrightarrow}\emptyset a\emptyset$ & rate $ 2.\frac{1}{4}=\frac{1}{2}$\\ & $ \stackrel{\text{1/8}}{\longrightarrow}a\emptyset\emptyset$ & rate $ 2.\frac{1}{8}=\frac{1}{4}$\\ & $ \stackrel{\text{1/8}}{\longrightarrow}\emptyset\emptyset a$ & rate $ 2.\frac{1}{8}=\frac{1}{4}$ \end{tabular } \label{zwappa}\ ] ] for @xmath228 and @xmath2 we find the effective dynamics to contain only diffusion and destruction of particles . because of the first reaction appearing in ( [ zwappa ] ) , the decay of particles is exponentially fast , meaning that for any finite value of @xmath72 this system is non - critical . for the case @xmath47 also process ( [ reac_4 ] ) has to be taken into account . as a consequence ,
two different neighbouring inactive domains remain active .
for example @xmath229 remains a process of the effective dynamics .
one can easily construct the complete effective dynamics , but this does not lead to much further insight in the phase diagram .
one can only conclude that because of the presence of the process ( [ ad ] ) , it is in principle possible that both active and inactive phases are present .
it should be noted that rigorously for the calculation of the exponent @xmath83 one should first extrapolate the value of @xmath230 to the thermodynamic limit @xmath115 to obtain an estimate of the critical value @xmath231 .
then perform another extrapolation @xmath232 .
however in the present model both numerical procedures give the same extrapolated value for @xmath83 .
for a more detailed discussion on this point see the appendix b of ref .
notice also that since the maxima of @xmath233 are not so pronounced one can locate accurately their ordinates , while their abscissas are somewhat less accurate .
this is the reason why extrapolated value of @xmath83 are more accurate than the location of the critical point itself . | we investigate the critical properties of a one dimensional stochastic lattice model with @xmath0 ( permutation symmetric ) absorbing states .
we analyze the cases with @xmath1 by means of the non - hermitian density matrix renormalization group . for @xmath2 and @xmath3
we find that the model is respectively in the directed percolation and parity conserving universality class , consistent with previous studies . for @xmath4 and @xmath5 ,
the model is in the active phase in the whole parameter space and the critical point is shifted to the limit of one infinite reaction rate .
we show that in this limit the dynamics of the model can be mapped onto that of a zero temperature @xmath0-state potts model . on the basis of our numerical and analytical results
we conjecture that the model is in the same universality class for all @xmath6 with exponents @xmath7 , @xmath8 and @xmath9 .
these exponents coincide with those of the multispecies ( bosonic ) branching annihilating random walks . for @xmath4
we also show that , upon breaking the symmetry to a lower one ( @xmath10 ) , one gets a transition either in the directed percolation , or in the parity conserving class , depending on the choice of parameters . 2 |
consider an oriented unknotted braid axis @xmath0 in @xmath1 whose complement fibers over @xmath2 with oriented disc fibers @xmath3 . for a given oriented topological link type @xmath4 with @xmath5 components we study
closed braid representatives @xmath6 which are embedded in @xmath7 and which transversely intersect the @xmath8-disc fibers positively .
our first result is the following : [ prop : annuli ] let @xmath9 be braided about a common unknotted braid axis in @xmath1 . then there exists two braids @xmath10 and @xmath11 whose @xmath5 components pairwise cobound @xmath5 embedded annuli , such that @xmath10 is obtained from @xmath12 via negative stabilizations , and @xmath11 is obtained from @xmath13 via positive stabilizations and braid isotopy .
we remark that this final braid isotopy will in general involve @xmath14 passing through @xmath15 so as to obtain the embedded annuli . for any closed braid @xmath16 ,
we denote its braid index by @xmath17 and its algebraic length by @xmath18 .
we use the above proposition to prove the following theorem , namely the generalized jones conjecture which relates the braid index and algebraic length of closed braids within a topological link type @xcite .
[ thm : jones ] let @xmath19 be two closed braids such that @xmath20 is minimal for its link type
. then @xmath21 as noted by others , the veracity of jones conjecture yields immediate applications to the study of transverse links in the contact 3-sphere , quasi - positive and strongly quasi - positive braids , representations of braid groups and polynomial invariants for links @xcite .
recently , dynnikov and prasolov provided a proof of jones conjecture by studying bypasses for rectangular diagrams representing legendrian links @xcite .
we present here an independent proof using an alternative approach .
the outline of the paper is as follows : in section [ sec : annuli ] we briefly review background , and then prove proposition [ prop : annuli ] using braid foliation techniques of birman - menasco ; in particular , our argument is inspired by the constructions in @xcite and 2 of @xcite . in section [ sec : jones ] we then establish theorem [ thm : jones ] by proving an equivalent statement proposed by kawamuro @xcite .
* acknowledgements . *
the authors would like to thank joan birman and keiko kawamuro for their helpful comments while reading a preliminary version of the current paper .
let @xmath16 be braided about an unknotted braid axis @xmath0 with braid fibration @xmath22 .
the _ braid index _ of @xmath16 , denoted @xmath17 , is the number of intersections of @xmath16 with any @xmath8-disc .
the _ algebraic length _ of @xmath16 , denoted @xmath18 , is the sum of the signed crossings in any regular braid projection of @xmath16 .
given @xmath16 , we may alter it through standard moves , namely : _
braid isotopy _ in the complement of @xmath0 which does not change @xmath23 nor @xmath24 ; _ exchange moves _ which change neither @xmath23 nor @xmath24 ( see the left side of figure [ fig : basicbraidmovesb ] ) ; and _ stabilization ( destabilization ) _ which increases ( decreases ) @xmath23 and either increases or decreases @xmath24 depending on whether the stabilization or destabilization is positive or negative ( see the right side of figure [ fig : basicbraidmovesb ] ) . throughout this paper
we will be studying braids using various embedded oriented annuli and bigon discs . by general position
we may assume that @xmath0 intersects our surface of interest @xmath25 transversely in finitely many points , called _ vertices _ , of either positive or negative parity , depending on whether the orientation of @xmath0 agrees or disagrees with the orientation of @xmath25 at that vertex .
provided the boundary components , or boundary arcs , of our surface @xmath25 are transverse to @xmath8-discs in the braid fibration , by standard braid foliation arguments @xcite we may also assume that there are finitely many points of tangency between @xmath25 and the @xmath8-disc fibers , all of which yield simple saddle singularities in the foliation of @xmath25 induced by the braid fibration .
we will generically refer to such points as _ singularities _ , and again these will be of positive or negative parity depending on whether the orientation of @xmath25 agrees or disagrees with the orientation of @xmath8 at those points . also by standard arguments @xcite ,
the non - singular leaves in the @xmath8-foliation of our discs and annuli will either be _ s - arcs _ whose endpoints are on two different boundary components of an annulus ( or different boundary arcs of a bigon disc ) ; _ a - arcs _ with one endpoint on a vertex and one endpoint on a boundary component or arc ; or _ b - arcs _ whose endpoints are on two different vertices of opposite parity .
furthermore , singular leaves can then be classified as to what non - singular leaves interact to form the saddle singularity ; specifically , in general we will have @xmath26- , @xmath27- , @xmath28- , @xmath29- and @xmath30-singularities .
these are depicted in figure [ fig : tilesb ] , where the oriented transverse boundary is given by bold black arrows ; we will refer to the gray shaded regions as @xmath26- , @xmath27- , @xmath28- , @xmath29- and @xmath30-tiles , classified by the singularity which they contain .
the _ valence _ of a vertex is the number of singular leaves for which it serves as an endpoint ; alternatively the valence is the number of ( one - parameter families of ) @xmath31-arcs or @xmath32-arcs to which the vertex is adjacent .
any vertex may therefore be labeled as type @xmath33 where @xmath34 and @xmath16 are the number of @xmath31-arcs and @xmath32-arcs to which the vertex is adjacent , with the valence of the vertex being @xmath35 .
as discussed in detail in @xcite and shown in figure [ fig : valence123b ] , valence-1 vertices of type @xmath36 indicate a destabilization of the braided boundary of a surface which results in the elimination of that vertex ; valence-2 vertices of type @xmath37 or @xmath38 indicate an exchange move with a corresponding elimination of two vertices ; and valence-3 vertices of type @xmath39 indicate the presence of two consecutive singularities of like parity adjacent to that vertex whose @xmath8-order can be interchanged using braid isotopy , thus reducing the valence of that vertex to 2 .
this re - ordering of consecutive singularities of like parity is referred to as a _ standard change of fibration _
see the bottom right picture in figure [ fig : valence123b ] . as a result , using destabilizations , exchange moves and braid isotopy we can reduce the total number of vertices provided there are such vertices of valence-1 , -2 or -3 . with this background
, we can now proceed with the proof of proposition [ prop : annuli ] , namely that given braids @xmath9 , then after negative stabilizations of @xmath12 and positive stabilizations of @xmath13 , we obtain two braids @xmath40 which cobound @xmath5 embedded annuli , one for each component of the link type @xmath4 .
we consider @xmath12 and @xmath13 braided about a common axis @xmath0 , and furthermore think of @xmath0 as the @xmath41-axis in @xmath42 , with @xmath43 and @xmath44 .
we begin with @xmath12 and consider a braided push - off @xmath45 of @xmath12 such that @xmath46 and @xmath47 cobound @xmath5 embedded annuli , one for each component of @xmath4 . now consider a regular braid projection of @xmath47 onto the @xmath48-plane in @xmath49 , and wherever there is a double point at which @xmath45 passes under @xmath12 , we imagine isotoping @xmath45 locally upwards in the @xmath41-direction through @xmath12 .
the result is @xmath50 which is in fact braid isotopic to @xmath45 , but which is now unlinked from @xmath12 , yet still in @xmath49 .
furthermore , for each of these @xmath51 crossing changes , we have a bigon disc @xmath52 , @xmath53 , cobounded by arcs of @xmath45 and @xmath50 and intersected once by @xmath12 ; see figure [ fig : passthroughb ]
. passes through @xmath12 to produce @xmath50 , and produces a bigon disc @xmath52 which is intersected once by @xmath12 .
, scaledwidth=75.0% ] we may now vertically braid isotop @xmath50 in the positive @xmath41-direction so that @xmath54 . then , since @xmath55 , there is an ambient isotopy of @xmath56 , relative to the @xmath57-plane , which takes @xmath50 to @xmath13 .
the braid @xmath45 will experience an induced isotopy to a link @xmath58 which continues to cobound with @xmath12 a total of @xmath5 embedded annuli .
similarly , the discs @xmath52 will experience an induced isotopy .
the boundary of each new @xmath52 consists of two arcs , namely @xmath59 and the boundary arc @xmath60 ; see figure [ fig : isotopyb ] .
observe that the braid @xmath12 intersects each disc @xmath52 once close to the @xmath61 boundary arc in @xmath49 . since @xmath59 it is transverse to the @xmath62-fibration .
however , the @xmath61 boundary arc will not necessarily be transverse to the @xmath62-fibration in @xmath56 , although it will still be transverse to the portions of @xmath8-discs in @xmath49 . and
discs @xmath52 after isotoping @xmath50 to @xmath13 in @xmath56.,scaledwidth=60.0% ] we now apply alexander s theorem for links in @xmath56 so as to make the @xmath61 boundary arcs for the discs @xmath52 transverse to the @xmath62-fibration , and we can do so without changing the conjugacy class of @xmath13 ( @xcite , see also 2 of @xcite ) .
this takes the link @xmath58 to a braid @xmath63 which continues to cobound with @xmath12 a total of @xmath5 embedded annuli .
moreover , we can now realize bigon discs @xmath52 whose braid foliations may be assumed to consist of bands of @xmath64-arcs alternating with regions tiled by @xmath26- , @xmath28- and @xmath27-tiles , with either a single @xmath29-tile serving as a transition between the @xmath64-band and the tiled region , or two @xmath30-tiles serving as the transition on either end of a bigon disc .
we orient each @xmath52 so as to agree with the orientation of the @xmath61 boundary arc , so that @xmath31-arcs along @xmath61 connect to positive vertices , and @xmath31-arcs along @xmath65 connect to negative vertices ; we then work to simplify the foliation of each @xmath52 .
first , consider all singular leaves which intersect the @xmath61 boundary arc ; by slightly perturbing @xmath52 if necessary we may assume @xmath12 does not intersect any such singular leaves .
if any of these singular leaves has its other endpoint on a vertex @xmath66 , we may stabilize the @xmath61 arc along that singular leaf so as to remove that vertex @xmath66 from the foliation of @xmath52 ( see figure [ fig : stabilizesingb ] ) .
along singular leaves if any of these three configurations occur.,scaledwidth=65.0% ] thus we may assume that all singular leaves which intersect the @xmath61 boundary arc in fact intersect it twice . as a result , if we consider the graph in @xmath52 which consists of singular leaves that connect vertices to vertices , or vertices to one of the boundary arcs @xmath61 or @xmath65 , this graph must be a forest , with the root of each tree on the @xmath65 boundary arc .
in other words , the foliation of each @xmath52 consists of a band of @xmath64-arcs with trees of @xmath26-tiles extending off of it along the @xmath61 boundary arc , with a single @xmath29-tile connecting each tree of the forest to the band of @xmath64-arcs see the left - hand side of figure [ fig : treeb ] .
whose vertex - singularity graph is a tree after stabilizing @xmath61 ; on the right is @xmath52 whose graph is a linear tree after destabilizing @xmath61.,scaledwidth=65.0% ] we can now destabilize the @xmath61 boundary arc along each outermost @xmath26-tile containing a valence-1 vertex , as long as it does not contain the point of intersection with @xmath12 ; the result is that each @xmath52 may be assumed to be a linear string of @xmath26-tiles as in the right - hand side of figure [ fig : treeb ] , where the outermost tile farthest from @xmath13 contains the intersection with @xmath12 .
observe that throughout this simplification of the foliation , both @xmath13 and @xmath12 are fixed , and the resulting braid @xmath63 continues to cobound with @xmath12 a total of @xmath5 embedded annuli .
we now examine the resulting linear foliation on a single @xmath52 ; it consists of positive vertices , along with a sequence of singularities .
the result is a twisted band , with the parity of the half - twists given by the signs of the singularities .
it is then evident that if anytime a negative singularity is consecutive with a positive singularity , we may perform an exchange move so as to re - order those two singularities see figure [ fig : exchangesb ] .
moreover , this exchange move involves an isotopy of @xmath52 which is performed in a regular neighborhood of the sub - disc of @xmath52 cobounded by @xmath61 and the singular leaves associated with the two singularities ; thus the isotopy fixes both @xmath12 , @xmath13 and the other @xmath52 .
-end of the @xmath52 linear tree .
the black dots represent intersections of the braid axis @xmath0 with @xmath52.,scaledwidth=50.0% ] in this way , we may arrange that all positive singularities are stacked at the @xmath13-end of the foliation of @xmath52 , and all negative singularities are stacked at the @xmath12-end of the foliation of @xmath52 .
we then can negatively destabilize @xmath61 through those singularities , therefore inducing negative stabilizations of @xmath12 to arrive at @xmath15 , as depicted in figures [ fig : destabstabb ] and [ fig : microflypeb ] .
( see ` microflypes ' in 2.3 of @xcite . )
we then observe that stabilizing @xmath13 along the remaining positive singularities and allowing it to pass through @xmath15 will yield @xmath14 which cobounds with @xmath15 a total of @xmath5 embedded annuli . along with corresponding negative stabilizations of @xmath12 which yield @xmath15.,scaledwidth=75.0% ] induces a stabilization of @xmath12.,scaledwidth=70.0% ]
before proving theorem [ thm : jones ] , we recall a reformulation of the generalized jones conjecture as described by kawamuro @xcite . to do so , observe that for any braid @xmath6 there is an ordered pair @xmath67 ; we then have the following definition : in the @xmath68-plane , the cone of @xmath16 indeed has the shape of a cone ; it consists of a lattice of points at or above @xmath67 which are contained within the region bounded by lines of slope + 1 and -1 passing through the point @xmath67 ; see figure [ fig : coneb ] .
suppose for contradiction that there is a @xmath71 whose cone contains points outside the cone of @xmath70 .
then in fact @xmath72 must be outside the cone of @xmath70 .
we assume for the moment that @xmath72 is such that @xmath73 for any @xmath16 in the cone of @xmath70 with @xmath74 .
in other words , as we look at the cone of @xmath70 , we see @xmath12 _ to the left _ of the cone of @xmath70 see figure [ fig : twoconesb ] .
by proposition [ prop : annuli ] there exists @xmath75 such that the @xmath5 components of each braid pairwise cobound @xmath5 embedded annuli , where @xmath76 is obtained from @xmath70 by positively stabilizing , and @xmath15 is obtained from @xmath12 by negatively stabilizing . as a result
, we observe that @xmath15 lies _ to the left and outside of _ the cone of @xmath70 , and @xmath76 lies _ to the right and outside of _ the cone for @xmath12 ; see figure [ fig : twoconesb ] .
our goal in the proof of the current proposition is to use the embedded annuli to find a braid @xmath77 which is obtained from @xmath76 by destabilizations , braid isotopy and exchange moves , and a braid @xmath78 which is obtained from @xmath15 by destabilizations , braid isotopy and exchange moves , such that the cone of @xmath77 equals the cone of @xmath78 . since neither braid isotopy nor exchange moves changes the algebraic length or braid index of a braid , the conclusion is that @xmath79 , which is our desired contradiction since @xmath70 is at minimum braid index .
it is then evident that if we begin with @xmath12 _ to the right _ of the cone of @xmath70 , we can reverse the roles of @xmath80 in proposition [ prop : annuli ] to achieve a similar contradiction .
so it remains to consider a representative annulus @xmath81 cobounded by a component of @xmath76 and a component of @xmath15 .
first , we observe that any @xmath26- or @xmath29- tile must separate from @xmath81 a subdisc @xmath82 which is cobounded by the associated singular leaf and a subarc of a single component of @xmath83 . by an euler characteristic calculation ( lemma 7 of @xcite ) , we are guaranteed that @xmath84 where @xmath85 is the number of vertices of type @xmath33 in @xmath82
. therefore , the tiling of @xmath82 either contains valence-1 vertices which we can remove by destabilizing , or valence-2 or valence-3 vertices which can then be eliminated after braid isotopy and exchange moves .
we may therefore assume that we can always eliminate @xmath26- or @xmath29- tiles .
the result then is an annulus @xmath81 whose foliation consists of @xmath64-bands alternating with bigon discs @xmath86 such that two @xmath30-tiles serve as the transition on either end of each bigon disc .
then by a related euler characteristic calculation ( lemma 6.3.1 of @xcite ) for our annulus @xmath81 , we know that @xmath87 where @xmath88 is twice the number of @xmath64-bands ; furthermore , if both sides equal zero we know that only vertices of type @xmath89 or @xmath90 appear .
if @xmath91 we may therefore eliminate all vertices in the foliation of @xmath81 using braid isotopy , exchange moves and destabilizations , and we obtain an annulus whose foliation consists entirely of @xmath64-arcs .
if this is the case for all of our @xmath5 embedded annuli , then we are done , since the resulting @xmath77 and @xmath78 will have the same braid index and algebraic length . otherwise , if @xmath92 for some annulus , then again using braid isotopy , exchange moves and destabilizations we either obtain an annulus entirely foliated by @xmath64-arcs ( in which case we are done ) , or an annulus whose foliation is tiled entirely by @xmath28- or @xmath27-tiles in which every vertex is either of type @xmath89 or @xmath90 .
if any of the remaining vertices of type @xmath89 or @xmath90 are adjacent to consecutive singularities of like parity on either end of a one - parameter family of @xmath32-arcs , then we may perform a standard change of fibration to obtain either a vertex of type @xmath37 or @xmath39 which we may then remove following braid isotopy and exchange moves .
we may therefore assume we obtain a tiling in which consecutive singularities around any vertex alternate parity ( see figure [ fig : checkerboardb ] ) .
the tiling of @xmath81 will then be composed of a subannulus of some @xmath51 number of @xmath28-tiles along @xmath77 , and a subannulus of the same @xmath51 number of @xmath28-tiles along @xmath78 , along with @xmath93 subannuli containing @xmath94 number of @xmath27-tiles which interpolate between the subannuli of @xmath28-tiles ; in figure [ fig : checkerboardb ] , @xmath95 and @xmath96 .
however , in this case the resulting braids @xmath77 and @xmath78 will then have the same braid index and algebraic length , and we achieve the desired contradiction . to justify this last statement , observe that if the annulus @xmath81 is oriented so as to agree with @xmath78 , we may stabilize @xmath78 along singular leaves in @xmath81 some @xmath97 number of times to remove all negative vertices in @xmath81 , and then destabilize along valence-1 vertices the same @xmath97 number of times to remove all positive vertices from @xmath81 ; moreover , the parity of all of these stabilizations and destabilizations are identical , and thus @xmath98 . | we show that after stabilizations of opposite parity and braid isotopy , any two braids in the same topological link type cobound embedded annuli .
we use this to prove the generalized jones conjecture relating the braid index and algebraic length of closed braids within a link type , following a reformulation of the problem by kawamuro . |
it has been known for a long time that the floer theoretic invariants of cotangent bundles should be expressible in terms of classical invariants of the base .
the prototypical such result is floer s proof in @xcite that the lagrangian floer cohomology groups of the zero section in a cotangent bundle agree with its ordinary cohomology groups .
closer to the subject of this paper , abbondandolo and schwarz proved in @xcite that the wrapped floer cohomology of a cotangent fibre is isomorphic to the homology of the based loop space .
the study of fukaya categories in the setting of homological mirror symmetry as well as some of its applications to lagrangian embeddings ( see @xcite ) requires understanding such floer theoretic invariants at the chain level .
building upon the results in this paper , we shall prove in @xcite that the wrapped fukaya category of a cotangent bundle is generated by a fibre , which makes it the most important object to study from the categorical point of view . in section [ sec : cotangent ]
we explain the proof of the following result .
[ thm : main ] if @xmath4 is a closed smooth manifold , there exists an @xmath2 equivalence @xmath5 between the homology of the space of loops on @xmath4 based at @xmath6 and the floer cohomology of the cotangent fibre at @xmath6 taken as an object of the wrapped fukaya category of @xmath7 with background class @xmath8 given by the pullback of @xmath9 . in appendix
[ sec : orientations ] , we shall use a hybrid of the methods appearing in @xcite and @xcite in order to define the coherent orientations of moduli spaces of holomorphic discs needed to prove this result . in lemma [ lem :
twisted_to_untwisted_sign_difference ] , we shall prove that the contribution of each holomorphic disc to an operation on floer cohomology for background class @xmath10 differs from the corresponding contribution for the trivial background class by a sign equal to the intersection number with an appropriate cycle poincar dual to @xmath10 .
this proves that if we consider floer cochains for the trivial background class , the count of holomorphic curves that we define also produces an @xmath2 equivalence @xmath11 where @xmath12 is the @xmath13-local system on the based loop space which is uniquely determined up to isomorphism by the property that the monodromy around a loop in @xmath14 is given by the evaluation of @xmath15 on the corresponding torus .
theorem [ thm : main ] relies on the fact that a certain map we shall construct inverts an isomorphism constructed by abbondandolo and schwarz in @xcite . in summer 2009 , the author was informed by schwarz that he , together with abbondandolo , can prove that these maps are indeed inverses and planned to write it out in an upcoming paper , leading the author to write down a proof that the maps are right inverses ( the proof is sketched in section [ sec : cotangent ] ) .
an interesting analytic problem seems to arise when trying to construct by hand the homotopy that would independently prove that the maps are also left inverses , but this is of course unnecessary as the right inverse to an isomorphism is also a left inverse .
[ rem : sign_error ] the original version of this paper claimed a result for general @xmath4 , but in fact assumed it to be @xmath16 in the proof , and did not specify that , in the non-@xmath0 case , the cotangent fibre should be considered as an object of a wrapped fukaya category with a non - trivial background class ( see section [ sec : at - level - objects ] and appendix [ sec : orientations ] for a discussion of this modification of the usual fukaya category ) . in the case of non-@xmath0 manifolds , the signed contribution of moduli spaces of holomorphic half - strips had not been analysed .
the fact that the symplectic literature failed to account for these necessary signs was revealed , in the setting of generating functions , by work of kragh @xcite in his generalisation of viterbo s restriction map , and verified in a computation of the symplectic cohomology of @xmath17 by seidel in @xcite relying on floer - theoretic methods largely independent from the ones used here .
theorem [ thm : main ] is a relatively straightforward application of the main construction of this paper ( using as well as abbondandolo and schwarz s main result from @xcite at the level of homology ) which concerns the wrapped fukaya category @xmath18 of a liouville manifold @xmath3 .
recall that a liouville manifold is an exact symplectic manifold which may be equipped with an end modelled after the positive half of the symplectisation of a contact manifold .
the wrapped fukaya category of such a manifold has as objects exact lagrangians modelled after legendrians along this cylindrical end .
such a category was defined in @xcite , but we shall give a different construction in section [ sec : revi - wrapp - fukaya ] .
in real dimension @xmath19 , liouville manifolds correspond to symplectic structures on punctured surfaces ; a choice of liouville form fixes a symplectomorphism from a neighbourhood of each puncture to @xmath20 and the collection of lagrangians we consider agree with radial lines in these coordinates ( see figure [ fig : exact ] ) .
[ fig : exact ] let us now assume that a closed manifold @xmath4 embeds as an exact lagrangian in @xmath3 , and that there exists a class @xmath21 whose restriction to @xmath4 agrees with the second stiefel - whitney class @xmath15 ; we say that such a lagrangian is _ relatively @xmath0 _ for the background class @xmath10 .
we associate to @xmath4 a category @xmath22 whose objects are points of @xmath4 and morphisms are chains on the spaces of paths between such points .
the endomorphism algebra of any object in this category is @xmath23 , which can be made into a differential graded algebra by using normalised cubical chains as explained in the next section .
given an exact lagrangian submanifold @xmath24 whose second stiefel - whitney class is also given by the restriction of @xmath10 , we construct an explicit module @xmath25 over the category @xmath22 , in fact a twisted complex , using the moduli spaces of holomorphic strips bounded on one side by @xmath4 and on the other by @xmath24 . with minor technical differences coming from choices of basepoints ,
such a module was constructed by barraud and cornea in @xcite .
we prove that this assignment extends to an @xmath2 functor @xmath26 where the left hand side is the wrapped fukaya category with respect to the background class @xmath10 , and the right hand side is the category of twisted complexes . applying this functor to @xmath27 and @xmath28 yields theorem [ thm : main ] .
moreover , using theorem [ thm : main ] we conclude [ cor : restriction ] if @xmath4 is an exact relatively @xmath0 lagrangian embedded in @xmath3 , there is an @xmath2 restriction functor @xmath29 we may split the map to obtain an @xmath2 equivalence @xmath30 the left hand side may be thought of as twisted complexes on @xmath31 which are built using only the object @xmath32 .
composing this equivalence with the inclusion of the right hand side in @xmath33 and the functor @xmath34 , we obtain the desired result .
the reader may want to compare this circuitous construction of a restriction map with the one defined in @xcite for an inclusion @xmath35 of a liouville subdomain .
the key differences are that the construction in @xcite can only be performed for lagrangians which intersect the boundary of @xmath36 in a very controlled way , and that for such lagrangians the image of the restriction functor lies in @xmath37 ( rather than twisted complexes thereon ) .
corollary [ cor : restriction ] leads us to make the following conjecture : any inclusion of a liouville subdomain @xmath35 induces an @xmath2 restriction functor @xmath38 it is important to note that this result can not hold if we do not pass to twisted complexes on the right hand side ( or derived categories if the reader prefers that language ) .
i would like to thank ivan smith for the invitation to visit cambridge university in the winter of 2008 , when we discussed wrapped fukaya categories in various setting , and ralph cohen for the invitation to lecture at the march 2008 stanford university workshop on string topology , where i presented a preliminary version of these results .
conversations , electronic and otherwise , with ralph cohen and andrew blumberg as well as peter albers and dietmar salamon helped , respectively on the topological and symplectic parts , to clarify the technical problems to be circumvented . finally , i would like to thank thomas kragh and paul seidel for discussions about the sign error mentioned in remark [ rem : sign_error ] .
for the parts of the paper dealing with the homological algebra of @xmath2 categories , we shall mostly use results which appear in the first part of @xcite .
in particular , our sign conventions for orienting moduli spaces of holomorphic discs are the same as those appearing in @xcite , as well as in @xcite
let @xmath4 be any path connected topological space . consider the topological category with objects the points of @xmath4 , and morphisms from @xmath39 to @xmath40 given by the moore path space @xmath41 \to
q | \gamma(0 ) = q^0 , \gamma(r ) = q^1 \}\ ] ] where @xmath42 is allowed to vary between @xmath43 and infinity .
the composition law is given by concatenating the domains and the maps : @xmath44 where @xmath45 is assumed to have domain @xmath46 $ ] .
it is well known that this formula defines an associative composition of paths . in order for this operation to induced the structure of a differential graded algebra on chains , we use normalised cubical chains throughout this paper .
recall , for example from @xcite , that a map from a cube to a topological space is said to be degenerate if it factors through the projection to a factor .
the graded abelian groups underlying the normalised chain complex are @xmath47^{i } , x ) ) \right ] } { { { \mathbb z}}\left[\textrm{degenerate maps}\right]}\end{aligned}\ ] ] writing @xmath48 is the inclusion of the face where the @xmath49@xmath50 coordinate is constant and equal to @xmath51 , we define a differential by the formula @xmath52 the key difference with the theory based on simplices is that a product of cubes is again a cube , so it is easy to define a map @xmath53 which may easily be checked to be associative in the appropriate sense . applying this to path spaces , we obtain a differential graded category . in order to be consistent with our sign conventions for the fukaya category ,
we denote by @xmath54 the @xmath2 category with objects points of @xmath4 , morphism spaces @xmath55 and differential and product @xmath56 note that the path - connectivity assumption on @xmath4 implies that all objects of this category are quasi - isomorphic .
in particular , the inclusion of any object defines a fully faithful @xmath2 embedding @xmath57 we shall need to consider an enlargement of @xmath22 to a triangulated @xmath2 category .
the canonical such enlargement is the triangulated closure of the image of @xmath22 in its category of modules under the yoneda embedding . in practice
, it shall be convenient to use the more explicit model of twisted complexes introduced by bondal and kapranov in @xcite .
first , we enlarge @xmath22 by allowing shifts of all objects ( by arbitrary integers ) and define @xmath58 , q^1[m_1 ] ) \equiv { \operatorname{hom}}_{*}(q^0,q^1)[m_1- m_0]\ ] ] with differential @xmath59 .
given a triple @xmath60,q^1[m_1],q^1[m_2])$ ] , and morphisms @xmath61 multiplication is defined via @xmath62 , \sigma_1 [ m_1 -m_0 ] ) = ( -1)^ { ( \deg(\sigma_2)+1)(m_1 -m_0 ) } \mu^{{{\mathrm p}}}_{2 } ( \sigma_2 , \sigma_1 ) [ m_2 - m_0].\ ] ] a _ twisted complex _ consists of the following data @xmath63 , q^j[m_j ] ) $ , such that } \\ \displaystyle { \mu^{{{\mathrm p}}}_{1}\delta_{i , j } + \sum_{k } \mu^{{{\mathrm p}}}_{2 } ( \delta_{k , j } , \delta_{i , k } ) = 0 } \end{array}\ ] ] we write @xmath64 for the matrix of morphisms @xmath65 , and @xmath66 , d)$ ] for such a twisted complex , and note that the equation imposed on @xmath67 can be conveniently encoded as @xmath68 given two such complexes @xmath69 , d^1)$ ] and @xmath70 , d^2)$ ] , we define the space of morphisms between them as a direct sum @xmath71.\ ] ] the differential of an element @xmath72 in this space is given for an elementary matrix by @xmath73 which can be written much more clearly in terms of matrix multiplication @xmath74 composition is also most clearly defined in terms of matrix multiplication @xmath75 the following result is essentially due to bondal and kapranov in @xcite : twisted complexes form a triangulated @xmath2 category .
let @xmath3 be a symplectic manifold equipped with a @xmath76-form @xmath77 whose differential @xmath78 is a symplectic form .
assume the existence of a compact ( codimension @xmath43 ) submanifold @xmath36 with boundary such that the restriction of @xmath77 to @xmath79 is a contact form .
we say that @xmath3 is a liouville manifold if we have a decomposition @xmath80 such that the liouville form is given by @xmath81 on the infinite end @xmath82 , where @xmath83 is the coordinate on @xmath84 . instead of studying all lagrangians in @xmath3
, we consider a _ closed _ exact lagrangian @xmath1 , together with a finite collection @xmath85 of exact properly embedded lagrangians such that @xmath86 the first condition is equivalent to the requirement that the intersection @xmath87 of @xmath24 with @xmath88 be legendrian , and that @xmath24 be obtained by attaching an infinite cylindrical end to the intersection of @xmath24 with @xmath36 @xmath89 we shall choose a primitive @xmath90 for the restriction of @xmath77 to each such lagrangian , which by the above conditions is locally constant away from a compact set , as well as a primitive @xmath91 for the restriction of @xmath77 to @xmath4 .
the exactness condition excludes bubbling of holomorphic discs ; a general lagrangian floer theory is developed by fukaya , oh , ohta and ono in @xcite , but a wrapped version of their theory has not been worked out .
it is also likely that the theory works under weaker assumptions on the properties of @xmath24 at infinity , but these would require more delicate estimates on the behaviour of solutions to the perturbed @xmath92 equations we shall study , which discourages us from pursuing this generalisation . with the condition imposed above , we may define a @xmath93-graded fukaya category over a field of characteristic @xmath19 . to obtain @xmath13-gradings and work over the integers ,
we assume that for each @xmath94 or for @xmath95 , @xmath96 the condition on @xmath97 incorporates the requirement that @xmath3 have a quadratic complex volume form when equipped with any compatible almost complex structure , and that the maslov index on @xmath98 vanish .
this class is well defined because the space of almost complex structures is contractible .
since @xmath24 is lagrangian , we may evaluate such a volume form at every point @xmath99 to obtain an element of @xmath100 .
the vanishing of the maslov index on @xmath101 implies that this function @xmath102 may be factored through the universal cover of @xmath103 .
the choice of such a lifting is called a _ grading _ on @xmath24 ( see section 12 of @xcite ) . on the other hand , the condition that @xmath10 agree , when restricted to @xmath24 , with @xmath104 allows us to choose a relative @xmath0 structure on @xmath4 as well as on the objects of @xmath105 . instead of following the approach using non - abelian cohomology
( see e.g. section ( 11i ) of @xcite and @xcite ) , we shall adopt a variant of the method used in chapter 9 of @xcite for orienting moduli spaces of holomorphic curves .
namely , we fix a triangulation of @xmath3 so that @xmath4 and the elements of @xmath106 are subcomplexes and fix an orientable vector bundle @xmath107 on the @xmath108-skeleton of this triangulation whose second stiefel - whitney class is the restriction of @xmath10 to the @xmath108-skeleton . condition , together with the fact that @xmath107 is orientable , implies that @xmath109 } \oplus e_{b } | l_{[3 ] } ) = 0\ ] ] where @xmath110}$ ] stands for the @xmath108-skeleton of @xmath24 with respect to the triangulation induced as a subcomplex of @xmath3 .
we refer the reader to section ( 11i ) of @xcite for a discussion of the double cover of the orthogonal group called @xmath0 , and the fact that the second stiefel - whitney class is precisely the obstruction to the existence of @xmath0 structures : a relative @xmath0 structure on @xmath24 is the choice of a @xmath0 structure on the vector bundle @xmath111 } \oplus e_{b } | l_{[3 ] } $ ] defined over the @xmath108-skeleton of @xmath24 .
[ def : brane ] a _ brane structure _ on a lagrangian is the choice of a grading , together with a relative @xmath0 structure .
let us fix once and for all such a structure on @xmath4 , as well as on each lagrangian @xmath94 , which is precisely the data needed to unambiguously assign gradings to floer groups and signs to operations thereon .
in the next section , we shall review the construction of a category we call the wrapped fukaya category , whose objects are @xmath112 .
to each object @xmath24 of the wrapped fukaya category , we assign a twisted complex over the pontryagin category of @xmath4 as follows : we write @xmath113 for the set of intersection points between @xmath4 and @xmath24 , which we assume are ordered by action , i.e. the difference between @xmath114 and @xmath115 .
moreover , we write @xmath116 for the space of almost complex structures on @xmath3 which are compatible with the symplectic form on @xmath3 , and such that @xmath117 on the cylindrical end .
this is a mild version of the contact - type property often imposed on almost complex structures on symplectic manifolds with contact boundary . fixing a family @xmath118 of such almost complex structures parametrised by @xmath119 $ ] , we consider the moduli spaces @xmath120 which are the quotients by the @xmath121 action of the space of solutions to the time - dependent @xmath92-equation @xmath122 with boundary conditions @xmath123 \longrightarrow m , \\ & u \left({{\mathbb r}}\times \ { 1 \ } \right ) \subset l , \\ & u \left({{\mathbb r}}\times \ { 0 \ } \right ) \subset q , \\ & \lim_{s \rightarrow - \infty } u(s , t ) = q^j ,
\\ & \lim_{s \rightarrow + \infty } u(s , t ) = q^i \end{aligned } \right.\ ] ] these boundary conditions are conveniently summarised in figure [ fig : strip ] . as the lagrangians @xmath4 and @xmath24 are both graded , we may assign an integer to each intersection point @xmath124 called the degree of @xmath6 and denoted @xmath125 . if we were trying to define the floer cohomology @xmath126 , this would be the degree of the corresponding generator of the floer complex . [
prop : moduli_spaces_strips_manifold ] for a generic family @xmath127 , the gromov compactification of @xmath128 is a topological manifold @xmath129 of dimension @xmath130 , possibly with boundary .
the boundary is stratified into topological manifolds with the closure of codimension @xmath76 strata given by the images of embeddings @xmath131 for all possible integers @xmath132 between @xmath133 and @xmath134 . for each pair @xmath135 of such integers
, we have a commutative diagram @xmath136 \ar[r ] & { \overline{{\mathcal h}}}(q^i , q^{i_1 } ) \times { \overline{{\mathcal h}}}(q^{i_1 } , q^{j } ) \ar[d ] \\ { \overline{{\mathcal h}}}(q^i , q^{i_2 } ) \times { \overline{{\mathcal h}}}(q^{i_2 } , q^{j } ) \ar[r ] & { \overline{{\mathcal h}}}(q^i , q^{j } ) . } \ ] ] in appendix [ app : manifold - with - boundary ] we explain how to prove the result we need using only `` standard results '' . in the appendix to @xcite ,
barraud and cornea provide an alternative construction . in order to define an evaluation map from the moduli space of holomorphic discs to the space of paths on @xmath4
, we must fix parametrisations .
one may inductively make choices of such parametrisations using the contractibility of the space of self - homeomorphisms of the interval .
alternatively , as suggested to the author by janko latschev , one may fix a metric on @xmath4 , and parametrise the boundary segments of a holomorphic disc by arc length .
indeed , it follows from elliptic regularity that each solution to restricts on the appropriate boundary component to a smooth map from @xmath121 to @xmath4 , and from theorem a of @xcite , that its derivatives decay exponentially in the @xmath137 topology at the ends .
in particular , the arc - length parametrisation of such a curve defines a continuous map from @xmath138 $ ] to @xmath4 , with @xmath42 the length of the image . as a sequence of holomorphic strips converge to a broken one , the length parametrisations of the boundary also converge : this follows from the usual proof of gromov compactness by using estimate ( 4.7.13 ) of @xcite which proves that , near the region where breaking occurs , the norm of the derivatives of a family of holomorphic strips decays exponentially , so that the resulting paths on @xmath139 converge to the concatenation of two curves .
this implies that this evaluation map extends continuous to the gromov compactification of the moduli space of discs .
note that the direction of the direction of the path , as indicated in figure [ fig : strip ] , goes from @xmath140 to @xmath141 .
[ lem : evaluate_discs_paths ] there exists a family of evaluation maps @xmath142 such that we have a commutative diagram @xmath143 \ar[r ] & { \overline{{\mathcal h}}}(q^i , q^j)\ar[d ] \\
\omega(q^i , q^k ) \times \omega(q^k , q^j ) \ar[r ] & \omega(q^i , q^j).}\ ] ] in which the top horizontal arrow is the inclusion of equation and the bottom arrow is given by concatenation of paths . before passing to chains , we recall that the choice of brane structure can be used to orient these moduli spaces .
our orientation conventions follow those of @xcite , and are discussed in appendix [ sec : orientations ] and [ sec : signs ] [ lem : product_orientations_strips ] the product orientation on @xmath144 differs from its orientation as a boundary of @xmath129 by a sign given by the parity of @xmath145 .
this result implies that we may choose fundamental chains @xmath146 $ ] in the cubical chain complex @xmath147 inductively .
we start by picking representatives of the fundamental cycles of all components of @xmath148 which are manifolds without boundary , compatibly with the orientations determined by our choices of brane data and the isomorphism of equation .
next , we proceed by induction , assuming that a class satisfying @xmath149
\\ = \sum_{k } ( -1)^ { |q^i|+ |q^k| } [ { \overline{{\mathcal h}}}(q^i , q^k ) ] \times [ { \overline{{\mathcal h}}}(q^k , q^{j'})]\ ] ] has been chosen for all moduli spaces of dimension smaller than some integer @xmath150 . in the induction step
, we note that the associativity of the cross product and the commutativity of the diagram imply that @xmath151 \times [ { \overline{{\mathcal h}}}(q^k , q^j ) ] \right ) = 0.\ ] ] as this is a closed class of codimension @xmath76 supported on the boundary of @xmath129 , we may choose a bounding class @xmath146 $ ] .
[ lem : definition_twisted_complex_lag ] each lagrangian @xmath152 determines a twisted complex @xmath25 in @xmath153 given by @xmath154 , \sum_{q^i , q^j } ( -1)^{|q^i| \left ( |q^j| + 1 \right ) } [ { \overline{{\mathcal h}}}(q^i , q^j ) ] \right).\ ] ] we must verify equation , which takes the form @xmath155 + \sum_{k } ( -1)^ * [ { \overline{{\mathcal h}}}(q^i , q^k ) ] \times [ { \overline{{\mathcal h}}}(q^k , q^j ) ] = 0,\ ] ] where the contributions to the signs in the second term are given by @xmath156 ignoring signs , we see that equation implies the desired result . to verify that the signed formula is correct , the reader should check that the sum of these contributions with @xmath157 is equal to @xmath158 using the formula for the dimension of the moduli spaces given in proposition [ prop : moduli_spaces_strips_manifold ] .
in @xcite , we defined the wrapped fukaya category of a liouville domain @xmath3 . in this paper
, we use a variant which does not use direct limits , following the construction introduced in @xcite in which we use a hamiltonian function growing quadratically at infinity to define floer cohomology groups .
recall that @xmath3 is a manifold equipped with a liouville @xmath76-form @xmath77 .
the liouville vector field @xmath159 defined by the equation @xmath160 agrees with the radial vector field @xmath161 along the cylindrical end of @xmath3 , and we write @xmath162 for the image of the negative liouville flow for time @xmath163 .
note that this flow may be written explicitly on the cylindrical end : @xmath164 let @xmath165 denote the set of smooth functions @xmath166 satisfying @xmath167 away from some compact subset of @xmath3 , and write @xmath168 for those which in addition vanish on @xmath4 .
we shall fix such a function for the purposes of defining floer cohomology , and let @xmath169 denote the hamiltonian flow of @xmath166 defined by the equation @xmath170 for each pair @xmath171 , we define @xmath172 to be the set of time-@xmath76 flow lines of @xmath169 which start on @xmath173 and end on @xmath174 , i.e. an element of @xmath175 is a map @xmath176
\to m $ ] such that @xmath177 we shall assume that @xmath178 it is convenient to remember that the elements of @xmath172 are in bijective correspondence with intersection points between @xmath179 and the image of @xmath173 under the time-@xmath76 hamiltonian flow of @xmath166 .
moreover , those chords which lie in the complement of @xmath88 are in bijective correspondence with reeb chords with endpoints on the @xmath180 and @xmath181 which , by condition , are legendrian submanifolds of @xmath88 .
in particular , non - degeneracy of chords lying in @xmath36 corresponds to transversality between @xmath179 and the image of @xmath173 under the time-@xmath76 hamiltonian flow of @xmath166 , while non - degeneracy in the cylindrical end corresponds to non - degeneracy of all reeb chords .
condition therefore holds after generic hamiltonian perturbations of the lagrangians in @xmath85 , and one may moreover assume that the perturbation preserves condition .
we shall therefore replace any collection @xmath182 by hamiltonian isotopic ones which satisfy condition . as lagrangian floer cohomology is invariant under hamiltonian perturbations , this results in no loss of generality .
to each element @xmath183 , we assign a maslov index @xmath184 coming from the gradings on @xmath173 and @xmath179 .
this grading agrees with the maslov index of the intersection of @xmath179 with the image of @xmath173 under the time-@xmath76 hamiltonian flow of @xmath166 which is defined , for example , in section ( 11h ) of @xcite .
we define the action of an @xmath169-chord starting on @xmath185 and ending on @xmath186 by the familiar formula ( note that our conventions on action differ by a sign from those of @xcite ) @xmath187 equation implies that any chord that intersects some slice @xmath188 is contained therein , and has action @xmath189 where @xmath190 and @xmath191 are the values of @xmath192 and @xmath193 on the ends of @xmath185 and @xmath186 .
[ lem : action_proper ] @xmath194 is a proper map from @xmath195 to @xmath121 .
non - degeneracy implies that there are only finitely many chords in any compact subset of @xmath3 , while equation implies that the action of a sequence of chords which escapes every compact set must go to @xmath196 .
let us fix once and for all a smooth map @xmath197 \to [ 0,1]\ ] ] such that @xmath198 is identically @xmath43 ( respectively @xmath76 ) in a neighbourhood of the appropriate endpoint . given a pair @xmath199 we define @xmath200 to be the moduli space of maps @xmath201 satisfying the following boundary and asymptotic conditions @xmath202 as solving floer s equation @xmath203 we shall write this equation in a coordinate free way as @xmath204 since equation is invariant under translation in the @xmath205-variable , the reals act on @xmath200 .
the analogue of theorem [ prop : moduli_spaces_strips_manifold ] holds as well [ lem : free_r_action_strips ] the moduli space @xmath200 is regular for a generic choice of almost complex structures @xmath206 and has dimension @xmath207 .
we write @xmath208 for the quotient of @xmath200 by the @xmath121 action whenever it is free , and declare it to be the empty set otherwise . by adding _ broken strips _ to this moduli space , we obtain a manifold with boundary @xmath209 whose strata are disjoint unions over all sequences starting with @xmath210 and ending with @xmath211 of the products of the moduli spaces @xmath212 since the lagrangians @xmath173 and @xmath179 have infinite ends , it does not immediately follow from gromov compactness that @xmath213 is compact .
[ lem : gromov_compactness ] if a solution to equation converges at @xmath196 to @xmath210 and at @xmath214 to @xmath211 , then @xmath215 moreover , all such solutions lie entirely within a compact subset of @xmath3 depending only on @xmath210 and @xmath211 .
the first part is a standard energy estimate using the positivity of @xmath166 . to prove the second
, we appeal to the argument given in lemma 7.2 of @xcite : consider any hypersurface @xmath216 separating @xmath210 and @xmath211 from infinity .
if a solution to equation escaped this region , we would find a compact surface @xmath217 , mapping to the cone @xmath218 and with boundary conditions the concave end @xmath219 and a collection of lagrangians on which @xmath77 vanishes identically .
applying stokes theorem , and using the fact that @xmath220 achieves its minimum on the boundary , we find that @xmath221 where @xmath222 is the inverse image of @xmath216 .
since the restriction of @xmath169 to @xmath219 is a multiple of the reeb flow , @xmath223 vanishes , so we conclude that @xmath224 on the other hand , if @xmath225 is a tangent vector compatible with the natural orientation of @xmath222 , @xmath226 is inward pointing along @xmath227 , so that @xmath228 points towards the infinite part of the cone .
the condition that @xmath229 be of contact type implies that @xmath230 yielding a contradiction .
together with lemma [ lem : action_proper ] , this result implies that for each chord @xmath211 , there is a compact subset of @xmath3 containing the union of all the moduli spaces @xmath231 . applying the usual gromov compactness and gluing theory results to this moduli space , we conclude : [ cor : compactification_strip_manifold ] for each chord @xmath211 , the moduli space @xmath232 is empty for all but finitely many choices of @xmath210 , and is a compact manifold with boundary of dimension @xmath233 whenever @xmath206 is a generic family of almost complex structure .
moreover , the boundary is covered by the closure of the images of the natural inclusions @xmath234 from now on , we shall fix an almost complex structure @xmath206 for which the conclusion of this corollary holds .
we define the graded vector space underlying the floer complex to be the direct sum @xmath235 here , @xmath236 is a certain real vector space of rank @xmath76 associated to every chord via a construction briefly reviewed in appendix [ sec : orientations ] , and @xmath237 is generated by the two possible orientations on it .
unless the reader wants to check the signs , there is no harm in assuming that we are simply taking the vector space freely generated by the set of chords .
the differential is then a count of solutions to equation .
more precisely , whenever @xmath238 , every element @xmath239 of the moduli space @xmath240 is rigid , and defines an isomorphism @xmath241 as defined in section [ sec : signs ] .
in particular , we may induce an orientation of @xmath242 from one on @xmath243 , and the associated map on orientation lines is denoted @xmath244 .
we define @xmath245 & \mapsto ( -1)^i \sum_{u } \mu_u([x_1 ] ) .\end{aligned}\ ] ] ignoring signs , we are indeed simply counting elements of the moduli spaces @xmath246 .
the proof that this count gives a well - defined differential is standard .
the only possibly new phenomenon is that corollary [ cor : compactification_strip_manifold ] implies that each chord can be the input of only finitely many solutions to , which guarantees that the image of the corresponding generator is a sum of only finitely many terms .
let us note that the graded vector space underlying @xmath247 depends on @xmath173 , @xmath179 , @xmath166 and @xmath78 , while the differential also depends on @xmath127 .
we write @xmath248 when the distinction is important as shall be the case in the next result . if @xmath249 satisfies @xmath250 for some non - zero constant @xmath251
, then we have a canonical isomorphism @xmath252 for any diffeomorphism @xmath253 , we obtain an isomorphism of chain complexes @xmath254 by composing every chord from @xmath173 and @xmath179 with @xmath253 to obtain a chord from @xmath255 to @xmath256 , and every solution to equation with @xmath253 to obtain an analogous solution for the family of almost complex structure @xmath257 .
the property that @xmath253 rescale @xmath78 implies that the hamiltonian flow of @xmath258 with respect to @xmath259 agrees with the flow of @xmath260 with respect to @xmath78 .
since only the hamiltonian flow appears in equation , we obtain an identification @xmath261 which proves the desired result . from now on , we define @xmath262 given a triple of lagrangians @xmath173 , @xmath179 and @xmath263 , we shall define a map @xmath264 where @xmath265 is the time @xmath266 negative liouville flow .
after composition with the inverse of the isomorphism of equation , we obtain the product @xmath267 in the wrapped fukaya category .
the map shall count solutions to a cauchy - riemann equation @xmath268 whose source is the surface @xmath269 obtained by removing @xmath108 points @xmath270 from the boundary of @xmath271 ( see figure [ fig : disc_2_inputs ] ) . in the above equation , @xmath272 is a closed @xmath76 form on @xmath269 while @xmath273 is the hamiltonian vector field of a function @xmath274 on @xmath3 which depends on @xmath269
. the most important condition on these data is the requirement that @xmath274 come from a map @xmath275 and restrict to @xmath166 near @xmath276 and @xmath277 and to @xmath278 near @xmath279 . to see that this makes sense
, we note that @xmath278 indeed lies in @xmath280 since @xmath166 agrees with @xmath281 on the cylindrical end .
to specify the remaining data required for equation , we choose strip - like ends for @xmath269 , i.e. we write @xmath282 and @xmath283 for the positive and negative half - strips in @xmath284 and choose embeddings @xmath285 which map @xmath286 to @xmath227 and converge to the respective marked points @xmath287 .
the closed @xmath76-form @xmath272 is required to vanish on @xmath227 , and to satisfy @xmath288 in addition , we choose a family of almost complex structures @xmath289 whose compositions with @xmath290 agrees with @xmath127 if @xmath291 , and with @xmath292 if @xmath293 .
we would like the lagrangian boundary conditions to be given by @xmath294 near @xmath295 , @xmath296 near @xmath277 , and @xmath297 near @xmath279 .
along the two segments of @xmath298 converging to @xmath279 , we can not therefore have a constant lagrangian condition , but we must interpolate between a lagrangian and its image under @xmath299 . technically , this puts us in the framework of _ moving lagrangian boundary conditions _ ( see section ( 8k ) of @xcite ) .
we shall choose the simplest such moving boundary condition by fixing a map @xmath300 from the boundary of @xmath271 to the interval @xmath301 $ ] such that @xmath302 the moduli space @xmath303 is the space of solutions to equation with boundary conditions @xmath304 and such that the image of @xmath239 converges to @xmath305 and @xmath306 at the corresponding incoming strip - like ends , and to @xmath307 at the outgoing strip - like end . by construction
, we have ensured that the pullback of equation by the strip - like end @xmath287 is given by @xmath308 with lagrangian boundary conditions @xmath294 near @xmath295 , @xmath296 near @xmath277 , and @xmath309 near @xmath279 .
for the inputs , we exactly recover equation , while for the output , we recover the same equation , but for the hamiltonian @xmath310 .
we conclude that the gromov bordification of the moduli space @xmath303 is obtained by adding the strata @xmath311 corresponding to the breaking of a holomorphic strip at any of the ends . the factor @xmath312 in the third stratum is obtained by applying the inverse of @xmath265 to the moduli space @xmath313 which would appear naturally from the point of view of breakings of holomorphic curves .
[ lem : regularity_compactness_2_inputs ] for a fixed pair @xmath314 , the moduli space @xmath315 is empty for all but finitely many choices of chords @xmath210 . for a generic family of almost complex structures @xmath316 and hamiltonians @xmath274 , @xmath317 is a compact manifold of dimension @xmath318 whose boundary is covered by the codimension @xmath76 strata listed in equations - .
the proof of transversality follows from a standard sard - smale argument going back to @xcite in the case of hamiltonian floer cohomology .
the proof of compactness relies an argument analogous to that of lemma [ lem : gromov_compactness ] , and which is explained in the proof of lemma 3.2 of @xcite . whenever @xmath319 , @xmath320 consists of finitely many elements , each of which defines an isomorphism @xmath321 coming from equation .
writing @xmath244 as before for the map induced on orientation lines we define a product on the wrapped floer complex by adding the appropriate signed contribution of each element of @xmath322 : @xmath323,[x_1 ] ) & = \sum _ { \substack { |x_0| = |x_1| + |x_2| \\ u \in { { \mathcal r}}_{2 } ( x_0;x_1,x_2 ) } } ( -1)^{|x_1| } \mu_u([x_2],[x_1]).\end{aligned}\ ] ] let us assign to each end @xmath324 of a punctured disc a _
@xmath325 which is a positive real number , with the convention that any @xmath92 operator we shall consider on such a disc must pull back , under a strip - like end near @xmath324 , to equation up to applying @xmath326 . in the case of a disc with @xmath19 incoming ends ,
the inputs have weights @xmath76 while the output has weight @xmath19 .
if we consider the pull back by @xmath299 of all the data used to define the moduli spaces @xmath327 in the previous section , we obtain a @xmath92 operator such that the weights are now equal to @xmath19 at the inputs , and @xmath328 at the output .
the space of solutions of the associated cauchy - riemann equation is a moduli space @xmath329 the count of whose elements defines a map @xmath330 which exactly agrees with @xmath331 if we pre - compose with @xmath332 on both factors the source and post - compose with @xmath333 . in the next few sections
, we shall prove that @xmath331 induces an associative product on cohomology by constructing a homotopy between the two possible compositions around the diagram @xmath334^{cw(\psi^{2 } ) \otimes\mu^{\psi^2}_{2 } } \ar[r]^ { \mu^{\psi^2}_{2 } \otimes cw(\psi^{2 } ) } & cw^{*}_{b}(\psi^{2 } ( l_1 ) , \psi^{2 } ( l_3 ) ) \otimes cw^{*}_{b}(\psi^{2 } ( l_0 ) , \psi^{2 } ( l_1 ) ) \ar[d]^ { \mu^{\psi^4}_{2 } } \\
cw^{*}_{b}(\psi^{2 } ( l_2 ) , \psi^{2 } ( l_3 ) ) \otimes cw^{*}_{b}(\psi^{2 } ( l_0 ) , \psi^{2 } ( l_2 ) ) \ar[r]^ { \mu^{\psi^4}_{2 } } & cw^{*}_{b}(\psi^{4 } ( l_0 ) , \psi^{4 } ( l_3 ) ) \ar[d]^{\cong}. \\ & cw^{*}_{b}(l_0 , l_3 ) } \ ] ] it is well known that the product @xmath335 is not in general associative at the chain level , and that such a homotopy should come from a moduli space of maps whose sources are @xmath328-punctured discs equipped with an arbitrary conformal structure .
we write @xmath336 for this moduli space , and recall that @xmath337 is an interval whose two endpoints are nodal discs obtained by gluing , in the two possible different ways as represented in the outermost surfaces in figure [ fig : three_puncture ] , two discs each with two incoming ends and one outgoing one . in order for this count to indeed define a homotopy , whatever equation we define on the moduli space @xmath337 is usually required to restrict on the boundary strata to equation on each component .
in particular , the cauchy - riemann equation imposed on a disc whose conformal equivalence class is close to the boundary of @xmath337 should be obtained by gluing the @xmath92 operators associated to equation at the node .
this only makes sense if the restriction of these @xmath92 operators to the two strip - like ends at the node agree , which is not the case for us , as even the lagrangian boundary conditions do not agree .
however , they agree up to applying @xmath338 , so we may glue two solutions to equation after applying @xmath338 to one of them .
this is the main reason for stating associativity in term of the diagram . as a final observation
, we note that the weights on the inputs of a @xmath108-punctured disc must be allowed to vary with its modulus , for they are given by @xmath339 at one of the endpoints of @xmath337 and @xmath340 at the other .
we could now choose auxiliary data of families of hamiltonians and almost complex structures on @xmath3 , and of @xmath76-forms on elements of @xmath337 which would define an operation @xmath341 providing the homotopy in diagram . as we shall have to repeat this procedure for an arbitrary number of inputs , and there is no simplifying feature in having only three , we proceed to give the general construction .
we write @xmath342 for the moduli space of abstract discs with one negative puncture denoted @xmath343 and @xmath344 positive punctures denoted @xmath345 which are ordered clockwise .
we write @xmath346 for the deligne - mumford compactification of this moduli space , and assume that a _
universal and consistent _ choice of strip - like end has been chosen as in section ( 9 g ) of @xcite .
this means that we have , for each surface @xmath269 and each puncture , a map @xmath347 whose source is @xmath283 if @xmath293 and @xmath282 otherwise , and that such a choice varies smoothly with the modulus of the surface @xmath269 in the interior of the moduli space
. moreover , near each boundary stratum @xmath348 of @xmath346 the strip - like ends are obtained by gluing in the following sense : if we write @xmath349 , then gluing the strip - like ends chosen on these lower dimensional moduli spaces defines an embedding @xmath350 a surface in the image of this gluing map is by definition covered by patches coming from surfaces in one of the factors of the product decomposition of @xmath348 .
in particular , each end of such a surface obtained by gluing comes equipped with strip - like ends induced from the choices of strip - like ends on the lower dimensional moduli spaces .
consistency of the choice of strip - like ends is the requirement that our choice on @xmath351 agree with this fixed choice in some neighbourhood of @xmath348 .
[ def : floer_datum_disc_1_output ] a _ floer datum _
@xmath352 on a stable disc @xmath353 consists of the following choices : 1 .
weights : a positive integer @xmath354 assigned to the @xmath49@xmath50 end such that @xmath355 2 . moving conditions : a map @xmath356 which agrees with @xmath357 near the @xmath49@xmath50 end .
basic @xmath76-form : a closed @xmath76-form @xmath358 whose restriction to the boundary vanishes and whose pullback under @xmath359 agrees with @xmath360 .
hamiltonian perturbations : a map @xmath361 which agrees with @xmath362 near the @xmath49@xmath50 end . 5 .
almost complex structure : a map @xmath363 whose pullback under @xmath364 agrees with @xmath365 . if we write @xmath366 for the hamiltonian flow of @xmath274 , then these data allow us to write down a cauchy - riemann equation @xmath367 where the @xmath368 part is taken with respect to @xmath369 .
the main reason of the long list of conditions in definition [ def : floer_datum_disc_1_output ] is the following conclusion [ lem : equation_restricts_to_ends ] the pullback of equation under @xmath370 is given by @xmath371 in particular , it agrees with equation up to applying @xmath372 . at the boundary of the moduli space @xmath373
, we should require that floer data be given by the choices performed on smaller dimensional moduli spaces .
when @xmath374 , we already noted in the previous section that the choice of floer data at the boundary can not be given exactly by the floer data for @xmath375 on each component , since these floer data can not be glued .
we shall consider the following notion of equivalence among floer data which is weaker than equality : we say that a pair @xmath376 and @xmath377 of floer data on a surface @xmath269 are _ conformally equivalent _ if there exists a constant @xmath378 so that @xmath379 and @xmath380 respectively agree with @xmath381 and @xmath382 , and @xmath383 we have already encountered the idea that rescaling by @xmath265 in equation gives an identification of moduli spaces .
this idea generalises as follows : composition with @xmath384 defines a bijective correspondence between solutions to equation for conformally equivalent floer data .
we can now state the desired compatibility between floer data at the boundary of the moduli spaces @xmath385 : a _ universal and conformally consistent _ choice of floer data @xmath386 for the @xmath2 structure , is a choice of floer data for every element of @xmath385 and every integer @xmath387 , which varies smoothly over the interior of the moduli space , whose restriction to a boundary stratum is conformally equivalent to the product of floer data coming from lower dimensional moduli spaces , and which near such a boundary stratum agrees to infinite order , in the coordinates , with the floer data obtained by gluing .
note that conformal consistency fixes , up to a constant depending on the modulus , the choice of floer data on @xmath388 once floer data on each irreducible component has been chosen .
for example , the choice of floer data for @xmath389 from section [ sec : comp - wrapp - fukaya ] determines the floer data on the boundary of @xmath390 as discussed in the previous section , where we chose to use the original floer data on the disc that does not contain the original end , and rescale it by @xmath265 on the other disc ( see figure [ fig : three_puncture ] ) . after introducing a perturbations of this floer data which vanishes to infinite order at the boundary , we extend it from a neighbourhood of the boundary to the remainder of @xmath336 .
one may proceed inductively to construct a universal datum @xmath391 using the fact that the boundary of @xmath385 is covered by the images of codimension @xmath76 inclusions @xmath392 and the contractibility of the space of floer data on a given surface .
the contractibility of this space also implies that we may extend any floer data chosen on a given surface to universal data : [ lem : abundance_floer_data ] the restriction map from the space of universal and conformally consistent floer data to the space of floer data for a fixed surface @xmath269 is surjective . in this section ,
we construct the @xmath2 structure on the wrapped fukaya category which is given by higher products @xmath393 coming from the count of certain solutions to equation . given a sequence of chords @xmath394 if @xmath395 and @xmath396 with @xmath397 objects of @xmath105 , we define @xmath398 to be the space of solutions to equation whose source is an arbitrary element @xmath399 with marked points @xmath400 , such that @xmath401 and with boundary conditions @xmath402 note that these conditions make sense because of lemma [ lem : equation_restricts_to_ends ] which shows that equation restricts to equation up to applying @xmath403 .
figure [ fig : boundary_conditions ] shows the asymptotic conditions for @xmath374 . with the exception of strips breaking at the ends , the virtual codimension @xmath76 strata of the gromov bordification @xmath404 lie over the codimension @xmath76 strata of @xmath385 .
the consistency condition imposed on @xmath386 implies that whenever a disc breaks , each component is a solution to equation for the floer data @xmath386 up to applying @xmath384 for some constant @xmath378 that depends on the modulus in @xmath385 .
since composition with @xmath384 identifies the solutions of this rescaled equation with the moduli space of the original equation , we conclude that for each integer @xmath49 between @xmath43 and @xmath405 and chord @xmath406 , we obtain a natural inclusion @xmath407 where the sequences of inputs in the respective factors are given by @xmath408 and @xmath409 as in figure [ fig : breaking ] .
applying the same arguments as would prove lemma [ lem : regularity_compactness_2_inputs ] , we conclude [ lem : boundary_moduli_spaces_a_infty ] the moduli spaces @xmath404 are compact , and are empty for all but finitely many @xmath210 once the inputs @xmath410 are fixed .
for a generic choice @xmath386 , they form manifolds of dimension @xmath411 whose boundary is covered by the images of the inclusions . whenever @xmath412 , there are therefore only finitely many elements of @xmath398 . via the procedure described in appendix
[ sec : signs ] , every such element @xmath413 induces an isomorphism @xmath414 we write @xmath415 for the induced map on orientation lines , and omitting composition with @xmath416 or its inverse from the notation , we define the @xmath344@xmath50 higher product @xmath393 as a sum @xmath417 , \ldots , [ x_1 ] ) = \sum_{\substack{|x_0| = 2 - d + \sum_{1 \leq k \leq d } |x_k| \\ u \in { { \mathcal r}}_{d}(x_0 ; { \vec{x}^{\ , \ ! } } ) } } ( -1)^{\dagger } \mu_{u}([x_d ] , \ldots , [ x_1])\ ] ] with sign given by the formula @xmath418 this is simply a complicated way of saying that we count the elements of @xmath419 with appropriate signs .
if we now consider the @xmath76 dimensional moduli spaces , lemma [ lem : boundary_moduli_spaces_a_infty ] asserts that their boundaries are given by the strata appearing in equation which are rigid and therefore correspond to the composition of operations @xmath420 .
if we take signs into account , we conclude : [ prop : a_infty_structure ] the operations @xmath420 define an @xmath2 structure on the category @xmath105
. in particular , the sum @xmath421 vanishes , where the value of the sign is given by @xmath422
in section [ sec : at - level - objects ] , we assigned to each lagrangian in @xmath105 an object @xmath25 in @xmath153 .
we shall now extend this assignment to an @xmath2 functor , and the first task is to construct a chain map @xmath423 on the half strip @xmath424 $ ] , let @xmath425 and @xmath426 , and consider the surface @xmath427 drawn in a non - standard way in figure [ fig : half - disc ] .
if we think of @xmath428 as the complement of @xmath108 marked point in a disc , we write @xmath295 for the puncture at infinity ; and call the segment between @xmath429 and @xmath430 outgoing . note that @xmath428 is biholomorphic to the surface @xmath269 shown in figure [ fig : disc_2_inputs ] , but we choose to use a different notation because maps from @xmath428 to @xmath3 will satisfy not satisfy the same cauchy - riemann equation we imposed on @xmath269 .
the most important difference is that we shall consider a map @xmath431 which near @xmath295 agrees with @xmath166 , and whose restriction to the interval between @xmath279 and @xmath432 takes values in @xmath168 ( i.e. vanishes on @xmath4 ) .
writing @xmath433 for the hamiltonian flow of @xmath434 , we shall consider the cauchy - riemann equation @xmath435 for maps @xmath436 . to specify the choice of a family of almost complex structure with respect to which we take the @xmath368 part , let us fix a positive strip - like end @xmath437 near @xmath279 and a negative end @xmath438 near @xmath432 , and choose a map @xmath439 which agrees with @xmath206 near @xmath295 , and whose pullback under @xmath437 and @xmath440 also agrees with @xmath206 . [ lem : restriction_half_disc_equation_ends ] the restriction of equation agrees with equation near @xmath276 , and its pullback under @xmath437 and @xmath440 agrees with equation
. given a chord @xmath441 , and intersection points @xmath442 ( respectively @xmath443 ) between @xmath173 ( or @xmath179 ) and @xmath4 , it therefore makes sense to define @xmath444 ( see figure [ fig : half - disc ] ) to be the moduli space of solutions @xmath239 to equation with boundary conditions @xmath445 and asymptotic conditions @xmath446 from the standard transversality results in floer theory we conclude : for generic choices of hamiltonians @xmath434 and almost complex structure @xmath447 , the moduli space @xmath448 is a smooth manifold of dimension @xmath449 we assume that a choice of @xmath450 and @xmath447 is now fixed , and proceed to analyse the boundary of the gromov compactification .
since @xmath428 is topologically a @xmath108-punctured discs , there is no modulus , and the only strata that we need to add to the gromov compactification are obtained by considering breakings of strips at the three ends . by lemma [ lem :
restriction_half_disc_equation_ends ] , the virtual codimension @xmath76 strata ( see figure [ fig : breaking_half_discs ] ) are @xmath451 the standard compactness result implies the gromov compactification @xmath452 is a compact manifold whose boundary is covered by the images of the strata - .
in addition , we shall need to compare the product orientations on the strata of @xmath453 .
the proof is delayed until appendix [ sec : signs ] .
[ lem : orientaiton_error_moduli_half_disc_1_puncture ] the difference between the orientation induced as a boundary of @xmath454 and the product orientation is given by signs whose parity is @xmath455 consider the evaluation map @xmath456 which takes every half disc @xmath239 to the path along @xmath4 between @xmath442 and @xmath457 obtained by restricting @xmath239 to the outgoing segment . as in section
[ sec : at - level - objects ] , we shall use the length parametrisation in order to remove any ambiguity .
in particular , if we consider the boundary strata , we obtain a commutative diagram @xmath458 \ar[d]^{{\operatorname{ev } } } & { \overline{{\mathcal h } } } ( q_0 , x , q_1 ) \ar[d]^{{\operatorname{ev } } } \\ { \mathcal h } ( q_0 , x_0 , q_1 ) \ar[r ] & \omega(q_0,q_1 ) & } \ ] ]
in which the left vertical arrow is projection to the first factor and the top map is the inclusion of a boundary stratum . while if we consider the strata and , we obtain diagrams @xmath459 \ar[d ] & { \overline{{\mathcal h } } } ( q_0 , x , q_1 ) \ar[d ] & \ar[l ] \ar[d ] { \mathcal h}(q_0 , q'_0 ) \times { \mathcal h } ( q'_0 , x , q_1 ) \\
\omega(q_0,q'_1 ) \times \omega(q'_1,q_1 ) \ar[r ] & \omega(q_0,q_1 ) & \ar[l ] \omega(q_0 , q'_0 ) \times \omega ( q'_0 , q_1 ) . }
\ ] ] where the arrows in the bottom row are obtained by concatenation .
[ lem : fundamental_chains_linear_map ] there exist fundamental chains @xmath460 \in c _ { * } ( { \overline{{\mathcal h}}}(q_0 , x , q_1 ) ) $ ] such that the assignment @xmath461 ) = \bigoplus_{q_0 , q_1 } ( -1)^{|x| + ( |q_0|+1)(|x| + |q_1| ) } { \operatorname{ev } } _ { * } ( [ { \overline{{\mathcal h}}}(q_0 , x , q_1)])\ ] ] defines the chain map described in equation . more precisely , the choice of fundamental chain depends up to sign on the element @xmath462 $ ] . as explained in appendix [ sec : orientations ] , the moduli space @xmath463 is oriented _ relative _ the line @xmath464 , which means that an orientation of @xmath464 induces an orientation of the moduli space , and the wrapped floer complex is precisely generated by such choices of orientations . the strategy is to start with the fundamental chains chosen for the moduli spaces @xmath465 in section [ sec : at - level - objects ] , and to follow the procedure described in that section to choose fundamental chains for the moduli spaces @xmath466 .
next , we choose fundamental cycles for those moduli spaces @xmath467 which do not have boundary compatibly with orientations .
let us consider moduli spaces
@xmath467 whose boundary only has codimension @xmath76 strata , which must therefore be products of closed manifolds .
by taking the product of the fundamental chains of the factors in each boundary stratum , we obtain a chain in @xmath468 @xmath469 \times [ { { \mathcal r}}(x_0,x ) ] + \sum_{q'_1 \in l_1 \cap q } [ { \mathcal h}(q_0 , q'_0 ) ] \times [ { \mathcal h } ( q'_0 , x , q_1 ) ] \\ + \sum_{q'_0 \in l_0 \cap q } ( -1)^ { 1 + |q_0| + |x| + | q'_1| } [ { \mathcal h } ( q_0 , x , q'_1 ) ] \times [ { \mathcal h}(q'_1 , q_1)]\end{gathered}\ ] ] where the sign in the first summation is @xmath470 if @xmath471 is rigid , and will not enter in any of our constructions otherwise .
lemma [ eq : half_disc_right_break ] implies that equation is a cycle .
we now define the fundamental chain @xmath472\ ] ] to be any chain whose boundary is equation .
consider one of the strata of @xmath473 listed in equation , for which the moduli space @xmath474 is not rigid . since the fundamental chain of this stratum is obtained by taking the products of the fundamental chains on each factor , its image in @xmath475 is degenerate , and hence vanishes .
we conclude that such strata do not contribute to @xmath476 ) $ ]
. consulting equation and the definition of @xmath25 in lemma [ lem : definition_twisted_complex_lag ] , we find that , up to signs which we shall ignore , proving that @xmath477 is a chain map is equivalent to proving that @xmath478 ) ) + \mu_{2}^{{{\mathrm p}}}\left({\mathcal f}^{1}[x ] , \sum_{q_{0 } , q'_{0 } } [ { \overline{{\mathcal h}}}(q_{0 } , q'_{0 } ) ] \right ) + \mu_{2}^{{{\mathrm p}}}\left ( \sum_{q_{1 } , q'_{1 } } [ { \overline{{\mathcal h}}}(q'_{1 } , q_{1 } ) ] , { \mathcal f}^{1 } [ x ] \right ) \\ = { \mathcal f}^{1}(\mu^{{{\mathrm f}}}_1 [ x]).\end{gathered}\ ] ] since it follows essentially by definition that @xmath479 ) = \mu_{1}^{{{\mathrm p } } } ( { \mathcal f}^{1}([x ] ) ) , \ ] ] we shall now interpret @xmath480 ) $ ] differently to account for the remaining terms in the equation for a chain map . going through the stratification of @xmath481 , we find that the stratum for rigid moduli spaces correspond to @xmath482)$ ] , and that the strata and respectively correspond to the second and third term in the left hand side of equation
. having constructed the chain map @xmath477 , we would need to check that it respects the product structure which on the source is given by the count of pairs of pants , and on the target is given by concatenation of paths .
as with the failure of associativity on the pair of pants product in the fukaya category , it is only the map induced by @xmath477 on cohomology that respects the product structure . at the chain level , there is a homotopy between the @xmath483 , [ x_1 ] ) ) $ ] and @xmath484 , { \mathcal f}^{1 } [ x_1 ] ) $ ] ; these two compositions are represented by the two outermost diagram in figure [ fig : breaking_half_disc_two_inputs ] .
we shall therefore have to introduce a moduli space @xmath485 which is @xmath76-dimensional , and whose boundary is represented by the two broken curves in figure [ fig : breaking_half_disc_two_inputs ] .
as in the proof of the homotopy associativity of the product in the fukaya category , we shall define a family of cauchy - riemann equations on this abstract moduli space , interpolating between the equations on the two broken curves , and the moduli space of solutions to this equation , with appropriate boundary conditions , shall define the desired homotopy .
since the notion of an @xmath2 functor requires the construction of a tower of such homotopies , we proceed with describing the construction in the general case : let us write @xmath486 for the moduli space of holomorphic discs with @xmath487 marked points , of which @xmath344 successive ones are distinguished as incoming ; the remaining marked points and the segment connecting them are called outgoing .
we identify @xmath488 with a point ( equipped with a group of automorphisms isomorphic to @xmath121 ) corresponding to the moduli space of strips .
in addition , we fix an orientation on the moduli space @xmath486 using the conventions for stasheff polyhedra in @xcite , and the isomorphism @xmath489 taking the incoming marked points on the source to the first @xmath344 incoming marked point on the target .
in particular , the usual deligne - mumford compactification of @xmath490 yields a compactification of @xmath491 by adding broken curves .
it shall be important , however , to understand the operadic meaning of the boundary strata of @xmath492 .
if breaking takes place away from the outgoing segment , the topological type of the broken curve is determined by sequences @xmath493 and @xmath494 , such that @xmath495 , and a fixed element @xmath49 in the first sequence .
these data determine a map @xmath496 and the simplest such curve is shown in the right of figure [ fig : breaking_half_disc_two_inputs ] .
if breaking occurs on the outgoing segment , it is determined by a partition @xmath497 .
letting @xmath498 , we obtain a map
@xmath499 the following definition is the direct extension of definition [ def : floer_datum_disc_1_output ] to the moduli space of half - discs . note that a universal choice of strip - like ends on the moduli spaces @xmath500 induces one on @xmath501 via the identification of equation : [ def : floer_datum_half_popsicle ] a _ floer datum _
@xmath502 on a stable disc @xmath503 consists of the following choices on each component : 1 .
weights : a positive real number @xmath504 associated to each end of @xmath428 which is assumed for be @xmath76 for @xmath505 .
2 . time shifting maps : a map @xmath506 which agrees with @xmath504 near @xmath507 .
hamiltonian perturbation : a map @xmath508 on each surface such that the restriction of @xmath509 to a neighbourhood of the outgoing boundary segment takes value in @xmath510 , and whose value near the @xmath287 for @xmath511 is @xmath512 4 .
basic @xmath76-form : a closed @xmath76-form @xmath513 whose restriction to the complement of the outgoing segment in @xmath514 and to a neighbourhood of @xmath279 and @xmath432 vanishes , and whose pullback under @xmath364 for @xmath511 agrees with @xmath515 . 5 .
almost complex structure : a map @xmath516 whose pullback under @xmath364 agrees with @xmath517 . note that the closedness of @xmath513 and the fact that it vanishes near the outgoing marked points implies that it does not vanish on the outgoing boundary segment . were it not for the fact that the outgoing segment is mapped to a _
lagrangian , this would cause a problem with compactness . as before , we write @xmath518 for the hamiltonian flow of @xmath434 and consider the differential equation @xmath519 with respect to the @xmath428-dependent almost complex structure @xmath520 .
the pullback of equation under the end @xmath370 agrees with equation if @xmath521 , and otherwise with @xmath522 in order for the count of solutions to equation to define the desired homotopies , we must choose its restriction to the boundary strata of the moduli spaces in a compatible way . in the case where @xmath389 , the moduli space is an interval , whose two endpoints may be identified with the products @xmath523 and @xmath524 ( see figure [ fig : breaking_half_disc_two_inputs ] ) .
our discussions in sections [ sec : comp - wrapp - fukaya ] and [ sec : moduli - spaces - half ] fix floer data respectively on the unique elements of @xmath525 and @xmath526 . in the case of @xmath527
the equations on both sides of the node agree , in the coordinates coming form strip - like ends , with equation , so we may glue the corresponding floer data .
on the other hand , the equations on the two sides of the node of the broken curve representing an element of @xmath524 only agree up to applying the liouville flow , since the weight on the input is @xmath76 while the weight of the output is @xmath19 .
our conventions that the weights @xmath528 and @xmath529 are always equal to @xmath76 imply that we must apply the liouville flow @xmath530 to the equation imposed on @xmath531 in order for gluing to make sense . in this way , we obtain floer data in a neighbourhood of @xmath532 , which may be extended to the interior of the moduli space . having fixed this choice , we proceed inductively for the rest of the moduli spaces : a _ universal and conformally consistent _ choice of floer data for the homomorphism @xmath34 , is a choice @xmath533 of such floer data for every integer @xmath534 , and every ( representative ) of an element of @xmath535 which varies smoothly over this compactified moduli space , whose restriction to a boundary stratum is conformally equivalent to the product of floer data coming from either @xmath386 or a lower dimensional moduli space @xmath535 , and which near such a boundary stratum agrees to infinite order with the floer data obtained by gluing .
the consistency condition implies that each irreducible component of a curve representing a point in the stratum carries the restriction of the datum @xmath391if it comes from the factor @xmath536 , and the restriction of the datum @xmath533 if it comes from @xmath537 , up to conformal equivalence which is fixed near by the requirement that @xmath538 and @xmath529 are both equal to @xmath76 .
given a sequence of chords @xmath394 if @xmath395 , and a pair of intersection points @xmath539 and @xmath540 , we define @xmath541 to be the moduli space of solutions to equation whose source is an arbitrary element @xmath542 , with boundary conditions @xmath543 and with asymptotic conditions @xmath544 the standard sard - smale type argument implies : for generic choices of universal floer data , @xmath545 is a smooth manifold of dimension @xmath546 moreover , the compactification of the abstract moduli space of half - discs extends to a gromov compactification of the moduli spaces @xmath545 .
the top strata of the boundary have a description analogous to the two types of boundary strata of the abstract moduli spaces ; assume in the first case that @xmath547 , that @xmath49 is an integer between @xmath43 and @xmath405 .
consider sequences of chords @xmath408 and @xmath409 with @xmath548 . by gluing a disc with inputs
@xmath549 and output @xmath550 to a half - disc with inputs
@xmath551 , we obtain a map @xmath552 similarly , given a partition of the inputs @xmath553 and @xmath554 , and an intersection point @xmath555 between @xmath556 and @xmath4 , we obtain a map @xmath557 using the parametrisation by arc length as in lemma [ lem : evaluate_discs_paths ] , we observe the existence of compatible families of evaluation maps to spaces of paths in @xmath4 : there exists a choice of parametrisations of the outgoing boundary segment of half - discs which yields maps @xmath558 such that in the setting equation we have a commutative diagram @xmath559 \ar[d ] & { \overline{{\mathcal h}}}(q_0 , { \vec{x}^{\ , \ ! } } , q_d ) \ar[d ] \\ { \overline{{\mathcal h}}}(q_0 , { \vec{x}^{\ , 1 } } , q_d ) \ar[r ] & \omega(q_0,q_d ) } \ ] ] while in the setting of equation the following diagram also commutes @xmath560 \ar[d ] & { \overline{{\mathcal h}}}(q_0 , { \vec{x}^{\ , \ ! } } , q_d ) \ar[d ] \\ \omega ( q_0 , q'_{d_1 } ) \times \omega ( q'_{d_1 } , q_d ) \ar[r ] & \omega(q_0,q_d ) . } \ ] ] the proof of the next result again follows from standard gluing techniques which are reviewed in appendix [ app : manifold - with - boundary ] . the fact that we are using hamiltonians which are not linear along the infinite cone might a priori imply that we lose compactness . however , the argument used in proving lemma [ lem : gromov_compactness ] applies here as well .
[ lem : topological_manifold_boundary_half_popsicle ] the moduli space @xmath561 is a compact topological manifold with boundary stratified by smooth manifolds .
the codimension @xmath76 strata are the images of the inclusion maps , . in order to define operations using the moduli space @xmath562 we must compare the orientation of the boundary of this moduli space with that of the interior . as with other sign - related issues , the proof of this result
is delayed to appendix [ sec : signs ] : [ lem : product_orientation_half_discs ] the product orientation on the strata differs from the boundary orientation by @xmath563 while the difference between the two orientations on the stratum is given by @xmath564 whenever @xmath565 is rigid and @xmath175 is the @xmath566@xmath567 element of @xmath551 .
following the same strategy as for the proof of lemma [ lem : fundamental_chains_linear_map ] , we may equip these moduli spaces with appropriate fundamental chains in the cubical theory : [ lem : good_choice_fundamental ] there exists a family of fundamental chains @xmath568 \in c_{*}({\overline{{\mathcal h}}}(q_0 , { \vec{x}^{\ , \ ! } } , q_d ) ) \ ] ] such that @xmath569 = \sum_{\substack { 0 \leq d_1 \leq d
\\ q'_{d_1 } \in l_{d_1 } \cap q } } ( -1)^{\flat } [ { \overline{{\mathcal h } } } ( q_0 , { \vec{x}^{\ , 1 } } , q'_{d_1 } ) ] \times [ { \overline{{\mathcal h } } } ( q'_{d_1 } , { \vec{x}^{\ , 2 } } , q_d ) ] + \\ \sum_{y \in { \euscript x}(l_{k } , l_{k+d_2 } ) } ( -1)^{\sharp } [ { \overline{{\mathcal h}}}(q_0 , { \vec{x}^{\ , 1 } } , q_d ) ] \times [ { \overline{{{\mathcal r}}}}(y,{\vec{x}^{\ , 2 } } ) ] .
\end{gathered}\ ] ] we now define a map @xmath570\otimes \cdots \otimes [ x_1 ] & \to \bigoplus_{\substack{q_0 \in l_0
\cap q \\ q_d \in l_d \cap q } } ( -1)^{\ddagger } { \operatorname{ev } } _ { * } ( [ { \overline{{\mathcal h}}}_{{\mathbf{p}}}(q_0 , { \vec{x}^{\ , \ ! } } , q_d ) ] ) .\end{aligned}\ ] ] where the sign is given by @xmath571 the collection of maps @xmath572 satisfy the @xmath2 equation for functors @xmath573 the discussion of signs is omitted as it is tedious , and does not differ in any significant way from the computation performed in section ( 9f ) of @xcite .
the correspondence between the terms of the @xmath2 equation and expressions for the boundary of the fundamental chain is particularly simple : ( i ) the left hand side of equation corresponds to the first term on the right hand side of equation ( ii ) the first sum on the right hand side of equation corresponds to the summation on the right hand side of , and ( iii ) the right hand side in equation cancels with the remaining terms from the right hand side of equation .
note that , a priori , the last sum in equation has contributions from moduli spaces of discs which are not necessarily rigid ; however , the image of such contribution under the evaluation map to @xmath22 vanishes as all such chains become degenerate ( i.e. , factor through a lower - dimensional chain ) .
let us now equip @xmath4 with a riemannian metric .
the cotangent bundle of @xmath4 is an exact symplectic manifold with primitive @xmath574 in local coordinates , and the points of @xmath575-norm less than @xmath76 form an exact symplectic manifold @xmath576 with contact type boundary . in particular
, we have a decomposition of the cotangent bundle as a union @xmath577 with @xmath578 the contact boundary of @xmath576 .
let us pick a function @xmath166 which agrees with @xmath579 on the infinite cone . choosing the metric generically
, we may ensure that such a hamiltonian has only non - degenerate chords with end points on @xmath580 .
abbondandolo and schwarz have already proved in @xcite that there is an isomorphism @xmath581 using a morse model for the left hand side . in this section
, we shall sketch the proof of [ eq : left_inverse ] the linear term @xmath582 of the functor @xmath34 descends on cohomology to an inverse for @xmath583 .
in particular , @xmath34 is an equivalence .
since the conventions of abbondandolo and schwarz for orientations on the floer side are the same , up to chain equivalence , to those we use in appendix [ sec : orientations ] for a trivial background class , the result of @xcite suffers from the sign issue alluded to in remark [ rem : sign_error ] .
the map constructed in @xcite becomes a chain map if we consider the morse homology of the based loop space with values in this local system .
alternatively , one can change the sign conventions on the floer side by twisting its generators by @xmath584 , and obtain a chain map from a twisted version of the wrapped floer cohomology to the ordinary homology of the based loop space .
lemma [ lem : twisted_to_untwisted_sign_difference ] implies that the resulting signs agree with those coming from considering the cotangent fibre as an object of the wrapped fukaya category with the background class @xmath10 which is the pullback of @xmath15 .
this allows us to use the results of @xcite whenever @xmath4 is orientable . if @xmath4 is not orientable then the pullback of @xmath585 under a based loop may not be trivial
. however , it admits two @xmath0 structure and @xmath586 is defined in this case to be the vector space generated by the two @xmath0 structures with the relation that their sum vanishes . in this way , the results of @xcite concerning the based loop space extend to the non - orientable case as noted in section 5.9 of @xcite ( though the relation between symplectic cohomology and the homology of the free loop space requires twisting the latter by the orientation line of the base ) .
since @xmath583 is known to be an isomorphism , it suffices to prove that @xmath34 is a one - sided inverse ; we shall prove that it is a right inverse . in order to pass to morse chains , we follow @xcite , and fix a morse function @xmath587 on @xmath588,0,1 ) , ( q , q , q))\ ] ] which is the lagrangian action functional for the legendre transform of @xmath166 .
the critical points of @xmath587 are non - degenerate by virtue of the analogous result for the critical points of @xmath194 , so we obtain a morse - smale flow on @xmath589 upon choosing a sufficiently generic metric .
let us write @xmath590 for the subgroup spanned by continuous chains of dimension @xmath591 which satisfy the following property : @xmath592 we call such chains _ weakly transverse_. note that weak transversality implies that a chain of dimension @xmath49 does not intersect the ascending manifolds of critical points of index greater than @xmath49 .
the inclusion of weakly transverse chains into all chains induces an isomorphism on cohomology .
moreover , for appropriate choices of fundamental chains on the moduli spaces of half - popsicle maps , the map @xmath593 factors through weakly transverse chains .
the first part follows by considering a smaller subcomplex consisting of all smooth chains which are transverse to all ascending manifolds of @xmath587 . as the inclusion of this subcomplex into @xmath594,0,1 ) , ( q , q , q ) ) \right)$ ] induces an isomorphism on cohomology , we conclude that the inclusion must induce a surjection on cohomology groups .
to obtain an injection , we must show that a weakly transverse chain is homologous to a transverse one through weakly transverse chains .
this follows from the fact that weakly transversality is stable under perturbations which are @xmath595 small near the intersections with ascending manifolds of codimension @xmath49 , since the result of a generic such perturbation is a homotopic transverse chain .
for the second part , we first observe that theorem 3.8 of @xcite implies that for a generic choice of almost complex structures the evaluation map @xmath596 is transverse to ascending manifolds of @xmath587 of codimension @xmath597 or higher for every @xmath169-chord @xmath175 .
the inverse image of such ascending manifolds consists of finitely many points in @xmath598 by gromov compactness . by induction
, we may therefore choose fundamental chains for @xmath599 satisfying the desired property .
the main reason for introducing @xmath600 is that the count of negative gradient flow lines starting on a given cell and ending on a critical point of @xmath587 defines a chain equivalence @xmath601 composing with we obtain a map @xmath602 where the right hand side is the morse complex of the function @xmath587 .
we end this section with a sketch of the proof of lemma [ eq : left_inverse ] .
in particular , we omit discussing signs .
abbondandolo and schwarz construct a chain homomorphism @xmath603 where @xmath604 is a count of maps @xmath605 \times [ 0,1 ] \to t^ * q\ ] ] solving equation which converge at infinity to @xmath175 with boundary condition @xmath606 along the semi - infinite segment @xmath607 \times \ { 0 \}$ ] and @xmath607 \times \ { 1 \}$ ] , and such that @xmath608 ) $ under the projection from $ t^*q$ to $ q$ is a path on $ q$ lying in the descending manifold of $ y$.}\ ] ] we shall prove that the composition @xmath609 is homotopic to the identity . to this end , for each positive real number @xmath610 and critical points @xmath550 and @xmath611 of @xmath587 , we consider a moduli space @xmath612 consisting of maps @xmath613 \times [ 0,1 ] \to t^ * q\ ] ] which solve the cauchy - riemann problem of equation , have boundary condition @xmath606 along the segments @xmath614 \times \ { 0 \}$ ] and @xmath614 \times \ { 1 \}$ ] , and such that @xmath615 ) $ is contained in the zero section and , considered as a path on $ q$ , lies on the ascending manifold of $ z$ , while the image of $ u ( \ { a \ } \times [ 0,1 ] ) $ under the projection from $ t^*q$ to $ q$ is a path on $ q$ lying in the descending manifold of $ y$.}\ ] ] for @xmath616 , any such solution is necessarily constant , so that the count of rigid elements of @xmath617 gives the identity map of @xmath618 .
let @xmath619 denote the gromov compactification of the union of all moduli spaces @xmath620 for @xmath621 .
letting @xmath610 go to @xmath214 , a family of maps @xmath622 defined on finite strips @xmath623 \times [ 0,1 ] $ ] breaks into two maps @xmath624 and @xmath625 defined on semi - infinite strips .
an inspection of the boundary conditions shows that the count of rigid elements of such a boundary stratum reproduces the composition of @xmath626 with @xmath583 as in equation .
the remaining boundary strata occur at finite @xmath610 when the projection of @xmath627 ) $ ] to @xmath4 escapes to the ascending manifold of a critical point @xmath628 which differs from @xmath550 .
similarly , the image of @xmath629 ) $ ] may converge to the descending manifold of a critical point @xmath630 .
such boundary strata are in bijective correspondence with @xmath631 where @xmath632 is the moduli space of gradient trajectories of @xmath587 ; the count of rigid element of these moduli spaces defines the differential on @xmath633 .
in particular , we conclude that counting rigid elements of @xmath634 is a homotopy between the identity and the composition of @xmath626 with @xmath583 completing the proof of lemma [ eq : left_inverse ] .
consider a pair of graded lagrangians @xmath24 and @xmath635 in a symplectic manifold @xmath3 , and a time-@xmath76 non - degenerate hamiltonian chord @xmath550 starting at @xmath24 and ending on @xmath635 .
let @xmath428 be a genus @xmath43 riemann surface obtained by removing a point from the boundary of @xmath271 , and equipped with a negative end near this puncture .
we choose a projection map @xmath636\ ] ] which extends the projection to the interval along the strip - like end , and a @xmath76-form @xmath637 on @xmath428 which vanishes near the boundary and agrees with the pullback of @xmath638 along the strip - like end .
using the almost complex structure @xmath127 on the tangent space of @xmath3 allows us to define a cauchy - riemann operator @xmath639 on sections of @xmath640 .
the linearisation of this operator has source sections of @xmath641 , and the data of gradings on @xmath24 and @xmath635 define a path of lagrangians @xmath642 in the restriction of this symplectic vector bundle to the boundary , uniquely determined up to homotopy by the requirement that it lift to a path of graded lagrangians interpolating between the graded structures on @xmath643 and @xmath644 ( see section ( 11 ) of @xcite ) on either side of the puncture
. if we were working in @xmath647 , and we replace @xmath24 by the zero section and @xmath635 by the graph of the differential of a morse function , then @xmath648 can be proved to be canonically isomorphic to the top exterior power of the descending manifold of the critical point associated to this lagrangian intersection .
the floer chain complexes we consider are direct sums of the orientation lines of @xmath648 ; i.e. the free abelian group generated by the two possible orientations of @xmath648 with the relation that their sum vanishes . in the case of exact lagrangian sections of cotangent bundles , such an orientation is equivalent to an orientation of the descending manifolds of the corresponding critical point , and it is well known that explicit signs in morse theory require choices of orientations on such descending manifolds . by working with orientation lines ,
we avoid making these choices , at the cost of being less explicit about the resulting pluses and minuses . for a thorough comparison between signs in morse and floer theory ,
see @xcite . imposing an additional condition on the choices made in section [ sec : at - level - objects ]
, we require that our chosen triangulation satisfy the following condition : @xmath649 note that while there are , a priori , infinitely many such chords , only finitely many of them intersect any given compact subset , which implies the existence of such a triangulation of @xmath3 . given a chord @xmath550 with endpoints on @xmath635 and @xmath24 ( and letting @xmath107 denote the vector bundle fixed in section [ sec : at - level - objects ] )
, we define a _ relative @xmath0 structure _ on @xmath646 to be a @xmath0 structure on @xmath650 whose restriction to an end of the chord agrees with the relative @xmath0 structure on either @xmath635 or @xmath24 . by thinking of an intersection point between @xmath24 and @xmath4 as a time-@xmath76 chord for the zero hamiltonian , we also associate a path @xmath651 to such a point , on which a relative @xmath0 structure may be chosen .
we now consider the following situation : @xmath652 is a collection of objects of @xmath653 and @xmath654 a sequence of chords with @xmath655 starting at @xmath656 and ending at @xmath657 .
in addition , we have either a chord @xmath396 or intersection points @xmath442 and @xmath658 of @xmath173 and @xmath659 with @xmath4 .
we shall only need to know the following lemma which is a minor generalisations of the standard result required to define operations in lagrangian floer theory away from characteristic @xmath19 . in our notation below , @xmath77 stands for the top exterior power of the tangent space of the appropriate manifold .
[ lem : orienting_mod_spaces ] choices of relative @xmath0 structures on the lagrangians @xmath657 and on the paths @xmath660 for each element of @xmath410 and for @xmath210 induce a canonical up to homotopy isomorphism @xmath661 whenever the moduli space is regular .
the additional choices of relative @xmath0 structures on @xmath662 and @xmath663 induce an isomorphism @xmath664 equation ( 12.8 ) of @xcite is the special case where the background class @xmath10 vanishes and proposition 4.2.2 of @xcite ( see , in particular equation ( 38 ) ) covers the relatively @xmath16 case .
we shall therefore limit ourselves to a discussion of the relevance of @xmath0 structures to sign considerations ; we shall see that they are introduced in order to resolve an ambiguity coming from the existence of homotopically distinct isotopies between a fixed pair of fredholm operators .
the next step is to deform @xmath670 through a family of smooth curves @xmath671}$ ] with the same asymptotic and boundary conditions to a curve @xmath672 mapping @xmath227 to the @xmath108-skeleton of @xmath3 .
because any homotopy of homotopies can be deformed so that the boundary still maps to the @xmath108-skeleton of @xmath3 , the construction we give below does not depend on the choice of @xmath673 ( see the analogous argument in chapter 8 of @xcite ) . from the homotopy
, we obtain an isomorphism @xmath674 the asymptotic conditions satisfied by @xmath675 near the positive punctures imply that we may glue to it the operators @xmath676 and obtain an operator on a disc with one negative puncture corresponding to @xmath442 which we denote : @xmath677 two salient fact about this operator will be relevant to us : first , gluing kernel and cokernel elements gives a natural identification @xmath678 as discussed , e.g. in section ( 11c ) of @xcite .
next , the fact that all lagrangians are _ graded _ implies that the boundary condition of this glued operator are isotopic to those of the operator @xmath679 through an isotopy which is constant near the end .
a _ choice _ of isotopy defines an isotopy of fredholm operators on the disc , and hence using equation , an isomorphism @xmath680 which , together with equations and , gives an isomorphism between the right spaces .
note that the choice of isotopy is parametrised , up to homotopy , by the fundamental group of the space of lagrangian paths between two fixed points , which is a group of order @xmath19 in dimensions greater than @xmath19 ; the two different choices give different signs in the above isomorphism and we must be able to make a coherent choice in order to obtain consistent operations away from characteristic @xmath19 ( see , e.g. lemma 11.17 of @xcite ) .
the choice of relative @xmath0 structures enters only in this choice of homotopy . indeed , definition [ def : brane ]
required that we choose a @xmath0 structure on the vector bundle @xmath681 } \oplus e_{b}|l_{[3]}$ ] over the @xmath108-skeleton of every object @xmath24 of @xmath18 .
this choice determines a @xmath0 structure on ( 1 ) the direct sum of the boundary condition of the operator @xmath679 with the trivial vector bundle with fibre @xmath682 and ( 2 ) the direct sum of the boundary condition for @xmath683 with the vector bundle @xmath684 over the boundary of the once - punctured disc obtained by gluing the pull - back of @xmath107 to @xmath227 with the pullbacks of @xmath107 to the one punctured disc @xmath428 under the compositions of the chords @xmath685 with the projection @xmath686 $ ] fixed in equation .
note that the pullback of @xmath107 by @xmath687 , and by @xmath688 , extends the vector bundle to the entire riemann surface , hence induces a canonical up to homotopy isomorphism between this glued vector bundle and the trivial vector bundle with fibre @xmath689 . with this in mind
, we now fix the path between the glued operator and @xmath679 to be the unique one , up to homotopy , such that direct sum of the boundary conditions with the trivial bundle @xmath690 admits a @xmath0 structure which restricts , at both ends , to the one fixed above .
[ rem : trivialise_r_action ] whenever @xmath691 in equation ( or @xmath43 in equation ) , it is useful to think of @xmath692 or @xmath693 as a manifold of dimension @xmath694 whose tangent space is `` generated '' by translation on the strip .
let us now consider the special case where @xmath3 is the cotangent bundle of @xmath4 , and the lagrangians @xmath695 agree with a cotangent fibre @xmath696 .
we choose a vector bundle @xmath107 on a @xmath108-skeleton for @xmath7 with stiefel - whitney class @xmath8 , and fix a submanifold @xmath697 poincar dual to @xmath10 , and which is disjoint from @xmath698 as well as from all time-@xmath76 hamiltonian chords of @xmath166 with endpoints on this cotangent fibre .
since it is contractible , the cotangent fibre @xmath696 is an object of the wrapped fukaya category of @xmath7 for all possible twistings , and we would like to compare the isomorphisms ( [ eq : orientation_higher_product ] ) for an arbitrary class @xmath10 , with the one coming from the trivial class . in order to do this , we choose a @xmath0 structure on the lagrangian path @xmath699 associated to any chord @xmath700 .
then , noting that the restriction of @xmath107 to the complement of @xmath701 in @xmath702 } $ ] admits a @xmath0 structure , we choose such a structure , and equip @xmath703 with the associated direct sum of @xmath0 structures . [ lem : twisted_to_untwisted_sign_difference ] given a curve @xmath704 , the two isomorphisms @xmath705 obtained from the sign conventions associated to @xmath696 either as a @xmath0 submanifold of @xmath7 , or as a relatively @xmath0 submanifold with respect to the background class @xmath706 differ by @xmath707 .
fix an isotopy of operators which induces the isomorphism ( [ eq : gluing_iso_dbar_disc ] ) in the untwisted case , i.e. such that the associated family of lagrangian boundary conditions admits a @xmath0 structure restricting at the two endpoints of the isotopy to the @xmath0 structure on @xmath708 , and on the glued boundary conditions . by definition
, this given isotopy induces the desired isomorphism for the twisted case if and only the direct sum with the pullback of @xmath107 admits a @xmath0 structure , i.e. if and only if the pullback of @xmath107 under @xmath672 is @xmath0 .
the obstruction is precisely the intersection number of @xmath672 with @xmath709 , which , because @xmath239 and @xmath687 have the same asymptotic and boundary conditions , agrees with the intersection number of @xmath239 with @xmath709 . to obtain operations from equations and
, one must choose orientations on the abstract moduli space @xmath710 and @xmath711 , or in the exceptional cases , a trivialisation of the @xmath121 action on the moduli spaces of strips .
the orientations on @xmath342 is fixed by identifying with the configuration space of @xmath712 ordered points on an interval , which correspond to the marked points @xmath713 ; this also fixes an orientation on @xmath714 via the isomorphism of equation . in the case of a strip
, we shall use @xmath715 as the generator of the action of @xmath121 . for simplicity ,
we shall also fix trivalisations of @xmath716 for every intersection point between a lagrangian @xmath717 and @xmath4 .
we shall use the following conventions in determining signs : every orientation on a manifold @xmath169 induces an orientation on its boundary via the canonical isomorphism @xmath718 where @xmath719 is the normal bundle of @xmath720 which is canonically trivialised by the outwards normal vector .
whenever @xmath169 is a moduli space , say for specificity @xmath721 , lemma [ lem : orienting_mod_spaces ] gives an orientation _ relative _ the orientations lines associated to the inputs and outputs ( and an orientation of the abstract moduli space ) . in all the cases we consider , the boundary of the moduli space is a product , and hence inherits two relative orientations , one coming from equation , the other from taking the product of the two relative orientations coming from lemma [ lem : orienting_mod_spaces ] .
the discrepancy between these two signs must be computed .
we illustrate the computation by providing the proof of some sign - related results .
the first is a lemma from section [ sec : at - level - objects ] : by rearranging the terms in equation , we obtain an isomorphism @xmath722 taking the tensor product of these isomorphisms for the pairs @xmath723 and @xmath724 , we obtain @xmath725 the key observation coming from gluing theory is that the first translation vector field corresponds to a vector pointing outwards along this boundary stratum of @xmath726 ( see section ( 12f ) of @xcite ) .
if we identify it with the normal vector @xmath719 , reorder the terms , and cancel the copy of @xmath727 we arrive at @xmath728 with a koszul sign equal to @xmath729 since @xmath526 is rigid , equation specialises to @xmath730 if we take the tensor product of equation with the corresponding isomorphism for @xmath731 , we obtain @xmath732 the translation vector @xmath733 gives rise to an outward pointing normal vector upon gluing , so the only difference between this and the boundary orientation comes from koszul signs . by assumption , @xmath734 is rigid , so there is no sign associated to moving @xmath735 past @xmath736 .
the total sign is therefore @xmath737 coming from moving @xmath738 to be the first factor .
next , we consider the situation where a strip breaks at the intersection point between @xmath179 and @xmath4 : @xmath739 after gluing , the vector @xmath740 gives an inward pointing normal vector , so the total sign difference between the two orientations is @xmath741 .
finally , when breaking occurs at the intersection point between @xmath173 and @xmath4 , the reader may easily check that the two relative orientations agree .
when breaking takes place on the outgoing segment , equation implies that the product orientation is given by an isomorphism @xmath742 as the dimension of @xmath743 is @xmath744 , the koszul sign introduced by moving @xmath745 to be adjacent to @xmath746 is @xmath747 .
the additional term in equation comes from the difference between the product orientation of @xmath748 and the orientation of @xmath711 . at the other type of boundary stratum
, the product orientation is given by @xmath749 since we re assuming that @xmath750 is rigid , there is no sign associated to reordering the factors of the right hand side as @xmath751 moving @xmath752 to be adjacent to @xmath753 introduces a koszul sign of @xmath754 .
the difference with equation again comes from the orientation of the boundary of the abstract moduli space of discs . in this section
we sketch the proof of proposition [ prop : moduli_spaces_strips_manifold ] .
this is a standard consequence of a gluing theorem , proved in slightly different ways by many different people , but we will describe a specific approach , expecting that expert readers will prefer to replace it by their own .
the method we explain in this appendix has the advantage of using only the most elementary version of the implicit function theorem in its proof , but the disadvantage of requiring a series of auxiliary geometric choices .
our approach closely follows that of @xcite .
first , the gromov compactification of @xmath755 is by definition the union of @xmath756 with the images of embeddings @xmath757 which satisfy .
an application of the sard - smale theorem in the appropriate banach space implies that @xmath756 is a smooth manifold of the appropriate dimension for generic choices of perturbation data .
an index computation implies that the difference between the dimension of @xmath756 and @xmath758 is @xmath83 ( the number of factor minus @xmath76 ) .
let @xmath759 denote a point in such a stratum .
we must prove that we may choose a neighbourhood @xmath760 of @xmath761 in this product of lower dimensional manifolds so that there is local homeomorphism ( the result of a gluing theorem ) @xmath762 in practice one uses the stratification on @xmath763 and constructs the above map separately for each stratum , checking ultimately that the resulting map is continuous ; we shall focus on the top stratum in which none of the gluing coordinates vanish . to construct these gluing map ,
we assume without loss of generality that each map @xmath764 has non - vanishing derivative at the origin ( this may be achieved by pre - composing @xmath764 by a translation ) , and set @xmath765 . in addition , writing @xmath766 for the dimension of @xmath767 , we choose a sequence of points @xmath768 which are regular points for @xmath769 and such that the derivative of the evaluation map at these points @xmath770 is an isomorphism at @xmath769 .
we fix a product of hypersurfaces @xmath771 which is transverse to the image of @xmath772 and such that the inverse image of each factor under @xmath773 is transverse to the lines obtained by fixing the @xmath774-coordinate on the strip . given a sequence @xmath775 of holomorphic maps sufficiently close to @xmath761
, there is a unique parametrisation @xmath776 which @xmath137 close to @xmath773 and such that @xmath777 .
having fixed this parametrisation , the inverse image under @xmath776 of the remaining factors of @xmath778 determines a unique sequence of points @xmath779 which are @xmath780 close to the sequence @xmath781 and such that the @xmath774-coordinates agree .
in particular , fixing a sufficiently small neighbourhood @xmath782 of @xmath776 in @xmath767 , we obtain an embedding @xmath783 to summarise this discussion , we have chosen a product of hypersurfaces such that the moduli space of holomorphic discs near @xmath773 is parametrised by the @xmath205-coordinates of the unique inverse image of @xmath785 whose @xmath774-coordinate agrees with @xmath774-coordinate of the original marked point @xmath786 .
the choice of @xmath778 determines a family of right inverses to the linearisation @xmath92 operator @xmath787 indeed , this operator becomes an isomorphism when restricted to those sections whose value at @xmath788 lies in @xmath789 .
the inverse of this isomorphism gives a right inverse to the linearisation of the @xmath92 operator in any reasonable sobolev completion of the space of smooth sections . to be precise , one must work with functions satisfying an exponential decay condition along the ends of the strip , which is reflected in weighted sobolev spaces when passing to completions .
we are now ready to proceed with gluing holomorphic curves .
first , we define a pre - gluing map @xmath790 obtained by removing the half - infinite strips @xmath791 \textrm { and } ( -\infty , \frac{-2}{\epsilon_\ell } ] \times [ -1,1]\ ] ] respectively from the domains of the curves @xmath792 and gluing the corresponding maps along the finite strips @xmath793 \times [ -1,1 ] \textrm { and } [ \frac{-2}{\epsilon_\ell } , \frac{-1}{\epsilon_\ell } ] \times [ -1,1]\ ] ] using a cutoff function .
this makes sense because these curves exponentially converge along the appropriate end to the same point in @xmath3 .
we assume that @xmath51 is small enough that the regions along which we glue the curves are disjoint from all marked points on the domains of @xmath794 and @xmath769 . as usual
, there is an @xmath121 ambiguity in the parametrisation of @xmath795 .
we resolve this ambiguity by making the origin of the domain of @xmath795 agree with the marked point coming from the origin of the domain of @xmath796 .
in other words , we start with the domain of @xmath796 , and glue the domains of the remaining curves to it . in particular ,
the pre - glued curve is equipped with @xmath797 marked points @xmath798 coming from the origins of the domains of the curves @xmath776 ( in particular , these marked points all have vanishing @xmath774-coordinate , and their @xmath205-coordinate is a sum of terms of the form @xmath799 ) .
moreover , there are @xmath800 marked points @xmath801 coming from the marked points on each curve @xmath776 .
the @xmath205 coordinate of of @xmath802 is simply the sum @xmath803 . letting @xmath344 stand for the dimension of @xmath804
, we obtain a sequence of @xmath344 marked points @xmath805 on the domain of @xmath795 with the convention that the last @xmath83 marked points correspond to the origins of the domains of @xmath806 , and we set @xmath807 to be the origin as before .
note that at each such marked point the image of @xmath795 intersects a distinguished hypersurface @xmath808 transversely ; these hypersurfaces are the ones we chose earlier in order to rigidify each curve @xmath776 .
as @xmath795 is not holomorphic , there is no natural linearisation of the @xmath92 operator acting on the tangent to the space of smooth maps at @xmath795 . to obtain such a linearisation , we choose for each curve @xmath776 , a family of metrics on @xmath3 parametrised by @xmath284 which is constant ( and independent of @xmath809 away from a compact set ) and such that @xmath810 is totally geodesic with respect to the metric corresponding to the marked point @xmath811 .
note that for @xmath51 sufficiently small , this choice naturally induces a choice of metrics in @xmath3 parametrised by the domain of @xmath795 such that @xmath808 is totally geodesic with respect to the metric corresponding to @xmath812 .
let us write @xmath813 for those sections of @xmath814 whose value at the @xmath49@xmath50 marked point lies in the tangent space to @xmath815 .
again , one must impose an appropriate decay condition at the ends in order to obtain a well - behaved elliptic problem . for @xmath51
sufficiently small , the linearisation of the @xmath92 operator becomes an isomorphism when restricted to @xmath816 .
moreover , in an appropriate sobolev completion of the space of smooth functions , the following quantities are uniformly bounded with respect to @xmath817 : 1 .
the norm of @xmath818 .
2 . the norm of the unique right inverse to the linearisation of the @xmath92 operator whose image is @xmath816 .
the difference between linearisation of the @xmath92 operator at @xmath795 and at points obtained by exponentiating vector fields of a fixed norm .
the most subtle part of the construction is the choice of a sobolev completion in which all these properties hold at once .
following @xcite , we would prove this result by choosing a family of weights depending on the parameter @xmath817 .
the properties are listed in what we believe is their order of difficulty : the first is an essentially elementary estimate while the second can be proved by constructing a pre - gluing map at the level of tangent vector field following @xcite . in general , one can not explicitly describe the result of pre - gluing tangent vectors in the image of the chosen right inverse along the component curves of @xmath806 .
however , for our explicit choices of right inverses , one can easily check that the result is the sobolev completion of @xmath816 .
the last property above is a version of the quadratic inequality ( see appendix b of @xcite for a proof in a related setting ) . the gluing map @xmath819 is defined by exponentiating this vector field at every curve obtained by pre - gluing . the proof that this map extends continuously upon letting some of the gluing parameters go to @xmath43 is a direct consequence of gromov compactness , and is omitted .
it remains therefore to explain why this map is a homeomorphism onto the intersection of a neighbourhood of @xmath761 in @xmath820 with @xmath804 . to prove this
, we first observe that , as a consequence of gromov compactness , there is a neighbourhood @xmath821 of @xmath761 in @xmath804 consisting of curves @xmath822 which , for each integer @xmath49 admit a unique point @xmath823 close to @xmath824 for some sequence @xmath817 , such that the @xmath774-coordinates of @xmath823 and @xmath824 agree , and @xmath825 in particular , taking the @xmath205-coordinate of each point @xmath826 defines a map @xmath827 as the right inverse to the @xmath92 operator consists of vector fields which , at @xmath828 point in the direction of @xmath815 , and as @xmath815 is totally geodesic with respect to the chosen metric , we conclude that @xmath829 intersects @xmath815 at @xmath828 .
this implies : the map @xmath42 is a right inverse to @xmath830 .
more formally , we have a commutative diagram @xmath831^{{\operatorname{g } } } \ar[dr ] & & w \ar[dl]^{r } \\ & { { \mathbb r}}^{d } & } \ ] ] where the map @xmath832 is the codimension @xmath43 embedding which takes @xmath833 to the @xmath205-coordinates of @xmath828 .
the existence of this a priori right inverse to the gluing map also helps in proving surjectivity . indeed ,
if the images of @xmath834 and @xmath835 under @xmath42 agree , then @xmath822 can be expressed as the exponential of a vector field along @xmath795 lying in @xmath816 . by the uniqueness part of the implicit function theorem , @xmath822 and @xmath835
must agree if we can prove that this vector field is bounded in the appropriate sobolev norm . for @xmath822 sufficiently close to @xmath761 , this follows from gromov compactness , and a standard estimate on the energy of long strips ( see e.g. section 5d of @xcite ) .
this completes our sketch of the proof of proposition [ prop : moduli_spaces_strips_manifold ] . | given an exact relatively @xmath0 lagrangian embedding @xmath1 , we construct an @xmath2 restriction functor from the wrapped fukaya category of @xmath3 to the category of modules on the differential graded algebra of chains over the based loop space of @xmath4 .
if @xmath3 is the cotangent bundle of @xmath4 , this functor induces an @xmath2 equivalence between the wrapped floer cohomology of a cotangent fibre and the chains over the based loop space of @xmath4 , extending a result proved by abbondandolo and schwarz at the level of homology . |
recent observations@xcite suggest that the universe consists of dark energy ( @xmath3 ) , dark matter ( @xmath4 ) and baryon matter ( @xmath5 ) . to understand the nature of the dark energy is one of essential tasks in cosmology .
the most simple form of the dark energy is the cosmological constant , which fits well to the observation data @xcite and which has the effective equation of state @xmath6 , _
i.e. _ @xmath7 .
if the dark energy is totally contributed by the cosmological constant , the universe is asymptotically de sitter one and we should study the de sitter spacetime and asymptotically de sitter spacetimes in all aspects , including in the framework of general relativity @xcite , in the framework of de sitter invariant special relativity @xcite , and locally de sitter invariant gauge theory of gravity @xcite , etc .
however , there is an evidence @xcite to show that the dark energy might evolve from @xmath8 in the past to @xmath9 today and cross @xmath0 in the intermediate redshift . obviously , the simple dynamical dark energy models considered vastly in the literature as well as the cosmological constant can not explain it . for example ,
@xmath10 in the quintessence models @xcite is always evolving in the range of @xmath11 . in the phantom models
@xcite , regardless its wrong - sign kinetic energy term , @xmath10 always varies below @xmath12 .
the parameter @xmath10 of the effective equation of state for the general k - essence models @xcite also fails to cross @xmath0 .
recently , many dark energy models with the property of crossing @xmath0 have been proposed @xcite . in the dynamical - scalar - field models of dark energy @xcite , either the malformed phantom fields or higher derivative term
are needed . in @xcite ,
the structures in extra dimension are introduced to realize the parameter @xmath10 crossing @xmath12 . in @xcite ,
the dark energy is supposed to be contributed from a vector field @xcite or a gas of fermion particles interacting with a condensate represented by a vector field @xmath13 @xcite .
the vector - field and fermion - field models of the dark energy do not need phantom fields , higher derivative terms and more complex structure in extra dimension and so on . in the present paper
, we shall propose a new single - dynamical - scalar - field model , as a toy , of dark energy , which may explain that the parameter @xmath10 crosses @xmath12 at @xmath1 and the the universe experience a phase transition form decelerating to accelerating at @xmath2 .
the basic idea is that the dark energy is not direct effect of a fundamental field , while it is only effectively described by a dynamical scalar field .
taking it into account , we may consider the following model of a single dynamical scalar field coupled to an _ a priori _ non - dynamical , background covariant vector field , [ action ] s_^eff = d^4x ( 12g^_,_-g^ _ , a_-v ( ) ) where the vector @xmath14 describes some unknown effects of the universe and is supposed to have constant zeroth - component in a comoving system .
in such a model , no phantom is needed to realize @xmath9 and @xmath10 crossing @xmath12 . in the following
, we shall introduce the model first .
and then we shall make some numerical analysis with an ordinary potential and a potential inspired from supergravity @xcite .
finally , we shall give some concluding remarks .
suppose that the universe is governed by the action s = s_eh+s_fluid+s_^eff , where @xmath15 is the einstein - hilbert action of gravitation , @xmath16 is the action of perfect fluid , describing the baryonic matter and cold dark matter , and @xmath17 is the effective action of dark energy , which has been given in eq.([action ] ) . since @xmath14 is introduced as a part of the effective model of dark energy , we suppose that it only couples to the scalar field but does not couple to the matter .
as the standard treatment in cosmology , we suppose that the universe is homogeneous and isotropic on large scale , and further suppose that the universe is flat for simplicity , so that the geometry of the universe is described by the robertson - walker metric [ rwmetric ] ds^2=dt^2-a^2(t)(dr^2 + r^2 d_2 ^ 2 ) , where @xmath18 is the line - element on a unit sphere . under the assumption of homogeneity and isotropy , all quantities , including @xmath19 , @xmath20 , etc . , depend on @xmath21 only .
in particular , the stress - energy tensor for the scalar field , defined by [ stress ] t_= 2 , reads [ stress - cp ] t_= ^2(t ) ^0_^0 _ -2(t ) a_0 ^0_^0_- 1 2 g _ ( ^2(t ) -2(t)a_0 -2v[(t ) ] ) . comparing the stress - energy tensor ( [ stress - cp ] ) with the one for perfect fluid , one may define the effective energy density and pressure of scalar field by _
^(t ) & = & t_00=1 2 ^2(t ) -(t ) a_0 + v[(t ) ] , + p_^(t ) & = & -g^11t_11= 1 2 ^2(t ) -(t ) a_0 -v[(t ) ] . as the effective description of dark energy
, @xmath22 should always be positive .
the parameter @xmath10 in the effective equation of state @xmath23 is then [ w ] w= .
it is obvious that for @xmath24 , w -1 , & 2a_0 0 , + < -1 , & 2a_0 > > 0 ^2 - 2a_0 + 2v ( ) > 0 , + , & 2a_0 > > 0 ^2 - 2a_0 + 2v ( ) 0 , and for @xmath25 , w -1 , & 0 2a_0 , + < -1 , & 0 > > 2a_0 ^2 - 2a_0 + 2v ( ) > 0 , + irrelevant , & 0 > > 2a_0 ^2 - 2a_0 + 2v ( ) 0 . namely , the parameter @xmath10 may cross @xmath12 in the evolution of the universe .
as mentioned in the previous section , we suppose that @xmath14 has a constant norm and constant zeroth - component in robertson - walker metric . under the assumption ,
the equation of motion for @xmath26 takes the form [ eomphi ] + 3h+ = 3ha_0 , where @xmath27 is the hubble parameter . comparing with the standard form of the equation of motion for a scalar field
, there exists an additional current due to the existence of @xmath28 .
now , the friedmann equation reads [ friedmann1 ] h^2 = 3(1 2 ^2 -a_0 + v ( ) + _ fluid^ ) , where the last term on the right - hand - side is the energy density of matter as a perfect fluid . in the present paper
, we are only interested in the late stage of the evolution of the universe , in which both the baryon matter and cold dark matter can be treated as pressureless perfect fluid .
thus , we may set = _ 0^a_0 ^ 3/a^3 , where @xmath29 is a constant , denoting the present value of the energy density of all matter .
the subscript 0 , except in @xmath28 and the following @xmath30 , represents to take the value today .
in order to solve eqs.([eomphi],[friedmann1 ] ) numerically , we recast them as the first - order differential equations with respect to the redshift @xmath31 , & = & - + & = & 3 + + h^2&=&1 2 ^2 -b_0+v ( ) + _ m0^(1+z)^3 , where = , = , b_0 = , h = h h_0 , v()= , _ m0^=. in terms of new variable , eq.([w ] ) becomes w= .
the initial values of integration are assigned to fit the present observation data @xcite : h_0=1,&_de0^:=.(1 2 ^2 -b_0+v ) |_0 = 0.73,&_m0^ = 0.27 , + w_0=-1.33 , & _
0^(_0^ -2b_0 ) = -0.24 , & v_0^ = 0.85 .
now , we consider several examples .
the first example is the universe with a quadratic potential [ potential ] v ( ) = 1 2 m^2 ^2 .
we may further take @xmath32 and w_1=.|_0 = 1 , then @xmath33 , @xmath34 , v()|_z=0^==-0.117 , @xmath35 and @xmath36 .
1 gives the evolution of @xmath10 for a flat universe with a quadratic potential .
it is obvious that the equation of state across @xmath0 at @xmath1 , which is consistent with the sne ia observation @xcite and that the transition redshift between the accelerating and decelerating phases occurs at @xmath37 , which is lower than the observational constraint on the transition redshift interval @xmath38 @xcite .
the limit of @xmath10 at large @xmath39 is 1 . for a flat universe with a quadratic potential with the initial values @xmath40 , @xmath41 , @xmath42 , @xmath35 , @xmath36 , and @xmath43 . the right figure is the expanded view of the left figure between @xmath44 and @xmath45.,title="fig : " ] for a flat universe with a quadratic potential with the initial values @xmath40 , @xmath41
, @xmath42 , @xmath35 , @xmath36 , and @xmath43 . the right figure is the expanded view of the left figure between @xmath44 and @xmath45.,title="fig : " ] for @xmath46 the behavior of @xmath10 is similar to the case @xmath42 . it can be shown that @xmath10 has the similar behavior for the quartic potential v ( ) = 1 4 ^4 , and the inverse power - law potential v ( ) = v_0 ^- .
the next example is the universe with the potential inspired from supergravity @xcite , [ supgravpot ] v ( ) = v_0 ^-e^4 g ^2 , where @xmath47 is a constant . for the potential
, we may take @xmath48 , @xmath49 , @xmath50 , thus @xmath51 , @xmath52 , @xmath53 .
3 gives the evolution of @xmath10 for the model .
the equation of state across @xmath0 at @xmath1 and the transition redshift between the accelerating and decelerating phases occurs at @xmath54 , which is consistent with the sne ia observation @xcite .
however , the dependence of @xmath10 on @xmath39 is not monotonic .
@xmath10 becomes less than @xmath12 between @xmath55 and @xmath56 again .
the universe experienced another accelerated expansion around @xmath57 .
for a flat universe with a quadratic potential with the initial values @xmath40 ,
@xmath50 , @xmath49 , @xmath48 and @xmath43 . the right figure is the expanded view of the left figure between @xmath58 and @xmath45.,title="fig : " ] for a flat universe with a quadratic potential with the initial values @xmath40 , @xmath50 , @xmath49 , @xmath48 and @xmath43 . the right figure is the expanded view of the left figure between @xmath58 and @xmath45.,title="fig : " ]
the single - dynamical - scalar - field model of dark energy may explain the parameter @xmath10 in the effective equation of state crossing @xmath12 during the evolution of the universe , as long as a fixed background vector field is introduced .
for some potentials , the parameter @xmath10 crossing @xmath12 and transition from decelerating to accelerating occur at @xmath1 and @xmath2 , respectively , which is consistent with the observations .
the dependence of the parameter @xmath10 on the redshift @xmath39 is far from the linear relation .
thus , it is needed to re - analyze the observation data based on more general form of @xmath59 . by the way , without the introduction of oscillation potential , the universe may also exhibit some oscillation behaviors .
the present model is in the almost standard framework of physics , _
e.g. _ in general relativity in 4-dimension .
there does not exist phantom in the model , which will lead to theoretical problems in field theory . instead ,
an _ a priori _ vector field , @xmath14 , is introduced , which is selected in an unspecified way by the cosmological background .
it will result in the violation of the lorentz invariance and might be contributed by the collective effects of the vacuum polarization of some fundamental fields .
it is needed to further explore the physical meaning of such a vector field .
the work is supported in part by the national natural science foundation of china under the grant no .
90403023 and 10375087 .
bennett c l _ et al _ , first year wilkinson microwave anisotropy probe ( wmap ) observations : preliminary maps and basic results , 2003 _ astrophys
. j. suppl . _ * 148 * 1 [ astro - ph/0302207 ] ; spergel d n _ et al .
_ , first year wilkinson microwave anisotropy probe ( wmap ) observations : determination of cosmological parameters , 2003 _ astrophys
. j. suppl . _ * 148 * 175 [ astro - ph/0302209 ] ; tegmark m _
et al . _ ,
cosmological parameters from sdss and wmap , 2004 _ phys.rev .
d _ * 69 * 103501 [ astro - ph/0310723 ] .
@xmath60 , wang y and tegmark m , new dark energy constraints from supernovae , microwave background and galaxy clustering , 2004 241302 [ astro - ph/0403292 ] ; seljak u _
_ , cosmological parameter analysis including sdss ly - alpha forest and galaxy bias : constraints on the primordial spectrum of fluctuations , neutrino mass , and dark energy , 2004 [ astro - ph/0407372 ] .
see , e.g. , gibbons g w and hawking s w , cosmological event horizons , thermodynamics , and particle creation , 1977 , 2738 ; spradlin m , strominger a and volovich a , les houches lectures on de sitter space , 2001 [ hep - th/0110007 ] ; huang c - g , liu l and wang b , thermodynamics of de sitter universes , 2002 _ phys . rev .
d _ * 65 * 083501 .
see , e.g. , guo h - y , huang c - g , xu z and zhou b , on beltrami model of de sitter spacetime , 2004 _ mod .
a _ * 19 * 1701 [ hep - th/0311156 ] ; on special relativity with cosmological constant 2004_phys .
a _ * 331 * 1 [ hep - th/0403171 ] ; guo h - y , huang c - g and zhou b , on horizon in de sitter invariant special relativity , hep - th/0404010 ; guo h - y _ et al _ , on de sitter invariant special relativity and cosmological constant as origin of inertia , hep - th/0405137 ; beltrami - de sitter spacetime and de sitter invariant special relativity , 2005 _ acta phys . sin . _ * 54 * 2494 ( in chinese ) and references therein .
wu y - s , li g - d and guo h - y , gravitational lagrangian and local de sitter invariance , 1974 _ kexue tongbao ( chinese sci . bull . )
_ * 19 * 509 ; guo h - y , the local de sitter invariance , 1976 _ ibid _ * 21 * 31 ; an y , chen s , zou z - l and guo h - y , a new theory of gravity gravitational theory with local de sitter invariance , 1976 _ ibid _ , * 21 * 379 ( all in chinese ) . townsend p , small - scale structure of spacetime as the origin of the gravitational constant , 1977 _ phys . rev .
_ * d15 * 2795 ; macdowell s w and mansouri f , unified geometric theory of gravity and supergravity , 1977 _ phys .
* 38 * 739 ; * 38 * 1376 .
@xmath60 , corasaniti p s _ et al _ , foundations of observing dark energy dynamics with the wilkinson microwave anisotropy probe , 2004 _ phys
d _ * 70 * ; 083006 , [ astro - ph/0406608 ] ; hannestad s and mortsell e , cosmological constraints on the dark energy equation of state and its evolution 2004 _ jcap _ 0409 001 , [ astro - ph/0407259 ] ; upadhye a , ishak m and steinhardt p j , dynamical dark energy : current constraints and forecasts 2004 [ astro - ph/0411803 ] .
wetterich c , cosmology and the fate of dilatation symmetry , 1988 _ nucl .
b _ * 302 * , 668 ; 1995 _ astron .
astrophys . _ * 301 * 321 ; ratra b and peebles p j , cosmological consequences of a rolling homogeneous scalar field , 1988 _ phys . rev .
* 37 * 3406 ; caldwell r r , dave r and steinhardt p j , cosmological imprint of an energy component with general equation of state , 1998 _ phys .
* 80 * , 1582 [ astro - ph/9708069 ] .
feng b , wang x and zhang x , dark energy constraints from the cosmic age and supernova , 2005 _ phys .
b _ * 607 * 35 [ astro - ph/0404224 ] ; feng b li m piao y - s and zhang x , oscillating quintom and the recurrent universe , 2004 [ astro - ph/0407432 ] ; wei h cai r - g and zeng d - g , hessence : a new view of quintom dark energy , 2005 _ class . quantum grav . _
* 22 * 3189 [ hep - th/0501160 ] .
cai r - g zhang h - s and wang a - z , crossing @xmath0 in gauss - bonnet brane world with induced gravity , 2005 to appear [ hep - th/0505186 ] ; li b j chu m - c cheung k c and tang a , dark energy as a signature of extra dimensions , 2005 [ astro - ph/0501367 ] .
zhang y , a model of dark energy for the accelerating universe , 2003 _ gen .
* 35 * 689 ; _ ibid _ dark energy coupled with relativistic dark matter in accelerating universe 2003 _ chin .
* 20 * 1899 ; _ ibid _ , relic gravitational waves and a model of dark energy , talk on ccast workshop on dark matter and dark energy in the universe 2005 ; armendriz - picn c , could dark energy be vector - like ?
2004 _ jcap _ 07 007 [ astro - ph/0405267 ] .
riess a g _ et al _ , the farthest known supernova : support for an accelerating universe and a glimpse of the epoch of deceleration , 2001 _ astrophys .
j. _ * 560 * 49 [ astro - ph/0104455 ] ; avelino p p de carvalho j p m and martins c j a p , supernova constraints on spatial variations of the vacuum energy density , 2001 _ phys .
d _ * 64 * 063505 [ astro - ph/0103075 ] ; avelino p p and martins c j a p , a supernova brane scan , 2002 _ astrophys.j _ * 565 * 661 [ astro - ph/0106274 ] ; zhu z - h and fujimoto m - k , constraints on cardassian expansion from distant type ia supernovae , 2003 _ astrophys . j _ * 585 * 52 [ astro - ph/0303021 ] . | a new single - dynamical - scalar - field model of dark energy is proposed , in which either higher derivative terms nor structures of extra dimension are needed . with the help of a fixed background vector field ,
the parameter for the effective equation of state of dark energy may cross @xmath0 in the evolution of the universe . after suitable choice of the potential , the crossing @xmath0 and transition from decelerating to accelerating occur at @xmath1 and @xmath2 , respectively . |
from the eigenvalue equation @xmath87 with @xmath88 and the corresponding characteristic polynomial @xmath89 one obtains @xmath90 , @xmath91 where @xmath92 for gain / loss parameters @xmath93 $ ] the eigenvalues are purely real , @xmath94 ; it holds @xmath95 so that @xmath96 and the @xmath97symmetry is exact . for @xmath98 the eigenvalues
@xmath99 are not real , but pairwise complex conjugate ; one finds @xmath100 so that @xmath101 and the @xmath97symmetry is spontaneously broken with @xmath102 phase transition at @xmath103 .
the hamiltonian @xmath57 can be brought into the more symmetric form eq .
( 5 ) via a similarity transformation @xmath55 given by equation ( 4 ) . to describe the dynamics in the sector of exact @xmath102 symmetry , @xmath93 $ ] , where @xmath104 , @xmath105 , @xmath94 we start from an ansatz @xmath106 , @xmath107 and impose the experimentally required reality constraint @xmath108 as @xmath109 , @xmath110 .
the solutions of the evolution equation ( 1 ) , ( 2 ) with initial condition on the gain side @xmath111 are @xmath112 the first passage time @xmath0 required for the evolution from the initial state @xmath111 to an orthogonal final state @xmath113 , @xmath114 with regard to the slowly evolving enveloping amplitude follows from @xmath115\sin\left[\frac{\bar\om\tau+\d_1+\d_2}2\right]$ ] as @xmath116 due to the invariance of equations ( 1 ) under simultaneous action of @xmath117 and @xmath118 the evolution from the lossy side @xmath119 to @xmath120 requires a @xmath0 obtainable from ( [ solution ] ) by sign change @xmath121 .
explicitly this yields eqs . | the beat time @xmath0 associated with the energy transfer between two coupled oscillators is dictated by the bandwidth theorem which sets a lower bound @xmath1 .
we show , both experimentally and theoretically , that two coupled active lrc electrical oscillators with parity - time ( @xmath2 ) symmetry , bypass the lower bound imposed by the bandwidth theorem , reducing the beat time to zero while retaining a real valued spectrum and fixed eigenfrequency difference @xmath3 .
our results foster new design strategies which lead to ( stable ) pseudo - unitary wave evolution , and may allow for ultrafast computation , telecommunication , and signal processing .
one of the fundamental principles of wave physics is the bandwidth theorem @xcite which in quantum mechanics takes the form of the celebrated energy - time heisenberg uncertainty relation @xcite .
a direct consequence of this principle is the fact that the time for evolution between two orthogonal states @xmath0 ( first passage time ) is bounded by @xmath1 @xcite . a basic example where this lower bound can be exhibited is the beat time associated with the energy transfer between two coupled oscillators . even though the validity of the uncertainty principle is undoubted , it has recently been suggested that possible extensions of quantum mechanics invoking non - hermitian @xmath4symmetric hamiltonians @xcite can generate arbitrarily fast state evolution referred to as brachistochrone dynamics @xcite .
the main characteristic of this class of hamiltonians @xmath5 @xcite is that they commute with an anti - linear operator @xmath2 , where the time - reversal operator @xmath6 is the anti - linear operator of ( generalized ) complex conjugation and @xmath7 is a ( generalized ) parity operator @xcite .
examples of such @xmath2-symmetric systems range from quantum field theories to solid state physics and classical optics @xcite . due to the anti - linear nature of the @xmath2 operator , the eigenstates of @xmath5 may or may not be eigenstates of @xmath2 . in the former case ,
all the eigenvalues of @xmath5 are strictly real and the @xmath2-symmetry is said to be _
exact_. otherwise the symmetry is said to be _ spontaneously broken_. in many physical realizations , the transition from the exact to the broken @xmath2 -symmetric phase is due to the presence of various gain / loss mechanisms that are controlled by some parameter @xmath8 of @xmath5 . at the same time , the brachistochrone evolution has a long standing history and is significant both in theory and in application .
it was one of the earliest problems posed in the calculus of variations , and in the framework of classical mechanics it dictates `` the curve down which a particle sliding from rest and accelerated by gravity will slip ( without friction ) from one point to another in the least time '' @xcite .
the quantum mechanical brachistochrone problem has recently been revived in the emerging fields of quantum computation and signal processing where one studies the possibility to use dynamical protocols in solving computational problems and enhancing signal transport respectively .
more specifically , the quantum brachistochrone problem can be formulated as follows : given two orthogonal states @xmath9 and @xmath10 , one wants to _ find _ the ( time - independent ) hamiltonian @xmath11 ( protocol ) that performs the transformation @xmath12 in the minimal time @xmath0 for a fixed difference @xmath13 of the eigenvalues @xmath14 of @xmath11 @xcite .
such a constraint is appropriate since a rescaling of the hamiltonian as @xmath15 , with @xmath16 , would make @xmath17 , and hence the transition rates , large .
this corresponds to the fact that physically only a finite amount of resources ( e.g. a finite magnetic field , bandwidth resolution , etc . )
are typically available .
the quantum brachistochrone was addressed by a number of researchers ( see for example @xcite ) who show that the minimal time to perform the required transformation is bounded by the energy - time uncertainty @xmath18 .
here we adopt these ideas into the realm of classical wave propagation and engineer a system comprised of two active @xmath19 circuits with @xmath2-symmetry having an arbitrarily low first passage time @xmath0 bypassing the lower bound imposed by the bandwidth theorem .
we call this limit - breaking wave phenomenon _ tachistochrone passage _
@xcite where the equivalent system of coupled @xmath2-symmetric oscillators with the same bandwidth @xmath3 can be obtained if one oscillator experiences attenuation with rate @xmath8 while its partner experiences an equivalent amplification rate @xmath8 . depending on the application point of the incident excitation , we observe unidirectional accelerated signal / energy transport , where the time @xmath20 can , in principle , become arbitrarily short .
although these results rely on the fact that the generated dynamics is non - unitary , we mark that due to the @xmath2-symmetry , the corresponding eigenfrequencies @xmath21 can be real , thus guaranteeing the stability of our system @xcite .
our results foster new design strategies based on active elements @xcite with @xmath2 symmetric arrangements that may allow for ultrafast computation , telecommunication , and signal processing .
-symmetric dimer .
the coils are inductively coupled with @xmath22 and @xmath23 providing access to the system variables .
the switch @xmath24 asserts the initial condition .
a resistor provides the damping ( green side ) while a negative resistance gain element ( red side ) is implemented by feedback from an lm356 voltage doubling amplifier .
current flows in the direction of the arrows proportional to the respective voltages @xmath22 and @xmath23 .
, width=312 ] our system , shown in fig .
[ fig : fig2 ] , consists of a pair of coupled @xmath25 circuits , one with amplification and the other with equivalent attenuation .
the circuit was shown in ref .
@xcite to be a simple realization of a @xmath2-symmetric dimer .
each inductor is wound with @xmath26 turns of # 28 copper wire on @xmath27 diameter pvc forms in a @xmath28 loose bundle for an inductance of @xmath29 .
the coils are mounted coaxially with a bundle separation adjusted for the desired mutual inductance @xmath30 .
the isolated natural frequency of each coil is @xmath31 .
the actual experimental circuit deviates from fig . [
fig : fig2 ] in the following ways : ( 1 ) a resistive component associated with coil wire dissipation is compensated by an equivalent gain component applied to each coil ; ( 2 ) a small capacitance trim is included to aid in circuit balancing ; and ( 3 ) additional lm356 voltage followers are used to buffer the voltages @xmath32 and @xmath33 , captured with a tektronix dpo2014 oscilloscope .
the linear nature of the system requires a balance of @xmath2 symmetry only to the extent that component stability over time allows for a measurement .
all circuit modes either exponentially grow to the nonlinearity limit of the buffers , or shrink to zero .
transient data is obtained respecting these time scales .
kirchhoff s laws lead to the following set of equations for the charge @xmath34 ( @xmath35 ) on the capacitor corresponding to the amplified ( lossy ) side : @xmath36 where @xmath37 , @xmath38 , @xmath39 is the gain / loss parameter , and @xmath40 is the rescaled mutual inductance
. inspection of eqs .
( [ kirchhoff2 ] ) reveals that they are invariant under a combined parity ( i.e. @xmath41 ) and time - reversal ( i.e. @xmath42 ) transformation .
-dimer eigenfrequencies vs. the gain / loss parameter @xmath8 , for different mutual inductances @xmath43 ( indicated in the figure ) .
the solid circles are experimental data .
the borders of the highlighted area ( marked by a black dashed line ) indicate a constant bandwidth path with fixed frequency difference @xmath44 . , width=288 ] the theoretical analysis ( see reference @xcite ) relies on a liouvillian formulation of eqs .
( [ kirchhoff2 ] ) which take the form @xmath45 where @xmath46 .
eq .
( [ liuvilian1 ] ) can be interpreted as a schrdinger equation with non - hermitian effective hamiltonian @xmath47 .
this hamiltonian is symmetric with respect to generalized @xmath48 transformations , i.e. @xmath49=0 $ ] , where @xmath50 and @xmath51 is the pauli matrix , @xmath52 is the @xmath53 identity matrix , and @xmath54 denotes the operation of complex conjugation . by a similarity transformation @xmath55 , @xmath56 @xmath57 can be related to a transposition symmetric , @xmath58symmetric hamiltonian @xmath11 .
specifically , @xmath59&=&0,\quad { \mathcal{p}}=\left ( \begin{array}{cccc } 0 & 0 & 0 & 1 \\ 0 & 0 & 1 & 0 \\ 0 & 1 & 0 & 0 \\ 1 & 0 & 0 & 0 \\ \end{array } \right ) = { \cal r}{\mathcal{p}}_0{\cal r}^{-1}\end{aligned}\ ] ] where @xmath60 and @xmath61 , @xmath62 .
this allows us to make contact with the brachistochrone studies of refs .
@xcite .
the eigenfrequencies @xmath63 of system ( [ kirchhoff2 ] ) ( or equivalently eq . ( [ liuvilian1 ] ) )
are shown as functions of the gain / loss parameter @xmath8 , in fig .
[ fig : fig3 ] ( solid lines ) . for @xmath64
the system is in the exact phase and thus the eigenfrequencies are real @xcite .
we investigate the signal / energy tachistochrone passage under the constraint of fixed bandwidth @xmath65 .
experimentally , the @xmath66 constraint is implemented in the @xmath19 dimer through adjustment of the mutual inductance .
figure [ fig : fig3 ] shows both theoretical and experimental results for the parametric evolution of the eigenfrequencies in the exact phase , for various @xmath43 values . the black dashed lines in fig .
[ fig : fig3 ] illustrate a path for fixed @xmath67 through the family of eigenfrequencies associated with different mutual inductances @xmath43 .
eqn .
( [ liuvilian1 ] ) can be solved either analytically or via direct numerical integration in order to obtain the temporal behavior of the capacitor charge @xmath68 and the displacement current @xmath69 in each of the two circuits of the @xmath2-symmetric dimer . for the investigation of the tachistochrone wave evolution
, we consider an initial displacement current in one of the circuits with all other dynamical variables zero .
the first passage time @xmath0 is then defined as the time interval needed to reach an orthogonal state . in our experiments
this corresponds to the condition that the envelope function of the current at the initially excited circuit is zero .
we find that the first passage time is asymmetric with respect to the initially excited circuit .
specifically we have that @xmath70\ ] ] ( see reference @xcite ) where the @xmath71 sign corresponds to an initial condition starting from the gain side while the @xmath72 sign corresponds to an initial condition starting from the lossy side . for @xmath73 , eq . ( [ tautheory ] ) takes the limiting values @xmath74 and @xmath75 respectively .
the latter case indicates the possibility of transforming an initial state to an orthogonal final one , or in more practical terms , transferring energy from one side to the other , in an _ arbitrarily _ short time interval . in the opposite limit of @xmath76
, we recover for both initial conditions the anandan - aharonov lower bound for the first passage time @xmath77 @xcite .
this is the time for which energy is transferred from the initial circuit to its partner according to the constraint of the bandwidth theorem .
geometrically , one can understand the relation ( [ tautheory ] ) in the following way : the time required for the evolution between two states induced by a hermitian hamiltonian is proportional to the length of the shortest geodesic connecting the two states in projective hilbert space @xcite .
non - hermitian @xmath2-symmetric hamiltonians in the exact @xmath2-symmetric domain can be similarity mapped to equivalent hermitian hamiltonians .
under such a similarity mapping the corresponding projective hilbert space undergoes a deformation obtaining a nontrivial metric .
this results in an effective contraction or dilation of the corresponding geodesic and with it of the corresponding evolution time @xcite . in fig .
[ fig : fig4 ] we present some typical measurements for the temporal behavior of displacement currents . in subfigure
[ fig : fig4]a we show @xmath78 for an initial condition corresponding to the case @xmath79 with all other dynamical variables zero .
the case where the initial current excitation is at the lossy side i.e. @xmath80 , is shown for contrast in fig .
[ fig : fig4]b . in both cases ,
agreement between the experiment ( circles ) and the simulations ( lines ) is observed . for comparison
, we also report with black line the temporal behavior of the displacement current for the case of a passive circuit ( i.e. @xmath76 ) with the same @xmath3-constraint .
we observe that the orthogonal target state is reached faster ( or slower ) depending on whether the initial excitation is applied to the lossy ( or gain ) side .
the above results can be verified in more cases by changing the inductive coupling @xmath43 and gain / loss parameter @xmath8 , while keeping constant the frequency difference @xmath81 .
a summary of our measured @xmath0 versus @xmath8 is presented in fig .
[ fig : fig4]c .
the experimental data show agreement with the theoretical prediction eq .
( [ tautheory ] ) .
a parallel analysis pertains directly to the study of the energy transport from one side to another . using the same initial conditions as above we investigate the temporal behavior of the energies @xmath82 of each @xmath83 circuit .
the first passage time can be defined as the time for which the two energies become equal for the first time i.e. @xmath84 . for passive ( i.e. @xmath76 ) coupled circuitry , this time is half of the beating time @xmath85 and it is insensitive to the initial preparation . in contrast , for the active @xmath2-symmetric dimer of fig .
[ fig : fig2 ] , we find that the energy transfer from the lossy ( gain ) side to the gain ( lossy ) one , is faster ( slower ) than the corresponding passive system with the same @xmath3 . in fig .
[ fig : fig4]d , we summarize our measurements for the @xmath86 versus @xmath8 under the constraint of fixed frequency bandwidth @xmath3 . a similar behavior as the one found for the displacement current is evident .
our results open a new direction towards investigating novel phenomena and functionalities of @xmath2-symmetric arrangements with pseudo - unitary _ spatio - temporal _ evolution . along these lines
, we envision @xmath2-symmetric ( nano)-antenna configurations and metamaterial or optical microresonator arrays with unidirectional ultra - fast communication capabilities .
these structures also have potential applications as delay lines , buffers and switches .
questions like the effects of non - linearity or the topological complexity of the @xmath2-symmetric structures in the tachistochrone dynamics are open and offer new exciting opportunities , yet to be discovered .
_ acknowledgments _ we thank d. christodoulides , d. cohen , l. hwel , v. kovanis , and t. morgan for useful discussions .
this research was supported by an afosr no .
fa 9550 - 10 - 1 - 0433 grant , by an nsf eccs-1128571 grant and by a wesleyan project grant . |
in the three articles @xcite , m. douglas , b. shiffman , and s. zelditch presented their research into the statistics of critical points of random holomorphic sections of line bundles over complex manifolds .
this line of research was initially motivated by the vacuum selection problem in string theory , and thus there was much discussion in these papers about the application of these statistics to string theory and related fields .
however , this research is also of independent interest from a geometric point of view .
the purpose of the present article is to prove the main conjecture in @xcite which concerns the purely geometric consideration of determining the metric dependence and asymptotic minimization of the expected number of critical points of random sections of positive line bundles .
while it is clear that the number of critical points of a given holomorphic section will vary with the metric , it is not clear whether the same is true for the expected number of critical points of a random section .
therefore , in @xcite , the authors performed an asymptotic analysis of the expected number @xmath0 of critical points of random holomorphic sections of @xmath4 in order to determine the metric dependence of this statistic .
they showed that the asymptotic expansion of @xmath0 is topologically invariant to at least two orders in @xmath1 , but that the third term in the expansion is a sum of a topological invariant and a universal constant , @xmath2 , times the calabi functional @xmath5 , where @xmath6 is the scalar curvature of the khler form @xmath7 .
the authors derived a complicated integral formula for @xmath2 and , based on their calculations in low dimensions and a heuristic argument , conjectured that @xmath8 for all @xmath3 . in this paper
we will prove their conjecture .
the fact that @xmath2 does not vanish proves that the expansion is non - topological , with a metric dependence in the third term for all @xmath3 .
the fact that @xmath2 is positive proves that , whenever it exists , the calabi extremal metric asymptotically minimizes @xmath9 .
it would seem likely that metrics which minimize the expected number of critical points would be ideally suited for the vacuum selection problem .
in this paper we consider the @xmath1th tensor power of a positive hermitian line bundle @xmath10 over a compact khler manifold of dimension @xmath3 .
the khler form is given by @xmath7 , where @xmath11 is the curvature form of the metric on the bundle .
for a holomorphic section @xmath12 , we consider the connection critical points given by @xmath13 , where @xmath14 is chosen to be the chern connection of @xmath15 .
we will use @xmath16 to denote the set of critical points of @xmath17 . in general , the critical point equation given above
is not holomorphic . in a local frame @xmath18
we can write @xmath19 and then @xmath20 , where @xmath21 is the khler potential . from this
we see that the critical point equation in the local frame is @xmath22 , which is holomorphic only when @xmath23 is .
therefore the cardinality of @xmath16 is a non - constant random variable on the space @xmath24 .
next , we endow the space @xmath24 with the gaussian measure @xmath25 given by @xmath26 where @xmath27 is lebesgue measure and @xmath28 is an orthonormal basis of @xmath24 relative to the inner product @xmath29 induced by @xmath15 on @xmath24 .
we define the expected distribution of critical points of @xmath30 with respect to @xmath25 to be @xmath31\,d{\gamma}_n(s),\ ] ] where @xmath32 is the dirac point mass at @xmath33 , and then the expected number of critical points is given by @xmath34 we recall that the morse index @xmath35 of a critical point of a real - valued function is given by the number of negative eigenvalues of its hessian at the point , and that for a positive line bundle @xmath36 @xcite .
since the critical points of @xmath17 with respect to @xmath14 are the same as those of the real - valued function @xmath37 , we can consider the morse indices of the critical points in @xmath16 . to this end , we let @xmath38 denote the expected number of critical points of morse index @xmath35 . it follows that @xmath39 in @xcite , the authors derived the complete asymptotic expansion of @xmath38 .
[ dsztheo1 ] let @xmath40 be a positive holomorphic line bundle on a compact khler manifold , with @xmath41 , then the expected number of critical points of morse index @xmath35 ( @xmath42 ) of random sections in @xmath24 has the asymptotic expansion @xmath43n^m + \left[\frac{\pi^m{\beta}_{1q}}{(m-1)!}\ , c_1(m)\cdot c_1(l)^{m-1}\right]n^{m-1 } \nonumber \\ & & \
+ \biggl[{\beta}_{2q}\int_m\rho_h^2d{{\operatorname{vol}}}_h + { \beta}'_{2q}\ , c_1(m)^2\cdot c_1(l)^{m-2 } + { \beta}''_{2q}\ , c_2(m)\cdot c_1(l)^{m-2}\biggr ] n^{m-2}+\cdots\,,\nonumber\end{aligned}\ ] ] where @xmath44 are universal constants depending only on the dimension @xmath3 , and @xmath6 is the scalar curvature of @xmath45 .
for the sake of clarity we state the obvious corollary : [ dsztheo2 ] under the same conditions as above , the expected number of critical points of random sections in @xmath24 has the asymptotic expansion @xmath46n^m + \left[\frac{\pi^m{\beta}_{1}}{(m-1)!}\ , c_1(m)\cdot c_1(l)^{m-1}\right]n^{m-1 } \nonumber \\ & & \ + \biggl[{\beta}_{2}\int_m\rho_h^2d{{\operatorname{vol}}}_h +
{ \beta}'_{2}\ , c_1(m)^2\cdot c_1(l)^{m-2 } + { \beta}''_{2}\ , c_2(m)\cdot c_1(l)^{m-2}\biggr ] n^{m-2}+\cdots\,,\nonumber\end{aligned}\ ] ] where @xmath47 denote the sum over @xmath35 of @xmath44 , respectively . in our previous paper ( @xcite )
we examined the constant @xmath48 in the leading term of the above expansion .
we started with an integral formula for @xmath49 , which was derived in @xcite , and simplified it using an argument based on the itzykson - zuber integral formula .
we then used a variant of the selberg integral formula to obtain an exact formula for the @xmath50 case and , finally , showed that the value of the integral decreased as @xmath35 increased using a change of variable argument .
this gave us upper and lower bounds on the leading term of the expansion of @xmath0 .
[ conj ] let @xmath51 denote the leading coefficient in the expansion of @xmath38 , and let @xmath52 , so that @xmath53 then @xmath54 and hence @xmath55 using similar techniques , we also derived the growth rate of @xmath0 as the dimension of the manifold gets large in the special case of positive line bundles over complex projective space . in this paper
we will focus our attention on the constant @xmath56 .
we see from corollary [ dsztheo2 ] that the first two terms in the expansion are topological and that the only non - topological part of the third term is @xmath57 times the calabi functional @xmath5 .
hence , in order to determine the metric dependence of the expansion we will need to evaluate @xmath57 . as the analysis is simplified by treating the contributions from critical points of different morse indices separately , the following integral formula for @xmath58 was derived in @xcite .
[ l1 ] @xmath59 where @xmath60\frac 1{\left(1 - \frac i2 \sum_j\xi_j\right)}+ \frac 2{(m+1)(m+3)\left(1 - \frac i2 \sum_j\xi_j\right)^2 } } { \left(1 - \frac i2 \sum_j\xi_j\right)\prod_{j\le k}\left [ 1 + \frac i2 ( \xi_j+\xi_k)\right ] } \;.\ ] ] here , @xmath61 , @xmath62 is the vandermonde determinant , @xmath63 is the diagonal matrix with diagonal entries @xmath64 , and @xmath65for ( hermitian ) @xmath66 matrices @xmath67 . the iterated @xmath68 integrals are defined in the distribution sense .
we ask the interested reader to refer to @xcite for additional background information .
in 2 we evaluate the @xmath69 integrals in the above formula for @xmath70 with an iterated residues technique to obtain the following simplification .
[ l2 ] @xmath71 where @xmath72 and , for @xmath73 , @xmath74 we then use an extension of the exponential selberg integral to calculate the exact formula for the @xmath50 case in 3 .
[ l3 ] for @xmath75 and @xmath50 , @xmath76 unfortunately , due to the presence of the additional exponential term in @xmath77 when @xmath73 , direct application of a selberg integral formula is not possible for the other cases .
although it would be possible to calculate the exact formulas for one or two more cases , these would be extremely complicated and we would be no closer to our goal of showing that @xmath78 for all @xmath3 .
instead we consider the sum over the other indices and denote @xmath79 from lemma [ l2 ] we immediately see that @xmath80 where @xmath77 is given by . in
4 we use a change of variable argument and then exploit the symmetries in the resulting integrand to show that : [ l2a ] @xmath81 @xmath82 and for @xmath83 , @xmath84 where @xmath85 in 5 , we prove our main result by showing that the above integrals are positive for all @xmath3 .
[ t1 ] the constant @xmath86 is strictly positive for @xmath75 .
we would like to point out that the positivity of @xmath58 , for each @xmath35 , was also conjectured in @xcite .
although it seems almost certain that this is true , this remains an open problem . using the above result we can extend theorem 1.6 and corollary 1.7 from @xcite in the @xmath0 case to all dimensions .
we let @xmath87 denote the class of positively curved metrics , i.e. metrics for which @xmath88 is a positive ( 1,1)-form . as was noted in @xcite
, we would not expect @xmath89 to have an upper bound as @xmath90 varies over @xmath87 , however it is bounded from below by @xmath91 .
it is therefore of interest to determine when a metric which minimizes @xmath89 , at least in an asymptotic sense , exists . to this end
, we make the definition : let @xmath92 be an ample holomorphic line bundle over a compact khler manifold . for @xmath93 , we say that @xmath89 is _ asymptotically minimal _
if for all @xmath94 in @xmath95 , there exists @xmath96 such that @xmath97 as we noted before , the first non - topological term in the asymptotic expansion of @xmath89 is @xmath98 .
thus , to find minimizers of @xmath89 we need to find minimizers of the calabi functional .
we note that by results in @xcite all critical points of the calabi functional are local minima .
those that obtain the absolute minimum in a fixed khler class are called calabi extremal metrics .
therefore , the following theorem is an immediate consequence of theorem [ t1 ] .
[ dsztheo3 ] let @xmath99 and let @xmath100 , then @xmath89 is asymptotically minimal whenever @xmath101 is an extremal calabi metric . by a result of calabi in @xcite , metrics with constant scalar curvature
are extremal , and by a result of donaldson in @xcite , there is at most one khler metric of constant scalar curvature in the cohomology class @xmath102 .
therefore , we have the corollary : [ dsztheo4 ] suppose that @xmath103 possesses a metric @xmath90 for which the scalar curvature of @xmath100 is constant , then @xmath90 is the unique metric on @xmath103 such that @xmath89 is asymptotically minimal.[calabi ] due to the technical nature of this paper and the many tedious calculations in the proofs below , we have provided a mathematica worksheet with the arxiv.org posting that provides verification in low dimensions of various intermediate steps .
it calculates the values of @xmath58 and @xmath104 in low dimensions using the original formula from @xcite and compares these with the values it calculates from the formulas given in lemma [ l2 ] , lemma [ l3 ] , lemma [ l2a ] , , and .
these results are part of the author s ongoing thesis research at the johns hopkins university which is being advised by steve zelditch .
the proof of this lemma will consist of evaluating the inner integrals in using an iterated residues technique . to this end ,
we rewrite @xmath70 as @xmath105 { \mathcal{i}}_{\lambda,2 } + \frac { 2 { \mathcal{i}}_{\lambda,3}}{(m+1)(m+3 ) } \right ) \ ; , \nonumber \end{aligned}\ ] ] where @xmath106}.\ ] ] then we make the change of variables @xmath107 to obtain @xmath108 where @xmath109 putting this together we have @xmath110 to complete the proof we apply the following lemma with @xmath111 and @xmath112 to and simplify to obtain the statement of lemma [ l2 ] . [ la1 ] let @xmath113 , @xmath114 , and @xmath115 , then for @xmath116 we have @xmath117 \displaystyle i^{m^2-s}\,\frac{\pi^m}{c^s } & \mbox{for } \
p = m \end{array}\right.\ , \ ] ] where @xmath118 \displaystyle 1-c{\lambda}_m & \mbox{for } \ s=2 \\[10pt ] \displaystyle \frac{c^2 { \lambda}_m^2 - 2 c { \lambda}_m + 2}{2 } & \mbox{for } \
s=3 \end{array}\right.\ .\ ] ] the case @xmath119 is essentially lemma 4.3 from @xcite .
we will extend the proof of that lemma to prove the other cases . to do so we let @xmath120 and @xmath121 so that @xmath122 when @xmath123 , we start by evaluating the @xmath124 integral .
we close the contour of integration in the upper half plane and pick up the poles at @xmath125 , and at @xmath126 for @xmath127 .
the pole at @xmath128 is below the contour .
the @xmath126 poles do not contribute to the integral . to see why , we compute the residue at the pole @xmath129 to obtain @xmath130 } 2t_2(t_2+t_3)\cdots(t_2+t_m)\delta(t_2,\dots , t_m ) } { ( t_3+\cdots + t_m+ci)^s\,2t_2(-t_2+t_3)\cdots(-t_2+t_m)\prod_{2\le j \le k\le m } ( t_j+t_k)}\\ = \frac { e^{i(\lambda_2-\lambda_1)t_2}}{2t_2}\ , { \mathcal{i}}({\lambda}_3,\dots,{\lambda}_m , t_3,\dots , t_m;s , c)\;.\end{gathered}\ ] ] it is easy to see that the integral of the above formula is zero , since to calculate the @xmath131 integral we would need to close the contour in the lower half plane and then the lone pole at @xmath132 would be above the contour . by the symmetry in @xmath133
we could have replaced @xmath131 in the above argument with any of the other @xmath134 s and obtained the same result .
this leaves only the pole at @xmath125 , and the residue of @xmath133 at this pole is @xmath135 if we apply recursively , we see that the integral with @xmath136 is reduced to the case with all @xmath137 s negative : @xmath138 when @xmath139 , we compute @xmath140 and see that @xmath141 to calculate the integral in , we start with the @xmath142 integral and close the contour in the lower half plane , picking up the pole of order @xmath17 at @xmath143 . these residues are @xmath144 and @xmath145 , \end{gathered}\ ] ] for @xmath146 and @xmath147 , respectively . next , we evaluate the @xmath148 integral .
we close the contour in the upper half plane and see that all of the denominatorial factors in which the summand @xmath149 appears either cancel out or give poles below the contour .
it can then be verified by a straightforward ( if somewhat tedious ) calculation that the poles @xmath150 do not contribute to the value of the integral . indeed ,
after computing the residue at the pole @xmath150 , consider the @xmath134 integral .
the coefficient of @xmath134 in the exponential will be @xmath151 , which is always negative , and thus the contour can be closed in the lower half plane .
all of the poles will be above the contour , since all of the denominatorial factors with an @xmath149 will have canceled out , and therefore the integral will be zero .
this leaves only the pole at @xmath152 , and we calculate that @xmath153 for @xmath120 .
we apply this argument recursively and then compute the residue of @xmath154 at @xmath155 to obtain @xmath156 and @xmath157 for @xmath146 and @xmath147 , respectively . substituting these formulas into and
simplifying gives the desired result .
we see from lemma [ l2 ] that @xmath158 using , we can rewrite this as @xmath159 where @xmath160 next , we see that making the change @xmath161 in the integrand above does not change its value on the region over which we are integrating . after doing this , we notice that the integrand is now symmetric under permutations of @xmath162 , allowing us to take the integral over @xmath163 and replace each of the sums with a multiple of one of the summands .
thus @xmath164 now we need a lemma , which will be proved in the subsection below .
[ l4 ] @xmath165 where @xmath166 is the rising factorial .
this is valid for integer @xmath167 with @xmath168 and complex @xmath169 and @xmath170 with re @xmath171 , re @xmath172 -min @xmath173
we apply this lemma to with @xmath174 , @xmath175 , and @xmath176 for each of the four cases : @xmath177 , @xmath178 , @xmath179 , and @xmath180 to obtain @xmath181 after applying gauss s multiplication formula and simplifying we have @xmath182 substituting in for @xmath183 gives the desired result .
first we let @xmath184 @xmath185 and @xmath186 then we take the partial derivative with respect to @xmath187 of the integrand in the formula for @xmath188 , noting that by the fundamental theorem of calculus this will then integrate to zero .
thus , @xmath189 here we used the fact that @xmath190 since @xmath191 is symmetric under permutations of the @xmath192 s , when @xmath193 , we see that @xmath194 continuing to utilize the symmetry , we make the transposition @xmath195 for each of the terms in the sum in . when @xmath196 , we see that this term is zero since @xmath197 when @xmath198 , we have @xmath199 and therefore @xmath200 finally , when @xmath201 , @xmath202 here we applied the transposition @xmath195 to the second integral and by the reasoning in the second term in this integrand disappears .
applying these results to we see that @xmath203 we then iterate this relation over @xmath204 to achieve @xmath205 next , we apply the same technique to @xmath206 , and we see that @xmath207 using the arguments above , we have @xmath208 thus , @xmath209 combining this with produces the relation @xmath210 to complete the proof , we note that @xmath211 is the so - called exponential selberg integral , for which we have the following formula ( see @xcite ) : [ selbergcor ] for any positive integer m , let @xmath212 then @xmath213 valid for complex @xmath169 , @xmath170 with re @xmath171 , re @xmath172 -min @xmath214 .
we recall that from lemma [ l2 ] we have @xmath80 where @xmath215 for each @xmath35 , we make the change of variables @xmath216 in the integral over @xmath217 above .
this change of variables is a composition of the following two changes of variables : @xmath218 and @xmath219 these changes take @xmath220 and @xmath221 , respectively , and thus all of the integrals will now be over a common region of integration , @xmath222 . under this change of variables we see that @xmath223 is unchanged and that @xmath224 @xmath225 and @xmath226 here we have let @xmath227 to simplify the notation .
therefore , since the absolute value of the jacobian is 1 , we have @xmath228 where @xmath229 and @xmath230 next we see that making the change @xmath161 in does not change the value of the integrand on the region over which we are integrating .
we make this change and now the integrand is symmetric under permutations of @xmath162 , so we can take the integral over @xmath163 .
thus @xmath231 the symmetry also allows us to replace any symmetric sum with a multiple of one of its summands .
therefore @xmath232 to obtain the desired formulas , we see that @xmath233 for @xmath234 , @xmath235 for @xmath236 , and finally for @xmath237 , we compute that @xmath238 where once again we have replaced sums with multiples of one of their summands .
in [ c1 ] we showed that @xmath239 was positive , so to prove this theorem we need to prove the positivity of @xmath240 . when @xmath234 , there is nothing to prove since the integrand in is clearly positive . for the other cases we need the following lemma which we will prove in [ e2 ] .
[ l5 ] let @xmath241 , then we have the following identities : @xmath242 @xmath243 @xmath244 where @xmath245 is the dirac delta function , @xmath246 @xmath247 @xmath248 the first three identities hold for @xmath249 and the last three for @xmath83 . when @xmath83 , we apply , , and to to obtain @xmath252 we then apply and to this and we have @xmath253 by the middle term vanishes and therefore @xmath254 it is now clear that @xmath255 is positive .
indeed , by computing the @xmath187 integral we see that @xmath256 for each of these identities the proof will consist of taking a partial derivative with respect to some @xmath257 inside the integral on the lhs of the equation . by the fundamental theorem of calculus
this will integrate to zero .
we compute the derivative and manipulate the result utilizing the symmetry in @xmath258 to achieve the desired formula .
we will also make use of the fact that @xmath259 so , for we have @xmath260 since @xmath258 is symmetric under permutations of @xmath162 , for @xmath261 , we make the transposition @xmath262 and see that @xmath263 and therefore @xmath264 combining and gives the desired result .
next we have @xmath265 in the last equality we used the fact that , for @xmath261 , @xmath266 here the second term in the second integral vanishes since the transposition @xmath262 just changes the sign of the integrand .
we then apply to to obtain . for the fourth identity
we have @xmath270 this time we applied and used the following facts which follow from the symmetry in @xmath258 as was demonstrated above : @xmath271 and @xmath272 once again we apply to to obtain .
e. calabi , _ extremal khler metrics _ , in `
seminar on differential geometry .
papers presented at seminars held during the academic year 19791980 ' , edited by s .-
annals of mathematics studies , 102 , princeton university press , princeton , n.j , 1982 , 259290 . | we prove the main conjecture from @xcite concerning the metric dependence and asymptotic minimization of the expected number @xmath0 of critical points of random holomorphic sections of the @xmath1th tensor power of a positive line bundle .
the first non - topological term in the asymptotic expansion of @xmath0 is the the calabi functional multiplied by the constant @xmath2 which depends only on the dimension of the manifold .
we prove that @xmath2 is strictly positive in all dimensions , showing that the expansion is non - topological for all @xmath3 , and that the calabi extremal metric , when it exists , asymptotically minimizes @xmath0 . |
fock and goncharov conjectured that the algebra of functions on a cluster variety has a canonical vector space basis parameterized by the tropical points of the mirror cluster variety .
unfortunately , as shown in @xcite by the first three authors of this paper , this conjecture is usually false : in general the cluster variety may have far too few global functions .
one can only expect a power series version of the conjecture , holding in the `` large complex structure limit , '' and honest global functions parameterized by a subset of the mirror tropical points .
for the conjecture to hold as stated , one needs further affineness assumptions . here
we apply methods developed in the study of mirror symmetry , in particular _
scattering diagrams _ , introduced by kontsevich and soibelman in @xcite for two dimensions and by gross and siebert in @xcite for all dimensions , _ broken lines _ , introduced by gross in @xcite and developed further by carl , pumperla and siebert in @xcite , and _ theta functions _ , introduced by gross , hacking ,
keel and siebert , see @xcite , @xcite , @xcite , and @xcite , to prove the conjecture in this corrected form .
we give in addition a formula for the structure constants in this basis , non - negative integers given by counts of broken lines .
definitions of all these objects , essentially combinatorial in nature , in the context of cluster algebras will be given in later sections .
here are more precise statements of our results . for basic cluster variety notions we follow the notation of @xcite , 2 , for convenience , as we have collected there a number of definitions across the literature ; nothing there is original .
we recall some of this notation in appendices [ ldsec ] and [ rdsec ] .
the various flavors of cluster varieties are all varieties of the form @xmath5 , where @xmath6 is a copy of the algebraic torus @xmath7\ ] ] over a field @xmath8 of characteristic zero , and @xmath9 is a lattice , indexed by @xmath10 running over a set of _ seeds _ ( a seed being roughly an ordered basis for @xmath11 ) .
the birational transformations induced by the inclusions of two different copies of the torus are compositions of _ mutations_. fock and goncharov introduced a simple way to dualize the mutations , and using this define the _ _ fock - goncharov dual _ _ , @xmath12 .
we write @xmath13 for the tropical semi - field of integers under @xmath14 .
there is a notion of the set of @xmath13-valued points of @xmath15 , written as @xmath16 .
this can also be viewed as being canonically in bijection with @xmath17 , the set of divisorial discrete valuations on the field of rational functions of @xmath15 where the canonical volume form has a pole , see [ tropsec ] .
each choice of seed @xmath10 determines an identification @xmath18 .
our main object of study is the _
@xmath19 cluster variety with principal coefficients _ , @xmath20 .
( see appendices [ ldsec ] and [ rdsec ] for notation . )
this comes with a canonical fibration over a torus @xmath21 , and a canonical free action by a torus @xmath22 .
we let @xmath23 .
the fibre @xmath24 ( @xmath25 the identity ) is the fock - goncharov @xmath19 variety ( whose algebra of regular functions is the fomin - zelevinsky upper cluster algebra ) .
the quotient @xmath26 is the fock - goncharov @xmath27 variety .
[ globalmonomialdef ] a _ global monomial _ on a cluster variety @xmath28 is a regular function on @xmath15 which restricts to a character on some torus @xmath6 in the atlas . for @xmath15
an @xmath19-type cluster variety a global monomial is the same as a cluster monomial .
one defines the _ upper cluster algebra _ @xmath29 associated to @xmath15 by @xmath30 , and the _ ordinary cluster algebra _
@xmath31 to be the subalgebra of @xmath29 generated by global monomials .
for example , @xmath32 is the original cluster algebra defined by fomin and zelevinsky in @xcite , and @xmath33 is the corresponding upper cluster algebra as defined in @xcite .
given a global monomial @xmath34 on @xmath15 , there is a seed @xmath10 such that @xmath35 is a character @xmath36 , @xmath37 . because the seed @xmath10 gives an identification of @xmath38 with @xmath39 , we obtain an element @xmath40 , which we show is well - defined ( independent of the open set @xmath6 ) , see lemma [ fgccthg2 ] .
this is the _
@xmath41-vector _ of the global monomial @xmath34 .
we show this notion of @xmath41-vector coincides with the notion of @xmath41-vector from @xcite in the @xmath19 case , see corollary [ gvecthm ] .
let @xmath42 be the set of @xmath41-vectors of all global monomials on @xmath15 .
finally , we write @xmath43 for the @xmath8-vector space with basis @xmath44 , i.e. , @xmath45 ( where @xmath46 for the moment indicates the abstract basis element corresponding to @xmath47 ) . fock and goncharov s dual basis conjecture says that @xmath43 is canonically identified with the vector space @xmath29 , and so in particular @xmath43 should have a canonical @xmath8-algebra structure .
note that such an algebra structure is determined by its structure constants , a function @xmath48 such that for fixed @xmath49 , @xmath50 for all but finitely many @xmath51 and @xmath52 with this in mind , we have : [ mainth ] let @xmath15 be one of @xmath53 .
the following hold : 1 .
there are canonically defined non - negative _ structure constants _ @xmath54 these are given by counts of _ broken lines _
, certain combinatorial objects which we will define .
the value @xmath55 is not taken in the @xmath27 or @xmath56 case .
2 . there is a canonically defined subset @xmath57 with @xmath58 such that the restriction of @xmath59 gives the vector subspace @xmath60 with basis indexed by @xmath61 the structure of an associative commutative @xmath8-algebra .
3 . @xmath62 , i.e. , @xmath61 contains the @xmath41-vector of each global monomial .
4 . for the lattice structure on @xmath44 determined by any choice of seed , @xmath57 is closed under addition .
furthermore , @xmath57 is saturated : for @xmath63 and @xmath64 , @xmath65 if and only if @xmath66 .
there is a canonical @xmath8-algebra map @xmath67 which sends @xmath46 for @xmath68 to the corresponding global monomial .
the image @xmath69 is a universal positive laurent polynomial ( i.e. , a laurent polynomial with non - negative integral coefficients in the cluster variables for each seed )
@xmath70 is injective for @xmath71 or @xmath72 .
furthermore , @xmath70 is injective for @xmath73 under the additional assumption that there is a seed @xmath74 for which all the covectors @xmath75 , @xmath76 , lie in a strictly convex cone .
when @xmath70 is injective we have canonical inclusions @xmath77 there is an analog to theorem [ mainth ] for @xmath78 ( the main difference is that the _ theta functions _ ,
i.e. , the canonical basis for @xmath79 , are only defined up to scaling each individual element , and the structure constants will not in general be integers ) .
injectivity in ( 7 ) holds for very general @xmath78 .
see theorem [ mainthax ] .
note that ( 5 - 6 ) immediately imply : [ poslp ] each cluster variable of an @xmath19-cluster algebra is a laurent polynomial with non - negative integer coefficients in the cluster variables of any given seed .
this was conjectured by fomin and zelevinsky in their original paper @xcite .
positivity was obtained independently in the skew - symmetric case by [ ls13 ] , by an entirely different argument . in our proof
the positivity in ( 1 ) and ( 6 ) both come from positivity in the scattering diagram , a powerful tool fundamental to the entire paper
. see theorem [ scatdiagpositive ] .
we conjecture that injectivity in ( 7 ) holds for all @xmath78 ( without the convexity assumption ) .
note ( 7 ) includes the linear independence of cluster monomials , which has already been established ( without convexity assumptions ) for skew - symmetric cluster algebras in @xcite , by a very different argument .
the linear independence of cluster monomials in the principal case also follows easily from our scattering diagram technology , as pointed out to us by greg muller .
see theorem [ properlaurentproperty ] .
when there are frozen variables , one obtains a partial compactification @xmath80 ( where the frozen variables are allowed to take the value @xmath81 ) for @xmath82 , @xmath56 or @xmath78 .
the notions of @xmath83 , @xmath84 , @xmath85 , and @xmath86 extend naturally to @xmath87 .
see construction [ fvss ] .
of course if @xmath88 , and we have injectivity in ( 7 ) , @xmath89 has a canonical basis @xmath61 with the given properties
. also , @xmath88 implies , under certain hypotheses , @xmath90 , see lemma [ midocalem ] .
such partial compactifications are essential for representation - theoretic applications : [ slrexample ] let @xmath91 .
choose a borel subgroup @xmath92 of @xmath93 , @xmath94 a maximal torus , and let @xmath95 $ ] be the unipotent radical of @xmath92 .
these choices determine a cluster variety structure ( with frozen variables ) on @xmath96 , with @xmath97 , the ring of regular functions on @xmath98 , see @xcite , 10.4.2 .
theorem [ mainth ] implies that these choices canonically determine a vector space basis @xmath99 .
each basis element is an @xmath100-eigenfunction for the natural ( right ) action of @xmath100 on @xmath98 .
for each character @xmath101 , @xmath102 is a basis of the weight space @xmath103 .
the @xmath104 are the collection of irreducible representations of @xmath93 , each of which thus inherits a basis , canonically determined by the choice of @xmath105 .
we give , combining our results with results of t. magee , much more precise results , see corollary [ slrcor ] below .
canonical bases for @xmath106 have been constructed by lusztig . here
we will obtain bases by a procedure very different from lusztig s , as a special case of the more general @xcite , conjecture 0.6 , which applies in theory to any variety with the right sort of volume form .
see remark [ clusterindependenceremark ] for further commentary on this .
the tools necessary for the proof of theorem [ mainth ] are developed in the first six sections of the paper , with the proof given in [ midclustersection ] . this material is summarized in more detail in [ intro1.2 ] .
the second part of the paper turns to criteria for the full fock - goncharov conjecture to hold .
precisely : [ fgconjdef ] we say the _ full fock - goncharov conjecture holds _ for a cluster variety @xmath15 if the map @xmath107 of theorem [ mainth ] is injective , @xmath108 note this implies @xmath109 .
we prove a number of criteria which guarantee the full fock - goncharov conjecture holds .
one such condition , which seems to be very natural in our setup and is implied , say , by the existence of a maximal green sequence , is : [ proposition [ egmscprop ] ] if the set @xmath110 of all @xmath41-vectors of global monomials of @xmath111 in @xmath112 is not contained in a half - space under the identification of @xmath112 with @xmath113 induced by some choice of seed , then the full fock - goncharov conjecture holds for @xmath56 , @xmath27 , very general @xmath78 and , if the convexity condition ( 7 ) of theorem [ mainth ] holds , for @xmath19 .
many of the results in the second part of the paper are proved using a generalized notion of convex function or convex polytope , see [ intro1.3 ] and [ intro1.4 ] for more details . in [ candsec ] , we turn to results on partial compactifications .
we first explain how convex polytopes in our sense give rise , under suitable hypotheses , to compactifications of @xmath19-type cluster varieties and toric degenerations of such .
this connects our constructions to the mirror symmetry picture described in @xcite , and in particular describes a partial compactification of @xmath56 as giving a degeneration of a family of log calabi - yau varieties to a toric variety .
partial compactifications via frozen variables are also important in representation theoretic applications , as already indicated in example [ slrexample ] .
we prove results for such partial compactifications which , combined with recent results of t. magee @xcite , @xcite , yield strong representation - theoretic results , see [ intro1.4 ] for more details .
we now turn to a more detailed summary of the contents of the paper .
[ scatdiagsection ] is devoted to the construction of the fundamental tool of the paper , _ scattering diagrams_. while @xcite defined these in much greater generality , here they are collections of walls living in a vector space with attached functions constructed canonically from a choice of seed data .
a precise definition can be found in [ defconstsection ] .
here we simply highlight the main new result theorem [ scatdiagpositive ] , whose proof , being fairly technical , is deferred to appendix [ scatappendix ] .
this says that the functions attached to walls of a scattering diagram associated to seed data have positive coefficients .
all positivity results in this paper flow from this fundamental observation , and indeed many of our arguments use this in an essential way . for the reader s convenience , we give in [ consistentscatdiagramsection ] an elementary construction of the relevant scattering diagrams , drawing on the method given in @xcite . since a scattering diagram depends on a choice of seed , [ mutinvsec ] shows how scattering diagrams associated to mutation equivalent seeds are related .
this shows that a scattering diagram has a chamber structure indexed by seeds mutation equivalent to the initial choice of seed . in [ tropsec ]
, we review some notions of tropicalizations of cluster varieties , showing that scattering diagrams naturally live in such tropicalizations . indeed , the scattering diagram which is associated to a cluster variety @xmath15 lives naturally in the tropical space of the fock - goncharov dual @xmath114 .
these tropicalizations , crucially , can only be viewed as piecewise linear , rather than linear , spaces , with a choice of seed giving an identification of the tropicalization with a linear space .
already the mutation combinatorics becomes apparent : [ fgcclemth ] for each seed @xmath74 of a @xmath19-cluster variety , the _
( fock - goncharov ) cluster chamber _ associated to @xmath10 is @xmath115 where @xmath116 denotes the tropicalization of the monomial @xmath117 , see [ tropsec ] .
the collection @xmath118 of such subsets of @xmath119 over all mutation equivalent seeds form the maximal cones of a simplicial fan , the _
( fock - goncharov ) cluster complex_. the fomin - zelevinsky exchange graph is the dual graph of this fan .
the collection of cones @xmath118 was introduced by fock and goncharov , who conjectured they formed a fan .
it is not at all obvious from the definition that the interiors of the cones can not overlap .
our description of the chamber structure induced by a scattering diagram in fact shows that part of the chamber structure coincides with the collection of cones @xmath118 .
this shows the fact they form a fan directly .
in addition , the set @xmath110 of theorem [ mainth ] consists of the integral points of the union of cones in @xmath118 . [ blsec ] gives the definition of broken line , the second principal combinatorial tool of the paper .
these were originally introduced in @xcite and developed further in @xcite as tropical replacements for maslov index two disks . in @xcite , they were used to define _ theta functions _ , which are , in principle , formal sums over all broken lines with fixed boundary conditions .
the relevance of theta functions for us comes in [ basection ] . here
we show the direct relationship between scattering diagrams and the @xmath19 cluster algebra .
we show that if we associate a suitable torus @xmath120 to each chamber of the scattering diagram associated to a mutation of the initial seed , then the walls separating the chambers can be interpreted as giving birational maps between these tori .
gluing together these copies of @xmath120 gives the @xmath19 cluster variety , see theorem [ rpth ] .
further , a theta function @xmath121 depends on a point @xmath122 .
if for a given choice of @xmath123 , @xmath121 is in fact a finite sum , then @xmath121 is a global function on @xmath19 .
we show that this holds in particular when @xmath123 lies in the cluster complex @xmath118 , and in this case @xmath121 agrees with the cluster monomial with @xmath41-vector given by @xmath123 .
because of the positivity result theorem [ scatdiagpositive ] , @xmath121 is in any event always a power series with positive coefficients .
thus we get positivity of the laurent phenomenon , theorem [ pilpth ] , as an easy consequence of our formalism .
in [ cgvectorsec ] we begin with what is another essential observation for our approach .
a choice of initial seed @xmath10 provides a partial compactification @xmath124 of @xmath56 by allowing the variables @xmath125 ( the principal coefficients ) to be zero .
these variables induce a flat map @xmath126 , with @xmath19 being the fibre over @xmath127 .
our methods easily show : [ cftth ] ( corollary [ mutcor ] , ( 1 ) ) the central fibre @xmath128 is the algebraic torus @xmath129 . though immediate from our scattering diagram methods , the result is not obvious from the original definitions : indeed , it is equivalent to the sign - coherence of @xmath130-vectors , see corollary [ sccveccor ] .
the last major ingredient in the proof of theorem [ mainth ] is a formal version of the fock - goncharov conjecture . as mentioned above
, this conjecture does not hold in general , but in [ formalfgcsec ] , we show that the fock - goncharov conjecture holds in a formal neighbourhood of the torus fibre of @xmath131 .
we show the structure constants given in theorem [ mainth ] , ( 1 ) , have a tropical interpretation and determine an associative product on @xmath132 , except that @xmath133 will in general be an infinite sum of theta functions .
further , canonically associated to each universal laurent polynomial @xmath134 is a formal power series @xmath135 which converges to @xmath41 in a formal neighbourhood of the central fibre . for the precise statement see theorem [ ffgth ] , which we interpret as saying that the fock - goncharov dual basis conjecture always holds in the _ large complex structure limit_. this is all one should expect from log calabi - yau mirror symmetry , in the absence of further affineness assumptions .
a crucial point , shown in the proof of theorem [ ffgth ] , is that the expansion of @xmath136 is independent of the choice of seed @xmath10 determining the compactification @xmath124 , i.e. , is independent of which degeneration is used to perform the expansion . in [ midclustersection ]
, we introduce the middle cluster algebra @xmath137 .
the idea is that while we do nt know that every regular function on @xmath56 can be written as a linear combination of theta functions , there is a set @xmath138 indexing those @xmath123 for which @xmath121 is a regular function on @xmath56 .
these in fact yield a vector space basis for a subalgebra of @xmath139 which necessarily includes all cluster monomials , hence includes the ordinary cluster algebra . with this in hand ,
theorem [ mainth ] becomes a summary of the results proved up to this point .
we then deduce the result for @xmath27 and @xmath19-type cluster varieties from the @xmath56 case .
we now turn to the use of convexity conditions to prove the fock - goncharov conjecture in a number of different situations , as covered in [ ppsec ] . to motivate the concepts ,
let us define a _ partial minimal model _ of a log calabi - yau variety @xmath15 .
this is an inclusion @xmath140 as an open subset such that the canonical volume form on @xmath15 has a simple pole along each irreducible divisor of the boundary @xmath141 .
for example , a partial minimal model for an algebraic torus is the same as a toric compactification .
we wish to extend elementary constructions of toric geometry to the cluster case .
for example , the partial compactification @xmath142 determined by frozen variables is a partial minimal model .
the generalisation of the cocharacter lattice @xmath143 of the algebraic torus @xmath144 is the tropical set @xmath145 of @xmath15 .
the main difference between the torus and the general case is that @xmath146 is not in general a vector space .
indeed , the identification of @xmath147 with the cocharacter lattices of various charts of @xmath15 induce piecewise linear ( but not linear ) identifications between the cocharacter lattices . as a result ,
a piecewise straight path in @xmath146 which is straight under one identification @xmath148 will be bent under another .
thus the usual notions of straight lines , convex functions or convex sets do not make sense on @xmath149 .
the idea for generalizing the notion of convexity is to instead make use of broken lines , which are piecewise linear paths in @xmath149 . using broken lines in place of straight lines
we can say which piecewise linear functions , and thus which polytopes , are convex , see definition [ imcdef ] .
each regular function @xmath150 has a canonical piecewise linear tropicalisation @xmath151 , which we conjecture is convex in the sense of definition [ imcdef ] , see conjecture [ dcon ] .
the conjecture is easy for @xmath152 , see proposition [ convcor ] .
each _ convex _
piecewise linear @xmath153 gives a _ convex _
polytope @xmath154 and a _ convex _
cone @xmath155 , where italics indicates convexity in our broken line sense .
we believe the existence of a bounded polytope is equivalent to the full fock - goncharov conjecture : [ ffgconj ] the full fock - goncharov conjecture holds for @xmath56 if and only if the tropical space @xmath156 contains a full dimensional bounded polytope , convex in our sense .
the examples of @xcite , 7 , show that for the full fock - goncharov conjecture to hold , we need to assume @xmath15 has enough global functions . in that case tropicalizing a general function gives ( conjecturally ) a bounded
_ convex _ polytope .
as we are unable to prove conjecture [ dcon ] except in the monomial case we use a restricted version ( which happily still has wide application ) : [ gmdef ] a cluster variety @xmath15 has _ enough global monomials _ ( egm ) if for each valuation @xmath157 there is a global monomial @xmath34 with @xmath158 .
the condition that @xmath15 has egm is equivalent to the existence of @xmath159 whose associated convex polytope @xmath160 is bounded , see lemma [ egmlem ] .
the following theorem demonstrates the value of the egm condition : [ egmth ] let @xmath15 be a cluster variety .
then : 1 . ( proposition [ cafin ] ,
corollary [ fgcor ] ) if @xmath161 satisfies the egm condition , then the multiplication rule on @xmath43 is polynomial , i.e. , for given @xmath162 , @xmath163 for all but finitely many @xmath164 .
this gives @xmath43 the structure of a finitely generated commutative associative @xmath8-algebra .
2 . ( proposition [ markscor ] )
if @xmath165 and @xmath15 satisfies the egm condition , then there are canonical inclusions @xmath166 3 .
( theorem [ fgscth ] , ( 2 ) ) for @xmath167 , if @xmath139 is generated by finitely many cluster variables , then the full fock - goncharov conjecture holds for @xmath15 ( definition [ fgconjdef ] ) .
4 . ( proposition [ egmscprop ] ) if the set @xmath110 of all @xmath41-vectors of global monomials of @xmath111 in @xmath112 is not contained in a half - space under the identification of @xmath112 with @xmath113 induced by some choice of seed , then the full fock - goncharov conjecture ( see definition [ fgconjdef ] ) holds for @xmath56 , @xmath27 , very general @xmath78 and , if the convexity condition ( 7 ) of theorem [ mainth ] holds , for @xmath19 .
[ markovremark ] we believe , based on calculations in @xcite , 7.1 , that the conditions of the theorem ( @xmath56 has egm , and @xmath168 ) hold for the cluster variety associated with the once punctured torus , see some details in examples [ standardexample ] and [ examples2 ] . however , the equality @xmath169 is expected to fail , and in particular in this case we expect the full fock - goncharov holds for @xmath170 and very general @xmath171 but not for @xmath19 .
we note that @xmath56 has enough global monomials in many cases : [ egmprop ] consider the following conditions on a cluster algebra @xmath19 : 1 .
the exchange matrix has full rank , @xmath33 is generated by finitely many cluster variables , and @xmath172 is a smooth affine variety .
@xmath19 has an acyclic seed .
3 . @xmath19 has a seed with a maximal green sequence .
4 . for some seed ,
the cluster complex @xmath173 is not contained in a half - space .
@xmath56 has enough global monomials .
then ( 1 ) implies ( 5 ) ( proposition [ egmprop1 ] ) .
furthermore , ( 2 ) implies ( 3 ) implies ( 4 ) implies ( 5 ) ( propositions [ mgsrem ] , [ gsprop ] , and [ ldegmprop ] ) . finally ( 4 ) implies the full fock - goncharov conjecture , for @xmath71 , @xmath27 , or very general @xmath78 , or , under the convexity assumption ( 7 ) of theorem [ mainth ] , for @xmath19 ( proposition [ egmscprop ] ) .
[ nex ] a recent paper @xcite of goodearl and yakimov announces the equality @xmath174 for all double bruhat cells in semi - simple groups . in this case
, yakimov has furthermore announced the existence of a maximal green sequence .
many cluster varieties @xmath19 associated to a marked bordered surface with at least two punctures also have a maximal green sequence , see @xcite , 1.3 for a summary of known results on this .
the recent @xcite , theorem 1.12,1.17 , shows that ( 4 ) holds for the fock - goncharov cluster varieties of @xmath175 local systems on _ most _ decorated surfaces . together with proposition [ egmprop ]
these results imply the full fock - goncharov theorem in any of these cases .
we note that for the cluster algebra associated to a marked bordered surface , a canonical basis of @xmath176 parameterized by @xmath177 has been previously obtained by fock - goncharov , @xcite , theorem 12.3 .
they show that the @xmath19 and @xmath27 varieties have natural modular meaning as moduli spaces of local systems .
they identify @xmath177 with a space of integer laminations ( isotopy classes of disjoint loops with integer weights ) and their associated basis element is a natural function given by trace of monodromy around a loop .
we checked , together with a. neitzke , that our basis agrees with the fock - goncharov basis of trace functions in the case of a sphere with four punctures , for primitive elements of the tropical set .
our theta function basis comes canonically from the cluster structure ( it does not depend on any modular interpretation ) .
[ clusterindependenceremark ] in general , we conjecture the bases we construct for rings of global functions on cluster varieties @xmath15 or partial compactifications @xmath87 are intrinsic to the underlying log calabi - yau variety @xmath15 , and do not depend on the particular cluster structure on @xmath15 .
this is a non - trivial statement : there exist varieties with multiple cluster structures ( in particular different atlases of tori for the same variety ) .
yan zhou will show in her ph.d .
thesis that the ( principal coefficient version of ) the cluster variety associated to the once - punctured torus is an example .
this conjecture is suggested by @xcite , conjecture 0.6 , and the results of @xcite , @xcite and @xcite prove this in the case of the @xmath27 cluster varieties where the skew - symmetric form has rank two , which includes the case of the sphere with four punctures .
thus we have the ( at least to us ) remarkable conclusion that in many cases where bases occur because of some extrinsic interpretation of the spaces , in fact this extrinsic interpretation is irrelevant .
for example , the theta functions given by trace functions above , which would appear to depend on the realization of the cluster variety as a moduli space of local systems , are actually intrinsic to the underlying variety . in the case of example
[ slrexample ] , where bases may arise from representation theory , our basis does not use the group - theoretic aspects of the spaces .
the suggestion that the canonical basis is independent of the cluster structure may surprise some , as understanding the canonical basis was the initial motivation for the fomin - zelevinsky definition of cluster algebras . returning to the role of convexity notions , we note that our formula for the structure constants @xmath59 of theorem [ mainth ] , ( 1 ) is given by counting broken lines .
as a result , our notion of convexity interacts nicely with the multiplication rule .
this allows us to generalize basic polyhedral constructions from toric geometry in a straight - forward way .
a polytope @xmath178 convex in our sense determines ( by familiar rees - type constructions for graded rings ) a compactification of @xmath15 .
furthermore , for any choice of seed , @xmath114 is identified with a linear space @xmath179 and @xmath180 with an ordinary convex polytope .
our construction also gives a flat degeneration of this compactification of @xmath15 to the ordinary polarized toric variety for @xmath181 .
see [ candsec ] .
we expect this specializes to a uniform construction of many degenerations of representation theoretic objects to toric varieties , see e.g. , @xcite , @xcite , and @xcite .
applied to the fock - goncharov moduli spaces of @xmath93-local systems , this will give for the first time compactifications of character varieties with nice ( e.g. , toroidal anti - canonical ) boundary .
see remark [ crems ] .
the polytope can be chosen so that the boundary of the compactification is very simple , a union of toric varieties . for example , let @xmath182 be the open subset where the frozen variables for the standard cluster structure are non - vanishing .
then the boundary @xmath183 consists of a union of certain schubert cells .
we obtain using a polytope an alternative compactification where the schubert cells ( which are highly non - toric ) are replaced by toric varieties .
see theorem [ toricbth ] .
the fock - goncharov conjecture is the cluster special case of @xcite , conjecture 0.6 , which says ( roughly ) that affine log cys with maximal boundary come in canonical dual pairs with the tropical set of one parameterizing a canonical basis of functions on the other .
we can view the conjecture as having two parts : first , that the vector space , @xmath85 , with this basis @xmath184 is naturally an algebra in a such a way that @xmath185 is an affine log cy . and then furthermore , that this log cy is the mirror in the cluster case the fock - goncharov dual ( it is natural to further ask if this is the mirror in the sense of hms but we do not consider this question here ) .
our deepest mirror theoretic result is the following weakening of the first part : assume @xmath186 has egm . then for @xmath187 or @xmath78 for very general @xmath188 , or , under the convexity assumption ( 7 ) of theorem [ mainth ] , @xmath19 , @xmath43 ( with structure constants as in [ mainth ] ) is a finitely generated algebra and @xmath189 is a log canonical gorenstein @xmath190-trivial affine variety of dimension @xmath191 . for the proof see theorem [ cfcor ] .
we turn to [ pcrtsec ] . here
we study features of partial compactifications coming from frozen variables .
as explained in example [ slrexample ] , these partial compactifications are often the relevant ones in representation - theoretic examples . in particular , for a partial minimal model @xmath142 , often the vector subspace @xmath192 is more important than @xmath33 itself .
for example there is a cluster structure with frozen variables for the open double bruhat cell @xmath1 in a semi - simple group @xmath93 .
then @xmath33 is the ring of functions on the open double bruhat cell and @xmath193 .
of course @xmath194 is the most important representation of @xmath93 .
however , one can not expect a canonical basis of @xmath195 , i.e. , one determined by the intrinsic geometry of @xmath196 .
for example , @xmath93 has no non - constant global functions which are eigenfunctions for the action of @xmath93 on itself .
but we expect , and in the myriad cases above can prove , that the affine log calabi - yau open subset @xmath142 has a canonical basis , @xmath61 , and we believe that @xmath197 , the set of theta functions on @xmath19 that extends regularly to all of @xmath196 , is a basis for @xmath195 , canonically associated to the choice of log calabi - yau open subset @xmath142 , see @xcite , remark 1.10 .
this is not a basis of @xmath93-eigenfunctions , but they are eigenfunctions for the associated maximal torus , which is the subgroup of @xmath93 that preserves @xmath1 .
this is exactly what one should expect : the basis is not intrinsic to @xmath93 , instead it is ( we conjecture ) intrinsic to the pair @xmath198 , see remark [ clusterindependenceremark ] .
we shall now describe in more detail what can be proved for partial compactifications of cluster varieties coming from frozen variables .
a key point is a technical but combinatorial hypothesis , that _ each variable has an _ optimized seed _ _ , see definition [ osdef ] and lemmas [ quiveroptimized ] and [ opslem ] . the main need for this hypothesis is proposition [ boundaryprop ] , which states that if a linear combination of theta functions extends across a boundary divisor then each theta function in the sum extends across the divisor .
thus the middle cluster algebra , in this case , behaves well with respect to boundary divisors .
happily , this condition holds for the cluster structures on the grassmannian , and , for @xmath4 , for the cluster structure on a maximal unipotent subgroup @xmath199 , the basic affine space @xmath200 , and the fock - goncharov cluster structure on @xmath201 , see remark [ osrem ] .
let us now work with the principal cluster variety @xmath56 .
consider the partial compactification @xmath202 by allowing the frozen variables to be zero .
each boundary divisor @xmath203 gives a point @xmath204 and thus ( in general conjecturally ) , a canonical theta function @xmath205 on @xmath206 .
we then define the _ potential _ @xmath207 as the sum of these theta functions .
we have its piecewise linear tropicalisation @xmath208 .
this defines a cone @xmath209 [ spth ] assume that each frozen index @xmath210 has an optimized seed . then : 1 .
@xmath211 and @xmath180 are convex in our sense .
2 . the set @xmath212 parameterizes a canonical basis of an algebra @xmath213 , and @xmath214 3
. now assume further that we have enough global monomials on @xmath186 .
if for some seed @xmath10 , @xmath180 is contained in the convex hull of @xmath61 ( which itself contains the convex hull of @xmath110 ) then @xmath215 , @xmath216 is finitely generated , and the integer points @xmath217 parameterize a canonical basis .
each choice of seed identifies @xmath218 with a lattice , and the cone @xmath219 with a rational polyhedral cone , described by canonical linear inequalities given by the tropicalisation of the potential .
note that @xmath180 is convex in our generalized sense .
we show , making use of recent results of magee and goncharov - shen , @xcite , @xcite , @xcite , that in the representation theoretic examples which were the original motivation for the definition of cluster algebras our polyhedral cones @xmath180 specialize to the piecewise linear parameterizations of canonical bases of berenstein and zelevinsky @xcite , knutson and tao @xcite , and goncharov and shen @xcite : [ slrcor ] let @xmath220 and let @xmath142 be the fomin - zelevinsky cluster variety for the basic affine space @xmath98 . 1 .
all the hypotheses , and thus the conclusions , of theorem [ spth ] hold .
in particular @xmath221 parameterizes a canonical theta function basis of @xmath222 .
our potential @xmath223 agrees with the ( representation theoretically defined ) potential function of berenstein - kazhdan , @xcite .
the maximal torus @xmath100 acts canonically on @xmath196 , preserving the open set @xmath142 .
each theta function is an @xmath100-eigenfunction , and there is a canonical map @xmath224 ( the target is the character lattice of @xmath100 ) , linear for the linear structure given by any seed , which sends an integer point to the @xmath100-weight of the corresponding theta function .
the slice @xmath225 parameterizes a canonical theta function basis of the eigenspace @xmath226 , the corresponding irreducible representation of @xmath93 . 5 . for a natural choice of seed
the _ cone _
@xmath180 is canonically identified with the gelfand - tsetlin cone .
[ conf3cor ] let @xmath142 be the fock - goncharov cluster variety for @xmath227 1 .
all the hypotheses , and thus the conclusions , of theorem [ spth ] hold .
in particular the cone @xmath221 parameterizes a canonical theta function basis of @xmath228 .
our potential function @xmath223 agrees with the ( representation theoretically defined ) potential function of goncharov - shen , @xcite .
@xmath229 acts canonically on @xmath196 , preserving the open subset @xmath142 .
each theta function is an @xmath100-eigenfunction , and there is a canonical map @xmath230 linear for the linear structure given by any seed , which sends an integer point to the @xmath229-weight of the corresponding theta function .
the slice @xmath231 parameterizes a canonical theta function basis of the eigenspace @xmath232 in particular the number of integral points in @xmath233 is the corresponding littlewood - richardson coefficient .
5 . for a natural choice of seed ,
the cone @xmath180 is canonically identified with the knutson - tao hive .
these corollaries are proven at the end of [ fullfgsection ] .
we stress here that the above representation theoretic results come _ for free _ from general properties of our mirror symmetry construction : any partial minimal model @xmath234 of an affine log calabi - yau variety with maximal boundary determines ( in general conjecturally ) a cone @xmath235 with the analogous meaning .
we are getting these basic representation theoretic results without representation theory !
we recover the remarkable gelfand - tsetlin and hive polytopes for a particular choice of seed .
different ( among the infinitely many possible ) choices of seed give in general combinatorially different cones , whose integer points parameterize the same theta function basis .
the canonical object is the _ convex cone _
@xmath236 cut out by @xmath211 , different ( by piecewise linear mutation ) identifications of @xmath237 with a vector space give different incarnations of @xmath180 as convex cones in the usual sense .
potentials were considered in the work of goncharov and shen , @xcite , which in turn built on work of berenstein and zelevinsky , @xcite and berenstein and kazhdan , @xcite , @xcite .
the potential constructed by goncharov and shen has a beautiful representation theoretic definition , and was found in many situations to coincide with known constructions of landau - ginzburg potentials . on the other hand ,
the construction of the potential in terms of theta functions coincides precisely with the construction of the mirror landau - ginzburg potential as carried out in @xcite,@xcite .
the latter work can be viewed as a tropicalization of the descriptions of the potential in terms of holomorphic disks in @xcite,@xcite .
thus our construction explains the emergence of the landau - ginzburg potentials in @xcite .
our potentials are determined by the cluster structure ( and conjecturally , just the underlying log calabi - yau variety ) , and in particular are independent of any modular or representation theoretic interpretation of the cluster variety .
this gives , as in remark [ clusterindependenceremark ] , the remarkable suggestion that e.g. , the representation theoretically defined goncharov - shen potential , which would seem to depend heavily on the modular interpretation of @xmath238 , is actually intrinsic to the partial minimal model @xmath142 .
_ acknowledgments _ : we received considerable inspiration from conversations with a. berenstein , v. fock , s. fomin , a. goncharov , b. keller , b. leclerc , j. kollr , g. muller , g. musiker , a. neitzke , d. rupel , m. shapiro , b. siebert , y. soibelman , and d. speyer .
the first author was partially supported by nsf grant dms-1262531 and a royal society wolfson research merit award , the second by nsf grants dms-1201439 and dms-1601065 , and the third by nsf grant dms-0854747 .
some of the research was conducted when the first and third authors visited the fourth at i.h.e.s , during the summers of 2012 and 2013 .
here we recall the basic properties of scattering diagrams , the main technical tool in this paper .
scattering diagrams appeared first in @xcite in two dimensions , and then in all dimensions in @xcite , with another approach in a more specific case in @xcite .
here we give a self - contained treatment restricted to the specific case needed in this paper .
we start with a choice of _ fixed data _ @xmath239 as defined in @xcite , which for the reader s convenience is described at the beginning of appendix [ ldsec ] . in brief
, this entails a lattice @xmath240 with dual lattice @xmath241 , a skew - symmetric form @xmath242 sublattices @xmath243 with @xmath244 a saturated sublattice and @xmath245 a sublattice of finite index with dual lattice @xmath246 , an index set @xmath247 with @xmath248 and a subset @xmath249 with @xmath250 , as well as positive integers @xmath251 , @xmath252 .
finally we also choose an initial seed @xmath10 , i.e. , a basis @xmath253 of @xmath240 .
see appendix [ ldsec ] for the precise properties that all this data must satisfy . for the construction of the scattering diagram associated to this data
, we will require * the injectivity assumption*. the map @xmath254 given by @xmath255 is injective .
while this does not hold for a general choice of fixed data , it does hold in the principal coefficient case ( see appendix [ rdsec ] ) , and results in this paper about arbitrary cluster varieties and algebras will be proved via the principal case . set @xmath256 choose a linear function @xmath257 such @xmath258 for @xmath259 . under the injectivity assumption , one can choose a strictly convex top - dimensional cone @xmath260 , with associated monoid @xmath261 , such that @xmath262 for all @xmath263 . here
@xmath264 is the group of units of the monoid @xmath265 .
this gives the monomial ideal @xmath266 $ ] in the monoid ring @xmath267 $ ] over a field @xmath8 of characteristic zero , and we write @xmath268}$ ] for the completion with respect to @xmath269 .
we define the _ module of log derivations _ of @xmath267 $ ] as @xmath270):={\bbbk}[p]\otimes_{{\mathbb{z } } } n^{\circ},\ ] ] with the action of @xmath271 on @xmath267 $ ] being given by @xmath272 so we write @xmath271 as @xmath273 .
let @xmath274)}$ ] denote the completion of @xmath275)$ ] with respect to the ideal @xmath269 . using this action , if @xmath276)}$ ] , then @xmath277})\ ] ] makes sense using the taylor series for the exponential .
we have the lie bracket @xmath278=z^{m+m ' } \partial_{\langle n , m'\rangle n'-\langle n',m\rangle n}.\ ] ] then @xmath279))$ ] can be viewed as a subgroup of the group of continuous automorphisms of @xmath268}$ ] which are the identity modulo @xmath269 , with the group law of composition coinciding with the group law coming from the baker - campbell - hausdorff formula .
define the sub - lie algebra of @xmath275)={\bbbk}[p]\otimes_{{\mathbb{z } } } n^{\circ}$ ] @xmath280 where @xmath281 is the one - dimensional subspace of @xmath275)$ ] spanned by @xmath282 .
we calculate that @xmath283 is in fact closed under lie bracket : @xmath284 = { } & z^{p_1^*(n+n')}\left ( \langle p_1^*(n'),n\rangle\partial_{n ' } -\langle p_1^*(n),n'\rangle\partial_{n}\right)\\ = { } & z^{p_1^*(n+n')}\left ( \{n ' , n\}\partial_{n'}-\{n , n'\}\partial_n\right)\\ = { } & \{n',n\ } z^{p_1^*(n+n')}\partial_{n+n'}. \end{split}\end{aligned}\ ] ] we have @xmath285 a lie subalgebra , and @xmath286 a nilpotent lie algebra .
we let @xmath287 be the corresponding nilpotent group .
this group , as a set , is just @xmath288 , but multiplication is given by the baker - campbell - hausdorff formula .
we set @xmath289 the corresponding pro - nilpotent group .
we have the canonical set bijections @xmath290 for @xmath291 we define @xmath292 note that by the commutator formula , @xmath293 , hence @xmath294 , is abelian . in
what follows , noting that @xmath93 is a subgroup of @xmath295})$ ] , we will often describe elements of @xmath296 as follows .
[ wcdef ] let @xmath291 , @xmath297 and @xmath298}$ ] .
define @xmath299 to be the automorphism of @xmath268}$ ] given by @xmath300 where @xmath301 is the generator of the monoid @xmath302 .
[ thetaflemma ] for @xmath303 , @xmath304})$ ] is the subgroup of automorphisms of the form @xmath305 for @xmath34 as in definition [ wcdef ] with the given @xmath291 .
more specifically , @xmath306 acts as the automorphism @xmath305 with @xmath307 , where @xmath308 is the smallest positive rational number such that @xmath309 .
let @xmath310})$ ] be the set of @xmath305 of the given form . then @xmath100 is a subgroup as @xmath311 . note that @xmath312 , where @xmath313 is as described in the statement .
the exponential of this vector field is easily seen to act as @xmath305 with @xmath314 .
hence @xmath315 . from this , we see also that if @xmath316 , then @xmath317 , and the latter lies in @xmath294 .
[ walldef ] a _ wall _ in @xmath318 ( for @xmath319 and @xmath283 ) is a pair @xmath320 such that 1 .
@xmath321 for some primitive @xmath291 .
@xmath322 is a @xmath323-dimensional convex ( but not necessarily strictly convex ) rational polyhedral cone .
the set @xmath324 is called the _ support _ of the wall @xmath320 . using lemma [ thetaflemma ]
, we often write a wall as @xmath325 for @xmath326}$ ] , necessarily of the form @xmath327 .
we shall use this notation interchangeably without comment .
[ ksscatdiagdef ] a _ scattering diagram @xmath328 for @xmath319 and @xmath283 _ is a set of walls such that for every degree @xmath329 , there are only a finite number of @xmath330 with the image of @xmath331 in @xmath332 not the identity .
if @xmath328 is a scattering diagram , we write @xmath333 for the support and singular locus of the scattering diagram .
if @xmath328 is a finite scattering diagram , then its support is a finite polyhedral cone complex .
joint _ is an @xmath334-dimensional cell of this complex , so that @xmath335 is the union of all joints of @xmath328
. we will often ( especially in appendix [ scatappendix ] ) want to use a slightly more general notion of scattering diagram , where the elements attached to walls lie in some other choice of group @xmath336 arising from an @xmath319-graded lie algebra @xmath337 . in this case
we talk about a scattering diagram for @xmath337 .
for example , any scattering diagram for @xmath283 induces a finite scattering diagram for @xmath338 by taking the image of the attached group elements under the projection @xmath339 . given a scattering diagram @xmath328
, we obtain the _ path - ordered product_. assume given a smooth immersion @xmath340\rightarrow m_{{\mathbb{r}}}\setminus{\operatorname{sing}}({\mathfrak{d}})\ ] ] with endpoints not contained in the support of @xmath328 .
assume @xmath341 is transversal to each wall of @xmath328 that it crosses .
for each degree @xmath329 , we can find numbers @xmath342 and elements @xmath343 with the image of @xmath344 in @xmath332 non - trivial such that @xmath345 @xmath346 if @xmath347 , and @xmath348 taken as large as possible .
( the @xmath349 are the times at which the path @xmath341 hits a wall .
we allow @xmath350 because we may have two different walls @xmath351 which span the same hyperplane . ) for each @xmath210 , define @xmath352 where @xmath303 with @xmath353 .
we then define @xmath354 if @xmath355 , then @xmath356 span the same hyperplane @xmath357 , hence @xmath358 .
thus , since this latter group is abelian , @xmath344 and @xmath359 commute , so this product is well - defined .
we then take @xmath360 we note that @xmath361 depends only on its homotopy class ( with fixed endpoints ) in @xmath362 .
we also note that the definition can easily be extended to piecewise smooth paths @xmath341 , provided that the path always crosses a wall if it intersects it .
two scattering diagrams @xmath328 , @xmath363 are _ equivalent _ if @xmath364 for all paths @xmath341 for which both are defined .
call @xmath365 _ general _ if there is at most one rational hyperplane @xmath357 with @xmath366 . for
@xmath367 general and @xmath328 a scattering diagram , let @xmath368 .
one checks easily : [ easyequivlemma ] two scattering diagrams @xmath369 are equivalent if and only if @xmath370 for all general @xmath367 .
a scattering diagram @xmath328 is _ consistent _ if @xmath361 only depends on the endpoints of @xmath341 for any path @xmath341 for which @xmath361 is defined .
we say a wall @xmath371 is _ incoming _ if @xmath372 otherwise , we say the wall is _ outgoing _
( note in any case @xmath373 lies in the span of the wall @xmath357 ) .
we call @xmath374 the _ direction _ of the wall .
( this terminology comes from the case @xmath375 , where an outgoing wall is then a ray containing its direction vector , thus one that points outward . )
we need one particular scattering diagram , determined by the fixed data and seed data . setting @xmath376 , @xmath263 , we start with the scattering diagram @xmath377 the main result on scattering diagrams , which follows easily from theorem [ kslemma ] , is the following .
a more general version of this was proved in two dimensions in @xcite , and in a much more general context in all dimensions in @xcite . a simpler argument which applies to the case at hand
was given in @xcite , which shall be reviewed in
[ consistentscatdiagramsection ] .
[ kslemmags ] there is a scattering diagram @xmath378 satisfying : 1 .
@xmath378 is consistent , 2 .
@xmath379 , 3 .
@xmath380 consists only of outgoing walls .
moreover , @xmath378 satisfying these three properties is unique up to equivalence .
the crucial positivity result satisfied by @xmath378 is now easily stated : [ scatdiagpositive ] the scattering diagram @xmath378 is equivalent to a scattering diagram all of whose walls @xmath325 satisfy @xmath381 for some @xmath382 and @xmath130 a positive integer . in particular , all nonzero coefficients of @xmath383 are positive integers . the proof is given in appendix [ scatappendix ] .
the basic idea is that the construction of the scattering diagram @xmath378 can be reduced to repeated applications of the following example : [ basicscatteringexample ] take @xmath384 , @xmath385 and the skew - symmetric form @xmath386 given by the matrix @xmath387 , where @xmath388 .
let @xmath389 be the dual basis of @xmath390 , and write @xmath391 , @xmath392 .
we get @xmath393 then one checks easily that @xmath394 see figure [ scat11 ] .
( see for example @xcite , example 1.6 . )
[ bcexample ] take @xmath395 , with basis @xmath390 , and take @xmath245 to be the sublattice generated by @xmath396 . further take @xmath397 , @xmath398 , where @xmath399 are two positive integers , and take the skew - symmetric form to be the same as in the previous example . then @xmath400 , @xmath401 . taking as before @xmath391 , @xmath392 , we get @xmath402 for most choices of @xmath403 and @xmath130 , this is a very complicated scattering diagram .
a very similar scattering diagram , with functions @xmath404 and @xmath405 , has been analyzed in @xcite , but it is easy to translate this latter diagram to the one considered here by replacing @xmath406 by @xmath407 and using the change of lattice trick , which is given in step iv of the proof of proposition [ basicpositivity ] .
all rays of @xmath408 are contained strictly in the fourth quadrant ( i.e. , in particular are not contained in an axis ) . without giving the details , we summarize the results .
there are two linear operators @xmath409 given by the matrices in the basis @xmath410 as @xmath411 then @xmath412 is invariant under @xmath413 and @xmath414 , in the sense that if @xmath415 , we have @xmath416 provided @xmath417 is contained strictly in the fourth quadrant .
it is also the case that applying @xmath414 to @xmath418 or @xmath413 to @xmath419 gives an element of @xmath412 .
further , @xmath378 contains a discrete series of rays consisting of those rays in the fourth quadrant obtained by applying @xmath413 and @xmath414 alternately to the above rays supported on @xmath420 and @xmath421 .
these rays necessarily have functions of the form @xmath422 or @xmath423 for various choices of @xmath59 and @xmath424 .
if @xmath425 , we obtain a finite diagram .
( moreover , the corresponding @xmath19 cluster variety is the cluster variety of finite type @xcite associated to the root system @xmath426 , @xmath427 , or @xmath428 for @xmath429 , @xmath430 , or @xmath431 respectively . ) if @xmath432 , these rays converge to the rays contained in the two eigenspaces of @xmath433 and @xmath434 .
these are rays of slope @xmath435 .
this gives a complete description of the rays outside of the cone spanned by these two rays .
the expectation is that every ray of rational slope appears in the interior of this cone , and the attached functions are in general unknown .
however , in the @xmath436 case , it is known @xcite that the function attached to the ray of slope @xmath437 is @xmath438 the chamber structure one sees outside the quadratic irrational cone is very well - behaved and familiar in cluster algebra theory . in particular , the interiors of the first , second and third quadrants are all connected components of @xmath439 , and there are for @xmath440 an infinite number of connected components in the fourth quadrant .
we will see in [ tropsec ] that this chamber structure is precisely the fock - goncharov cluster complex .
on the other hand , it is precisely the rich structure inside the quadratic irrational cone which scattering diagram technology brings into the cluster algebra picture . in this subsection
we give more details about the construction of scattering diagrams , and in particular give results leading to the proof of theorem [ kslemmags ] .
this material can be skipped on first reading , but is recommended before reading the more difficult material on scattering diagrams in appendix [ scatappendix ] .
let @xmath328 be a scattering diagram .
if we set @xmath441 then since any wall spans a hyperplane @xmath357 , for some @xmath291 , @xmath442 in particular , if @xmath328 is a consistent scattering diagram , then @xmath361 for @xmath341 a path with initial point in @xmath443 and final point in @xmath444 is independent of the particular choice of path ( or endpoints in @xmath445 ) .
thus we obtain a well - defined element @xmath446 which only depends on the scattering diagram @xmath328 .
[ maximscorrespondence ] the assignment of @xmath447 to @xmath328 gives a one - to - one correspondence between equivalence classes of consistent scattering diagrams and elements @xmath446 .
this is a special case of @xcite , 2.1.6 .
for the reader s convenience we include the short proof : we need to show how to construct @xmath328 given @xmath446 .
to do so , choose any @xmath303 primitive and a point @xmath448 general .
then we can determine @xmath449 as follows , noting by lemma [ easyequivlemma ] that this information for all such @xmath450 and general @xmath367 determines @xmath328 up to equivalence .
we can write @xmath451 with @xmath452 each of these subspaces of @xmath283 are closed under lie bracket , thus defining subgroups @xmath453 of @xmath93 .
note by the generality assumption on @xmath367 , we in fact have @xmath454 .
this splitting induces a unique factorization @xmath455 for any element @xmath456 .
applying this to @xmath447 gives a well - defined element @xmath457 .
we need to show that the set of data @xmath458 determines a scattering diagram @xmath328 such that @xmath459 for all general @xmath365 . to do this
, one needs to know that to any finite order @xmath460 , the hyperplane @xmath357 is subdivided into a finite number of polyhedral cones @xmath461 such that the image of @xmath458 in @xmath332 is constant for @xmath462 .
this is clear because the number of @xmath463 with @xmath464 is finite , as then the decomposition varies discretely with @xmath367 to order @xmath460 .
we need to show that @xmath328 satisfies the condition that @xmath465 for any path @xmath341 from the positive to the negative chamber and that @xmath361 only depends on endpoints of @xmath341 .
to do so , we work modulo @xmath466 for any @xmath460 , so we can assume @xmath328 has a finite number of walls .
choose a general point @xmath467 .
take a general two - dimensional subspace of @xmath318 containing @xmath468 , and after choosing a metric , let @xmath341 be a semi - circle in the two - dimensional subspace with endpoints @xmath468 and @xmath469 and center @xmath81 .
then @xmath470 for points @xmath471 contained in walls crossed by @xmath341 and @xmath472 the element of @xmath473 determined by the factorization of @xmath447 above . note that if @xmath474 lies in the hyperplane @xmath475 , all the wall - crossing automorphisms of walls traversed by @xmath341 _ before _ crossing @xmath475 lie in @xmath476 and all those from walls traversed by @xmath341 _
after _ crossing @xmath475 lie in @xmath477 .
it then follows inductively that the factorization of @xmath447 given by @xmath474 takes the form @xmath478 for some @xmath479 .
indeed , for @xmath480 , this just follows from the definition of @xmath481 , while if true for @xmath482 , then we have @xmath483 is a decomposition of @xmath447 induced by the splitting @xmath484 , and the claim then follows by the definition of @xmath472 .
in particular , for @xmath485 , taking @xmath486 and noting that @xmath487 , one sees that @xmath488 .
next we show the independence of path for the @xmath328 we have constructed , again modulo @xmath466 .
it is sufficient to check @xmath489 as an element of @xmath332 for any small loop @xmath490 around any joint @xmath491 of @xmath328 .
take @xmath492 a general point in @xmath491 , @xmath463 such that @xmath493 , and choose @xmath494 to be points in @xmath495 near @xmath492 on either side of the joint @xmath491 .
let @xmath496 be two semi - circular paths with endpoints @xmath468 and @xmath469 and passing through @xmath494 respectively .
then up to orientation @xmath497 is freely homotopic to @xmath490 in @xmath498 .
thus @xmath499 .
thus we have established the one - to - one correspondence between consistent scattering diagrams @xmath328 and elements of @xmath93 .
following @xcite , we give an alternative parameterization of @xmath93 , as follows . for any @xmath303 primitive
, we get the splitting @xmath500 where @xmath501 these give rise to subgroups @xmath502 of @xmath93 .
we drop the @xmath450 when it is clear from context . again , this allows us to factor any @xmath456 as @xmath503 with @xmath504 , @xmath505 .
we can further decompose @xmath506 , where @xmath507 , while @xmath508 involves those summands of @xmath509 coming from @xmath510 not proportional to @xmath450 .
note that @xmath511\subseteq { \mathfrak{g}}_0^{\perp}$ ] .
indeed , if @xmath512 with @xmath513 for @xmath514 , we then have @xmath515 so that @xmath516 = 0 $ ] by ( [ liebracketform ] )
. thus we have a projection homomorphism @xmath517 with kernel @xmath518 .
in particular , the factorization @xmath503 yields an element @xmath519 via this projection .
we then have a map ( of sets ) @xmath520 [ gparam ] @xmath521 is a set bijection @xmath521 is induced by an analogous map to order @xmath460 , @xmath522 one checks easily that this is a bijection order by order .
[ kslemma ] let @xmath328 be a consistent scattering diagram corresponding to @xmath523 .
the following hold : 1 . for each @xmath303 , to any fixed finite order
, there is an open neighbourhood @xmath524 of @xmath525 such that @xmath526 for all general @xmath527 . here
@xmath528 denotes the component of @xmath529 indexed by @xmath450 .
@xmath328 is equivalent to a diagram with only one wall in @xmath357 containing @xmath525 for each @xmath303 , and the group element attached to this wall is @xmath530 .
3 . set @xmath531 then @xmath328 is equivalent to a consistent scattering diagram @xmath363 such that @xmath532 and @xmath533 consists only of outgoing walls .
furthermore , up to equivalence @xmath363 is the unique consistent scattering diagram with this property .
the equivalence class of a consistent scattering diagram is determined by its set of incoming walls .
we note first that ( 3 ) of theorem [ kslemma ] implies theorem [ kslemmags ] .
indeed , let @xmath534 for @xmath263 be the group element corresponding to @xmath535 , so that the initial scattering can be written as @xmath536 . by proposition [ gparam ]
there is a unique element @xmath537 with @xmath538 now apply theorem [ kslemma ] with @xmath539 .
first note that statement ( 1 ) implies ( 2 ) .
further , ( 1 ) , along with theorem [ maximscorrespondence ] and proposition [ gparam ] , implies ( 4 ) , which in turn gives the uniqueness in ( 3 ) .
note ( 1 ) implies that , to the given finite order , @xmath328 is equivalent to a diagram having only one incoming wall contained in @xmath357 , and the attached group element is @xmath530 .
now we can replace this single wall by an equivalent collection of walls consisting of @xmath540 and a number of outgoing walls contained in @xmath357 with attached group element @xmath541 .
this gives the existence in ( 3 ) .
thus it suffices to prove ( 1 ) .
we work modulo @xmath466 , so we may assume @xmath328 is finite , and compare the splittings coming from a choice of @xmath448 near @xmath525 and , after replacing @xmath283 with @xmath542
. for each @xmath463 there exists an open neighbourhood @xmath543 of @xmath525 such that @xmath544 ( resp .
@xmath545 ) implies @xmath546 ( resp .
@xmath545 ) for all @xmath547 .
since @xmath548 is now a finite sum of @xmath281 s , we can find a single @xmath1 so that @xmath549 for all @xmath527 . if @xmath367 is general inside @xmath357 we also have @xmath550 .
now write @xmath551 as in
. then we can further factor @xmath552 as in .
note @xmath553 , @xmath554 . since the projection @xmath555 is a group homomorphism with kernel @xmath518 ,
the image of @xmath556 in @xmath296 is @xmath557 , which thus coincides with @xmath530 by definition of the latter .
we have @xmath558 which is then the ( unique ) factorisation from .
thus @xmath559 for any general @xmath527 .
we now study how the scattering diagram @xmath378 constructed from seed data defined in the previous subsection changes under mutation .
this is crucial for uncovering the chamber structure of these diagrams and giving the connection with the exchange graph and cluster complex .
thus let @xmath560 and @xmath561 be the mutated seed ( see e.g. , @xcite , ( 2.3 ) ) . to distinguish the two lie algebras involved , we write @xmath562 and @xmath563 for the lie algebras arising from these two different seeds .
we recall that the injectivity assumption is independent of the choice of seed .
[ halfspacedef ] [ tkdef ] we set @xmath564 for @xmath560 , define the piecewise linear transformation @xmath565 by , for @xmath566 , @xmath567 as we will explain in [ tropsec ] , @xmath568 is the tropicalisation of @xmath569 viewed as a birational map between tori .
we will write @xmath570 and @xmath571 to be the linear transformations used to define @xmath568 in the regions @xmath572 and @xmath573 respectively .
define the scattering diagram @xmath574 to be the scattering diagram obtained by : 1 . for each wall @xmath575 , where @xmath576 , we have one or two walls in @xmath574 given as @xmath577 throwing out the first or second of these if @xmath578 or @xmath579 , respectively . here for @xmath580 linear , we write @xmath581 for the formal power series obtained by applying @xmath582 to each exponent in @xmath383 .
@xmath574 also contains the wall @xmath583 .
the main result of this subsection is : [ adiagrammutations ] suppose the injectivity assumption is satisfied .
then @xmath574 is a consistent scattering diagram for @xmath584 and @xmath585 .
furthermore , @xmath586 and @xmath574 are equivalent . the main point in the proof , which is not at all obvious from the definition ,
is that @xmath574 is a scattering diagram for @xmath563 , @xmath587 , where @xmath561 .
formally , consistency will be easy to check using consistency of @xmath378 .
it will follow easily that by construction @xmath588 and @xmath574 have the same incoming walls , so the theorem will then follow from the uniqueness in theorem [ kslemmags ] .
the main problem to overcome is that the functions attached to walls of @xmath378 and @xmath588 live in two different completed monoid rings , @xmath268}$ ] and @xmath589}$ ] , for @xmath265 a monoid chosen to contain @xmath590 , and @xmath591 a monoid chosen to contain @xmath592 .
we need first a common monoid @xmath593 containing both @xmath265 and @xmath591 .
[ pbardef ] let @xmath594 be a top - dimensional cone containing @xmath595 , @xmath263 , and @xmath596 , and such that @xmath597 .
set @xmath598 , and @xmath599 .
given such a choice of @xmath593 , we can find @xmath265 , @xmath591 contained in @xmath593 .
however , we have an additional problem that @xmath378 is not trivial modulo @xmath269 . indeed , @xmath600 , while one of the initial walls of @xmath601 is @xmath602 .
in particular , the wall - crossing automorphism associated to @xmath603 is not an automorphism of the ring @xmath604}$ ] , but rather of the localized ring @xmath604}_{1+z^{v_k}}$ ] .
( here the hats denote completion with respect to @xmath269 . )
this kind of situation is dealt with in @xcite , see especially 4.3 .
however the current situation is quite a bit simpler , so we will give the complete necessary arguments here and in appendix [ scatappendix ] .
we will use the notation @xmath605 for the automorphism of @xmath604}_{1+z^{v_k}}$ ] associated to crossing the wall @xmath606 from @xmath572 to @xmath573 .
explicitly , @xmath607 in this situation , define @xmath608 we note that by the definition of the mutated seed @xmath609 , @xmath610 , so we indicate it by @xmath611 .
we now extend the definition of scattering diagram .
[ escdef ] a _ wall _ for @xmath593 and ideal @xmath269 is a pair @xmath612 with @xmath613 as in definition [ walldef ] , but with @xmath614 , and @xmath615}$ ] congruent to @xmath616 mod @xmath269 .
the _ slab _ for the seed @xmath10 means the pair @xmath617 .
note since @xmath618 this does not qualify as a wall .
now a scattering diagram @xmath328 is a collection of walls and possibly this single slab , with the condition that for each @xmath329 , @xmath619 for all but finitely many walls in @xmath328 .
note that crossing a wall or slab @xmath325 now induces an automorphism of @xmath604}_{1+z^{v_k}}$ ] of the form @xmath620 ( with the localization only needed when a slab is crossed ) .
the following is proved in appendix [ scatappendix ] : [ kslemma2 ] there exists a scattering diagram @xmath621 in the sense of definition [ escdef ] such that 1 .
@xmath622 , 2 .
@xmath623 consisting only of outgoing walls , and 3 .
@xmath361 as an automorphism of @xmath604}_{1+z^{v_k}}$ ] only depends on the endpoints of @xmath341 .
furthermore , @xmath624 with these properties is unique up to equivalence . finally , @xmath624 is also a scattering diagram for the data @xmath625 , and as such is equivalent to @xmath378 .
[ noekwalls ] note in particular that the theorem implies @xmath380 does not contain any walls contained in @xmath626 besides @xmath606 .
indeed , no wall of @xmath627 is contained in @xmath626 : only the slab @xmath606 is contained in @xmath626 .
we write @xmath561 , @xmath628 .
we first note that we can choose representatives for @xmath378 , @xmath588 which are scattering diagrams in the sense of definition [ escdef ] , by theorem [ kslemma2 ] .
furthermore , @xmath574 is also a scattering diagram in the sense of definition [ escdef ] for the seed @xmath609 : this follows since if @xmath629 for some @xmath210 , we also have @xmath630 .
thus by the uniqueness statement of theorem [ kslemma2 ] , @xmath574 and @xmath588 are equivalent if ( 1 ) these diagrams are equivalent to diagrams which have the same set of slabs and incoming walls ; ( 2 ) @xmath574 is consistent .
we carry out these two steps .
_ step i. up to equivalence , @xmath574 and @xmath588 has the same set of slabs and incoming walls . _
if @xmath631 is outgoing , the wall @xmath613 contributes to @xmath574 and is also outgoing , so let us consider the incoming walls of @xmath632 . setting @xmath633 ,
already @xmath634 contains the slab for @xmath609 @xmath635 which lies in @xmath632 by construction .
next consider the wall @xmath636 , for @xmath637 .
we have three cases to consider , based on whether @xmath638 is zero , positive or negative .
first if @xmath639 , then @xmath568 takes the plane @xmath640 to itself ( in a piecewise linear way ) , and @xmath641 .
thus the wall @xmath642 contributes two walls @xmath643 whose union is the wall @xmath644 , as @xmath645 and @xmath646 in this case . up to equivalence
, we can replace these two walls with the single wall @xmath647 .
if @xmath648 , then consider the wall @xmath649 this wall contains the ray @xmath650 , so this is an incoming wall .
note that if @xmath651 , we have , with @xmath652 as given in , @xmath653_+e_k , m+v_k\langle d_ke_k , m\rangle\rangle\\ = { } & \{e_k , e_i\}\langle d_ke_k , m\rangle + d_k\{e_i , e_k\}\langle e_k , m\rangle\\ = { } & 0.\end{aligned}\ ] ] thus @xmath654 is a half - space contained in @xmath655 , and furthermore @xmath656 since @xmath657 thus we see that the wall @xmath658 of @xmath632 is half of the wall @xmath659 of @xmath634 .
if @xmath660 , then the wall @xmath661 coincides with @xmath662 , and @xmath572 also contains @xmath663 , so @xmath664 is an incoming wall .
but also @xmath646 , @xmath645 in this case .
thus @xmath664 is again half of the wall @xmath659 . in summary
, we find that after splitting some of the walls of @xmath634 in two , @xmath632 and @xmath634 have the same set of incoming walls , and thus , making a similar change to @xmath588 , we see that @xmath574 and @xmath588 have the same set of incoming walls .
_ step ii . _
@xmath665 for any loop @xmath341 for which this automorphism is defined . indeed , the only place a problem can occur is for @xmath341 a loop around a joint of @xmath378 contained in the slab @xmath606 , as this is where @xmath568 fails to be linear . to test this ,
consider a loop @xmath341 around a joint contained in @xmath606 .
assume that it has basepoint in the half - space @xmath572 and is split up as @xmath666 , where @xmath667 immediately crosses @xmath606 , @xmath668 is contained entirely in @xmath573 , crossing all walls of @xmath378 which contain @xmath491 and intersect the interior of @xmath573 , @xmath669 crosses @xmath606 again , and @xmath670 then crosses all relevant walls in the half - space @xmath572 .
let @xmath605 , @xmath671 be the wall - crossing automorphisms for crossing @xmath606 or @xmath672 passing from @xmath673 to @xmath674 , as in .
then by remark [ noekwalls ] , @xmath675 and @xmath676 .
let @xmath677\rightarrow{\bbbk}[m^{\circ}]$ ] be the automorphism induced by @xmath571 , i.e. , @xmath678 then note that @xmath679 thus to show @xmath680 , it is enough to show that @xmath681 but @xmath682 as desired . [ chamberstructureremark ] suppose given fixed data @xmath239 satisfying the injectivity assumption and seed data @xmath10 .
we then obtain for every seed @xmath609 obtained from @xmath10 via mutation a scattering diagram @xmath588 . in each case
we will choose a representative for the scattering diagram with minimal support .
note by construction and remark [ noekwalls ] , irrespective of the representative of @xmath378 used , @xmath378 contains walls whose union of supports is @xmath683 .
furthermore , we have @xmath684 given by , which can be written more explicitly as @xmath685 then @xmath686 are the closures of connected components of @xmath687 . similarly , we see that taking @xmath688 to be the chambers where all @xmath689 are positive ( or negative ) , we have that @xmath688 is the closure of a connected component of @xmath690 , so that @xmath691 is the closure of a connected component of @xmath687 .
note that the closures of @xmath692 and @xmath693 have a common codimension one face given by the intersection with @xmath626 .
this gives rise to the following chamber structure for a subset of @xmath694 .
recall from appendix [ ldsec ] the infinite oriented tree @xmath695 ( or @xmath696 ) used for parameterizing seeds obtained via mutation of @xmath10 . in particular , for any vertex @xmath697 of @xmath695 , there is a simple path from the root vertex to @xmath697 , indicating a sequence of mutations @xmath698 and hence a piecewise linear transformation @xmath699 note that @xmath700 is defined using the basis vector @xmath701 of the seed @xmath702 , not the basis vector @xmath701 of the original seed @xmath10 . by applying theorem [ adiagrammutations ] repeatedly
, we see that @xmath703 ( where @xmath704 applied to the scattering diagram @xmath378 is interpreted as the composition of the actions of each @xmath700 ) and @xmath705 is the closure of a connected component of @xmath687 .
note that the map from vertices of @xmath695 to chambers of @xmath706 is never one - to - one . indeed , if @xmath697 is the vertex obtained by following the edge labelled @xmath460 twice starting at the root vertex , one checks that @xmath707 , even though @xmath708 ( see @xcite , remark 2.5 ) .
thus we have a chamber structure on a subset of @xmath318 ; in general , the union of the cones @xmath709 do not form a dense subset of @xmath318 .
since we will often want to compare various aspects of this geometry for different seeds , we will write the short - hand @xmath710 for an object parameterized by a vertex @xmath697 where the root of the tree is labelled with the seed @xmath10 . in particular : [ fgchamberdef ] we write @xmath711 for the chamber of @xmath706 corresponding to the vertex @xmath697 . we write @xmath712 for the set of chambers @xmath711 for @xmath697 running over all vertices of @xmath696 .
we call elements of @xmath713 _ cluster chambers_.
we now explain that the chamber structure of construction [ chamberstructureremark ] coincides with the fock - goncharov cluster complex .
to do so , we first recall the basics of tropicalisation . for a lattice @xmath240 with @xmath241 , let @xmath714 be the subset of elements of the field of fractions of @xmath715 = h^0(t_n,\co_{t_n})$ ] which can be expressed as a ratio of laurent polynomials with non - negative integer coefficients .
then @xmath716 is a semi - field under ordinary multiplication and addition . for any semi - field @xmath265 ,
restriction to the monomials @xmath717 gives a canonical bijection @xmath718 where the first @xmath719 is maps of semi - fields , @xmath720 means the multiplicative group of @xmath265 , and in the last tensor product we mean @xmath720 viewed as @xmath721-module .
following @xcite we define the @xmath265-valued points of @xmath722 to be @xmath723 . a _ positive _ birational map @xmath724 means a birational map for which the pullback @xmath725 induces an isomorphism on @xmath714 .
obviously it gives an isomorphism on @xmath265-valued points .
thus it makes sense to talk about @xmath726 for any variety @xmath727 with a positive atlas of tori , for example many of the various flavors of cluster variety .
there are two equally good semi - field structures on @xmath728 , the max - plus and the min - plus structures . here
addition is either maximum or minimum , and multiplication is addition .
we notate these as @xmath729 and @xmath730 respectively , thinking of capital @xmath582 for the max - plus tropicalization and little @xmath188 for the min - plus tropicalization .
we similarly define @xmath731 and @xmath732 . thus taking @xmath733 or @xmath730
, we obtain the sets of tropical points @xmath734 or @xmath735 .
the former is the convention used by fock and goncharov in @xcite , so we refer to this as the fock - goncharov tropicalization .
the latter choice in fact coincides with @xmath736 as defined in @xcite , def .
1.7 , defined as a subset of the set of discrete valuations .
we refer to this as the geometric tropicalization .
it will turn out both are useful .
there is the obvious isomorphism of semi - fields @xmath737 from @xmath738 .
this induces a canonical sign - change identification @xmath739 .
given a positive birational map @xmath740 , we use @xmath741 and @xmath742 to indicate the induced maps @xmath743 and @xmath744 respectively . for the geometric tropicalization
, this coincides with the map on discrete valuations induced by pullback of functions , see @xcite , 1 . for cluster varieties
the two types of tropicalisation are obviously equivalent .
the geometric tropicalisation has the advantage that it makes sense for any log calabi - yau variety , while the fock - goncharov tropicalisation is restricted to ( fock - goncharov ) _ positive spaces _ , i.e. , spaces obtained by gluing together algebraic tori via positive birational maps .
we will use both notions , @xmath745 because in many cases it is more natural to think in terms of valuations / boundary divisors , and @xmath746 because , as we indicate below , the scattering diagram for building @xmath56 lives naturally in @xmath747 ( because of already established cluster sign conventions ) .
one computes easily that for the basic mutation @xmath748 the fock - goncharov tropicalisation is @xmath749_{+ } n\ ] ] while the geometric tropicalisation ( see @xcite , ( 1.4 ) ) is @xmath750_- n.\ ] ] thus : [ fgtprop ] @xmath751 defined in is the fock - goncharov tropicalisation of @xmath752 a rational function @xmath34 on a cluster variety @xmath15 is called _ positive _ if its restriction to each seed torus is positive , i.e. , can be expressed as a ratio of sums of characters with positive integer coefficients .
we can then define its fock - goncharov tropicalisation @xmath753 by @xmath754 .
similarly , for @xmath34 positive , we have its geometric tropicalisation @xmath755 which for each @xmath756 has value @xmath757 . using the identification of @xmath758 with @xmath17 ,
@xmath697 is interpreted as a valuation and @xmath759 coincides with @xmath760 , the value of @xmath697 on @xmath34 . in particular , this value is defined regardless of whether @xmath34 is positive .
we have a commutative diagram @xmath761 where @xmath210 is the canonical isomorphism determined by the sign change isomorphism .
the definition of @xmath762 in terms of valuations extends the definition of @xmath762 , and hence @xmath763 via this diagram , to any non - zero rational function .
we note that @xmath764 where @xmath765 is the canonical restriction isomorphism
. we will almost always leave @xmath766 out from the notation .
[ lplemma ] 1 . for a positive laurent polynomial @xmath767 ( i.e. , @xmath768 ) , and @xmath769 @xmath770 where @xmath51 is the canonical isomorphism .
if @xmath771 is a divisorial discrete valuation , and @xmath772 is any laurent polynomial ( so now @xmath773 ) , then @xmath774 by definition @xmath775 this gives the first statement . for the second
, we can assume @xmath776 is primitive , so part of a basis .
then the statement reduces to an obvious statement about the @xmath777 degree of a linear combination of monomials in @xmath778 $ ] .
note the mutations @xmath779 are precisely the mutations between the tori in the atlas for @xmath780 ( see appendix [ ldsec ] for the definition of the fock - goncharov dual @xmath780 , and @xcite , ( 2.5 ) for the mutations between @xmath27 tori in our notation ) .
thus by theorem [ adiagrammutations ] and proposition [ fgtprop ] , the support of @xmath378 viewed as a subset of @xmath781 under the identification @xmath782 ( induced canonically from the open set @xmath783 ) is independent of seed .
in particular it makes sense to talk about @xmath784 as being completely canonically defined without choosing any seed . for any seed
the chambers @xmath785 are connected components of @xmath786 .
we recall from @xcite : suppose given fixed data @xmath239 and suppose given an initial seed . for a seed @xmath787 obtained by mutation from the initial seed , the _ fock - goncharov cluster chamber associated to @xmath10 _
is the subset @xmath788 identified with @xmath789 via @xmath210 .
( fock - goncharov ) cluster complex _
@xmath118 is the set of all such chambers .
[ fgcclem ] suppose given fixed data @xmath239 satisfying the injectivity assumption and suppose given an initial seed . for a seed @xmath74 obtained by mutation from the initial seed ,
the chamber @xmath790 ( also identified with @xmath791 via @xmath210 ) is the fock - goncharov cluster chamber associated to @xmath10 .
hence the fock - goncharov cluster chambers are the maximal cones of a simplicial fan ( of not necessarily strictly convex cones ) .
in particular @xmath118 is identified with @xmath792 for any choice of seed @xmath10 giving an identification of @xmath793 with @xmath794 .
the identification of the chamber is immediate from the definition .
the result then follows from the chamber structure of remark [ chamberstructureremark ] and the fact that the @xmath568 are the fock - goncharov tropicalizations of the mutations @xmath569 for @xmath795 .
it s obvious each maximal cone is simplicial , and each adjacent pair of maximal cones meets along a codimension one face of each . hence we obtain a simplicial fan . [ sdprin ] see appendix [ rdsec ] for a review of the cluster variety with principal coefficients , @xmath56 .
any seed @xmath10 gives rise to a scattering diagram @xmath796 living in @xmath797 the second equality by proposition [ ldpprop ] , ( 3 ) .
indeed in this situation , the injectivity assumption is satisfied , since the form @xmath798 on @xmath799 is non - degenerate ( which is the reason we use @xmath56 instead of @xmath19 or @xmath27 ) . indeed
, the vectors @xmath800 are linearly independent .
note by theorem [ kslemma ] , @xmath801 contains the scattering diagram @xmath802 recall from proposition [ ldpprop ] that we have a canonical map @xmath803 which is defined on cocharacter lattices by the canonical projection @xmath804 , see .
thus the tropicalization @xmath805 coincides with this projection , which can be viewed as the quotient of an action of translation by @xmath240 . by definition [ walldef ] ,
walls of @xmath806 are of the form @xmath807 for @xmath259 .
thus all walls are invariant under translation by @xmath240 , and thus are inverse images of walls under @xmath808 .
so even though @xmath19 may not satisfy the injectivity assumption necessary to build a scattering diagram , we see that @xmath809 is the inverse image of a subset of @xmath793 canonically defined independently of the seed . in particular , note that the fock - goncharov cluster chamber in @xmath793 associated to the seed @xmath10 ( where @xmath810 for all @xmath263 ) pulls back to the corresponding fock - goncharov cluster chamber in @xmath811 . the following was conjectured by fock and goncharov , @xcite , 1.5 : [ fgccth ] for any initial data the fock - goncharov cluster chambers in @xmath781 are the maximal cones of a simplicial fan .
when the injectivity assumption holds , this follows from lemma [ fgcclem ] .
in particular it holds for @xmath56 .
now the general case follows by the above invariance of @xmath812 under the translation by @xmath240 .
[ standardexample ] consider the rank three skew - symmetric cluster algebra given by the matrix @xmath813 then projecting the walls of @xmath812 to @xmath814 via @xmath808 , one obtains a collection of walls in a three - dimensional vector space .
one can visualize this by intersecting the walls with the affine hyperplane @xmath815 .
the collection of resulting rays and lines appears on the first page of @xcite .
while fock and goncharov were not aware of scattering diagrams in this context , in fact there the picture represents the same slice of the cluster complex , and hence coincides with the scattering diagram .
the cluster complex in fact fills up the half - space @xmath816 .
there is no path through chambers connecting @xmath817 and @xmath693 .
this example is particularly well - known in cluster theory , and gives the cluster algebra associated with triangulations of the once - punctured torus .
we will explain how a scattering diagram determines a class of piecewise straight paths which will allow for the construction of theta functions .
the notion of broken line was introduced in @xcite , and developed from the point of view of defining canonical functions in @xcite and @xcite .
we choose fixed data @xmath239 and a seed @xmath10 as described in appendix [ ldsec ] , and assume it satisfies the injectivity assumption .
this gives rise to the group @xmath93 described in [ defconstsection ] which acts by automorphisms of @xmath268}$ ] for a choice of monoid @xmath265 containing @xmath818 and with @xmath819 .
the group @xmath93 also acts on the rank one free @xmath268}$]-module @xmath820}$ ] for any @xmath821 , with a log derivation @xmath273 acting on @xmath822 as usual to give @xmath823 .
we then have : [ blde ] let @xmath328 be a scattering diagram in the sense of definition [ ksscatdiagdef ] , @xmath824 and @xmath825 .
a _ broken line _ for @xmath826 with endpoint @xmath827 is a piecewise linear continuous proper path @xmath828\rightarrow m_{{\mathbb{r}}}\setminus{\operatorname{sing}}({\mathfrak{d}})$ ] with a finite number of domains of linearity .
this path comes along with the data of , for each domain of linearity @xmath829 $ ] of @xmath341 , a monomial @xmath830 $ ] .
this data satisfies the following properties : 1 .
if @xmath11 is the first ( and therefore unbounded ) domain of linearity of @xmath341 , then @xmath832 .
3 . for @xmath188 in a domain of linearity @xmath11 , @xmath833 .
4 . let @xmath834 be a point at which @xmath341 is not linear , passing from domain of linearity @xmath11 to @xmath835 .
let @xmath836 then @xmath837 is a term in the formal power series @xmath838 . note that since a broken line does not pass through a singular point of @xmath328 , we can write @xmath839 where @xmath840 is primitive , vanishes on each @xmath841 , and @xmath842 is positive by item ( 3 ) of the definition of broken line . it is an important feature of broken lines that we never need to invert .
[ ifdef ] let @xmath328 be a scattering diagram , @xmath824 , @xmath825 .
for a broken line @xmath341 for @xmath826 with endpoint @xmath827 , define @xmath843 ( @xmath844 is for initial ) , @xmath845 and @xmath846 to be the monomial attached to the final ( @xmath847 is for final ) domain of linearity of @xmath341 .
define @xmath848 where the sum is over all broken lines for @xmath826 with endpoint @xmath827 . for @xmath849
, we define for any endpoint @xmath827 @xmath850 in general , @xmath851 is an infinite sum , but makes sense formally : [ formalthfun ] @xmath852}$ ] .
it is clear by construction that for any broken line @xmath341 with @xmath853 , we have @xmath854 $ ] .
so it is enough to show that for any @xmath329 , there are only a finite number of broken lines @xmath341 such that @xmath853 , @xmath855 , and @xmath856 . first note by the assumption that @xmath857 , there are only a finite number of choices for @xmath858 such that @xmath859 .
fix a choice @xmath860 for @xmath858 .
second , to test that there are finitely many broken lines with @xmath853 , @xmath855 and @xmath861 , we can throw out any wall @xmath862 with @xmath863 , so we can assume @xmath328 is finite .
third , no broken line @xmath341 with @xmath864 can bend more than @xmath460 times .
thus there are only a finite number of possible ordered sequences of walls @xmath865 at which @xmath341 can bend .
fix one such sequence .
one then sees there are at most a finite number of broken lines with @xmath855 , @xmath861 bending at @xmath865 .
indeed , one can start at @xmath827 and trace a broken line backwards , using that the final direction is @xmath866 . crossing a wall @xmath867 and passing from domain of linearity @xmath11 ( for smaller @xmath188 ) to domain of linearity @xmath835 ( for larger @xmath188 )
, one sees that knowing the monomial attached to @xmath835 restricts the choices of monomial on @xmath11 to a finite number of possibilities .
this shows the desired finiteness .
the most important general feature of broken lines is the following : [ blinvth ] let @xmath328 be a consistent scattering diagram , @xmath868 , @xmath869 two points with all coordinates irrational .
then for any path @xmath341 with endpoints @xmath827 and @xmath870 for which @xmath361 is defined , we have @xmath871 this is a special case of results of 4 of @xcite . the condition that @xmath827 and @xmath870 both have irrational coordinates guarantees that we do nt have to worry about broken lines which pass through joints ( which we are nt allowing ) .
let us next consider how broken lines change under mutation .
so let @xmath10 be a seed , @xmath872 as in definition [ pbardef ] .
[ brlineinvmut ] @xmath568 defines a one - to - one correspondence between broken lines for @xmath826 with endpoint @xmath827 for @xmath378 and broken lines for @xmath873 with endpoint @xmath874 for @xmath586 .
in particular , depending on whether @xmath875 or @xmath673 , we have @xmath876 where the superscript indicates which scattering diagram is used to define the theta function , and @xmath877 acts linearly on the exponents in @xmath878 .
[ blrem ] by the proposition and proposition [ fgtprop ] , when the injectivity assumption holds , broken lines make sense in @xmath781 independent of a choice of seed .
given a broken line @xmath341 for @xmath378 , we define @xmath879 to have underlying map @xmath880\rightarrow m_{{\mathbb{r}}}$ ] .
subdivide domains of linearity of @xmath341 so that we can assume any domain of linearity @xmath11 satisfies @xmath881 or @xmath673 .
in the two cases , the attached monomial @xmath882 becomes @xmath883 or @xmath884 respectively .
we show that @xmath879 is a broken line for @xmath873 with endpoint @xmath874 , with respect to the scattering diagram @xmath586 , which is equal to @xmath574 , by theorem [ adiagrammutations ] .
indeed , the only thing to do is to analyze what happens when @xmath341 crosses @xmath626 .
so suppose in passing from a domain of linearity @xmath885 to a domain of linearity @xmath886 , @xmath341 crosses @xmath626 , so that @xmath887 is a term in @xmath888 assume first that @xmath341 passes from @xmath673 to @xmath674 .
then @xmath889 is a term in @xmath890 showing that @xmath879 satisfies the correct rules for bending as it crosses the slab @xmath891 of @xmath574 . if instead @xmath341 crosses from @xmath674 to @xmath673 , then @xmath892 is a term in @xmath893 so again @xmath879 satisfies the bending rule at the slab @xmath672 .
the map @xmath568 on broken lines is then shown to be a bijection by observing @xmath894 , similarly defined , is the inverse to @xmath568 on the set of broken lines .
the following , which shows that cluster variables are theta functions , is the key observation for proving positivity of the laurent phenomenon .
[ oneblprop ] let @xmath895 be a base - point and let @xmath896 .
then @xmath897 .
this says the only broken line with asymptotic direction @xmath860 and basepoint @xmath827 is @xmath898 , with attached monomial @xmath36 . to see this ,
suppose we are given a broken line @xmath828\rightarrow m_{{\mathbb{r}}}$ ] with asymptotic direction @xmath860 which bends successively at walls @xmath899 . for each @xmath210
, there is an @xmath900 such that @xmath901 .
multiplying @xmath902 by a positive integer if necessary , we can assume that the monomial attached to @xmath341 upon crossing the wall @xmath867 changes by a factor @xmath903 .
now if @xmath904 is the image of the @xmath905 linear segment of @xmath341 , we show inductively that @xmath906 indeed , @xmath907 for some @xmath908 , so initially @xmath885 is contained on the positive side of @xmath909 , i.e. , @xmath910 is positive on @xmath885 , and hence after bending at @xmath909 , we see @xmath911 .
next , assume true for @xmath912 .
then @xmath913 , and if @xmath914 is the time when @xmath341 bends at the wall @xmath606 , we have @xmath915 and @xmath916 by the induction hypothesis .
thus @xmath917 .
in addition , the derivative @xmath918 of @xmath341 along @xmath919 is @xmath920 , and @xmath921 by skew - symmetry of @xmath798 and @xmath922 .
thus @xmath923 .
since @xmath924 for all @xmath210 , any broken line with asymptotic direction @xmath860 which bends can not terminate at the basepoint @xmath925 .
this shows that there is only one broken line for @xmath860 terminating at @xmath926 .
[ monocor ] let @xmath927 be a cluster chamber , and let @xmath928 , @xmath929
. then @xmath897 .
note @xmath930 for some vertex @xmath697 of @xmath696 , with associated seed @xmath931 .
there is then a piecewise linear map @xmath932 with @xmath933 , see .
then the result follows by applying proposition [ oneblprop ] to @xmath934 , @xmath935 and proposition [ brlineinvmut ] . in the next section
, we will identify theta functions which are polynomials with universal laurent polynomials , i.e. , elements of the cluster algebra associated to the fixed and seed data .
it will follow from the above corollary that cluster monomials are in fact theta functions .
[ brokenlineexamples ] figures [ brokenline11 ] and [ brokenline22 ] show some examples of broken lines in the case of example [ bcexample ] with @xmath936 .
in the first figure , we take @xmath937 , and in the second , @xmath938 .
neither of these lie in the cluster complex : the union of all cones in the cluster complex is @xmath939 . in this case
the only bends occur on the original lines of @xmath940 , as any bending along the additional rays of the scattering diagram will result in the broken line shooting back out , unable to reach the first quadrant containing the basepoint @xmath827 . in the figures ,
the final line segment is labelled with its attached monomial , so that the theta function is a sum of these labels .
one finds @xmath941 in @xcite , it was shown that for any @xmath399 , with @xmath827 lying in the first quadrant , the @xmath942 with @xmath860 ranging over all elements of @xmath860 coincides with the greedy basis @xcite .
throughout this section we work with initial data @xmath239 satisfying the injectivity assumption , so we obtain the cluster chamber structure @xmath792 from @xmath378 described in construction [ chamberstructureremark ] . in particular , this condition holds for initial data @xmath943 , see appendix [ rdsec ] . in what follows
, we will often want to deal with multiple copies of @xmath944 etc . indexed either by vertices @xmath697 of @xmath696 or chambers @xmath945 . to distinguish these ( identical ) copies , we will use subscripts @xmath697 or @xmath946 , e.g. , the scattering diagram @xmath947 lives in @xmath948 , and chambers in @xmath947 give , under the identification @xmath949 , the fock - goncharov cluster complex @xmath118 , by lemma [ fgcclem ] .
in particular the cluster chambers of @xmath947 and @xmath950 are in canonical bijection .
[ capcon ] fix a seed @xmath10 .
we use the cluster chambers to build a positive space .
we attach a copy of the torus @xmath951 to each cluster chamber @xmath927 . given any two cluster chambers @xmath952 of @xmath792
, we can choose a path @xmath341 from @xmath953 to @xmath946 .
we then get an automorphism @xmath954 } \rightarrow\widehat{{\bbbk}[p]}$ ] which is independent of choice of path .
if we choose the path to lie in the support of the cluster complex , then by remark [ noekwalls ] ( which shows in particular that the scattering functions on walls of the cluster complex are polynomials , as opposed to formal power series ) , the wall crossings give birational maps of the torus and hence we can view @xmath361 as giving a well - defined map of fields of fractions @xmath955 this induces a birational map @xmath956 which is in fact positive .
we can then construct a space @xmath957 by gluing together all the tori @xmath958 , @xmath927 via these birational maps , see proposition 2.4 of @xcite .
we call this space ( with its atlas of tori ) @xmath959
. we write @xmath960 if we need to make clear which seed @xmath10 is being used .
we check first that mutation equivalent seeds give canonically isomorphic spaces .
we recall first something of the construction of @xmath19 .
fix a seed @xmath10 .
then we have positive spaces @xmath961 where each atlas is parameterized by vertices @xmath697 of the infinite tree @xmath696 .
we write e.g. , @xmath962 for the open subset parameterized by @xmath697 .
if we obtain a seed @xmath963 by mutation from @xmath10 , then we can think of the tree @xmath964 as a subtree of @xmath696 rooted at @xmath697 , and thus we obtain natural open immersions @xmath965 these are easily seen to be isomorphisms . under this immersion , the open cover of @xmath966
is identified canonically with the subcover of @xmath967 indexed by vertices of @xmath964 ( but in either atlas there are many tori identified with the same open set of the union ) . because of this we view @xmath19 as independent of the choice of seed in a given mutation equivalence class .
given vertices @xmath968 of @xmath696 , we have birational maps @xmath969 induced by the inclusions @xmath970 and @xmath971 respectively . in
what follows , we use the same notation for the restriction of a piecewise linear map to a maximal cone on which it is linear and the unique linear extension of this restriction to the ambient vector space .
[ rpprop ] let @xmath10 be a seed .
let @xmath697 be the root of @xmath696 , @xmath972 any other vertex .
consider the fock - goncharov tropicalisation @xmath973 of @xmath974 .
its restriction @xmath975 to each cluster chamber @xmath976 is a linear isomorphism onto the corresponding chamber @xmath977 .
the linear map @xmath978 induces an isomorphism @xmath979 these glue to give an isomorphism of positive spaces @xmath980 . in view of the proposition
, we can view @xmath981 as independent of the seed in a given mutation class .
it is enough to treat the case where @xmath972 is adjacent to @xmath697 via an edge labelled with @xmath560 , so that @xmath982 , as in general @xmath983 is the inverse of a composition of mutations @xmath984 .
note in this special case @xmath985 by proposition [ fgtprop ] , the definition of @xmath986 in appendix [ ldsec ] , and the formula for the @xmath987-cluster mutation @xmath569 ( see e.g. , @xcite , ( 2.5 ) ) .
so @xmath988 is the isomorphism determined by the linear map @xmath989
. the proposition amounts to showing commutativity of the diagram , for @xmath990 , @xmath991 , @xmath992 , @xmath993^{t_{v',\sigma}}\ar@![d]_{{{\mathfrak{p}}}_{\sigma , \tilde\sigma } } & t_{n^{\circ } , \sigma'\in { { \bf s } } ' } \ar@![d]^{{{\mathfrak{p}}}_{\sigma',\tilde\sigma'}}\\ t_{n^{\circ},\tilde \sigma \in { { \bf s}}}\ar[r]_{t_{v',\tilde\sigma } } & t_{n^{\circ},\tilde\sigma'\in { { \bf s}}'}}\ ] ] where in the left column @xmath994 indicates wall crossings in @xmath378 while in the right column the wall crossings are in @xmath588 .
if @xmath946 and @xmath995 are on the same side of the wall @xmath626 , then commutativity follows immediately from theorem [ adiagrammutations ] .
so we can assume that @xmath946 and @xmath995 are adjacent cluster chambers separated by the wall @xmath626 , and further without loss of generality that @xmath996 is non - negative on @xmath946 .
now by remark [ noekwalls ] there is only one wall of @xmath378 ( @xmath588 ) contained in @xmath626 , with support @xmath626 itself and attached function @xmath997 ( resp .
@xmath998 ) .
now it is a simple calculation : @xmath999 this gives the desired commutativity .
next we explain how to identify @xmath1000 with @xmath19 .
recall for each vertex @xmath697 of @xmath696 there is an associated cluster chamber @xmath1001 in the cluster complex .
while the atlas for @xmath959 is parameterized by chambers of @xmath792 , we can use a more redundant atlas indexed by vertices of @xmath696 , equating @xmath1002 with @xmath1003 . the open sets and the gluing maps in this redundant atlas are the same as in the original , but in the redundant atlas a given open set might be repeated many times
. [ rpth ] fix a seed @xmath10 .
let @xmath697 be the root of @xmath696 , @xmath972 any other vertex .
let @xmath1004 be the linear map @xmath1005 .
let @xmath1006 be the associated map of tori .
these glue to give an isomorphism of positive spaces @xmath1007 furthermore , the diagram @xmath1008 is commutative , where the right - hand vertical map is the isomorphism of proposition [ rpprop ] , the left - hand vertical map the isomorphism given in , and the horizontal maps are the isomorphisms just described .
let @xmath1009 .
the desired isomorphism is equivalent to commutativity of the diagram @xmath1010^{\psi_{v , v'}}\ar@![d]_{\mu_{v',v '' } } & t_{n^{\circ},v'}\subset{\mathcal{a}}_{{\mathrm{scat}},{{\bf s}}}\ar@![d]^{{{\mathfrak{p}}}_{{\mathcal{c}}^+_{v'},{\mathcal{c}}^+_{v''}}}\\ t_{n^{\circ},v''}\subset{\mathcal{a}}_{{{\bf s}}}\ar[r]_{\psi_{v , v '' } } & t_{n^{\circ},v''}\subset{\mathcal{a}}_{{\mathrm{scat}},{{\bf s } } } } \ ] ] where the right - hand vertical arrow is given by wall crossings in @xmath378 between the cluster chambers for @xmath1011 . for this we may assume there is an oriented path from @xmath972 to @xmath1012 in @xmath696 , and thus that @xmath1013 .
the commutativity of is equivalent to the commutativity of @xmath1014 where the right - hand vertical map is the restriction of the isomorphism @xmath1015 of proposition [ rpprop ] .
we argue the commutativity of first , and then show that this implies the commutativity of .
each map in is an isomorphism , induced by the restrictions of tropicalizations of various @xmath1016 to various cluster chambers .
explicitly , on character lattices , we have the corresponding diagram @xmath1017 which is obviously commutative as tropicalization is functorial and @xmath1018 .
now for the commutativity of .
it is enough to check the case when there is an oriented edge from @xmath972 to @xmath1012 in @xmath696 labelled by @xmath1019 .
we claim we may also assume @xmath1020 .
indeed , assume we have proven commutativity in this case .
we draw a cube , whose back vertical face is the diagram , and whose front vertical face is the analogous diagram for @xmath1021 , which is commutative by assumption .
the top and bottom horizontal faces are instances of , and the right - hand vertical face is the commutative diagram of atlas tori giving the isomorphism @xmath1015 of proposition [ rpprop ] .
finally the left - hand vertical face consists of equality of charts or birational maps coming from inclusions of these tori in @xmath967 or @xmath1022 , thus commutative .
now commutativity of the back vertical face , follows . finally , to show when @xmath1023 , i.e. , @xmath1024 , we note @xmath1025 is automatically the identity , and @xmath1026 is also the identity , by definition [ tkdef ] , and the identification of @xmath568 as fock - goncharov tropicalisation of the birational map of tori @xmath1027 .
thus the commutativity amounts to showing that the wall - crossing automorphism @xmath1028 of fraction fields , given by crossing the wall @xmath626 from the negative to the positive side , is the pullback on rational functions of the birational mutation @xmath1029 .
note the only scattering function on the wall is @xmath997 , so this follows from the coordinate free formula for the birational mutation , see e.g. , @xcite , ( 2.6 ) .
we can now make precise the relationship between theta functions and cluster monomials mentioned at the end of [ blsec ] .
first recall : [ clustermonodef ] given fixed and initial data @xmath1030 , if a seed @xmath1031 is given , with @xmath1032 the dual basis and @xmath1033 , a _ cluster monomial _ in this seed is a monomial on @xmath1034 of the form @xmath36 with @xmath1035 and the @xmath1036 non - negative for @xmath263 . by the laurent phenomenon @xcite ,
such a monomial always extends to a regular function on @xmath111 .
a _ cluster monomial on @xmath111 _ is a regular function which is a cluster monomial in some seed .
[ thetaequalsmonomial ] let @xmath239 be fixed data satisfying the injectivity assumption and @xmath10 an initial seed .
let @xmath1037 and @xmath1038 for some @xmath1039 .
then @xmath942 is a positive laurent polynomial which expresses a cluster monomial of @xmath19 in the initial seed @xmath10 .
further , all cluster monomials can be expressed in this way . by theorem [ rpth ] and proposition [ rpprop ]
we have a canonical isomorphism of positive spaces @xmath1040 .
let @xmath697 be the root of @xmath1041 and @xmath972 any vertex of @xmath696 .
then we have @xmath1042 , and the cluster monomials for the seed @xmath1021 are just the monomials @xmath36 on @xmath1043 with @xmath1044 . by theorem [ rpth ] , this is identified with the monomial @xmath1045 on @xmath1046 , as @xmath1047 takes @xmath1048 to @xmath1049 by proposition [ rpprop ] .
so the cluster monomials for the chart indexed by @xmath972 in @xmath1050 are of the form @xmath36 with @xmath1051 . furthermore ,
if for each vertex @xmath153 of @xmath696 , @xmath1052 is a general basepoint , we have @xmath1053 for @xmath1051 by corollary [ monocor ] . by the definition of @xmath959 in construction [ capcon ] , the corresponding rational function on the open set @xmath1054 is @xmath1055 , where @xmath341 is a path from @xmath1056 to @xmath1037 lying in the support of @xmath792 . but
@xmath1057 by theorem [ blinvth ] .
finally @xmath942 is a positive laurent series by theorem [ scatdiagpositive ] and the definition of broken lines . by the laurent phenomenon
, it is also a polynomial .
we can now remove the injectivity assumption to prove : [ pilpth ] each cluster variable of an @xmath19-cluster algebra is a laurent polynomial with non - negative integer coefficients in the cluster variables of any given seed .
since , as explained in proposition [ cvextrem ] , each cluster variable lifts canonically from @xmath19 to @xmath56 , we can replace the initial data @xmath239 with @xmath943 , for which the injectivity assumption holds .
the result then immediately follows from theorem [ thetaequalsmonomial ] .
[ capconpc ] when fixed and initial data @xmath239 , @xmath10 has frozen variables there is a partial compactification of cluster varieties @xmath142 , see construction [ fvss ] .
we have an analogous partial compactification @xmath1058 , given by an atlas of toric varieties @xmath1059 .
the choice of fans is forced by the identifications of proposition [ rpprop ] : for @xmath697 the root of @xmath696 , @xmath1060 ( @xmath1061 as in construction [ fvss ] ) and then @xmath1062 . now proposition [ rpprop ] and theorem [ rpth ] ( and their proofs ) extend to the partial compactifications without change .
one checks easily that all mutations in the positive spaces @xmath1063 , and all the linear isomorphisms between corresponding tori in the atlases for @xmath1064 preserve the monomials @xmath1065 , @xmath1066 ( these are the frozen cluster variables ) , so that all the spaces come with canonical projection to @xmath1067 , preserved by the isomorphisms between these positive spaces .
we shall see in the next section that in the special case of the partial compactification of @xmath56 , the relevant fans are particularly well - behaved .
we begin with some philosophy concerning log calabi - yau varieties following on from the discussion of @xcite , 1 .
suppose @xmath1068 are both log calabi - yau and @xmath15 is a zariski open subset of @xmath1 , both having maximal boundary ( @xcite , definition 1.5 ) .
the tropical sets ( which are expected to parameterize the theta function basis of functions on the mirror ) of @xmath1 and @xmath15 are canonically equal and we expect the mirror @xmath1069 to degenerate to the mirror @xmath161 . in particular when @xmath1070 is an algebraic torus , we expect a canonical degeneration of @xmath1069 to the dual torus @xmath1071 , under which the theta functions degenerate to monomials ( i.e. , characters ) .
when @xmath1072 is an @xmath19-cluster variety , and @xmath1073 is a cluster torus , it turns out this degeneration has a purely cluster construction : the choice of seed @xmath10 determines a canonical partial compactification @xmath1074 of @xmath21 , see proposition [ ldpprop ] and remark [ aprinremark ] .
the main point of this section is to show that @xmath1075 , corollary [ mutcor ] .
this degeneration is central to what follows in this paper .
for example , we prove linear independence of theta functions by showing they restrict to different characters on @xmath1076 , and the fock - goncharov conjecture , false in general , is true in a formal neighborhood of this fibre
. there are analogous degenerations ( identified with this one when the fock - goncharov conjecture holds ) for e.g. , @xmath1077 , and here they are even more central , being the main tool we have for proving properties of this algebra ( e.g. , that its spectrum gives a gorenstein log calabi - yau of the right dimension ) , see theorem [ cfcor ] . the equality @xmath1075 , while not at all obvious from the cluster atlas , is immediate using the alternative description @xmath959 of the previous section , as we now explain .
further , there are some immediate benefits , such as sign coherence of @xmath130-vectors . for the remainder of the paper ,
the only scattering diagram we will ever consider is @xmath806 , see construction [ sdprin ] .
so we will often omit the superscript from the notation .
[ atlascon ] fix a seed @xmath10 for fixed data @xmath239 . by construction [ capcon ] , the scattering diagram @xmath1078 gives an atlas for the space @xmath959 .
( technically , we should write @xmath1079 to indicate we are constructing something isomorphic to @xmath56 , however this will make the notation even less readable . ) this was constructed by attaching a copy @xmath1080 of the torus @xmath1081 to each cluster chamber @xmath927 , and ( compositions of ) wall crossing automorphisms give the birational maps between them . by theorem [ rpth ] this space is canonically identified with @xmath56 : @xmath56 has an atlas of tori @xmath1082 parameterized by vertices @xmath153 of @xmath696 , and we have canonical isomorphisms @xmath1083 for each vertex @xmath153 which induce the isomorphism @xmath1084 . in what follows , if @xmath153 is a vertex of @xmath696 , we write @xmath1085 for the seed obtained by mutating @xmath1086 ( see ) via the sequence of mutations dictated by the path from the root @xmath697 of @xmath696 to @xmath153 .
as described in remark [ aprinremark ] , the initial seed @xmath10 determines the partial compactification @xmath1087 , given by the atlas of toric varieties @xmath1088 where @xmath1089 is the cone generated by the subset of basis vectors of @xmath1090 corresponding to the second copy of @xmath844 . by remark [ capconpc ]
, the seed @xmath10 also determines a partial compactification @xmath1091 ( the superscript , thus the seed _ close _ to the overline in the notation , is responsible for the partial compactification ) , given by an atlas of toric varieties .
explicitly , if @xmath153 is a vertex of @xmath696 , the fan @xmath1089 yields the partial compactification of @xmath1092 in @xmath124 , and this is identified with @xmath1093 via @xmath1094 under the isomorphism @xmath1095 of theorem [ rpth ] .
thus the fan giving the partial compactifaction of @xmath1093 is @xmath1096 in fact , this fan is easily calculated : [ clem ] the cones @xmath1097 , and thus the toric varieties in the atlas for the partial compactification @xmath1091 , are the same for all @xmath153 .
each is equal to the cone spanned by the vectors @xmath1098 , where @xmath74 and @xmath1099 denotes the dual basis .
@xmath1100 is the given cone , by definition of the seed @xmath1086 . by construction [ fvss ]
the other fans are given by applying the geometric tropicalisation of the birational gluing of the tori in the atlas for @xmath959 .
these birational maps are given by wall crossings in @xmath378 .
but for each wall between cluster chambers the wall crossing is a standard mutation @xmath1101 , notation as in [ tropsec ] , for some @xmath1102 .
the attached scattering function is @xmath1103 for some @xmath510 in the convex hull of @xmath1104 , and @xmath1105 .
but then @xmath1106
. thus the geometric tropicalisation @xmath1107 fixes all the @xmath1108 by , and so the fan is constant . [ mutcor ] fix a seed @xmath10 , and let @xmath697 be the root of @xmath696 .
the following hold : 1 .
the fibre of @xmath1109 over @xmath81 is @xmath22 .
( see proposition [ ldpprop ] for the definition of @xmath1110 ) .
the mutation maps @xmath1111 for the atlas of toric varieties defining @xmath1112 are isomorphisms in a neighborhood of the fibre over @xmath1113 .
3 . for the partial compactification @xmath1114 with atlas corresponding to cluster chambers of @xmath378 ,
the corresponding mutation map between two charts ( which by lemma [ clem ] has the same domain and range ) is an isomorphism in a neighborhood of the fibre @xmath1113 and restricts to the identity on this fibre .
it is clear that ( 3 ) implies ( 2 ) implies ( 1 ) . for ( 3 ) , the scattering diagram @xmath378 is trivial modulo the @xmath1115 ( which pulls back to @xmath1116 ) , because this holds for the initial walls , with attached functions @xmath1117 .
now for any adjacent vertices @xmath1118 , the birational gluing map @xmath1119 is given on the level of monomials by @xmath1120 for a regular function @xmath34 on @xmath1121 and some @xmath1122 and any @xmath1123 , and by the above @xmath34 is identically @xmath616 when restricted to the torus where the @xmath1115 are zero . on the other hand
, this birational map gives an isomorphism between the open subsets of @xmath1124 and @xmath1121 where @xmath34 is non - zero .
in particular , the gluing maps are isomorphisms in the neighbourhood of the fibre where all @xmath1115 vanish and are the identity on that fibre .
the proof of the corollary shows the utility of constructing @xmath1125 as the positive space @xmath1091 associated to the cluster chambers in the scattering diagram @xmath378 .
next we show sign coherence of @xmath130-vectors follows easily from the corollary . in what follows , given a seed @xmath1126 obtained via mutation from @xmath1086 , we write @xmath1127 for the @xmath1128 exchange matrix for this seed , with @xmath1129 the @xmath130-vectors of this seed are the rows of the right - hand @xmath1130 submatrix .
[ sccveccor ] for any vertex @xmath153 of @xmath696 and fixed @xmath460 satisfying @xmath1131 , either the entries @xmath1132 , @xmath1133 are all non - positive , or these entries are all non - negative .
the result follows directly from corollary [ mutcor ] by writing down the mutation in cluster coordinates . following the notation given in appendix [ rdsec ]
, we have the fixed seed @xmath1134 which determines @xmath1087 and the family @xmath1074 .
the corresponding initial seed for @xmath124 is @xmath1135 and the coordinate @xmath1115 on @xmath1136 pulls back to @xmath1116 on @xmath124 .
these are the frozen cluster variables for @xmath124 .
note @xmath1137 where @xmath1138 is the dual basis to the basis @xmath1139 of @xmath1140 .
a vertex @xmath1141 corresponds to a seed @xmath1142 for @xmath240 with corresponding seed @xmath1143 for @xmath1144 , with @xmath1145 obtained from @xmath1086 by a sequence of mutations .
the @xmath1146 are no longer necessarily given by the @xmath1147 .
write @xmath1148 , @xmath1149 for the corresponding basis of @xmath1150 .
the cluster variables on the corresponding torus @xmath1151 are @xmath1152 .
say @xmath1153 is a vertex of @xmath696 adjacent to @xmath1141 along an edge labelled by @xmath460 .
then @xmath1154 and the cluster coordinates are @xmath1155 . since the last @xmath510 cluster variables are frozen , @xmath1156 , @xmath1157 .
the fan @xmath1158 determining a toric variety in the atlas for @xmath124 consists of a single cone spanned by @xmath1159 , and @xmath1160 similarly @xmath1161 the mutation @xmath1162 is given by the exchange relation @xcite ( 2.15 ) ( see @xcite , ( 2.8 ) in our notation ) which is , with @xmath1163 , @xmath1164_+ } + \prod_{i=1}^{2n } ( a_i')^{-[\tilde \epsilon_{ki}]_- } ) \\
= { } & ( a_k')^{-1 } ( p_k^+ \prod_{i=1}^{n } ( a_i')^{[\tilde\epsilon_{ki}]_+ } + p_k^{- } \prod_{i=1}^{n } ( a_i')^{-[\tilde\epsilon_{ki}]_- } ) \\ \mu_{k}^*(x_i)= { } & x_i\end{aligned}\ ] ] where @xmath1165 now @xmath569 fails to be an isomorphism exactly along the vanishing locus of @xmath1166_+ } + p_k^{- } \prod_{i=1}^{n } ( a_i')^{-[\tilde\epsilon_{ki}]_-}.\ ] ] this locus is disjoint from the central fibre @xmath1113 by corollary [ mutcor ] . on the other hand
it is disjoint from the central fibre if and only if exactly one of @xmath1167 is the empty product , i.e. , the constant monomial @xmath616 .
sign coherence is the statement that at least one of @xmath1168 , @xmath1169 is the empty product .
recall from definition [ clustermonodef ] the notion of cluster monomial , and also recall from proposition [ ldpprop ] , ( 2 ) the @xmath1076-action on @xmath1170 .
[ gvecdef ] by proposition [ cvextrem ] , the choice of seed @xmath10 provides a canonical extension of each cluster monomial on @xmath111 to a cluster monomial on @xmath1170 .
each cluster monomial on @xmath56 is a @xmath22-eigenfunction under the above @xmath1076 action .
the _ @xmath41-vector _ with respect to a seed @xmath10 ( see @xcite , ( 6.4 ) ) associated to a cluster monomial of @xmath19 is the @xmath22-weight of its lift determined by @xmath10 .
we now give an alternative description of @xmath41-vectors , which will lead to a more intrinsic definition of @xmath41-vector ( definition [ intrinsicga ] ) .
this in turn generalises to all the different flavors of cluster varieties ( definition [ fgccthg0 ] ) .
[ gprop ] fix a seed @xmath10 , giving the partial compactification @xmath1087 and @xmath22-equivariant @xmath1074 .
the central fibre @xmath1171 is a @xmath22-torsor .
let @xmath1172 be a cluster monomial on @xmath1173 and @xmath1174 the corresponding lifted cluster monomial on @xmath124 .
this restricts to a regular non - vanishing @xmath22-eigenfunction along @xmath1171 , and so canonically determines an element of @xmath1175 ( its weight ) .
this is the @xmath41-vector associated to @xmath1172 .
let @xmath1176 determine the seed in which @xmath1172 is defined as a monomial . by corollary [ mutcor ] , all mutations are isomorphisms near the central fibre of @xmath1110 , so it s enough to check that @xmath1172 is regular on the toric variety @xmath1177 , and restricts to a character on its central fibre .
but this is true by construction : if the seed @xmath1178 is @xmath1179 , then the cluster variables for the seed @xmath1085 on the torus @xmath1082 are @xmath1180 and @xmath1089 is the fan with rays spanned by the @xmath1181 .
thus the lift @xmath1174 of @xmath1172 is regular on @xmath1182 , and hence is regular in a neighbourhood of @xmath1183 .
furthermore , it is non - zero on @xmath1171 since the canonical lift only involves monomials @xmath1184 , which are non - vanishing on the strata of @xmath1182 .
the final statement follows since the restriction of the variable to the central fibre will have the same @xmath22-weight , as the map @xmath1110 is @xmath22-equivariant , and @xmath22 fixes @xmath1185 .
[ intrinsicga ] writing @xmath1186 , let @xmath1172 be a cluster monomial of the form @xmath36 on a chart @xmath1187 , @xmath1188 .
note that @xmath1189 for all @xmath210 , so after identifying @xmath793 with @xmath1190 , @xmath860 yields a point in the fock - goncharov cluster chamber @xmath1191 , as defined in lemma [ fgcclem ] .
we define @xmath1192 to be this point of @xmath1193 .
[ gvecthm ] let @xmath1172 be a cluster monomial on @xmath19 , and fix a seed @xmath10 giving an identification @xmath1194 .
then under this identification , @xmath1192 is the @xmath41-vector of the cluster monomial @xmath1172 with respect to @xmath10 .
we first note that if @xmath1172 is a monomial @xmath36 on the chart @xmath1187 with @xmath1195 , @xmath1196 , then the image of @xmath1192 under the identification @xmath1197 is @xmath1198 , where as usual @xmath1199 is the rational map induced by the inclusions @xmath1200 .
the choice of the seed @xmath10 gives the lift of @xmath1172 to a cluster monomial @xmath1174 on @xmath1170 . using the identification of @xmath1170 with @xmath957 ,
@xmath1174 is identified with a monomial of the form @xmath1201 on the chart @xmath1092 ( or @xmath1202 , depending on how one chooses to parametrize charts of @xmath957 ) .
let @xmath697 be the root of @xmath696 .
by lemma [ clem ] , the corresponding chart of @xmath1203 is the toric variety defined by the fan @xmath1097 . by proposition [ gprop ] , @xmath1174 is a regular function on @xmath1124 which is non - vanishing along @xmath1171 .
the @xmath1076 weight is the @xmath41-vector . since @xmath1097 is the cone spanned by @xmath1204 in @xmath1205 , where @xmath787 , one sees that @xmath1206 . thus to show the corollary , it is enough to show that @xmath1207 .
note however a similar statement is already true at the level of @xmath1170 .
indeed , in the chart @xmath1092 of @xmath1170 , the monomial @xmath1174 takes the form @xmath1208 for some @xmath1209 , and @xmath1210 lies in the positive chamber of @xmath1211 .
but @xmath1212 is the image of this positive chamber under the map @xmath1213 , where now @xmath1214 is the map induced by the inclusions @xmath1215 .
now @xmath1216 by theorem [ rpth ] , and @xmath1217 , so we see that @xmath1218 . now because there is a well - defined map @xmath1219 by proposition [ ldpprop ] , ( 4 ) , with @xmath808 given by projection onto @xmath113 , this projection @xmath808 is compatible with the tropicalizations @xmath1220 and @xmath1221 , i.e. , @xmath1222 .
thus @xmath1223 , as desired .
this corollary shows us how to generalize the notion of @xmath41-vector to any cluster variety : [ fgccthg0 ] let @xmath5 be a cluster variety , suppose that @xmath34 is a global monomial ( see definition [ globalmonomialdef ] ) on @xmath15 , and let @xmath10 be a seed such that @xmath1224 is the character @xmath36 , @xmath1225 .
define the _
@xmath41-vector _ of @xmath34 to be the image of @xmath860 under the identifications of [ tropsec ] : @xmath1226 we write the @xmath41-vector of @xmath34 as @xmath1227 .
note that the definition as given is not clearly independent of the choice of seed @xmath10 , but for a cluster variety of @xmath19 type , the previous corollary shows this .
this independence will be shown in general in lemma [ fgccthg2 ] .
by @xcite , the sign coherence for @xmath130-vectors ( proved in corollary [ sccveccor ] here ) , implies a sign coherence for @xmath41-vectors .
here we give a much shorter proof using the above description of @xmath41-vectors .
[ scgvec ] fix initial seed @xmath1228 , with @xmath1229 as usual . given any mutation equivalent seed @xmath609 , the @xmath1230-coordinates of the @xmath41-vectors for the cluster variables of this seed , expressed in the basis @xmath1231 , are either all non - negative , or all non - positive . by corollary [ gvecthm ] ,
the @xmath41-vectors in question are the generators of a chamber in the cluster complex of @xmath10 , defined as the images of the cluster chambers of @xmath806 under the projection @xmath808 , by theorem [ fgccth ] .
the hyperplanes @xmath640 are thus walls in the cluster complex . in particular
, @xmath1232 is either non - negative everywhere on a chamber , or non - positive everywhere on a chamber .
the theorem follows . for future reference
, we record the relationship between @xmath130-vectors and the cluster chambers in the case of no frozen variables . fix a seed @xmath10 . by lemma [ fgcclem
] , each mutation equivalent seed @xmath1195 has an associated cluster chamber @xmath1233 .
this is a full dimensional strictly simplicial cone , generated by a basis of @xmath113 consisting of @xmath41-vectors of the cluster variables @xmath1234 of @xmath1235 .
the facets of @xmath1212 are thus in natural bijection with the elements of @xmath10 ( or the indices in @xmath1236 ) .
[ facelem ] the facet of @xmath1212 corresponding to @xmath1237 is the intersection of @xmath1212 with the orthogonal complement of the @xmath130-vector for the corresponding element of @xmath1238 ( the corresponding mutation of the langlands dual seed @xmath1239 , see appendix [ ldsec ] ) .
furthermore , each @xmath130-vector for @xmath1238 is non - negative on @xmath1212 .
this is the content of @xcite , theorem 1.2 , the condition ( 1.8 ) of @xcite holding by our corollary [ sccveccor ] .
the @xmath41-vectors used in @xcite are precisely the @xmath41-vectors of the cluster variables @xmath1240 .
we close this section by identifying the cluster complex @xmath118 with the fomin - zelevinsky exchange graph .
the identification yields a proof of part of @xcite , conjecture 4.1 .
this was originally proven in @xcite , but we include our alternative argument as the statements follow quite easily from our techniques .
[ cdsth ] a seed for a cluster algebra without frozen variables is uniquely determined by its cluster , i.e. , the ordered collection of global functions @xmath1241 determines the exchange matrix @xmath652 with entries @xmath1242 .
the vertices of the exchange graph can be identified with the clusters , with two clusters adjacent if and only if their intersection has cardinality @xmath1243 .
[ rbijlem ] if for two mutation equivalent seeds @xmath1244 the open subsets @xmath1245 are the same , then the open subsets @xmath1246 are the same , and the open subsets @xmath1247 are the same .
we first claim that a chart in the atlas for @xmath56 ( or any @xmath19-type cluster variety ) is the complement of the union of zero loci of global functions which restrict to characters on the chart .
indeed , let @xmath1 be such a chart and @xmath15 be the complement of the union of such zero loci .
clearly @xmath1248 , since characters have no zeroes on @xmath1 .
on the other hand , by the laurent phenomenon , the cluster variables on @xmath1 extend to global monomials , and are non - vanishing on @xmath15 .
these functions define a map @xmath1249 which is clearly the identity when restricted to @xmath1 .
since @xmath1170 is separated ( see @xcite , theorem 3.14 ) , it follows that @xmath1250 .
let @xmath1251 be the vertices of @xmath695 corresponding to the seeds @xmath10 and @xmath609 .
for the statement about @xmath56 it s now enough to show that @xmath1252 is a character for any global monomial @xmath1174 which restricts to a character on @xmath1253 .
now @xmath1254 , and this is a character by the assumption that @xmath1255 inside @xmath111 .
by positivity of the laurent phenomenon , corollary [ pilpth ] , @xmath1252 is a laurent polynomial with non - negative integer coefficients . since its restriction to some subtorus is a character , it must be a character .
this completes the proof for @xmath56 .
this implies the result for @xmath27 , which is the quotient of @xmath56 by @xmath1076 , see proposition [ ldpprop ] , ( 2 ) .
we have the following alternative description of @xmath130-vectors in the case of no frozen variables .
fix a seed @xmath10 . by lemma [ fgcclem
] , each mutation equivalent seed @xmath1195 has an associated cluster chamber @xmath1233 .
this is a full dimensional strictly simplicial cone , generated by a basis of @xmath113 consisting of @xmath41-vectors of the cluster variables @xmath1234 of @xmath1235 .
the facets of @xmath1212 are thus in natural bijection with the elements of @xmath10 ( or the indices in @xmath1236 ) . [ facelem ]
the facet of @xmath1212 corresponding to @xmath1237 is the intersection of @xmath1212 with the orthogonal complement of the @xmath130-vector for the corresponding element of @xmath1238 ( the corresponding mutation of the langlands dual seed @xmath1239 , see appendix [ ldsec ] ) .
furthermore , each @xmath130-vector for @xmath1238 is non - negative on @xmath1212 .
this is the content of @xcite , theorem 1.2 , the condition ( 1.8 ) of @xcite holding by our corollary [ sccveccor ] .
the @xmath41-vectors used in @xcite are precisely the @xmath41-vectors of the cluster variables @xmath1240 .
we introduce notation for the @xmath130-vectors as follows . for fixed seed @xmath10
we write @xmath1256 for the @xmath130-vector , with fixed seed @xmath10 , whose @xmath1257 entry is @xmath1258 , see .
[ cdeexlem ] fix a seed @xmath10 .
if we view the @xmath130-vectors as vectors of @xmath240 expressed in the basis @xmath10 , then @xmath1259 is the @xmath905 basis vector of @xmath1260 , i.e. , the @xmath130-vectors for a seed adjacent to the initial fixed seed @xmath10 are just the basis elements of this seed .
these @xmath130-vectors , for all seeds adjacent to @xmath10 , determine the exchange matrix for @xmath10 . also , the chambers @xmath1261 together with the adjacent chambers determine the exchange matrix for @xmath10 .
the explicit formula for the @xmath130-vectors is an easy exercise applying the definition , see e.g. , @xcite , ( 2.4 ) .
the statement that the @xmath130-vectors for seeds adjacent to @xmath10 determines the exchange matrix is obvious from the formula for mutations , e.g. , @xcite , ( 2.3 ) .
the statement that the given chambers determine the exchange matrix for @xmath10 then follows .
indeed , the chambers determine the @xmath130-vectors for @xmath1239 and @xmath1262 by lemma [ facelem ] , and hence determines the exchange matrix for @xmath1239 .
but that itself determines the exchange matrix for @xmath10 because langlands duality is an involution .
suppose two seeds @xmath1244 have the same cluster .
we show the exchange matrices are the same .
the cluster determines the open subset in the atlas for @xmath19 ( it s the locus where the cluster variables are all non - zero , c.f .
the proof of lemma [ rbijlem ] ) , thus @xmath1263 and so by lemma [ rbijlem ] , @xmath1264 .
thus the associated birational mutation @xmath1265 is a regular isomorphism , and thus the piecewise linear mutation @xmath1266 is a linear isomorphism . because the clusters are equal it takes @xmath693 to @xmath1267 ( the chamber is determined by the tropicalisations of the cluster variables ) .
note these are chambers for the scattering diagram for @xmath1268 .
now by lemma [ cdeexlem ] the exchange matrices for @xmath1239 and @xmath1269 are the same .
this implies the exchange matrices for @xmath10 and @xmath609 are the same ( see appendix [ ldsec ] for a review of langlands duality ) .
suppose there are no frozen variables .
viewed as an abstract fan , the cluster complex @xmath118 in @xmath237 is the dual to the fomin - zelevinsky exchange graph : the maximal cones are in bijection with equivalence classes of seeds ( where two seeds are defined to be equivalent if , possibly after permutation of the indices @xmath844 respecting the @xmath251 , the cluster variables and pairings @xmath1270 coincide ) , with their edges corresponding to the cluster variables of the seed . by the fock - goncharov definition of the cluster complex , each cluster determines a maximal cone of @xmath118 , and we know @xmath118 forms a fan in @xmath237 by lemma [ fgcclem ] .
thus we have a map from the fomin - zelevinsky exchange graph to the dual graph to @xmath118 , which is obviously surjective .
but the chamber corresponding to a seed determines the cluster .
indeed , by proposition [ oneblprop ] , the theta functions associated to points of the chamber are the cluster monomials for this cluster , and the theta functions corresponding to primitive generators of the rays of the chamber ( which we recall is a simplicial cone ) are the variables of the cluster .
now the cluster determines the seed by theorem [ cdsth ] .
in this section we associate in a canonical way to every universal laurent polynomial @xmath41 on @xmath56 a formal sum @xmath1271 , @xmath1272 , which , roughly speaking , converges to @xmath41 at infinity in each partial compactification @xmath1087 .
to give such an expression for a single @xmath10 is quite easy , see propositions [ thetabasislemma ] and [ hatinfun ] .
the crucial point is that remarkably these coefficients are independent of @xmath10 , see theorem [ ffgth ] , our alternative to the fock - goncharov conjecture ( which fails in general ) .
this establishes the connection between @xmath139 and @xmath132 , and is key to one of our main technical results , see the proof of proposition [ markscor ] .
choose a seed @xmath1134 .
we let @xmath1273 , @xmath1274 $ ] , set @xmath1275/i_{{{\bf s}}}^{k+1 } , \quad \oca^{{{\bf s}}}_{{\operatorname{prin}},k}=\oca_{{\operatorname{prin}}}^{{{\bf s}}}\times_{{\mathbb{a}}^n_{x_1,\ldots , x_n } } { \mathbb{a}}^n_{(x_1,\ldots , x_n),k},\ ] ] and write the map induced by @xmath1276 also as @xmath1277 we use the notation @xmath1278 for a variety @xmath0 , so that e.g. , @xmath1279 is the upper cluster algebra .
we define @xmath1280 note that for any @xmath1281 , @xmath1282 for some monomial @xmath1283 in the @xmath1115 .
this induces a canonical inclusion @xmath1284 } { \bbbk}[n]\ ] ] where @xmath1285 is the monoid generated by @xmath1286 .
let @xmath1287 be the projection , and set @xmath1288 recall a choice of seed @xmath74 determines a scattering diagram @xmath1289 with initial walls @xmath1290 for @xmath263 .
we let @xmath1291 be the monoid generated by @xmath1292 .
we have the cluster complex @xmath792 of cones in @xmath1293 , with cones @xmath1294 for each vertex @xmath697 of @xmath696 . similarly to the above discussion
, @xmath1295 induces maps @xmath1296 the isomorphism between @xmath1297 and @xmath1298 discussed in construction [ atlascon ] restricts to give an isomorphism between @xmath1299 and @xmath1300 .
furthermore , as @xmath1298 is described by gluing charts isomorphic to @xmath1301 with @xmath1302 the cone generated by @xmath1303 for every chart by lemma [ clem ] , in fact @xmath1300 is described by gluing charts parameterized by @xmath927 isomorphic to @xmath1304 note for @xmath1305 , the birational map @xmath1306 between the charts of @xmath1298 indexed by @xmath946 and @xmath953 restrict to isomorphisms @xmath1307 : this is implied by corollary [ mutcor ] , ( 3 ) .
we choose a generic basepoint @xmath1308 for each @xmath1309 .
then for any @xmath1310 , by proposition [ formalthfun ] , we obtain as a sum over broken lines a well - defined series @xmath1311}\ ] ] satisfying by theorem [ blinvth ] @xmath1312 the following definition will yield the structure constants for the @xmath1313 : [ structuredef ] let @xmath1314 .
let @xmath1315 be chosen generally .
there are at most finitely many pairs of broken lines @xmath1316 with @xmath1317 , @xmath1318 and @xmath1319 ( see definition [ ifdef ] for this notation ) .
we can then define @xmath1320 the integers @xmath1321 are non - negative . by definition of scattering diagram for @xmath378 ,
all walls @xmath1322 have @xmath1323}$ ] . note also that because @xmath1324 comes from a strictly convex cone , any element of @xmath1324 can only be written as a finite sum of elements in @xmath1324 in a finite number of ways . in particular , as @xmath1325 , we can write @xmath1326 for @xmath1327 . thus we have @xmath1328 and there are only a finite number of possible @xmath1329 , @xmath1330 .
so with @xmath1331 fixed , there are only finitely many possible monomial decorations that can occur on either @xmath1332 . from this finiteness
is clear , c.f .
the proof of proposition [ formalthfun ] .
the non - negativity statement follows from theorem [ scatdiagpositive ] , which implies @xmath1333 for any broken line @xmath341 .
[ ckdef ] for a monoid @xmath1334 a lattice , we write @xmath1335 for the subset of elements which can be written as a sum of @xmath460 non - invertible elements of @xmath1336 . [ thetabasislemma ] notation as above .
the following hold : 1 . for @xmath1337
, @xmath1338 is a regular function on @xmath1339 , and the @xmath1338 as @xmath946 varies glue to give a canonically defined function @xmath1340 .
2 . for each @xmath1341 and @xmath1342 , we have @xmath1343 , and thus the @xmath1344 for @xmath1345 canonically define @xmath1346}{\bbbk}[n].\ ] ] the @xmath46 are linearly independent , i.e. , we have a canonical inclusion of @xmath8-vector spaces @xmath1347 } { \bbbk}[n].\ ] ] 3 . @xmath1348 } { \bbbk}[n]\ ] ] for @xmath1349 chosen sufficiently close to @xmath908 .
in particular , @xmath1321 is independent of the choice of @xmath1350 sufficiently near @xmath908 , and we define @xmath1351 for @xmath1350 chosen sufficiently close to @xmath908 .
@xmath1352 restrict to bases of @xmath1353 as @xmath8-vector space and @xmath1354/i_{{{\bf s}}}^{k+1}$]-module respectively .
using the isomorphism @xmath1297 with @xmath1203 , the basic compatibility theorem [ blinvth ] gives the gluing statement ( 1 ) . to prove ( 4 ) , using the @xmath240-linearity , it is enough to prove the given @xmath46 restrict to basis as @xmath1355/(x_1,\dots , x_n)^{k+1}$]-module . by corollary [ mutcor ] ,
the central fibre of @xmath1356 is the torus @xmath22 .
if @xmath1357 , the only broken lines with @xmath1358 and @xmath1359 are straight lines .
thus these @xmath1360 restrict to the basis of characters on the central fibre .
now the result follows from the nilpotent nakayama lemma ( see @xcite , pg .
58 , theorem 8.4 ) .
\(2 ) follows immediately from ( 1 ) and ( 4 ) . for ( 3 ) ,
it is enough to prove the equality in @xmath1361 for each @xmath460 .
the argument is the same as the proof of the multiplication rule in @xcite , theorem 2.38 , which , as it is very short , we recall for the reader s convenience .
we work with the scattering diagram @xmath378 modulo @xmath1362 , which has only finitely many walls with non - trivial attached function . expressing @xmath1363 in the basis @xmath1364 of ( 4 ) , we examine the coefficient of @xmath46 . we choose a general point @xmath1365 very close to @xmath908 , so that @xmath1366 lie in the closure of the same connected component of @xmath1367 ( where @xmath1368 denotes the finitely many walls non - trivial modulo @xmath1362 ) . by definition of @xmath1369 , @xmath1370 now observe first that there is only one broken line @xmath341 with endpoint @xmath1350 and @xmath1371 : this is the broken line whose image is @xmath1372 .
indeed , the final segment of such a @xmath341 is on this ray , and this ray meets no walls , other than walls containing @xmath908 , so the broken line can not bend .
thus the coefficient of @xmath1373 can be read off by looking at the coefficient of the monomial @xmath1374 on the right - hand side of the above equation .
this gives the desired formula to order @xmath460 .
the finiteness argument of definition - lemma [ structuredef ] then shows that for any given @xmath908 , @xmath1350 chosen sufficiently close to @xmath908 will work for all @xmath460 . by the proposition
, each @xmath1375 has a unique expression as a convergent formal sum @xmath1376 , with coefficients @xmath1377 .
this immediately implies : [ hatinfun ] notation as in proposition [ thetabasislemma ] . there is a unique inclusion @xmath1378 } { \bbbk}[n ] \hookrightarrow { \operatorname{hom}}_{\operatorname{sets}}({{\mathcal a}^{{\vee}}_{{\operatorname{prin}}}(\bz^t)}= \tm^\circ_{{{\bf s}}},{\bbbk})\ ] ] given by @xmath1379 we have @xmath1380 for all @xmath1381 .
[ defs ] for @xmath1382 , write @xmath1383 on the torus chart @xmath1384 of @xmath56 corresponding to a seed @xmath10 .
we also have a formal expansion @xmath1385 as @xmath1386 for some @xmath1387 . set @xmath1388
if @xmath1324 is the monoid generated by @xmath1389 , it is easy to check from the construction of theta functions that @xmath1390 note that @xmath1391 is constructed by gluing together various @xmath1392/(x_1,\ldots , x_n)^{k+1}$ ] via isomorphisms , and hence @xmath1393/(x_1,\ldots , x_n)^{k+1}$ ] . thus by proposition [ thetabasislemma ] , ( 4 ) , a collection of the @xmath46 yield a basis for regular functions on @xmath1394/(x_1,\ldots , x_n)^{k+1}$ ] with the property that this basis restricts to a monomial basis on the underlying reduced space .
it remains a mystery about theta functions in general whether they satisfy some other interesting characterizing properties , such as the heat equation satisfied by theta functions on abelian varieties . [ defs ] we need to make algorithmically explicit the map @xmath1395 of proposition [ hatinfun ] when restricted to elements of @xmath139 .
let @xmath1281 be a universal laurent polynomial .
fix a seed @xmath74 .
let @xmath1396 be the monoid generated by the @xmath1397 , so that monomials in functions attached to walls of @xmath806 all lie in @xmath1398 $ ] .
let @xmath1399 be the corresponding torus in the atlas for @xmath56 .
let @xmath1400 be the laurent polynomial @xmath1401\ ] ] representing @xmath41 in this chart , and let @xmath1402 .
we inductively define power series @xmath1403 , @xmath1404 , whose non - zero monomial terms lie in @xmath1405 .
let @xmath827 be a general basepoint in @xmath693 .
set @xmath1406 .
having defined @xmath1403 we set @xmath1407 set @xmath1408 .
note @xmath1409 where the higher order terms are monomials in @xmath1410 .
thus each term of @xmath1411 is in @xmath1412 as required .
we thus have for all @xmath460 @xmath1413 where the higher order terms are monomials in @xmath1412 .
we set @xmath1414 this indexes the @xmath1415 sum of theta functions , and @xmath1416 .
note the union is disjoint by construction .
associated to each @xmath1417 we have a coefficient @xmath1418 .
now defining @xmath1419 by @xmath1420 gives the @xmath1395 of proposition [ hatinfun ] .
the main point of the following theorem is that on @xmath1421 , @xmath1395 is independent of the seed @xmath10 .
[ ffgth ] there is a unique function @xmath1422 with all the following properties : 1 .
@xmath59 is compatible with the @xmath1423$]-module structure on @xmath139 and the @xmath240-translation action on @xmath1424 in the sense that @xmath1425 for all @xmath1281 , @xmath1381 , @xmath1426 .
2 . for each choice of seed @xmath10
, the formal sum @xmath1271 converges to @xmath41 in @xmath1427}{\bbbk}[n]$ ] .
3 . if @xmath1428 then @xmath1429 unless @xmath1430 , and @xmath1431 and the coefficients @xmath1432 are the coefficients for the expansion of viewed as an element of @xmath1353 in the basis of theta functions from proposition [ thetabasislemma ] .
4 . for any seed @xmath609 obtained via mutations from @xmath10 , @xmath59 is the composition of the inclusions @xmath1433 } { \bbbk}[n ] \subset { \operatorname{hom}}_{\operatorname{sets}}({{\mathcal a}^{{\vee}}_{{\operatorname{prin}}}(\bz^t)}= \tm^\circ_{{{\bf s}}'},{\bbbk})\ ] ] given by and proposition [ hatinfun ] .
this sends a cluster monomial @xmath1434 to the delta function @xmath1435 for its @xmath41-vector @xmath1436 . in the notation of definition [ defs ] , @xmath1437 for any seed @xmath609 .
in particular the sets @xmath1438 of definition [ defs ] are independent of the seed , depending only on @xmath41 .
it is easy to see from proposition [ thetabasislemma ] that @xmath1395 is the unique function which satisfies conditions ( 1 - 3 ) of the theorem for the given seed @xmath10 .
moreover it satisfies ( 4 ) for @xmath1439 .
thus it is enough to show that @xmath1395 is independent of the choice of seed .
the basic idea is that @xmath1395 expresses @xmath41 as a sum of theta functions .
as the theta functions are linearly independent , the expression is unique .
but as the sums can be infinite , we make the argument in the appropriate formal neighborhood . for a seed @xmath1440
we write @xmath1061 for the fan in @xmath1441 with rays spanned by the @xmath1442 .
we write @xmath1443 for the fan in @xmath1444 with rays spanned by the @xmath1445 .
clearly for the invariance it is enough to consider two adjacent seeds , say @xmath1440 and @xmath1446 obtained , without loss of generality , by mutation of the first basis vector @xmath1447 .
we consider the union of the two tori @xmath1448 in the atlas for @xmath56 , glued by the mutation @xmath1449 , which we recall is given by @xmath1450 @xmath1451 , see @xcite , ( 2.6 ) .
we will partially compactify this union by gluing the toric varieties @xmath1452 writing @xmath1453 under this gluing .
note this union of toric varieties is _ not _ part of the atlas for either @xmath124 or @xmath1454 ( for either of these , the fans determining the atlases for the toric compactifications are related by geometric tropicalisation of the birational mutation , but here @xmath1455 , and thus @xmath1456 ) .
note @xmath1457 for @xmath1458 , while @xmath1459_+ d_jf_j$ ] , ( see e.g. , @xcite , ( 2.3 ) ) .
thus the two cones @xmath1460 share a codimension one face and form a fan , @xmath1461 .
let @xmath1462 . by construction the rational map @xmath1463 is regular , and the same holds for the seed @xmath609 .
observe that @xmath1449 commutes with the second projection @xmath1464 . from this
it follows that @xmath1465 is regular .
note the toric boundary @xmath1466 has a unique complete one - dimensional stratum @xmath1467 and two zero strata @xmath1468 , @xmath1469 , whose complements in the @xmath1467 we write as @xmath1470 respectively .
we write e.g. @xmath1471 for the @xmath1415 order neighborhood of this curve , and e.g. @xmath1472 for the scheme - theoretic inverse image @xmath1473 .
finally , let @xmath1474 we will show theta functions give a basis of functions on these formal neighborhoods . to make the computation transparent we introduce coordinates .
we let @xmath1475 , @xmath1476 , observing that @xmath1477 for all @xmath1381 . in particular
@xmath1478 , @xmath1479 .
since there is a map of fans from @xmath1461 to the fan defining @xmath1480 by dividing out by the subspace spanned by @xmath1481 , there is a map @xmath1482 .
we can pull back @xmath1483 to @xmath15 , getting a line bundle with monomial sections @xmath1484 pulled back from @xmath1480 with @xmath1485 in the above notation .
the open subset of @xmath1 where @xmath1486 is given explicitly up to codimension two by the hypersurface @xmath1487 where @xmath1488 and @xmath1489 .
note the points @xmath1490 lie in the chambers of @xmath792 corresponding to @xmath10 and @xmath609 respectively , and thus by proposition [ oneblprop ] these points determine theta functions in @xmath139 , which are of course the corresponding cluster monomials @xmath1491 .
we have the exactly analogous description for the open subset @xmath1492 .
next observe that all but one of the functions attached to walls in @xmath378 is trivial modulo the ideal @xmath1493 . indeed , the unique non - trivial wall is @xmath1494 .
it follows from theorem [ kslemma2 ] that modulo @xmath1495 the scattering diagram @xmath378 has only finitely many non - trivial walls , and @xmath942 is regular on @xmath1472 , for @xmath827 a basepoint in the distinguished chamber @xmath693 , so long as @xmath1496 , @xmath1287 the projection .
let @xmath1497 , @xmath1498 .
noting @xmath1499 , we can set @xmath1500 observe @xmath1501 is the subscheme of @xmath1 defined by the ideal @xmath1495 in the open subset @xmath1502 .
note that the open subset of @xmath1 defined by @xmath1503 is the union of the two tori @xmath1448 .
[ basislem ] the following hold : 1 .
the collection @xmath942 , @xmath1504 , @xmath1505 forms a @xmath8-basis of the vector space @xmath1506 .
2 . the collection @xmath1507 , @xmath1508 , forms a basis of @xmath1506 as an @xmath1509-module .
3 . the collection @xmath942 , @xmath1510 , forms a @xmath8-basis of @xmath1511 .
\(2 ) implies ( 1 ) using the @xmath240-linearity of the scattering diagram and multiplication rule with respect to the @xmath240-translation .
similarly , ( 2 ) implies ( 3 ) by inverting @xmath777 . for the second claim , by lemma 2.30 of @xcite , we need only prove the statement for @xmath1512 . to prove linear independence it is enough to show linear independence modulo @xmath1513 for all @xmath51 . for this ,
again by lemma 2.30 of @xcite , it is enough to check just over the fibre @xmath1514 .
this is the torus @xmath1076 and the theta functions restrict to the basis of characters .
so it remains only to show the given theta functions generate modulo @xmath269 .
here we use the explicit description of the open subset of @xmath1 where @xmath1486 above .
this is an affine variety , and the ring of functions is clearly generated by the @xmath1515 as a @xmath1516$]-algebra . on the other hand , by the explicit description of @xmath378
modulo the ideal @xmath269 , for @xmath1517 , @xmath1518 this shows theta functions generate @xmath1519 as an @xmath1520-module , hence the result .
of course there is an analogous claim for @xmath609 .
now we can prove @xmath1521 for @xmath1522 . by the @xmath240-translation action on @xmath1424 ( and the corresponding @xmath240-linearity of the scattering diagrams ) , to prove the equality , we are free to multiply @xmath41 by a monomial from the base of @xmath1523 . multiplying by a monomial in the @xmath1115 , @xmath1524 , we can then assume @xmath41 is a regular function on the open subset of @xmath1 where @xmath1525 .
now in the notation of definition [ defs ] , @xmath1526 for @xmath1527 or @xmath1528 .
it follows now from the fact that @xmath378 is finite modulo @xmath1495 for any @xmath460 that each @xmath1529 for @xmath1530 is a finite laurent polynomial modulo @xmath1495 . here
@xmath1531 are basepoints in the chambers indexed by @xmath10 and @xmath609 respectively .
[ claimtwo ] modulo @xmath1495 , the sums @xmath1532 , @xmath1533 are finite , and coincide with @xmath41 in the charts indexed by @xmath10 and @xmath609 respectively .
by symmetry it s enough to treat @xmath10 .
we can multiply both sides by a power of @xmath777 , and so may assume @xmath41 is regular on @xmath1486 , and @xmath1534 for each @xmath1535 .
note @xmath1536 , thus by construction modulo @xmath1537 for any @xmath51 the sum @xmath1538 is finite and equal to @xmath41 . by claim [ basislem ] ,
( 1 ) , we have a ( finite ) expression modulo @xmath1495 @xmath1539 thus , for fixed @xmath460 and arbitrary @xmath1540 we have modulo @xmath1541 , @xmath1542 by the linear independence these expressions are the same , for all @xmath51 , thus the expressions are the same modulo @xmath1495 .
note that by theorem [ blinvth ] , for @xmath1543 , @xmath942 and @xmath1544 induce the same regular function @xmath1545 on @xmath1501 .
thus we have by claim [ claimtwo ] that @xmath1546 now by ( 3 ) of claim [ basislem ] ( varying @xmath460 ) the coefficients in the sums are the same .
the theorem implies that the theta functions are a topological basis for a natural topological @xmath8-algebra completion of @xmath139 : for @xmath1547 , let @xmath1548 denote precomposition by the action of translation by @xmath510 on @xmath1424 .
let @xmath1549 be the vector subspace of functions , @xmath34 , such that for each seed @xmath10 , there exists @xmath1547 for which the restriction of @xmath1550 to @xmath1551 has finite support for all @xmath1552 .
then we have @xmath1553 } { \bbbk}[n ] \subset { \operatorname{hom}}_{\operatorname{sets}}({{\mathcal a}^{{\vee}}_{{\operatorname{prin}}}(\bz^t)},{\bbbk}).\ ] ] @xmath1554 is a complete topological vector space under the weakest topology so that each inclusion @xmath1555 } { \bbbk}[n]$ ] is continuous .
let @xmath1556 be the delta function associated to @xmath1557 .
the @xmath46 are a topological basis for @xmath1554 .
there is a unique topological @xmath8-algebra structure on @xmath1554 such that @xmath1558 with structure constants given by definition [ structuredef ] .
in this section , we prove one of the main theorems of the paper , theorem [ mainth ] .
this is done in two steps .
first , it follows from the results of the previous section and properties of theta functions in the @xmath56 case .
this is easiest since the scattering diagram technology works best for @xmath56 .
second , we descend to the @xmath19 or @xmath78 case and the @xmath27 case , with the @xmath19-type varieties appearing as fibres of @xmath1559 and the @xmath27 variety as a quotient of @xmath56 by @xmath1076 , using the @xmath56 case to deduce the result for these other cases .
recall from definition [ fgchamberdef ] that @xmath792 is the collection of chambers forming the cluster complex .
abstractly , by lemma [ fgcclem ] , this can be viewed as giving a collection of chambers @xmath118 in @xmath1560 .
[ posprop ] choose @xmath1561 .
if for some generic basepoint @xmath1562 there are only finitely many broken lines @xmath341 with @xmath1563 and @xmath1564 , then the same is true for any generic @xmath1565 .
in particular , @xmath1566 $ ] is a positive universal laurent polynomial . by positivity of the scattering diagram , theorem [ scatdiagpositive ] , for any basepoint @xmath827
, @xmath851 has only non - negative coefficients ( though it may have infinitely many terms ) . also , we know that for basepoints in different chambers , the @xmath851 are related by wall - crossings by theorem [ blinvth ] , which in turn are identified with the mutations of tori in the atlas for @xmath56 .
so the @xmath851 determine a universal positive laurent polynomial if and only if we have finiteness of broken lines ending at any @xmath827 in any chamber of @xmath792 .
if we vary @xmath827 in the chamber , @xmath851 does not change .
so it s enough to check that if @xmath851 is a polynomial , the same is true of @xmath1567 for @xmath870 in an adjacent chamber @xmath953 to @xmath946 close to the wall @xmath1568 .
we can work in some seed .
let the wall be contained in @xmath357 , @xmath1569 , with @xmath1570 , and denote the wall - crossing automorphism from @xmath827 to @xmath870 as @xmath994 . recall that @xmath1571 , where for walls between cluster chambers , @xmath34 is some positive laurent polynomial ( in fact it has the form @xmath1572 for some @xmath1573 ) .
monomials @xmath1574 are then divided into three groups , according to the sign of @xmath1575 .
this sign is preserved by @xmath994 , as @xmath450 takes the same value on each exponent of a monomial term in @xmath1576 as @xmath450 takes on @xmath860 .
monomials with @xmath1577 are then invariant under @xmath994 , so these terms in @xmath1567 coincide with those in @xmath851 .
hence there are only a finite number of such terms in @xmath1567 .
the sum of terms of the form @xmath1578 in @xmath851 with @xmath1579 , which we know form a laurent polynomial , is , by the explicit formula for @xmath994 , sent to a polynomial .
so it only remains to show that there are only finitely many terms @xmath1578 in @xmath1567 with @xmath1580 .
suppose the contrary is true .
the direction vector of each broken line contributing to such terms at @xmath870 is towards the wall @xmath1568 , and so we can extend the final segment of any such broken line to obtain a broken line terminating at some point @xmath1581 ( depending on @xmath860 ) in the same chamber as @xmath827 . as there are no cancellations because of the positivity of all coefficients and @xmath851 does not depend on the location of @xmath827 inside the chamber by theorem [ blinvth ] , we see that @xmath851 has an infinite number of terms , a contradiction .
[ thetadef ] let @xmath1582 be the collection of @xmath826 such that for some ( or equivalently , by proposition [ posprop ] , any ) generic @xmath1583 there are only finitely many broken lines @xmath341 with @xmath1563 , @xmath1564 . [ icdef ]
we call a subset @xmath1584 _ intrinsically closed under addition _ if @xmath1585 and @xmath1586 implies @xmath1587 .
[ iclemma ] let @xmath1588 be intrinsically closed under addition .
the image of @xmath1589 in @xmath1590 ( under the identification @xmath1591 induced by the seed @xmath10 ) is closed under addition for any seed @xmath10 . if for some seed @xmath1592 is a toric monoid ( i.e. , the integral points of a convex rational polyhedral cone ) , then this holds for any seed .
choose a seed @xmath10 .
then straight lines in definition - lemma [ structuredef ] show @xmath1593 .
this gives closure under addition .
now suppose @xmath1592 is a toric monoid , generating the convex rational polyhedral cone @xmath1594 .
then @xmath1595 is a rational polyhedral cone with integral points @xmath1596 . as this set of integral points
is closed under addition , @xmath1597 is convex , and so its integral points are a toric monoid .
recall from the introduction the definition of global monomial ( definition [ globalmonomialdef ] ) .
[ goodalgebra ] let @xmath1598 be the set of integral points in chambers of the cluster complex
. then 1 .
2 . for @xmath1600 @xmath1601 is a finite sum ( i.e. , @xmath1602 for all but finitely many @xmath51 ) with non - negative integer coefficients . if @xmath1603 , then @xmath1604 .
the set @xmath61 is intrinsically closed under addition . for any seed @xmath10 , the image of @xmath1605 is a saturated monoid .
the structure constants @xmath1606 of definition [ structuredef ] make the @xmath8-vector space with basis indexed by @xmath61 , @xmath1607 into an associative commutative @xmath1423$]-algebra
. there are canonical inclusions of @xmath1423$]-algebras @xmath1608 } { \bbbk}[n].\end{aligned}\ ] ] under the first inclusion a cluster monomial @xmath1609 is identified with @xmath1610 for @xmath1611 its @xmath41-vector . under the second inclusion
each @xmath46 is identified with a universal positive laurent polynomial .
\(1 ) is immediate from corollary [ monocor ] . for ( 2 ) ,
first note that the coefficients @xmath1612 are non - negative by definition - lemma [ structuredef ] .
suppose @xmath1600 .
take a generic basepoint @xmath827 in some cluster chamber .
then @xmath1613 is the product of two laurent polynomials , thus a laurent polynomial .
it is equal to @xmath1614 by ( 3 ) of proposition [ thetabasislemma ] , and hence this sum must be finite , as it involves a positive linear combination of series with positive coefficients .
further , each @xmath1615 appearing with @xmath1616 must be a laurent polynomial for the same reason .
thus @xmath1604 by definition [ thetadef ] . ( 2 ) then immediately implies @xmath61 is intrinsically closed under addition . for ( 4 ) ,
note each @xmath1617 , @xmath1618 is a universal positive laurent polynomial by proposition [ posprop ] . for @xmath1619
, @xmath1620 is the corresponding cluster monomial by ( 4 ) of theorem [ ffgth ] .
the inclusions of algebras , and the associativity of the multiplication on @xmath86 follow from proposition [ thetabasislemma ] .
finally we complete the proof of ( 3 ) by checking that @xmath61 is saturated .
assume @xmath1621 for some integer @xmath1622 .
take @xmath827 to be a generic basepoint in some cluster chamber .
we show that the set of final monomials @xmath1623 for broken lines @xmath341 with @xmath1624 , @xmath1564 is finite . by assumption
( and the positivity of the scattering diagram ) , this holds with @xmath908 replaced by @xmath1625 .
so it is enough to show @xmath1626 implies @xmath1627 .
indeed , it is easy to see that for every broken line @xmath341 for @xmath908 ending at @xmath827 , there is a broken line @xmath918 for @xmath1625 with the same underlying path , such that for every domain of linearity @xmath11 of @xmath341 , the exponents @xmath1628 and @xmath1629 of the monomial decorations of @xmath11 for @xmath341 and @xmath918 respectively satisfy @xmath1630 .
this completes the proof of ( 3 ) , hence the theorem .
the above theorem immediately implies : theorem [ mainth ] is true for @xmath1631 .
the following shows our theta functions are well - behaved with respect to the canonical torus action on @xmath56 .
[ thetaeigenfunctions ] let @xmath1632 .
then @xmath1633 is an eigenfunction for the natural @xmath1634 action on @xmath56 ( see proposition [ ldpprop ] , ( 2 ) ) , with weight @xmath1635 given by the canonical map @xmath1636 ( the map being dual to the inclusion @xmath1637 ) . in particular @xmath46 is an eigenfunction for the subtorus @xmath1638 with weight @xmath1635 where @xmath1639 is given by @xmath1640 . pick a seed @xmath10 , giving an identification @xmath1641 with @xmath1150 .
pick also a general basepoint @xmath1037 .
we need to show that for any broken line @xmath341 in @xmath1642 for @xmath908 with endpoint @xmath827 , @xmath1643 is a semi - invariant for the @xmath1634 action with weight @xmath1635 .
the @xmath1634 action on the seed torus @xmath1644 is given on cocharacters by the natural inclusion @xmath1637 . by definition of @xmath1645
we have @xmath1646 , @xmath76 , so every monomial appearing in any function @xmath383 for @xmath1647 is in the kernel of @xmath153 .
the result for @xmath1634 follows .
the statement for @xmath1076 now follows from the definitions . with more work
, we will define the middle cluster algebra for @xmath1648 or @xmath27 .
we now show how the various structures we have used to understand @xmath1170 induce similar structures for @xmath78 and @xmath27 .
by @xcite , 3 , each seed @xmath10 ( in the @xmath27 , @xmath19 and @xmath56 cases ) gives a _ toric model _ for @xmath15 .
the seed specifies the data of a fan @xmath1649 , consisting only of rays ( so the boundary @xmath1650 of the associated toric variety is a disjoint union of tori ) .
the seed also then specifies a blowup @xmath1651 with codimension two center , the disjoint union of divisors @xmath1652 in each of the disjoint irreducible components @xmath1653 .
if @xmath1654 is the proper transform of @xmath1655 , then there is a birational map @xmath1656 .
this map is an isomorphism outside of codimension two between @xmath1657 and the upper bound ( see @xcite , remark 3.13 , @xcite , def .
1.1 ) @xmath1658 , which we recall is the union of @xmath6 with @xmath1659 for the _ adjacent seeds _ , @xmath1660 . in the case @xmath1661 or @xmath78 for very general @xmath188 ,
the inclusion @xmath1658 is an isomorphism outside codimension two .
we have @xmath1662 from these toric models it is easy to determine the global monomials : [ gmlemma ] notation as immediately above . for @xmath1663 , the character @xmath36 on the torus
@xmath1664 is a global monomial if and only if @xmath36 is regular on the toric variety @xmath1665 , which holds if and only if @xmath1666 for the primitive generator @xmath510 of each ray in the fan @xmath1649 . for @xmath19-type cluster varieties
a global monomial is the same as a cluster monomial , i.e. , a monomial in the variables of a single cluster , where the non - frozen variables have non - negative exponent .
we have a surjection @xmath1667 by construction of @xmath0 , and thus a monomial @xmath36 is regular on @xmath1665 if and only if its pull - back to @xmath0 is regular . certainly such a function is also regular on @xmath1668 .
conversely , suppose @xmath36 is not regular on @xmath1665 .
then it has a pole on some toric boundary divisor @xmath1669 .
now the support of @xmath1670 is given by @xmath1671 ( resp .
@xmath1672 ) in the @xmath19 ( resp .
@xmath27 ) case , as explained in @xcite , 3.2 . in particular for @xmath1673 , @xmath1674 is non - empty , in the @xmath19 case because of the assumption that @xmath1675 for @xmath263 stated in appendix [ ldsec ] .
as @xmath36 has no zeros on the big torus , the divisor of zeros of @xmath36 will not contain the center @xmath1670 .
it follows that @xmath36 has a pole along the exceptional divisor @xmath1676 over @xmath1674 . since @xmath1677
, @xmath36 is not regular on @xmath1657 .
thus we conclude that @xmath36 is regular on @xmath1678 if and only if @xmath36 is regular on @xmath1665 . of course , @xmath36 is regular on @xmath1665 if and only if @xmath1679 for all primitive generators @xmath510 of rays of @xmath1649 .
now the rational map @xmath1680 to the upper bound is an isomorphism outside codimension two , so the two varieties have the same global functions . in the @xmath27 ( or @xmath56 ) case ,
the inclusion @xmath1658 is an isomorphism outside codimension two as well .
this gives the theorem for @xmath27 or @xmath56 , and the forward direction for @xmath78 .
the reverse direction for @xmath78 follows from the laurent phenomenon .
indeed , the final statement of the lemma simply describes the monomials regular on @xmath1681 , and a monomial of the given form is the same as a cluster monomial and these are global regular functions by the laurent phenomenon .
recall from proposition [ ldpprop ] , ( 4 ) , the canonical maps @xmath1682 and @xmath1683 whose tropicalizations are @xmath1684 note @xmath808 identifies @xmath112 with the quotient of @xmath1685 by the natural @xmath240-action . since @xmath1686 identifies @xmath1687 with the fibre over @xmath1688 of @xmath1689 , @xmath1690 identifies @xmath1691 with @xmath1692 , where @xmath1693 is the weight map given by @xmath1694 .
[ fgccthg1 ] let @xmath5 be a cluster variety .
define @xmath1695 to be the set of @xmath41-vectors ( see definition [ fgccthg0 ] ) for global monomials which are characters on the seed torus @xmath1664 , and @xmath1696 to be the union of all @xmath1697 .
[ fgccthg2 ] 1 .
for @xmath15 of @xmath19-type @xmath1697 is the set of integral points of the cone @xmath1698 in the fock - goncharov cluster complex corresponding to the seed @xmath10 .
2 . in any case
@xmath1699 is the set of integral points of a rational convex cone @xmath693 , and the relative interiors of @xmath693 as @xmath10 varies are disjoint .
the @xmath41-vector @xmath1700 depends only on the function @xmath34 ( i.e. , if @xmath34 restricts to a character on two different seed tori , the @xmath41-vectors they determine are the same ) .
3 . for @xmath1701 ,
the global monomial @xmath1545 on @xmath56 is invariant under the @xmath1076 action and thus gives a global function on @xmath1702 .
this is a global monomial and all global monomials on @xmath27 occur this way , and @xmath1703 .
\(1 ) in the @xmath19 case , @xmath693 is the fock - goncharov cone by lemma [ fgcclem ] and lemma [ gmlemma ] .
these cones form a fan by theorem [ fgccth ] , and the fan statement implies that @xmath1704 depends only on @xmath34 .
the @xmath19 case of ( 2 ) immediately follows also from the discussion in [ cgvectorsec ] .
the @xmath27 case follows from the @xmath19-case ( applied to @xmath56 ) by recalling that there is a map @xmath1705 making @xmath56 into @xmath1076-torsor over @xmath27 , see proposition [ ldpprop ] , ( 2 ) .
this map is defined on monomials by @xmath1706 . pulling back a monomial for @xmath27 under @xmath1707
gives a @xmath1076-invariant global monomial for @xmath56 .
thus there is an inclusion @xmath1708 by proposition [ thetaeigenfunctions ] .
conversely , if @xmath1709 and @xmath1710 for a global monomial @xmath34 on @xmath1170 , then there is some seed @xmath787 where @xmath34 is represented by a monomial @xmath36 on @xmath1384 . because @xmath1387 lies in @xmath1692
it is of the form @xmath1711 for some @xmath1547 . by lemma [ gmlemma ] ,
@xmath860 is non - negative on the rays @xmath1712 of @xmath1713 , hence @xmath510 is non - negative on the rays @xmath1714 of @xmath1715 .
hence @xmath1283 defines a global monomial on @xmath987 .
thus @xmath1716 .
furthermore , one then sees that the fock - goncharov cones for @xmath56 yield the cones for @xmath27 by intersecting with @xmath1692 .
this gives the remaining statements of ( 2 ) in the @xmath27 case , as well as ( 3 ) .
[ blxa ] _ the @xmath27 case . _ note that every function @xmath383 attached to a wall in @xmath801 is a power series in @xmath1717 for some @xmath510 , thus @xmath153 is zero on all exponents appearing in these functions .
thus broken lines with both @xmath1718 and initial infinite segment lying in @xmath1692 remain in @xmath1692 .
in particular @xmath1719 , and all their monomial decorations , e.g. , @xmath858 , are in @xmath1692 .
we define these to be the broken lines in @xmath1720 .
induces a collection of walls with attached functions , @xmath1721 , living in @xmath1722 , just by intersecting each wall with @xmath1692 and taking the same scattering function .
this is a consistent scattering diagram , and we are getting exactly the broken lines for this diagram .
we will not use this diagram , as we can get whatever we need from @xmath1723 . ] _ the @xmath19 case .
_ we define broken lines in @xmath781 to be images of broken lines in @xmath1560 under @xmath808 ( applying the derivative @xmath1724 to the decorating monomials ) .
we define 1 .
@xmath1726 where the superscript denotes the invariant part under the group action .
[ xcorollary ] theorem [ mainth ] holds for @xmath1727 .
this follows immediately from the @xmath1170 case by taking @xmath1076-invariants . moving on to the @xmath19 case , the following
is easily checked : [ thacasedef ] 1 .
define @xmath1728 noting that @xmath1729 is invariant under @xmath240-translation , we have @xmath1730 .
furthermore , any choice of section @xmath1731 of @xmath808 induces a bijection @xmath1732 .
2 . define @xmath1733 } { \bbbk}$ ] , where @xmath1423 \twoheadrightarrow { \bbbk}$ ] is given by @xmath1734 .
given a choice of @xmath1302 , the collection @xmath1735 gives a @xmath1423$]-module basis for @xmath137 and thus a @xmath8-vector space basis for @xmath79 .
for @xmath1736 the basis is independent of the choice of @xmath1302 , while for @xmath79 it is independent up to scaling each basis vector ( i.e. , the decomposition of the vector space @xmath1736 into one dimensional subspaces is canonical ) .
the variety @xmath78 is a space @xmath1737 with the tori glued by birational maps which vary with @xmath188 .
it is then not so clear how to dualize these birational maps to obtain @xmath1738 as it is not obvious how to deal with these parameters .
however , the tropicalisations of these birational maps are all the same ( independent of @xmath188 ) and thus the tropical sets @xmath1739 should all be canonically identified with @xmath1740 .
so we just take : @xmath1741 .
[ mainthax ] for @xmath1742 the following modified statements of theorem [ mainth ] hold . 1 .
there is a map @xmath1743 depending on a choice of a section @xmath1744 .
this function is given by the formula @xmath1745 if this sum is finite ; otherwise we take @xmath1746 .
this sum is finite whenever @xmath1747 .
2 . there is a canonically defined subset @xmath57 given by @xmath1748 such that the restriction of the structure constants give the vector subspace @xmath60 with basis indexed by @xmath61 the structure of an associative commutative @xmath8-algebra .
3 . @xmath1749 , i.e. , @xmath61 contains the @xmath41-vector of each global monomial .
4 . for the lattice structure on @xmath44 determined by any choice of seed , @xmath57
is closed under addition .
furthermore @xmath61 is saturated .
there is a @xmath8-algebra map @xmath67 which sends @xmath121 for @xmath1750 to a multiple of the corresponding global monomial .
there is no analogue of ( 6 ) of theorem [ mainth ] because the coefficients of the @xmath1617 will generally not be integers .
@xmath70 is injective for very general @xmath188 , and for all @xmath188 if the vectors @xmath595 , @xmath76 , lie in a strictly convex cone .
when @xmath70 is injective we have canonical inclusions @xmath77 taking @xmath1751 gives theorem [ mainth ] for the @xmath73 case . for ( 1 ) , note that for @xmath1752 , we have @xmath1753 and on @xmath56 @xmath1754 using @xmath1755 .
note that the sums are finite because @xmath1756 .
restricting to @xmath78 gives the formula of ( 1 ) .
the remaining statements follow easily from the definitions except for the injectivity of ( 7 ) . to see this , fix a seed @xmath10 , which gives the second projection @xmath1757 .
choose the section @xmath1302 of @xmath808 to be @xmath1758 .
note the collection of @xmath121 , @xmath1759 are a @xmath1423$]-basis for @xmath137 . by the choice of @xmath1302 , the @xmath121
restrict to the basis of monomials on the central fibre @xmath22 of @xmath1760 .
it follows that for any finite subset @xmath1761 there is a zariski open set @xmath1762 such that @xmath121 , @xmath1763 restrict to linearly independent elements of @xmath1764 , @xmath1765 .
this gives the injectivity of @xmath70 for very general @xmath188 .
now suppose the @xmath1766 , @xmath76 span a strictly convex cone .
we can then pick an @xmath1767 such that @xmath1768 for all @xmath210 . now pick @xmath1769 such that @xmath1770 for @xmath1771 , and set @xmath1772 , notation as in ( 2 ) of proposition [ ldpprop ] . by construction the second projection @xmath1773 lies in the interior of the orthant generated by the dual basis @xmath1108 .
take the one - parameter subgroup @xmath1774 .
now , by proposition [ ldpprop ] , the map @xmath1775 is @xmath1634-equivariant , where the action on @xmath1136 is given by the map of cocharacters @xmath1776 .
thus @xmath582 has a one - dimensional orbit whose closure contains @xmath1777 .
so @xmath81 is in the closure of the orbit @xmath1778 for all @xmath1779 . in particular for all @xmath367 and all @xmath1589
there is some @xmath1780 with @xmath1781 .
now from the @xmath1782-equivariance of the construction , proposition [ thetaeigenfunctions ] , the linear independence holds for all @xmath188 .
changing @xmath1302 will change the @xmath1423$]-basis for @xmath137 , multiplying each @xmath121 by some character @xmath1283 , @xmath1381 .
the restrictions to @xmath79 are then multiplied by the values @xmath1783 .
theorem [ mainth ] for @xmath73 now follows from setting @xmath1751 , where @xmath1784 for all @xmath510 .
it is natural to wonder : does the equality @xmath1785 always hold ?
our guess is no , but we do not know a counterexample .
certainly @xmath1786 in general , for this implies @xmath1787 , which is defined to be @xmath1788 , coincides with @xmath1789 , while we know that in general @xmath27 has many fewer global functions , see @xcite , 7 .
so we look for conditions that guarantee @xmath1790 , and @xmath1791 .
we turn to this in the next section .
[ examples2 ] in the cases of example [ bcexample ] , the convex hull of the union of the cones of @xmath118 in @xmath1792 is all of @xmath1792 .
indeed , the first three quadrants already are part of the cluster complex .
it then follows from the fact that @xmath61 is closed under addition and is saturated that @xmath1793 . in the case of example
[ standardexample ] , we know that @xmath1794 it then follows again from the fact that @xmath61 is closed under addition that either @xmath1795 or @xmath1793 .
we believe , partly based on calculations in @xcite , 7.1 , that in fact the latter holds .
we show the analogue of proposition [ thetaeigenfunctions ] for the @xmath19 variety : [ thetaeigenfunctions2 ] if @xmath1796 , then @xmath1797 is an eigenfunction for the natural @xmath1798 action on @xmath19 .
this is essentially the same as the proof of proposition [ thetaeigenfunctions ] , noting that the monomials @xmath1799 are invariant under the @xmath1798 action , as @xmath1800 by definition of @xmath1801 .
we end this section by showing that linear independence of cluster monomials follows easily from our techniques .
this was pointed out to us by gregory muller . in the skew - symmetric case ,
this was proved in @xcite .
[ properlaurentproperty ] for any @xmath19 cluster variety , there are no linear relations between cluster monomials and theta functions in @xmath1802 . more precisely , if there is a linear relation @xmath1803 in @xmath33 , then @xmath1804 for all @xmath1805 .
in particular the cluster monomials in @xmath1806 are linearly independent .
suppose given such a relation .
we choose a seed @xmath10 and a generic base point @xmath1807 .
the seed gives an identification @xmath1808 .
we first show that if @xmath1805 with @xmath1809 , then @xmath1810 satisfies the _ proper laurent property _ , i.e. , every monomial @xmath1811 appearing in @xmath1810 has @xmath1812 for some @xmath210 .
indeed , fix a section @xmath1813 as in definition - lemma [ thacasedef ] . as restriction to @xmath1814
gives a bijection between the cluster variables for @xmath56 and the cluster variables for @xmath19 , between the theta functions @xmath1360 , @xmath1815 and the theta functions for @xmath19 , and between the corresponding local expressions @xmath1810 , it is enough to prove the claim in the @xmath56 case .
this follows immediately from the definition of broken line .
indeed , if @xmath341 is a broken line ending at @xmath827 and @xmath1816 with @xmath1817 for all @xmath210 , then @xmath341 must be wholly contained in @xmath693 .
but the unbounded direction of @xmath341 is parallel to @xmath1818 , so it follows that @xmath1819 .
we then have the relation @xmath1820\ ] ] which we rearrange as @xmath1821 the collection of @xmath1810 for @xmath1822 are exactly the distinct cluster monomials for the seed @xmath10 .
in particular all of their exponents are non - negative .
thus both sides of the equation are zero .
since the cluster monomials for @xmath10 are linearly independent , we conclude @xmath1823 for all @xmath1824 .
varying @xmath10 the result follows .
as explained in the introduction , the fock - goncharov conjecure is in general false , as a cluster variety @xmath15 has in general too few functions .
the conjectured theta functions only exist formally , near infinity , in the sense of [ formalfgcsec ] .
the failure of convergence in general manifests itself in the existence of infinitely many broken lines with a given incoming direction , and fixed basepoint , and non - finiteness of the multiplication rule ( for fixed @xmath49 infinitely many @xmath51 with @xmath1586 ) .
the remainder of the paper is devoted to the question of finding conditions on cluster varieties which guarantee the conjecture holds as stated .
one can only expect a theta function basis for @xmath29 in cases when @xmath15 has enough functions ; more precisely , when @xmath29 is finitely generated , and the natural map @xmath1825 is an open immersion note the second condition is automatic by the laurent phenomenon when @xmath15 is @xmath19-type .
our main ( and simple ) idea is that we can replace the assumption of enough functions by the existence of a bounded _
polytope in @xmath1826 , cut out by the tropicalisation of a regular function .
however , our notion of convexity is delicate , as @xmath149 a priori only has a piecewise linear structure .
the correct notion is explained in [ convexityconditionssection ] .
polytopes will play several roles .
the existence of bounded convex polytopes implies various results on convergence of theta functions .
we get finiteness of the multiplication rule , and so an algebra structure on @xmath43 , see proposition [ fmprop ] . for technical reasons ,
see remark [ mandelremark ] , we often have to replace enough global functions by enough global monomials ( egm ) , and can make optimal use only of convex polytopes cut out by the tropicalisations of global monomials .
egm implies @xmath1827 , see proposition [ cafin ] , and proposition [ markscor ] .
convex polytopes give partial ( full in the bounded polytope case ) compactifications of @xmath189 , by copying the familiar toric construction , see [ candsec ] .
these compactifications then degenerate to toric compactifications under ( the analog ) of the canonical degeneration @xmath1828 of @xmath19 to @xmath129 .
we use these degenerations to prove @xmath189 is log cy , see theorem [ cfcor ] .
convex cones in @xmath1826 are intimately related with partial minimal models @xmath80 , and potential functions , see [ fullfgsection ] , and corollary [ spcor ] .
finally our methods give several sufficient conditions to guarantee the full fock - goncharov conjecture , see proposition [ egmscprop ] and corollary [ ffgcor ] .
these statements are weaker , and more technical , than one would hope the reason is our inability to prove in full generality that egm ( or better , the existence of a bounded convex polytope ) implies @xmath1829 .
we have only optimal control over the subset @xmath1830 ( the cluster complex ) , and the technical statements are various ways of saying @xmath118 is sufficiently big .
the first issue is to make sense of the notion of convexity in @xmath146 .
the following is elementary : [ mcdef ] by a piecewise linear function on a real vector space @xmath223 we mean a continuous function @xmath1831 piecewise linear with respect to a finite fan of ( not necessarily strictly ) convex cones . for a piecewise linear function @xmath1832
we say @xmath34 is _ min - convex _ if it satisfies one of the following three equivalent conditions : 1 . there are finitely many linear functions @xmath1833 such that @xmath1834 for all @xmath1835 .
@xmath1836 for all @xmath1837 and @xmath1838 .
the differential @xmath1839 is decreasing on straight lines .
in other words , for a directed straight line @xmath11 with tangent vector @xmath697 , and @xmath1840 general , then @xmath1841 where @xmath1842 is general and the subscript denotes the point at which the differential is calculated . in the case that @xmath223 is defined over @xmath1843 , then in condition ( 3 ) we can restrict to lines of rational slope .
we now define convexity for functions on @xmath149 for @xmath15 a cluster variety by generalizing the third condition above , using broken lines instead of straight lines : [ imcdef ] 1 . a _ piecewise linear function _
@xmath1844 is a function which is piecewise linear after fixing a seed @xmath10 to get an identification @xmath1845 .
if the function is piecewise linear for one seed it is clearly piecewise linear for all seeds . 2 .
let @xmath1846 be piecewise linear , and fix a seed @xmath10 , to view @xmath1847 .
we say @xmath34 is _ min - convex for @xmath15 _ ( or just _ min - convex _
if @xmath15 is clear from context ) if for any broken line for @xmath15 in @xmath1848 , @xmath1839 is increasing on exponents of the decoration monomials ( and thus decreasing on their negatives , which are the velocity vectors of the underlying directed path ) .
we note that this notion is independent of mutation , by the invariance of broken lines , proposition [ brlineinvmut ] , and thus an intrinsic property of a piecewise linear function on @xmath146 .
we have a closely related notion , instead defined using the structure constants for multiplication of theta functions . [ decdef ]
we say that a piecewise linear @xmath1849 is _ decreasing _ if for @xmath1850 , with @xmath1603 , @xmath1851 . here
@xmath1606 are the structure constants of theorem [ mainth ] .
these two notions are not quite equivalent : [ diplem1 ] 1
. if @xmath1844 is min - convex , then @xmath34 is decreasing .
2 . if @xmath1852 is decreasing , then for any seed @xmath10 , we have @xmath1853 min - convex in the sense of definition - lemma [ mcdef ] .
\(1 ) let @xmath667 , @xmath668 be broken lines .
assume @xmath34 is min - convex and that @xmath1350 very close to @xmath51 is the endpoint of each broken line , with @xmath1854 .
then @xmath1855 where @xmath1856 .
thus @xmath34 is decreasing .
\(2 ) suppose @xmath34 is decreasing .
for any @xmath1857 , and the linear structure on @xmath1858 determined by any choice of seed @xmath10 , the contribution of straight lines in definition - lemma [ structuredef ] ( and item ( 1 ) of theorem [ mainthax ] in the @xmath19 case ) shows @xmath1859 for all @xmath1860 .
thus @xmath1861 for all positive integers @xmath1862 and @xmath403 . by rescaling ,
the same is true for all positive rational numbers @xmath1862 and @xmath403 and @xmath1863 .
min - convexity in the sense of definition [ mcdef ] then follows by continuity of @xmath34 .
we have a closely related concept , capturing the generalization of the notion of a convex polytope . for @xmath1864 a closed subset ,
define the _ cone _
of @xmath180 @xmath1865 note the closure is only necessary if @xmath180 is not compact , in which case @xmath1866 is an asymptotic form of @xmath180 .
denote @xmath1867 which we view as a subset of @xmath147 .
note for @xmath1868 , @xmath1869 agrees with the obvious notion @xmath1870 .
we will call a closed subset @xmath1871 _ positive _ if for any non - negative integers @xmath1872 , any @xmath1873 , @xmath1874 , and any @xmath1875 with @xmath1876 , we have @xmath1877 . note that if @xmath180 is a cone , i.e. , invariant under rescaling , then this definition agrees with definition [ icdef ] . for a piecewise linear function @xmath1878 ,
let @xmath1879 by lemma [ diplem1 ] , if @xmath34 is min - convex in the sense of definition [ imcdef ] ( or more generally , decreasing in the sense of definition [ decdef ] ) , then under any identification @xmath1880 given by any seed , @xmath1881 is a convex polytope .
[ sminlem ] if a piecewise linear function @xmath1844 is decreasing , then @xmath1882 is positive .
furthemore , @xmath1882 is compact if and only if @xmath1844 is strictly negative away from @xmath81 .
note that @xmath1883 . thus if @xmath34 is decreasing and @xmath1884 with @xmath1885 , then @xmath1886 and thus @xmath1887 , so @xmath1888 .
the second statement is obvious .
we collect here a couple of results comparing these convexity conditions on @xmath1889 and @xmath795 .
recall from proposition [ ldpprop ] , ( 4 ) , the natural map @xmath1890 with @xmath1891 being the canonical projection @xmath1892 , the quotient by the @xmath240 translation action .
[ mcxcase ] a piecewise linear function @xmath34 on @xmath237 is min - convex ( resp . decreasing ) if and only if @xmath1893 on @xmath1894 is min - convex ( resp . decreasing ) .
broken lines in @xmath793 are by definition images of broken lines on @xmath1894 under @xmath808 , which gives the min - convex statement .
the decreasing statement follows from the formula for the structure constants for @xmath111 of theorem [ mainthax ] , ( 1 ) . to understand the relationship between positive polytopes in @xmath1895 and @xmath119
, we need to understand how broken lines behave under the canonical @xmath240-translation on @xmath1895 : [ eqslem ] for @xmath1896 general and @xmath1381 there are natural bijections between the following sets of broken lines @xmath1897 if @xmath1586 , then @xmath1898 and @xmath1899 .
the implications for @xmath59 follow from the equality of the sets using definition - lemma [ structuredef ] . to get the bijections between the sets , we first recall that every wall of @xmath1723 is invariant under the canonical @xmath240-translation and is contained in a hyperplane @xmath807 for some @xmath1900 .
thus @xmath240 acts on broken lines , by translation on the underlying path , keeping the monomial decorations the same .
this gives the bijection between the second and third sets .
for bijection between the first and second sets , we need to translate the decorations on each straight segment of @xmath341 by @xmath510 .
this will change the slopes of each line segment . to do this precisely ,
take @xmath341 in the first set , say with straight decorated segments @xmath1901 taken in reverse order , with @xmath1902 the infinite segment .
suppose the monomial attached to @xmath1903 is @xmath1904 with @xmath1905 .
say the bends are at points @xmath1906 along a wall contained the hyperplane @xmath1907 so that @xmath1903 is parameterized ( in the reverse direction to that of definition [ blde ] ) by @xmath1908 , @xmath1909 .
then we define @xmath1910 observe that @xmath1911 .
let @xmath1912 be the segment @xmath1913 , @xmath1914 , with attached monomial @xmath1915 .
then @xmath1916 form the straight pieces of a broken line @xmath918 in the second set .
this gives the desired bijection .
[ rescalelem ] let @xmath1917 , @xmath460 a positive integer .
if @xmath1918 , then @xmath1919 . this follows immediately from definition - lemma [ structuredef ] and the argument given in the proof of saturatedness in theorem [ goodalgebra ] .
this latter argument shows that if there is a broken line @xmath341 with @xmath1920 , @xmath1921 , then there is a broken line @xmath918 with @xmath1922 , @xmath1923 .
[ tqprop ] suppose @xmath1924 is a positive polytope defined over @xmath1843 ( i.e. , all faces span rationally defined affine spaces )
. then @xmath1925 is positive .
suppose @xmath1926 , and @xmath1603 .
we can always write @xmath1927 with @xmath1928 and @xmath1929 by the rationality assumption .
let @xmath460 be a positive integer such that @xmath1930 and @xmath1931 are all integral for @xmath514 . then because @xmath1932 , @xmath1933 by lemma [ rescalelem ] , and thus @xmath1934 by lemma [ eqslem ] .
as @xmath1935 , positivity of @xmath180 implies @xmath1936 and thus @xmath1937 .
the chief difficulty now lies in constructing min - convex functions or positive polytopes .
we turn to this next .
the following would be a powerful tool for construction min - convex functions on cluster varieties : [ dcon ] if @xmath1938 is a regular function on a log calabi - yau manifold @xmath15 with maximal boundary , then @xmath1939 is min - convex . here
@xmath1940 for the valuation @xmath34 .
[ mandelremark ] to make sense of the conjecture one needs a good theory of broken lines , currently constructed in @xcite in dimension two , and here for cluster varieties of all dimension . in dimension two ,
the conjecture has been proven by travis mandel , @xcite .
also , it is easy to see that in any case , for each seed @xmath10 and regular function @xmath34 , that @xmath1941 , see , is min - convex in the sense of definition [ mcdef ] .
indeed this is the standard ( min ) tropicalisation of a laurent polynomial .
we hope to eventually give a direct geometric description of broken lines ( without reference to a scattering diagram ) , for any log cy , as tropicalisations of some algebraic analog of holomorphic disks .
we expect the conjecture to follow easily from such a description . in fact
, we can prove conjecture [ dcon ] for global monomials , which gives our main tool for constructing min - convex functions ( our inability to prove the conjecture in general is the main reason we use the condition egm rather than the more natural condition of enough global functions ) : [ convcor ] for a global monomial @xmath34 on @xmath161 , the tropicalisation @xmath763 is min - convex , and in particular , by lemma [ diplem1 ] , decreasing .
[ diplem ] let @xmath1942 be a piecewise linear function . 1 .
if @xmath1631 or @xmath27 and for some choice of seed @xmath34 is the minimum of a collection of linear functionals @xmath1943 , each of which is non - negative on all the initial scattering monomials of @xmath812 ( resp .
, for @xmath82 , the pullbacks @xmath1944 are non - negative on the initial scattering monomials of @xmath812 ) then @xmath34 is min - convex .
2 . for @xmath165 ,
if @xmath34 is linear in a neighborhood of every wall of @xmath812 , then @xmath34 is min - convex if and only if each @xmath1943 is non - negative on each of the initial scattering monomials .
[ evalrem ] recall from [ blgcss ] that for any choice of seed , the scattering monomials in @xmath806 lie in @xmath1945 .
so it makes sense to evaluate functions defined only on @xmath1946 on scattering monomials for @xmath1723 .
choose a seed and suppose @xmath34 is the minimum of the @xmath1943 .
we consider a broken line @xmath341 , with two consecutive monomial decorations @xmath1947 . possibly refining the linear segments
, we can assume @xmath34 is given by @xmath1948 along the first segment , and @xmath1949 along the second .
let @xmath1950 be points in the domain of @xmath341 in the two segments .
then @xmath1951 the last inequality comes from the fact that @xmath1952 and @xmath860 lies on the side of the wall crossed by @xmath341 where @xmath1953 .
now @xmath1954 is some positive multiple of the scattering monomial .
this gives ( 1 ) .
if @xmath34 is linear near any bend of the broken line , then @xmath1955 , the inequality is an equality , and the right- ( and left- ) hand side is just @xmath1956 , which gives the equivalence of ( 2 ) . first consider the case @xmath165 .
suppose @xmath34 is a global monomial which is a character on a chart indexed by @xmath10 .
then by lemma [ gmlemma ] , this character is regular on @xmath1957 , i.e. , it is a character whose geometric tropicalisation has non - negative value on each ray in the fan @xmath1958 .
these rays are spanned by @xmath1959 , @xmath76 , the negatives of the initial scattering monomials for @xmath812 .
thus , because of the sign change between geometric and fock - goncharov tropicalisation , see , @xmath763 is non - negative on the initial scattering monomials .
thus @xmath763 is min - convex by lemma [ diplem ] .
the same argument then applies in the @xmath72 case , see remark [ evalrem ] . for the @xmath73 case
, a global monomial on @xmath795 pulls back to a global monomial on @xmath186 via the map @xmath1960 , and then the result follows from the @xmath165 case by lemma [ mcxcase ] .
we need a slight refinement of proposition [ convcor ] for the proof of finite generation of the canonical algebra in the next subsection .
[ blblem ] let @xmath15 be a cluster variety and let @xmath123 be an integral point in the interior of a maximal dimensional cone @xmath1961 ( see definition [ fgccthg1 ] ) .
then @xmath1962 evaluated on monomial decorations strictly increases at any non - trivial bend of a broken line in @xmath1963 .
it is enough to treat the case @xmath1964 , because global monomials on @xmath27 or @xmath19 are given either by @xmath1076-invariant global monomials on @xmath56 or by restriction of global monomials on @xmath56 to @xmath19 respectively . furthermore , broken lines in @xmath1895 yield broken lines in @xmath1965 and @xmath119 .
the integral points of the cluster cone @xmath1966 correspond to characters of @xmath1967 which extend to global regular functions on @xmath206 .
just as in the proof of proposition [ convcor ] , these are the characters @xmath36 with @xmath860 non - negative on the rays spanned by @xmath1959 , @xmath76 , the negatives of the initial scattering monomials for @xmath812 .
thus , because of the sign change between geometric and fock - goncharov tropicalisation , see , @xmath1968 in this case @xmath1962 is strictly increasing on monomial decorations as in the statement .
we now introduce our key assumption necessary for proving strong results about theta functions and the algebras they generate .
we say that @xmath161 has _ enough global monomials _
if for any @xmath1969 , @xmath1970 , there is a global monomial @xmath1971 such that @xmath1972 .
[ egmlem ] under any of the identifications @xmath1973 induced by a choice of seed , the set @xmath1974 is a closed convex subset of @xmath114 .
the following are equivalent : 1 .
@xmath161 has enough global monomials .
@xmath1975 is bounded , or equivalently , the intersection of all sets @xmath1976 for @xmath1977 equals @xmath1978 .
3 . there exists a finite number of points @xmath1979 such that @xmath1980 is bounded , or equivalently , the intersection of all sets @xmath1981 for @xmath1982 equals @xmath1978 .
4 . there is function @xmath1983 whose associated polytope @xmath1984 is bounded . by remark [ mandelremark ] , @xmath1975 is the intersection of closed rational convex polygons ( with respect to any seed ) , and hence is a closed convex set .
the equivalence of ( 1 ) and ( 2 ) is immediate from the definitions , while ( 3 ) clearly implies ( 2 ) . for the converse ,
let @xmath1589 be a sphere in @xmath1985 centered at the origin .
for each @xmath1986 there is a global monomial @xmath1987 such that @xmath1988 , and thus there is an open subset @xmath1989 on which @xmath1990 is negative .
the @xmath1991 form a cover of @xmath1589 , and hence by compactness there is a finite subcover @xmath1992 .
taking @xmath1993 gives the desired collection of @xmath1994 .
finally we show the equivalence of ( 3 ) and ( 4 ) .
the @xmath121 , @xmath1995 are exactly the global monomials on @xmath161 , thus generators of @xmath1996 ) .
now for any finite collection of functions @xmath1138 , @xmath1997 and for the @xmath1138 positive universal laurent polynomials , for example for global monomials , we have equality . thus given ( 3 ) , we take @xmath1998 .
conversely , an element @xmath41 of @xmath1999 is a linear combination of some collection of @xmath2000 .
then @xmath2001 is contained in @xmath2002 , so if the latter is bounded , so is the former .
we note that the property of egm is preserved by fock - goncharov duality : [ ldegmprop ] let @xmath239 be fixed data , and @xmath2003 the langlands dual data .
we write e.g. @xmath2004 for the corresponding lattice for the data @xmath2003 as in appendix [ ldsec ] . for each seed
@xmath10 , the canonical inclusion @xmath2005 commutes with the tropicalization of mutations , and induces an isomorphism @xmath2006 for @xmath2007 , the monomial @xmath1283 on @xmath2008 is a global monomial if and only if @xmath2009 on @xmath2010 is a global monomial . finally , @xmath1889 has egm if and only if @xmath56 has egm .
the statement about tropical spaces is immediate from the definitions .
( note that a similar statement does not hold at the level of tori , so there is no isomorphism between @xmath2011 and @xmath2012 . )
the statement about global monomials is immediate from lemma [ gmlemma ] .
now the final statement follows from the definition of egm , the isomorphism @xmath2013 of proposition [ ldpprop ] , ( 1 ) , and the equality @xmath2014 of proposition [ ldpprop ] , ( 3 ) .
in we introduced @xmath43 as a @xmath8-vector space . in the presence of suitable convex objects on @xmath114
, we can in fact put an algebra structure on @xmath43 using the structure constants given by @xmath59 .
if further the egm condition holds , then @xmath43 contains @xmath29 , and is a finitely generated algebra .
this often makes it easier to work with @xmath43 , as it is a more geometric object . precisely : [ fmprop ] for @xmath2015 or @xmath27 , suppose there is a compact positive polytope @xmath2016 .
assume further that @xmath180 is top - dimensional , i.e. , @xmath2017 . then for @xmath2018 , there are at most finitely many @xmath51 with @xmath1586 .
these give structure constants for an associative multiplication on @xmath2019 if there is a compact positive polytope @xmath2020 then the same conclusion holds for the structure constants ( which are all finite ) and multiplication rule of @xmath2021 , for all @xmath188 . for @xmath56 or @xmath27 ,
the structure constants are defined in terms of broken lines .
the finiteness is then immediate from lemma [ sminlem ] .
indeed , given @xmath2022 , we have @xmath2023 , @xmath2024 for some @xmath1872 , by the fact that @xmath180 is top - dimensional , and thus if @xmath1918 , then @xmath51 lies in the bounded polytope @xmath2025 .
the algebra structure is associative by ( 3 ) of proposition [ thetabasislemma ] .
the @xmath78 case follows from the @xmath56 case and the definitions of the structure constants and multiplication rule for @xmath2021 , see theorem [ mainthax ] .
[ cafin ] for @xmath2015 or @xmath27 assume @xmath2026 has enough global monomials . for @xmath1742 assume
@xmath186 has enough global monomials .
then @xmath59 defines a @xmath8-algebra structure on @xmath43 .
the case of @xmath1742 follows from the case of @xmath56 so we may assume @xmath15 is either @xmath27 or @xmath56 . using the enough global monomials hypothesis and lemma [ egmlem ] , we can find a finite collection @xmath2027 such that the intersection of the finite collection of polytopes @xmath2028 is bounded .
but since @xmath2029 is min - convex by proposition [ convcor ] , each of these polytopes is positive by lemma [ sminlem ] .
thus the result follows from proposition [ fmprop ] .
finite generation of @xmath43 is a special case of a much more general result .
[ fgprop ] let @xmath165 or @xmath27 , assume @xmath161 has enough global monomials , and let @xmath2016 be a positive polytope , which we assume is rationally defined and not necessarily compact
. then @xmath2030\ ] ] is a finitely generated @xmath8-subalgebra .
note that @xmath2031 is a subalgebra of @xmath2032 $ ] by the definition of positive polytope . as in the proof of corollary [ cafin ]
, we can choose @xmath2033 so that @xmath2034 is a compact positive polytope . moreover , because boundedness of the intersection is preserved by small perturbation of the functions , we can assume that each @xmath1994 is in the interior of some maximal dimensional cluster cone @xmath2035 .
note the seed @xmath2036 is then uniquely determined by @xmath1994 , by lemma [ fgccthg2 ] .
it follows that @xmath2037 is strictly increasing on the monomial decorations at any non - trivial bend of any broken line in @xmath2038 by lemma [ blblem ] .
for @xmath2039 , we consider @xmath2040 here the second sum is over all @xmath2041 satisfying the stated condition .
this is a subalgebra of @xmath2042 by proposition [ convcor ] .
similarly , we define @xmath2043\ ] ] to be the vector subspace spanned by all @xmath2044 , @xmath2045 where @xmath2046 for @xmath2047 , and @xmath2048 .
then @xmath2049 is a graded subalgebra of @xmath2050 $ ] by proposition [ convcor ] ( graded by @xmath1-degree ) and @xmath2051 is the degree zero part .
the result then follows from the following claim , noting that @xmath2052 and for @xmath2053 there is a natural surjection @xmath2054 by sending @xmath2055 , @xmath2056 .
@xmath2057 is a finitely generated @xmath8-algebra . _
proof_. we argue first that @xmath2058 is finitely generated .
we work on @xmath2059 , so that the multiplication rule is defined using broken lines for @xmath2060 , as described by proposition [ thetabasislemma ] , ( 3 ) and definition - lemma [ structuredef ] .
note @xmath2061 is linear on @xmath2062 .
if @xmath2063 , then modulo @xmath1 , @xmath2064 unless @xmath2065 , for otherwise @xmath2066 . by lemma [ blblem ]
, @xmath2061 is strictly increasing on monomial decorations at any non - trivial bend of a broken line and thus the only broken lines that will contribute ( modulo @xmath1 ) to @xmath2067 are straight , thus modulo @xmath1 , @xmath2068 ( addition here in @xmath2069 ) .
thus @xmath2058 is the monoid ring associated to the rational convex cone @xmath2070 and is thus finitely generated .
now by decreasing induction on the @xmath1-degree , to show @xmath2057 is finitely generated , it is enough to show that its degree @xmath81 subalgebra , @xmath2071 , is finitely generated .
this is obvious if the set @xmath2072 is @xmath1978 , and by assumption this is true for sufficiently large @xmath2073 . by the above , in any case , to prove @xmath2071
is finitely generated , it is enough to show @xmath2074 is finitely generated .
so we are done by induction .
[ fgcor ] for @xmath165 or @xmath27 , suppose that @xmath161 has enough global monomials
. for @xmath1648 , assume @xmath186 has enough global monomials .
then @xmath43 is a finitely generated @xmath8-algebra . in the @xmath165 or @xmath27 cases ,
apply the theorem with @xmath2075 or @xmath2076 , which are trivially positive .
then @xmath132 is the degree @xmath81 part of the finitely generated ring @xmath2077 with respect to the @xmath582-grading .
since @xmath2021 is a quotient of @xmath132 by construction of @xmath2078 in theorem [ mainthax ] , ( 1 ) , @xmath2021 is also finitely generated .
[ markscor ] assume @xmath2079 has egm .
then for each universal laurent polynomial @xmath41 on @xmath56 , the function @xmath2080 of theorem [ ffgth ] has finite support ( i.e. , @xmath2081 for all but finitely many @xmath1557 ) , and @xmath2082 gives inclusions of @xmath8-algebras @xmath2083 } { \bbbk}[n].\ ] ] let @xmath41 be a universal laurent polynomial on @xmath1170 . by theorem [ ffgth ] the sets @xmath2084 of definition [ defs ] are independent of the seed @xmath10 .
we claim that for each global monomial @xmath121 on @xmath2085 , there is a constant @xmath2086 such that @xmath2087 to see that this is sufficient to prove the proposition , note that by lemma [ egmlem ] , there are a finite number of @xmath1994 such that the intersection of the sets where @xmath2088 is the origin in @xmath2089 .
thus , if the claim is true , @xmath2090 , the support of @xmath2080 , is a finite set .
the inclusion of algebras follows by proposition [ thetabasislemma ] .
so it s enough to establish the claim .
let @xmath121 be a global monomial which is a character on the seed torus for @xmath10 .
we follow the notation of definition [ defs ] .
thus @xmath2091 , where @xmath2092 itself depends on the seed @xmath10 and @xmath41 .
the tropicalization @xmath1962 of global monomials @xmath121 which restrict to a character on the seed torus @xmath2093 are identified with integer points of the dual cone @xmath2094 ( i.e. , elements non - negative on each of the initial scattering monomials ) , see the proof of proposition [ convcor ] .
note @xmath1962 is linear on @xmath2095 ( though for our purposes , its min - convexity will be enough ) .
since @xmath2092 is a finite set , for any such @xmath2096 , there is constant @xmath2086 such that @xmath2097 this completes the proof .
we begin by showing some standard conditions in cluster theory , namely acyclicity of the quiver or existence of a maximal green sequence both imply a weaker condition which in turn implies both the egm condition and the full fock - goncharov conjecture .
this suggests that this weaker condition is perhaps a more natural one in cluster theory .
this point has been explored in @xcite .
[ lccdef ] we say a cluster variety @xmath19 has _ large cluster complex _ if for some seed @xmath10 , @xmath2098 is not contained in a half - space .
[ egmscprop ] if @xmath19 has large cluster complex , then @xmath2099 has egm , @xmath2100 , and the full fock - goncharov conjecture ( see definition [ fgconjdef ] ) holds for @xmath56 , @xmath27 , very general @xmath78 and , if the convexity condition ( 7 ) of theorem [ mainth ] holds , for @xmath19 .
assume egm fails for @xmath56 .
then we have some point @xmath2101 and @xmath2102 for all @xmath2103 .
take any seed @xmath10 .
we can compute @xmath2104 by using the corresponding positive laurent polynomial @xmath2105 $ ] , for @xmath827 a point in the distinguished chamber @xmath693 of @xmath806 . thus using lemma [ lplemma ] ( leaving the canonical isomorphism @xmath51 out of the notation ) , @xmath2106 here
the minimum is over all broken lines
@xmath341 contributing to @xmath1617 and the final inequality comes from the fact that one of the broken lines is the obvious straight line .
thus @xmath110 is contained in the half - space @xmath2107 .
since @xmath2108 is the inverse image of @xmath2109 under the map @xmath2110 , the egm statement follows .
now @xmath2111 since @xmath2112 and @xmath61 is saturated and intrinsically closed under addition , see theorem [ goodalgebra ] . since @xmath56 satisfies egm , so does @xmath186 by proposition [ ldegmprop ] , and the full fock - goncharov conjecture for @xmath56 then follows from corollary [ cafin ] and proposition [ markscor ] .
the @xmath2113 and @xmath19 cases then follow as in the proofs of corollary [ xcorollary ] and theorem [ mainthax ] .
[ gsprop ] let @xmath2114 be a seed .
suppose for some vertex @xmath153 of @xmath696 the cluster chamber @xmath2115 meets the interior of @xmath817 .
then the following hold : 1 .
@xmath19 has large cluster complex .
obviously ( 1 ) implies ( 2 ) .
( 1 ) follows from the fact that each cluster chamber is a chamber in @xmath1723 , and each @xmath2117 is a union of walls from @xmath806 , see theorem [ fgccth ] .
[ mgsrem ] consider the following conditions on a skew - symmetric cluster algebra @xmath19 : 1 .
@xmath19 has an acyclic seed .
2 . @xmath19 has a seed with a maximal green sequence .
( for the definition , see @xcite , def . 1.8 . )
@xmath19 has large cluster complex .
then ( 1 ) implies ( 2 ) implies ( 3 ) .
\(1 ) implies ( 2 ) is @xcite , lemma 1.20 . for ( 2 )
implies ( 3 ) , let @xmath10 be an initial seed and @xmath609 the seed obtained by mutations in a maximal green sequence . by definition , the @xmath130-vectors for @xmath609 have non - positive entries
. by lemma [ facelem ] the @xmath130-vectors are the equations for the walls of the cluster chamber @xmath1267 , thus the hypothesis of proposition [ gsprop ] holds .
we make an aside here connecting with work of reineke @xcite in the acyclic skew - symmetric case . in this case
@xmath2118 has a natural interpretation in terms of moduli of quiver representations .
consider skew - symmetric fixed and initial data with no frozen variables .
set @xmath2119 , and let @xmath2120}$ ] be the completion of the polynomial ring @xmath2121 $ ] with respect to the maximal monomial ideal .
let @xmath2122 be a monoid as in 1 containing all @xmath1397 , so that @xmath93 , the pro - nilpotent group of [ defconstsection ] ( in the @xmath56 case ) acts by automorphisms of @xmath268}$ ] as usual .
note there is an embedding @xmath2120 } \hookrightarrow \widehat{{\bbbk}[p]}$ ] given by @xmath2123 .
the action of @xmath93 on @xmath268}$ ] then induces an action on @xmath2120}$ ] . indeed ,
one checks immediately that an automorphism ( for @xmath2124 ) @xmath2125 induces the automorphism on @xmath2120}$ ] given by @xmath2126 [ reinekeprop ] suppose given fixed skew - symmetric data with no frozen variables along with an acyclic seed @xmath787 .
let @xmath827 be the associated quiver . that @xmath1675 for any @xmath263 , the quiver @xmath827 has no isolated vertex .
] each @xmath365 gives a _ stability _ in the sense of @xcite .
assume there is a unique primitive @xmath2127 with @xmath2128 .
for each @xmath1237 let @xmath2129 where @xmath2130 is the framed moduli space ( framed by the vector spaces @xmath2131 with @xmath2132 unless @xmath2133 , in which case @xmath2134 ) of semi - stable representations of @xmath827 with dimension vector @xmath2135 and @xmath367-slope zero , ( see @xcite,5.1 ) and @xmath2136 denotes topological euler characteristic .
let @xmath2137 for some @xmath2138 .
then @xmath2139 depends only on @xmath827 and @xmath367 ( i.e. , is independent of the vertex @xmath1237 ) . furthermore , for arbitrary @xmath2140 , @xmath2141 ( see lemma [ easyequivlemma ] ) acts on @xmath2120}$ ] by @xmath2142 and on @xmath268}$ ] by @xmath2143 the equality of the @xmath2144 follows from the argument in the proof of @xcite , lemma 3.6 .
if @xmath2145 for some @xmath210 then one checks easily that @xmath2146 is a point and @xmath2147 for @xmath2148 or @xmath2149 .
thus @xmath2150 and the formula for @xmath2151 holds by remark [ noekwalls ] .
let @xmath93 be the pro - nilpotent group of [ defconstsection ] ( in the @xmath56 case ) associated to the completion of the lie algebra @xmath2152 the group @xmath93 acts faithfully on @xmath268}$ ] but need not act faithfully via restriction on @xmath2120}$ ] .
it turns out , however , that there is a subgroup @xmath2153 which does act faithfully on @xmath2120}$ ] and that all automorphisms attached to walls in @xmath378 are elements of @xmath336 .
we see this as follows .
consider the subspace @xmath2154 by the commutator formula ( [ liebracketform ] ) we have @xmath2155 \subset { \mathfrak{g}}'$ ] , and in particular @xmath337 is a lie subalgebra of @xmath283 .
let @xmath2156 denote the associated pro - nilpotent subgroup .
because of the assumption that @xmath2157 for any unfrozen @xmath210 , all automorphisms associated to initial walls of @xmath378 lie in @xmath336 , and thus by the inductive construction of the scattering diagram in [ subseckslemalg ] , all automorphisms associated to outgoing walls of @xmath378 also lie in @xmath336 .
let @xmath2158 denote the pro - nilpotent group acting faithfully on @xmath2120}$ ] associated to the completion of the lie algebra @xmath2159 then the restriction of the action of @xmath93 on @xmath268}$ ] to @xmath2120}$ ] is given by the group homomorphism @xmath2160 associated to the lie algebra homomorphism @xmath2161 this homomorphism restricts to an isomorphism @xmath2162 , and in particular the restriction of the @xmath336 action on @xmath268}$ ] to @xmath2120}$ ] is faithful .
assume now that the indices are ordered so that @xmath827 has arrows from the vertex with index @xmath210 to the vertex with index @xmath2073 only if @xmath2163 .
we compute @xmath2164 , the automorphism associated to a path from the positive to the negative chamber , in two different ways .
first , there is a sequence of chambers connecting @xmath693 to @xmath817 via the mutations @xmath2165 , @xmath2166 .
indeed , it is easy to check that the @xmath130-vectors obtained by mutating @xmath2167 @xmath2168 are precisely @xmath2169 , and the chamber corresponding to this sequence of mutations is precisely the dual of the cone generated by the @xmath130-vectors , see lemma [ facelem ] .
thus in particular , we can find a path @xmath341 from @xmath693 to @xmath817 which only crosses the walls @xmath2170 in order .
note that the element of @xmath336 attached to the wall @xmath2171 acts on @xmath2120}$ ] by @xmath2172 , which agrees with the automorphism in @xcite written as @xmath2173 ( noting that @xcite uses the opposite sign convention for the skew form @xmath798 associated to the quiver ) . from this
we conclude that @xmath2174 , the left - hand side of the equality of theorem 2.1 of @xcite . on the other hand , choose a stability condition @xmath367 and consider the path @xmath341 from @xmath2175 to @xmath817 parameterized by @xmath2176 , with @xmath2177 , with domain sufficiently large so the initial and final endpoints lie in the positive and negative orthants respectively .
then a dimension vector has @xmath2178-slope @xmath81 if and only if it has @xmath367-slope @xmath2176 . thus if the description in the statement of the theorem of @xmath2179 is correct , then @xmath2180 coincides with the right - hand side of the equality of theorem 2.1 of @xcite . by the uniqueness of the factorization of @xmath447 from the proof of theorem [ maximscorrespondence ] and the faithful action of @xmath336 on @xmath2120}$ ] shown above ,
we obtain the result .
because non - negativity of euler characteristics for the quiver moduli spaces appearing in the above statement is known ( @xcite ) , this gives an alternate proof of positivity of the scattering diagram in this case .
since the initial version of this paper was released , bridgeland @xcite developed a hall algebra version of scattering diagrams in the context of quiver representation theory , and the above result follows conceptually from his results . [ finite_type_scattering ]
let @xmath827 be a quiver given by an orientation of the dynkin diagram of a simply - laced finite dimensional simple lie algebra .
then the dimension vectors of the indecomposable complex representations of @xmath827 are the positive roots of the associated root system @xmath2181 ( gabriel s theorem ) . moreover , for each positive root @xmath2135 , there is a unique indecomposable representation @xmath15 with dimension vector @xmath2135 , and @xmath2182 .
see e.g. @xcite .
the @xmath19 cluster variety associated to @xmath827 is the cluster variety of finite type associated to the root system @xmath2181 @xcite . using proposition [ reinekeprop ]
we can give an explicit description of the scattering diagram @xmath328 for @xmath56 as follows .
first we observe that a representation of @xmath827 that contributes to @xmath328 is a direct sum of copies of an indecomposable representation .
let @xmath2183 be a primitive vector and @xmath365 be such that @xmath2184 .
suppose @xmath223 is an @xmath367-semistable representation of @xmath827 with dimension vector a multiple of @xmath2135 , and consider the decomposition of @xmath223 into indecomposable representations .
by @xmath367-semistability and our assumption @xmath2184 , each factor must have dimension vector a multiple of @xmath2135 . by gabriel s theorem
, we see that @xmath2135 is a positive root and @xmath223 is a direct sum of copies of the associated indecomposable representation .
we see that the walls of @xmath328 are in bijection with the positive roots of @xmath2181 .
let @xmath2185 be a positive root and @xmath15 the indecomposable representation with dimension vector @xmath2135 .
let @xmath2186 be the locus of @xmath365 such that @xmath15 is @xmath367-semistable of @xmath367-slope zero , that is , @xmath2187 and @xmath2188 for @xmath2189 the dimension vector of any subrepresentation of @xmath15 .
then @xmath613 is a rational polyhedral cone in @xmath2190 , and is non - empty of real codimension @xmath616 .
indeed , there exists @xmath2128 such that @xmath15 is @xmath367-stable by @xcite , remark 4.5 and @xcite , theorem 6.1 , and this is an open condition on @xmath2128 .
now let @xmath2191 be a point such that @xmath2184 .
then the @xmath367-semistable representations of @xmath367-slope zero are the direct sums of copies of @xmath15 .
let us now examine the moduli space @xmath2192 .
an object in this moduli space is a direct sum @xmath2193 of @xmath460 copies of the unique indecomposable representation of dimension vector @xmath2135 , along with the framing , a choice of a vector @xmath2194 .
such an object is stable if and only if @xmath697 is not contained in a proper subrepresentation of @xmath2195 of the form @xmath2196 for some subspace @xmath2197 . in order for this to be the case
, the @xmath2198 must be linearly independent elements of @xmath2199 , and hence span a @xmath460-dimensional subspace of @xmath2199 .
the automorphism group of @xmath2195 is @xmath2200 , which has the effect of changing the basis of the subspace spanned by @xmath2198 .
now it follows easily from the definitions that , for each @xmath2201 and @xmath1237 such that @xmath2202 , the moduli space @xmath2203 of @xmath367-semistable representations with framing at vertex @xmath210 is isomorphic to the grassmannian @xmath2204 .
so , in the notation of proposition [ reinekeprop ] , @xmath2205 and @xmath2206 thus the wall of @xmath328 associated to @xmath2185 is @xmath2207 for example , suppose @xmath827 is the quiver with vertices @xmath2208 , and arrows from 1 to 2 and 2 to 3 .
this is an orientation of the dynkin diagram @xmath2209 .
we have the following isomorphism types of indecomposable representations : @xmath2210 @xmath2211 ( here the numbers denote the dimension of the vector space at the vertex , and the symbol @xmath2212 over an arrow indicates that the corresponding linear transformation is an isomorphism . )
we write @xmath2213 and @xmath2214
. then the walls of @xmath328 are @xmath2215 @xmath2216 @xmath2217 @xmath2218 for example , the indecomposable representation with dimension vector @xmath2219 has subrepresentations with dimension vectors @xmath2220 and @xmath2221 .
so the associated wall has support @xmath2222 defined by the inequalities @xmath2223 and @xmath2224 .
this gives the last wall in the list .
[ acyclic_scattering ] kac generalized gabriel s theorem to the case of an arbitrary quiver @xmath827 ( without edge loops ) as follows @xcite,@xcite : let @xmath283 be the kac - moody algebra associated to the underlying graph of @xmath827 .
then the dimension vectors of indecomposable complex representations of @xmath827 are the positive roots of @xmath283 .
the roots @xmath2181 of @xmath283 are divided into real and imaginary roots .
the real roots are the translates of the simple roots @xmath1286 under the action of the weyl group .
let @xmath2225 be the asymmetric bilinear form defined by @xmath2226 then for representations @xmath15 and @xmath2227 of @xmath827 with dimension vectors @xmath2135 and @xmath2189 , @xmath2228 let @xmath2229 be a positive root
. we have @xmath2230 if @xmath2135 is real and @xmath2231 if @xmath2135 is imaginary .
the indecomposable representations of dimension vector @xmath2135 depend on @xmath2232 parameters .
we say @xmath2135 is a _
root if there exists a representation @xmath15 of @xmath827 with dimension vector @xmath2135 such that @xmath2182 .
assume that @xmath827 is acyclic .
let @xmath328 be the scattering diagram for the associated @xmath56 cluster variety .
we study @xmath328 using proposition [ reinekeprop ] .
we show that each wall of @xmath328 is contained in @xmath2233 for @xmath2135 a primitive schur root .
first , as in example [ finite_type_scattering ] , each wall is contained in @xmath2233 for @xmath2135 a primitive positive root ( note that the set of roots is saturated by @xcite , proposition 1.2 ) .
it remains to show that @xmath2135 is necessarily a schur root .
otherwise , a representation @xmath15 with dimension vector @xmath2135 deforms to a decomposable representation @xmath2227 @xcite , proposition 1(b ) .
then , for @xmath365 such that @xmath2184 , @xmath2227 is @xmath367-unstable and so @xmath15 is @xmath367-unstable ( as @xmath367-semistability is an open condition ) . a representation with dimension vector a multiple of @xmath2135 is @xmath367-unstable for the same reason , using @xcite , theorem 3.8 .
now by proposition [ reinekeprop ] we see that there does not exist a wall of @xmath328 contained in @xmath2233 . for a real schur root @xmath2135
there is a unique wall contained in @xmath2233 which can be described explicitly as in example [ finite_type_scattering ] .
we remark that @xmath2135 is a real schur root iff there is an indecomposable representation @xmath15 of @xmath827 with dimension vector @xmath2135 such that @xmath2182 and @xmath2234 .
( moreover , @xmath15 is uniquely determined by @xmath2135 . )
such a representation @xmath15 is an exceptional object in the category of representations of @xmath827 in the sense of @xcite . for an imaginary schur root
the associated walls involve contributions from positive dimensional moduli spaces of semistable representations of @xmath827 .
the case of the imaginary root @xmath2235 for the quiver @xmath827 with vertices @xmath2236 and two arrows from 1 to 2 is described in @xcite , 6.1 .
( this is the case @xmath936 of example [ bcexample ] . )
we next give another condition for the egm condition to hold . while this may appear very technical , it is in fact very important for group - theoretic examples , see remark [ egmrem ] .
[ mcexprop ] 1 .
let @xmath2237 be an affine variety over a field @xmath8 , and @xmath2238 generators of @xmath1174 as a @xmath8-algebra .
for each divisorial discrete valuation @xmath2239 ( where @xmath2240 denotes the function field of @xmath1 ) which does not have center on @xmath1 ( or equivalently , for each boundary divisor @xmath2241 in any partial compactification @xmath2242 ) , @xmath2243 for some @xmath210 .
2 . suppose @xmath15 is a cluster variety , @xmath2244 is a smooth affine variety , and @xmath2245 is an open immersion .
let @xmath2238 generate @xmath29 as a @xmath8-algebra .
then @xmath2246 is strictly negative on @xmath2247 .
\(1 ) let @xmath2248 be an open immersion with complement an irreducible divisor @xmath2249 .
suppose each @xmath2250 is regular along @xmath2249 .
then the inclusion @xmath2251 is an equality .
thus the inverse birational map @xmath2252 is regular , which implies @xmath2253 .
thus ( 1 ) follows .
\(2 ) since the restriction @xmath2254 to the open subset @xmath2255 is an isomorphism , it follows that @xmath2256 has codimension at least two .
thus @xmath1 itself is log calabi - yau by @xcite , lemma 1.4 , and the restriction @xmath2257 of the holomorphic volume form is a scalar multiple of @xmath2258 .
in addition @xmath2259 .
now ( 2 ) follows from ( 1 ) .
[ egmprop1 ] if the canonical map @xmath2260 is surjective , then 1 .
@xmath2261 is isomorphic to @xmath2262 .
2 . we can choose @xmath2263 so that the induced map @xmath2264 is an isomorphism .
the map induced by the choice of @xmath2265 in ( 2 ) , @xmath2266 , is finite .
4 . if furthermore for each @xmath2267 we can find a cluster variable @xmath1174 with @xmath2268 , then @xmath19 ( and @xmath56 ) has enough global monomials .
this final condition holds if @xmath2269 is finitely generated and @xmath172 is a smooth affine variety .
\(1 ) is lemma [ taprinlem ] .
( 3 ) follows from ( 2 ) .
so we assume @xmath2270 is surjective and show we can choose @xmath2265 to have finite cokernel , or equivalently , so @xmath2265 is injective .
we follow the notation of @xcite , 2.1 . by the assumed surjectivity ,
@xmath2265 is injective iff the induced map @xmath2271 is injective . we can replace @xmath2265 by @xmath2272 for any map @xmath2273 which vanishes on @xmath244 , i.e. , factors through a map @xmath2274 .
note by the assumed surjectivity that @xmath190 and @xmath2275 have the same rank , and moreover the restriction @xmath2276 ( which is unaffected by the addition of @xmath59 ) is injective . in particular
@xmath2277 is injective .
thus we can choose @xmath2278 , vanishing on @xmath2279 ( i.e. , factoring through a map @xmath2280 ) so that @xmath2281 is injective . by viewing the determinant of @xmath2282 for @xmath860 an integer as a polynomial in @xmath860 , we see that @xmath2283 is injective for all but a finite number of @xmath860 . for
sufficiently divisible @xmath860 , @xmath2284 extends to @xmath2274 under the natural inclusion @xmath2285 .
now @xmath2286 is injective as required .
this shows ( 2 ) . for ( 4 )
, when @xmath1523 is a trivial bundle , it follows that @xmath2287 so we have enough global monomials so long as we can find cluster variables on @xmath19 with the given condition .
the final statement of ( 4 ) follows from proposition [ mcexprop ] .
[ egmrem ] every double bruhat cell is an affine variety by @xcite , prop .
2.8 and smooth by @xcite , theorem 1.1 .
the surjectivity condition in the statement of proposition [ egmprop1 ] holds for all double bruhat cells by @xcite , proposition 2.6 ( the proposition states that the exchange matrix has full rank , but the proof shows the surjectivity ) .
so by the proposition , @xmath56 has enough global monomials for double bruhat cells for which the upper and ordinary cluster algebras are the same .
this holds for the open double bruhat cell of @xmath93 and the @xmath98 ( @xmath2288 maximal unipotent ) for @xmath2289 by @xcite , remark 2.20 , and is announced in @xcite for all double bruhat cells of all semi - simple @xmath93 . in this subsection
, we will use positive polytopes in @xmath2290 to create partial compactifications of @xmath2291 which fibre over an affine space @xmath1136 .
the fibre over @xmath81 will be a toric variety , and the general fibre is log calabi - yau .
fix seed data for a cluster variety , @xmath74 and let @xmath2292 be the monoid generated by the @xmath1232 .
similarly , let @xmath2293 be the cone generated by the @xmath1232 .
the choice of seed gives an identification @xmath2294 and in particular determines a second projection @xmath2295 ( which depends on the choice of seed ) .
we have the canonical inclusion @xmath2296 given in each seed by @xmath2297 , and canonical translation action of @xmath240 on @xmath1424 making @xmath132 into a @xmath1423$]-module . now assume given a compact , positive , rationally defined top - dimensional polytope @xmath2298 .
we let @xmath2299 . by proposition [ fmprop ] and the compactness of @xmath180
, @xmath1589 is a @xmath1423$]-algebra with @xmath8-algebra structure constants @xmath1606 .
the set @xmath2300 is a positive polytope .
denote by @xmath2301 the degree @xmath81 part of the ring @xmath2302 defined in theorem [ fgprop ] .
then @xmath2301 is a finitely generated @xmath2303$]-algebra .
positivity follows from the fact that @xmath2304 for each scattering monomial @xmath860 in @xmath806 .
the finite generation statement then follows from theorem [ fgprop ] .
let @xmath2305 , and @xmath2306 then @xmath2307 is positive by proposition [ tqprop ] , and as the intersection of two positive sets is positive , @xmath2308 is positive . hence the associated graded rings @xmath2309 and @xmath2310 ( graded by @xmath582 ) defined via theorem [ fgprop ] are finitely generated .
note that @xmath2301 is the set of homogeneous elements of degree @xmath81 in the localization @xmath2311 .
thus we have an inclusion @xmath2312 an open subset , with complement the zero locus of @xmath2313 .
the inclusion of @xmath2314=\vartheta_0{\bbbk}[n_{{{\bf s}}}^{\oplus}]$ ] in the degree 0 part of @xmath2077 induces a morphism @xmath2315 ) = \ba^n_{x_1,\dots , x_n}$ ] .
this morphism is flat , since @xmath2077 is a free @xmath2303$]-module .
[ polylem ] the central fibre of @xmath2316 is the polarized toric variety @xmath2317 given by is only a rationally defined polyhedron rather than a lattice polyhedron , we can still define @xmath2318 . ]
the polyhedron @xmath2319 where @xmath2320 is the natural map of proposition [ ldpprop ] , ( 4 ) .
this follows from the multiplication rule . indeed , since all the scattering monomials project under @xmath2321 into the interior of @xmath2322
, @xmath2323 vanishes modulo the maximal ideal of @xmath2303 $ ] for any broken line @xmath341 that bends , see e.g. the proof of corollary [ cafin ] . thus @xmath2324 } ( { \bbbk}[n^{\oplus}_{{{\bf s}}}]/(x_1,\dots , x_n ) ) = \bigoplus_{d \geq 0 } \bigoplus_{q \in d \cdot \oxi } { \bbbk}\cdot \vartheta_q \cdot t^d\ ] ] with multiplication induced by @xmath2325 ( addition in @xmath1175 ) .
this is the coordinate ring of @xmath2326 consider the fixed data and seed data given in example [ basicscatteringexample ] .
the scattering diagram for @xmath56 in this case has three walls , pulled back from the walls of the scattering diagram for @xmath19 as given in example [ basicscatteringexample ] , with attached functions @xmath2327 , @xmath2328 and @xmath2329 . here , with basis @xmath390 of @xmath240 and dual basis @xmath389 of @xmath1444 , we have @xmath2330 and @xmath2331 . take @xmath2332 to be the pentagon with vertices ( with respect to the basis @xmath410 ) @xmath2333 , @xmath2334 , @xmath2335 , @xmath2336 , and @xmath2337 , which we write as @xmath2338
. then @xmath2339 pulls back to @xmath1792 to give a polytope @xmath180 .
it is easy to see that @xmath180 is a positive polytope .
further , write @xmath2340 , @xmath2341
. then it is not difficult to describe the ring @xmath2042 determined by @xmath2308 as the graded ring generated in degree @xmath616 by @xmath2342 , with relations @xmath2343 these equations define a family of projective varieties in @xmath2344 , parameterized by @xmath2345 . for @xmath2346 , we obtain a smooth del pezzo surface of degree @xmath2347 . the boundary ( where @xmath2348 ) is a cycle of five projective lines . when @xmath2349 , we obtain a toric surface with two ordinary double points
. [ cfcor ] assume that @xmath186 has enough global monomials , @xmath180 is given as above , and that @xmath8 is an algebraically closed field of characteristic zero .
let @xmath15 be one of @xmath2350 or @xmath56 .
we note @xmath43 has a finitely generated @xmath8-algebra structure by proposition [ cafin ] .
define @xmath2351 .
define @xmath2352 ( constructed above ) in case @xmath2015 , and for @xmath2353 , take instead its fibre over @xmath1734 ( we are not defining @xmath0 in the @xmath2354 case ) , so by construction we have an open immersion @xmath2242 .
define @xmath2355 .
the following hold : 1 . in all cases
@xmath1 is a gorenstein scheme with trivial dualizing sheaf .
2 . for @xmath2015 , @xmath27 , or @xmath78 for @xmath188 general ,
@xmath1 is a @xmath190-trivial gorenstein log canonical variety .
3 . for @xmath2015 or @xmath78 for @xmath188 general , or all @xmath78 assuming there exists a seed @xmath2356 and a strictly convex cone containing all of @xmath1766 for @xmath76 , we have @xmath2242 is a minimal model . in other words , @xmath0 is a ( in the @xmath56 case relative to @xmath2357 ) projective normal variety , @xmath2358 is a reduced weil divisor , @xmath2359 is trivial , and @xmath2360 is log canonical .
first we consider the theorem in the cases @xmath2361 .
note that ( 3 ) implies ( 2 ) by restriction .
we consider the family @xmath2362 constructed above , where @xmath92 is the divisor given by @xmath2363 with its reduced structure .
using lemma [ kollarslemma ] below , the condition that on a fibre @xmath1609 , @xmath2364 is a minimal model ( in the sense of the statement ) is open , and it holds for the central fibre as it is toric by theorem [ polylem ] .
thus the condition holds for fibres over some non - empty zariski open subset @xmath2365 .
this gives ( 3 ) for @xmath78 with @xmath188 general .
the convexity condition ( on the @xmath595 ) implies there is a one - parameter subgroup of @xmath22 which pushes a general point of @xmath2366 to @xmath81 ( see the proof of theorem [ mainthax ] ) , and now ( 3 ) for @xmath78 for all @xmath188 follows by the @xmath22-equivariance .
now note given seed data @xmath239 the convexity assumption holds for the seed data @xmath943 .
thus the final paragraph applies with @xmath2367 and so in particular @xmath2291 is gorenstein with trivial dualizing sheaf . the same
then holds for the fibres of the flat map @xmath2368 , which are @xmath2369 ( for arbitrary @xmath1734 ) .
this gives ( 1 ) .
finally we consider the case @xmath2370 .
the graded ring construction above applied with seed data @xmath943 gives a degeneration of a compactification of @xmath2371 ( which is now a fibre of the family ) to a toric compactification of @xmath2372 .
the torus @xmath1076 acts on the family , trivially on the base , and the quotient gives an isotrivial degeneration of an analogously defined compactification of @xmath2373 to a toric compactification of @xmath2357 .
we leave the details of the construction ( which is exactly analogous to the construction of @xmath2374 above ) to the reader .
now exactly the same openness argument applies .
we learned of the following result , and its proof , from j. kollr .
[ kollarslemma ] let @xmath8 be an algebraically closed field of characteristic zero .
let @xmath2375 be a proper flat morphism of schemes of finite type over @xmath8 , and @xmath2376 a closed subscheme which is flat over @xmath1589 .
let @xmath2377 denote the fiber of @xmath2378 over a closed point @xmath2379 .
assume that @xmath1589 is regular and for @xmath2380 the following hold : 1 .
@xmath2381 is normal and cohen macaulay .
2 . @xmath2382 is a reduced divisor .
the pair @xmath2383 is log canonical .
5 . @xmath2385 then the natural morphism @xmath2386 is an isomorphism , and there exists a zariski open neighbourhood @xmath2387 such that the conditions ( 1 - 5 ) hold for all @xmath2388 . in particular , @xmath2389 is a @xmath190-trivial gorenstein log canonical variety for all @xmath2388 .
we are free to replace @xmath1589 by an open neighbourhood of @xmath2379 and will do so during the proof without further comment . by assumption @xmath2390 and @xmath2391
is cohen macaulay .
so @xmath2392 is cohen - macaulay by @xcite , corollary 2.71 , p. 82 .
it follows that @xmath2393 is cohen
macaulay by @xcite , corollary 2.63 , p. 80 .
the base @xmath1589 is regular by assumption , so @xmath2379 is cut out by a regular sequence . since @xmath2391 and @xmath2393 are cohen - macaulay , and @xmath2394 is proper and flat , we may assume that @xmath727 and @xmath92 are cohen macaulay .
now @xmath2395 is cohen
macaulay by @xcite , corollary 2.63 , and @xmath2396 is cohen
macaulay by @xcite , corollary 2.71 .
the relative dualizing sheaf @xmath2397 is identified with @xmath2398 , so @xmath2399 is also cohen macaulay .
it follows that @xmath2400 is cohen macaulay , and so in particular satisfies serre s condition @xmath414 .
the natural map @xmath2386 is an isomorphism in codimension @xmath616 ( because @xmath2391 is smooth in codimension @xmath616 ) and both sheaves are @xmath414 , hence the map is an isomorphism .
now @xmath2390 implies that we may assume @xmath2401 using @xmath2402 .
the conditions ( 1),(2 ) , and ( 5 ) are open conditions on @xmath2403 because @xmath2394 is proper and flat .
so we may assume they hold for all @xmath2403 .
we established above that @xmath2399 is invertible .
it follows that condition ( 3 ) is also open on @xmath2403 by @xcite , corollary 4.10 , p. 159 , and that condition ( 4 ) is open on @xmath1589 ( using ( 5 ) ) .
[ crems ] note that directly from its definition , with the multiplication rule counting broken lines , it is difficult to prove anything about @xmath43 , e.g. , that it is an integral domain , or determine its dimension .
but the convexity , i.e. , existence of a convex polytope in the intrinsic sense , gives this very simple degeneration from which we get many properties , at least for very general @xmath78 , for free .
there have been many constructions of degenerations of flag varieties and the like to toric varieties , see @xcite and references therein .
we expect these are all instances of theorem [ polylem ] .
many authors have looked for a nice compactification of the moduli space @xmath2404 of ( say ) rank two vector bundles with algebraic connection on an algebraic curve @xmath727 . we know of no satisfactory solution .
for example , in @xcite the case of @xmath727 the complement of @xmath2405 points in @xmath1467 is considered , a compactification is constructed , but the boundary is rather nasty ( it lies in @xmath2406 , but this anti - canonical divisor is not reduced ) .
this can be explained as follows : @xmath2404 has a different algebraic structure , the @xmath2407 character variety , @xmath15 ( as complex manifolds they are the same ) .
note @xmath2408 is covered by affine lines ( the space of connections on a fixed bundle is an affine space ) , thus it is not log calabi - yau .
rather , it is the log version of uniruled , and there is no mori theoretic reason to expect a natural compactification .
@xmath15 however is log calabi - yau , and then by mori theory one expects ( infinitely many ) nice compactifications , the minimal models , see @xcite , 1 , for an introduction to these ideas . when @xmath727 has punctures , @xmath15 is a cluster variety , see @xcite and @xcite . in the case of @xmath2409 with @xmath2405 punctures , @xmath15 is the universal family of affine cubic surfaces ( the complement of a triangle of lines on a cubic surface in @xmath2410 ) .
see @xcite , example 6.12 .
each affine cubic has an obvious normal crossing minimal model , the cubic surface .
this compactification is an instance of the above , for a natural choice of polygon @xmath180 .
the same procedure will give a minimal model compactification for any @xmath2411 character variety ( of a punctured riemann surface ) by the above simple procedure that has nothing to do with teichmller theory .
for the remainder of this section we will assume that @xmath1889 has enough global monomials .
by lemma [ egmlem ] , there are global monomials @xmath2412 with @xmath2413 such that @xmath2414 is min - convex with @xmath2415 being compact .
thus we have seeds @xmath2416 ( possibly repeated ) such that @xmath2000 is a character on @xmath2417 , so that @xmath2037 is linear after making the identification @xmath2418 .
furthermore , as in the proof of theorem [ fgprop ] , we can assume @xmath1994 is in the interior of the cone @xmath2035
. we will now observe that with these assumptions the irreducible components of the boundary in the compactification of @xmath1170 induced by @xmath180 are toric .
note for each @xmath1994 there is at least one seed where @xmath2037 is linear .
we assume the collection of @xmath1994 is minimal for defining @xmath180 , and thus @xmath2419 is a union of maximal faces of @xmath180 , a non - empty closed subset of codimension @xmath616 .
writing @xmath2420 , let @xmath2421 be the graded algebra of theorem [ fgprop ] , again a finitely generated algebra .
then @xmath2422 is a projective variety and @xmath2363 gives a cartier ( but not necessarily reduced ) boundary @xmath2423 .
[ toricbth ] in the above situation , the irreducible components of @xmath1654 are projective toric varieties .
more precisely , for each @xmath1994 we have a seed @xmath2036 such that @xmath2000 is a character on @xmath2424 .
then @xmath2425 is a bounded polytope .
the associated projective toric variety is an irreducible component of @xmath1654 , and all irreducible components of @xmath1654 occur in this way . for each @xmath210
consider the vector subspace @xmath2426 with basis @xmath2427 with @xmath2428 and @xmath2429 for @xmath2430 .
note that @xmath2431 is an ideal of @xmath2077 .
indeed , the fact that @xmath1994 lies in the interior of its cone of the cluster complex for @xmath2085 implies by lemma [ blblem ] that @xmath2029 is strictly increasing at bends on monomial decorations of broken lines . now if @xmath2432 , @xmath2433 , and @xmath2434 appears in @xmath133 , then @xmath2435 , and thus @xmath2436
now the definitions imply @xmath2437 .
so it is enough to show that @xmath2438 is the projective toric variety given by the polytope @xmath2439 .
now @xmath2440 has basis @xmath2441 , @xmath2442 . by the multiplication rule , and the fact again that @xmath2029 is strictly increasing at bends on monomial decorations of broken lines ,
the only broken line that contributes to @xmath2443 is the straight broken line , and the multiplication rule on @xmath2444 is given by lattice addition , i.e. , @xmath2438 is the projective toric variety given by the polytope @xmath2445 .
the result is ( at least to us ) surprising in that many cluster varieties come with a natural compactification , where the boundary is not at all toric .
for example , order the columns of a @xmath2446 matrix and consider the open subset @xmath2447 where the @xmath510 consecutive plcker coordinates ( the determinant of the first @xmath460 columns , columns @xmath2448 , etc . ) are non - zero .
this is a cluster variety .
its boundary in the given compactification @xmath2449 is a union of schubert cells ( which are not toric ) .
this has egm by proposition [ egmprop1 ] .
then generic compactifications given by bounded polytopes @xmath180 gives an alternative compactification in which we replace all these schubert cells by toric varieties .
we do not know , e.g. , how to produce such a compactification by birational geometric operations beginning with @xmath2449 .
as discussed in the introduction , many basic objects in representation theory , e.g. , a semi - simple group @xmath93 , are not log calabi - yau , and we can not expect that they have a canonical basis of regular functions .
however , in many cases the basic object is a _ partial minimal model _ of a log calabi - yau variety , i.e. , contains a zariski open log calabi - yau subset whose volume form has a pole along all components of the complement .
for example , the group @xmath93 will be a partial compactification of an open double bruhat cell , and this is a partial minimal model .
we have a canonical basis of functions on the cluster variety , and from this , we conjecture one can get a canonical basis on the partial compactification ( the thing we really care about ) in the most naive possible way , by taking those elements in the basis of functions for the open set which extend to regular functions on the compactification .
we are only able to prove the conjecture under rather strong assumptions , see corollary [ spcor ] .
happily these conditions hold in many important examples .
note that a frozen variable for @xmath19 ( or @xmath56 ) canonically determines a valuation , a point of @xmath2450 , namely the boundary divisor where that variable becomes zero .
see construction [ fvss ] .
while we have myriad ( and near optimal ) sufficient conditions guaranteeing a canonical basis @xmath61 for @xmath33 , we can only prove our conjecture that @xmath2451 is a basis of @xmath2452 under a much stronger condition ( which happily holds in the most important representation theoretic examples ) : [ osdef ] we say a seed @xmath2453 is _ optimized _ for @xmath2454 if @xmath2455 where @xmath2456 is the composition of canonical identifications defined in [ tropsec ] .
if instead @xmath2457 , we say @xmath10 is _ optimized _ for @xmath510 if @xmath2458 for all @xmath560 .
we say @xmath10 is optimized for a frozen index if it is optimized for the corresponding point of @xmath2459 .
for the connection between optimal seed and our conjecture on @xmath2460 see proposition [ boundaryprop ] and conjecture [ nosq ] .
[ quiveroptimized ] in the skew - symmetric case , a seed is optimized for a frozen index if and only if in the quiver for this seed all arrows between unfrozen vertices and the given frozen vertex point towards the given frozen vertex . under the identification @xmath2461 ( which is just @xmath240 in the skew - symmetric case ) ,
the valuation corresponding to the divisor given by the frozen variable indexed by @xmath2462 is simply @xmath1232 .
thus the seed is optimized for this frozen variable if @xmath2463 for all @xmath560 ; this is the number of arrows from @xmath460 to @xmath210 in the quiver , with sign telling us that they are incoming arrows .
[ opslem ] 1 .
the seed @xmath10 is optimized for @xmath2464 if and only if the monomial @xmath2465 on @xmath2466 is a global monomial . in this case @xmath2467 and the global monomial @xmath2465 is the restriction to @xmath2468 of @xmath2469 . in the @xmath56 case , for @xmath2470 primitive , this holds if and only if each of the initial scattering monomials @xmath2471 in @xmath806 is regular along the boundary divisor of @xmath56 corresponding to @xmath510 under the identification @xmath2472 .
@xmath2454 has an optimized seed if and only if @xmath510 lies in @xmath2473 . for ( 1 )
, the rays for the fan @xmath2474 giving the toric model for @xmath795 are @xmath2475 for @xmath560 .
note that @xmath2476 , see .
now the statement concerning @xmath19 follows from lemma [ gmlemma ] and lemma [ fgccthg2 ] .
the additional statement in the @xmath56 case is clear from the definitions . for ( 2 )
, one notes that the forward implication is given by ( 1 ) , while for the converse , if @xmath2477 , then @xmath2478 for some seed @xmath10 , and then @xmath510 is optimized for that seed . [ osprop ] for the standard cluster algebra structure on @xmath2479 ( the affine cone over @xmath2449 in its plcker embedding ) every frozen variable has an optimized seed . as was pointed out to us by lauren willams , for @xmath2449 , the initial seed in @xcite , figure 4.4 ,
is optimized for one frozen variable ( the _ special _ upper right hand vertex for the initial quiver ) .
the result follows from the cyclic symmetry of this cluster structure .
[ osrem ] b. leclerc , and independently l. shen , gave us an explicit sequence of mutations that shows the proposition holds as well for the cluster structure of @xcite , @xcite on the maximal unipotent subgroup @xmath2480 , and the same argument applies to the fock - goncharov cluster structure on @xmath2481 , @xmath220 .
the argument appears in @xcite .
[ linindlemma ] let @xmath11 be a lattice and @xmath2482 a submonoid with @xmath2483 .
for any subset @xmath2484 and collection of elements @xmath2485 such that @xmath2486 $ ] , the subset @xmath2487 $ ] is linearly independent over @xmath8 .
suppose @xmath2488 for @xmath2489 all non - zero and @xmath2490 a finite set .
let @xmath2491 be minimal with respect to the partial ordering on @xmath11 given by @xmath265 ( where @xmath2492 means @xmath2493 for some @xmath2494 ) .
the coefficient of @xmath2495 in the sum , expressed in the basis of monomials , must be zero .
but the minimality of @xmath2496 implies the monomial @xmath2495 does not appear in any of the @xmath2497 , @xmath2498 .
thus the coefficient of @xmath2495 is just @xmath2499 , a contradiction .
[ boundaryprop ] suppose a valuation @xmath2500 has an optimized seed . if @xmath2501 , then @xmath2502 for all @xmath908 with @xmath2503 .
let @xmath74 be optimized for @xmath697 .
let @xmath1336 be the strictly convex cone spanned by the exponents of the initial scattering monomials @xmath2504 .
let @xmath2505 .
take @xmath827 a basepoint in the positive chamber of @xmath378 . by definition @xmath2506 where @xmath2507 is a finite sum of monomials . by ( 1 ) of lemma [ opslem ] we have @xmath2508 , and thus by ( 2 ) of lemma [ lplemma ] , @xmath2509 .
let @xmath51 be the minimum of @xmath2510 over all @xmath908 with @xmath2503 , and suppose @xmath2511 .
since @xmath2512 , necessarily @xmath2513.\ ] ] note this is the sum of all the monomial terms in @xmath2514 which have the maximal order of pole , @xmath2515 , along @xmath697 .
this contradicts lemma [ linindlemma ] .
we believe the assumption of an optimized seed is not necessary : [ nosq ] the proposition holds for any @xmath2516 . any finite set @xmath2517 of primitive elements gives a partial compactification ( defined canonically up to codimension two ) @xmath2518 , with the boundary divisors of this partial compactification in one - to - one correspondence with the elements of @xmath1589 ( this is true for any finite collection , @xmath1589 , of divisorial discrete valuations on the function field of a normal variety @xmath19 : there is always an open immersion @xmath2519 , with divisorial boundary @xmath2520 corresponding to @xmath1589 , and @xmath2519 is unique up to changes in codimension greater than or equal to two ) .
we then define @xmath2521 and @xmath2522 the vector subspace with basis @xmath2523 .
similarly we define @xmath2524 to be the subalgebra of @xmath2525 generated by those cluster variables that are regular ( generically ) along all @xmath2526 .
[ spdef ] each choice of seed @xmath10 gives a pairing @xmath2527 which is just the dual pairing composed with the identifications @xmath2528 [ midocalem ] 1 .
@xmath2522 is a subalgebra containing @xmath2524 . if @xmath2529 then @xmath2530 2 .
assume each @xmath2531 has an optimized seed .
then @xmath2532 if @xmath1785 then @xmath2533 .
if each @xmath2531 has an optimized seed and @xmath10 is optimized for @xmath2526 , the piecewise linear function @xmath2534 is min - convex , and for all @xmath2535 , @xmath2536 where @xmath2537 is the global monomial on @xmath2079 corresponding to @xmath2538 ( which exists by lemma [ opslem ] ) .
there are pairings @xmath2539 which are much more natural then definition [ spdef ] .
indeed , @xmath2540 gives a canonical function @xmath2541 and , since @xmath2542 , a valuation on @xmath2543 .
the analogous statements apply to @xmath2544 .
so we could define a pairing by either @xmath2545 we conjecture these two pairings are equal .
lemma [ midocalem ] , ( 3 ) , gives the result when one of @xmath2546 lies in the cluster complex .
one can pose the same symmetry conjecture for mirror pairs of affine log cys ( with maximal boundary ) in general , the two - dimensional case having been shown in @xcite .
suppose the symmetry conjecture holds , and furthermore @xmath2547 .
then ( see the proof of lemma [ midocalem ] below ) the cone of cut out by the tropicalisation of the potential function is @xmath2548 if furthermore conjecture [ nosq ] holds and @xmath1785 , then @xmath180 gives a basis of @xmath2525 , canonically determined by the open set @xmath2549 ( together with its cluster structure , though we conjecture the basis is independent of the cluster structure ) .
see corollary [ spcor ] .
the subalgebra statement of ( 1 ) follows from the positivity ( both of structure constants and the laurent polynomials @xmath1810 ) just as in the proof of theorem [ goodalgebra ] .
every cluster variable is a theta function , so the inclusion @xmath2550 is clear .
now obviously if @xmath2529 then both are equal to @xmath86 .
the intersection expression of ( 2 ) for the middle algebra follows from proposition [ boundaryprop ] .
now obviously if @xmath1785 then @xmath2533 . for ( 3 ) , we work with the scattering diagram @xmath378
. then @xmath2538 is the @xmath41-vector of the global monomial @xmath2537 , with @xmath2551 , by lemma [ opslem ] . using @xmath2552
, one sees that @xmath2553 is linear on @xmath2554 , so obviously min - convex in the sense of definition - lemma [ mcdef ] . since it is the tropicalisation of a global monomial it is also min - convex in the sense of definition [ imcdef ] , by proposition [ convcor ] .
now fix a base point @xmath2555 , and consider @xmath1810 , @xmath2556 . by lemma [ opslem ] , ( 1 ) , each scattering function is regular along the boundary divisor corresponding to @xmath2557 . by definition @xmath2558 , where @xmath2559 is a linear combination of monomials @xmath2560 with @xmath2495 regular along the boundary divisor corresponding to @xmath697
thus @xmath2561 by lemma [ lplemma ] .
since @xmath2537 is the monomial @xmath2562 on @xmath2563 , @xmath2564 this completes the proof of ( 3 ) .
recall a choice of seed gives a partial compactification @xmath1125 and a map @xmath2565 .
the boundary @xmath2566 has @xmath510 irreducible components , primitive elements of @xmath2567 , the vanishing loci of the @xmath1115 .
[ gsoplem ] the seed @xmath10 is optimized for each of the boundary divisors of @xmath1087 . if @xmath787 , the corresponding seed for @xmath56 is @xmath2568 and the boundary divisors correspond to the @xmath2569 .
but @xmath2570 hence the claim .
we adjust slightly the notation @xmath2571 of the previous subsection to this case : let @xmath2572 be the subset of points @xmath908 such that @xmath46 remains regular on the partial compactification @xmath2573 , i.e. , such that @xmath2574 [ productlem ] under the identification @xmath2575 , we have @xmath2576 and @xmath2577 . @xmath61 is invariant under translation by @xmath2578 , and thus @xmath2576 . by lemma [ clem ]
we construct @xmath2579 from the atlas of toric compactifications @xmath2580 parameterized by the cluster chambers in @xmath792 .
now take @xmath2556 , and consider @xmath1810 for some basepoint in the cluster complex .
this is a positive sum of monomials , so it will be regular on the boundary of @xmath2579 iff each summand is .
one summand is @xmath1374 , so if @xmath46 is regular on @xmath2581 then @xmath2582 .
thus @xmath2583 . but
now suppose @xmath2584 , some @xmath2585 and @xmath2586 .
then @xmath1374 is regular on the boundary .
since the initial scattering monomials are @xmath1397 , any bend in a broken line multiplies the decorating monomial by a monomial regular on the boundary .
thus @xmath2587 .
this completes the proof .
we define @xmath2588 recall @xmath2589 are the cluster and upper cluster algebras with principal coefficients respectively , with the frozen variables inverted .
on the other hand , @xmath2590 are the cluster and upper cluster algebras with principal coefficients respectively , with the frozen variables not inverted . by lemma [ midocalem ] , @xmath2591 is a subalgebra , and @xmath2592 . by lemma [ midocalem ] and lemma [ gsoplem ]
, we have [ improp ] if @xmath1785 then @xmath2593 .
here is another sufficient condition for the full fock - goncharov conjecture to hold , which will prove immediately useful below : [ scprop ] suppose there is a min - convex function @xmath2594 , such that @xmath2595 implies @xmath1618 , and such that @xmath2595 for some @xmath123 .
suppose also that there is a bounded positive polytope in @xmath1424 ( which holds for example if @xmath2079 has enough global monomials )
. then @xmath2596 .
take any @xmath2597 and @xmath1557 with @xmath2598 .
then consider any @xmath2434 appearing in @xmath2599 , for @xmath2600 . by lemma [ diplem1 ] , @xmath2601 for @xmath860 sufficiently large .
in particular @xmath2602 , so for each @xmath2434 that appears , @xmath1615 is a universal positive laurent polynomial , for any basepoint @xmath827 in the cluster complex .
the existence of the bounded positive polytope implies @xmath2599 is a finite sum of @xmath2434 .
thus the product @xmath2603 is also a universal positive laurent polynomial , and thus by the positivity of the scattering diagram , @xmath1617 must be a finite positive laurent polynomial .
thus @xmath1618 .
if there are frozen variables , there is a canonical candidate for @xmath153 in the proposition .
the cluster algebras related to double bruhat cells are of this sort , and we hope that this will give a way of completing the proof of the full fock - goncharov conjecture in these cases . when we have frozen variables , this gives a partial compactification @xmath142 . in this case , let us change notation slightly and write a seed @xmath10 as @xmath2604 with @xmath2605 and @xmath2606 , and the @xmath1146 are frozen . in this case
the elements @xmath2607 give @xmath2608 canonical boundary divisors for a partial compactification @xmath142 , and an analogous @xmath2609 .
an atlas for @xmath2609 is given by gluing the partial compactification @xmath2610 , where @xmath2474 is the fan consisting of the rays @xmath2611 .
[ spcor ] assume that for each @xmath2612 , @xmath2613 has an optimized seed , @xmath2614 .
let @xmath2615 be the _ ( landau - ginzburg ) potential _ , the sum of the corresponding global monomials on @xmath2079 given by lemma [ opslem ] .
then : 1 .
the piecewise linear function @xmath2616 is min - convex and @xmath2617 is a positive polytope .
@xmath180 has the alternative description : @xmath2618 3 .
the set @xmath2619 parameterizes a canonical basis of @xmath214 this is immediate from lemma [ midocalem ] .
[ ffgcor ] assume we have enough global monomials on @xmath186 , and every frozen variable has an optimized seed .
let @xmath223 and @xmath180 be as in corollary [ spcor ] . if for some seed @xmath10 , @xmath180 is contained in the convex hull @xmath2620 of @xmath61 ( which itself contains the integral points of the cluster complex @xmath110 ) then @xmath168 , @xmath216 is finitely generated , and the integer points @xmath2621 parameterize a canonical basis of @xmath2622 . by definition @xmath2623 , and @xmath211 is min - convex by lemma [ opslem ] .
thus @xmath2624 by proposition [ scprop ] .
now the result follows from the inclusions @xmath2625 of corollary [ markscor ] the corollary applies in important representation theoretic examples : the hyptheses of theorem [ spth ] are proven in @xcite , using proposition [ scprop ] applied to the tropicalisation of our potential @xmath223 .
the agreement of @xmath223 with the berenstein - kazhdan potential is given in @xcite .
theorem [ spth ] is stated for @xmath56 .
but in this case it is shown in @xcite that the exchange matrix has full rank , i.e. the equivalent conditions of lemma [ taprinlem ] hold .
now the results for @xmath56 imply the analogous result for @xmath19 using @xmath1634 equivariance , as in the proof of ( 7 ) of theorem [ mainthax ] .
the @xmath100 action is identified with the action of @xmath2626 on @xmath795 , the various statements about @xmath100-weights now follow immediately from the equivariance , proposition [ thetaeigenfunctions ] .
we check the conditions of theorem [ spth ] : the existence of an optimized seed is proven in @xcite , following suggestions of l. shen and b. leclerc .
the cluster variety has large cluster complex ( definition [ lccdef ] ) by @xcite , theorems 1.12 and 1.17 .
this gives the hypotheses of theorem [ spth ] , ( 3 ) .
the equality of our @xmath223 with the goncharov - shen potential is given in @xcite .
it is shown in @xcite that the exchange matrix has full rank , in the sense of lemma [ taprinlem ] . now
the @xmath56 results imply the analogous statements for @xmath56 as in the proof directly above .
the @xmath229 action is identified with the @xmath2626 action , which gives the weight statements as above .
we first review basic cluster variety notation as adopted in @xcite .
none of this is original to @xcite , but we follow that source for consistency of notation . as in @xcite ,
2 , _ fixed data _ @xmath239 means * a lattice @xmath240 with a skew - symmetric bilinear form @xmath2627 * an _ unfrozen sublattice _
@xmath2628 , a saturated sublattice of @xmath240 .
if @xmath2629 , we say the fixed data has no frozen variables .
* an index set @xmath844 with @xmath248 and a subset @xmath249 with @xmath250 .
* positive integers @xmath251 for @xmath252 with greatest common divisor @xmath616 . * a sublattice @xmath2630 of finite index such that @xmath2631 , @xmath2632 .
* @xmath241 , @xmath246 .
here we modify the definition slightly , and include in the _ fixed data _
@xmath2633 $ ] a mutation class of seed .
recall a seed @xmath74 is a basis of @xmath240 satisfying certain properties , see @xcite , 2 , for the precise definitions , including that of mutation .
in particular , we write @xmath1099 for the dual basis and @xmath1229 . we write @xmath2634 we shall also assume that if @xmath263 , then the linear functional @xmath75 is non - zero .
( if this happens , then one can view @xmath1232 as frozen . )
we have two natural maps defined by @xmath798 : @xmath2635 we also choose a map @xmath2636 such that ( a ) @xmath2276 and ( b ) the composed map @xmath2637 agrees with @xmath2638 . different choices of @xmath2265 differ by a choice of map @xmath2639 .
further , if there are no frozen variables , i.e. , @xmath2640 , then @xmath2641 is canonically defined .
we also define @xmath2642 following our conventions in @xcite , let @xmath695 be the infinite oriented rooted tree with @xmath2643 outgoing edges from each vertex , labelled by the elements of @xmath2644 .
let @xmath697 be the root of the tree .
attach some choice of initial seed @xmath2645 $ ] to the vertex @xmath697 .
( we write @xmath696 if we want to record this choice of initial seed . )
now each simple path starting at @xmath697 determines a sequence of mutations , just mutating at the label attached to the edge . in this way we attach a seed to each vertex of @xmath695 .
we write the seed attached to a vertex @xmath153 as @xmath1235 , and write @xmath2646 etc . for the corresponding tori .
mutations define birational maps between these tori , and the associated fock - goncharov @xmath111 , @xmath987 cluster varieties are defined by @xmath2647 this parameterization of torus charts is very redundant , with infinitely many copies of the same chart appearing .
in particular , given a vertex @xmath153 of @xmath695 , one can consider the subtree @xmath2648 rooted at @xmath153 , with initial seed @xmath1235 .
this tree can similarly be used to define @xmath2649 , and the obvious inclusion between these two versions of @xmath2649 is in fact an isomorphism , as can be easily checked . as one expects the mirror of a variety obtained by gluing charts of the form @xmath2650 to be obtained by gluing charts of the form @xmath1076
, the mirror of @xmath19 is not @xmath27 , as the latter is obtained by gluing charts of the form @xmath722 . to get the correct mirrors of @xmath19 and @xmath27 , one follows @xcite in defining the _ langlands dual _
cluster varieties .
this is done by , given fixed data @xmath239 , defining fixed data @xmath2651 to be the fixed data : @xmath2652 where @xmath2653 the lattice , with its finite index sublattice , is @xmath2654 and the @xmath2655-valued skew - symmetric form on @xmath2656 is @xmath2657 for each @xmath2658 $ ] , we define @xmath2659 one checks easily that @xmath2660 gives a bijection between @xmath2633 $ ] and @xmath2661 $ ] .
note that for skew - symmetric cluster algebras , i.e. , when all the multipliers @xmath2662 , langlands duality is the identity , @xmath2663 .
[ fgddef ] we write @xmath2664 and @xmath2665 . note in the skew - symmetric case , that @xmath2666 .
one observes the elementary given fixed data @xmath239 , the double langlands dual data @xmath2667 is canonically isomorphic to the data @xmath239 via the map @xmath2668 given by @xmath2669 .
we recall briefly the construction of principal fixed data from @xcite , construction 2.11 . for fixed data @xmath239 , the data for the cluster variety with principal coefficients @xmath943
is defined by : * @xmath2670 with the skew - symmetric bilinear form @xmath2671 * @xmath2672 * the sublattice @xmath2673 is @xmath2674 . *
the index set @xmath844 is now the disjoint union of two copies of @xmath844 , with the @xmath251 taken to be as in @xmath239 .
the set of unfrozen indices @xmath2644 is just the original @xmath2644 thought of as a subset of the first copy of @xmath844 . * given an initial seed @xmath787 , we define @xmath2675 we then take the mutation class @xmath2676 $ ] . note
that @xmath2676 $ ] depends on the choice of @xmath10 : it is not true that if @xmath609 is obtained by mutation from @xmath10 then @xmath2677 is obtained from the same set of mutations applied to @xmath1086
. nevertheless , the cluster varieties @xmath2678 are defined independently of the seed @xmath10 .
this is a very important point , which we shall revisit in remark [ initialseedindremark ] .
the following summarizes all of the important relationships between the various varieties which will be made use of in this paper .
[ ldpprop ] giving fixed data @xmath239 , we have : 1 .
there is a commutative diagram where the dotted arrows are only present if there are no frozen variables ( i.e. , @xmath2629 ) : @xmath2679\ar[d]&{\mathcal a}_{{\operatorname{prin}}}\ar@/^24pt/[rr]^p\ar[r]^{\tilde p}\ar[d]_{\pi } & \cx\ar[d]_{\lambda}&\cx_{{\operatorname{prin}}}\ar[l]_{\rho}\ar[d]_w&{\mathcal a}\ar[l]_{\xi}\ar[d]\\ t\ar[r]&t_m\ar@{-->}[r]&t_{k^*}&\ar@{-->}[l]t_m&\ar[l ] e } \ ] ] with @xmath2680 any point , @xmath2681 the identity , and with the left- and right - hand squares cartesian and @xmath123 an isomorphism , canonical if there are no frozen variables .
2 . there are torus actions @xmath2682 here @xmath2683 is the kernel of the map @xmath2684 furthermore @xmath1076 and @xmath2685 act on @xmath2357 so that the map @xmath2686 is @xmath1076- and @xmath2685-equivariant .
the map @xmath2687 is a @xmath1076-torsor .
there is a map @xmath2688 such that the map @xmath1707 is also compatible with the actions of these two tori on @xmath56 and @xmath27 respectively , so that @xmath2689 is a @xmath2685-torsor .
@xmath2690 and @xmath2691 are isomorphic data , so we can define @xmath2692 4 .
there is a commutative diagram @xmath2693\ar[d]&\cx_{{\operatorname{prin}}}^{\vee}\ar@/^24pt/[rr]^p\ar[r]^{\tilde p } \ar[d]_{\pi } & { \mathcal a}^{\vee}\ar[d]_{\lambda}&{\mathcal a}^{\vee}_{{\operatorname{prin } } } \ar[l]_{\rho}\ar[d]_w&\cx^{\vee}\ar[l]_{\xi}\ar[d]\\ e\ar[r]&t_{m^{\circ}}\ar@{-->}[r]&t_{(k^{\circ})^*}&\ar@{-->}[l]t_{m^{\circ}}&\ar[l ] e } \ ] ] we consider the diagram of ( 1 ) .
the maps with names are given as follows on cocharacter lattices : @xmath2694 note @xmath2695 is the transpose of the inclusion @xmath2696 . in the case
there are no frozen variables , the two dotted horizontal lines are just given on cocharacter lattices by @xmath2695 again .
one checks commutativity from these formulas at the level of individual tori , and one checks the maps are compatible with mutations .
note the left - hand diagram defines @xmath78 , see @xcite , definition 2.12 .
the statements that @xmath1707 , @xmath1110 and @xmath2695 are compatible with mutations are in 2 of @xcite , as well as the commutativity of the second square in case of no frozen variables .
it is clear that @xmath123 induces an isomorphism of lattices , hence an isomorphism of the relevant tori .
this isomorphism is canonical in the no frozen variable case because @xmath2265 is well - defined in this case .
the fact the right - hand square is cartesian follows from the fact that @xmath2697 .
note the signs in the definition of @xmath1686 are necessary to be compatible with mutations .
this gives ( 1 ) . for ( 2 ) ,
the first action is specified on the level of cocharacter lattices by @xmath2698 while the last three are given by the inclusions @xmath2699 one checks easily that the induced actions are compatible with mutations .
the action of @xmath1076 and @xmath2685 on @xmath2357 are induced by the maps @xmath2700 and @xmath2701 respectively , in order to achieve the desired equivariance .
the map @xmath2702 is given by @xmath2703 the other statements are easily checked . for ( 3 ) ,
from the definitions , the lattices playing the role of @xmath2630 are : @xmath2704 these are isomorphic under the map @xmath2705 .
furthermore , the pairings in the two cases are given by @xmath2706 respectively .
the isomorphism given preserves the pairings , hence the isomorphism .
\(4 ) is the same as ( 1 ) , but for the langlands dual data @xmath2003 . for reference ,
the maps are given as follows : @xmath2707 whenever the lattice @xmath2708 appears in dealing with the langlands dual data , we will always identify this with @xmath240 in the obvious way .
simple linear algebra gives : [ eglem ] the choice of the map @xmath2265 gives an inclusion @xmath2709 ( see proposition [ ldpprop ] , ( 2 ) ) given by @xmath2710 .
we also have @xmath2275 ( a sublattice of @xmath1444 ) included in @xmath2683 via @xmath2711 .
these inclusions induce an isomorphism @xmath2712 .
[ taprinlem ] the map @xmath2713 induced by the composition of the inclusion and projection @xmath2714 is a split surjection if and only if the map @xmath2715 is surjective .
this holds if and only if in some seed @xmath2716 , the @xmath2717 matrix with entries for @xmath263 , @xmath2718 , @xmath2719 gives a surjective map @xmath2720 . in this case
@xmath21 is isomorphic to the trivial bundle @xmath2721 .
for the first statement , note using lemma [ eglem ] that the map @xmath2722 is surjective if and only if the map @xmath2723 given by @xmath2724 is surjective , and this is the case if and only if the induced map @xmath2725 is surjective .
the given matrix is the matrix for @xmath2726 in the given bases , so the second equivalence is clear .
the final statement follows from the @xmath1782-equivariance of @xmath1110 ( the trivialization then comes by choosing a splitting of @xmath2727 ) .
[ initialseedindremark ] in general , a seed is defined to be a basis of the lattice @xmath240 ( or @xmath1144 ) , but to define the seed mutations @xcite , ( 2.2 ) and the union of tori , all one needs are elements @xmath2728 , @xmath76 ( the definitions as given make sense even if the @xmath1232 are dependent , or fail to span ) .
if one makes the construction in this greater generality , the characters @xmath1273 on @xmath2729 will not be independent ( if the @xmath1232 are not ) and unless we take a full basis , we can not define the cluster variables @xmath2730 on @xmath2731 , as the @xmath2250 are defined as the dual basis to the basis @xmath2732 for @xmath2733 . in the case of the principal data , given a seed @xmath787 for @xmath239 , we get a seed @xmath2734 in this modified sense for the data @xmath943 .
we also write this seed as @xmath10 . on the other hand , in @xcite , the seed @xmath1086 for @xmath943
is defined in the more traditional sense to be the basis @xmath2735 .
it is not the case that if @xmath609 is obtained from @xmath10 via a sequence of mutations , then @xmath2677 is obtained from @xmath1086 by the same sequence of mutations .
in particular , the set @xmath2676 $ ] of seeds mutation equivalent to @xmath1086 depends not just on the mutation equivalence class of @xmath10 , but on the original seed @xmath10 .
however , using the seed @xmath10 as a seed for @xmath943 in this modified sense , we can build @xmath1170 , and this depends only on the mutation class of @xmath10 .
thus @xmath56 does not depend on the initial choice of seed , but only on its mutation equivalence class .
however , as we shall now see , the choice of initial seed does give a partial compactification .
this is a more general phenomenon when there are frozen variables .
[ fvss ] when the cluster data @xmath239 includes frozen variables , @xmath19 comes with a canonical partial compactification @xmath142 , given by partially compactifying each torus chart via @xmath2736 , where for @xmath2114 , @xmath2737 .
thus the dual cone @xmath2738 is cut out by the half - spaces @xmath2739 , @xmath1066 .
note that the monomials @xmath2730 , @xmath1066 are invariant under mutation . these give a canonical map @xmath2740 , where @xmath2741 is the number of unfrozen variables .
note that the basis elements @xmath1232 for @xmath2742 , though they have frozen indices , can change under mutation .
what is invariant is the associated boundary divisor with valuation given by @xmath2743 .
these are the boundary divisors of @xmath142 .
we remark that like @xmath19 , @xmath196 is also separated , with the argument given in @xcite , theorem 3.14 working equally well for @xmath196 .
here is another way of seeing the same thing . given any cluster variety @xmath2744 and a single fan @xmath2745 for a toric partial compactification @xmath2746 for some @xmath2747 , there is a canonical way to build a partial compactification @xmath2748 we let @xmath2749 and @xmath2750 , where @xmath2751 is the birational map given by the composition @xmath2752 and @xmath2753 is the geometric tropicalisation , see [ tropsec ] .
[ aprinremark ] we now return to the discussion of @xmath56 .
note that the frozen variables for @xmath56 are indexed by @xmath2754 in the first copy of @xmath844 , along with all indices in the second copy of @xmath844 .
however , we can apply construction [ fvss ] taking only the second copy of @xmath844 as the set of frozen indices , with the initial choice of seed @xmath10 determining a partial compactification of @xmath56 .
in this case , we indicate the partial compactification by @xmath1087 .
it is important to keep in mind the dependence on @xmath10 .
fixing @xmath10 fixes @xmath1086 , and hence cluster variables @xmath2330 , @xmath2331 .
the variables @xmath1115 can then take the value @xmath81 in the compactification .
in particular , we obtain an extension of @xmath21 to @xmath1074 , @xmath1273 pulling back to @xmath2331 .
note that the seeds in @xmath2633 $ ] and @xmath2755 $ ] are in one - to - one correspondence . given any seed @xmath2756
$ ] , and seed @xmath2757 $ ] obtained via the same sequence of mutations we have @xmath2758 for some @xmath2759 .
these two seeds give rise to coordinates @xmath1240 on the chart of @xmath19 indexed by @xmath609 and coordinates @xmath2760 on the chart of @xmath56 indexed by @xmath2677 . as @xmath19 is the fibre of @xmath1110 over the point of @xmath1136 with all coordinates @xmath616 , the coordinate @xmath1240 on the chart of @xmath56 restricts to the coordinate @xmath1240 on the chart of @xmath19 .
this gives a one - to - one correspondence between cluster variables on @xmath19 and @xmath1174-type cluster variables on @xmath56 .
to summarize : [ cvextrem ] the cluster variety @xmath2761 depends only on the mutation class @xmath2633 $ ] .
but the choice of a seed @xmath10 determines : 1 . a partial compactification @xmath2762 ; 2 . the canonical extension of each cluster variable on any chart of @xmath19 to a cluster variable on the corresponding chart of @xmath2763 .
this appendix is devoted to giving proofs of theorems [ kslemma2 ] and [ scatdiagpositive ] .
the proof of [ kslemma2 ] is essentially given in @xcite , but the special case here is considerably simpler than the general case covered there , and it is likely to be very difficult for the reader to extract the needed results from @xcite .
in addition , the details of the proof of [ kslemma2 ] will be helpful in proving theorem [ scatdiagpositive ] .
[ kslemalg ] there is a simple order by order algorithm , introduced in @xcite in the two - dimensional case and in @xcite in the higher dimensional case , for producing the diagram @xmath2764 of theorem [ kslemma ] , which we will describe shortly after a bit of preparation .
this is useful both from a computational point of view and because a more complicated version of this will be necessary in the remainder of this appendix .
we first introduce some additional terminology . for any scattering diagram @xmath328 for @xmath2765 , and any @xmath1552 we let @xmath2766 be the ( by definition , finite ) set of @xmath320 with @xmath331 non - trivial in @xmath332 . a scattering diagram for @xmath319
, @xmath283 induces a scattering diagram for @xmath319 , @xmath338 in the obvious way , viewing @xmath2767 for a wall @xmath320 .
we say two scattering diagrams @xmath328 , @xmath363 are equivalent to order @xmath460 if they are equivalent as scattering diagrams for @xmath338 .
[ pjdef ] let @xmath491 be a joint of the scattering diagram @xmath2768 .
either every wall containing @xmath491 has direction tangent to @xmath491 ( where the direction of a wall contained in @xmath495 is @xmath2769 ) , or every wall containing @xmath491 has direction _ not _ tangent to @xmath491 . in the first case we call the joint _
parallel _ and in the second case _ perpendicular_.
suppose @xmath491 spans the subspace @xmath2770 .
then the direction of any wall containing @xmath491 is of the form @xmath2771 for some @xmath2772 .
if this is tangent to @xmath491 , then @xmath2773 for @xmath514 , and hence @xmath2774 . from this
it follows that @xmath2775 for all @xmath2776 , and hence the direction of any wall containing @xmath491 is tangent to @xmath491 .
a joint @xmath491 is a codimension two convex rational polyhedral cone .
let @xmath2777 be the set of integral tangent vectors to @xmath491 .
this is a saturated sublattice of @xmath113 .
then we set @xmath2778 this is closed under lie bracket . if @xmath491 is a parallel joint , then @xmath2779 is abelian , since if @xmath2780 with @xmath2781 , @xmath2782 , so @xmath516=0 $ ] .
we denote by @xmath2783 the corresponding group .
we will build a sequence of finite scattering diagrams @xmath2784 , with the property that @xmath2785 is equivalent to @xmath328 to order @xmath460 .
taking @xmath2786 , we obtain @xmath2787 equivalent to @xmath328 .
let @xmath2788 denote the subset of @xmath940 consisting of walls which are non - trivial in @xmath332 .
we start with @xmath2789 if @xmath491 is a joint of a finite scattering diagram , we write @xmath490 for a simple loop around @xmath491 small enough so that it only intersects walls containing @xmath491 . in particular , for each joint @xmath491 of @xmath2790 , @xmath2791 .
indeed , @xmath2792 is abelian and by the form given for @xmath940 in the statement of theorem [ kslemma ] , all walls containing @xmath491 are hyperplanes . thus the automorphism associated to crossing each wall and its inverse occurs once in @xmath2793 , and hence cancel .
now suppose we have constructed @xmath2794 . for every perpendicular joint @xmath491 of @xmath2794
, we can write uniquely in @xmath2795 @xmath2796 where @xmath2797 and @xmath2798 . such an expression holds because all wall - crossing automorphisms for walls containing @xmath491 lie in @xmath2799 , so that @xmath2800 can be viewed as an element of @xmath2795 .
furthermore , by the inductive hypothesis ,
this element is trivial in @xmath2801 .
because @xmath491 is perpendicular , we never have @xmath2802 .
now define @xmath2803:=\{\big({\mathfrak{j}}-{\mathbb{r}}_{\ge 0}p^*(\alpha),\exp(\pm g_{\alpha})\big ) \,|\ , \alpha \in s\},\ ] ] where the sign is chosen so that the contribution to crossing the wall indexed by @xmath59 in @xmath2804}$ ] is @xmath2805 .
note the latter element is central in @xmath2806 .
thus @xmath2804}={{\mathfrak{p}}}_{\gamma_{{\mathfrak{j } } } , \tilde{\mathfrak{d}}_k}^{-1}$ ] and @xmath2807 } = { { \mathfrak{p}}}_{\gamma_{{\mathfrak{j}}},\tilde{\mathfrak{d}}_k}\circ{{\mathfrak{p}}}_{\gamma_{{\mathfrak{j}}},{\mathfrak{d}}[{\mathfrak{j } } ] } = { \operatorname{id}}\ ] ] in @xmath2806 .
consider a perpendicular joint @xmath491 of @xmath2809 . if @xmath491 is contained in a joint @xmath2811 of @xmath2794
, @xmath2811 is the unique such joint , and we constructed @xmath2812 $ ] above . if @xmath491 is not contained in a joint of @xmath2794 , we define @xmath2812 $ ] to be the empty set .
there are three types of walls @xmath613 in @xmath2809 containing @xmath491 : 1 .
@xmath2813 $ ] .
2 . @xmath2814)$ ] , but @xmath2815 .
this type of wall does not contribute to @xmath2816 , as the associated automorphism is central in @xmath2806 , and in addition this wall contributes twice to @xmath2817 , with the two contributions inverse to each other .
3 . @xmath2818)$ ] and @xmath2819 . since each added wall is of the form @xmath2820 for some joint @xmath2821 of @xmath2794 , where @xmath866 is the direction of the wall , the direction of the wall is parallel to @xmath491 , contradicting @xmath491 being a perpendicular joint .
thus this does not occur . 1
. @xmath328 and @xmath2787 are equivalent to order @xmath460 .
@xmath328 is consistent to order @xmath2810 .
@xmath2824 is the identity for every perpendicular joint @xmath491 of @xmath2787 to order @xmath2810 .
4 . @xmath328 and @xmath2787 have the same set of incoming walls .
we work with scattering diagrams in the group @xmath2806 .
there is a finite scattering diagram @xmath363 with the following properties : ( 1 ) @xmath2825 is equivalent to @xmath328 ; ( 2 ) @xmath363 consists only of walls trivial to order @xmath460 but non - trivial to order @xmath2810 .
indeed , @xmath363 can be chosen so that @xmath2826 for any general point @xmath367 in any @xmath495 , @xmath463 .
note that @xmath363 is finite because the same is true of @xmath328 and @xmath2787 .
thus to show @xmath328 and @xmath2787 are equivalent , it is sufficient to show that @xmath363 is equivalent to the empty scattering diagram .
to do so , replace @xmath363 with an equivalent scattering diagram with minimal support .
let @xmath491 be a perpendicular joint of @xmath2827 . then in @xmath2806 , @xmath2828 , since @xmath2829 and automorphisms in @xmath363 are central in @xmath2806
however , this implies that for each @xmath291 with @xmath2830 and @xmath2831 two points in @xmath357 on either side of @xmath491 , the automorphisms associated with crossing @xmath357 in @xmath363 through either @xmath367 or @xmath492 must be the same in order for these two automorphisms to cancel in @xmath2832 . from this
it is easy to see that @xmath363 is equivalent to a scattering diagram such that for every wall @xmath2833 , each facet of @xmath613 is a parallel joint of @xmath363 , i.e. , the direction @xmath2834 is tangent to every facet of @xmath613 .
however , such a wall must be incoming , contradicting , if @xmath363 is non - empty , the fact that @xmath2787 and @xmath328 have the same set of incoming walls by assumption .
we fix the notation of theorem [ kslemma2 ] , and in addition make use of the notation @xmath2835 of definition [ halfspacedef ] and @xmath605 as in the map associated to crossing the slab @xmath2836 from @xmath673 to @xmath674 .
we define the lie algebra @xmath2837 and set @xmath2838 , @xmath2839 as usual , with the degree function @xmath2840 given by @xmath2841 .
we note that @xmath2842 acts on @xmath604}$ ] as usual , and if @xmath328 is a scattering diagram in the sense of definition [ escdef ] , then all automorphisms associated to crossing walls ( rather than slabs ) lie in @xmath2842 . besides the lie algebra @xmath2843
just defined , recall we also have @xmath2844 as usual .
we have the degree map @xmath2845 given by @xmath2846 , but we also have @xmath2847 given by the restriction of @xmath2840 .
we use the notation @xmath2848 and @xmath2849 to distinguish between the two possibilities for @xmath2850 determined by the two choices of degree map . then @xmath2851 and we define @xmath2852 . note that @xmath2853 both act faithfully on @xmath268}$ ] , where the completion is respect to the maximal monomial ideal @xmath2854 , and @xmath2855 act faithfully on @xmath604}$ ] .
there are inclusions @xmath2856 and @xmath2857 .
only the second inclusion holds at finite order , i.e. , @xmath2858 . for each of the above lie algebras @xmath337 , we can now also talk about scattering diagrams for @xmath337 using definitions [ walldef ] and [ ksscatdiagdef ] , replacing @xmath283 with @xmath337 in those definitions .
we need to understand the interaction between elements of @xmath93 and the automorphism associated to crossing the slab ( see lemma 2.15 of @xcite ) .
recall the notation @xmath2799 from construction [ kslemalg ] ; this is applied also to the various assorted groups above .
[ conjugationlemma ] let @xmath2865 ( resp .
@xmath319 ) and let @xmath2866 ( resp .
@xmath2867 ) be an automorphism of the form @xmath2868 for @xmath2869 .
let @xmath2870 . if @xmath2871 , then @xmath2872 while if @xmath2873 , then @xmath2874 here , we view @xmath2875 or @xmath2876 as automorphisms of @xmath604}_{1+z^{v_k}}$ ] , and @xmath2842 or @xmath2861 as subgroups of the group of automorphisms of this ring .
let us prove the first statement , the second being similar .
it is enough to check that @xmath2877 but , with @xmath2878 , @xmath2879 noting that @xmath2880 , and in addition @xmath2881 , we see that as a derivation , writing @xmath2882 , @xmath2883 of course @xmath2884 by definition of @xmath491 , so the derivation @xmath2885 lives in @xmath2886 ( resp .
@xmath2887 ) . _ step i. strategy of the proof .
_ we will first construct @xmath624 using essentially the same algorithm as the one given in construction [ kslemalg ] , but working with the group @xmath2861 .
the algorithm is slightly more complex because of the slab , and needs to be carried out in two steps . to show that the diagram constructed is consistent at each step , we compare it with the scattering diagram @xmath378 for the group @xmath93 which we know exists , using @xmath2888 as an intermediary group . because @xmath2889 , we obtain a consistent scattering diagram for @xmath2842 and @xmath93 . while @xmath624 is equivalent to @xmath378 as a scattering diagram for @xmath93 by construction , this does not show uniqueness of @xmath624 , as there may be a different choice with wall crossing automorphisms in @xmath2890 but not in @xmath2861 , so it can not be compared with @xmath378 .
thus , the final step involves showing uniqueness directly for the group @xmath2842 , again as part of the inductive proof .
we will proceed by induction on @xmath2073 , constructing for each @xmath2073 a finite scattering diagram @xmath2891 for @xmath2861 containing @xmath601 such that the following induction hypotheses hold : 1 . for every joint @xmath491 of @xmath2891 ,
there is a simple loop @xmath490 around @xmath491 small enough so that it only intersects walls and slabs containing @xmath491 and such that @xmath2892 , as an automorphism of @xmath604}_{1+z^{v_k}}$ ] , lies in @xmath2861 and is trivial in @xmath2888 , or equivalently , by the inclusion @xmath2893 , trivial in @xmath2894 .
2 . if @xmath2895 is a scattering diagram for @xmath2842 which has the same incoming walls as @xmath2891 and satisfies ( 1 ) ( with @xmath2861 replaced by @xmath2842 everywhere ) , then @xmath2895 is equivalent to @xmath2891 in @xmath2894
. _ step iii . from @xmath2891 to @xmath2899 : adding walls associated to joints not contained in @xmath626 .
_ now assume we have found @xmath2900 satisfying the induction hypotheses .
we need to add a finite number of walls to get @xmath2899 .
we will carry out the construction of @xmath2899 in two steps , following construction [ kslemalg ] .
first , let @xmath491 be a perpendicular joint of @xmath2891 with @xmath2901 .
let @xmath2777 be the set of integral tangent vectors to @xmath491 .
if @xmath490 is a simple loop around @xmath491 small enough so that it only intersects walls containing @xmath491 , we note that every wall - crossing automorphism @xmath2902 contributing to @xmath2903 lies in @xmath2860 .
thus as in , in @xmath2904 we can write @xmath2905 with @xmath2906 , and @xmath2907 with @xmath2908 as @xmath2892 is the identity in @xmath2888 by the induction hypothesis . finally , @xmath2909 because the joint is perpendicular .
let @xmath2803:=\{({\mathfrak{j}}-{\mathbb{r}}_{\ge 0}p^*(n_i),(1+z^{p^*(n_i)})^{\pm c_i } ) \,|\,i=1,\ldots , s \}.\ ] ] here @xmath2910 makes sense as a power series .
the sign is chosen in each wall so that its contribution to @xmath2804}$ ] is @xmath2911 to @xmath2912-order @xmath2913 .
_ step iv . from @xmath2891 to @xmath2899 : adding walls associated to joints contained in @xmath626 .
_ if we did nt have a slab , @xmath2895 constructed above would now do the job as in the proof of lemma [ ksalglemma ] .
however , the elements of @xmath2861 trivial in @xmath2888 do not commute with @xmath605 to order @xmath2913 as automorphisms of @xmath604}_{1+z^{v_k}}$ ] in any reasonable sense . as a consequence
, we will need to add some additional walls coming from joints in @xmath626 , some of which have arisen as the intersection of @xmath626 with walls added in step iii .
consider a perpendicular joint @xmath2915 of @xmath2916 .
necessarily the linear span of @xmath491 is @xmath2917 for some @xmath463 .
furthermore , we can choose @xmath510 so that any wall containing @xmath491 then has linear span @xmath2918 for some @xmath2919 non - negative rational numbers .
the direction of such a wall is positively proportional to @xmath2920 .
we now distinguish between two cases .
note that @xmath2921 as the joint is not parallel , so we call the joint @xmath491 _ positive _ or _ negative _ depending on the sign of @xmath2922 .
note that if the joint is positive ( negative ) then @xmath2923 is positive ( negative ) for all @xmath2924 .
if the joint is positive , then choose @xmath490 so that the first wall crossed is @xmath606 , passing from @xmath673 to @xmath674 .
we can write @xmath2925 where @xmath2926 are compositions of wall - crossing automorphisms .
it then follows from @xmath2927 for all @xmath2928 , @xmath2924 and lemma [ conjugationlemma ] that @xmath2929 , hence @xmath2930 .
if the joint is negative , then we use a slightly different loop : without changing the orientation of the loop @xmath490 , change the endpoints so that @xmath490 now starts and ends in @xmath674 , crossing @xmath606 just before its endpoint .
then @xmath2931 and again by lemma [ conjugationlemma ] , @xmath2930 .
thus in both cases , @xmath2932 and is the identity in @xmath2888 .
thus we still have and we can produce a scattering diagram @xmath2933 $ ] in the same way as for the joints @xmath491 not contained in @xmath626 .
we then set @xmath2934,\ ] ] where the union is over perpendicular joints of @xmath2916 contained in @xmath626 .
_ step v. ( 1 ) of the induction hypothesis is satisfied . _
consider a perpendicular joint @xmath491 of @xmath2899 .
first suppose @xmath2901 .
we proceed as in the proof of lemma [ ksalglemma ] .
if @xmath491 is contained in a joint of @xmath2891 , there is a unique such joint , say @xmath2811 , and we constructed @xmath2812 $ ] above . if @xmath491 is not contained in a joint of @xmath2891 , we define @xmath2812 $ ] to be the empty set .
there are three types of walls @xmath613 in @xmath2899 containing @xmath491 : 1 .
@xmath2935 $ ] .
2 . @xmath2936)$ ] , but @xmath2815 .
this type of wall does not contribute to @xmath2937 in @xmath2904 .
indeed , the associated automorphism is in the center of @xmath2904 and this wall contributes twice to @xmath2938 , with the two contributions inverse to each other , so the contribution cancels .
3 . @xmath2939)$ ] and @xmath2819 . since each added wall is of the form @xmath2820 for some joint @xmath2821 of @xmath2891 , where @xmath866 is the direction of the wall , the direction of the wall is parallel to @xmath491 , contradicting @xmath491 being a perpendicular joint .
thus this does not occur .
on the other hand , suppose @xmath491 is a perpendicular joint of @xmath2899 contained in @xmath626 . then since no wall of @xmath2940 is contained in @xmath626 , by definition of @xmath611 ,
in fact @xmath491 is a joint of @xmath2916 .
thus we see again by construction of @xmath2933 $ ] that to order @xmath2913 , @xmath2937 is the identity for @xmath490 the loop around @xmath491 described in step iv . recall the choice of loop depends on whether the joint is positive or negative .
now we show that @xmath2899 satisfies the induction hypothesis ( 1 ) , using the above existence of a @xmath490 such that @xmath2941 for each perpendicular joint @xmath491 .
note that there is a map @xmath2942 for any @xmath2943 .
the slab automorphism @xmath605 can be viewed as an element of @xmath2944 for any @xmath2943 , and hence @xmath2899 can be viewed as a scattering diagram for @xmath2944 in the sense of definition [ ksscatdiagdef ] .
we will first show that @xmath2899 is consistent as a diagram for @xmath2944 inductively on @xmath2943 .
the base case is @xmath2945 .
all walls of @xmath2946 are trivial to @xmath2912-order @xmath2073 and hence to @xmath2135-order @xmath2073 .
now @xmath2891 satisfies the main induction hypothesis ( 1 ) at order @xmath2073 , which implies via the natural map @xmath2947 that @xmath2899 is consistent as a diagram for @xmath2948 .
indeed , as @xmath2899 is a finite scattering diagram , it is enough to check that @xmath2937 is the identity in @xmath2949 for any small loop @xmath490 around any joint @xmath491 . by the hypothesis ( 1 ) , this is the case for some loop @xmath490 , and hence for all loops .
note that by uniqueness of consistent scattering diagrams with the same incoming walls , we also record for future use : @xmath2950 the induction step follows from lemma [ equivalencetonextorder ] , applied to @xmath2951 , @xmath2952 , and the group being @xmath2944 ( a quotient of @xmath93 , so the argument of lemma [ equivalencetonextorder ] still applies ) .
indeed , if we assume @xmath2899 is consistent in @xmath2944 , then it is equivalent to @xmath378 as a scattering diagram in @xmath2944 .
furthermore , @xmath378 is consistent to all orders by theorem [ kslemmags ] , and has the same set of incoming walls as @xmath2899 by construction .
finally , @xmath2937 is the identity in @xmath2953 for any perpendicular joint @xmath491 , as shown above
. thus @xmath2899 and @xmath378 are equivalent in @xmath2953 , and in particular @xmath2899 is consistent in @xmath2953 . thus taking the inverse limit
, we see that @xmath2899 is consistent as a scattering diagram for @xmath2954 .
this almost completes the proof of the induction hypothesis ( 1 ) in degree @xmath2913 . indeed , as @xmath2904 is a subgroup of @xmath2954 , certainly @xmath2937 is the identity for any joint not contained in @xmath626 , including the parallel joints . for a perpendicular joint contained in @xmath626 ,
if we choose @xmath490 as given in step iv , @xmath2937 lies in @xmath2861 and is the identity in @xmath2904 by the construction of @xmath2933 $ ] in step iv .
finally , for a parallel joint @xmath491 contained in @xmath626 , note that all wall and slab - crossing automorphisms associated to walls containing @xmath491 commute , and in particular the contribution of @xmath605 and @xmath2955 in @xmath2937 as an automorphism of @xmath604}_{1+z^{v_k}}$ ] cancel , so that the latter automorphism lies in @xmath2861 .
hence the image of this automorphism in @xmath2956 must also be trivial .
this gives the induction hypothesis ( 1 ) .
_ step vi . uniqueness .
_ suppose we have constructed two scattering diagrams @xmath2957 for @xmath2842 from @xmath2891 which satisfy the inductive hypothesis ( 1 ) to @xmath2912-order @xmath2913 , but with the group @xmath2861 replaced with @xmath2842 . by the induction hypothesis ( 2 ) , these two scattering diagrams are equivalent to @xmath2912-order @xmath2073 , and we wish to show they are equivalent to @xmath2912-order @xmath2913 .
one first constructs a finite scattering diagram @xmath328 consisting only of outgoing walls whose attached functions are of the form @xmath2958 with @xmath2959 and @xmath2960 , with the property that @xmath2961 is equivalent to @xmath2962 to @xmath2912-order @xmath2913 .
this is done precisely as in the proof of lemma [ equivalencetonextorder ] .
we need to show @xmath328 is equivalent to the empty scattering diagram to @xmath2912-order @xmath2913 .
to show this , first note that for any loop @xmath341 which does not cross the slab @xmath606 , @xmath2963 to @xmath2912-order @xmath2913 implies that @xmath2964 to @xmath2912-order @xmath2913 .
indeed , all wall - crossing automorphisms of @xmath328 are central in @xmath2965 . now if @xmath2865 with @xmath2960 , let @xmath2966 be the set of walls in @xmath328 with attached functions of the form @xmath2958 .
note all wall - crossing automorphisms of @xmath328 , viewed as elements of @xmath2965 , lie in @xmath2967 , which as a group coincides with the additive group structure on @xmath2968 .
thus for any path @xmath341 not crossing @xmath606 , we obtain a unique decomposition @xmath2969 from the @xmath611-grading on @xmath2970 , and if @xmath361 is the identity , so is each @xmath2971 .
fixing @xmath510 as above , replace @xmath2972 with an equivalent scattering diagram with smallest possible support , and let @xmath2973 .
so if @xmath2974 is a general point , @xmath2975 if and only if @xmath2976 is not the identity
. assume first that @xmath2977 .
we shall show @xmath2978 .
assume not .
taking a general point @xmath2979 , it is not possible for the ray @xmath2980 to be contained in @xmath2981 .
this is because @xmath328 consists of only a finite number of walls , none of which are incoming .
let @xmath2982 , and @xmath2983 .
this makes sense as @xmath2984 is in the set over which we are taking the maximum , as we are assuming @xmath2975 .
then necessarily @xmath2985 is in a joint @xmath491 of @xmath2972 , and every wall of @xmath2972 containing @xmath491 is contained in @xmath2986 .
furthermore , since @xmath2987 , @xmath2988 , it follows that @xmath2989 and @xmath491 is not contained in @xmath626 . thus given a loop @xmath490 around @xmath491 ,
@xmath2990 is the identity .
this implies that in fact to @xmath2912-order @xmath2913 , @xmath2991 in particular , a point @xmath2992 for small @xmath652 is contained in precisely those walls of @xmath2972 containing @xmath491 .
but then @xmath2993 , contradicting minimality of @xmath2981 .
thus one finds that @xmath2994 .
similarly , if @xmath2995 , then @xmath2996 . in particular ,
if @xmath2997 , @xmath2998 , but there are no walls contained in @xmath626 , so in this case @xmath2999 . now consider a joint @xmath491 of @xmath2962 contained in @xmath626 .
there are three cases : either @xmath491 is perpendicular and positive , perpendicular and negative , or parallel .
consider the first case .
take a loop @xmath490 around @xmath491 as in the positive case in step iv .
because of positivity , if a wall @xmath613 of @xmath328 contains @xmath491 , then with @xmath510 chosen so that @xmath3000 , we must have @xmath3001 and hence @xmath613 is contained in @xmath673 .
thus we have that to @xmath2912-order @xmath2913 , @xmath3002 as in , where @xmath3003 and @xmath3004 denote the contributions coming from the scattering diagrams @xmath328 and @xmath2899 and the pieces of @xmath490 not crossing @xmath626 .
the same argument works for negative joints , while a parallel joint can not contain any wall of @xmath328 , ( as we showed above that @xmath2999 if @xmath2997 ) , so that @xmath489 trivially . we can now repeat the argument of the previous paragraph , taking for any @xmath510 a general point @xmath2975 rather than @xmath3005 .
this allows us to conclude that @xmath2999 for all @xmath510 , proving uniqueness .
_ step vii .
finishing the proof of theorem [ kslemma2 ] . _
having completed the induction step , we take @xmath3006 . we need to check it satisfies the stated conditions in theorem [ kslemma2 ] .
certainly conditions ( 1 ) and ( 2 ) hold by construction . for ( 3 ) , first recall that because by construction @xmath627 can be viewed as a scattering diagram for @xmath2861 , it can also be viewed as a scattering diagram for @xmath93 via the inclusion @xmath2856 , and in addition @xmath3007 , so that @xmath627 is viewed as a scattering diagram for @xmath93 in the sense of definition [ ksscatdiagdef ] , i.e. , with no slab .
now as a scattering diagram for @xmath93 , @xmath627 is equivalent to @xmath378 by . by consistency of @xmath378
, @xmath3008 is independent of the endpoints of @xmath341 as an element of @xmath93 .
now suppose @xmath3009 are two automorphisms of @xmath604}_{1+z^{v_k}}$ ] which induce automorphisms of @xmath268}$ ] , ( i.e. , for @xmath3010 , @xmath3011}$ ] , giving a map @xmath3012}\rightarrow \widehat{{\bbbk}[p]}$ ] which is an automorphism ) and agree as automorphisms of the latter ring .
then @xmath3009 agree as automorphisms of @xmath604}_{1+z^{v_k}}$ ] .
thus in particular , @xmath3013 is independent of the endpoints of @xmath341 as an automorphism of @xmath604}_{1+z^{v_k}}$ ] .
this gives condition ( 3 ) .
the uniqueness of @xmath627 with these properties then follows from the induction hypothesis ( 2 ) . indeed ,
if @xmath3014 satisfies conditions ( 1)-(3 ) of theorem [ kslemma2 ] , then working by induction on the order @xmath2073 , the induction hypothesis ( 1 ) holds for @xmath3014 ( the existence of @xmath490 with @xmath3015 only being an issue for joints contained in @xmath626 , and step iv explains how to choose the loop @xmath490 ) .
thus by induction hypothesis ( 2 ) , @xmath627 and @xmath3016 are equivalent to order @xmath2073 .
the key point of the proof is just the positivity of the simplest scattering diagram as described in example [ basicscatteringexample ] , which we use to analyze general two - dimensional scattering diagrams . we will consider a somewhat more general setup , but only in two dimensions , than considered in the rest of this paper . in particular
, we will follow the notation of @xcite , 6.3.1 , taking @xmath3017 , @xmath3018 , and assume given a monoid @xmath265 with a map @xmath3019 , @xmath3020 .
we will consider scattering diagrams @xmath328 for this data as in @xcite , def .
6.37 , consisting of rays and lines which do not necessarily pass through the origin . given any scattering diagram @xmath940 , the argument of kontsevich and soibelman from @xcite ( see @xcite , theorem 6.38 for an exposition of this particular case ) adds rays to @xmath940 to obtain a scattering diagram @xmath3021 such that @xmath3022 is the identity for every loop @xmath341 .
this diagram is unique up to equivalence .
[ basicpositivity ] in the above setup , suppose given @xmath3023 , @xmath3024 , with @xmath3025 , and positive integers @xmath3026 .
consider the scattering diagram @xmath3027 @xmath3028 .
we can choose @xmath328 within its equivalence class so that for any given ray @xmath3029 , we have @xmath3030 for @xmath130 a positive integer and the @xmath902 non - negative integers with at least two of them non - zero .
_ step i. the change of monoid trick_. note that if the @xmath3031 generate a rank one sublattice of @xmath1444 , then all the wall - crossing automorphisms of @xmath940 commute and @xmath3032 , so we are done .
so assume from now on that the @xmath3031 generate a rank two sublattice of @xmath1444 .
let @xmath3033 , generated by @xmath3034 , define a map @xmath3035 by @xmath3036 , and a map @xmath3037 by @xmath3038 .
we extend @xmath2741 to a map @xmath3039}\rightarrow\widehat{{\bbbk}[p]}$ ] , and define , for a scattering diagram @xmath328 for the monoid @xmath591 , @xmath3040 . clearly if @xmath2964 , then @xmath3041 . thus if @xmath3042 , then @xmath3043 is equivalent to @xmath3021 , by uniqueness of @xmath3044 up to equivalence .
so it is sufficient to show the result with @xmath3045 , @xmath3046 .
_ step ii .
everything but the positivity of the exponents_. we can construct @xmath328 specifically using the original method of @xcite , already explained here in steps iii and iv of the proof of theorem [ kslemma2 ] : we construct @xmath328 order by order , constructing @xmath3047 so that @xmath3048 is the identity modulo @xmath3049 for @xmath341 a loop around the origin . given a description @xmath3050 with the @xmath902 primitive and the @xmath3051 all distinct , we add a collection of rays @xmath3052 for some @xmath2906 .
however , inductively , we can show the @xmath3053 can be taken to be integers .
indeed , if all rays in @xmath3047 have this property , then @xmath3048 is in fact an automorphism of @xmath3054}$ ] , and thus the @xmath3053 appearing in of @xmath3048 are also integers .
next let us show that any exponent @xmath3051 is of the form @xmath3055 with at least two of the @xmath3056 non - zero .
the pro - nilpotent group @xmath3057 in which all automorphisms live is given by the lie algebra @xmath3058),\ ] ] following the notation of @xcite , pp .
290 - 291 .
this contains a subalgebra @xmath3059 where the sum is taken over all @xmath3060 not proportional to one of the @xmath1232 .
then clearly @xmath3061\subseteq { \mathfrak{v}}'$ ] , so the corresponding pro - nilpotent group @xmath3062 is normal in @xmath3057 .
furthermore , @xmath3063 is abelian , hence so is @xmath3064 . for any loop @xmath341 ,
the image of @xmath3065 is thus the identity in @xmath3064 , as every wall in @xmath940 contributes twice to @xmath3066 , but with inverse automorphisms .
assume inductively that @xmath3067 only contains rays whose attached functions @xmath3068 have @xmath3051 not proportional to any @xmath3069 .
then the wall - crossing automorphisms associated to these rays lie in @xmath3062 , so @xmath3048 is the identity in @xmath3064 , i.e. , lies in @xmath3062 .
thus the expression @xmath3070 of lies in @xmath3059 , hence the inductive step follows .
_ step iii .
the perturbation trick_. we will now show the result for all monoids @xmath3072 for all @xmath59 , all choices of @xmath3019 , all choices of @xmath3073 with @xmath3025 , and all positive choices of @xmath251 .
( note by step i this is a bit more than we need , as we do nt take the @xmath1994 to necessarily be generators of @xmath265 ) .
all cases are dealt with simultaneously by induction .
we define for @xmath3074 the _ order _
@xmath3075 , which is the unique @xmath3076 such that @xmath3077 . for a ray @xmath3078
, we write @xmath3079 , and say @xmath613 is a ray of order @xmath3080 .
we will go by induction on the order , showing that a ray @xmath3078 in @xmath328 of order @xmath3081 for any choice of data has @xmath130 positive .
this is obviously the case for @xmath3082 , as all elements of @xmath3083 have order at least @xmath430 .
so assume the induction hypothesis is true for all orders @xmath3084 , and we need to show rays added of order @xmath460 have positive exponent .
we will use the _ perturbation trick _ repeatedly . given a scattering diagram @xmath940 for which we would like to compute @xmath3085 , choose general @xmath3086 for each @xmath3087 .
define @xmath3088 ; this is the perturbed diagram .
we can then run the kontsevich - soibelman algorithm for @xmath3089 , for example as described in @xcite , theorem 6.38 .
this gives a scattering diagram @xmath3090 with the property that @xmath3091 is the identity for every loop @xmath341 .
this is the case in particular for @xmath341 a very large loop around the origin which contains all singular points of @xmath363 .
we can assume as usual that @xmath363 has been constructed only by adding rays of the form @xmath3092
. then up to equivalence , @xmath328 can be obtained from @xmath363 by taking the _ asymptotic scattering diagram of @xmath363 _ , i.e. , just translate each line of @xmath363 so it passes through the origin and each ray of @xmath363 so its endpoint is the origin .
1.4 of @xcite for more details .
if after performing this translation , we obtain a number of rays with the same support of the form @xmath3093 , @xmath210 in some index set , we can replace all these rays with a single ray @xmath3094 without affecting the equivalence class .
thus if we want to show positivity of the exponents for @xmath328 , it is enough to show the desired positivity for @xmath363 .
we will typically use an induction hypothesis to show positivity for @xmath363 .
indeed , for each order , we will run the kontsevich - soibelman algorithm at each singular point , and the behaviour at each singular point is equivalent to a scattering diagram of the general type being considered . indeed ,
if @xmath123 is a singular point of some @xmath3095 constructed to order @xmath2135 , we obtain a local version @xmath3096 of the scattering diagram at @xmath123 by replacing each @xmath613 with @xmath3097 with @xmath3098 , and replacing such translated rays with the line spanned by the ray if the translated ray does not have the origin as its endpoint .
as long as all attached functions of rays and lines passing through @xmath123 are of the form @xmath3092 with @xmath130 a positive integer , we are back in the original situation of the proposition .
we shall write @xmath3099 for the set of lines in @xmath3096 .
we first observe that using the perturbation trick it is enough to show the induction hypothesis for order @xmath460 when at most two of the @xmath1994 have @xmath3100 . indeed , after perturbing , the lines of @xmath3089 only intersect pairwise , but as more rays are added as the kontsevich - soibelman algorithm is run , one might have more complicated behaviour at singular points .
however , any ray added has order @xmath3101 .
thus we only have to analyze initial scattering diagrams @xmath3102 with at most two lines of order @xmath616 .
next we observe the induction hypothesis allows us to show the result only for @xmath3103 , with both lines having order @xmath616 . indeed , write @xmath3102 as @xmath3104 and order the @xmath1994 so that @xmath3105 .
apply step i , getting a map @xmath3035 with @xmath3036 .
we are trying to prove that rays with @xmath265-order @xmath460 have positive exponent . but consider a ray @xmath3106 which is the image under @xmath2741 of a ray @xmath3107 appearing in @xmath3108 with @xmath3109 and at least one of @xmath3110 non - zero .
then @xmath3111 , so by the induction hypothesis , we can assume @xmath130 is positive . on the other hand , rays of the form @xmath3107 with @xmath3112 for @xmath3113 appearing in @xmath3108 already appear in @xmath3114 , as follows easily by working modulo the ideal in @xmath591 generated by the @xmath3115 .
thus we are only concerned about rays which arise from scattering the two order @xmath616 lines .
thus it is sufficient to show the result when @xmath3103 .
_ step iv .
the change of lattice trick_. to deal with the case where @xmath940 consists of two lines , we use the _ change of lattice _ trick to reduce to a simpler expression for the scattering diagram . by step
i , we can take @xmath3116 , @xmath3046 .
let @xmath3117 be the sublattice generated by @xmath3118 .
note as in step i we can assume that this is a rank @xmath430 sublattice , as otherwise the automorphisms associated to the two lines commute . then @xmath3119 is a superlattice of @xmath240 , with dual basis @xmath3120 . in what follows , we will talk about scattering diagrams defined using both the lattice @xmath1444 and @xmath113 .
bear in mind that a wall @xmath325 could be interpreted using either lattice , and the automorphism induced by crossing such a wall depends on which lattice we are using , as primitive vectors in @xmath240 differ from primitive vectors in @xmath245 . to see the relationship between these automorphisms , for @xmath3121 ,
let @xmath3122 then a wall @xmath325 for @xmath1444 induces a wall - crossing automorphism of @xmath268}$ ] which is the same as the automorphism induced by the wall @xmath3123 for @xmath113 , where @xmath3124 is primitive and annihilates @xmath613 .
consider @xmath3125 as a scattering diagram for the lattice @xmath113 .
let @xmath3126 .
let @xmath363 be the scattering diagram for @xmath1444 obtained by replacing every wall @xmath3127 with @xmath3128 .
thus the wall - crossing automorphism for each wall in @xmath363 as a scattering diagram for the lattice @xmath1444 is the same automorphism for the corresponding wall in @xmath3129 .
then @xmath3091 is the identity .
thus by uniqueness of the scattering process up to equivalence , @xmath363 is equivalent to @xmath3021 .
( note this implies that @xmath3130 also , as @xmath3021 only involves integer exponents . )
thus it is enough to prove the desired positivity for the scattering diagram @xmath3129 . to do so
, we use a variant of the perturbation trick , factoring the two lines in @xmath3131 .
we choose general @xmath3132 , with @xmath3133 , @xmath3134 .
define @xmath3135 again , we initially only have pair - wise intersections .
the first stage of this algorithm will then only involve points where two lines of the form @xmath3136 and @xmath3137 intersect .
the algorithm only adds one ray in the direction @xmath3138 with endpoint the intersection point and attached function @xmath3139 , as follows from example [ basicscatteringexample ] .
this now accounts for all new rays of order @xmath430 .
we continue to higher degree , but now we can use the induction hypothesis at every singular point @xmath123 as we did in step iii , because every line in @xmath3099 has order @xmath3140 except for possibly one or two of the given lines of order @xmath616 , and we have already accounted for all rays produced by collisions of two lines of order @xmath616 .
[ positivitywithcoefs ] in the situation of proposition [ basicpositivity ] , suppose instead that @xmath3141 where now @xmath3142 , the ground field .
choosing @xmath3085 up to equivalence , we can assume that each ray @xmath3143 satisfies @xmath3144 for some choice of non - negative integers @xmath1036 and where @xmath130 is a positive integer .
this follows easily from from proposition [ basicpositivity ] .
first , using the change of monoid trick ( step i of the proof of proposition [ basicpositivity ] ) , we may assume @xmath3045 and @xmath3046 .
consider the automorphism @xmath3145}\rightarrow \widehat{{\bbbk}[p]}$ ] defined by @xmath3146 .
applying @xmath70 to the function attached to each wall of @xmath3147 gives a scattering diagram @xmath363 whose incoming walls are precisely those of @xmath940 , and @xmath3148 for @xmath341 a loop around the origin
. thus we can take @xmath3149 and the result follows from proposition [ basicpositivity ] .
_ proof of theorem [ scatdiagpositive ] .
_ in fact one can use the @xmath378 as constructed explicitly in the algorithm of the proof of theorem [ kslemma2 ] .
the only issue is that we need to know that the walls added at each joint have the desired positivity property .
note that the statement of theorem [ scatdiagpositive ] involves scattering diagrams without slabs , while the proof of theorem [ kslemma2 ] given involves a slab .
so for the purpose of this discussion , we can ignore all issues concerning the slab in the proof of theorem [ kslemma2 ] , and the only thing we need to do is look at the procedure for producing @xmath2933 $ ] in step ii of the proof of theorem [ kslemma2 ] .
for a perpendicular joint @xmath491 of @xmath3047 , we can split @xmath3150 , where @xmath3151 is a rank two lattice . for each wall
@xmath3152 containing @xmath491 , we can inductively assume that @xmath381 for some positive integer @xmath130 , and split @xmath3153 , with @xmath3154 and @xmath3155 . because @xmath491 is perpendicular , we have @xmath3156
. we will apply corollary [ positivitywithcoefs ] to the case where the monoid @xmath265 is the one being used in theorem [ kslemma ] , and @xmath3157 is the projection .
we can then view the computation at the joint as a two - dimensional scattering situation in the lattice @xmath3151 over the ground field @xmath3158 , the quotient field of @xmath3159 $ ] . to obtain the relevant two - dimensional scattering diagram
we replace each wall @xmath325 with @xmath3160 with @xmath3161 in @xmath3162 .
we are then in the situation of corollary [ positivitywithcoefs ] , and the result follows .
g. cerulli irelli , b. keller , d. labardini - fragoso , and p. plamondon , _ linear independence of cluster monomials for skew - symmetric cluster algebras _ , compos .
* 149 * ( 2013 ) , 17531764 .
cheung , m. gross , g. muller , g. musiker , d. rupel , s. stella , h. williams , _ the greedy basis equals the theta basis _ ,
jour . of combinatorial
theory , series a , ( 2017 ) , 150171 .
m. inaba , k. iwasaki , and m. saito , _ moduli of stable parabolic connections , riemann - hilbert correspondence and geometry of painlev equation of type vi .
i. _ publ .
inst . math .
* 42 * ( 2006 ) , 9871089 .
m. kontsevich and y. soibelman , _ affine structures and non - archimedean analytic spaces _ , in : _ the unity of mathematics _
( p. etingof , v. retakh , i.m .
singer , eds . ) , 321385 , progr .
math . * 244 * , birkhuser 2006 .
m. kontsevich and y. soibelman , _ wall - crossing structures in donaldson - thomas invariants , integrable systems and mirror symmetry _ , in _ homological mirror symmetry and tropical geometry _ , 197308 , lect .
notes unione mat .
* 15 * , springer , cham , 2014 . | in @xcite , conjecture 0.6 , the first three authors conjectured that the ring of regular functions on a natural class of affine log calabi - yau varieties ( those with maximal boundary ) has a canonical vector space basis parameterized by the integral tropical points of the mirror .
further , the structure constants for the multiplication rule in this basis should be given by counting broken lines ( certain combinatorial objects , morally the tropicalisations of holomorphic discs ) .
here we prove the conjecture in the case of cluster varieties , where the statement is a more precise form of the fock - goncharov dual basis conjecture , @xcite , conjecture 4.3 .
in particular , under suitable hypotheses , for each @xmath0 the partial compactification of an affine cluster variety @xmath1 given by allowing some frozen variables to vanish , we obtain canonical bases for @xmath2 extending to a basis of @xmath3 .
each choice of seed canonically identifies the parameterizing sets of these bases with integral points in a polyhedral cone .
these results specialize to basis results of combinatorial representation theory .
for example , by considering the open double bruhat cell @xmath1 in the basic affine space @xmath0 we obtain a canonical basis of each irreducible representation of @xmath4 , parameterized by a set which each choice of seed identifies with integral points of a lattice polytope .
these bases and polytopes are all constructed essentially without representation theoretic considerations . along the way
, our methods prove a number of conjectures in cluster theory , including positivity of the laurent phenomenon for cluster algebras of geometric type . |
when the characteristic dimensions of a metallic wire are shrunk all the way down to the atomic scale , its electronic transport properties change dramatically as a consequence of the appearance of quantum mechanical effects .
this has been nicely illustrated in recent years with the help of metallic atomic contacts fabricated by means of scanning tunneling microscopes ( stms ) and break - junction techniques @xcite .
thus for instance , it has been shown that transport properties such as the conductance @xcite , shot noise @xcite , or thermopower @xcite , differ markedly from those of macroscopic wires , while transport phenomena such as joule heating @xcite or magnetoresistive effects @xcite take place in a very different manner in these nanowires . in all cases ,
these dramatic differences can be traced back to the fact that the transport in metallic atomic contacts is mainly quantum coherent .
the central goal of this work is to show that the nature of electrical conduction in ferromagnetic atomic contacts is clearly at variance with the established picture in macroscopic wires .
our present understanding of electrical conduction in macroscopic wires made of ferromagnetic metals like fe , co , and ni is largely based on the semiclassical model put forward by sir nevill mott in the 1930s @xcite . in that simple model , sometimes referred to as the two - current model , the current is carried by the electrons of two independent spin bands , the majority - spin electrons and the minority - spin electrons .
moreover , it is assumed that the conduction bands have an @xmath2 character , while the @xmath1-electrons are localized and the corresponding spin - split bands are responsible for the net magnetization .
mott described the conductivity of a ferromagnetic metal in terms of the drude formula , where the key parameters are the density of conduction electrons , assumed to be spin - independent , and the spin - dependent inelastic scattering time ( or relaxation time ) .
while the relaxation time for the majority - spin electrons is similar to that of a nonmagnetic metal , the scattering of minority - spin electrons from the @xmath2 states into the partially occupied @xmath1 states reduces considerably the corresponding relaxation time and , in turn , the electrical conductivity in ferromagnetic metals .
this explains , for instance , why fe , co , and ni are less conductive than the noble metals ( au , ag , and cu ) .
moreover , this model predicts that the conduction is dominated by the majority - spin electrons , leading to a positive spin polarization of the current .
this simple picture turns out to be quite accurate , as it has been very recently demonstrated using ultrafast terahertz spectroscopy @xcite .
however , as we shall show in this work , it severely fails to explain the transport properties in ferromagnetic atomic contacts . since the advent of stm - based and break - junction techniques in the 1990s , many experimental studies of transport through atomic contacts made of ferromagnetic metals
have been reported @xcite . some of the early reports focused on the observation of half - integer quantization ,
_ i.e. _ , on the observation of peaks in the conductance histograms at half - integer multiples of the quantum of conductance @xmath3 @xcite . in the landauer picture of the coherent transport through these ferromagnetic nanowires ,
the low - temperature linear conductance is given by @xmath4 , where @xmath5 is the transmission coefficient at the fermi energy of the @xmath6th available conduction channel for a spin @xmath7 .
thus , these observations were interpreted as an evidence indicating that only spin - split fully open channels contribute to the conductance in these atomic - scale ferromagnetic wires .
this interpretation was questioned by untiedt _
et al . _
@xcite who measured the conductance for atomic contacts of fe , co , and ni using break junctions under cryogenic vacuum conditions .
contrary to the experiments mentioned above , they reported the absence of fractional conductance quantization .
instead , they observed conductance histograms that show broad peaks above @xmath8 .
moreover , they suggested that the observation of peaks in the conductance histograms at half - integer values of @xmath9 could be due to contamination .
we note that half - integer conductance values are occasionally detected for arbitrarily realized atomic - scale contacts of both ferromagnetic and non - ferromagnetic metals .
another controversy in the context of ferromagnetic atomic contacts has to do with the observation of an anomalous anisotropic magnetoresistance ( amr ) .
amr is a spintronic effect in which the resistance varies as a function of the relative orientation between the magnetization and the current directions and it originates from spin - orbit interaction .
et al . _
@xcite reported that the magnitude of the amr of permalloy atomic contacts can be considerably larger than in bulk samples and that it exhibits an anomalous angular dependence .
similar observations were reported by viret _
et al . _
@xcite in fe contacts , but they also reported the occurrence of conductance jumps upon the rotation of the magnetization .
similar step - wise variations in the conductance were later found in co nanocontacts @xcite .
these jumps were tentatively interpreted as a manifestation of the so - called ballistic amr ( bamr ) @xcite .
this approach suggested that in a ballistic contact the rotation of the magnetization could result in variations in band crossing at the fermi energy , leading to an abrupt change in the conductance on the order of @xmath10 .
however , as discussed above , ferromagnetic contacts are not expected to be ballistic ( _ i.e. _ to exhibit only fully - open channels ) , and therefore , the interpretation in terms of bamr is highly questionable @xcite .
in fact , shi and ralph @xcite suggested that these jumps might originate from sudden atomic rearrangements @xcite .
half - integer quantization and bamr are not inherently expected in ferromagnetic atomic contacts in view of the established picture of the conduction in these atomic - scale wires @xcite . within this picture ,
based on the landauer approach to coherent transport , the conduction channels in a nonmagnetic metal are spin - degenerate and are determined by the orbital structure of the metal and the geometry of the atomic contact . in particular , the number of conduction channels of a single - atom contact is expected to be limited by the number of valence orbitals , as it was experimentally verified with the use of superconductivity to extract the channel content @xcite .
moreover , it was established that conductance quantization is only expected in few - atom contacts of monovalent metals , whereas in the case of multivalent metals the conductance is essentially determined by several partially - open channels and there are no signs of conductance quantization .
the controversies discussed above and the great interest in spin - dependent transport at the nanoscale have led to numerous theoretical groups to investigate the transport properties of ferromagnetic atomic contacts @xcite .
apart from the analysis of the conductance and the amr , these studies have also addressed many different aspects of the physics of these contacts such as the electronic structure of ideal systems , like monoatomic wires , the influence of domain walls on electronic and transport properties , or the magnetic structure .
the ensemble of the reported theoretical results clearly suggest that the @xmath1 orbitals play a fundamental role in the transport properties of these atomic - scale wires , contrary to their macroscopic counterparts , and there is no fundamental reason for expecting either conductance quantization or full spin polarization .
however , there is still no generally accepted picture for electrical conduction in ferromagnetic atomic contacts for two main reasons .
first , a systematic one - to - one comparison between experiment and theory for the conductance has never been established . this is a difficult task since it requires from the theory side to obtain a simultaneous description of the detailed atomic structure and transport properties of the contact .
second and more important , transport measurements performed so far in ferromagnetic atomic contacts have been restricted to the conductance , which does not contain information about the individual transmission coefficients .
the access to this information was the key breakthrough that finally enabled to elucidate the nature of the conduction in nonmagnetic atomic contacts , something that became possible thanks to the use of superconductivity @xcite .
however , such a method is not possible in the case of ferromagnetic contacts . in this work
we revisit the nature of the electrical conduction in ferromagnetic atomic contacts by providing a very detailed experimental study of both conductance and shot noise in fe , co , and ni atomic contacts at cryogenic temperature and vacuum conditions .
we present the corresponding results for cu contacts for comparison .
in particular , we show how shot noise measurements provide an insight into the electrical conduction of these ferromagnetic nanowires , which is crucial to understand how they differ from nonmagnetic atomic - scale contacts and macroscopic ferromagnetic wires .
moreover , we supplement our experimental study with a comprehensive theoretical analysis based on molecular dynamics ( md ) and quantum mechanical calculations of the transport properties that allows us to establish a direct comparison with our experiments for the different materials under study . from the experimental side ,
our main results for the ferromagnetic contacts are : ( i ) the confirmation of the lack of any kind of conductance quantization , ( ii ) we find no signs of pronounced shot noise suppression at any conductance value , contrary to noble metals , which shows that the transport is dominated by partially - open channels for any contact size , ( iii ) the analysis of shot noise shows that at least several channels contribute to transport even in smallest ( presumably one - atom ) contacts , again in clear contrast with noble metals , and ( iv ) the fano factor saturates for contacts with conductances higher than @xmath11 to a value that is clearly larger than for monovalent metals .
our theoretical results reproduce very satisfactorily all these basic observations . in all cases
, the origin of these observations can be traced back to the fact that the @xmath0 orbitals play a major role in the conduction of these ferromagnetic atomic contacts .
the picture that emerges is that in a ferromagnetic contact the majority - spin electrons behave as in a noble metal , where the conduction is dominated by the @xmath2 conduction band , while the minority - spin electrons behave as in a transition metal , where both @xmath2 and @xmath1 orbitals contribute decisively to the transport properties . in particular , the @xmath1 orbitals corresponding to the minority - spin bands build up partially open channels which , in turn , are responsible for the experimental observations described above .
moreover , these additional conduction channels are responsible for the higher conductance of ferromagnetic few - atom contacts as compared with noble metals and they give rise to a negative spin polarization of the current in all three ferromagnetic @xmath0 metals , in clear contrast to the positive spin polarization in macroscopic wires .
thus , our study provides a consistent general picture of the conduction in ferromagnetic atomic contacts and their unique transport properties compared to macroscopic ferromagnetic wires .
the rest of the manuscript is organized as follows . in section [ sec - methods ]
we describe the experimental and theoretical techniques that we have used to investigate both the conductance and the shot noise of ferromagnetic atomic contacts .
we present in section [ sec - histograms ] our results for the conductance histograms of the different materials studied in this work .
section [ sec - noise ] is devoted to the description of our experimental and theoretical results for the shot noise and fano factor .
we present a detailed discussion on the origin of our results in terms of the nature of the conduction channels in section [ sec - channels ] .
section [ sec - further - discussions ] is devoted to further discussion of our results .
then , we summarize our main conclusions in section [ sec - conclusions ] .
finally , in appendix a we show the results for co and fe that are not presented in the main text .
in this section we describe the basic experimental and theoretical methods that we have employed in this work to study both the conductance and shot noise of ferromagnetic atomic contacts .
our measurements were performed using a mechanical controllable break junction ( mcbj ) @xcite composed of a notched metal wire attached to a flexible substrate ( see fig . [ exp - setup ] ) .
the wires used consist of a high purity metal ( @xmath12 99.9% ) and have macroscopic dimensions ( 0.1 mm diameter ) .
the sample is positioned in a vacuum chamber that is initially pumped to @xmath13 mbar and cooled by liquid helium to @xmath14 k. once the sample is cold , a mechanical push screw is used to bend the substrate until the wire breaks in its weakest spot , forming two atomically - sharp tips .
since the junction is held under cryogenic vacuum , the freshly exposed tips are kept with minimum exposure to impurities .
this is extremely important to avoid oxidation or other contaminations , which can have a considerable effect on transport @xcite .
the displacement between the tips is then controlled with sub - angstrom resolution by the precise movement of a piezoelectric element .
the tips can be pushed back towards each other to reform an atomic contact , whose minimal cross section can be varied by changing the voltage applied on the piezo element .
the electronic conductance across the atomic contact is measured by connecting measurement wires to the two sides of the thinned metallic wire , which essentially serve as leads to the atomic - scale junction .
figure [ exp - setup ] shows the electronic circuit used for conductance and shot noise measurements .
two computer - controlled relays are used to toggle between conductance ( marked in red ) and noise ( blue ) measurement modes . for dc conductance measurement , a bias voltage
is provided to the junction from a differential voltage source . the response current is amplified with a current preamplifier ( sr 570 ) and recorded by a 24bit digitizer ( ni 4461 ) .
ac conductance measurements are performed by driving a small sinusoidal signal ( 2 mvpp , @xmath153 khz ) added to the bias voltage .
the differential conductance @xmath16 is obtained using the lock - in technique , performed digitally with the data acquisition software . for noise measurements ,
the sample is current biased and the voltage response is amplified by low - noise voltage amplifiers .
the amplified voltage signal is recorded by a fast digitizer ( ni 5992 ) and the power spectrum is calculated by digital fourier transform .
two different configurations of voltage amplifiers were used for the experiments . in one configuration ,
a nf li-75a was used , followed by a signal recovery 5184 .
the overall voltage amplification is @xmath17 and the input voltage noise is 1.5nv/@xmath18 . in a second configuration ,
a specially designed amplifier ( janascard ) with amplification of @xmath19 and voltage noise of 0.9nv/@xmath18 is used .
the lower voltage noise enabled a higher signal to noise ratio , which was important for shot noise measurements at high conductance values ( @xmath20-@xmath21 ) .
the sample and amplifiers are located inside a specially - designed faraday cage in order to minimize noise pickup from environmental radiation .
the piezo voltage is supplied by a differential voltage source and amplified by a factor of four using a piezomechanik svr-150 piezo driver . in order to compute the conductance and
shot noise of the atomic contacts studied experimentally , we have combined classical molecular dynamics ( md ) simulations of the formation of the contacts , a tight - binding description of the electronic structure , and nonequilibrium green s function techniques .
our methodology proceeds along the lines of refs .
@xcite . in the following
we briefly describe our approach .
nm . in both panels
we have indicated the partitioning of the contact into the upper and lower electrodes and the central wire , as used for the md and transport calculations . ]
_ molecular dynamics simulations_. in metallic atomic contacts there is a crucial interplay between mechanical and transport properties .
thus , in order to establish a direct comparison with our experiments , it is necessary to describe the formation process of these nanowires . for this purpose
we have carried out md simulations using the open source program package lammps @xcite . within lammps , we employed the embedded atom method with the semi - empirical potentials from ref .
@xcite for ni and cu , ref .
@xcite for co , and ref .
@xcite for fe to model the interactions between the atoms .
these potentials take into account the possibility to have an atomic coordination different from bulk . to generate the geometrical configurations , we started with an ideal fcc lattice for co , ni , and cu and an ideal bcc lattice for fe where the crystal direction @xmath22 lies parallel to the @xmath23 axis , coinciding with the transport and elongation direction .
for the simulations , we first divided the geometry into three parts : two electrodes and a central wire , attached to them ( see fig .
[ fig - structure ] ) .
the electrodes consist of 661 ( 321 ) atoms for fcc ( bcc ) each that are kept fixed during the md calculations .
the wire is made up of 563 ( 275 ) atoms that follow the newtonian equations of motion .
we assume a canonical ensemble and use the velocity verlet integration scheme @xcite .
the simulated wires have an initial length of 0.57 nm for fe , 0.75 nm for co , and 0.73 nm for ni and cu .
the starting velocities of the atoms in the wire were chosen randomly with a gaussian distribution to yield an average temperature of @xmath24 k. because of this randomness , every elongation calculation evolves differently , while a nos - hoover thermostat keeps the temperature fixed @xcite . to relax the system
, the wire gets equilibrated for 0.1 ns at the beginning of the elongation process .
finally , the elongation process is simulated by separating one electrode from the other at a constant velocity of 0.4 m / s . during this process ,
every 10 ps the geometry is recorded .
a stretching process needs a total simulation time of about 5 ns , until the contact breaks . _
transport calculations_. once the geometries of the atomic wires are determined through the md simulations , we use them to calculate the conductance and the shot noise within the landauer - bttiker formalism . within this formalism
all linear response transport properties are determined by the transmission function . to compute this function
one needs , first of all , a proper description of the electronic structure of the metallic atomic contacts . for this purpose ,
we have employed the non - orthogonal slater - koster tight - binding parameterization of refs .
@xcite , which has been quite successful in determining a variety of properties of these atomic wires @xcite and it is known to accurately reproduce the band structure and total energy of bulk ferromagnetic materials @xcite . in this parameterization one takes into account the relevant valence orbitals of the material under study . for the four materials studied in this work ( fe , co , ni , and cu ) , the atomic basis includes the @xmath0 , @xmath25 , and @xmath26 orbitals . for the ferromagnetic materials ,
the on - site energies and hopping matrix elements depend on the electron spin and the model describes two independent sets of spin bands . in other words , there is no spin mixing in our model , as in the classical two - current model , which means in particular that we do not consider the spin - orbit interaction .
this also means that we do not consider here the possibility of having magnetic domains in the atomic contacts . within this tight - binding model ,
the hopping and overlap matrix elements are parametric functions of the distance between the atoms , which allows us to combine it with our md simulations .
we compute the transmission in the framework of our tight - binding model by making use of nonequilibrium green s function techniques , as described in refs .
briefly , as in the md simulations , the system is divided into three regions for the transport calculations , _ i.e. _ , the upper and lower electrodes and the central wire ( see fig . [ fig - structure ] ) . as the local environment of the atoms in the central part is very different from that in the bulk ,
we enforce the charge neutrality for all the atoms of the wire @xcite .
such a neutrality condition is typically a good approximation for metallic systems .
the electrodes are considered to be semi - infinite perfect crystals .
their surface green s functions are computed with the help of a decimation technique @xcite , and we use the same tight - binding parameterization as for the central part to determine their electronic structure .
it is worth stressing that the green s function techniques allow us to compute not only the transmission function , but also the spin - resolved transmission eigenvalues , @xmath27 , the analysis of which provides additional physical insight . within the landauer - bttiker formalism
the conductance @xmath28 can be expressed in terms of @xmath29 at the fermi energy as @xcite @xmath30 where @xmath31 is the spin . at low temperatures and within the linear regime ,
the zero - frequency shot noise is given by @xmath32 , where @xmath33 is the bias current .
the fano factor @xmath34 describes the noise suppression with respect to its full poissonian value of @xmath35 , and is given by @xmath36 to conclude this section , let us introduce the concept of minimum cross section ( mcs ) , which provides a measure of the contact size and the number of atoms in the narrowest part of the wire .
as we shall see , this is a useful concept , but it must be acknowledged that there is no unambiguous way to define this quantity in an atomic - scale wire . in our simulations , we define and obtain this quantity as follows .
we first superimpose our atomic structure with a fine - meshed grid with a grid spacing of about 0.02 nm . in analogy to a sandglass
, we compute the total flow through the contact , which consists of the flows through all ( one or more ) bottlenecks connecting the two ends of the wire . dividing the total flow by the possible flow through one atom
, one gets the sum of the areas of all bottlenecks as function of atomic cross sections .
to characterize the differences in transport and structural properties of fe , co , ni and cu atomic contacts , we have measured their dc conductance ( @xmath37 ) as function of applied piezo voltage , during repeated breaking and reforming of the contact .
figure [ cond - histo](a ) shows examples of conductance traces recorded in the last stages of the breaking process of ni contacts . as the minimal cross section of the contact
is reduced , the conductance exhibits a sequence of plateaus separated by sudden jumps to lower conductance values .
the plateaus correspond to relatively stable atomic configurations , which are separated by jumps due to sudden atomic rearrangements that occur once sufficient stress is accumulated @xcite .
further elongation leads to rupture of the atomic contact and to a corresponding drop in the conductance .
the small , though non - vanishing measured conductance arises from electron tunneling between the broken atomic tips .
conductance traces obtained from our simulations of ni contacts are presented in fig .
[ cond - histo](b ) .
as one can see , they nicely reproduce the slope of the conductance plateaus as well as the transition to the tunneling regime .
the observed variation from trace to trace stems from the different atomic geometries probed during each breaking cycle . to obtain a statistical picture of the typical conductance values characterizing each metal ,
we construct conductance histograms by collecting conductance values from thousands of experimental conductance traces .
figures [ cond - histo](c - f ) show the conductance histograms obtained from experimental traces ( blue lines ) .
the histograms show different sets of peaks , which are interpreted as the conductance of frequently occurring atomic configurations @xcite . for all ferromagnetic metals
we find in common that the conductance peaks are located above the quantum of conductance ( @xmath9 ) and that they exhibit a considerable width ( fwhm @xmath38 ) .
these observations are in good agreement with previously reported measurements in inert environments by untiedt _
in contrast , in the case of the monovalent metal cu , the conductance histogram ( fig .
[ cond - histo](f ) ) exhibits a series of relatively narrow peaks close to multiples of @xmath9 .
our results therefore indicate the absence of any kind of conductance quantization ( either integer or half - integer ) for ferromagnetic contacts . for comparison with the experimental data , we have computed conductance histograms from 100 breaking simulations for every metal ( grey areas in fig .
[ cond - histo](c - f ) ) .
as one can see , there is generally a good agreement between theory and experiment in the locations and widths of the peak features for the different metals .
one exception is the observed shift in the peak position of the simulated fe histogram with respect to experiment . a possible explanation for this difference could be that the fe potential , created for the crystalline phase with defects and for the liquid phase , is not able to provide a good enough description of the low - coordinated environment near the ending of the rupture process , despite its advantages from the embedded atom method ( eam ) formalization .
we tried different eam potentials and different elongation directions , but the agreement with the experimental data did not improve significantly .
nevertheless , the reproduction of most of the experimental features indicates that our simulations provide a good description of the structural and transport properties of the examined atomic contacts .
this is reinforced by the good agreement found for the shot noise results , as we shall discuss in the next section .
to understand the features observed in the conductance histograms , we turn to the analysis of the relation between conductance and structure in the simulated atomic contacts .
an important structural parameter that is correlated with the transport properties is the minimal cross section ( mcs ) of the contact , whose precise definition was provided at the end of section [ sec - methods - thy ] .
figures [ cond - vs - mcs](a , b ) show 2d density plots of the mcs and conductance obtained from the contact geometries probed in our simulations for ni and cu . from these figures , we can see that the conductance of the last peak can be mainly ascribed to a contact with a single - atom constriction . for ni
, the conductance of the last peak ( @xmath39-@xmath40 ) corresponds to a mcs of @xmath41-@xmath42 , while for cu the conductance range ( @xmath43-@xmath44 ) corresponds approximately to mcs values of @xmath41-@xmath42 .
we have obtained similar conclusions for fe and co contacts . for all ferromagnetic contacts
we find a large variance in the conductance for a given mcs value as compared with cu .
this observation indicates that the conductance of ferromagnetic contacts is more sensitive to the exact atomic configuration than in noble metals .
moreover , notice that for a given value of the mcs , the conductance of ni contacts ( and also fe and co contacts , not shown here ) is higher than that of cu contacts , which is clearly at variance with what happens in macroscopic wires of these materials .
the calculated conductance is indicated next to each contact . ]
the lower panels of fig .
[ cond - vs - mcs ] show three typical configurations of ni contacts exhibiting conductance values in the range of the main conductance peak .
we find two types of configurations which can be defined as single - atom contacts : a single atom bonded to two or more atoms on each side ( left contact ) and contacts that show two under - coordinated atoms with a single bond on one side ( central contact ) , which we will refer to as monomer and dimer , respectively .
note that configurations that can not be defined as a single - atom contact ( see right contact ) also contribute to the conductance peak . in any case
, the fact that these three different geometries contribute to the ni conductance peak shows that it is not straightforward to establish a one - to - one correspondence between conductance peaks and atomic structures , especially in the case of multivalent metals .
a similar variability is also found in the atomic configurations corresponding to the peaks in fe and co. in the case of co , for which two peaks are observed , we find that the examined configurations contributing to the lower conductance peak can be mainly identified as single - atom contacts ( _ i.e. _ monomer and dimer configurations ) , while also other configurations can contribute to the higher conductance peak .
interestingly , figs .
[ cond - vs - mcs](a , b ) show a clear linear relation between the conductance and mcs , reflecting the dependence expected from sharvin s conductance formula for a ballistic constriction @xcite .
the observation that the linear relation extends almost all the way down to the single - atom level may seem quite surprising since , as we have seen above , the conductance is highly sensitive to small changes in the atomic geometry .
the results of figs .
[ cond - vs - mcs](a , b ) also show that there is a considerable variance of the conductance for a given mcs value .
thus for instance , the conductance for a ni contact with a mcs of 1 atom can vary between @xmath8 and @xmath45 .
therefore , the scaling of the mcs should be understood for the average conductance of many contact geometries .
) curve measured for a ni atomic contact before ( blue ) and after ( red ) shot noise measurements .
( b ) a series of noise spectra recorded at different bias voltages .
black lines indicate the frequency window which was selected for obtaining the average noise power .
( c ) average noise power as function of bias voltage ( d ) dependence of the reduced values @xmath46 ( blue ) calculated for the measured noise in ( c ) , and a linear fit ( red ) , giving @xmath47 according to eq .
( [ eq - y ] ) . ] to further investigate the transport properties of ferromagnetic contacts , we have conducted measurements of the electronic shot noise generated by the contacts .
let us remind that within the landauer - bttiker framework , shot noise depends on the number of open conduction channels and their transmission probabilities , see section iib .
the overall zero - frequency current noise generated by a quantum conductor ( including the thermal noise ) can be expressed as @xcite @xmath48 , \ ] ] where @xmath28 and @xmath34 are the conductance and fano factor , respectively [ see eqs .
( [ eq - g - s ] ) and ( [ eq - fano ] ) ] , and @xmath49 describes the ratio between bias voltage @xmath50 and temperature @xmath51 . near equilibrium conditions ( @xmath52 )
, @xmath53 reduces to the johnson - nyquist expression for thermal noise @xmath54 . at high bias ( @xmath55 )
, the current noise depends linearly on the bias current , @xmath56 .
figure [ fig - noise ] shows an example shot noise measurement for a ni atomic contact .
differential conductance spectra ( @xmath16 ) of the contact , see fig .
[ fig - noise](a ) , are recorded before and after the noise measurement in order to confirm that the junction has remained stable during the measurement .
the zero bias conductance of the junction is determined from the average differential conductance in the window of @xmath57 mv .
figure [ fig - noise](b ) shows a series of noise spectra for different applied bias .
each spectrum is obtained from the fourier transform of voltage fluctuations produced by the junction , and averaged for 5000 consecutive measurements . the voltage noise produced by the amplifier
was measured separately and subtracted from the recorded spectra .
the spectra were corrected to account for low pass filtering due to capacitance of the cabling and amplifier input capacitance ( total capacitance of @xmath58 pf ) .
the noise power is averaged in a frequency window , which is selected to be high enough to reduce @xmath59 noise contributions .
figure [ fig - noise](c ) shows the average noise power as function of bias voltage across the junction .
following ref .
[ ] , eq .
( [ eq - si ] ) can be expressed as @xmath60 where @xmath61 and @xmath62 / s_i ( 0)$ ] .
the fano factor is obtained by calculating the reduced parameters @xmath46 and obtaining a linear fit of @xmath63 according to eq .
( [ eq - y ] ) , as shown in figure [ fig - noise](d ) .
figure [ fig - fano ] shows the distribution of @xmath34 as function of zero - bias conductance @xmath28 obtained for fe , co , ni , and cu atomic contacts . for cu , a suppression of the fano factor towards its lower limit for spin - degenerate transmission ( light shaded area )
is observed .
this behavior is similar to the results obtained for the monovalent metals au and ag , indicating conductance quantization , _
i.e. _ , that in multiples of @xmath9 the conductance takes place through nearly fully - open channels @xcite .
conversely , a distinct qualitative picture arises for ferromagnetic contacts .
the measured values of @xmath34 are found to be significantly higher compared to cu . in particular , the large suppression at multiples of @xmath9 observed for cu is clearly missing for the ferromagnetic contacts .
the conductance and fano factor depend on the distribution of conduction channels carrying the current , in accordance with eqs .
( [ eq - g - s ] ) and ( [ eq - fano ] ) .
this dependency allows us to draw several conclusions regarding the set of transmission probabilities @xmath64 .
first , one can determine the minimum number of channels contributing to transport according to the position of the measurement in the @xmath65 space .
the solid lines starting from @xmath66 and ending at @xmath67 in fig .
[ fig - fano](a - c ) indicate the maximum fano factor that can be obtained for @xmath68 spin channels ( _ i.e. _ having a maximum conductance of @xmath69 ) .
therefore , given a combination of @xmath65 one can obtain a lower bound for the number of transmitting channels ( indicated within the regions between each two lines in fig .
[ fig - fano](c ) ) .
for ni , one can see that a minimum of @xmath70-@xmath71 spin channels contribute to transport for contacts with conductance values corresponding to configurations with single - atom cross section ( _ i.e. _ @xmath39-@xmath40 , see fig .
[ cond - histo](e ) ) .
similarly , the results for co and fe indicate that the current through a single - atom contact is carried through a multiple number of channels .
to study the transmission values @xmath64 , we use a numerical analysis introduced in ref . [ ] .
the analysis is based on enumerating the possible combinations of @xmath64 up to a certain number of channels @xmath68 ( here , @xmath72 was chosen to account for the main channels ) , and identifying the range of solutions that give the measured @xmath34 and @xmath28 . in many cases , this analysis allows us to determine the transmission probabilities with reasonable accuracy .
we note that since eqs .
( [ eq - g - s ] ) and ( [ eq - fano ] ) are symmetric with respect to spin , the spin direction can not be determined using this method .
the transmission probabilities are labeled by a single index @xmath73 , and are ordered according to decreasing transmission .
figure [ fig - fano](d ) shows the possible values of @xmath64 for three different combinations of @xmath34 and @xmath28 for ni contacts , which are indicated in fig .
[ fig - fano](c ) . for combination 1 ( bottom panel ) ,
in which the fano factor is near its minimum limit , two fully - open spin channels carry most of the current . in the case of cu , for which spin polarization is not expected , measurements located at this position indicate transport dominated by two channels ( one for each spin type ) with identical transmission . as can be seen from examples 2 and 3 ( middle and upper panels in fig .
[ fig - fano](d ) ) , an increase in @xmath34 results in a larger number of partly - open channels carrying the current .
this trend appears for any conductance range ; a low fano factor near the minimum limit indicates a channel distribution with a minimal number of channels , and maximal number of open channels ( @xmath74 ) , while a higher @xmath34 results in a distribution with a larger number of channels and lower transmission values .
thus , the large spread in @xmath34 observed for fe , co , and ni implies that there is no conductance quantization , while the number of channels and their transmission probabilities vary significantly between different contact realizations .
our analysis of conductance traces and shot noise measurements indicates that the conductance and channel distribution in ferromagnetic contacts are highly sensitive to atomic geometry .
this sensitivity makes the comparison between theory and experiment challenging .
interestingly , the structure - conductance analysis obtained from our simulations ( fig .
[ cond - vs - mcs ] ) , shows that the relative variance of the conductance in respect to the mean value ( @xmath75 ) is greatly reduced with increasing mcs .
thus , the study of larger contacts could be advantageous as the sensitivity to atomic geometry is significantly reduced .
unfortunately , the conductance histograms for ferromagnetic contacts do not show clear features at values above @xmath76 . on the other hand
, the study of shot noise at larger conductance values could potentially provide new information , allowing a more straightforward comparison with theory . to investigate this possibility ,
we have extended our measurements and simulations to contacts with higher conductance values , focusing on ni and cu for comparison between a ferromagnetic metal and a monovalent metal .
figure [ fig - fano - sat](a ) shows the distribution of the fano factor as function of conductance measured up to @xmath21 , for ni and cu contacts .
remarkably , for both metals , the variance in the fano factor is strongly reduced for larger contacts , and the fano factor saturates to a nearly constant value of @xmath77 and @xmath78 for ni and cu , respectively .
figure [ fig - fano - sat](b ) shows the @xmath65 distribution obtained from our simulations .
the results are found to be in very good agreement with the experimental values , both in the large variance at low conductance , and in the saturation at higher conductance values .
the simulated saturation values obtained yield @xmath79 and @xmath80 for ni and cu , respectively .
our simulations for fe and co result in similar behavior to that obtained for ni , both giving the same saturation value of @xmath81 , see appendix a. for au , having a monovalent orbital structure similar to cu , the fano factor obtained by our simulation reaches a value of @xmath82 .
we note that previous measurements for au contacts have indicated an average @xmath34 of @xmath83 for @xmath84 @xcite .
thus , the clearly larger fano factor values encountered in the ferromagnetic metals as compared to the noble ones seem to be a generic feature , which reflects the different orbital structure of these two kinds of metals .
this observation will be further discussed in the following section .
the goal of this section is to elucidate the transport mechanism in ferromagnetic atomic contacts by analyzing the electronic structure and conduction channels of our simulated atomic contacts .
we shall show that the fundamental differences between noble and ferromagnetic metals are related to the significant role that @xmath1 orbitals play in the transport properties in the latter case .
furthermore , we find that the exchange - splitting of the @xmath1 orbitals results in distinct transport for majority - spin and minority - spin electrons . for didactic reasons , in this section we focus on the comparison between ni and cu .
we find that the conclusions drawn for ni are qualitatively valid for fe and co. our theoretical results for these metals are shown in appendix a. figure [ fig - dos ] shows a comparison of the energy dependence of both the density of states ( dos ) and the transmission function for ni and cu , as calculated from our tight - binding model .
the upper panels show the bulk dos , while the middle panels display the local dos projected onto the central atom for a single - atom contact configuration .
the contributions of @xmath0 , @xmath25 and @xmath26 orbitals to the dos are shown in separate curves . in the case of ni , we show separately the contributions of the majority - spin ( or spin - up ) electrons and minority - spin ( or spin - down ) electrons to both dos and transmission . comparing these figures , one can immediately see that the main difference between ni and cu is the large contribution of @xmath1 orbitals at the fermi energy for ni .
this contribution is particularly large in the case of minority - spin electrons , for which the @xmath1 states are located higher in energy due to the spin - splitting for the bulk @xmath1 bands .
the lower panels of fig . [ fig - dos ] show the total transmission as a function of energy for four different contacts with different cross sections obtained in an individual simulation of the breaking process of a ni and a cu wire .
the corresponding contact geometries are shown in the left ( ni ) and right ( cu ) columns .
as one can see in fig .
[ fig - dos](c , f ) , the corresponding spin - resolved total transmission follows closely the energy dependence of the total dos .
thus for instance , the total transmission at the fermi energy , @xmath85 , for the minority - spin electrons is significantly higher than that of the majority - spin electrons , in correlation with the higher dos at @xmath86 in the former case .
these results strongly suggest that @xmath1 orbitals are responsible for the significantly higher transmission for minority - spin electrons .
importantly , we find that the key role of the @xmath1 orbitals in transport properties also extends to atomic contacts with larger contact sizes . to further study the transport mechanism in ferromagnetic contacts , we examine the distribution of conduction channels .
figure [ fig - channels ] shows the average value of the transmission probabilities as a function of conductance as obtained from 100 contact stretching simulations for ni ( spin up and down ) and cu . in the case of ni
, one can immediately see that there is a larger number of channels contributing to transport for minority - spin electrons .
for example , for conductance values related to a single - atom ni contact ( _ i.e. _ @xmath39-@xmath40 ) , there are approximately 4 - 7 minority - spin channels compared to 2 - 4 majority - spin channels .
the larger number of minority - spin channels can be correlated with the higher number of the minority - spin @xmath1 states at @xmath86 ( fig .
[ fig - dos](d , e ) ) , which are available for conduction electrons . in the case of cu ( fig .
[ fig - channels](c ) ) , we see that transport is dominated by channels with nearly perfect transparency which open one by one with increasing conductance .
this behavior naturally explains the suppression of shot noise at multiples of @xmath9 observed for this metal .
as the conductance increases ( or equivalently the contact size ) more partially - open channels appear , explaining why the fano factor does not vanish for contacts with conductance above @xmath87 .
interestingly , a tendency for transport through nearly - open channels is also observed for ni spin - up electrons ( fig .
[ fig - channels](a ) ) .
however , as the transport is mainly carried by spin - down channels with intermediate transmission values , one does not observe any strong suppression of the fano factor at any particular value of the conductance . .
we show the results for both majority - spin ( spin - up ) electrons and minority - spin ( spin - down ) electrons .
( b ) the corresponding results for a cu with @xmath88 . in both panels the dotted line shows the bimodal distribution of eq .
( [ eq - bimodal ] ) . ] to understand the origin of the saturation of the fano factor and the differences in the saturation values between cu and the ferromagnetic metals , we show in fig .
[ fig - distr ] histograms of the transmission eigenvalues for ni and cu contacts with a conductance of @xmath21 . as one can see in fig .
[ fig - distr](a ) , the spin - up and spin - down transmission coefficients are distributed very differently . in the former case ,
the distribution shows a preference to open channels . in contrast , for minority spins , the distribution is mainly composed by partially open channels . as discussed in section [ sec - noise ] , the fano factor increases as the conductance is determined by a larger number of partially - open channels .
thus , from the transmission histograms , we can conclude that the lower value of the fano factor at saturation in the case of cu results from the dominant role of nearly fully open channels , while for ni this value is significantly larger due to the presence of numerous partially open ( minority - spin ) channels .
the results presented so far indicate that transport for ni spin - up and spin - down electrons differs not only in the total transmission value , but exhibits fundamentally distinct behaviors .
the transmission distribution for the majority spins clearly resembles that of the cu contacts ( fig .
[ fig - distr](b ) ) , in which transport is dominated by highly transmissive channels .
this behavior can be traced back to the fact that in both cases ( cu and ni spin up ) , the states available for transport originate mainly from the @xmath2 valence orbitals . on the other hand , the channel distribution of the ni minority spins is very similar to that of the transition metal pt @xcite , with dominant @xmath1 valence orbitals near the fermi energy .
thus , we see that ni behaves in some sense as a combination of monovalent and transition metal conductors in parallel .
as it is well known in mesoscopic physics , the fano factor of a metallic diffusive wire is universal ( it does not depend on the degree of disorder or on the wire geometry ) and adopts a value equal to @xmath89 @xcite . this value can be understood in terms of the distribution of transmission coefficients , which in the diffusive wire case reduces to the so - called bimodal distribution @xcite @xmath90 where @xmath91 is the probability density to find a given value of the transmission @xmath92 and @xmath28 is the wire conductance . to understand the differences between our atomic contacts and the diffusive wire case , we show in the histograms of fig .
[ fig - distr ] the bimodal distribution of eq .
( [ eq - bimodal ] ) as a solid line . as one can see in fig .
[ fig - distr](a ) , the distribution of the ni minority spins follows the bimodal distribution .
indeed , the fano factor for the minority - spin electrons @xmath93 yields 0.28 , 0.30 and 0.34 for ni , co and fe , respectively .
these values are very close to the universal value of 1/3 .
thus , our results suggest that ferromagnetic atomic contacts show a diffusive - like behavior for minority - spin electrons .
on the other hand , the transmission distributions for the ni majority spins and for cu differ markedly from the bimodal distribution , the main difference being the presence of a larger number of channels with very high transmissions .
this fact leads to a reduction of the fano factor as compared to the diffusive case . in this respect , we find that the fano factor for the ni majority - spin electrons is @xmath94 , while @xmath95 for co , and @xmath96 for fe .
these values are closer to the saturation value for cu , @xmath97 .
we will further discuss these results in the following section .
another aspect of the conduction across ferromagnetic atomic contacts that differs dramatically from the macroscopic case is the spin polarization of the current .
this quantity is defined as @xmath98 where @xmath99 is the current carried by the majority - spin electrons and @xmath100 is the current carried by the minority - spin electrons . in the case of macroscopic wires ,
mott s model predicts a positive sign for the spin polarization because transport is dominated by majority - spin electrons , as explained in the introduction .
the positive spin polarization was found to be consistent with studies of sub - gap structure measurements of ferromagnet - superconductor interfaces @xcite .
in contrast , following our discussion above , it is clear that in the case of atomic - size contacts , the current is dominated by minority - spin electrons.the influence of this property is illustrated in fig .
[ fig - spinpol ] , showing the current spin polarization as a function of the conductance .
as one can see , the spin polarization is negative , irrespective of the conductance value , and it saturates to a value of @xmath101 for large contacts . interestingly , while the presence of spin - down @xmath1 states at the fermi energy reduces the minority - spin conductivity in the bulk case , for atomic contacts the situation is precisely the opposite : additional minority - spin channels with @xmath1 character become available for transport , resulting in an opposite sign of the spin polarization .
our results from both experiment and theory clearly indicate that @xmath1 orbitals are responsible for fundamental differences in transport properties of ferromagnetic contacts compared to monovalent contacts .
ferromagnetic contacts clearly differ from cu ( and other monovalent metals ) by : ( i ) higher conductance , ( ii ) larger variance in the conductance and shot noise ( for small contacts ) , ( iii ) existence of multiple partly - transmitting channels , and ( iv ) nearly double fano factor saturation value for large contacts .
as our channel analysis reveals , transport in ferromagnetic contacts can be viewed to occur in parallel through spin - up channels , behaving similar to a monovalent metal , and spin - down channels , acting as a transition metal .
thus , the aforementioned properties can be mainly attributed to the contribution of minority - spin electrons , for which @xmath1 states are available for transport .
we will therefore divide our discussion into @xmath2-dominated transport ( _ e.g. _ cu , ni spin up ) and transport through mixed @xmath102 states ( _ e.g. _ ni spin down ) . in the case of @xmath2-dominated transport , conductance quantization , _
i.e. _ , transport through fully - open channels , takes place only for small contacts ( @xmath103 ) , whereas for larger contacts , an increasing contribution of partially - open channels is observed .
this behavior was reasonably reproduced by a model of a free - electron constriction with disorder , connecting two gradually narrowing contacts @xcite .
the mentioned disorder can be associated with local impurities , scattering at the contact surfaces or lack of periodical lattice structure .
indeed , one could naturally expect that as the constriction size increases , the series resistance induced by disorder will play a more dominant role in transport , obscuring the measurement of quantized values of conductance .
we argue that the saturation of the fano factor could be viewed as an interplay between the quantization of the contact on one hand and some amount of disorder which results in imperfect transmissions .
these arguments explain why the average value of the fano factor ( @xmath15 0.15 ) is significantly lower than the value of 1/3 expected for a diffusive contact . in the case of transport with @xmath102-character
, our results suggest that disorder is significantly higher compared to a monovalent conductor .
we find this expressed in the bimodal distribution observed for spin - down ni channels ( and the corresponding @xmath104 ) , being a characteristic of diffusive transport .
a higher disorder in ferromagnetic contacts can be related to the inherent anisotropy of @xmath1 orbitals compared to the high - symmetry of @xmath2 orbitals .
the interplay between the disorder in diffusive wires and a ballistic quantum point contact has been studied theoretically by beennakker and melsen @xcite .
the model analyzed in that work was able to recover the saturation of the fano factor to 1/3 with increasing contact size , however the channel distribution obtained for narrow contacts is at variance with the partially - open channel distribution we obtain .
this discrepancy indicates the need to consider the specific orbital structure of the contacts .
we note that riquelme _
et al . _
@xcite reported an experimental analysis of the transmission distribution of pb atomic contacts with conductance values ranging between @xmath8 and @xmath21 .
the study was based on the use of superconductivity and the analysis of the sub - gap current , in the spirit of refs .
the authors concluded that as the contact size increased , the transmission distribution approached very quickly the universal bimodal distribution .
those authors also presented a theoretical analysis based on ideal geometries and a tight - binding model similar to the one employed here that suggested that this behavior could be associated with the @xmath105 valence orbital structure of pb that gives rise to a significant contribution of partially - open channels even in the absence of atomic disorder .
this conclusion , together with our results , suggests that such behavior may be a general characteristic of multivalent metals .
when discussing diffusive transport in atomic contacts , it is important to consider the conductor dimensions .
the strict definition of diffusive transport requires that the length of the conductor satisfies @xmath106 , where @xmath107 is the elastic mean free path . in the case of atomic contacts made of the monovalent metal au , different estimates of @xmath107
give a value of approximately 4 - 5 nm @xcite . in comparison ,
the length of the contacts in our simulations is up to 3 nm in total . since the saturation of the fano factor
is captured in our simulated structures , we infer that this region is sufficient to describe the main transport characteristics . thus , it is rather unlikely that the condition @xmath106 for diffusive transport is met .
yet , we can find two reasons why diffusive behavior is seen nevertheless .
first , the increased disorder originating from the contribution of @xmath1 orbitals could lead to a considerably lower value for @xmath107 compared to au .
second , theoretical studies have shown that it is sufficient to have a small number of tunneling barriers in series to result in fano factor near 1/3 @xcite .
these considerations can explain why the bimodal channel distribution can appear even in contacts of several atoms in cross section .
to summarize , we have presented a comprehensive experimental and theoretical study of the conductance and shot noise in ferromagnetic atomic contacts made of ni , co , and fe , and we have compared the results with those for the nobel metal cu .
our experimental results reveal clear differences between the ferromagnetic contacts and those made of cu such as the absence of any type of conductance quantization and the larger values of the fano factor for any conductance value ( including the saturation value for large contacts ) in the ferromagnetic case .
our theoretical results , which are able to satisfactorily reproduce our main experimental observations , clearly show that the transport properties of the ferromagnetic contacts can be explained in the framework of quantum coherent transport .
our results demonstrate that the @xmath1 orbitals ( especially those of the minority - spin electrons ) play a fundamental role in the transport through ferromagnetic atomic contacts .
the contribution of these @xmath1 orbitals leads to the appearance of partially - open channels , which explains the absence of conductance quantization and the large values of the fano factor , as compared to noble metals .
moreover , the contribution of minority - spin channels makes the ferromagnetic contacts more conductive than the noble metallic contacts and leads to negative values of the spin polarization , both observations in stark contrast with the behavior of macroscopic metallic wires .
thus this work provides a textbook example of how the transport properties of metallic wires can drastically change upon shrinking their characteristic dimensions to the atomic scale , a change that is due to modification in the dominant transport mechanism .
the authors thank a. halbritter for fruitful discussions . m.m . and p.n .
acknowledge financial support from the sfb767 and computer time granting from the nic and the bwhpc framework program of the state of baden - wrttemberg .
j.c.c . acknowledges financial support from the spanish ministry of economy and competitiveness ( contract no .
fis2014 - 53488-p ) .
thanks the harold perlman family for their support and acknowledges funding by the israel science foundation and the minerva foundation .
for completeness , we display in fig .
[ fig - co - fe ] our theoretical results for the fano factor and the spin - resolved channel distributions for the co and fe contacts .
as explained in the main text , these results are qualitatively similar to those of ni and confirm the general conclusions drawn on the nature of the transport properties of ferromagnetic atomic contacts .
we also show in fig .
[ fig - spinpol - co - fe](a , b ) the corresponding results for the current spin polarization in co and fe contacts .
notice that the co case is very similar to the ni ( fig .
[ fig - spinpol ] ) , while in the fe case the current spin polarization exhibits a sign change as a function of the contact size and it adopts rather small values for large contacts .
an analysis of the channel distribution for fe , see fig .
[ fig - co - fe](e , f ) , suggests that this behavior is due to a reduced contribution of the minority - spin @xmath1 bands , as compared with the other two ferromagnetic metals .
another feature in these results that is worth remarking is the fact that , as in the ni case , the fluctuations in the current polarization are particularly large in the tunnel regime ( @xmath108 ) , when the contacts are already broken . | with the goal to elucidate the nature of spin - dependent electronic transport in ferromagnetic atomic contacts , we present here a combined experimental and theoretical study of the conductance and shot noise of metallic atomic contacts made of the @xmath0 ferromagnetic materials fe , co , and ni . for comparison , we also present the corresponding results for the noble metal cu .
conductance and shot noise measurements , performed using a low - temperature break junction setup , show that in these ferromagnetic nanowires : ( i ) there is no conductance quantization of any kind , ( ii ) transport is dominated by several partially - open conduction channels , even in the case of single - atom contacts , and ( iii ) the fano factor of large contacts saturates to values that clearly differs from those of monovalent ( nonmagnetic ) metals .
we rationalize these observations with the help of a theoretical approach that combines molecular dynamics simulations to describe the junction formation with nonequilibrium green s function techniques to compute the transport properties within the landauer - bttiker framework .
our theoretical approach successfully reproduces all the basic experimental results and it shows that all the observations can be traced back to the fact that the @xmath1 bands of the minority - spin electrons play a fundamental role in the transport through ferromagnetic atomic - size contacts .
these @xmath1 bands give rise to partially open conduction channels for any contact size , which in turn lead naturally to the different observations described above .
thus , the transport picture for these nanoscale ferromagnetic wires that emerges from the ensemble of our results is clearly at variance with the well established conduction mechanism that governs the transport in macroscopic ferromagnetic wires , where the @xmath1 bands are responsible for the magnetism but do not take part in the charge flow .
these insights provide a fundamental framework for ferromagnetic - based spintronics at the nanoscale . |
the quest for neutrinoless double beta decay ( @xmath1 ) has experienced an interesting history @xcite . in 1939 , combining the idea of double beta decay ( @xmath2 ) proposed by goeppert - mayer four years ago and the notion of majorana fermions formulated by majorana two years ago , furry for the first time discussed the possibility of observing two electrons but without any neutrinos during the transformation of the @xmath3 nucleus to the @xmath4 nucleus , i.e. , @xmath5 . at that time , simply based on the phase space argument , this @xmath1 mode was believed to have a half - life around @xmath6 , much shorter than that of the corresponding @xmath2 mode , @xmath7 . if such an estimation were true , one would have discovered @xmath1 quite shortly , maybe by the 1950s .
the first turning point in the @xmath1 history is the discovery of parity violation in 1957 and the establishment of v - a theory , which concludes that if neutrinos were massless , @xmath1 would be exactly forbidden .
such a discouraging result declared that the experimental search for @xmath1 entered the middle ages " @xcite . although @xmath1 experiments were still performed in the 1960s and 1970s , the main goal was to test the lepton number conservation law . only until the early 1980s
the search for @xmath1 embraced its renaissance " period @xcite , mainly because the grand unified theories developed around that time can naturally generate a majorana mass term for neutrinos , and neutrinos with tens of ev masses were thought to be possible for being a good dark matter candidate in cosmology .
in the current contemporary ages " @xcite , the observation of neutrino oscillation , which indicates that neutrinos have masses , really boomed this field .
several experiments have been built , or are under construction , and dozens of proposals are made for future investigations .
such dedicated experimental efforts may enable us to extract rich physics results from @xmath1 in a near future .
for example , * pinning down the nature of neutrinos . according to schechter - valle theorem @xcite , the observation of @xmath1 would imply a majorana mass term for neutrinos , proving the majorana nature of neutrinos . * constraining the absolute neutrino mass scale and lepton cp - violating phases .
this is because that if @xmath1 is mediated by the light neutrino majorana mass term , the decay half - life @xmath8 can then be written as @xmath9 where @xmath10 and @xmath11 stand for the phase - space factor @xcite and the nuclear matrix element ( nme ) , respectively , and @xmath12 is the so - called effective neutrino mass .
although there still exist large uncertainties in the calculation of nme @xcite , in principle , one can obtain the former two accurately within nuclear theory .
then , the measured limit on the half - life @xmath8 would directly yield a constraint on @xmath12 , whose definition is given by @xmath13 where @xmath14 s are neutrino masses , and @xmath15 s and @xmath16 s are the mixing angles and cp - violating phases in the lepton mixing matrix , respectively . * determining the neutrino mass ordering .
this is due to the presence of a lower bound for the above effective neutrino mass @xmath12 in the inverted neutrino mass ordering , according to the current global fit results @xcite . *
restricting new physics contributions .
as @xmath1 is a rare process , any new physics that contributes to it should be constrained by the obtained data . in this
talk we focus on the question of determining neutrino mass ordering in the next generation @xmath1 experiments @xcite .
our goal is to provide a quantitative description of experimental power to distinguish between normal neutrino mass ordering ( no ) and inverted neutrino mass ordering ( io ) . for simplicity ,
we only consider @xmath0ge - type experiments , as their background is found to be nearly flat in the signal region .
moreover , we adopt bayesian analysis as our statistical method . *
large amount of @xmath17-decaying isotopes .
the number of signal events @xmath18 is directly related to the fiducial mass of @xmath17-decaying isotopes .
taking @xmath0ge for an example , we have @xmath19 where @xmath20 is the avogadro s constant , @xmath21 is the molar mass of @xmath0ge , and @xmath22 and @xmath23 are the exposure and detection efficiency , respectively . since the exposure @xmath22 is defined as the product of the fiducial mass and the experiment operating time , increasing the mass of isotopes is helpful to reduce the detection time . in table
[ tb : classification ] we list the typical sizes of the fiducial mass in the previous , current and next generation @xmath1 experiments . *
good energy resolution .
ideally , all the above signal events should show up at the q - value of the corresponding @xmath1 process . for @xmath0ge ,
such a q - value is 2039 kev .
however , because of a finite energy resolution , these signal events spread out and can be easily contaminated by background events , especially those from the intrinsic @xmath2 process . because @xmath0ge - type experiments employ the
detector = studied substance " scheme , a very high energy resolution can be reached . in this work
we take fwhm ( full width half maximum ) as 3 kev . * low background .
it was found that reducing the level of background is very helpful to improve the half - life sensitivity @xcite . moreover ,
if possible , one should pursue the zero background limit , so that a single @xmath1 event would be enough to pin down the majorana nature of neutrinos .
for the next generation @xmath0ge - type experiments we will choose @xmath24 counts / kev / kg / yr , with bi standing for background index .
bi s of other generations of @xmath1 experiments are also given in table [ tb : classification ] .
* good detection efficiency .
@xmath1 happens so rare that any signal event would be precious .
here we take the typical value of @xmath25 for @xmath0ge - type experiments . *
good scalability . by scalability we mean the ability to scale the experimental set - up to a larger size .
this is very important for building a large @xmath1 experiment stage by stage .
currently , we have two commissioning @xmath0ge - type @xmath1 experiments , i.e. , gerda and majorana demonstrator @xcite . in the first phase of gerda ,
a lower limit of @xmath26 yr at @xmath27 confidence level was reported for an exposure of 21.6 kg @xmath28 yr and a bi about @xmath29 @xcite .
the gerda phase - ii and majorana demonstrator are expected to increase the exposure to about 200 kg @xmath28 yr each , and to reduce bi to the @xmath30 level at the same time .
more excitingly , these two collaborations are also discussing the possibility of building a future large scale @xmath0ge ( lsge ) experiment together @xcite , which may eventually reach an exposure of around @xmath31 kg @xmath28 yr , and @xmath32 .
it is such a possibility that motivates us to perform a detailed statistical analysis , so as to find out at which level the currently undetermined neutrino mass ordering can be resolved then .
we begin with generating pseudo - data samples that are to be used in the later statistical analysis . following gerda s analysis on its first phase data @xcite
, we consider a region of spectrum that spans from 2022 kev to 2061 kev with a bin size of 1 kev .
see fig .
[ fg : spectrum ] for an example of the simulated event spectrum .
red curve represents the total un - binned signal and background events , while the binned results are given by the gray histogram , where the poisson statistics is assumed . for the @xmath33-based experiment , where the red solid curve represents the expected gaussian distribution of signal events on top of a constant background and typical values of the half life @xmath34 , the exposure @xmath35 , the energy resolution @xmath36 , and the efficiency @xmath25 have been used .
the black - dashed horizontal line corresponds to the background with @xmath37 .
the gray histograms stand for the total number of both signal and background events , which is randomly generated according to the poissonian distribution in each energy bin .
taken from ref .
[ 6 ] @xcite . ] in ref .
[ 6 ] @xcite , we indeed adopt the above generated spectrum , which incorporates statistical fluctuation , to perform analysis on the exclusion limit and discovery potential in the upcoming @xmath0ge - type experiments .
however , for the current discussion on the determination of neutrino mass orderings , using the above statistical - fluctuation - included spectrum would be very time - consuming .
this is because for each set of input parameters , one needs to generate and then analyze hundreds of pseudo - data samples so as to obtain enough statistics . for simplicity
, here we instead take the so - called asimov data set @xcite , which is obtained by simply assuming the _ expected _ number of events for each bin in the spectrum . in terms of the example given in fig .
[ fg : spectrum ] , the asimov data set just means the events under the red curve .
apparently , no statistical fluctuations are included in this asimov data set .
however , it was found that the analysis of the asimov data yields a good approximation to the _ median _ projection of experiments @xcite . for our current discussion
such a median projection is sufficient .
we adopt bayesian analysis as our statistical method .
bayesian analysis resides on the well - known bayes theorem , and describes the degree of belief in a certain hypothesis @xmath38 , given the data set @xmath39 .
specifically , bayes theorem states that @xmath40 where @xmath41 and @xmath42 are the prior and posterior probabilities of the hypothesis @xmath38 , respectively .
the probability of obtaining the data @xmath39 , given the hypothesis @xmath43 to be true , is denoted as @xmath44 , and is also called the evidence @xmath45 .
lastly , @xmath46 stands for the overall probability of observing the data @xmath39 , and it is equal to @xmath47 , because of the normalization condition @xmath48 . following the above formalism , we take our two competing hypotheses as @xmath49 and @xmath50 , and the above discussed pseudo - data samples would be our data set @xmath39 . to find out the more favorable hypothesis , we compute the following posterior odds @xmath51 where the ratio of evidences @xmath52 is termed bayes factor .
if _ a prior _ we have no preference for a particular mass ordering , the above posterior odds is then directly reflected by the bayes factor @xmath53 .
furthermore , to interpret the value of this posterior odds or the bayes factor , we adopt the jeffreys scale @xcite given in table [ tb : jeffreys ] .
the computation of @xmath54 is performed via @xmath55 ( similar computation for @xmath56 ) .
@xmath57 stands for the parameters in the hypothesis @xmath49 , and here they are three lepton mixing angles @xmath58 , two cp - violating phases @xmath59 , two neutrino mass - squared differences and the lightest neutrino mass @xmath60 .
their prior probability distributions are collectively denoted as @xmath61 . for the mixing angles and mass - squared differences , we assume gaussian priors with the central values and @xmath62 errors taken from ref.[5 ] @xcite . for cp - violating phases ,
uniform priors are chosen , and for the lightest neutrino mass @xmath60 we adopt two different priors , i.e. , a uniform prior within @xmath63~\text{ev}$ ] and a logarithmic prior on @xmath64~\text{ev}$ ] . finally , we obtain @xmath65 by fitting the pseudo - data set with the above sampled parameters in the realm of @xmath49 .
calculated by comparing between no with io , where the thick solid ( dashed ) curves correspond to the low nme @xmath66 ( the high nme @xmath67 ) .
the orange , black and blue curves represent @xmath68 , @xmath69 and @xmath70 , respectively . taken from ref . [
6 ] @xcite . ] in fig .
[ fg : nh_ih ] , the left and right panels correspond to the uniform and logarithmic priors on the lightest neutrino mass @xmath60 , respectively .
given a true value of @xmath71 and an exposure @xmath72 , we can calculate the bayes factor @xmath73 , and show its contours as orange , black and blue curves for @xmath68 , @xmath69 and @xmath70 , respectively .
thick solid and dashed curves distinguish two different choices of nme , i.e. , @xmath66 and @xmath74 for the former and later , respectively . for reference , we also plot the lower bound of @xmath75 in io , i.e. , the red horizontal line represents the result calculated by using the best - fit values of neutrino mixing parameters , while the dark ( light ) band is for the result by using @xmath62 ( @xmath76 ) ranges . * an exposure of at least @xmath77 ( or @xmath78 ) kg @xmath28 yr is needed in order to report a weak ( or moderate ) evidence for no . according to the jeffreys scale in table [ tb : jeffreys ]
, the weak and strong evidence should be understood as a degree of belief of @xmath79 and @xmath80 , respectively .
* in the next generation of @xmath0ge - based @xmath1 experiments with an exposure of @xmath81 , we are only able to reach a value of @xmath82 , indicating a moderate evidence for no . to obtain a strong evidence ( @xmath83
, one needs to further increase the exposure or reduce the level of background .
the question of whether neutrino is its own anti - particle , namely , the nature of neutrino , leads to a long history of searching for @xmath1 experimentally .
in the course of such a dedicated search , one also realizes that studying @xmath1 may help us distinguish the currently unresolved neutrino mass ordering .
motivated by the latter fact , we perform a detailed quantitative analysis , aiming at finding out at which statistical level one can exclude the inverted neutrino mass ordering in the next generation @xmath0ge - type @xmath1 experiments .
our finding indicates that the next generation @xmath0ge - type @xmath1 experiments indeed has some sensitivity to the discrimination of neutrino mass orderings , however , not in a very decisive manner .
the author thanks prof .
shun zhou for the fruitful collaboration on the subject .
he is also grateful to prof .
harald fritzsch for the kind invitation , and local organizers for the warm hospitality .
this work was supported in part by the innovation program of the institute of high energy physics under grant no .
y4515570u1 .
a. s. barabash , phys . atom .
nucl . * 74 * , 603 ( 2011 ) [ arxiv:1104.2714 [ nucl - ex ] ] .
j. schechter and j. w. f. valle , phys .
d * 25 * , 2951 ( 1982 ) .
j. kotila and f. iachello , phys .
c * 85 * , 034316 ( 2012 ) [ arxiv:1209.5722 ] .
j. engel , j. phys .
g * 42 * , no . 3 , 034017 ( 2015 ) .
m. c. gonzalez - garcia , m. maltoni and t. schwetz , jhep * 1411 * , 052 ( 2014 ) [ arxiv:1409.5439 [ hep - ph ] ] .
j. zhang and s. zhou , phys .
d * 93 * , no . 1 , 016008 ( 2016 ) [ arxiv:1508.05472 [ hep - ph ] ] . j. j. gomez - cadenas , j. martin - albo , m. mezzetto , f. monrabal and m. sorel , riv .
nuovo cim . * 35 * , 29 ( 2012 ) [ arxiv:1109.5515 [ hep - ex ] ] .
m. agostini _ et al .
_ [ gerda collaboration ] , phys .
* 111 * , no .
12 , 122503 ( 2013 ) [ arxiv:1307.4720 ] .
neutrinoless double beta decay report to the nuclear science advisory committe , apirl 24 , 2014 ( @xmath84 ) . | motivated by recent intensive experimental efforts on searching for neutrinoless double beta decays , we present a detailed quantitative analysis on the prospect of resolving neutrino mass ordering in the next generation @xmath0ge - type experiments . |
gamma - ray bursts ( grb ) are the brightest transient phenomena in the universe ( the isotropic equivalent of their total energy reaches @xmath5 erg ) . along with actual bursts , variable sources related to them
are also observed in different ranges : radio , infrared , optical , and x - ray .
they are less intensive but powerful enough , with luminosities reaching @xmath6 ergs@xmath7 .
a division of all these manifestations into two types of transients is generally accepted : emission synchronous with gamma - ray ( prompt ) , continuing at least until @xmath8 ( the time till which 90% of the gamma - ray burst energy is released ) , and the afterglows , usually starting after the burst turns off .
the most popular description of their population is a model which combines the inner and outer shock waves produced while the fragments of the relativistic outflow decelerate due to the collisions inside of it ( prompt ) and while it decelerates in the outer interstellar medium ( afterglow ) @xcite .
most of the studies are devoted to examining the properties of individual gamma - ray bursts and to attempts of verification of phenomenon models in general .
as time has shown , in this approach its self - consistent physical picture is far from being constructed , especially for the optical components of the bursts .
it is still unclear whether there occur reverse shock waves , what energy structure does a relativistic jet have , what is the density distribution in the ambient interstellar medium surrounding the burst , etc . in these circumstances ,
the search for statistical relationships between various parameters of bursts in their subsamples formed according to various criteria and with the minimal use of model representations remains a topical task . in its context
, a particularly important role has the analysis of the properties of the sample of gamma - ray bursts with measured redshifts , numbering about 250 objects @xcite .
this amount of data allows us to perform a statistical analysis of the parameters of gamma - ray bursts in the proper frame of reference , revealing empirical relations between intrinsic physical characteristics @xcite . at the same time ,
similar relations are also analyzed for the optical components of the bursts @xcite , the host galaxies @xcite , and the characteristics of the local interstellar medium in the grb progenitors @xcite .
the main problem of all such studies is to find an optimum ratio between the object selection criteria and the sample size : the latter decreases for a more rigorous selection , and subsequently the significance of established statistical dependences falls , while the role of selection effects increases .
we study the correlations between the parameters of grbs with measured redshifts and peaked optical light curves .
the latter condition isolates a specific physical condition within any given model : deceleration of either the ejecta colliding inside the jet in the case of prompt optical emissions or the ejected matter in the outer interstellar medium .
notice that it is not known in advance which kind of physical processes would determine the occurrence of this peak .
the division of objects into subsamples was carried out by comparing the time interval between the beginning of the optical brightness increase and the time of the trigger with the duration of the gamma - ray emission itself ( prompt optical emissions , afterglows , and afterglows accompanied by residual gamma - ray radiation ) , while the identification of the physical type of optical emissions ( prompt or afterglow ) stems from the statistical analysis of the entire subsample .
this way , some objects , initially classified as the prompt optical emissions were subsequently found to be close to the afterglows by their statistical properties . notice that the use of a sample of grbs with known redshifts for the statistical analysis allows us to pose the problem of the cosmological evolution of the properties of both the bursts themselves and the interstellar medium in the region of their localization .
section 2 of this paper gives the definitions and techniques of obtaining different characteristics of the objects , and describes the methods of statistical analysis of the initial data and the obtained results .
sections 3 , 4 , 5 discuss some of the most important correlations between the grb parameters we found : the relationship between the optical luminosity at the light curve peaks and the redshift , between the optical luminosity and the burst energy in the gamma - ray range , and between the luminosity and the time of the peak .
section 6 briefly summarizes the results obtained .
fifty - four gamma - ray bursts with known redshifts and peaked optical light curves were selected for the analysis .
the time dependences of the detected radiation flux both in the gamma - ray and optical ( in stellar magnitudes ) ranges served as the data sources for determining grb parameters . in the latter case , these are , as a rule , the light curves in the @xmath0 band , @xmath9 , sometimes with modified filters , @xmath10 , @xmath11 , and @xmath12 , in one case in the @xmath13 band ( grb050730 ) , and in two cases in the white light ( grb100906a and grb080810 ) . with the use of calibrations from @xcite , spectral flux densities in the respective @xmath14 bands in the observer s reference frame
were found .
the @xmath14 values were converted into the spectral flux densities @xmath15 at the effective @xmath0-band wavelength in the proper frame of reference .
if there were no data on the optical spectral index @xmath16 , @xmath17 , the mean value @xcite for grb optical components was used , and the following values were determined : the flux in the observer s frame of reference @xmath18 where @xmath19 and @xmath20 are the frequency boundaries of a photometric band in the observer s frame of reference ; the integral optical flux @xmath21 ( the result of numerical integration of @xmath22 over time ) ; the time of the optical peak relative to the trigger time of a gamma - ray burst @xmath23 ; the increase and decrease indices of the optical radiation intensity , @xmath24 , .
the parameters of the gamma - ray emission the peak @xmath25 and integral @xmath26 fluxes in the range of 15350 kev , the time of the release of 90% of the grb energy @xmath27 , and the spectral index of the energy distribution @xmath28were adopted from butler et al .
table 1 lists the data on the gamma - ray bursts and their parameters in the observer s frame of reference . taking into account the extinction in the galaxy @xcite , the brightness of the host galaxy ( if available ) , and the extinction in it , the following grb parameters were determined for the standard cosmological model with @xmath29 , @xmath30 , in the proper frame of reference : the maximal optical @xmath0 band luminosity @xmath31 where @xmath32 and @xmath33 are the frequency boundaries of the standard @xmath0 band , and @xmath34 is the photometric distance ; the total optical energy @xmath35 for the same range ( the time integral of @xmath36 ) ; the time - related parameters taking into account cosmological contraction .
the isotropic equivalent of the gamma - ray burst energy in the extended range of @xmath37 was adopted from @xcite , the peak luminosity @xmath38 was derived from @xmath25 as @xmath39 if the extinction @xmath40 in the host galaxy was not measured , then based on the data on @xmath41 for 76 bursts from @xcite , we determined the average of these values at a corresponding redshift and transformed them into @xmath40 using the model of extinction in the magellanic clouds @xcite .
the data on the parameters of the gamma - ray bursts in the proper frame of reference are presented in table 2 .
all the optical transients have been classified into one of four following types based on the ratio between the duration of gamma - ray emission itself and the lag of the beginning of the optical brightness increase relative to the time of the trigger .
( prompt ) ; thirteen optical transients are classified here whose brightness begins to increase before the attenuation of gamma - ray radiation .
this type includes four events , referred to as p ? , whose belonging to p is ambiguous ; they may belong to the a(u ) type ( see below ) .
@xmath42 a ( afterglow ) ; 34 transients whose brightness begins to increase after the gamma - ray activity is ended .
@xmath42 a(u ; seven transients whose brightness is increasing in the presence of underlying residual gamma - ray radiation after 90% of the total burst energy is released .
for the statistical analysis , seven samples were formed containing the following types of objects : \(1 ) p and p ?- type transients ; \(2 ) only afterglows a ; \(3 ) only the transients with underlying continuing gamma - ray activity a(u ) ; \(4 ) all 54 optical transients ; \(5 ) afterglows a and transients with underlying continuing gamma - ray activity a(u ) ;
\(6 ) only prompt optical emissions p ; \(7 ) the a - type transients along with a(u ) and p ? .
unweighted pearson correlation coefficients @xmath0 and their significance levels @xmath43 were determined between all possible pairs of grb parameters @xmath44 and @xmath45 from these samples by the formulas @xmath46 @xmath47 where @xmath48 and @xmath49 are the average values of @xmath44 and @xmath45 , @xmath50 is the euler gamma function , @xmath51 is the number of points @xmath52 .
correlations between the parameters with and the significance levels of were assumed to be significant , and the linear regression coefficients were then determined for these parameters ( table 3 ) .
among all the correlation relationships established in this study , the @xmath54 dependence ( fig . [ fig1:beskin_n ] ) , discovered for the first time , in our opinion presents the greatest interest .
in particular , its presence leads to the concept of probable cosmological evolution of the local interstellar medium in the gamma - ray burst birthplaces .
notice that this correlation is highly significant for all object types except for the p - type sample objects , which is corroborating evidence that transients belong to afterglows ( a and objects , whose characteristics virtually coincide in the discussed correlation ) . to be completely confident that the discovered correlation is real , the possible role of selection effects in its occurrence has to be analyzed .
notice that to determine the effect of selection on the parameters of the bursts in the samples of different types , it is first of all necessary to examine the whole set of brightness estimates for optical companions of the bursts at the time of their discovery . to this end , we built the dependences of the observed magnitudes on @xmath55 of optical companions at the times of their discovery and the light curve peak ( fig .
[ fig2:beskin_n ] ) . based on the data of figs . 1 and 2
, the following conclusions can be formulated .
given a relatively equal luminosity at different redshifts in the proper reference frames , the observed flux should drop quadratically with @xmath55 , and therewith at maximum redshifts reach the typical detection limit at around @xmath56 , estimated in our sample from the total light curves .
this effect would lead to a lack of low - luminosity objects in the proper frame of reference , whose observed brightness is fainter than @xmath57 , and as a consequence , to a luminosity increase with growing redshift .
nevertheless , no regular radiation flux decrease is observed either in the whole set of brightness estimates for optical transients at the moment of their detection or in the peak brightness . on the other hand , both at @xmath58 and
@xmath59 the brightness of objects varies within close limits from @xmath60@xmath61 to @xmath62@xmath63 , by far exceeding the detection limit of @xmath56 . in other words ,
the selection by brightness is absent in our data .
we compared the correlation coefficients and linear regression parameters for the group of transients from the a + a(u ) + p ?
sample , localized in different redshift ranges ( table 4 ) .
it is easy to see that even if we exclude the transients with @xmath66 and @xmath67 , the luminosity increase with increasing redshift remains detectable and significant . clearly , if we narrow the redshift range , the correlation coefficients decrease , but nevertheless the linear regression parameters for different groups of transients remain the same within their errors ; in particular , the exponents are close to 45
. .characteristics of the redshift dependence of the a+a(u)+p ?
sample peak optical luminosity in different @xmath65 ranges : the number of transients in the sample @xmath68 , the pearson correlation coefficient @xmath0 , the linear regression parameters @xmath69 [ cols="^,^,^,^,^,^",options="header " , ] | we present the statistical analysis of the properties of gamma - ray bursts with measured host galaxy redshifts and peaked optical light curves in proper frames of reference .
the optical transients are classified by comparing the time lag of the optical peak relative to the grb trigger with the duration of the gamma - ray emission itself .
the results of the correlation analysis of all possible pairs of energy , spectral , and temporal characteristics of both gamma - ray and optical emissions are given .
we specify the pairs of the parameters with correlation coefficients greater than 50% at significance levels better than 1% .
the following empirical relations , obtained for the first time , are specifically discussed : a correlation between the peak optical afterglow @xmath0 band luminosity and redshift @xmath1 and a correlation between the peak luminosity of the prompt optical emissions and the time of the peak @xmath2 .
we also analyze the similarity of the relationships between the peak optical luminosity and the isotropic equivalent of the total energy of gamma - ray bursts for afterglows ( @xmath3 ) and for prompt optical emissions ( @xmath4 ) . |
sequence comparison is an important step in many basic tasks in bioinformatics , from phylogenies reconstruction to genomes assembly .
it is often realized by sequence alignment techniques , which are computationally expensive , requiring quadratic time in the length of the sequences .
this has led to increased research into _ alignment - free _ techniques .
hence standard notions for sequence comparison are gradually being complemented and in some cases replaced by alternative ones @xcite .
one such notion is based on comparing the words that are absent in each sequence @xcite .
a word is an _ absent word _ ( or a forbidden word ) of some sequence if it does not occur in the sequence .
absent words represent a type of _ negative information _ : information about what does not occur in the sequence . given a sequence of length @xmath1 , the number of absent words of length at most @xmath1 is exponential in @xmath1 .
however , the number of certain classes of absent words is only linear in @xmath1 .
this is the case for _ minimal absent words _ , that is , absent words in the sequence whose all proper factors occur in the sequence @xcite .
an upper bound on the number of minimal absent words is known to be @xmath2 @xcite , where @xmath3 is the size of the alphabet @xmath4 .
hence it may be possible to compare sequences in time proportional to their lengths , for a fixed - sized alphabet , instead of proportional to the product of their lengths . in what follows ,
we consider sequences on a _ fixed - sized alphabet _ since the most commonly studied alphabet is @xmath5 .
an @xmath6-time and @xmath6-space algorithm for computing all minimal absent words on a fixed - sized alphabet based on the construction of suffix automata was presented in @xcite .
the computation of minimal absent words based on the construction of suffix arrays was considered in @xcite ; although this algorithm has a linear - time performance in practice , the worst - case time complexity is @xmath7 .
new @xmath6-time and @xmath6-space suffix - array - based algorithms were presented in @xcite to bridge this unpleasant gap .
an implementation of the algorithm presented in @xcite is currently , and to the best of our knowledge , the fastest available for the computation of minimal absent words .
a more space - efficient solution to compute all minimal absent words in time @xmath6 was also presented in @xcite . in this article , we consider the problem of comparing two sequences @xmath8 and @xmath9 of respective lengths @xmath10 and @xmath1 , using their sets of minimal absent words . in @xcite ,
chairungsee and crochemore introduced a measure of similarity between two sequences based on the notion of minimal absent words .
they made use of a length - weighted index to provide a measure of similarity between two sequences , using sample sets of their minimal absent words , by considering the length of each member in the symmetric difference of these sample sets .
this measure can be trivially computed in time and space @xmath11 provided that these sample sets contain minimal absent words of some bounded length @xmath12 . for unbounded length ,
the same measure can be trivially computed in time @xmath13 : for a given sequence , the cumulative length of all its minimal absent words can grow _ quadratically _ with respect to the length of the sequence . the same problem can be considered for two _ circular _ sequences .
the measure of similarity of chairungsee and crochemore can be used in this setting provided that one extends the definition of minimal absent words to circular sequences . in section [ sec : circ_seq_comp ] , we give a definition of minimal absent words for a circular sequence from the formal language theory point of view .
we believe that this definition may also be of interest from the point of view of symbolic dynamics , which is the original context in which minimal absent words have been introduced @xcite . * our contribution . * here we make the following threefold contribution : a ) : : we present an @xmath11-time and @xmath11-space algorithm to compute the similarity measure introduced by chairungsee and crochemore by considering _ all _ minimal absent words of two sequences @xmath8 and @xmath9 of lengths @xmath10 and @xmath1 , respectively ; thereby showing that it is indeed possible to compare two sequences in time proportional to their lengths ( section [ sec : seq_comp ] ) .
b ) : : we show how this algorithm can be applied to compute this similarity measure for two circular sequences @xmath8 and @xmath9 of lengths @xmath10 and @xmath1 , respectively , in the same time and space complexity as a result of the extension of the definition of minimal absent words to circular sequences ( section [ sec : circ_seq_comp ] ) .
c ) : : we provide an open - source code implementation of our algorithms and investigate potential applications of our theoretical findings ( section [ sec : imp_app ] ) .
we begin with basic definitions and notation .
let @xmath14y[1]\dd y[n-1]$ ] be a _ word _ of _ length _
@xmath15 over a finite ordered _ alphabet _ @xmath4 of size @xmath16 . for two positions @xmath17 and @xmath18 on @xmath9 , we denote by @xmath19=y[i]\dd y[j]$ ] the _ factor _ ( sometimes called _ substring _ ) of @xmath9 that starts at position @xmath17 and ends at position @xmath18 ( it is empty if @xmath20 ) , and by @xmath21 the _ empty word _ , word of length 0 .
we recall that a prefix of @xmath9 is a factor that starts at position 0 ( @xmath22 $ ] ) and a suffix is a factor that ends at position @xmath23 ( @xmath24 $ ] ) , and that a factor of @xmath9 is a _ proper _ factor if it is not @xmath9 itself .
the set of all the factors of the word @xmath9 is denoted by @xmath25 .
let @xmath8 be a word of length @xmath26 .
we say that there exists an _ occurrence _ of @xmath8 in @xmath9 , or , more simply , that @xmath8 _ occurs in _ @xmath9 , when @xmath8 is a factor of @xmath9 .
every occurrence of @xmath8 can be characterised by a starting position in @xmath9 .
thus we say that @xmath8 occurs at the _ starting position _
@xmath17 in @xmath9 when @xmath27 $ ] .
opposingly , we say that the word @xmath8 is an _ absent word _ of @xmath9 if it does not occur in @xmath9 .
the absent word @xmath8 of @xmath9 is _ minimal _ if and only if all its proper factors occur in @xmath9 .
the set of all minimal absent words for a word @xmath9 is denoted by @xmath28 .
for example , if @xmath29 , then @xmath30 . in general ,
if we suppose that all the letters of the alphabet appear in @xmath9 of length @xmath1 , the length of a minimal absent word of @xmath9 lies between @xmath31 and @xmath32 .
it is equal to @xmath32 if and only if @xmath9 is the catenation of @xmath1 copies of the same letter .
so , if @xmath9 contains occurrences of at least two different letters , the length of a minimal absent word for @xmath9 is bounded from above by @xmath1 .
a _ language _ over the alphabet @xmath4 is a set of finite words over @xmath4 .
a language is _ regular _ if it is recognized by a finite state automaton .
a language is _ factorial _ if it contains all the factors of its words .
a language is _ antifactorial _ if no word in the language is a proper factor of another word in the language . given a word @xmath8 , the language _ generated _ by @xmath8 is the language @xmath33 .
the _ factorial closure _ of a language @xmath34 is the language @xmath35 .
given a factorial language @xmath34 , one can define the ( antifactorial ) language of minimal absent words for @xmath34 as @xmath36 .
notice that @xmath37 is not the same language as the union of @xmath38 for @xmath39 .
we denote by the _ suffix array _ of @xmath9 of length @xmath1 , that is , an integer array of size @xmath1 storing the starting positions of all ( lexicographically ) sorted suffixes of @xmath9 , i.e. for all @xmath40 we have @xmath41 \dd n-1 ] < y[{\textsf{sa}}{}[r ] \dd n - 1]$ ] @xcite .
let @xmath42 denote the length of the longest common prefix between @xmath43 \dd n - 1]$ ] and @xmath44 \dd n - 1]$ ] for all positions @xmath45 , @xmath46 on @xmath9 , and @xmath47 otherwise .
we denote by the _ longest common prefix _ array of @xmath9 defined by @xmath48={\textsf{lcp}}{}(r-1 , r)$ ] for all @xmath49 , and @xmath50 = 0 $ ] .
the inverse of the array is defined by @xmath51 = r$ ] , for all @xmath52 .
it is known that @xcite , , and @xcite of a word of length @xmath1 can be computed in time and space @xmath6 . in what follows , as already proposed in @xcite , for every word
@xmath9 , the set of minimal words associated with @xmath9 , denoted by @xmath28 , is represented as a set of tuples @xmath53 , where the corresponding minimal absent word @xmath8 of @xmath9 is defined by @xmath54=a$ ] , @xmath55 , and @xmath56 = y[i \dd j]$ ] , where @xmath57 .
it is known that if @xmath58 and @xmath59 , then @xmath60 @xcite . in @xcite , chairungsee and crochemore
introduced a measure of similarity between two words @xmath8 and @xmath9 based on the notion of minimal absent words .
let @xmath61 ( resp .
@xmath62 ) denote the set of minimal absent words of length at most @xmath12 of @xmath8 ( resp .
@xmath9 ) .
the authors made use of a length - weighted index to provide a measure of the similarity between @xmath8 and @xmath9 , using their sample sets @xmath61 and @xmath62 , by considering the length of each member in the symmetric difference @xmath63 of the sample sets . for sample sets
@xmath61 and @xmath62 , they defined this index to be @xmath64 this work considers the following generalized version of the same problem .
we also consider the aforementioned problem for two circular words .
a circular word of length @xmath10 can be viewed as a traditional linear word which has the left- and right - most letters wrapped around and stuck together in some way . under this notion ,
the same circular word can be seen as @xmath10 different linear words , which would all be considered equivalent .
more formally , given a word @xmath8 of length @xmath10 , we denote by @xmath65x[0 \dd i-1]$ ] , @xmath66 , the @xmath17-th _ rotation _ of @xmath8 , where @xmath67 . given two words @xmath8 and
@xmath9 , we define @xmath68 if and only if there exist @xmath17 , @xmath69 , such that @xmath70 .
a _ circular word _
@xmath71 is a conjugacy class of the equivalence relation @xmath72 .
given a circular word @xmath71 , any ( linear ) word @xmath8 in the equivalence class @xmath71 is called a _
linearization _ of the circular word @xmath71 .
conversely , given a linear word @xmath8 , we say that @xmath71 is a _ circularization _ of @xmath8 if and only if @xmath8 is a linearization of @xmath71 .
the set @xmath73 of factors of the circular word @xmath71 is equal to the set @xmath74 of factors of @xmath75 whose length is at most @xmath76 , where @xmath8 is any linearization of @xmath71 .
note that if @xmath77 and @xmath78 are two rotations of the same word , then the factorial languages @xmath79 and @xmath80 coincide , so one can unambiguously define the ( infinite ) language @xmath81 as the language @xmath82 , where @xmath8 is any linearization of @xmath71 . in section [ sec : circ_seq_comp ] , we give the definition of the set @xmath83 of minimal absent words for a circular word @xmath71 .
we will prove that the following problem can be solved with the same time and space complexity as its counterpart in the linear case .
the goal of this section is to provide the first linear - time and linear - space algorithm for computing the similarity measure ( see section [ sec : prem ] ) between two words defined over a fixed - sized alphabet . to this end , we consider two words @xmath8 and @xmath9 of lengths @xmath10 and @xmath1 , respectively , and their associated sets of minimal absent words , @xmath38 and @xmath28 , respectively .
next , we give a linear - time and linear - space solution for the maw - sequencecomparison problem .
it is known from @xcite and @xcite that we can compute the sets @xmath38 and @xmath28 in linear time and space with respect to the two lengths @xmath10 and @xmath1 , respectively .
the idea of our strategy consists of a merge sort on the sets @xmath38 and @xmath28 , after they have been ordered with the help of suffix arrays . to this end
, we construct the suffix array associated to the word @xmath84 , together with the implicit @xmath85 array corresponding to it .
all of these structures can be constructed in time and space @xmath86 , as mentioned earlier .
furthermore , we can preprocess the array for range minimum queries , which we denote by @xmath87 @xcite . with the preprocessing complete ,
the longest common prefix @xmath88 of two suffixes of @xmath89 starting at positions @xmath90 and @xmath0 can be computed in constant time @xcite , using the formula @xmath91
+ 1,\textsf{isa}[q])].$ ] using these data structures , it is straightforward to sort the tuples in the sets @xmath38 and @xmath28 lexicographically .
that is , two tuples @xmath92 , are ordered according to the letter following their longest common prefix , or when it is not the case , with the one being the prefix , coming first . to do this
, we simply go once through the suffix array associated to @xmath89 and assign to each tuple in @xmath38 , respectively @xmath28 , the rank of the suffix starting at the position indicated by its second component , in the suffix array . since sorting an array of @xmath1 distinct integers , such that each is in @xmath93 $ ] , can be done in time @xmath6 ( using bucket sort , for example ) , we can sort now each of the sets of minimal absent words , taking into consideration the letter on the first position and these ranks .
thus , from now on , we assume that @xmath94 where @xmath95 is lexicographically smaller than @xmath96 , for @xmath97 , and @xmath98 , where @xmath99 is lexicographically smaller than @xmath100 , for @xmath101 .
provided these tools , we now proceed to do the merge .
thus , considering that we are analysing the @xmath102th tuple in @xmath38 and the @xmath103th tuple in @xmath28 , we note that the two are equal if and only if @xmath104=y_j[0]$ ] and @xmath105 , |x|+y_j[1])\geq \ell , \mbox { where } \ell = x_i[2]-x_i[1]=y_j[2]-y_j[1].\ ] ] in other words , the two minimal absent words are equal if and only if their first letters coincide , they have equal length @xmath106 , and the longest common prefix of the suffixes of @xmath89 starting at the positions indicated by the second components of the tuples has length at least @xmath12
. such a strategy will empower us with the means for constructing a new set @xmath107 . at each step , when analysing tuples @xmath95 and @xmath99 we proceed as following : @xmath108 observe that the last condition is saying that basically each common tuple is added only once to their union .
furthermore , simultaneously with this construction we can also calculate the similarity between the words , given by @xmath109 , which is initially set to @xmath47 .
thus , at each step , when comparing the tuples @xmath95 and @xmath99 , we update @xmath110 we impose the increment of both @xmath17 and @xmath18 in the case of equality as in this case we only look at the symmetric difference between the sets of minimal absent words .
as all these operations take constant time , once per each tuple in @xmath38 and @xmath28 , it is easily concluded that the whole operation takes in the case of a fixed - sized alphabet time and space @xmath86 .
thus , we can compute the symmetric difference between the _ complete _ sets of minimal absent words , as opposed to @xcite , of two words defined over a fixed - sized alphabet , in linear time and space with respect to the lengths of the two words .
we obtain the following result .
[ the : maw ] problem maw - sequencecomparison can be solved in time and space @xmath86 .
next , we discuss two possible definitions for the minimal absent words of a circular word , and highlight the differences between them . we start by recalling some basic facts about minimal absent words . for further details and references the reader
is recommended @xcite .
every factorial language @xmath34 is uniquely determined by its ( antifactorial ) language of minimal absent words @xmath37 , through the equation @xmath111 . the converse is also true , since by the definition of a minimal absent word we have @xmath112 .
the previous equations define a bijection between factorial and antifactorial languages .
moreover , this bijection preserves regularity . in the case of a single ( linear ) word @xmath8 ,
the set of minimal absent words for @xmath8 is indeed the antifactorial language @xmath113 .
furthermore , we can retrieve @xmath8 from its set of minimal absent words in linear time and space @xcite .
recall that given a circular word @xmath71 , the set @xmath73 of factors of @xmath71 is equal to the set @xmath74 of factors of @xmath75 whose lengths are at most @xmath76 , where @xmath8 is any linearization of @xmath71 . since a circular word @xmath71 is a conjugacy class containing all the rotations of a linear word @xmath8 , the language @xmath73 can be seen as the factorial closure of the set @xmath114 .
this leads to the first definition of the set of minimal absent words for @xmath71 , that is the set @xmath115 .
for instance , if @xmath116 , we have @xmath117 the advantage of this definition is that we can retrieve uniquely @xmath71 from @xmath118 .
however , the total size of @xmath118 ( that is , the sum of the lengths of its elements ) can be very large , as the following lemma suggests .
let @xmath71 be a circular word of length @xmath119 .
the set @xmath118 contains precisely @xmath12 words of maximal length @xmath120 , where @xmath12 is the number of distinct rotations of any linearization @xmath8 of @xmath71 , that is , the cardinality of @xmath114 .
let @xmath121x[1 ] \dd x[m-1]$ ] be a linearization of @xmath71 .
the word obtained by appending to @xmath8 its first letter , @xmath54x[1 ] \dd x[m-1]x[0]$ ] , belongs to @xmath118 , since it has length @xmath120 , hence it can not belong to @xmath73 , but its maximal proper prefix @xmath122 and its maximal proper suffix @xmath123 \dd x[m-1]x[0]$ ] belong to @xmath73 .
the same argument shows that for any rotation @xmath124x[i+1 ] \dd x[m-1]x[0]\dd x[i-1]$ ] of @xmath8 , the word @xmath125x[i+1 ] \dd x[m-1]x[0]\dd x[i-1]x[i]$ ] , obtained by appending to @xmath126 its first letter , belongs to @xmath118 .
conversely , if a word of maximal length @xmath120 is in @xmath118 , then its maximal proper prefix and its maximal proper suffix are words of length @xmath10 in @xmath73 , so they must be consecutive rotations of @xmath8 . therefore , the number of words of maximal length @xmath120 in @xmath118 equals the number of distinct rotations of @xmath8 , hence the statement follows .
this is in sharp contrast with the situation for linear words , where the set of minimal absent words can be represented on a trie having size linear in the length of the word .
indeed , the algorithm mf - trie , introduced in @xcite , builds the tree - like deterministic automaton accepting the set of minimal absent words for a word @xmath8 taking as input the factor automaton of @xmath8 , that is the minimal deterministic automaton recognizing the set of factors of @xmath8 .
the leaves of the trie correspond to the minimal absent words for @xmath8 , while the internal states are those of the factor automaton .
since the factor automaton of a word @xmath8 has less than @xmath127 states ( for details , see @xcite ) , this provides a representation of the minimal absent words of a word of length @xmath1 in space @xmath128 .
this algorithmic drawback leads us to the second definition .
this second definition of minimal absent words for circular strings has been already introduced in @xcite .
first , we give a combinatorial result which shows that when considering circular words it does not make sense to look at absent words obtained from more than two rotations . [ lem : general ] for any positive integer @xmath129 and any word @xmath130 , the set @xmath131 is empty .
this obviously holds for all words @xmath130 of length 1 .
assume towards a contradiction that this is not the case in general .
hence , there must exist a word @xmath132 of length @xmath10 that fulfills the conditions in the lemma , thus @xmath133 and @xmath134 .
furthermore , since the length @xmath135 prefix and the length @xmath135 suffix of every minimal absent word occur in the main word at non - consecutive positions , there must exist positions @xmath136 such that @xmath137=u^{k+1}[i+1\dd i+m-2]=u^{k+1}[j+1\dd j+m-2].\end{aligned}\ ] ] obviously , following equation ( [ eq:1 ] ) , since @xmath138 , we have that @xmath139 $ ] is @xmath140-periodic .
but , we know that @xmath141 $ ] is also @xmath1-periodic . thus , following a direct application of the periodicity lemma we have that @xmath141 $ ] is @xmath142-periodic .
but , in this case we have that @xmath130 is @xmath90-periodic , and , therefore , @xmath143=u[j]$ ] , which leads to a contradiction with the fact that @xmath132 is a minimal absent word , whenever @xmath17 is defined .
thus , it must be the case that @xmath144 . using the same strategy and looking at positions @xmath145 $ ] and
@xmath146 $ ] , we conclude that @xmath147 . therefore , in this case , we have that @xmath148 , which is a contradiction with the fact that the word @xmath132 fulfills the conditions of the lemma .
this concludes the proof .
observe now that the set @xmath149 consists in fact of all extra minimal absent words generated whenever we look at more than one rotation , that do not include the length arguments .
that is , @xmath149 does not include the words bounding the maximum length that a word is allowed , nor the words created , or lost , during a further concatenation of an image of @xmath130 .
however , when considering an iterative concatenation of the word , these extra elements determined by the length constrain cancel each other . as observed in section
[ sec : prem ] , two rotations of the same word @xmath8 generate two languages that have the same set of factors .
so , we can unambiguously associate to a circular word @xmath71 the ( infinite ) factorial language @xmath81 .
it is therefore natural to define the set of minimal absent words for the circular word @xmath71 as the set @xmath150 .
for instance , if @xmath151 , then we have @xmath152 this second definition is much more efficient in terms of space , as we show below . in particular
, the length of the words in @xmath150 is bounded from above by @xmath76 , hence @xmath150 is a finite set .
recall that a word @xmath8 is _ a power _ of a word @xmath9 if there exists a positive integer @xmath153 such that @xmath8 is expressed as @xmath129 consecutive concatenations of @xmath9 , denoted by @xmath154 .
conversely , a word @xmath8 is _ primitive _ if @xmath154 implies @xmath155 .
notice that a word is primitive if and only if any of its rotation is .
we can therefore extend the definition of primitivity to circular words .
the definition of @xmath150 does not allow one to uniquely reconstruct @xmath71 from @xmath150 , unless @xmath71 is known to be primitive , since it is readily verified that @xmath156 and therefore also the minimal absent words of these two languages coincide .
however , from the algorithmic point of view , this issue can be easily managed by storing the length @xmath76 of a linearization @xmath8 of @xmath71 together with the set @xmath150 . moreover , in most practical cases , for example when dealing with biological sequences , it is highly unlikely that the circular word considered is not primitive .
the difference between the two definitions above is presented in the next lemma .
[ lem : twodef ] @xmath157 clearly , @xmath158 .
the statement then follows from the definition of minimal absent words .
based on the previous discussion , we set @xmath159 , while the following corollary comes straightforwardly as a consequence of lemma [ lem : general ] .
[ lem : circ ] let @xmath71 be a circular word
. then @xmath160 .
corollary [ lem : circ ] was first introduced as a definition for the set of minimal absent words of a circular word in @xcite . using the result of corollary [ lem : circ ]
, we can easily extend the algorithm described in the previous section to the case of circular words .
that is , given two circular words @xmath71 of length @xmath10 and @xmath161 of length @xmath1 , we can compute in time and space @xmath86 the quantity @xmath162 .
we obtain the following result .
[ the : cmaw ] problem maw - circularsequencecomparison can be solved in time and space @xmath86 .
we implemented the presented algorithms as programme to perform pairwise sequence comparison for a set of sequences using minimal absent words .
uses programme @xcite for linear - time and linear - space computation of minimal absent words using suffix array .
was implemented in the @xmath163 programming language and developed under gnu / linux operating system .
it takes , as input argument , a file in multifasta format with the input sequences , and then any of the two methods , for _ linear _ or _ circular _ sequence comparison , can be applied .
it then produces a file in phylip format with the distance matrix as output .
cell @xmath164 $ ] of the matrix stores @xmath165 ( or @xmath162 for the circular case ) .
the implementation is distributed under the gnu general public license ( gpl ) , and it is available at http://github.com/solonas13/maw , which is set up for maintaining the source code and the man - page documentation . notice that _ all _ input datasets and the produced outputs referred to in this section are publicly maintained at the same web - site .
an important feature of the proposed algorithms is that they require space linear in the length of the sequences ( see theorem [ the : maw ] and theorem [ the : cmaw ] ) .
hence , we were also able to implement using the open multi - processing ( openmp ) pi for shared memory multiprocessing programming to distribute the workload across the available processing threads without a large memory footprint . * application . *
recently , there has been a number of studies on the biological significance of absent words in various species @xcite . in @xcite ,
the authors presented dendrograms from dinucleotide relative abundances in sets of minimal absent words for prokaryotes and eukaryotic genomes .
the analyses support the hypothesis that minimal absent words are inherited through a common ancestor , in addition to lineage - specific inheritance , only in vertebrates . very recently , in @xcite , it was shown that there exist three minimal words in the ebola virus genomes which are absent from human genome .
the authors suggest that the identification of such species - specific sequences may prove to be useful for the development of both diagnosis and therapeutics . in this section ,
we show a potential application of our results for the construction of dendrograms for dna sequences with circular structure .
circular dna sequences can be found in viruses , as plasmids in archaea and bacteria , and in the mitochondria and plastids of eukaryotic cells .
circular sequence comparison thus finds applications in several contexts such as reconstructing phylogenies using viroids rna @xcite or mitochondrial dna ( mtdna ) @xcite .
conventional tools to align circular sequences could yield an incorrectly high genetic distance between closely - related species . indeed ,
when sequencing molecules , the position where a circular sequence starts can be totally arbitrary . due to this _ arbitrariness _ , a suitable rotation of one sequence would give much better results for a pairwise alignment @xcite .
in what follows , we demonstrate the power of minimal absent words to pave a path to resolve this issue by applying corollary [ lem : circ ] and theorem [ the : cmaw ] .
next we do not claim that a solid phylogenetic analysis is presented but rather an investigation for potential applications of our theoretical findings .
we performed the following experiment with synthetic data .
first , we simulated a basic dataset of dna sequences using indelible @xcite .
the number of taxa , denoted by @xmath166 , was set to @xmath167 ; the length of the sequence generated at the root of the tree , denoted by @xmath168 , was set to 2500bp ; and the substitution rate , denoted by @xmath169 , was set to @xmath170 .
we also used the following parameters : a deletion rate , denoted by @xmath171 , of @xmath172 _ relative _ to substitution rate of @xmath173 ; and an insertion rate , denoted by @xmath174 , of @xmath175 _ relative _ to substitution rate of @xmath173 .
the parameters were chosen based on the genetic diversity standard measures observed for sets of mtdna sequences from primates and mammals @xcite .
we generated another instance of the basic dataset , containing one _ arbitrary _ rotation of each of the @xmath166 sequences from the basic dataset .
we then used this randomized dataset as input to by considering @xmath162 as the distance metric .
the output of was passed as input to @xcite , an efficient implementation of neighbor - joining @xcite , a well - established hierarchical clustering algorithm for inferring dendrograms ( trees ) .
we thus used to infer the respective tree @xmath176 under the neighbor - joining criterion .
we also inferred the tree @xmath177 by following the same pipeline , but by considering @xmath165 as distance metric , as well as the tree @xmath178 by using the _ basic _ dataset as input of this pipeline and @xmath162 as distance metric . hence , notice that @xmath178 represents the original tree .
finally , we computed the pairwise robinson - foulds ( rf ) distance @xcite between : @xmath176 and @xmath178 ; and @xmath177 and @xmath178 .
let us define _ accuracy _ as the difference between 1 and the relative pairwise rf distance .
we repeated this experiment by simulating different datasets @xmath179 and measured the corresponding accuracy .
the results in table [ tab : accuracy ] ( see @xmath176 vs. @xmath178 ) suggest that by considering @xmath162 we can always re - construct the original tree even if the sequences have first been arbitrarily rotated ( corollary [ lem : circ ] ) .
this is not the case ( see @xmath177 vs. @xmath178 ) if we consider @xmath165 .
notice that @xmath180 accuracy denotes a ( relative ) pairwise rf distance of 0 .
in this article , complementary to measures that refer to the composition of sequences in terms of their constituent patterns , we considered sequence comparison using minimal absent words , information about what does not occur in the sequences .
we presented the first linear - time and linear - space algorithm to compare two sequences by considering _
all _ their minimal absent words ( theorem [ the : maw ] ) . in the process
, we presented some results of combinatorial interest , and also extended the proposed techniques to circular sequences .
the power of minimal absent words is highlighted by the fact that they provide a tool for sequence comparison that is as efficient for circular as it is for linear sequences ( corollary [ lem : circ ] and theorem [ the : cmaw ] ) ; whereas , this is not the case , for instance , using the general edit distance model @xcite .
finally , a preliminary experimental study shows the potential of our theoretical findings . our immediate target is to consider the following _ incremental _ version of the same problem : given an appropriate encoding of a comparison between sequences @xmath8 and @xmath9 ,
can one incrementally compute the answer for @xmath8 and @xmath181 , and the answer for @xmath8 and @xmath182 , efficiently , where @xmath183 is an additional letter ?
incremental sequence comparison , under the edit distance model , has already been considered in @xcite .
in @xcite , the authors considered a more powerful generalization of the @xmath0-gram distance ( see @xcite for definition ) to compare @xmath8 and @xmath9 .
this generalization comprises partitioning @xmath8 and @xmath9 in @xmath168 blocks each , as evenly as possible , computing the @xmath0-gram distance between the corresponding block pairs , and then summing up the distances computed blockwise to obtain the new measure .
we are also planning to apply this generalization to the similarity measure studied here and evaluate it using real and synthetic data .
we warmly thank alice heliou for her inestimable code contribution and antonio restivo for useful discussions .
gabriele fici s work was supported by the prin 2010/2011 project `` automi e linguaggi formali : aspetti matematici e applicativi '' of the italian ministry of education ( miur ) and by the `` national group for algebraic and geometric structures , and their applications '' ( gnsaga indam ) .
robert merca s work was supported by the p.r.i.m.e .
programme of daad co - funded by bmbf and eu s 7th framework programme ( grant 605728 ) .
solon p. pissis s work was supported by a research grant ( # rg130720 ) awarded by the royal society .
barton , c. , iliopoulos , c.s .
, kundu , r. , pissis , s.p . ,
retha , a. , vayani , f. : accurate and efficient methods to improve multiple circular sequence alignment . in : sea , lncs ,
vol . 9125 , pp .
247258 ( 2015 ) | sequence comparison is a prerequisite to virtually all comparative genomic analyses .
it is often realized by sequence alignment techniques , which are computationally expensive .
this has led to increased research into alignment - free techniques , which are based on measures referring to the composition of sequences in terms of their constituent patterns .
these measures , such as @xmath0-gram distance , are usually computed in time linear with respect to the length of the sequences . in this article , we focus on the complementary idea : how two sequences can be efficiently compared based on information that does not occur in the sequences .
a word is an _ absent word _ of some sequence if it does not occur in the sequence .
an absent word is _ minimal _ if all its proper factors occur in the sequence . here
we present the first linear - time and linear - space algorithm to compare two sequences by considering _
all _ their minimal absent words . in the process
, we present results of combinatorial interest , and also extend the proposed techniques to compare circular sequences . |
two years before irving langmuir coined the term `` plasma '' as in plasma physics @xcite , he and mott - smith laid down a foundational theory on the charging of a spherical and cylindrical probe in a laboratory plasma @xcite , which was necessary for interpreting the measurement of what became later known as langmuir probes @xcite . in the ensuing decades , the charging and dynamics of solid particulates immersed in plasmas built the foundation for a new discipline in plasma physics
dusty plasmas @xcite , in which the collective behavior of a group of dust particles in a plasma environment is studied .
the physics of both dust in a plasma and dusty plasmas finds applications in space @xcite , astrophysics @xcite , and laboratory @xcite .
the fundamental dust - plasma interaction , which is also essential to understand the collective behavior in a dusty plasma , includes ( 1 ) charging of the dust by absorption of plasma particles so it becomes subject to the electric field ( @xmath0 ) via the electrical force @xmath1 ( @xmath2 is the dust charge ) ; ( 2 ) heating of the dust via the collection of plasma electron and ion energy fluxes so the dust particulate can melt , evaporate , or simply sublimate ; ( 3 ) dragging of the dust via a frictional force enhanced by the coulomb interaction with the flowing background plasma . to understand the dust dynamics and its change of state
, one must resort to a dust charging theory .
the most - widely used of such is the orbital - motion - limited ( oml ) theory , which dates back to the work of mott - smith and langmuir @xcite . in the 1960s ,
alpert et al @xcite and laframboise @xcite completed the current formulation .
the oml theory is known to predict accurately the dust potential for a small @xcite and not so small dust grain @xcite , despite its simplifying assumption on collisionless ion orbit that misses the absorption radius effect away from the dust surface @xcite .
surprisingly , the existing formulation can not be used for predicting the plasma response , namely solving the plasma potential @xmath3 , as shown in allen et al @xcite . as the result , oml theory to date can not predict the fundamental quantity of dust charge . in practice ,
one has been using an idealized dust charge - potential relationship due to whipple @xcite , who computed the dust capacitance using the conventional debye shielding potential that does not take into account the constraint of angular momentum conservation in setting the plasma density near the dust surface .
the purpose of this paper is to present a revised oml formulation that , for the first time , is able to predict the plasma response and hence the dust charge .
the oml predictions will then be contrasted with the whipple approximation to elucidate the missing physics in dust charging for a spherical dust of comparable size with the debye length , which is of special importance to dust transport / survivability in tokamaks @xcite .
this new oml formulation is also important for electron - emitting dust particulates as it provides the basis for extending the oml theory to positively charged dust @xcite . as an approximation of the orbital motion ( om ) theory which follows the collisionless particle orbit via conservation of energy and angular momentum ,
the oml theory will inevitably introduce discrepancies in the evaluation of dust current / heat collection , and in the plasma ion density evaluation .
the former will give rise to a discrepancy in dust potential , while the latter in plasma potential and hence dust charge , in comparison with those predicted by the om theory .
the usefulness of the oml formulation is due to its simplicity , and the relatively high accuracy for a variety of applications where the dust size is not large compared with the plasma debye length .
the rest of the paper is organized as follows . in section [ sec :
oml ] , we briefly recall the key components of the oml theory and its inability to predict dust charge in its current formulation . a corrected formulation on the oml ion density is shown in section [ sec : corrected - oml ] , which completes the oml formulation for calculating both dust potential and dust change . in section
[ sec : oml - result ] , we apply the new oml model to evaluate the plasma response and the dust charge , and contrast the results with the widely used whipple approximation . the findings are summarized in section [ sec : summary ] .
there are three essential ideas underlying oml theory .
first , the collection of plasma electrons and ions by the dust is governed by their collisionless orbit .
these are subject to two conservation laws : energy [ @xmath4 conservation and angular momentum ( @xmath5 ) conservation . here
@xmath6 ( @xmath7 ) and @xmath3 are the mass ( charge ) of the particles of species @xmath8 ( @xmath9 labels electrons and ions , respectively ) and the plasma potential , while @xmath10 , @xmath11 and @xmath12 are the radial distance , radial velocity and tangential velocity in a spherical reference frame centered on the dust grain .
the radial motion of the plasma particles is governed by a one degree - of - freedom hamiltonian with effective potential @xmath13 @xmath14 second , whether the plasma particle reaches the dust is governed by @xmath15 for a typical negatively charged dust , @xmath3 is negative and monotonically increasing with @xmath16 so that @xmath17 is positive and monotonically decreasing with @xmath10 for electrons . as the result
, only electrons with @xmath18 from far away can reach the dust of radius @xmath19 the final and third idea , which is a simplifying approximation , is what delineates oml from the more complete but rarely used orbital motion ( om ) theory @xcite .
the oml theory approximates the @xmath20 for ions also as a monotonic function of @xmath16 which results in the simplification that ions with @xmath21 from far away can reach and charge the dust .
this is a remarkable simplification since one does not need @xmath22 to evaluate the ion and the electron current collected by the dust , which are @xmath23 here @xmath24 is the elementary charge , @xmath25 is the boltzmann constant , @xmath26 ( @xmath27 ) is the electron ( ion ) density away from the dust , @xmath28 , @xmath29 is the electron ( ion ) temperature , @xmath30 , @xmath31 , and @xmath32 setting @xmath33 , one can solve for the dust potential as a function of @xmath34 ion charge state @xmath35 and @xmath36 for an electron - proton plasma with @xmath37 oml predicts @xmath38 which is in remarkable agreement with particle - in - cell simulations @xcite that do not make the oml assumption of ion @xmath20 being monotonic . the oml approximation of a monotonic ion @xmath20 is known to be violated for ions with a certain range of @xmath39 this comes about @xcite because the centrifugal potential energy @xmath40 decreases with @xmath10 at precisely @xmath41 but the electrical potential energy @xmath42 increases at a faster rate ( exponential in @xmath10 ) in the debye shielding region ,
before it eventually asymptotes to @xmath43 for large @xmath44 this can produce one or multiple extrema in @xmath20 away from the dust surface , fig .
[ fig : om - potential ] , which can turn back some ions in a range of @xmath45 at @xmath46 in the literature this is known as the absorption radius ( at @xmath47 ) effect for certain ions . by neglecting this subtlety , oml approximation would over - estimate the ion current to the dust .
interestingly , for dust size small and even comparable to debye length , this correction appears to be small and oml prediction of dust potential remains reliable . is shown for ions in a hydrogen plasma with different values of @xmath45 or @xmath48 @xmath49 can have extrema at @xmath47 that reflects some ions ( @xmath50 ) .
the oml approximation ignores this absorption radius effect , which can over - estimate the ion current ( @xmath51 ) and the ion density between dust surface and the absorption radius ( @xmath52 ) . ]
a far more serious problem was identified by allen , annaratone , and de angelis in 2000 @xcite that the oml theory can not predict the plasma response to the presence of a dust particle , which requires the solution of the oml poisson equation for the plasma potential @xmath53 where @xmath54 is vacuum permittivity .
allen et al .
@xcite used the well - known oml ion and electron density given by alpert , gurevich , and pitaevskii in 1965 @xcite , @xmath55 \exp\left[\frac{\varphi-\varphi_d}{z^2 - 1}\right ] \right\ } \exp \left(\varphi\right ) .
\label{eq : ne - oml}\\ \frac{n_i(z)}{n_{i0 } } = & \sqrt{-\frac{z\beta\varphi}{\pi } } \left[1 + \sqrt{1 - \frac{\varphi_d}{z^2\varphi}}\right ] + \frac{e^{-z\beta\varphi}}{2 } \left [ 1 - { \rm erf}\left(\sqrt{-z\beta\varphi}\right)\right ] + \frac{\sqrt{1- z^{-2}}}{2 } e^{- z\beta\tilde{\varphi } } \left [ 1 - { \rm erf}\left(\sqrt{-z\beta\tilde{\varphi}}\right ) \right ] , \label{eq : ni - oml - alpert}\end{aligned}\ ] ] where @xmath56 the failure of oml theory manifests in an imaginary ion density when @xmath57 for the initially faster than @xmath43 increase in @xmath22 due to debye shielding , which is the cause for ion absorption radius at @xmath47 as noted previously for a negatively charged dust , allen et al . concluded that the contradiction between @xmath22 and an imaginary @xmath58 in eq .
( [ eq : ni - oml - alpert ] ) would be an intrinsic defect which prevents the oml theory from predicting the plasma response . since the dust charge is related to the normal electric field at the dust surface , which requires the solution of @xmath59 one reaches the inevitable position from ref .
@xcite that oml theory can not predict the dust charge , despite its success on dust potential as re - affirmed in ref .
intuitively it is quite puzzling that the oml simplification of ignoring the absorption radius effect , which contributes a small error in the ion charging current , would produce an imaginary ion density , as revealed in allen et al.s analysis .
physically , a maxima in ion @xmath61 at @xmath47 would pose a barrier that turn back ions with a range of @xmath39 ignoring this effect with the oml approximation should manifest in an over - estimation of the ion density for @xmath62 so why allen et al .
discovered an apparently inherent contradiction in the oml theory between @xmath22 and @xmath63 following the physical picture just given , one is tempted to conclude that the resolution has to come from a revision of the oml expression for @xmath64 as given in eq .
( [ eq : ni - oml - alpert ] ) .
we find that this is indeed the case .
the cause is a change in integration bound for the oml ion density when the plasma potential transitions from @xmath65 to @xmath66 to understand this subtlety , which has been evidently elusive for the past five decades , we recall that since oml assumes a monotonically varying @xmath67 the ions at @xmath10 have a collisionless orbit that will either intercept the dust particle or be reflected by the effective potential before it can reach the dust surface . in the canonical case that @xmath68 the birth energy of the background plasma ion far away
must have @xmath69 if this ion reaches @xmath16 it must have , at @xmath16 that @xmath70 for a negatively charged dust which has @xmath71 a plasma ion of such unbounded orbit ( meaning that the ion orbit connects to infinity ) must have higher kinetic energy as it approaches the dust , @xmath72 as illustrated in fig .
[ fig : oml - integration - bounds ] , this is outside a circle in @xmath73 space , which intercepts the @xmath74 axis at @xmath75 not all of these ions can reach the dust surface ( @xmath76 ) due to angular momentum conservation . in the oml approximation , the effective potential @xmath61 peaks at @xmath77 @xmath78 the ions with @xmath79 will be reflected by the effective potential before they can reach @xmath19 at @xmath80 these reflected ions satisfy , after explicitly writing out @xmath81 @xmath82 for a negatively charged dust with @xmath83 the reflected ions are bounded by a parabola in @xmath73 space , @xmath84 as illustrated in fig .
[ fig : oml - integration - bounds ] , it intercepts the @xmath74 axis at @xmath85 as we shall see , this should be compared with the intercept of eq .
( [ eq : unbounded - orbit ] ) with @xmath74 axis , i.e. @xmath86 in eq .
( [ eq : vt - unbounded ] ) .
to evaluate the ion density at @xmath87 which is the solution of the vlasov equation , we integrate the maxwellian distribution , which is assumed for the background plasma far away , in the velocity space region where they are allowed @xcite . for ions with @xmath88
their population lies in the region of the @xmath89 velocity space given by @xmath90 for ions with @xmath91 only the reflected ions are present so they are in a region of the velocity space given by @xmath92 these give rise to integration bounds for @xmath93 and @xmath94 separately .
the one for @xmath93 is straightforward , see fig .
[ fig : oml - integration - bounds ] , and the previous oml form @xcite is correct .
the one for @xmath94 is complicated by the possibility that @xmath95 which implies @xmath96 if this is the case , the integration has two zones , separated by a set of critical @xmath97 and @xmath98 which are the intercept of the two constraints , @xmath99 and @xmath100 @xmath101 as shown in the left - side diagram of fig .
[ fig : oml - integration - bounds ] , the integration can be carried out in two zones , depending on whether @xmath102 or not .
the lower zone ( denoted as i ) is @xmath103 \,\,\ , \ & \,\,\ , v_t \in [ \sqrt{-({2ze}/{m_i})\phi - v_r^2 } , \infty).\end{aligned}\ ] ] the upper zone ( denoted as ii ) is @xmath104 the ion density is given by @xmath105 which yields eq .
( [ eq : ni - oml - alpert ] ) , _ now valid only for @xmath106_. ion density @xmath107 and electron density @xmath108 are plotted as functions of @xmath109 for three cases of @xmath110 ( black ) , @xmath111 ( red ) , and @xmath112 ( green ) . ] in the case that @xmath113 i.e. , @xmath114 the integration is restricted to a single zone given by @xmath115 which is illustrated in the right - side diagram of fig .
[ fig : oml - integration - bounds ] .
the ion density then takes the form @xmath116 + \frac{\sqrt{1- z^{-2}}}{2 } e^{- z\beta\tilde{\varphi}},\,\,\,\,\,\ , \phi\ge\phi_d / z^2 , \label{eq : ni - oml - corrected}\end{aligned}\ ] ] where @xmath117 while eq .
( [ eq : ni - oml - alpert ] ) should be used for @xmath118 contrasting eq .
( [ eq : ni - oml - corrected ] ) with eq .
( [ eq : ni - oml - alpert ] ) , one sees that the previously - known imaginary oml ion density for @xmath119 is removed in the corrected oml theory for @xmath120
with the corrected oml ion density , the oml poisson equation ( [ eq : oml - poisson ] ) can be solved for the plasma response .
it is important to note that the radial electric field at the dust surface @xmath121 is related to the dust charge according to gauss s law , @xmath122 since the oml theory to date can not predict @xmath123 users of the oml charging theory have been using a simple relation for spherical dust capacitance due to whipple @xcite , @xmath124 with @xmath125 the electron debye length .
whipple s idea is to expand the plasma density at @xmath126 as in the standard debye shielding calculation . applying this approach to oml s poisson equation
, one finds @xmath127 and hence @xmath128 introducing the so - called linearized debye length @xmath129 and defining the normalized dust radius as @xmath130 one finds the solution to the poisson equation , @xmath131.\label{eq : oml - debye - shielding}\end{aligned}\ ] ] hence the dust charge is @xmath132 unlike the original whipple formula [ eq .
( [ eq : wipple - formula ] ) ] which assumes a uniform ion density , eq .
( [ eq : oml - wipple - formula ] ) takes into account the ion density response .
it is interesting to note that for a small dust @xmath133 the debye shielding contribution , which is usually small , can be enhanced substantially if @xmath134 and/or @xmath135 we will call the charge - potential relationship given in eq .
( [ eq : oml - wipple - formula ] ) the generalized whipple approximation , to distinguish it from the well - known whipple formula in eq .
( [ eq : wipple - formula ] ) . the actual oml density for @xmath136 can deviate significantly from the asymptotic expansion valid when @xmath137 see fig .
[ fig : oml - plasma - profile ] .
this is already evident from the analytical form of @xmath108 and @xmath58 in oml theory . since the dust charge can be alternatively computed by integrating the net charge of the plasma , @xmath138 due to overall charge conservation , one suspects that the oml prediction of dust charge can be significantly different from that in eq .
( [ eq : oml - wipple - formula ] ) . from the numerically evaluated @xmath108 and @xmath58 shown in fig .
[ fig : oml - plasma - profile ] , we find that sharper deviation occurs closer to @xmath139 as the dust size becomes smaller , but there is a greater spatial extent of the deviation for large dust size .
since the dust charge corresponds to the spatial integration of electron and ion density , one finds that such deviation from the simple debye shielding calculation is proportional to @xmath140 in fig .
[ fig : oml - scan - dust - size ] , we plot @xmath141 as a function of @xmath140 the specific case has @xmath142 and @xmath143 so the generalized whipple approximation , eq .
( [ eq : oml - wipple - formula ] ) , is simply @xmath144 the deviation of the oml prediction of dust charge from whipple approximation can be significantly greater if the dust size is approaching the debye shielding length , which is consistent with the results of ref .
this is ultimately due to the effect of angular momentum conservation on the plasma density near the dust particle .
obviously a significant correction in @xmath145 implies a substantial change in the electrical force @xmath146 which can impact the dust dynamics , for example , in a tokamak reactor @xcite .
in conclusion , we have resolved a long - standing issue in the oml charging theory , and by doing so , obtained a complete oml theory that predicts both dust potential and dust charge .
this is enabled by a revised oml ion density formula for the case of @xmath147 which is given in eq .
( [ eq : ni - oml - corrected ] ) .
we also provide the first calculation of the plasma potential and the dust charge using the oml theory .
our results show that for applications where the dust particulate radius is much smaller than the plasma debye length , the whipple approximation is in good agreement with the oml prediction of the dust charge - potential relationship . when the dust size is comparable to or larger than the debye shielding length , which is a case of importance to magnetic fusion , we discover significant deviation from the whipple approximation of the dust charge - potential relationship .
this is attributed to the fundamental role of angular momentum conservation in setting the plasma electron and ion density near the dust particle .
the deviation from eq .
( [ eq : oml - wipple - formula ] ) becomes large when the dust size becomes comparable or greater than the debye length . ]
this research was supported by the u.s .
department of energy office of science , office of fusion energy sciences , under the auspices of the national nuclear security administration of the u.s .
department of energy by los alamos national laboratory , operated by los alamos national security llc under contract de - ac52 - 06na25396 .
27ifxundefined [ 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 10.1073/pnas.14.8.627 [ * * , ( ) ] link:\doibase 10.1103/physrev.28.727 [ * * , ( ) ] @noop _ _
( , , ) @noop _ _ ( , , ) @noop _ _ ( , , ) link:\doibase 10.1103/revmodphys.81.1353 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) link:\doibase 10.1146/annurev.astro.41.011802.094840 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop _ _ , ( , ) http://permalink.lanl.gov/object/view?what=info:lanl-repo/isi/000171401000002 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) link:\doibase 10.1088/0031 - 8949/45/5/013 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1007/s10894 - 010 - 9295-x [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1017/s0022377803002265 [ * * , ( ) ] @noop * * , ( )
as pointed out by allen et al @xcite , the oml ion density can become imaginary if one is to solve the poisson equation , eq . ( [ eq : oml - poisson ] ) , for the plasma potential using the previously known oml ion density formula by alpert et al @xcite .
this suggests a fundamental breakdown of the oml theory for plasma potential and hence dust charge calculation .
allen et al investigated this problem from the angle of the oml approximation , i.e. , the negelect of the ion absorption radius effect . in the main text
, we give a physics argument that ignoring the ion absorption radius effect should only introduce a discrepancy in ion density compared to om prediction , but not an unphysical imaginary number .
our corrected calculation of the oml ion density removes the possibility of an imaginary density .
here we give a more detailed account of how the imaginary ion density comes about in the previous oml formulation .
the objective is to further clarify the contrast between ( 1 ) a physical approximation ( oml ) that introduces quantitative discrepency which vanishes in its limit of applicability , and ( 2 ) an invalid theoretical formulation that produces unphysical results . the condition for @xmath148 to become imaginary ( hence unphysical ) in eq .
( [ eq : ni - oml - alpert ] ) is to have @xmath149 for a negatively charged dust ( @xmath150 and @xmath151 ) , this implies a plasma potential @xmath152 that rises faster than @xmath153 with @xmath154 the normalized distance from the dust surface . as long as debye shielding is in action ( i.e. the plasma transport to the dust surface is not intrinsically ambipolar ) , we know that @xmath155 would rise exponentially as a function of @xmath156 within the debye shielding sphere . the cross - over point after which eq .
( [ eq : cross - over ] ) is satisfied can be illustrated using the debye shielding potential solution of eq .
( [ eq : oml - debye - shielding ] ) by setting @xmath157 the cross - over point @xmath158 is the solution of @xmath159 = 1 .
\label{eq : zc}\end{aligned}\ ] ] here @xmath160 is the dust radius normalized against the debye length and @xmath154 is the radial distance from the dust surface normalized against the dust size . for @xmath161
one finds @xmath162 or 6.48 deybe length away . if @xmath163 we have @xmath164 or 11.7 debye length away . in other words , the breakdown of the ion density formula , eq .
( [ eq : ni - oml - alpert ] ) , is the direct result of the debye shielding physics , which is independent of how small the dust size is compared to the debye length . of course , the debye shielding potential as given in eq .
( [ eq : oml - debye - shielding ] ) is an analytical approximation , so the actual cross - over point @xmath158 would likely differ in its exact location from that predicted by eq .
( [ eq : zc ] ) . when the oml poisson equation is numerically solved , the failure of the oml formulation using the ion density given in eq .
( [ eq : ni - oml - alpert ] ) manifests in a not - a - number floating point violation for all small @xmath160s we have attempted , as one would expect by taking the squared root of a negative number .
this is to be contrasted with the corrected oml formulation which produces the numerical solutions shown in section [ sec : oml - result ] . | the foundational theory for dusty plasmas is the dust charging theory that provides the dust potential and charge arising from the dust interaction with a plasma .
the most widely used dust charging theory for negatively charged dust particles is the so - called orbital motion limited ( oml ) theory , which predicts the dust potential and heat collection accurately for a variety of applications , but was previously found to be incapable of evaluating the dust charge and plasma response in any situation . here
we report a revised oml formulation that is able to predict the plasma response and hence the dust charge .
numerical solutions of the new oml model show that the widely - used whipple approximation of dust charge - potential relationship agrees with oml theory in the limit of small dust radius compared with plasma debye length , but incurs large ( order - unity ) deviation from the oml prediction when the dust size becomes comparable with or larger than plasma debye length .
this latter case is expected for the important application of dust particles in a tokamak plasma . |
inflation has been the backbone of the high - energy cosmology over the past 3 decades @xcite .
the most simple source for inflation is a minimally coupled scalar field @xmath0 ( `` inflaton '' ) with a slowly varying potential @xmath1 @xcite the spectra of density perturbations generated from the quantum fluctuations of inflaton are consistent with the temperature anisotropies observed in the cosmic microwave background ( cmb ) @xcite . from the amplitude of the observed cmb anisotropies @xcite the typical mass scale of inflation
is known to be around @xmath2 gev @xcite .
this is much larger than the electroweak scale ( @xmath3gev ) , which suggests the requirement of new physics beyond the standard model of particle physics @xcite . in other words , for the potential @xmath4 with @xmath5 gev
, the coupling @xmath6 is constrained to be @xmath7 from the cmb normalization @xcite , but this is much smaller than the coupling constant @xmath8 of the higgs boson @xcite . there have been attempts to accommodate the higgs field for inflation .
one is to use a non - minimal field coupling @xmath9 with the ricci scalar @xmath10 @xcite ( see also refs .
@xcite ) . if @xmath11 the self coupling @xmath6 can be as large as @xmath12 from the cmb normalization @xcite .
although this scenario is attractive , it is plagued by the unitary - violation problem associated with graviton exchange in @xmath13 scalar scattering around the energy scale @xmath14 ( where @xmath15 gev is the reduced planck mass ) @xcite .
since @xmath16 is around the energy scale of inflation , some strong coupling effect can give rise to additional corrections to the inflaton potential .
another attempt is to employ a field derivative coupling with the einstein tensor @xmath17 , i.e. @xmath18 , where @xmath19 is a constant having a dimension of mass @xcite ( see also ref .
@xcite for the original work ) . in the regime where the hubble parameter @xmath20 is larger than @xmath19
the field evolves more slowly relative to the case of standard inflation due to a gravitationally enhanced friction .
hence it is possible to reconcile steep potentials such as @xmath21 ( @xmath8 ) with the cmb observations . in refs .
@xcite it was shown that , for a slow - rolling scalar field satisfying the condition @xmath22 , the strong coupling scale @xmath16 of the derivative coupling theory is around @xmath23 in a homogeneous and isotropic cosmological background .
provided that @xmath20 and @xmath19 are below the planck scale , the theory is in a weak coupling regime with suppressed quantum corrections .
the property of the high cut - off scale @xmath16 around @xmath23 is associated with the fact that whenever the non - minimal derivative coupling to gravity dominates over the canonical kinetic term the theory possesses an asymptotic local shift symmetry for @xmath24 @xcite .
this symmetry is related to the galilean symmetry @xmath25 in minkowski space - time @xcite , but the difference is that the coordinate @xmath26 in the derivative coupling theory on curved backgrounds is linked to the covariantly constant killing vectors @xcite . in the presence of a slowly varying inflaton potential such a local symmetry is only softly broken , so that the potential can be protected against quantum corrections during inflation .
the field self - interaction of the form @xmath27 @xcite , which satisfies the galilean symmetry in the limit of minkowski space - time , also leads to the slow evolution of @xmath0 along the inflaton potential @xcite ( see also refs .
@xcite ) .
a nice feature of the non - minimal derivative coupling with the einstein tensor is that the mechanism of the gravitationally enhanced friction works for general steep potentials .
for instance , let us consider the potential of natural inflation , @xmath28 $ ] , where @xmath29 characterizes the scale of the breaking of a global shift symmetry @xcite . in order for this potential to be consistent with the cmb observations , we require that @xmath29 is larger than @xmath30 in conventional slow - roll inflation @xcite .
then the global symmetry is broken above the quantum gravity scale , in which case quantum field theory may be invalid @xcite . in the presence of the non - minimal derivative coupling to gravity , however , the scale @xmath29 can be much smaller than @xmath23 because of the gravitationally enhanced friction @xcite .
another example is the exponential potential @xmath31 , whose dominance leads to the power - law expansion of the universe ( with the scale factor @xmath32 , where @xmath33 is cosmic time ) @xcite . in higher - dimensional gravitational theories ,
exponential potentials often arise as the curvature of internal spaces related with the geometry of extra dimensions @xcite . in such cases
the constant @xmath34 is usually larger than the order of unity , so that it is difficult to realize sufficient amount of inflation .
as we will see later , this problem can be circumvented by taking into account the non - minimal derivative coupling . moreover , unlike the standard case , inflation comes to end with gravitational particle production . in order to test the viability of inflationary models with the non - minimal derivative coupling it is important to estimate the power spectra of density perturbations relevant to the cmb anisotropies . in refs .
@xcite the authors computed the inflationary observables such as the scalar spectral index @xmath35 , the tensor - to - scalar ratio @xmath36 , and the nonlinear parameter @xmath37 of the equilateral scalar non - gaussianities ( see also refs .
since the scalar propagation speed is close to the speed of light during inflation , the scalar non - gaussianities are suppressed to be small ( @xmath38 ) .
hence @xmath35 and @xmath36 are the two main observables to distinguish between different inflaton potentials .
in this paper we shall place observational constraints on a number of representative inflationary models in the presence of the field derivative coupling with the einstein tensor .
we use the bounds derived from the joint data analysis of wmap7 @xcite , baryon acoustic oscillations ( bao ) @xcite , and the hubble constant measurement ( hst ) @xcite . note that some constraints on higgs inflation and natural inflation have been discussed in refs .
@xcite without the cmb likelihood analysis . in ref .
@xcite the author carried out the cosmological monte - carlo simulation to test higgs inflation with the field derivative coupling to gravity .
our analysis based on the recent observational data is general enough to cover a wide variety of models such as chaotic inflation @xcite , inflation with exponential potentials @xcite , natural inflation @xcite , and hybrid inflation @xcite .
we show that the gravitationally enhanced friction mechanism can make most of the models compatible with the current observations .
we start with the following 4-dimensional action s = d^4x , [ action ] where ^=g^ -g^. [ delta ] here @xmath39 is a determinant of the space - time metric @xmath40 , @xmath10 is the ricci scalar , @xmath17 is the einstein tensor , @xmath19 is a constant having a dimension of mass , and @xmath1 is the potential of a scalar field @xmath0 . the action ( [ action ] ) belongs to a class of the most general scalar - tensor theories having second - order equations of motion ( which is required to avoid the ostrogradski instability ) @xcite .
the lagrangian in such general horndeski s theories is the sum of the terms @xmath41 , @xmath42 , @xmath43 $ ] , and @xmath44 $ ] , where @xmath45 , @xmath46 ( @xmath47 ) are functions of @xmath0 and @xmath48 , and @xmath49 @xcite .
the conditions for the avoidance of ghosts and laplacian instabilities were derived in refs .
these conditions can be used to restrict the functional forms of @xmath45 , @xmath46 ( @xmath47 ) to construct theoretically consistent models of inflation .
the non - minimal derivative coupling in eq .
( [ action ] ) is recovered in the horndeski s lagrangian by choosing the function @xmath50 after integration by parts .
the sign in front of the term @xmath51 in eq .
( [ delta ] ) is chosen to avoid the appearance of ghosts in the scalar sector @xcite . in ref .
@xcite it was found that in a manifold having integrable ( covariantly constant ) killing vectors @xmath52 the field lagrangian @xmath53 in eq .
( [ action ] ) is invariant under the ( curved - space ) galilean transformation @xmath54 , where @xmath55 , @xmath56 , @xmath57 are constants and @xmath58 is a space - time coordinate .
the existence of the galilean symmetry has an advantage that the theory can be quantum mechanically under control @xcite . imposing the above galilean symmetry in the curved background with integrable killing vectors , germani _ et al . _
@xcite showed that the second - order lagrangians are restricted to take the forms @xmath53 or @xmath59 ( plus a field derivative coupling with the double dual riemann tensor ) . in the small derivative regime
in which the condition @xmath60 is satisfied ( e.g. , during inflation ) , an approximate infinitesimal shift symmetry @xmath61 ( where @xmath62 is an arbitrary function of space - time coordinates @xmath58 ) emerges for the lagrangian @xmath18 , provided that the metric is shifted appropriately @xcite .
the existence of such a gauge symmetry can allow the theory ( [ action ] ) to be protected against quantum corrections even up to the planck scale .
note that the term @xmath59 does not possess such a general gauge shift symmetry .
the scale of unitarity violation for the theory ( [ action ] ) was estimated in ref .
@xcite in the context of higgs inflation . in standard model
we can consider the @xmath63 scattering via graviton exchange , where @xmath0 is one of the real scalar degrees of freedom for the higgs doublet .
we expand the metric in the form @xmath64 , where @xmath65 is the metric on the flat friedmann - lematre - robertson - walker ( flrw ) background ( @xmath66 is the scale factor with cosmic time @xmath33 ) .
we are interested in the high - friction regime in which the hubble parameter @xmath67 ( a dot represents a derivative with respect to @xmath33 ) is much larger than @xmath19 . in this regime
the field @xmath0 is expanded as @xmath68 , where @xmath69 is a canonically normalized field perturbation . the first non - renormalizable operator associated with the interaction between gravitons and scalars is given by @xcite i= ^2 h^ _ _ .
[ inter ] a power counting analysis gives the unitary bound @xmath70 .
for the suppression of higher dimensional operators we require the condition @xmath71 . on using the relation @xmath72
this condition translates into @xmath73 , which is satisfied during inflation .
the discussion of the unitary bound given above can be applied to the multi - field inflationary models in which one of the fields is not necessarily responsible for the cosmic acceleration . in single field models where only one field @xmath0 leads to inflation
the unitary bound can be as close as the planck scale @xmath23 in the regime where the condition @xmath60 is satisfied @xcite . in this case
the slow - roll evolution of the field @xmath0 suppresses the interaction ( [ inter ] ) below the planck scale . in
the following let us study the background dynamics for the theory described by the action ( [ action ] ) .
in the flat flrw background the equations of motion following from the action ( [ action ] ) are & & 3 m_pl^2 h^2= 12 ( 1 + 9 ) + v ( ) , [ hubbleeq ] + & & + v_,=0 , [ jeq ] where @xmath74 . in order to solve the dynamical equations numerically , it is convenient to introduce the following dimensionless variables x=,y=,z=. differentiating eq .
( [ hubbleeq ] ) with respect to @xmath33 and using eq .
( [ jeq ] ) to eliminate @xmath75 , we obtain the second - order equation for the field @xmath0 .
it then follows that & = & y , [ dxeq ] + & = & - , [ dyeq ] + z & = & , [ zeq ] where = m t,_,= , = . we are interested in slow - roll inflation in which cosmic acceleration is mainly driven by the potential energy @xmath1 . in this case
( [ hubbleeq ] ) and ( [ jeq ] ) reduce to & & 3m_pl^2 h^2 v ( ) , [ sloweq1 ] + & & 3h a + v _
, 0 , [ sloweq2 ] where = 1 + 3 .
[ cala ] we define the following slow - roll parameters & & = -,_= , + & & _ x= , _ d=. for the validity of the slow - roll approximation we require that @xmath76 . taking the time - derivative of eq .
( [ sloweq1 ] ) and using eq .
( [ sloweq2 ] ) , we have _
x+6_d , [ epre ] where _ = ( ) ^2 .
[ epv ] this shows that @xmath77 for @xmath78 and hence the evolution of the field @xmath0 slows down relative to that in standard slow - roll inflation .
the field value @xmath79 at the end of inflation is known by solving @xmath80 , i.e. _ ( _ f ) ^-1=1 .
[ epv2 ] the number of e - foldings from the time @xmath33 during inflation to the time @xmath81 at the end of inflation is defined by @xmath82 .
on using eqs .
( [ sloweq1 ] ) and ( [ sloweq2 ] ) , it follows that n _ _
f^ ( 1 + ) d. [ efold ] if @xmath78 ( i.e. @xmath83 ) during inflation , we can neglect the first term inside the bracket of eq .
( [ efold ] ) relative to the second one .
this is not the case for inflation in which the transition from the regime @xmath84 to the regime @xmath85 occurs prior to the onset of reheating . versus @xmath86 for ( i ) @xmath87 and ( ii ) @xmath88 .
the initial conditions at @xmath89 are chosen by using the slow - roll equations ( [ sloweq1 ] ) and ( [ sloweq2 ] ) , i.e. , ( i ) @xmath90 , @xmath91 , and ( ii ) @xmath92 , @xmath93 .
[ fig1],width=326,height=307 ] as an example , let us consider chaotic inflation @xcite with the potential v()= ^n , where @xmath6 and @xmath94 are constants .
in this case eqs .
( [ efold ] ) and ( [ epv2 ] ) read & & n = - , [ chaotinre ] + & & 2 x_f^2 ( 1 + x_f^n ) = n^2 , [ chaotinre2 ] where = , x_f=. for the quadratic potential @xmath95 ( i.e. @xmath96 and @xmath97 ) one has @xmath98 and x^2= ( -2 ) / , [ xslow ] where @xmath99 . in the general relativistic ( gr ) limit ( @xmath100 ) this gives @xmath101 . in the high - friction limit ( @xmath102 ) one has @xmath103 , which means that the field value is smaller than that in standard chaotic inflation . in order to confirm the accuracy of the slow - roll approximation we solve the full equations of motion ( [ dxeq])-([zeq ] ) numerically . in fig .
[ fig1 ] the evolution of the field @xmath0 is plotted for the potential @xmath95 , i.e. @xmath104 and @xmath105 in eqs .
( [ dyeq ] ) and ( [ zeq ] ) .
we choose the initial conditions of @xmath58 and @xmath106 at @xmath89 by using the values derived under the slow - roll approximation .
the numerical simulations labeled as ( i ) and ( ii ) in fig . [ fig1 ] correspond to the parameters @xmath87 and @xmath88 , respectively . in the case ( i )
the numerical value of @xmath107 at the end of inflation ( @xmath108 ) is @xmath109 , which means that the solution is in the high - friction regime during inflation . in this case
@xmath20 drops below @xmath19 at the reheating stage .
after inflation there is a transient period with @xmath84 in which the slow - roll condition is violated . in this regime some quantum corrections may come into play to the action ( [ action ] ) . as long as such corrections are unimportant in the field equations ( [ hubbleeq ] ) and ( [ jeq ] )
, we find that the inflaton oscillation is not disturbed during the transient period ( see fig .
[ fig1 ] ) . in the case
( ii ) we have @xmath110 at @xmath111 and hence the system enters the regime @xmath85 during inflation . in this case
the qualitative behavior for the oscillation of inflaton at reheating is not much different from that in standard inflation .
we confirmed that the difference between the numerical and analytic values of the number of e - foldings acquired during inflation is usually less than a few percent .
this shows that the slow - roll approximation employed in eqs .
( [ sloweq1 ] ) and ( [ sloweq2 ] ) can be trustable .
the spectra of scalar and tensor perturbations generated in the theories given by the action ( [ action ] ) were derived in refs .
@xcite . here
we briefly review their formulas in order to apply them to concrete inflationary models in sec .
[ obsercon ] .
the perturbed metric about the flat flrw background is given by @xcite ds^2 & = & -(1 + 2a)dt^2 + 2_i bdtdx^i + & & + a^2(t)dx^i dx^j , [ permet ] where @xmath112 , @xmath113 , and @xmath114 are scalar metric perturbations , and @xmath115 are tensor perturbations which are transverse and traceless .
the spatial part of a gauge - transformation vector @xmath116 is fixed by gauging away a perturbation @xmath117 appearing as a form @xmath118 in the metric ( [ permet ] ) .
we decompose the inflaton field into the background and inhomogeneous parts , as @xmath119 . in the following we choose the uniform - field gauge characterized by @xmath120 , which fixes the time - component of the vector @xmath116 . expanding the action ( [ action ] ) up to second order in perturbations and using the hamiltonian and momentum constraints
, we obtain the second - order action for scalar perturbations @xcite s_s^(2)=dtd^3x a^3 q_s , where q_s & = & , [ qdef ] + c_s^2 & = & , and & & w_1=m_pl^2 ( 1 - 2_d),w_2=2 h m_pl^2 ( 1 - 6_d ) , + & & w_3=-3h^2 m_pl^2 ( 3-_x -36 _ d ) , + & & w_4=m_pl^2 ( 1 + 2_d ) . in order to avoid the appearance of scalar ghosts and laplacian instabilities we require that @xmath121 and @xmath122 . picking up the dominant contributions to @xmath123 and @xmath124 under the slow - roll approximation
, we obtain q_s & & m_pl^2 ( _ x+6_d ) m_pl^2m_pl^2 , [ qs ] + c_s^2 & & 1- , [ cs ] which mean that @xmath125 .
the power spectrum of the curvature perturbation @xmath114 , which is evaluated at @xmath126 ( @xmath127 is a comoving wavenumber ) , is given by _
s= ( 1 + ) , [ cobe ] where in the last approximate equality we used eqs .
( [ sloweq1 ] ) , ( [ cala ] ) , ( [ epv ] ) , ( [ qs ] ) , and ( [ cs ] ) .
the scalar spectral index is n_s-1 & = & |_c_sk = ah + & & - , [ ns ] where @xmath128 is defined in eq .
( [ epv ] ) , and _ = m_pl^2 . in the high - friction limit ( @xmath78 )
one has @xmath129 with @xmath130 , whereas @xmath131 in the gr limit ( @xmath132 ) .
the intrinsic tensor perturbation @xmath115 can be decomposed into two independent polarization modes , i.e. @xmath133 . in fourier space
we normalize the two modes , as @xmath134 ( where @xmath135 ) and @xmath136 .
then the second - order action for tensor perturbations can be written as @xcite s_t^(2)=_pdtd^3x a^3q_t , where q_t & = & w_1/4 = m_pl^2 ( 1 - 2_d)/4 , + c_t^2 & = & w_4/w_1 = 1 + 4_d+o ( ^2 ) .
this shows that , unlike the scalar propagation speed squared , @xmath137 is slightly larger than 1 during inflation . since lorentz invariance is explicitly broken on the flrw background , the superluminal mode does not necessarily imply a violation of causality .
the tensor power spectrum is given by _
t= , which is evaluated at @xmath138 .
the tensor spectral index is n_t = |_c_tk = ah -2the tensor - to - scalar ratio is r = |_k ah 16 , [ ratio ] from which we obtain the consistency relation r -8 n_t .
[ consistency ] this relation is the same as that in conventional inflation at leading order in slow - roll . the running spectral indices @xmath139 and @xmath140 are second order in slow - roll parameters . they are set to be 0 in the cmb likelihood analysis . the consistency relation ( [ consistency ] )
reduces the inflationary observables to three , i.e. , @xmath35 , @xmath36 , and @xmath141 .
these observables are varied in the likelihood analysis with the pivot wavenumber @xmath142mpc@xmath143 , by assuming the flat @xmath144-cold - dark - matter model . since @xmath125 , the general formula for the equilateral non - gaussianities of scalar perturbations @xcite shows that the nonlinear parameter @xmath37 is of the order of @xmath145 @xcite .
hence the scalar non - gaussianities do not provide additional constraints to those derived from the linear perturbations .
in the presence of the field derivative coupling to the einstein tensor we place observational constraints on a number of models such as ( i ) chaotic inflation , ( ii ) inflation with exponential potentials , ( iii ) natural inflation , and ( iv ) hybrid inflation .
our analysis covers most of the representative inflaton potentials proposed in literature .
we start with chaotic inflation characterized by the potential v()= ^n , where , for @xmath97 , we use the notation @xmath96 as in the previous section . under the slow - roll approximation
the dimensionless field @xmath146 is related with the number of e - foldings @xmath147 as eq .
( [ chaotinre ] ) . from eqs .
( [ ns ] ) and ( [ ratio ] ) the scalar spectral index and the tensor - to - scalar ratio are given , respectively , by n_s & = & 1- , [ nschao ] + r & = & , [ rchao ] where @xmath148 .
in the ( @xmath149 ) plane with the three different numbers of e - foldings ( @xmath150 ) .
we evaluate the theoretical values of @xmath35 and @xmath36 in the range @xmath151 .
the thin solid curves correspond to the 1@xmath152 ( inside ) and 2@xmath152 ( outside ) observational contours constrained by the joint data analysis of wmap7 , bao , and hst . for larger values of @xmath99 the tensor - to - scalar ratio @xmath36 gets smaller , whereas the scalar spectral index increases .
[ fig2],width=326,height=316 ] in the limit that @xmath100 one has @xmath153 from eqs .
( [ chaotinre ] ) and ( [ chaotinre2 ] ) , in which case @xmath35 and @xmath36 are n_s & = & 1- , [ chao1d ] + r & = & .
[ chao1 ] these values correspond to those for standard chaotic inflation . if @xmath89 , then @xmath154 , @xmath155 for @xmath97 and @xmath156 , @xmath157 for @xmath158 . in another limit
@xmath102 one has @xmath159n^2/(2\alpha)$ ] and n_s & = & 1- , [ chao2d ] + r & = & .
[ chao2 ] if @xmath89 , then @xmath160 , @xmath161 for @xmath97 and @xmath162 , @xmath163 for @xmath158 . in the intermediate values of @xmath164 between @xmath165
we need to solve eq .
( [ chaotinre ] ) for @xmath58 by using eq .
( [ chaotinre2 ] ) . when @xmath97 the field value @xmath58 can be expressed as eq .
( [ xslow ] ) , in which case @xmath35 and @xmath36 are known from eqs .
( [ nschao ] ) and ( [ rchao ] ) for given values of @xmath99 and @xmath147 . in fig .
[ fig2 ] we plot the theoretical values of @xmath35 and @xmath36 for @xmath97 as a function of @xmath164 ( between @xmath166 ) with three different values of @xmath167 , together with the @xmath168 and @xmath169 observational contours constrained by the joint data analysis of wmap7 @xcite , bao @xcite , and hst @xcite . for @xmath170
these observables are close to the values estimated by eqs .
( [ chao1d ] ) and ( [ chao1 ] ) , whereas for @xmath171 they are close to those given by eqs .
( [ chao2d ] ) and ( [ chao2 ] ) . for larger
@xmath164 , @xmath36 gets smaller whereas @xmath35 increases , so that the quadratic potential shows better compatibility with observations in the presence of the field derivative coupling to gravity . in the range
@xmath172 with the three different values of @xmath147 . the @xmath168 and @xmath169 observational contours are the same as those in fig . [ fig2 ]
. while the standard case ( @xmath173 ) is outside the @xmath169 bound , the field derivative coupling to gravity can make the quartic potential compatible with observations .
[ fig3],width=326,height=316 ] in fig .
[ fig3 ] the theoretical values of @xmath35 and @xmath36 are plotted for @xmath158 as a function of @xmath174 between @xmath175 with @xmath150 . in the limit that @xmath100
the quartic potential is outside the @xmath169 observational contour for @xmath147 smaller than 70 . in the presence of the field derivative coupling to gravity
the model can be compatible with the current observations due to the suppressed tensor - to - scalar ratio and the larger spectral index . for @xmath89
the parameter @xmath164 is constrained to be > 3 10 ^ -5 ( 95% cl ) , and @xmath176 ( 68 % cl ) .
for @xmath177 the constraints are @xmath178 ( 95 % cl ) and @xmath179 ( 68 % cl ) .
using the scalar power spectrum ( [ cobe ] ) , the cmb normalization by wmap @xcite corresponds to ( ) ^2 x_60^n+2 ( n+x_60^n ) 2.4 10 ^ -9 , [ cmbnor ] where @xmath180 is the value of @xmath58 at @xmath89 . in the regime @xmath181 one
has @xmath159n^2/(2\alpha)$ ] , so that eq .
( [ cmbnor ] ) gives & & 1.5 10 ^ -10 n=2 , + & & 5.9 10 ^ -32 ( ) ^4 n=4 . in the case of the quartic potential it is possible to realize @xmath182 for @xmath183 .
let us proceed to the exponential potential v()=v_0e^/m_pl , where @xmath184 and @xmath34 are constants . in this case one
has @xmath185 and @xmath186 .
hence in standard slow - roll inflation we require the condition @xmath187 .
moreover this corresponds to the power - law inflation without the graceful exit @xcite . in the presence of the field derivative coupling to gravity , however , the slow - roll parameter @xmath188 can be smaller than 1 even for steep exponential potentials with @xmath189 .
inflation ends when @xmath145 grows to the order of unity , which is followed by reheating with gravitational particle production .
this situation is analogous to that in braneworld inflation where the dominance of the density squared term ( @xmath190 ) can lead to cosmic acceleration for steep exponential potentials @xcite . in the regime @xmath78 for @xmath150 .
the @xmath168 and @xmath169 observational contours are the same as those in fig . [ fig2 ] .
even the steep exponential potentials with @xmath34 larger than 1 can be allowed from the current observational data .
[ fig4],width=326,height=316 ] we focus on the case in which the condition @xmath191 is satisfied during the whole stage of inflation . using eq .
( [ epv2 ] ) the field value @xmath79 at the end of inflation can be estimated as , which implies that @xmath192 . from eq .
( [ efold ] ) it follows that n+12 . in the regime
@xmath78 the scalar spectral index ( [ ns ] ) and the tensor - to - scalar ratio ( [ ratio ] ) reduce to n_s & & 1- , [ expod ] + r & & , [ expo ] which correspond to taking the limit @xmath193 in eqs .
( [ chao2d ] ) and ( [ chao2 ] ) . in fig .
[ fig4 ] we plot the theoretical values of @xmath35 and @xmath36 for @xmath150 as well as the 1@xmath152 and 2@xmath152 observational contours .
this shows that even the steep exponential potentials with @xmath192 are compatible with the current observational data . using the wmap normalization @xmath194 at @xmath89 in the regime @xmath191
, we obtain the constraint 8.8 10 ^ -6 .
[ lamcon ] for @xmath195 one has @xmath196 .
natural inflation @xcite is described by the potential v()=^4 , [ natural ] where @xmath144 and @xmath29 are constants having the dimension of mass . in the absence of the field derivative coupling to gravity
the above potential can be compatible with observational data only for @xmath197 @xcite .
then a global symmetry associated with the pseudo - nambu - goldstone - boson is broken above the quantum gravity scale , in which case standard quantum field theory may not be applicable .
if the potential ( [ natural ] ) originates from the string axion , the regime @xmath198 is not generally realized @xcite .
this problem can be circumvented by taking into account the field derivative coupling to gravity in large - field axion models . ]
@xcite . in the following we focus on the case
in which the condition @xmath78 is satisfied during the whole stage of inflation ( @xmath199 ) . from eq .
( [ epv2 ] ) the end of inflation is characterized by _
f -1 + , [ chif ] where @xmath200 and = .
the number of e - foldings is given by n , [ efoldnatural ] where @xmath201 and ( ) = 2 ( - ) + 2()+ .
the scalar spectral index and the tensor - to - scalar ratio are n_s & = & 1- , [ nsnatural ] + r & = & .
[ rnatural ] for given @xmath202 the field value @xmath203 is known from eq .
( [ chif ] ) . since @xmath69 is determined by eq .
( [ efoldnatural ] ) for given values of @xmath202 and @xmath147 , we can numerically evaluate @xmath35 and @xmath36 as functions of @xmath202 for several different numbers of e - foldings . in fig .
[ fig5 ] we plot @xmath35 and @xmath36 in the range @xmath204 , together with the 1@xmath152 and @xmath169 observational contours . for increasing @xmath202 , both @xmath35 and @xmath36
get larger .
$ ] in the range @xmath205 with @xmath78 for three different values of @xmath147 .
the @xmath168 and @xmath169 observational contours are the same as those in fig . [ fig2 ] . for larger @xmath202 ,
both @xmath35 and @xmath36 increase toward the values @xmath206 and @xmath207 .
[ fig5],width=326,height=316 ] in the limit that @xmath208 it is possible to estimate @xmath35 and @xmath36 analytically . from eq .
( [ chif ] ) one has @xmath209 and hence @xmath203 is close to @xmath210 , as @xmath211 . using eq .
( [ efoldnatural ] ) , it follows that @xmath212 .
then eqs .
( [ nsnatural ] ) and ( [ rnatural ] ) reduce to n_s & & 1- , [ nans ] + r & & .
[ nar ] these values correspond to @xmath97 in eqs .
( [ chao2d ] ) and ( [ chao2 ] ) .
this comes from the fact that in the regime @xmath208 inflation occurs around the potential minimum at @xmath213 . as in the case of the quadratic potential @xmath95 with @xmath214 , natural inflation with @xmath208 is compatible with the current observations .
for @xmath89 the parameter @xmath202 is constrained to be > 7.5 ( 95% cl ) , and @xmath215 ( 68 % cl ) .
for @xmath177 the constraints are @xmath216 ( 95 % cl ) and @xmath217 ( 68 % cl ) .
in the regime @xmath208 the wmap normalization at @xmath89 gives f 8 10 ^ 4 ^1/4 m . if @xmath218 , then one has @xmath219 .
finally we study hybrid inflation with the potential v()=v_0 + 12 m^2 ^2 , where @xmath184 and @xmath220 are constants .
inflation ends at a bifurcation point given by @xmath221 due to the appearance of a tachyonic instability driven by another field @xmath69 . as in the case of the original hybrid inflation @xcite we focus on the regime @xmath222 .
note that in another regime @xmath223 the situation is similar to that in chaotic inflation discussed in sec .
[ chaoticsec ] .
using eq .
( [ efold ] ) the field value can be estimated as _
c ( n ) , where @xmath224 and @xmath225 are positive constants defined by = , = . from eqs . (
[ ns ] ) and ( [ ratio ] ) we obtain n_s & & 1 + , [ bybridns ] + r & & ( ) ^2e^(n_s-1)n , [ bybridr ] which mean that the scalar power spectrum is blued - tilted ( @xmath226 ) .
compared to the standard hybrid inflation , the presence of the field derivative coupling to gravity ( @xmath227 ) leads to @xmath35 close to 1 as well as the suppressed tensor - to - scalar ratio . in the limit that @xmath228 one has @xmath229 and @xmath230 , i.e. the harrison - zeldovich ( hz ) spectrum .
the hz spectrum is under the observational pressure @xcite , but this property is subject to change depending on the assumptions about the reionization scenario @xcite .
the future high - precision observations will provide a more concrete answer about this issue . in the regime @xmath228 and @xmath231 one has @xmath232 for @xmath233 . from the wmap normalization
it follows that 5.3 10 ^ -4 . if @xmath234 , @xmath235 , @xmath236 , for example , @xmath237 gev . in this case
( [ bybridns ] ) and ( [ bybridr ] ) give @xmath238 and @xmath239 , which is close to the hz spectrum .
in this paper we have studied observational constraints on a number of representative inflationary models with a field derivative coupling to the einstein tensor , i.e. @xmath18 . such a non - minimal derivative coupling has an asymptotic local shift symmetry for a slow - rolling scalar field satisfying the condition @xmath240 .
since the strong coupling scale of the theory is around the planck scale for @xmath24 , quantum corrections to the inflaton potential can be suppressed during inflation .
the non - minimal derivative coupling to gravity leads to a gravitationally enhanced friction for the scalar field .
this property allows us to accommodate steep potentials with theoretically natural model parameters for realizing inflation .
not only the quartic potential @xmath21 with @xmath8 but the potential @xmath241 $ ] with @xmath242 gives rise to cosmic acceleration consistent with the amplitude of the cmb temperature anisotropies .
moreover the exponential potential @xmath243 , which often appears after the compactification of extra dimensions in higher - dimensional theories , can lead to inflation even for @xmath192 in the presence of the field derivative coupling to gravity . for the potential @xmath244 of chaotic inflation the tensor - to - scalar ratio @xmath36 decreases for larger @xmath148 , whereas the scalar spectral index @xmath35 increases . in the limit that @xmath102 , @xmath35 and @xmath36 approach the values given in eqs .
( [ chao2d ] ) and ( [ chao2 ] ) . as we see in figs .
[ fig2 ] and [ fig3 ] , for both @xmath97 and @xmath158 , the asymptotic values of @xmath35 and @xmath36 are within the @xmath168 observational bound derived by the joint data analysis of wmap7 , bao , and hst . for the quartic potential with @xmath89 we found that the parameter @xmath174 is constrained to be @xmath245 at the 95 % confidence level .
for the exponential potential @xmath31 the asymptotic values of @xmath35 and @xmath36 in the regime @xmath191 are given by eqs .
( [ expod ] ) and ( [ expo ] ) , with @xmath34 constrained as eq .
( [ lamcon ] ) from the wmap normalization .
figure [ fig4 ] shows that steep exponential potentials with @xmath192 can be compatible with the current observational data .
in natural inflation with the potential @xmath241 $ ] the observables can be parametrized by the parameter @xmath246 in the regime @xmath191 . for larger @xmath202 ,
both @xmath35 and @xmath36 tend to increase toward the values given in eqs .
( [ nans ] ) and ( [ nar ] ) .
these asymptotic values correspond to those derived in eqs .
( [ chao2d ] ) and ( [ chao2 ] ) for @xmath97 with @xmath181 .
this property comes from the fact that for @xmath208 inflation occurs around the potential minimum at @xmath213 .
the observational bound on the parameter @xmath202 is found to be @xmath247 for @xmath89 ( 95 % cl ) . from the wmap normalization
the symmetry breaking scale @xmath29 is constrained to be @xmath248 , which can be smaller than @xmath23 for @xmath218 . in hybrid inflation with the potential @xmath249 ( where @xmath222 ) the field derivative coupling to gravity leads to the blue - tilted scalar power spectrum close to @xmath250 .
compared to standard hybrid inflation , the power spectrum approaches the hz one , i.e. @xmath250 and @xmath251 .
the hz spectrum is in tension with observations , but we have to caution that this property is affected by the assumption of the reionization scenario .
if future observations such as planck @xcite can constrain the tensor - to - scalar ratio at the level of @xmath252 , this will allow us to place tighter bounds on the inflationary models with the field derivative coupling to gravity .
we hope that we can discriminate between a host of inflationary models within the next few years .
this work is supported by the grant - in - aid for scientific research fund of the fund of the jsps no 30318802 and scientific research on innovative areas ( no .
21111006 ) .
the author thanks antonio de felice for warm hospitality during his stay in naresuan university .
the author is also grateful to cristiano germani for useful correspondence .
v. f. mukhanov and g. v. chibisov , jetp lett . *
33 * , 532 ( 1981 ) ; a. h. guth and s. y. pi , phys .
lett . * 49 * , 1110 ( 1982 ) ; s. w. hawking , phys .
b * 115 * , 295 ( 1982 ) ; a. a. starobinsky , phys .
b * 117 * , 175 ( 1982 ) ; j. m. bardeen , p. j. steinhardt and m. s. turner , phys .
d * 28 * , 679 ( 1983 ) .
d. s. salopek , j. r. bond and j. m. bardeen , phys .
d * 40 * , 1753 ( 1989 ) ; n. makino and m. sasaki , prog .
phys . * 86 * , 103 ( 1991 ) ; d. i. kaiser , phys .
d * 52 * , 4295 ( 1995 ) ; e. komatsu and t. futamase , phys .
d * 59 * , 064029 ( 1999 ) ; s. tsujikawa and b. gumjudpai , phys . rev .
d * 69 * , 123523 ( 2004 ) .
c. p. burgess , h. m. lee and m. trott , jhep * 0909 * , 103 ( 2009 ) ; j. l. f. barbon and j. r. espinosa , phys .
d * 79 * , 081302 ( 2009 ) ; m. p. hertzberg , jhep * 1011 * , 023 ( 2010 ) ; g. f. giudice and h. m. lee , phys .
b * 694 * , 294 ( 2011 ) ; r. n. lerner and j. mcdonald , arxiv:1112.0954 [ hep - ph ] .
t. kobayashi , m. yamaguchi and j. yokoyama , phys . rev .
lett . * 105 * , 231302 ( 2010 ) ; c. burrage , c. de rham , d. seery and a. j. tolley , jcap * 1101 * , 014 ( 2011 ) ; s. mizuno and k. koyama , phys . rev .
* d82 * , 103518 ( 2010 ) ; a. de felice and s. tsujikawa , phys .
lett . * 105 * , 111301 ( 2010 ) ; phys .
d * 84 * , 124029 ( 2011 ) ; p. creminelli _ et al .
, _ jcap * 1102 * , 006 ( 2011 ) .
a. naruko and m. sasaki , class .
grav . * 28 * , 072001 ( 2011 ) ; a. de felice and s. tsujikawa , jcap * 1104 * , 029 ( 2011 ) ; t. kobayashi , m. yamaguchi and j. yokoyama , phys . rev . * d83 * , 103524 ( 2011 ) ; a. de felice , s. tsujikawa , j. elliston and r. tavakol , jcap * 1108 * , 021 ( 2011 ) .
f. lucchin and s. matarrese , phys .
d * 32 * , 1316 ( 1985 ) ; j. j. halliwell , phys .
b * 185 * , 341 ( 1987 ) ; j. yokoyama and k. i. maeda , phys .
b * 207 * , 31 ( 1988 ) ; a. b. burd and j. d. barrow , nucl .
b * 308 * , 929 ( 1988 ) .
m. b. green , j. h. schwarz and e. witten , _ superstring theory _ , cambridge university press ( 1987 ) ; k. a. olive , phys .
190 , 308 ( 1990 ) ; e. bergshoeff , m. de roo , m. b. green , g. papadopoulos and p. k. townsend , nucl . phys .
b * 470 * , 113 ( 1996 ) ; p. kanti and k. a. olive , phys . rev .
d * 60 * , 043502 ( 1999 ) .
j. m. bardeen , phys .
d * 22 * , 1882 ( 1980 ) ; h. kodama and m. sasaki , prog . theor .
suppl . * 78 * , 1 ( 1984 ) ; v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .
rept . * 215 * , 203 ( 1992 ) ; b. a. bassett , s. tsujikawa and d. wands , rev .
* 78 * , 537 ( 2006 ) .
j. e. kim , h. p. nilles and m. peloso , jcap * 0501 * , 005 ( 2005 ) ; s. dimopoulos , s. kachru , j. mcgreevy and j. g. wacker , jcap * 0808 * , 003 ( 2008 ) ; l. mcallister , e. silverstein and a. westphal , phys . rev .
d * 82 * , 046003 ( 2010 ) . | a field kinetic coupling with the einstein tensor leads to a gravitationally enhanced friction during inflation , by which even steep potentials with theoretically natural model parameters can drive cosmic acceleration . in the presence of this non - minimal derivative coupling we place observational constraints on a number of representative inflationary models such as chaotic inflation , inflation with exponential potentials , natural inflation , and hybrid inflation .
we show that most of the models can be made compatible with the current observational data mainly due to the suppressed tensor - to - scalar ratio . |
in the last few years , tools from the field of quantum information theory have found extensive use in the study of the phase diagrams of quantum systems .
one such technique , the fidelity approach to quantum phase transitions ( qpts ) has been successfully applied to various systems possessing quantum critical points @xcite ( see @xcite for a review ) .
this technique can be generalized to finite - temperature systems @xcite , classical phase transitions @xcite , and topological phase transitions @xcite . in a recent letter
@xcite we have studied the fidelity in the context of disordered quantum systems . the physics peculiar to disordered quantum systems is reflected in the properties of the fidelity , a quantity not previously used to investigate such systems . here
we study the scaling behavior and provide details concerning the zero - temperature phase diagram of the disordered quantum xy model in a transverse field , a prototypical model in the context of disordered quantum systems .
the paper is organized as follows .
section [ sec : met_mod ] is devoted to defining the model , along with a review of known results about its phase diagram and the basics of the fidelity approach .
section [ sec : res ] presents the numerical results of our study and discusses the main features of the fidelity for disordered quantum chains .
our conclusions are presented in section [ sec : conclusion ] .
it is known that disorder can have interesting effects on a system s phase diagram @xcite . in particular , griffiths phases may arise as a result of the randomness @xcite . here
we study the disordered anisotropic quantum xy spin chain in a random transverse field , a model where the disorder - free case can be analytically solved @xcite and for which some exact results are known in the disordered case @xcite .
its hamiltonian is given by @xmath0 where @xmath1 are pauli matrices , and the fields @xmath2 and anisotropies @xmath3 are independent gaussian - distributed random variables .
the average field and anisotropy are denoted by @xmath4 and @xmath5 , respectively .
the variance is taken to be the same for both the field and anisotropy distributions .
the jordan - wigner transformation maps this system onto quasi - free spinless fermions @xcite . neglecting the boundary term and taking the system to be closed in the fermion index ,
we obtain a hamiltonian of the form @xmath6 where @xmath7 and @xmath8 are symmetric and antisymmetric real @xmath9 matrices , respectively . explicitly : @xmath10 , @xmath11 and @xmath12 , @xmath13
. the hamiltonian may be rewritten in terms of the matrix @xmath14 , which contains all information about the system . performing the polar decomposition of @xmath15 we obtain the matrices @xmath16 and @xmath17 such that @xmath18 , where @xmath16 is a positive semi - definite matrix and @xmath17 is unitary . from the eigenvalues of @xmath16 one obtains the single - particle energy spectrum @xcite . for systems at zero temperature , the fidelity is simply the absolute value of the overlap between ground states corresponding to nearby points in parameter space . near a quantum critical point
the ground state changes rapidly for small shifts in the tuning parameters , an effect which is reflected in a corresponding decrease of the fidelity
. the ground state fidelity can be cast in terms of the unitary matrix @xmath17 in the following way @xcite @xmath19 where @xmath17 and @xmath20 are respectively the unitary parts of the matrices @xmath21 and @xmath22 , evaluated at the model parameters @xmath23 and @xmath24 .
the corresponding fidelity susceptibility is defined as @xcite @xmath25 and can be written in terms of the unitary matrix @xmath17 as @xmath26 where @xmath27 is the frobenius norm . for a derivation of eq .
( [ eq : chifrob ] ) see the appendix .
we evaluate the fidelity susceptibility using ( [ eq : chifrob ] ) by performing a singular value decomposition of @xmath15 @xmath28 where @xmath29 and @xmath30 are unitary matrices .
note that the fidelity susceptibility is defined for infinitesimally separated points along any chosen direction in parameter space . before considering the effects of disorder in the xy chain ,
let us first recall the behavior of the fidelity susceptibility for the disorder - free case , where @xmath31 @xcite .
the system can then be found in one of three phases .
for @xmath32 it is paramagnetic , and for @xmath33 and @xmath34 ( @xmath35 ) the system is ferromagnetic along the @xmath36- direction ( @xmath37-direction ) .
the boundary between any two of these phases is a quantum - critical line corresponding to a second - order quantum phase transition . here
, we refer to the transition driven by the magnetic field as the ising transition , and to the transition driven by the anisotropy coupling as the anisotropy transition . at the quantum - critical points
there is an avoided level crossing between the ground state and the first excited state . as shown in figs .
[ fig : fig1](a ) and [ fig : fig1](b ) , one observes a maximum of the fidelity susceptibility at both the ising and anisotropy critical lines . and system size @xmath38 .
( b ) near the anisotropy transition , with @xmath39 and system size @xmath38 .
( c ) finite - size scaling dimension of the fidelity susceptibility near the ising transition , with @xmath40 .
the sizes considered range from @xmath41 to @xmath42 , with @xmath43 .
( d ) finite - size scaling of the fidelity susceptibility near the anisotropy transition , with @xmath44 .
the sizes considered range from @xmath41 to @xmath45 . ]
moreover , the finite - size scaling dimension of the fidelity susceptiblity in figs .
[ fig : fig1](c ) and [ fig : fig1](d ) show it to be extensive away from criticality and superextensive ( scaling quadratically with @xmath46 ) at the critical points .
this scaling behavior holds for both the ising and anisotropy critical lines .
the apparent subextensive scaling in the immediate vicinity of the anisotropy critical point is a numerical artifact due to the narrowing of the fidelity susceptibility peak as the system size grows .
the ising transition does not show this behavior , since the narrowing appears to occur more slowly than for the anisotropy transition .
it has been shown @xcite , for translationally invariant systems , that superextensive finite - size scaling of the fidelity susceptibility implies a vanishing gap and therefore quantum criticality . as a result , for clean systems the points in parameter space corresponding to superextensive scaling of this quantity mark quantum critical regions .
when randomness is introduced , translational invariance is lost , and hence superextensive scaling of the fidelity susceptibility does not necessarily imply quantum criticality . as discussed before , we find that locations of superextensive scaling reveal more general behavior beyond quantum criticality , namely griffiths phenomena @xcite .
the effects of disorder on the physics of quantum magnets has been studied mainly using the strong - disorder renormalization group technique ( sdrg ) @xcite .
a different approach has been used in the work of mckenzie and bunder @xcite , where the critical behavior of the disordered xy chain has been studied using a mapping to random - mass dirac equations .
the properties of the solutions of these equations imply the disappearance of the anisotropy transition in the presence of disorder .
furthermore , griffiths phases are predicted to appear both around the ising critical line and the anisotropy @xmath47 line .
these results , together with the analysis performed by fisher @xcite , are significant since they analytically show the drastic effects that disorder can have on the critical properties of a quantum system . at fixed @xmath5 the xy random chain
is closely related to the random transverse - field ising chain ( rtfic ) , which is another prototypical model for disordered quantum systems @xcite .
since the rtfic is representative of the universality class of ising transitions for all values of @xmath48 , let us review what is known for this model .
the hamiltonian of the rtfic is @xmath49 , $ ] where @xmath50 and @xmath51 are random couplings and fields respectively .
the system is critical when the average value of the field equals the average value of the coupling . using the sdrg one obtains that , at the quantum critical point , the time scale @xmath52 and the length scale @xmath46 are related by @xmath53 .
this results in an infinite value for the dynamical exponent @xmath54 at criticality @xcite .
the distribution of the logarithm of the energy gap @xmath55 at criticality broadens with increasing system size , in accordance with the scaling relation @xmath56 @xcite . in the vicinity of the critical point
the distribution of relaxation times is broad due to the presence of a griffiths phase , characterized by a non - universal dynamical exponent @xmath57 depending on the distance from the critical point .
this dependence can be used as an indicator for the griffiths phase . in @xcite a study of the phase diagram of a random xy spin chain in a random transverse field
was performed using the fidelity approach .
there it was shown that superextensive finite - size scaling of @xmath58 signals the presence of a quantum phase transition close to the ising critical line , while the minimum in @xmath58 close to the anisotropy @xmath59 line is consistent with the absence of a phase transition in that parameter region , see figs . [
fig : fig2](a ) and ( b ) . ,
@xmath60 , and @xmath61 , ( b ) average @xmath62 near the @xmath47 line , for @xmath38 , @xmath63 , and @xmath61 , ( c ) scaling dimension of @xmath62 for the same set of parameters near the ising line , considering @xmath64 , ( d ) scaling dimension of @xmath62 for the same set of parameters near the @xmath47 line , considering @xmath64 . ]
the griffiths phases of the model manifest themselves in a non - universal dependence of the finite - size scaling dimension @xmath65 of the fidelity susceptibility on the distance from the disorder - free critical point , see figs .
[ fig : fig2](c ) and ( d ) .
the relation between the dynamical scaling exponent @xmath57 and the scaling dimension of @xmath58 @xcite @xmath66 establishes the connection between the fidelity susceptibility and the griffiths phase . in eq.([eq : scaling_formula ] ) @xmath67 is the scaling dimension of the relevant operator driving the transition .
( [ eq : scaling_formula ] ) implies a non - universal scaling dimension for the fidelity susceptibility @xmath68 , provided that the behavior of the unknown scaling dimension @xmath67 does not exactly cancel that of the dynamical exponent . in the following
we would like to study , using other methods , the extent of the griffiths phase for this model .
in particular , we would like to verify that the entrance into a griffiths phase is indeed reflected by a changing scaling behavior of the fidelity susceptibility .
in our numerical analysis we consider system sizes @xmath69 and for each system size we compute @xmath70 disorder realizations .
we take the external fields ( anisotropies ) to be independent and identically distributed gaussian random variables with standard deviation @xmath71 and mean @xmath4 ( @xmath5 ) .
we consider the range of values @xmath72 for the standard deviation @xmath71 .
this disorder strength can be considered strong with respect to the value of the other parameters .
we denote with @xmath73_{\textrm{ave}}$ ] the arithmetic mean over all @xmath70 disorder realizations .
the width of the griffiths phase for the xy model with weak gaussian disorder in the continuum limit is known due to the work of mckenzie @xcite .
let us denote the distance from criticality with @xmath74 , where @xmath75 corresponds to the points in parameter space where the pure system is critical . from @xcite one can compute that near the ising transition , where the field @xmath4 drives the transition , @xmath76 , while near the anisotropy line @xmath77 .
for the so - called commensurate case , which corresponds to the ising transition for this system since we have disorder in both the field and anisotropy , mckenzie showed that the disorder - averaged density of states @xmath78_{\rm{ave } } / \rho_0 $ ] diverges at zero energy within the range @xmath79 away from the critical point @xcite . here
@xmath80 is the high - energy " density of states .
this divergence implies that the gap distribution function @xmath81 also diverges at zero energy , since a large density of states means a vanishingly small gap .
note that a diverging probability of having a vanishing gap does not necessarily imply that the density of states is also divergent , since the gap distribution only provides information about the position of the first excited energy level relative to the ground state energy .
however , since a diverging gap distribution should be expected to occur as a result of a divergence of the low - energy density of states , we will use it to give a rough estimate of the extent of the griffiths phase . for the incommensurate case , which includes the anisotropy transition , @xmath82_{\rm{ave } } / \rho_0 $ ]
is of the order of unity for some range of parameters about @xmath83 , implying effective gaplessness , and @xmath82_{\rm{ave } } / \rho_0 $ ] is much smaller than unity for @xmath84 , giving an effectively finite gap @xcite .
note that the boundary of the griffiths phase in this region is expected to be less defined than near the ising transition , since the zero - energy density of states does not diverge at any value of @xmath74 .
these results apply for the case of weak disorder , but we consider a range of moderate to strong disorder strengths . in order to compare with the results of mckenzie for the griffiths phase extent , we propose a rough criterion for determining the extent of the griffiths phase using the gap distribution .
assume that the griffiths phase lies within the range of parameter values for which the distribution @xmath81 has a maximum for @xmath85 .
at some point away from criticality the distribution maximum moves away from zero , eventually becoming approximately gaussian far from the disorder - free critical point .
we have determined the range of parameters for which @xmath81 has a maximum at zero gap .
the width of this range of parameters is independent of system size and scales with the variance of the coupling or field distributions , in accordance with the result of @xcite .
however , this criterion gives an extent several times larger than the mckenzie value , around both the ising and @xmath47 lines .
this difference may be due to strong disorder or , rather , the result of an overestimation of the griffiths phase extent since the distribution @xmath81 having a maximum at zero does not directly imply a divergent zero - energy density of states . nonetheless , consistency between different estimations of the griffiths phase extent support the validity , at least qualitatively , of this approach .
fig.([fig : fig3 ] ) shows the gap distribution in the vicinity of the @xmath47 line . for small anisotropies
the distribution has a pronounced peak at zero gap . moving away from the anisotropy line
, the distribution develops a peak slightly away from @xmath86 , and for larger anisotropies the distribution becomes gaussian with a vanishing probability of having zero gap .
similar behavior holds for the ising transition as the average field @xmath4 is adjusted away from the finite - size pseudocritical point . + , @xmath61 , and system size @xmath38 .
plotted distributions are for @xmath87 and @xmath88 . ]
in computing the _ average _ of some physical quantity , namely the arithmetic mean over many disorder realizations , any rare but large values will significantly affect the result .
on the other hand , the geometric mean over disorder realizations gives a more representative measure of the _ typical _ values of the physical quantity .
recall that the arithmetic mean provides an upper bound for the geometric mean when averaging over a set of positive values , and the two are equal only when taking the mean of a constant set of values . to observe the presence of large fluctuations in the fidelity susceptibility , we plot in fig .
[ fig : fig4](a ) the disorder - averaged as well as typical fidelity susceptibility in the vicinity of the ising critical point for a disorder strength @xmath61 .
notice that in the vicinity of the critical point the average becomes significantly larger than the typical value , indicating that there are instances of large fidelity susceptibilities that skew the arithmetic average towards a greater value .
near the anisotropy line , as shown in fig .
[ fig : fig4](b ) , there are also regions where the average fidelity susceptibility becomes much larger than the typical value , but now the positions of largest difference do not correspond to a critical point . indeed , at the point @xmath47 , which in the disorder - free case is critical , the average - typical difference is much smaller than it is at the two offset peaks .
this is evidence for the disappearance of the anisotropy transition as a result of the disorder . for translationally invariant systems , which are disorder - free
, it has been shown that superextensive finite - size scaling of the fidelity susceptibility implies quantum criticality @xcite .
the disordered xy chain does not have translational invariance , so locations of superextensive scaling do not necessarily imply criticality
. however , it is still useful to consider finite - size scaling , since comparison with the disorder - free case may suggest in what way the phase diagram changes as a result of disorder .
fig.([fig : fig5 ] ) shows the finite - size scaling dimension of the typical fidelity susceptibility near the ising transition .
the locations of the maxima of the typical fidelity susceptibility and finite - size scaling dimension coincide , and are shifted from the pure pseudocritical point due to finite - size effects . also , the maximum scaling dimension obtained with disorder is smaller than the pure case of quadratic scaling in @xmath46 , and this maximum value decreases with increased disorder strength .
+ , @xmath38 , and @xmath61 ( b ) the @xmath47 line , for @xmath63 , @xmath38 , and @xmath61 . ] , for @xmath89 .
scaling fit considered system sizes @xmath90 , with @xmath91_{typ}}}$ ] assumed . ] in fig.([fig : fig6 ] ) , the finite - size scaling dimension of @xmath62 around the anisotropy line @xmath47 is shown for the same set of disorder strengths .
notice that the scaling depends on distance from the @xmath47 line and at @xmath47 the scaling is approximately extensive , as it is when far from the anisotropy line .
for sufficiently small disorder and system size there may appear to be only a single peak in the typical fidelity susceptibility at @xmath47 , suggesting that for these finite systems emergent criticality is still felt even though the quantum phase transition in the thermodynamic limit disappears as a result of the disorder .
however , increasing the system size reveals a double peak with a peak offset which grows with the strength of the disorder . in both the ising and anisotropy regions , the width of the parameter interval giving scaling dimensions larger than a particular value scales approximately with the variance of the disorder distribution .
this scaling behavior agrees with that given by the previously mentioned gap distribution criterion for the griffiths phase . in fig.([fig :
fig7 ] ) the disorder - averaged gap is plotted for the four disorder strengths .
the behavior of the gap corresponds closely with that of the fidelity susceptibility , in that the average gap minima have the same location as the typical @xmath62 maxima .
note that the vicinity of the ising critical line and the @xmath47 line are regions of effective gaplessness which we associate with quantum criticality or griffiths phases . + , for @xmath89 . scaling fit considered system sizes @xmath92 , with @xmath91_{typ}}}$ ] assumed . ] , @xmath38 , and @xmath60 .
( b ) disorder - averaged gap near the @xmath47 line for the same range of disorder strengths , @xmath38 , and @xmath63 . ]
an experiment on such a disordered system would consider only one particular realization of disorder , and as a result would not necessarily observe the disorder - averaged value of an observable but rather a typical value .
here , we would like to study whether a measurement of the fidelity susceptibility for a large system coincides with the average value , and to do this we must see for what conditions @xmath62 is a _ self - averaging _
quantity @xcite .
consider the quantity @xmath93/[\chi(x)]_{\rm{ave}}^2 $ ] .
we expect that @xmath94 for fixed @xmath36 will scale as a power law in the system size @xmath46 , @xmath95 .
if @xmath96 then we say @xmath62 is self - averaging , if @xmath97 then @xmath62 is weakly self - averaging , and if @xmath98 then @xmath62 is not self - averaging . in fig.([fig : fig8 ] ) we indicate the regions for which @xmath62 is self - averaging , weakly self - averaging and non - self - averaging for various disorder strengths near the ising transition as well as the @xmath47 line .
is weakly self - averaging ( blue ) and non - self - averaging ( orange ) , as a function of disorder strength @xmath71 . outside these regions
@xmath62 is self - averaging . ]
far from the ising critical line , the distribution of the fidelity susceptibility is gaussian , see fig.([fig : fig9 ] ) .
however , in the vicinity of the griffiths phase and the critical point the distribution is non - gaussian , developing a slowly - decaying tail towards large fidelity susceptibilities , as shown in fig.([fig : fig10 ] ) .
this tail reflects the presence of rare but large fidelity susceptibilities , and is expected to arise either in a griffiths phase or in a quantum critical region .
+ far from the ising critical point , with @xmath99 ( blue ) .
overlaid is a gaussian distribution function with the same mean and variance ( red ) .
( b ) distribution of @xmath100 for the same parameters ( blue ) along with the corresponding distribution function of the logarithm of a gaussian random variable . here
@xmath101 , @xmath38 and @xmath60 .
this distribution is well described as gaussian . ]
overlaid is a gaussian distribution function with the same mean and variance ( red ) .
( b ) distribution of @xmath100 for the same parameters ( blue ) along with the corresponding distribution function of the logarithm of a gaussian random variable . here
@xmath101 , @xmath38 , and @xmath60 .
clearly this distribution is not gaussian , as can be seen by the much slower dropoff of the tail towards large fidelity susceptibilities . ]
now we explore how the distribution of the logarithm of the fidelity susceptibility changes as the system size is varied , with all other parameters fixed .
far from the ising transition the distribution of @xmath102 narrows with increasing system size , as shown in fig .
[ fig : fig11](a ) .
the position of the peak of the distribution remains fixed for a rescaling @xmath103 ( see fig .
[ fig : fig11](b ) ) , but the width of the distribution decreases slightly .
choosing a more general scaling assumption @xmath104 allows for an improved collapse of the distributions in this region , indicated in fig .
[ fig : fig11](c ) .
the fit parameter @xmath105 essentially translates the distribution , while the parameter @xmath106 adjusts the width .
such a scaling would imply a size dependence @xmath107 , where @xmath108 .
however , for asymptotically large system sizes this would lead to subextensive scaling , and would thus not be expected to hold for all @xmath46 . it appears that this apparent non - power - law scaling for the range of sizes we have considered may be due to finite - size effects , and we speculate that an assumption of extensive scaling would lead to an improved collapse for sufficiently large system sizes . at the ising pseudocritical point
the distribution @xmath109 broadens significantly with increasing system size , and a rescaling @xmath110 gives a good collapse for a value of the fit parameter @xmath111 ( see fig.([fig : fig12 ] ) ) .
this collapse and value of fit parameter holds for the pseudocritical points corresponding to all four disorder cases we have considered .
a rescaling of this kind suggests that the fidelity susceptibility scales as a stretched exponential of the system size at the critical point rather than quadratically as in the pure xy chain . for the random transverse field ising chain @xcite
, it is known that the energy gap vanishes as @xmath112 at the ising critical point .
recalling the alternative expression for the fidelity susceptibility @xmath113 @xcite , we expect that the first term in this series would dominate , and that the fidelity susceptibility might scale as @xmath114 .
however , this crude argument appears not to be consistent with the scaling of the energy gap of the rtfim , perhaps because of a lack of universality in the power of @xmath46 in the stretched exponential . at @xmath115 for @xmath116 .
( b ) distribution @xmath117 .
( c ) rescaled distribution , @xmath118 . here
@xmath119 , @xmath60 , and the fit parameters are @xmath120 and @xmath121 . including this exponential correction improves the collapse of the curves .
this scaling would suggest @xmath107 . ] at the pseudocritical point @xmath122 for @xmath116 .
( b ) distribution after a power - law rescaling @xmath123 , where @xmath124 .
( c ) distribution of @xmath125 for @xmath111 and the same parameters . here
@xmath119 and @xmath60 , though this collapse applies for all other disorder strengths considered . ]
just like for the ising case , far from the anisotropy line the distribution @xmath126 is well - approximated as a gaussian .
closer to the @xmath47 line the distribution looks much like the distribution of @xmath127 in the vicinity of the ising line , becoming non - gaussian with a slowly - decaying tail towards large values of @xmath62 .
fig.([fig : fig13 ] ) shows the distribution @xmath128 for @xmath47 and noise strength @xmath61 , and fig.([fig : fig14 ] ) shows the same quantity for @xmath129 , the value of average anisotropy coinciding with the peak in the typical value of @xmath62 for that magnitude of disorder .
considering the point @xmath47 , as the system size increases the distribution @xmath109 does not change width , so a rescaling @xmath130 gives a good collapse , see fig.([fig : fig15 ] ) .
this scaling also agrees with the extensive scaling of the average fidelity susceptibility at @xmath47 . moving @xmath5 away from this point in either direction ,
soon the distribution begins to shift superextensively , as shown in fig.([fig : fig16 ] ) .
for all values of @xmath5 in this peak region , a rescaling of the form @xmath131 gives a good collapse , where @xmath105 is the fit value of the finite - size scaling dimension of the corresponding typical fidelity susceptibility . continuing to move @xmath5 away from the peak in the typical fidelity susceptibility ,
the distribution begins to narrow slightly as in the off - critical ising case .
however , a rescaling @xmath130 appears to give a good collapse for @xmath5 sufficiently large in magnitude .
+ at the anisotropy line @xmath47 ( blue ) .
overlaid is a gaussian distribution function with the same mean and variance ( red ) .
( b ) distribution of @xmath100 for the same parameters ( blue ) along with the corresponding distribution function of the logarithm of a gaussian random variable . here
@xmath61 , @xmath38 , and @xmath63 .
the distribution is non - gaussian with a power - law tail towards large fidelity susceptibilities . ]
( near the value of @xmath5 coinciding with the peak in @xmath132_{typ}$ ] ) ( blue ) . overlaid is a gaussian distribution function with the same mean and variance ( red ) .
( b ) distribution of @xmath100 for the same parameters ( blue ) along with the corresponding distribution function of the logarithm of a gaussian random variable . here
@xmath61 , @xmath38 , and @xmath63 .
note that there does not appear to be a power - law tail towards large fidelity susceptibilities . ] at @xmath47 for @xmath116 .
( b ) rescaled distribution , @xmath133 for the same parameters . here
@xmath61 and @xmath63 , though this collapse applies for all other disorder strengths considered . ] at @xmath134 , the position of the maximum of @xmath132_{typ}$ ] for @xmath61 .
system sizes @xmath116 are plotted .
( b ) rescaled distribution , @xmath123 for the same parameters , with @xmath135 . here
@xmath61 and @xmath63 . ]
in this work we have studied the effect of random transverse fields and couplings on the phase diagram of the quantum xy chain . by examining the finite - size scaling of the typical fidelity susceptibility and the fidelity susceptibility distribution for a range of disorder strengths and system sizes , we find agreement with earlier analytic results pertaining to the limit of weak randomness .
the introduction of disorder clearly removes the anisotropy quantum critical line , replacing it with an extended griffiths phase .
there , the typical fidelity susceptibility s finite - size scaling dimension depends strongly on the average value of the anisotropy parameter , and appears to become extensive at the line of vanishing anisotropy . at the ising critical line ,
the stretched exponential scaling of the fidelity susceptibility distribution is consistent with what is expected of an infinite randomness fixed point , while a griffiths phase is observed to form in the vicinity .
remarkably , the scaling of the fidelity susceptibility distribution at the ising critical line is universal , in that all disorder strengths give the same scaling behavior . in the griffiths
phase the fidelity susceptibility is not self - averaging . however , self - averaging behavior returns sufficiently far from the disorder - free critical lines .
these detailed results suggest that the fidelity susceptibility may be a useful tool for the study of other disordered systems .
we would like to thank n. bray - ali for helpful comments .
computation for the work described in this paper was supported by the university of southern california center for high performance computing and communications .
we acknowledge financial support by the national science foundation under grant dmr-0804914 .
here we derive an expression for the fidelity susceptibility @xmath62 in terms of the unitary matrix @xmath17 : @xmath136\right\},\end{aligned}\ ] ] where @xmath137 .
this leads to @xmath138\right\}}\nonumber \\ &
= & \exp\left\{-\frac{1}{8}\vert\partial_{x}t\vert_{f}^{2}dx^2-\frac{1}{16}\textrm{tr } \left [ t^{\dagger}\partial_{x}tt^{\dagger}\partial_{x}t \right]dx^2\right\}\nonumber \\ & = & \exp\left\{-\frac{1}{8}\vert\partial_{x}t\vert_{f}^{2}dx^2-\frac{1}{16}\textrm{tr } \left [ -\partial_{x}t^{\dagger}tt^{\dagger}\partial_{x}t \right]dx^2\right\}\nonumber \\ & = & \exp\left\{-\frac{1}{8}\vert\partial_{x}t\vert_{f}^{2}dx^2-\frac{1}{16}\textrm{tr } \left [ -\partial_{x}t^{\dagger}\partial_{x}t \right]dx^2\right\}\nonumber \\ & = & \exp\left\{-\frac{1}{16}\vert\partial_{x}t \vert_{f}^{2}dx^{2}\right\},\end{aligned}\ ] ] where we have used the anti - symmetry of @xmath139 which implies @xmath140=-\vert\partial_{x}t \vert_{f}^2 $ ] , with @xmath27 the frobenius norm . from ( [ eq : chi ] ) it follows that @xmath141 99 p. zanardi and n. paunkovi , phys . rev .
e * 74 * , 031123 ( 2006 ) h.q .
zhou and j.p .
barjaktarevic , arxiv : cond - mat/0701608 h.q .
zhou , j.h .
zhao and b. li , arxiv:0704.2940 h.q .
zhou , arxiv:0704.2945 s.j .
gu , arxiv:0811.3127v1 p. zanardi , h.t .
quan , x. wang , and c.p .
sun , phys .
a * 75 * , 032109 ( 2007 ) p. zanardi , l. campos venuti , p. giorda , phys .
a * 76 * , 062318 ( 2007 ) h. t. quan , f. m. cucchietti , phys .
e * 79 * , 031101 ( 2009 ) a. hamma , w. zhang , s. haas and d.a .
lidar , phys .
b * 77 * , 155111 ( 2008 ) d.f .
abasto , p. zanardi , phys .
a * 79 * , 012321 ( 2009 ) d.f .
abasto , a. hamma , p. zanardi , phys . rev
. a * 78 * , 010301(r ) ( 2008 ) s. yang , s.j .
sun and h.q .
lin , phys .
a , * 78 * , 012304 ( 2008 ) s. garnerone , d.f .
abasto , s. haas , and p. zanardi , phys .
a * 79 * , 032302 ( 2009 ) erik eriksson and henrik johannesson , arxiv:0902.3848 s. garnerone , n.t .
jacobson , s. haas , and p. zanardi , phys .
lett . , * 102 * , 057205 ( 2009 ) .
f. igloi and c. monthus , phys . rep . *
412 * 277 ( 2005 ) r.b .
griffiths , phys .
lett * 23 * , 17 ( 1969 ) e. lieb , t. schultz and d. mattis , ann.phys . * 16 * , 407 ( 1961 ) | the phase diagram of a quantum xy spin chain with gaussian - distributed random anisotropies and transverse fields is investigated , with focus on the fidelity susceptibility , a recently introduced quantum information theoretical measure .
monitoring the finite - size scaling of the probability distribution of this quantity as well as its average and typical values , we detect a disorder - induced disappearance of criticality and the emergence of griffiths phases in this model .
it is found that the fidelity susceptibility is not self - averaging near the disorder - free quantum critical lines . at the ising critical point the fidelity susceptibility scales as a disorder - strength independent stretched exponential of the system size , in contrast with the quadratic scaling at the corresponding point in the disorder - free xy chain . along the line where the average anisotropy vanishes the fidelity susceptibility appears to scale extensively , whereas in the disorder - free case
this point is quantum critical with quadratic finite - size scaling . |
recently the bes collaboration @xcite reported a near - threshold enhancement in the proton - antiproton ( @xmath0 ) invariant mass spectrum , observed in the @xmath1 decay .
signs for a low mass @xmath0 enhancement had been already seen earlier by the belle collaboration in their study of the @xmath7 decay @xcite as well as in the reaction @xmath8 @xcite .
but because of the large statistical uncertainties of the belle data it was difficult to draw concrete , quantitative conclusions about the extent of the near threshold @xmath0 enhancement . the new data by bai et al .
@xcite , however , are of rather high statistical accuracy and therefore provide very precise information about the magnitude and the energy dependence of the @xmath0 mass spectrum very close to threshold . the bes collaboration @xcite fitted their @xmath0 invariant mass spectrum below 1.95 gev by a breit - wigner resonance function .
assuming that the @xmath0 system is in an @xmath3-wave resulted in a resonance mass of @xmath9 mev and a total width of @xmath1030 mev .
a comparable fit to the data could be achieved with a @xmath11-wave breit - wigner function with @xmath12 mev and @xmath13 mev .
the proximity of these resonance masses to the @xmath6 reaction threshold ( which is at 1876.54 mev ) nourished speculations that the observed strong enhancement could be a signal of an @xmath2 bound state .
while theoretical considerations of such @xmath2 bound states ( or of baryonia , in general ) abound in the literature @xcite there is so far hardly any undisputed experimental information on the existence of such states @xcite .
thus , the supposition that one has found here independent and possibly even more convincing evidence in support of @xmath2 bound states is certainly appealing . an alternative explanation put forward by the bes collaboration invokes similarities of the observed enhancement in the @xmath6 mass spectrum near threshold with the strong energy dependence of the electromagnetic form factor of the proton around @xmath14 , in the timelike region , as determined in the reaction @xmath15 @xcite . in the latter case
it was argued that the sharp structure seen in the experiment could be caused by a narrow , near - threshold vector - meson ( @xmath16=@xmath17 ) resonance @xcite with @xmath18 mev and @xmath19 mev .
( see also refs .
@xcite for a pertinent discussion . )
one should keep in mind , however , that the quantum numbers of the @xmath20 particle ( @xmath16=@xmath17 ) would restrict such a resonance to occur in a pseudoscalar ( @xmath21 ) or scalar ( @xmath22 ) state should it be indeed responsible for the enhancement of the near - threshold @xmath6 mass spectrum in the decay @xmath1 .
an entirely different and much more conventional interpretation of the observed enhancement was suggested in several recent works @xcite .
these authors argue that the enhancement is primarily due to the final state interaction ( fsi ) between the produced proton and antiproton .
specifically , it was shown within the scattering length approximation @xcite that a calculation with a complex @xmath3-wave scattering length extracted from an effective - range analysis of @xmath0 scattering data can reproduce the shape of the @xmath0 mass distribution close to the threshold . in the present paper we analyse the near threshold enhancement in the @xmath0 invariant mass spectrum reported by the bes collaboration utilizing a realistic model of the @xmath2 interaction @xcite .
the elastic part of this model is the @xmath23-parity transform of the bonn meson - exchange @xmath24 potential , supplemented by a phenomenological complex potential to account for @xmath2 annihilation . as just mentioned , the investigations of kerbikov et al .
@xcite as well as those of bugg @xcite rely on the scattering length approximation .
but it remains unclear over which energy range this approximation can provide a reliable representation of the energy dependence of the @xmath6 amplitude .
the enhancement seen in the bes data extends up to invariant masses of @xmath252 gev , which corresponds to center - of - mass energies of around 120 mev in the @xmath6 system .
it is obvious that the simple scattering length approximation can not be valid over such a large energy region . using the @xmath6 amplitude of our interaction model ( which describes the available @xmath2 scattering data up to center - of - mass energies of 150 mev @xcite )
we can examine to what extent the scattering length approximation can indeed reproduce the energy dependence of the @xmath0 scattering amplitude .
moreover , we can compare the energy dependence induced by the full @xmath6 amplitude with the bes data over the whole energy range where the enhancement was observed .
a microscopic @xmath2 model has a further advantage .
it yields predictions for all possible spin- and isospin channels .
thus , we can investigate whether the energy dependence of the @xmath26 and @xmath27 @xmath2 scattering amplitudes in the @xmath28 and @xmath4 isospin channels is compatible with the bes data .
( we use the nomenclature @xmath29 but omit the isospin index when referring to both isospin channels or to isospin averaged results . ) the analysis of kerbikov et al .
utilizes scattering lengths obtained from a spin - averaged effective - range fit to the low - energy @xmath2 data .
those values might be strongly influenced or even dominated by the @xmath30 @xmath2 partial wave , a channel which can not contribute to the fsi in the @xmath1 decay because of charge - conjugation invariance .
let us finally mention that the work of zou and chiang @xcite does not employ the scattering length approximation but uses the @xmath31-matrix formalism .
however , their interaction consists only of one - pion exchange and is therefore not realistic because it does not take into account the most striking feature of low - energy @xmath2 scattering , namely annihilation .
recall that near the @xmath6 threshold the annihilation cross section is twice as large as the elastic @xmath6 cross section @xcite .
the paper is structured in the following way : in sect .
ii we briefly review the @xmath2 potential model that is used in the present analysis and examine the reliability of the scattering length approximation on the @xmath2 amplitude for the partial waves relevant to the analysis of the bes data .
the possible influence of the coulomb interaction is discussed as well . in sect .
iii we provide details of our calculation of the near - threshold @xmath6 mass spectrum for the reaction and we compare our results with the measurement of the bes collaboration .
iv is devoted to a discussion of possible signals of @xmath2 bound states or ( sub @xmath32 threshold ) meson resonances in the @xmath6 mass spectrum .
the paper ends with concluding remarks .
in the present investigation we use one of the @xmath2 models developed by the jlich group @xcite .
specifically , we use the bonn obe ( one - boson - exchange ) model introduced in ref .
@xcite whose results are also discussed in ref .
@xcite . in the latter
work the model is called a(obe ) and we also adopt this name in the present paper .
the model a(box ) is based on the obe version of the full bonn potential derived in time - ordered perturbation theory , i.e. obept in ref .
the g - parity transform of this interaction model constitutes the elastic part of the @xmath2 potential .
a phenomenological , isospin- and energy - independent complex potential of gaussian form is added to account for the @xmath2 annihilation . with only three free parameters ( the range and the strength of the real and imaginary parts of the annihilation potential ) a good overall description of the low- and intermediate energy @xmath2 data was achieved .
results for the total and the integrated elastic and charge - exchange cross sections as well as angular dependent observables can be found in refs .
@xcite . here in fig .
[ pbar7 ] we show only the elastic @xmath6 cross section .
we also utilized the most complete @xmath2 model of the jlich group , model d published in ref .
the elastic part of this interaction model is derived from the g - parity transform of the full bonn @xmath24 potential and the annihilation is described in part microscopically by @xmath33 2 meson decay channels - see ref .
@xcite for details .
but since the results turned out to be qualitatively rather similar to the ones obtained with a(obe ) we refrain from showing them here .
the differential cross sections for @xmath0 scattering @xcite already indicate a substantial contribution from higher partial waves at around 10 mev above the @xmath0 threshold .
this feature is also reflected in the predictions of the @xmath2 model . the dashed line in fig .
[ pbar7 ] shows the contribution from the @xmath26 partial wave .
the dotted and dash - dotted lines illustrate the total contributions from @xmath34 and @xmath35 waves , respectively . since the scattering length approximation to the @xmath2 amplitude
was used in two of the analyses of the bes data @xcite we want to investigate the validity of this approximation . in the following
we ignore the proton - neutron mass difference and also the coulomb interaction in the @xmath6 system , in order to simplify the discussion .
but we will come back to these issues later . in the scattering length approximation the @xmath3-wave @xmath0 scattering amplitude @xmath36
is given by @xmath37 where the scattering length @xmath38 is a complex number because of inelastic channels ( annihilation into multimeson states ) that are open already at the @xmath0 threshold .
the proton momentum in the c.m .
system is @xmath39 , where @xmath40 is the proton mass .
results for @xmath41 for the @xmath26 partial wave are presented in the upper part of fig .
[ pbar6 ] .
the solid lines are the result for the full amplitude while the dashed lines are based on the scattering length approximation given by eq.([scat1 ] ) .
note that the scattering lengths predicted by the @xmath2 model , which we use for the @xmath26 partial wave are @xmath42 fm and @xmath43 fm for the isospin @xmath28 and @xmath4 channels , respectively .
it is evident that the scattering length approximation does not reproduce the energy dependence of the scattering amplitude that well . for the @xmath4 channel the difference at an excess energy of 50 mev amounts as much as 50 % .
the difference is even more pronounced for the @xmath28 channel , where we already observe large deviations from the full result at rather low energies .
this strong failure of the scattering length approximation is due to the much smaller scattering length predicted by our model for the @xmath28 partial wave .
results for the @xmath30 partial wave are shown in the lower part of fig .
[ pbar6 ] . here
the scattering lengths predicted by the @xmath2 model are @xmath44 fm and @xmath45 fm for the @xmath28 and @xmath4 channels , respectively .
as pointed out above , this partial wave can not contribute to the reaction @xmath1 .
but it is still interesting to see how well the scattering length approximation works in this case .
obviously we also see similar shortcomings here .
thus , it is clear from fig .
[ pbar6 ] that the use of the scattering length approximation allows only a very rough qualitative estimate of the effects of fsi but it is definitely not reliable for a more quantitative analysis of the bes data . nevertheless , one has to realize that the main uncertainty in estimating @xmath6 fsi effects does not come from the scattering length approximation but from our poor knowledge of the @xmath6 @xmath26 amplitudes near threshold and of the @xmath1 reaction mechanism .
for example , the scattering lengths employed by kerbikov et al .
are spin - averaged values .
since the @xmath30 partial wave contributes with a weighting factor 3 to the @xmath6 cross sections ( and there are no spin - dependent observables at low energies that would allow one to disentangle the spin - dependence ) it is obvious that their value should correspond predominantly to the @xmath30 amplitude .
thus , it is questionable whether it should be used for analyzing the bes data at all because the contribution of the @xmath30 partial wave to the decay @xmath1 is forbidden by charge - conjugation invariance .
but even the availability of genuine @xmath26 amplitudes ( which are necessarily model dependent ) does not solve the problem .
the reaction @xmath1 can have any isospin combination in the final @xmath6 state .
a dominant reaction mechanism involving an intermediate isoscalar meson resonance ( @xmath46 , @xmath47 , ... ) , cf .
[ diag]c , would yield a pure @xmath28 final @xmath6 state , whereas an intermediate isovector meson resonance ( @xmath48 , @xmath49 , ... ) leads to pure @xmath4 . on the other hand , mechanisms like the ones depicted in fig .
[ diag]a , b generate the usual equal weighting of the two isospin amplitudes . in this context
let us mention that the spin - averaged @xmath6 scattering length predicted by the model a(box ) , @xmath50 fm , is in rough agreement with the value extracted from the level shifts of antiproton - proton atoms , cf .
table 8 in ref .
@xcite . on the other hand , this averaged value differs significantly from the scattering length in the @xmath51 partial wave which is one of the possible final states in the decay @xmath1 . at first glance
one would tend to believe that the unusually small real part of @xmath52 is basically accidental and reflects the fact that the @xmath26 state enters with smaller statistical weight into the @xmath6 cross section and , therefore , is much less constrained by the @xmath6 data .
however , exploratory calculations with the other @xmath2 models of the jlich group revealed that the real part of the @xmath51 scattering length is always small ( and often of a different sign from the other @xmath3 waves ) . for example , the more involved @xmath2 model d published in ref .
@xcite , yields @xmath53 fm .
moreover , the @xmath2 models of dover - richard also and kohno - weise predict rather small values for the real part of the @xmath51 scattering length , as is evident from the results presented in ref .
the paris @xmath2 potential , on the other hand , predicts the real parts of the @xmath54 scattering lengths to be of similar magnitude - cf .
table iv in ref .
@xcite . but
11 of that paper suggests that the central potential in the @xmath55 , @xmath56 channel has been drastically modified inside 1 fm as compared to what follows from the @xmath23-parity transformation of the paris @xmath24 potential . in the jlich @xmath57 models no such changes are introduced when deriving the elastic part of the @xmath57 interaction via @xmath23-parity !
a closer examination of our @xmath2 models disclosed that here the smallness of @xmath58 is definitely caused by the one - pion exchange contribution , i.e. by long - range physics .
switching off this contribution always led to a large scattering length similar to the one in the @xmath4 channel , whereas modifications of the short - range contributions to the elastic part of the @xmath2 potential had hardly any qualitative influence on the scattering lengths .
thus , it would be interesting to obtain experimental constraints on the @xmath54 scattering lengths .
corresponding measurements could be performed at the future gsi facility flair @xcite , where it is possible to have a polarized antiproton beam @xcite , as required for inferring the spin - triplet and spin - singlet amplitudes from the data .
the @xmath1 decay rate is given as @xcite @xmath59 where the function @xmath60 is defined by @xmath61 @xmath62 is the invariant mass of the @xmath0 system , @xmath63 is the proton angle in that system , while @xmath64 is the photon angle in the @xmath20 rest frame .
after averaging over the spin states and integrating over the angles , the differential decay rate is @xmath65 where @xmath66 is the total @xmath1 reaction amplitude . note that @xmath66 is dimensionless .
the differential rate @xmath67 , integrated over the range from @xmath68 to @xmath69 , yields the partial @xmath1 decay width , @xmath70 kev @xcite . the solid line in fig .
[ pbar5 ] is obtained with eq .
( [ trans ] ) by using a constant @xmath71 .
this result corresponds to the so - called phase space distribution .
the actual value of @xmath71 was adjusted to the data at @xmath72 gev , i.e. around the @xmath47 resonance region .
the circles in fig .
[ pbar5 ] show data for the @xmath1 decay published by the bes collaboration @xcite .
evidently , close to the @xmath0 threshold the data deviate substantially from the phase space distribution .
since close to the threshold the phase space factor introduces a strong but trivial energy dependence of the mass distribution it is convenient to divide the experimental @xmath0 spectrum by the kinematical factors that appear on the right side of eq.([trans ] ) and extract the invariant amplitude @xmath71 from the data .
corresponding results for the bes data @xcite are shown by the squares in fig .
[ pbar4 ] .
it is obvious that the experimental reaction amplitude exhibits a very strong mass dependence near the @xmath0 threshold . in recent investigations @xcite
this feature has been attributed to a strong final state interaction ( fsi ) between the outgoing protons and antiprotons .
indeed , an enhancement and/or a strong energy dependence of near - threshold cross sections has been seen in many other reactions involving hadrons and is commonly seen as to be caused by fsi effects , cf . @xcite .
thus , it is plausible and it even has to be expected that fsi effects play a role in the decay @xmath1 as well . the interesting issue is primarily whether the proton - antiproton interaction is really strong enough to induce such a sizeable enhancement in the invariant mass spectrum .
a very simple and therefore also very popular treatment of fsi effects is due to watson @xcite and migdal @xcite .
these authors suggested that the reaction amplitude for a production and/or decay reaction which is of short - ranged nature can be factorized in terms of an elementary production amplitude @xmath73 and the @xmath6 scattering amplitude @xmath36 of the particles in the final state , @xmath74 where @xmath73 is given by diagrams like those shown in fig .
[ diag ] and @xmath75 is a normalization factor .
if the production mechanism is of short range then the amplitude @xmath73 depends only very weakly on the energy and the near - threshold energy dependence of the reaction amplitude is driven primarily by the scattering amplitude , @xmath36 , of the outgoing particles .
this means that in the case of the reaction @xmath1 the near threshold mass dependence of the @xmath0 spectrum should be dominated by the energy dependence of the @xmath0 scattering amplitude . in the analyses of the bes data by kerbikov
@xcite and by bugg @xcite the above treatment of fsi effects was adopted . however , the prescription of watson
migdal is only valid for interactions that yield a rather large scattering length , like the @xmath26 @xmath24 partial wave where the scattering length @xmath38 is in the order of 20 fm , and even then only for a relatively small energy range , as was pointed out in several recent papers @xcite .
therefore , in the present case , where the scattering lengths are in the order of 1 fm one should be cautious with the interpretation of results obtained from applying eq.([fsi ] ) . rather one should start from the more general expression for the reaction amplitude , @xmath76 which corresponds to a distorted wave born approximation .
if one assumes that the production amplitude , @xmath77 , has only a very weak energy- and momentum dependence it can be factorized and one obtains @xmath78 = a_0 \psi^{(-)*}_{q_p}(0 ) , \label{fsidw}\end{aligned}\ ] ] where @xmath79 is a suitably normalized @xmath2 continuum wave function and @xmath80 is nothing else but the inverse of the jost function @xmath81 , i.e. @xmath82 @xcite .
eq.([fsidwba ] ) itself can be cast into the form @xcite @xmath83 , \label{fsidw2}\end{aligned}\ ] ] where @xmath84 is , in general , a complex number that represents the principal - value integral over the half - off - shell extension of the production amplitude @xmath73 and the @xmath6 scattering amplitude that appears on the very right hand side of eq.([fsidwba ] ) .
obviously , eq.([fsidw2 ] ) is formally equivalent to the watson - migdal prescription of eq.([fsi ] ) if @xmath85 is large . in case of the @xmath24 @xmath26 partial wave
@xmath85 is indeed large , as has been demonstrated in ref .
@xcite . in considering the @xmath3-wave interaction in the final @xmath0 system
one should account for the centrifugal barrier between the photon and the @xmath0 state given by the factor @xmath86^{l_\gamma},\end{aligned}\ ] ] with @xmath87=1 being the orbital momentum between the photon and the @xmath0 system . around the threshold , @xmath88
, the centrifugal correction depends only weakly on the invariant mass of the @xmath0 system and basically does not modify the mass dependence of the fsi . within the range @xmath89 gev the factor @xmath90 varies between 0.23 and 0.27 , which is not large enough to counterbalance the contribution from the strong @xmath3-wave fsi over the same invariant - mass range . note also that the watson - migdal formula ( [ fsi ] ) has to be modified when used for @xmath0 fsi effects in a @xmath11-wave . since in the production reaction
only the final momentum is on - shell one has to divide the on - shell @xmath36 matrix by the factor @xmath91 in order to impose the correct threshold behavior of the production amplitude .
let us first discuss calculations based on the watson - migdal approach given by eq .
( [ fsi ] ) .
the solid lines in fig . [ pbar4 ] show the @xmath0 invariant scattering amplitudes squared for the @xmath26 and @xmath27 partial waves and the @xmath28 and @xmath4 channels .
we consider the isospin channels separately because , as mentioned above , the actual isospin mixture in the final @xmath0 system depends on the reaction mechanism and is not known .
note that all squared @xmath6 scattering amplitudes @xmath41 were normalized to the bes data at the invariant mass @xmath92=50 mev by multiplying them with a suitable constant .
the results indicate that the mass dependence of the bes data can indeed be described with fsi effects induced by the @xmath26 scattering amplitude in the @xmath4 isospin channel .
the @xmath28 channel leads to a stronger energy dependence which is not in agreement with the bes data
. we can also exclude dominant fsi effects from the @xmath27 partial waves . here the different threshold behaviour due to the @xmath11-wave nature can not be brought in line with the data points very close to threshold .
it should be clear , of course , that a suitable combination of several partial waves might as well reproduce the experimental results on the @xmath0 invariant mass spectrum .
note that we do not include the coulomb interaction in our model calculation and we also ignore the difference in the @xmath6 and @xmath93 thresholds .
judging from the results shown by kerbikov et al .
@xcite their influence is noticeable only for excess energies below say 5 mev .
accordingly we do not consider the lowest data point of the bes experiment , which lies in the first ( 5 mev ) energy bin , in our discussion .
our results support the conjecture of kerbikov @xcite and bugg @xcite that the enhancement seen in the near - threshold @xmath6 invariant mass spectrum of the decay @xmath1 could be primarily due to fsi effects in the @xmath6 channel .
but we should also consider the @xmath6 invariant mass spectrum of the decay @xmath94 which was presented by the bes collaboration in the same paper @xcite . here
the near - threshold mass distribution does not show any enhancement as compared to the phase - space distribution though one would likewise expect strong fsi effects in the @xmath6 channel . in this reaction isospin
is preserved and the possible partial waves in the final @xmath6 state , @xmath95 and @xmath96 , differ from those available in the decay @xmath1 .
but the @xmath95 partial wave amplitude of the jlich @xmath2 model , shown in the lower part of fig . [ pbar6 ] , leads to fsi effects that are comparable to those of the @xmath97 state and in contradiction with the experimental @xmath98 mass distribution - as illustrated in fig .
we should mention in this context that the issue of the @xmath98 mass distribution is not discussed in the work of bugg , while kerbikov argues that in the effective range parameterization that he employs @xmath99 is much smaller than @xmath100 and therefore there should be much smaller fsi effects in the @xmath4 channel .
however , our experience is that @xmath6 amplitudes with a small @xmath101 often yield an even stronger energy dependence and therefore larger fsi effects - see the case of the @xmath51 partial wave in fig .
[ pbar6 ] . in our opinion these seemingly contradictory results can only be reconciled if we assume that the treatment of fsi effects by means of eq.([fsi ] ) is oversimplified and one should use eq.([fsidw2 ] ) instead .
results based on the latter equation are shown by the solid lines in fig .
[ be8 ] where we set @xmath102 .
it might be possible that a fine tuning of @xmath84 could allow one to better reproduce the bes data for the @xmath103 channels , though in view of the rather large statistical variations in the data we refrain from doing so . in any case it is obvious that the more refined treatment of the fsi effects based on eq .
( [ fsidw2 ] ) leads , in general , to a weaker energy dependence of the @xmath6 mass spectrum .
it is important to note that the near - threshold contribution to the @xmath1 decay rate is relatively large .
the contribution to @xmath1 up to invariant @xmath6 masses of @xmath104 2 gev is roughly 5 times larger than the rate for the reaction @xmath105 followed by the @xmath106 decay .
taking into account that the latter rate was recently published @xcite as br=@xmath107 we can estimate that @xmath108 where @xmath109 indicates the near - threshold contribution of this partial wave which is indeed a large fraction of the total @xmath1 branching rate that amounts to @xmath110 .
this suggests that in this energy region the reaction @xmath1 could be indeed dominated by a resonance .
such a resonance should lie below the @xmath6 threshold but , unlike the resonance state which emerged from the breit - wigner fit of the bes collaboration @xcite , its mass could be significantly below the @xmath0 threshold and it could have a large width , like the @xmath49 or @xmath111 resonances that are listed in the particle physics booklet @xcite .
we demonstrate the scenario in fig .
[ pbar3 ] . for illustration
we consider the @xmath48(1800 ) meson with mass and width 1801 and 210 mev , respectively .
the corresponding resonance amplitude squared is shown by the dashed line in fig .
[ pbar3 ] .
its energy dependence is , of course , too weak and does not agree with the near - threshold @xmath6 mass spectrum found by the bes collaboration .
however , when the @xmath48(1800 ) resonance decays into a proton and antiproton , a final state interaction should occur .
the solid line in fig .
[ pbar3 ] shows the total reaction amplitude squared , taken now as the product of the resonance amplitude and the @xmath0 scattering amplitude in the @xmath97 partial wave as is given by eq.([fsi ] ) .
the corresponding result is roughly in line with the bes data .
though this line of reasoning is certainly more a plausibility argument , rather than a solid calculation , we believe that it makes clear that one has to be rather cautious when trying to extract resonance properties by fitting the near threshold @xmath0 mass spectrum with a resonance amplitude , because any final state interaction will necessarily and substantially distort the production amplitude .
a different and alternative scenario consists in assuming the formation of many multimesonic intermediate states in the @xmath20 decay that couple strongly to the @xmath0 system where again a strong fsi occurs . the branching rates of the @xmath20 radiative decays into the @xmath112 , @xmath113 , @xmath114 and @xmath115 channels are larger than @xmath116 and the transition of those mesonic states into the @xmath0 final state could be sufficiently strong @xcite .
thus , such a scenario can not be excluded by the available data .
finally , let us come back to the other proposed explanation for the enhancement in the @xmath0 mass spectrum , namely near - threshold @xmath2 bound states .
clearly a description of the experimental mass spectrum in terms of @xmath0 fsi effects does not contradict the existence of such states . indeed , in case of the @xmath24 interaction the strong fsi effects are interconnected with the existence of a near - threshold bound state ( deuteron ) or anti bound - state in the corresponding @xmath30 and @xmath26 partial waves .
however , the @xmath2 model that we used in the present study does not lead to any near - threshold bound states .
in fact , we found only two bound states for the model a(obe ) within 100 mev from the threshold , namely at @xmath117 = 104@xmath118413 mev in the @xmath51 partial wave and at @xmath117 = 24.2@xmath118107 mev in the @xmath119 partial wave . obviously both bound states lie rather far away from the real axis and therefore should have practically no influence on the @xmath0 scattering amplitude near threshold .
we have investigated suggested explanations for the near - threshold enhancement in the @xmath0 invariant mass spectrum of the @xmath1 decay , reported recently by the bes collaboration .
in particular , we showed that the near - threshold enhancement in the @xmath0 mass spectrum can , in principle , be understood in terms of a final state interaction in the outgoing proton - antiproton system . within our model calculation
it can be described with the scattering amplitude in the @xmath97 @xmath0 partial wave but disagrees with the energy dependence that follows from the @xmath51 and both ( @xmath120 ) @xmath27 amplitudes .
we showed that the scattering length approximation can not be used for a more quantitative evaluation of the energy dependence of bes data .
in general , it does not reproduce the energy dependence of the scattering amplitude reliably enough within the @xmath0 mass range required . at the same time , one has to concede that the lack of knowledge of the reaction mechanism for @xmath1 and the insufficient information on the @xmath0 interaction near threshold precludes any more quantitative conclusions , even when microscopic models are used - as in the case of the present study .
we argued also that the use of the simple watson
migdal prescription for treating fsi effects has to be considered with caution for the present case in view of the small @xmath6 scattering lengths , whose real parts are typically in the order of only one fermi
. our suspicion is nourished , in particular , by the observation that the experimental @xmath0 mass spectrum of the comparable decay @xmath5 does not show any obvious sign of a @xmath0 fsi , while applicaton of the watson
migdal prescription would yield a similar , strong , near - threshold enhancement as for the @xmath121 channel for any of the @xmath2 models we utilized .
we demonstrated that a consistent qualitative description of the @xmath0 invariant mass spectra from the decay reactions @xmath1 and @xmath5 can be achieved , however , within a more refined treatment of fsi effects as it follows from a dwba approach .
though our study shows that the enhancement seen in the decay @xmath1 could indeed be a result of the fsi in the @xmath0 system one has to admit that due to the uncertainties mentioned above and the controversal situation in the @xmath122 channel explanations other than final state effects can not be ruled out at the present stage .
since the available data on the reaction @xmath5 are afflicted by large error bars it would be desirable to obtain improved experimental information here that allows one to quantify the extent of @xmath6 fsi effects in this reaction .
as already mentioned in the introduction , other reactions involving the @xmath6 system in the final state , such as @xmath7 decay @xcite or @xmath8 @xcite do show some indications for @xmath6 fsi effects .
but here too the quality of the available data is too poor to permit any reliable conclusions . in this context
let us emphasize that @xmath0 mass spectra for the reactions @xmath123 or @xmath124 would be rather interesting for clarifying the role of @xmath6 fsi effects in the decay of the @xmath20 meson .
both reactions restrict the isospin in the @xmath0 system to be zero so that one could explore the fsi for specific @xmath6 partial waves , namely @xmath51 and @xmath125 .
bai et al . , phys .
* 91 * , 022001 ( 2003 ) .
k. abe et al . , phys .
. lett . * 88 * , 181803 ( 2002 ) k. abe et al .
lett . * 89 * , 151802 ( 2002 ) f. myhrer and a.w .
thomas , phys .
b * 64 * , 59 ( 1976 ) .
i.s . shapiro , phys . rep . * 35 * , 129 ( 1978 ) .
jaffe , phys .
d * 17 * , 1444 ( 1978 ) c.b .
dover and j.m .
richard , ann . phys .
* 121 * , 70 ( 1979 ) .
dalkarov , v.m .
kolybasov , i.s .
shapiro and d.v .
voronov , phys .
b * 392 * , 229 ( 1997 ) .
richard , nucl .
. suppl . *
86 * , 361 ( 2000 ) .
d. bridges et al .
b * 180 * , 313 ( 1986 ) .
l. gray et al . , phys .
* 21 * , 1091 ( 1973 ) .
j. reidlberger et al .
c * 40 * , 2717 ( 1989 ) .
g. bardin et al . , nucl .
b * 411 * , 3 ( 1994 ) .
a. antonelli et al . , nucl .
b * 517 * , 3 ( 1998 ) .
williams , s. krewald , and k. linen , phys .
c * 51 * , 566 ( 1995 ) .
hammer , ulf - g .
meiner , and d. drechsel , phys .
b * 385 * , 343 ( 1996 ) .
b. kerbikov , a. stavinsky , and v. fedotov , phys .
c * 69 * , 055205 ( 2004 ) .
bugg , phys .
b * 598 * , 8 ( 2004 ) .
b.s . zou and h.c .
chiang , phys .
d * 69 * , 034004 ( 2004 ) .
t. hippchen , k. holinde , w. plessas , phys .
c * 39 * , 761 ( 1989 ) .
t. hippchen , j. haidenbauer , k. holinde , v. mull , phys .
c * 44 * , 1323 ( 1991 ) ; v. mull , j. haidenbauer , t. hippchen , k. holinde , phys . rev .
c * 44 * , 1337 ( 1991 ) .
v. mull , k. holinde , phys .
c * 51 * , 2360 ( 1995 ) .
r. machleidt , k. holinde , ch .
elster , phys . rep . * 149 * , 1 ( 1987 ) .
w. bruckner et al . , phys .
b * 166 * , 113 ( 1986 ) w. bruckner et al .
b * 169 * , 302 ( 1986 ) l. linssen et al .
a * 469 * , 726 ( 1987 ) d. got ta , prog .
. phys . * 52 * , 133 ( 2004 ) .
j. carbonell , j .-
richard , and s. wycech , z. phys .
* a 343 * , 325 ( 1992 ) .
m. pignone , m. lacombe , b. loiseau , and r. vinh mau , phys .
c * 50 * , 2710 ( 1994 ) .
e. widmann , proceedings of the 42nd international winter meeting on nuclear physics , bormio , italy , january 25 - february 1 , 2004 , in print ( nucl - ex/0406018 ) ; http://www - linux.gsi.de/`~`flair/. f. rathmann et al . , physics/0410067 ; p. lenisa et al . , _ antiproton - proton scattering experiments with polarization _ , .
e. byckling and k. kajantie , particle kinematics , john willey and sons ( 1973 ) .
a. gasparian , j. haidenbauer , c. hanhart , l. kondratyuk and j. speth , phys .
b * 480 * , 273 ( 2000 ) .
a. sibirtsev et al . , phys .
c * 65 * , 044007 ( 2002 ) .
a. sibirtsev et al . , phys .
c * 65 * , 067002 ( 2002 ) .
k. nakayama , j. haidenbauer , c. hanhart and j. speth , phys .
c * 68 * , 045201 ( 2003 ) . v. baru et al .
c * 67 * , 024002 ( 2003 ) . c. hanhart , phys .
rept . * 397 * , 155 ( 2004 ) .
f. hinterberger and a. sibirtsev , eur .
j. a. * 21 * , 313 ( 2004 ) .
watson , phys . rev . * 88 * , 1163 ( 1952 ) .
migdal , jetp * 1 * , 2 ( 1955 ) .
a. gasparyan , j. haidenbauer , c. hanhart and j. speth , phys .
c * 69 * 034006 ( 2004 ) . c. hanhart and k. nakayama , phys .
b * 456 * , 107 ( 1999 ) .
baru , a.m. gasparian , j. haidenbauer , a.e .
kudryavtsev and j. speth , phys . atom .
* 64 * , 579 ( 2001 ) [ yad
* 64 * , 633 ( 2001 ) ] .
m. l. goldberger and k. m. watson , collision theory ( wiley , new york 1964 ) , chapter 9.3 .
bai et al . , phys .
b * 578 * , 16 ( 2004 ) .
particle data group , phys .
b * 592 * , 1 ( 2004 ) . | we investigate the nature of the near - threshold enhancement in the @xmath0 invariant mass spectrum of the reaction @xmath1 reported recently by the bes collaboration . using the jlich @xmath2 model
we show that the mass dependence of the @xmath0 spectrum close to the threshold can be reproduced by the @xmath3-wave @xmath0 final state interaction in the isospin @xmath4 state within the watson - migdal approach .
however , because of our poor knowledge of the @xmath2 interaction near threshold and of the @xmath1 reaction mechanism and in view of the controversal situation in the decay @xmath5 , where no obvious signs of a @xmath6 final state interaction are seen , explanations other than final state interactions can not be ruled out at the present stage . |
proton and neutron ( @xmath12 ) correlation is one of the key phenomena to understand properties of @xmath1 odd nuclei ( see ref .
@xcite and references therein ) .
in particular , a deuteron - like @xmath16 @xmath12 pair plays an important role in low - lying states of light @xmath1 odd nuclei .
recently , a three - body model calculation of two nucleons with a doubly magic core nucleus has been performed to study low - lying states of @xmath1 odd nuclei , and the result indicates that deuteron - like @xmath11 and di - neutron - type @xmath9 pairs ( @xmath15-coupling @xmath12 pairs ) are predominantly formed at the surface of double magic cores such as @xmath17 @xcite .
moreover , in our previous work , we have studied @xmath12 correlation in @xmath18 and found the low - lying @xmath16 and @xmath19 states dominantly have the @xmath11 and @xmath9 pairs around a @xmath20 core , respectively @xcite .
it indicates that @xmath15-coupling scheme is better than @xmath21-coupling scheme to understand @xmath12 pairs in light @xmath1 odd nuclei even though the @xmath15-coupling @xmath12 pairs may change to @xmath21-coupling @xmath12 pairs in heavy - mass systems because of the spin - orbit mean potential .
gamow - teller transition is one of the useful observables to verify the @xmath15-coupling @xmath12 pairs because it is sensitive to the spin - isospin configuration . for light @xmath1 odd nuclei ,
the gt operator flips nucleon spins and isospins of a pair and changes the di - neutron - type @xmath10 pair to the deuteron - like @xmath12 pair .
this type of gt transitions correspond to the collectivity of the proton - neutron pair and dominate the gt sum rule if the core parts are spin - isospin saturated systems and give no contribution to the gt transition .
these modes are different from so - called gamow - teller giant resonances which are contributed by collectivity of the excess neutrons .
recently , the super - allowed gt transitions in the low - energy region have been observed and discussed in relation to the @xmath12 correlations @xcite .
the @xmath15-coupling @xmath12 pairs may play an important role to the low - energy super - allowed gt transitions .
however , there are a few theoretical works to systematically investigate @xmath15-coupling @xmath12 pairs in light @xmath1 odd nuclei , though proton - neutron pairing correlations in medium and heavy mass @xmath1 odd nuclei have been discussed in @xmath21-coupling scheme with mean field approaches @xcite .
the authors have a constructed new framework , isospin - projected @xmath3-constraint antisymmetrized molecular dynamics ( @xmath22-amd ) , which is useful in description of a @xmath12 pair in deformed or clustered systems @xcite . in this paper
, we investigate the gt transitions and @xmath12 pairs in @xmath2-shell nuclei applying @xmath22-amd to @xmath23 , @xmath18 , and @xmath24 .
we discuss strong gt transitions in terms of @xmath13 pair in @xmath15-coupling scheme and propose an interpretation of the initial and final states as spin - isospin partners .
a particular attention is paid on the role of non - zero intrinsic spin ( @xmath25 ) of the @xmath16 @xmath12 pair and its coupling with the orbital angular momentum of @xmath12 center of mass motion and that of core rotation .
the paper is organized as follows .
we briefly explain our framework in sect .
[ method ] .
we show the results of nuclear properties of energy spectra , @xmath26 , @xmath27 and @xmath28 in sect .
[ results ] .
we discuss the strong gt transitions in terms of @xmath13 pair in @xmath15-coupling scheme in sect .
[ discussion ] by analyzing the obtained wavefunctions .
a summary and an outlook are given in sect .
[ summary ] .
for @xmath1 odd nuclei , we apply @xmath22-amd @xcite in order to deal with the @xmath12 pair formation as well as nuclear deformation and clustering . for @xmath0 nuclei , we use @xmath3-constraint amd @xcite , which has been used for structure studies of light neutron - rich nuclei as well as @xmath29 even nuclei .
we here briefly explain the formulation of @xmath22-amd .
details of two methods , @xmath22-amd and @xmath3-amd are described in refs .
@xcite . in the original framework of amd ,
a basis wavefunction is written by a slater determinant of gaussian wave packets , @xmath30,\ ] ] @xmath31 \ket{\bm{\xi}_i}\ket{\tau_i}.\ ] ] in the present work , we use @xmath32 for @xmath2-shell nuclei as used in refs . @xcite . in @xmath22-amd , we perform parity and isospin ( @xmath33 ) projections before variation as @xmath34 where @xmath35 and @xmath36 are parity projection operator and isospin projection operator , respectively . for the @xmath33-projected amd wavefunction , we perform energy variation under the constraint on quadrupole deformation parameters @xmath3 and obtain the optimum solution for each set of @xmath37 and @xmath38 values . in order to obtain wavefunctions for the @xmath39th @xmath40 state ( denoted by @xmath41 )
, we superpose the angular momentum eigenstates projected from the obtained wavefunctions @xmath42 , @xmath43 where @xmath44 is the angular momentum projection operator . here
, the parameters , @xmath37 and @xmath38 , are treated as generator coordinates in the generator coordinate method ( gcm ) , and the @xmath45-mixing is taken into account .
we call this method @xmath22-amd+gcm .
we use the hamiltonian @xmath46 where @xmath45 is the kinetic energy , @xmath47 is the kinetic energy of the center of mass , and @xmath48 , @xmath49 , and @xmath50 are the central , spin - orbit , and coulomb forces , respectively . for the central and spin - orbit forces , we use the effective nucleon - nucleon ( @xmath13 ) forces same as those used for @xmath18 in the previous work @xcite .
namely , we use the volkov no . 2 force of the central force with the majorana exchange parameter @xmath51 and the g3rs force of the spin - orbit force with the strength parameters @xmath52 mev . for the bartlett and heisenberg parameters , @xmath53 and @xmath54 , of the volkov no
. 2 force , we use @xmath55 for @xmath23 , which reproduce the @xmath56-wave @xmath13 scattering lengths in the @xmath16 and @xmath19 channels . for @xmath18 and @xmath24 , we adopt a parameterization @xmath57 phenomenologically modified so as to describe energy difference between the lowest @xmath16 and @xmath19 states in each nucleus . the parameters @xmath53 and @xmath54 control the ratio ( @xmath58 ) of the central force in the @xmath16 channel to that in the @xmath19 channel .
the present choices , @xmath55 and @xmath57 , give the ratios @xmath59 and @xmath60 , respectively .
the decrease of @xmath58 is consistent with the naive expectation that the @xmath16 interaction is somewhat suppressed by a nuclear medium effect .
we should comment that , even though relative position between @xmath19 and @xmath16 spectra is sensitive to @xmath53 and @xmath54 , we obtain almost the same energy spectra in each isospin channel and also qualitatively similar results for structure properties of @xmath18 and @xmath24 in the cases of @xmath55 and @xmath57 .
calculated energy spectra of @xmath23 , @xmath18 , and @xmath24 obtained by @xmath22-amd+gcm are shown in fig . [ figure1 ]
compared with experimental spectra .
the present calculation reasonably reproduces the low - energy spectra of these nuclei .
the calculated binding energies , magnetic dipole moments ( @xmath61 ) , electric quadrupole moments ( @xmath62 ) , and @xmath63 and @xmath64 transition strengths of @xmath23 , @xmath18 , and @xmath24 are listed in table [ property ] together with experimental data .
the present calculation quantitatively or qualitatively reproduces the experimental data of these properties .
@xmath61 moments and @xmath26 as well as the gt transition strengths are observables that sensitively reflect spin configurations . the calculated @xmath61 moments of the ground states , @xmath65 , @xmath66 , and @xmath67 , and that of @xmath68 agree well to the experimental data . for the @xmath64 transition in @xmath23
, the remarkably large @xmath69 is well reproduced by the calculation . for @xmath18 ,
the @xmath64 transitions between @xmath19 and @xmath16 states are qualitatively described although quantitative reproduction is not satisfactory in the present calculation . for @xmath24 ,
the present calculation describes the general trend of the relatively strong @xmath64 transitions for @xmath70 and @xmath71 compared with those for other transitions . for @xmath18 ,
the calculated @xmath62 moment and @xmath72 are large consistently with the experimental data because of the prolate deformation .
the prolate deformation of @xmath18 is caused by formation of a @xmath20 core as shown later . .
binding energies , @xmath61 and @xmath62 moments , and @xmath64 and @xmath63 transition strengths of @xmath23 , @xmath18 , and @xmath24 .
the calculated values obtained by @xmath22-amd+gcm are shown .
experimental data are taken from @xcite . [
cols="^,^,^",options="header " , ]
we have studied the gamow - teller transitions from @xmath0 neighbors to @xmath1 odd nuclei in the @xmath2-shell region by using @xmath22-amd+gcm .
we have obtained that the strong gt transitions exhausting more than 50% of the sum rule for @xmath73 , @xmath74 , and @xmath75 . we have also found the concentration of the gt strengths of the transitions from @xmath76 states , @xmath77 , @xmath78 , @xmath79 , and @xmath80 .
these states connected with the strong gt transitions can be interpreted as `` spin - isospin partner '' states . for further analysis ,
we have introduced two - nucleon - pair densities to visualize @xmath13 pair distributions , and found that @xmath9 @xmath10 pairs and @xmath11 @xmath12 pairs are dominantly formed in the @xmath0 and @xmath1 odd nuclei , respectively .
we have studied the spin and orbital configurations of the @xmath13 pairs in @xmath15-coupling scheme and discussed the behaviors of the @xmath15-coupling @xmath13 pairs in relation to the gt transitions .
the ground states of @xmath0 nuclei , @xmath81 , @xmath82 , and @xmath83 , and their partner states , @xmath84 , @xmath85 , and @xmath86 , have major @xmath87 components , in which both the @xmath13 pairs and the core nuclei are in @xmath87 states .
the excited states , @xmath88 , @xmath89 , and @xmath90 , and their partner states have dominantly @xmath91 components mainly contributed by the @xmath13 rotation around the core , whereas @xmath92 and its spin - isospin partners have @xmath91 components with the deformed @xmath20 core rotating in @xmath91 .
based on the @xmath15-coupling @xmath13 pairs , the strong gt transitions between spin - isospin partners can be understood as spin - isospin - flip phenomena from the @xmath9 @xmath10 pairs in the @xmath0 initial states to @xmath11 @xmath12 pairs in the @xmath1 odd final states .
namely , the transitions @xmath6 , @xmath7 and @xmath8 are spin - flip phenomena of the @xmath13 pairs with @xmath93_{l=0}$ ] , whereas @xmath94 , @xmath95 , and @xmath96 are those with @xmath97_{l=2}$ ] . in the latter cases ,
the spectra of three final states with @xmath98 are split because of the spin - orbit interaction for the @xmath11 @xmath12 pairs in @xmath99 wave .
this spin - orbit splitting plays an important role in the low - energy spectra of the @xmath100 odd nuclei . on the other hand ,
the spectra of @xmath101 partnered with @xmath92 show small splitting because these states have dominant @xmath102_{l=2}$ ] component , in which the spin - orbit interaction does not affect the @xmath11 @xmath12 pairs in @xmath103 wave . in comparison with experimental data , the magnetic moments @xmath61 and the magnetic dipole transition strengths @xmath26 are reasonably reproduced in the present calculation .
moreover , relatively enhanced @xmath28 for @xmath6 , @xmath7 , and @xmath8 show consistent features with the present results .
the present calculation also succeeds in describing the concentrations of the gt strengths from the @xmath104 excited states : @xmath105 and @xmath106 .
the present framework , @xmath22-amd+gcm , is a useful tool to systematically study the @xmath12 pair correlations in @xmath107 , @xmath108 , and @xmath109 nuclei . with this method
, we can deal with nuclear deformations of the core nuclei and @xmath13 pair formation in the same footing .
this is one of the great advantages superior to three - body models with a spherical inert core .
as mentioned above , the strong gt transitions can be understood in terms of the transitions @xmath14 in @xmath15-coupling scheme .
it means that the gt transitions is a good probe to clarify the dynamics of the @xmath12 pairs in @xmath1 odd nuclei through the connection with the @xmath10 pairs in the neighboring nuclei .
we should comment that anomalous suppression of the gt transition @xmath110 is not reproduced in the present calculation and it is still a remaining problem . in light mass nuclei ,
the @xmath15-coupling @xmath12 pairs are formed . however , for heavier nuclei , the description of @xmath12 pair correlation in @xmath15-coupling scheme is no longer valid because @xmath21-coupling @xmath12 pairs and also the @xmath12 pair condensation are expected because of the spin - orbit interactions
. further investigations of @xmath1 odd nuclei in a wide mass number region from light to heavy mass nuclei are required for deeper understanding of @xmath12 pair correlations .
the computational calculations of this work were performed using the supercomputers in the yukawa institute for theoretical physics , kyoto university .
this work was supported by jsps kakenhi grant numbers 16j05659 and 26400270 .
99 s. frauendorf , a. o. macchiavelli , prog .
nucl . phys .
* 78 * ( 2014 ) .
y. tanimura , h. sagawa and k. hagino , prog .
2014 * , 053d02 ( 2014 ) .
h. morita and y. kanada - enyo , prog .
2016 * , 103d02 ( 2016 )
. y. fujita _ et al .
c * 91 * , 064316 ( 2015 ) .
h. wolter , a. faessler , p. sauer , nucl .
* a167 * , 108 ( 1971 ) .
a. gezerlis , g. f. bertsch and y. l. luo , phys .
lett . * 106 * , 252502 ( 2011 ) .
k. yoshida , phys .
c * 90 * , 031303(r ) ( 2014 ) . c. qi , j. blomqvist , t. bck , b. cederwall , a. johnson , r. j. liotta , and r. wyss , phys .
c * 84 * , 021301(r ) ( 2011 ) .
b. cederwall _ et al .
_ , nature * 469 * , 68 ( 2011 )
. t. suhara and y. kanada - enyo , prog .
. phys . * 123 * , 303 ( 2010 ) .
n. itagaki , t. otsuka , k. ikeda , and s. okabe , phys .
lett . * 92 * , 142501 ( 2004 ) .
n. itagaki , s. okabe , k. ikeda , and i. tanihata , phys . rev .
c * 64 * , 014301 ( 2001 ) .
s. okabe and y. abe , prog .
61 * , 1049 ( 1979 ) .
n. itagaki and s. okabe , phys .
c * 61 * , 044306 ( 2000 ) .
t. suhara and y. kanada - enyo , phys .
c * 82 * , 044301 ( 2010 ) .
d. r. tilley , j. h. kelley , j. l. godwin , d. j. millener , j. e. purcell , c. g. sheu , and h. r. weller , nucl .
a * 745 * , 155 ( 2004 ) .
d. r. tilley , c. m. cheves , j. l. godwin , g. m. hale , h. m. hofmann , j. h. kelley , c. g. sheu , h. r. weller , nucl . phys . * a708 * , 3 ( 2002 ) . f. ajzenberg - selove , nucl . phys .
a * 523 * , 1 ( 1991 ) .
f. ajzenberg - selove , nucl .
a * 490 * , 1 ( 1988 ) .
i. daito et al . , phys .
b * 418 * , 27 ( 1998 ) .
a. negret , t. adachi , b. r. barrett , c. baumer , a. m. van den berg , g. p. a. berg , p. von brentano , d. frekers et al .
lett . * 97 * , 062502 ( 2006 ) . j. rapaport , p. w. lisowski , j. l. ullmann , r. c. byrd , t. a. carey , j. b. mcclelland , l. j. rybarcyk , t. n. taddeucci , r. c. haight , n. s. p. king , g. l. morgan , d. a. clark , d. e. ciskowski , d. a. lind , r. smythe , c. d. zafiratos , d. prout , e. r. sugarbaker , d. marchlenski , w. p. alford , and w. g. love , phys .
c * 39 * 1929 ( 1989 ) .
p. maris , j. p. vary , p. navr@xmath111til , w. e. ormand , h. nam , and d. j. dean phys .
106 * , 202502 ( 2011 ) .
y. kanada - enyo and t. suhara , phys .
c * 89 * , 044313 ( 2014 ) . | we have studied the gamow - teller ( gt ) transitions from @xmath0 neighbors to @xmath1 odd nuclei in @xmath2-shell region by using isospin - projected and @xmath3-constraint antisymmetrized molecular dynamics combined with generator coordinate method . the calculated gt transition strengths from @xmath4 states to @xmath5 states such as @xmath6 , @xmath7 , and @xmath8 exhaust more than 50% of the sum rule . these @xmath0 initial states and
@xmath1 odd final states are found to dominantly have @xmath9 @xmath10 pairs and @xmath11 @xmath12 pairs , respectively .
based on two - nucleon ( @xmath13 ) pair picture , we can understand the concentration of the gt strengths as the spin - isospin - flip transition @xmath14 in @xmath15-coupling scheme .
the gt transition can be a good probe to identify the spin - isospin partner states with @xmath10 pairs and @xmath12 pairs of @xmath0 and @xmath1 odd nuclei , respectively . |
in a study of the radioactive decay of various actinide fractions @xcite separated from a cern w target which had been irradiated with 24-gev protons @xcite , very unusual phenomena which could not be understood from nuclear systematics , had been observed .
first @xcite , isomeric states with t@xmath4 @xmath5 0.6 y and @xmath6 30 d ( 10@xmath2 - 10@xmath7 times longer than the expected half - lives of the corresponding ground states ) were found in neutron - deficient @xmath8am and @xmath8bk nuclei , respectively . about 3x10@xmath7 atoms of @xmath8am and 4x10@xmath2 atoms of @xmath8bk
were produced in the isomeric states , and decayed by the @xmath9 or electron capture processes .
the character of these states was not clear : they are far from closed shells where high spin isomers are usually found , and they have very long lifetimes as compared to the known shape isomers .
secondly , long - lived fission activities with half - lives of several years @xcite , have been seen in all the actinide sources from am up to fm , and perhaps also in the md - no and the no - lr sources . also here
the origin of such long - lived fission activities is not known .
another observation was very low energy , 3.0 and 4.0 mev , particle groups , which were seen in the am source in coincidence with l@xmath10 x - rays in the am region @xcite .
the estimated lifetimes @xcite deviate by about 23 and 12 orders of magnitude respectively from those obtained from the systematic relationship between @xmath0-particle energies and their lifetimes @xcite .
it therefore was assumed @xcite that the observed particle groups were protons of unknown origin rather than @xmath0-particles .
perhaps the most surprising phenomenon was the relatively low energy and very enhanced @xmath0-particle groups , observed in various actinide sources which were separated from the w target @xcite .
for instance , a 5.14 mev group with a half - life of about 4 y has been seen in the bk source , and a 5.53 mev group with a half - life of about 26 d , has been seen in the lr - no source .
one is faced here with the following problems : \a ) the long lifetimes of the nuclei as compared to corresponding ground state systematics : it is expected and in fact evident from the identification of the @xmath8bk and @xmath8am nuclei , that neutron - deficient actinide nuclei can be produced via the secondary reactions in the w target .
lifetimes of several years for neutron - deficient nuclei around bk , and of about a month in the lr - no region , are very much longer than the expected minutes or hours in the first case , and seconds in the second case . here
also one is led to the conclusion that long - lived isomeric states of unknown character , which decay by the observed @xmath0-activities , were produced in the reaction .
\b ) the problem of the * low * energy of the @xmath0-particles : an energy of 5.14 mev in the neutron - deficient bk region is a * low * energy as compared to typically 6 - 7 mev , and 5.53 mev is * very low * for lr - no nuclei , where the typical transition energies are above 8 mev .
the penetrability factor for 6 mev @xmath0-particles in the bk region is about 5 orders of magnitude larger than for 5.14 mev , and for 8 mev in the no region is about 13 orders of magnitude larger than for 5.53 mev @xcite .
naturally , one assumes that an @xmath0-decay from an isomeric state is of even higher energy than normal , because of the larger available transition energy .
one is therefore faced with the problem , what is causing the nucleus to decay with very low - energy @xmath0-particles , when in principle much higher transition energy is available with correspondingly many orders of magnitude larger penetrability factor .
\c ) the problem of the very enhanced character @xcite of the @xmath0-decay : according to systematics @xcite , the half - lives for 5.14 mev @xmath0-particles for bk and nearby nuclei , and for 5.53 mev @xmath0-particles in lr - no , are around 10@xmath7 to 10@xmath11 y. the question is how the @xmath0-particle decay rate can be * enhanced * by 5 to 7 orders of magnitude .
\d ) the problem of the large cross section of the heavy - ion reaction : the production yield in the secondary reaction experiment depends on two cross sections : i ) the cross section for production of suitable fragments with high enough kinetic energies , which may interact with other w nuclei in the target , and lead to the formation of the various actinide nuclei .
ii ) the heavy ion reaction , or the evaporation residue , cross section , which depends first on the fusion cross section between such a fragment and another w nucleus in the target , and secondly on the competition with the fission process during the cooling of the produced compound nucleus . using reasonable estimates about the first process , evaporation residue cross sections in the region of mb
are needed in order to explain the production of about 10@xmath2 actinide atoms seen experimentally .
these are unexpectedly large cross sections , particularly when the production of nuclei in the lr - no region is considered where the typical evaporation residue cross sections are below 1 @xmath12b @xcite .
recently @xcite some similar abnormal decay phenomena have been observed , for the first time , in several ordinary heavy - ion reaction studies . in a study of the @xmath13o + @xmath14au reaction at e@xmath15 = 80 mev @xcite , an isomeric state has been found in @xmath16fr which decays by emitting a relatively low energy @xmath0-particle group , with e@xmath17 = 5.20 mev and t@xmath4 @xmath5 90 m. since this half - life is longer than the known half - life of the ground state of @xmath16fr ( t@xmath4 = 3.18 m ) , it was concluded that a long - lived isomeric state had been formed in this nucleus .
a t@xmath4 of 90 m for 5.20 mev @xmath0-particles in @xmath16fr is enhanced by a factor of 3x10@xmath7 as compared to normal transitions @xcite .
however , this @xmath0-group was observed in coincidence with @xmath18-rays which fit predictions for a superdeformed band @xcite .
therefore the effect of large deformations of the nucleus on the @xmath0-particle decay was calculated and found @xcite to be consistent with the observed enhancement .
it was argued @xcite that since the isomeric state decays to a high spin state , it should also have high spin , and since it decays by enhanced @xmath0-particle emission to state(s ) in the second well of the potential , it should be in the second well itself . in the same reaction
two long - lived proton activities with half - lives of about 6 h and 70 h were found @xcite with proton energies of 1.5 4.8 mev with a sharp line at 2.19 mev . a possible interpretation in terms of production of long - lived isomeric states in the second well of the potential of the parent nucleus which decay by protons to the normal states of the daughter
was given @xcite .
very recently @xcite , long - lived isomeric states , produced by the @xmath19si + @xmath20ta reaction at bombarding energies of 125 and 135 mev , which decay by strongly hindered @xmath0-particle and proton activities , have been discovered .
in particular , a very retarded 8.6 mev @xmath0-particle group with 40 d @xmath21 t@xmath4 @xmath21 2.1 y ( as compared to less than a 1 @xmath12s , typical for a normal transition of such high energy ) was found in coincidence with @xmath18-rays of a superdeformed band .
this group has been consistently interpreted @xcite in terms of production of a long - lived isomeric state in the third ( hyperdeformed ) well probably of @xmath22hg , which decays by very retarded @xmath0-particles , to the second ( superdeformed ) well in @xmath23pt . in the present work
it is shown that the low energy and very enhanced @xmath0-particle groups , seen in the heavy actinide , bk , es and lr - no sources , can be consistently interpreted as due to formation of long - lived isomeric states in the second and third wells of the potential energy surfaces in the parent nuclei , which decay to the corresponding wells in the daughters .
furthermore , it is pointed out that according to theoretical predictions @xcite , the third minima are the actual ground states of the very heavy actinide nuclei .
the experimental procedure has been published before @xcite and will be briefly reviewed .
actinide fractions from am up to lr - no were separated @xcite from our ( w3 ) @xcite tungsten target ( 6 cm long , 30 g , 99.95% pure ) which was irradiated in cern with about 1x10@xmath24 ( @xmath25 protons of 24 gev energy .
1 shows a block diagram of the chemical procedure . the main aim in the chemical separation was to separate the heavy actinides , from am up to lr , from the rare - earth isotopes and from @xmath0-emitting nuclei such as po , ra , ac , th , u and pu . several anion
( dowex-1 ) and cation ( dowex-50 ) exchange columns were used for this purpose , and at the end two fractions were obtained . the first included the remainder of the rare - earth elements , am and part of the cm .
the second fraction included the other part of cm and all the heavier actinides .
these two fractions were then passed through dowex-50 cation exchange columns with ammonium @xmath0-hydroxyisobutyrate acid at 87@xmath26 c , in order to separate individual actinides .
the elution curves obtained from these last separations are shown in fig . 2 .
the x - rays of the rare - earth elements , measured with a si(li ) detector were used throughout the whole procedure to identify the contents of the fractions .
actinide sources were prepared as indicated in fig .
2 using the measured elution positions of the rare - earth elements , and known information on the actinides @xcite .
the sources were prepared on thin glass discs which were then heated to about 600@xmath26 c to remove any organic material present .
the decontamination factor for th and pu was estimated to be @xmath6 10@xmath27 @xcite ; the decontamination factor from a typical rare earth element like gd , determined by a comparison of the residue number of @xmath28gd nuclei seen in the bk and cf sources @xcite with the total number of produced @xmath28gd nuclei , deduced from its known production cross section @xcite , was about 9x10@xmath2 .
the intensity of @xmath28gd was found to be about 4000 times larger in the cf and bk sources ( where it should preferentially be ) compared to the adjacent es and cm sources @xcite .
@xmath0-particle spectra were measured from the various sources using 450 mm@xmath29 si surface barrier detectors . in the present paper we concentrate on the 5.14 , 5.27 and 5.53 mev groups which have been seen with the bk , es and lr - no sources , respectively .
the uncertainty in the energies was about @xmath140 kev .
3 ( top ) shows a typical @xmath0-particle spectrum obtained with the bk source .
1f of ref .
[ 2 ] represents a corresponding background spectrum ) .
a pronounced peak at 5.14 mev is seen in this spectrum .
( the other pronounced peak at 5.74 mev was identified to be due to the decay of @xmath8pu , obtained after several @xmath30@xmath31 or ec transitions from an isomeric state in @xmath8bk @xcite ) .
fig . 3 ( bottom )
gives the decay curve of the 5.14 mev group , measured during about 15 y. it is seen in this figure that the intensity first grows with a half - life of [email protected] y and then decays with a half - life of [email protected] y. fig .
4 shows similar plots obtained with the es source , where a 5.27 mev @xmath0-particle group with a half - life of about 625 d is seen . in fig . 5 two spectra obtained with the lr - no source
are shown . in the first spectrum ( top ) a pronounced peak at 5.53 mev
is seen , while in the second one ( bottom ) , which was measured about 3 months later , very little of this peak remained . from these two
spectra a half - life of 26@xmath17 d is obtained .
1e of ref .
[ 2 ] is the corresponding background spectrum for fig .
4 ( top ) and fig . 5 ) . from the measured intensities and lifetimes of the @xmath0-particle groups
, it was estimated that the number of produced atoms were 3.3x10@xmath2 , 1.4x10@xmath2 and 1.4x10@xmath2 , in the bk , es and lr - no sources , respectively .
these number are correct to within @xmath150% .
it is impossible to identify the above mentioned @xmath0-particle groups with any known activity in the whole nuclear chart .
we shell now show in some detail that one can rule out the possibility that the 5.14 mev group is due to @xmath32po ( e@xmath17 = 5.12 mev ; t@xmath4 = 2.898 y ) , the 5.27 mev group is due to @xmath16po ( e@xmath17 = 5.30 mev ; t@xmath4 = 138.4 d ) , and the 5.53 mev group is due @xmath33rn ( e@xmath17 = 5.49 mev ; t@xmath4 = 3.83 d ) .
\1 ) the 5.14 mev group in fig .
3 is not due to @xmath32po since : \a ) in the chemical procedure ( fig .
1 ) there were 4 anion exchange columns where po should have been strongly adsorbed .
\b ) po is very volatile .
the last stage in the preparation of the sources , as mentioned above , was heating them to 600@xmath26c .
if some po was left after the chemical separation , then it would have evaporated during this heating .
\c ) as mentioned above the half - life of @xmath32po is 2.9 years and it does not have a long - lived parent . if the 5.14 mev is from @xmath32po , then the intensity at the beginning in fig .
3 ( bottom ) should have been about twice as much as the intensity seen at around 3 y , namely 160@xmath112 and not 20@xmath120 as seen .
in addition the intensity at the last point in fig . 3 ( bottom ) should have been [email protected] and not 20@xmath16 as seen .
\2 ) the 5.27 mev group seen with the es source in fig .
4 is not due to @xmath16po since : \a ) ( 1a ) and ( 1b ) above .
\b ) the deduced half - life is 625@xmath185 d which is about 4.5 times longer than the known half - life of @xmath16po of 138.4 d. this group is also not due to @xmath16po which is fed from @xmath16pb because : \c ) in the chemical separation procedure ( fig .
1 ) there were one 1.5 m hcl anion exchange and three cation exchange columns where pb should have been adsorbed .
\d ) the half - life of @xmath16pb is 22.3 y and of @xmath16po is 138.4 d. by normalizing to the data around 120 d and using mother - daughter relationship for these half - lives one gets that the intensity in the first measurement shown in fig .
4 ( bottom ) should have been 23@xmath13 counts , and in the last measurement 290@xmath132 .
these figures are in complete contradiction to the observed data where the corresponding numbers are 135@xmath122 and 40@xmath15 , respectively .
\3 ) the 5.53 mev group seen with the lr - no source in fig .
5 is not due to @xmath33rn ( which is introduced to the counters with air ) for the following reasons : \a ) if the 5.53 mev group is from @xmath33rn then groups from its short - lived daughters @xmath34po and @xmath35po , of about the same intensity , should have been seen at 6.00 mev and 7.69 mev in fig .
5 ( top ) .
nothing is seen in these energy regions .
\b ) a group from the air should have been very narrow , about 25 kev , and not about 130 kev , as seen experimentally in fig .
\c ) one does not see such a group of similar intensity in other spectra which were measured under identical conditions like the one displayed in fig . 5 ( bottom ) , or in the relevant background spectrum which is shown in fig .
1e of ref .
it is also not due to @xmath33rn which is fed from @xmath36ra , since one does not see in fig .
5 ( top ) the 4.78 mev @xmath0-particle group of @xmath36ra , the half - life of @xmath36ra is 1600 y much longer than the measured half - life of 26@xmath17 d , and also because of ( 3a ) above .
+ @xmath0-particle decays with similar energies but with very different half - lives of a few seconds are known in very neutron - deficient yb to pt nuclei .
for instance , @xmath37re and @xmath38pt decay with @xmath0-particles of about 5.14 mev ; @xmath39ir , @xmath40ir , @xmath41re , @xmath42os , @xmath43ta , @xmath44hf , @xmath45yb and @xmath46ta decay with @xmath0-particles around 5.3 mev , and @xmath47lu with a 5.57 mev group . in principle
one may assume that the activities seen by us are due to such decays which are fed by unknown long - lived isomeric states .
this possibility has been studied by us in some detail .
six of the above mentioned isotopes , those above w , can in principle be produced by either ( p , xn ) or secondary reactions , where the number of evaporated neutrons is between 9 to 17 .
for instance , @xmath37re can be reached by the @xmath48w(p,16n)@xmath37re reaction , and @xmath40ir by @xmath48w(@xmath11li,14n)@xmath40ir reaction .
additionally , five of the above mentioned isotopes , those below w , can in principle be produced by spallation reactions where 12 - 13 neutrons have to be evaporated beyond the most neutron deficient isotopes which are known to be produced by such reactions . by using very modest assumptions about the production cross sections of these nuclei , re from the @xmath48w(p,16n)@xmath37re reaction
, a cross section of 1.5 mb was assumed for the production of @xmath49re , like in the production of @xmath50w from @xmath20ta @xcite , and this was reduced by a factor of 5 for each additional evaporated neutron ( fig .
3a in ref .
this factor is very modest
. it may be much larger for very neutron - deficient nuclei where the proton binding energy decreases and the neutron binding energy increases .
( the binding energy for a proton in @xmath37re is about 260 kev and for a neutron it is about 11 mev ) . for the secondary reactions , the cross sections for the required high energy ( above the corresponding q - values ) heavy ions , like @xmath0-particles , @xmath11li and @xmath3be ,
were estimated from ref .
@xcite , and evaporation residue cross sections of 1 @xmath12b , 1 nb and 1 pb were assumed for 9n , 14n , and 16n reactions , respectively .
these are rather modest assumptions . for the spallation reactions ,
known cross sections @xcite for the production of the most neutron - deficient nucleus were used , and then reduced by a factor of 5 for each additional ( 12 or 13 ) evaporated neutron ( see above ) . ] and very modest assumptions or measurements of the decontamination factors of our sources from these elements , precipitate before the main chemical separation took place ( see fig .
1 ) . for the other elements it was assumed that the decontamination factor due to adsorption in an anion or a cation exchange column is at least as large as the decontamination factor of yb in the es source which was determined to be about 1200 ( see fig .
2 ( top ) ) . ]
it was estimated that the probabilities that all these isotopes , except @xmath45yb and @xmath47lu , will be produced and present in our sources are extremely small , factor of 10@xmath7 to 10@xmath13 below the experimental values .
( in addition , there are other arguments which prove that some of these isotopes are not those present in our spectra , such as the missing of competing @xmath0-particle groups or characteristic @xmath0-groups of the daughters decays . for instance , the 5.27 mev group seen in the es source can not be due to @xmath39ir , since one does not see in fig .
4 its 7.3 times stronger @xmath0-group at 5.48 mev ) . as far as @xmath45yb and @xmath47lu are concerned , the 5.27 mev group seen in the es source is not due to @xmath45yb , for the following reasons : a ) it is not seen in the lr - no source , where yb should be present with about 400 times stronger intensity , and b ) one does not see in fig .
4 the decay of its descendant , @xmath51dy , at 4.23 mev with a total estimated intensity of about 11 counts .
( only 2 counts , which may also be background events , are seen in this energy region . the probability to see 2 events when 11 are expected
is 1x10@xmath52 ) .
likewise , the 5.53 mev group seen in the lr - no source is not due to @xmath47lu , since one does not see in fig .
5 the decay of its descendant , @xmath53er , at 4.8 mev , with 90% intensity as compared to the 5.53 mev group .
thus it is seen that one may rule out the possibility that the experimentally observed @xmath0-particle groups are due to unknown long - lived isomeric states in the neutron - deficient yb to pt nuclei .
another possibility may perhaps be that the observed @xmath0-particle groups are due to unknown long - lived isomeric states in the rare - earth region .
this possibility is very unlikely for the following reasons : \a ) the decay properties of the rare - earth nuclei were studied very thoroughly , including measuring of extremely small branching ratios of @xmath0-particles , using reactions which were particularly chosen for the production of each isotope by itself .
there is no known long - lived isomeric state which decays by relatively high - energy @xmath0-particles ( 5 - 5.5 mev as compared to around 3 mev ) in the whole rare - earth region .
it is very difficult to see how such unknown activities will be produced and be present in our sources , which were separated from the rare - earth elements quite thoroughly ( the decontamination factor of the bk and es sources from their rare - earth homologs was about 10@xmath7 ( see section ii above ) , and that of the lr - no source from yb and lu was about 4.5x10@xmath29 ( fig .
2 , top and bottom ) , when they were not seen in the experiments in which the rare - earth isotopes were specifically studied .
\b ) by using the formulas of ref . 7 , the expected half - lives for such high energy @xmath0-particles in the rare - earth region are very short , 2.0x10@xmath54 s for 5.14 mev in for instance @xmath28eu , 5.6x10@xmath54 s for 5.27 mev in @xmath53 tb , and 2.4 s for 5.53 mev in @xmath55lu .
( eu , tb and lu are the chemical homologs of bk , es and lr - no , respectively ( fig .
the measured half - lives are longer by factors of 6x10@xmath56 , 9.6x10@xmath27 and 9.4x10@xmath7 , respectively .
high spin isomers can explain such large retardations only if @xmath57l@xmath58 is equal to about 15 for the first and second cases , and about 12 for the third case @xcite .
taking into account the large availability of states with various spins in the rare - earth nuclei , far from closed shells , and the various possible decay modes , it is very difficult to see a situation where a state with at least such a high spin exists , and does not have any other way to decay , but by very long - lived @xmath0 s with @xmath57l@xmath58 values of 12 or 15 .
\c ) another alternative could in principle be the existence , in rare - earth nuclei , of shape isomeric states , for instance long - lived isomeric states in the second minimum of the potential in the parent nuclei , which decay , by very retarded @xmath0-particles , to the normal states of the daughter .
( such states have been discovered by us @xcite in os , ir and hg isotopes using the reaction between the deformed @xmath19si and @xmath20ta nuclei at bombarding energy below and around the coulomb barrier ) .
however , no super - deformed minima at low spins have been theoretically found in rare - earth nuclei with z@xmath662 and n@xmath59126 @xcite .
superdeformed minima have been produced at high spins in the a @xmath5 150 region using heavy - ion reactions .
this region has been studied very carefully .
the minima are at high energies and the de - excitation is via multi - step or even single step @xmath18-ray transitions @xcite .
it is very difficult to see how unknown high - spin isomeric states will be produced in the second minima via spallation reactions and be present in our sources , taking into account that high spins are not produced preferentially by such reactions , and that our sources were separated from nuclei with a @xmath5 150 , by a factor of about 1x10@xmath7 .
since it is impossible to identify the observed @xmath0-particle groups with any known activity in the whole nuclear chart , and , in addition , as shown above , they are not due to unknown long - lived isomeric states in the neutron - deficient yb to pt nuclei , nor due to unknown long - lived isomeric states in the rare - earth region , one may try to interpret them as due to some new unknown long - lived isomeric states in the actinide region .
as mentioned in the introduction , the production of actinides via secondary reactions in the w target have been seen before @xcite . @xmath8am and @xmath8bk ( in long - lived isomeric states )
have been produced and positively identified by the decay of their daughter , after @xmath30@xmath31 or ec decay(s ) , @xmath8pu @xcite .
the measured energies of the @xmath0-particle groups are low as compared to the known energies in the heavy actinides @xcite , and their lifetimes are enhanced by a factor of 10@xmath2 to 10@xmath3 as compared to the systematics @xcite of @xmath0-particle decay in this region of nuclei . from the identification of the isomeric states in the @xmath8am and @xmath8bk nuclei @xcite , it may be deduced that neutron - deficient actinide nuclei are produced via secondary reactions in the w target . since the intensity of the 5.14 mev group in the bk source grew at the beginning
, one may conclude that this group is emitted from a daughter nucleus , produced by the @xmath30@xmath31 or ec process , rather than from an isotope of bk itself .
( one may exclude @xmath0-particle decay from the produced bk isotope to the daughter nucleus which decays by the 5.14 mev group , since no group of @xmath0-particles was seen in the first measured spectrum , taken about two and a half years before the one seen in fig . 3 ( ref .
@xcite , fig .
tables i and ii contain measured and calculated @xmath0-decay energies and lifetimes which we use to identify those nuclides emitting these unidentified @xmath0-groups .
curium and americium are the immediate @xmath9/ec decay products of berkelium . in column 1 of table 1
the @xmath0-particle energies @xcite for the ground state to ground state transitions of several neutron - deficient cm and am nuclei are given .
these energies of 5.6 to 6.5 mev ( some from experiment and some from systematics ) , are substantially higher than the observed 5.14 mev .
it is seen in table 2 that the half - life of 3.8 y is enhanced by a factor of 1.5x10@xmath7 or 3.8x10@xmath2 as compared to the calculated values @xcite for such energy from two typical parent nuclei , @xmath60am and @xmath60 cm , respectively .
it is also shown in column 2 of table 1 that the ground state to ground state transition energies ( mostly from systematics ) in neutron - deficient es isotopes are in the range of 7.4 to 8.2 mev .
these energies are much higher than the observed 5.27 mev .
furthermore , as seen in table 2 , a half - life of 625 d for such low energy @xmath0-particles in es isotopes is too fast by about a factor of 2.8x10@xmath11 .
the typical ground state to ground state @xmath0-energies for lr - no isotopes are 8 9 mev and the observed 5.53 mev is very low compared to these values .
( for a reason which will become clear in sec .
iii.c , @xmath0-energies for fm isotopes rather than for lr and no are given in table 1 ) .
also , a half - life of 26 d for 5.53 mev @xmath0-particles in the lr - no region , is enhanced by a factor of about 8.5x10@xmath11 ( see table 2 ) . from this discussion
it is obvious that previously unknown isomeric states , rather than the normal ground states , were produced in the reaction . as mentioned in the introduction
, long - lived isomeric states have recently been observed in the second @xcite and third @xcite wells of the potential energy surfaces . in
what follows , the unidentified @xmath0-groups are analyzed in terms of production of such isomeric states . in table 1 , predicted @xcite @xmath0-particle energies are given for several neutron - deficient actinide nuclei , for transitions from the second minima in the parent nuclei to the ground states and to the second and third minima in the daughters , and from the third minima in the parents to the third and second minima in the daughters .
table 2 gives calculated @xcite half - lives for various transitions from the second and the third minima in the parents , to the corresponding minima in the daughters .
the effect of large deformations on the @xmath0-decay was calculated with the potential parameters of igo @xcite but with a deformed radius + r(@xmath61 ) = c(@xmath30)r@xmath62[1 + @xmath30@xmath63y@xmath64(@xmath61 ) + @xmath30@xmath65y@xmath66(@xmath61 ) + @xmath30@xmath67y@xmath68(@xmath61 ) ] ( 1 ) + where c(@xmath30 ) is determined from the volume conservation condition and r@xmath62 = r@xmath62 x a@xmath69 .
a formation probability for the @xmath0-particle of @xcite 25(4/a)@xmath70 was used in the calculations . only the decay out of the tip
( @xmath61 = 0 ) of the emitting nucleus was directly calculated .
solid angle averaging increases the lifetimes by about a factor of 3 .
a 4% increase in the radius parameter ( from r@xmath62 = 1.17 to r@xmath62 = 1.22 fm ) , reasonable for @xmath0-decay from a highly excited isomeric state , reduces the lifetimes by about a factor of 4 .
@xmath30@xmath63 and @xmath30@xmath67 values for the different potential minima were deduced from the @xmath71@xmath63 and @xmath71@xmath67 values given in ref .
@xcite , using fig . 2 of ref .
for third minima transitions , where @xmath71@xmath63 and @xmath71@xmath65 values are given in ref .
@xcite , three calculations were performed . in the first one a @xmath30@xmath63 value was obtained as described above and @xmath30@xmath65
was taken as zero . in the second calculation
@xmath30@xmath65 was taken equal to @xmath71@xmath65 , and in the third calculation the parameters obtained by wiok et al .
@xcite for the hyperdeformed state in @xmath72th of @xmath30@xmath63 = 0.85 , @xmath30@xmath65 = 0.35 and @xmath30@xmath67 = 0.18 , were used .
it is seen in table 1 that from the energy point of view the 5.14 mev @xmath0-particle group may correspond to a ii@xmath73 to ii@xmath73 transition from @xmath60am or @xmath74am , or to a iii@xmath73 to iii@xmath73 transition from @xmath60 cm , @xmath74 cm or @xmath75am .
table 2 shows that the observed lifetime is consistent with a ii@xmath73 to ii@xmath73 transition .
it is also consistent with a iii@xmath73 to iii@xmath73 transition , if @xmath30@xmath65 = 0 .
however , the observed half - life is retarded by 1 to 3 orders of magnitude if various octupole deformations are considered .
it is seen in fig . 6
that according to the predictions of howard and mller @xcite the second minimum in @xmath60am is more pronounced than the third minimum in the same nucleus or in @xmath60 cm which favors the ii@xmath73 to ii@xmath73 transition . in conclusion ,
the 5.14 mev group seen in the bk source may be explained by an unhindered ii@xmath73 to ii@xmath73 transition from @xmath60am or @xmath74am . however , the possibility that it is due to a hindered iii@xmath73 to iii@xmath73 transition from @xmath60 cm , @xmath74 cm or @xmath75am can not be ruled out .
this is so because retardation of @xmath0-particle decay may be due to small overlap between the initial and final wave functions , even for transitions between states which belong to the same potential well .
furthermore , the third minimum in nuclei around @xmath60 cm may perhaps be more pronounced than seen in fig
. 6 if high spins are considered.pu or @xmath76 cm where two shape isomeric states have been observed , and in fact the higher excited state lives longer than the lower one .
( d. n. poenaru , m. s. ivacu , and d. mazilu in _ particle emission from nuclei _ , eds .
d. n. poenaru and m. s. ivacu , crc press , vol .
iii , p. 41
( 1989 ) . ]
let us now discuss the 5.27 mev group seen in the es source .
as seen in table 1 , columns 6 and 7 , it is consistent , from the energy point of view , with a iii@xmath73 to iii@xmath73 transition or iii@xmath73 to ii@xmath73 transition , from @xmath77es or @xmath78es .
it is seen in table 2 that the measured half - life is enhanced by about a factor of 70 as compared to the calculated value , for a ii@xmath73 to ii@xmath73 transition . in a iii@xmath73 to ii@xmath73 transition
the @xmath0-particle sees the potential of the daughter nucleus in the second minimum .
because of the change in the shape in a transition from the iii@xmath73 to the ii@xmath73 , such a transition will be hindered as compared to a ii@xmath73 to ii@xmath73 transition .
( in the case of the 8.6 mev iii@xmath73 to ii@xmath73 @xmath0-transition from @xmath22hg @xcite , the hindrance factor as compared to a ii@xmath73 to ii@xmath73 transition is in the range of 10@xmath13 ) .
this means that the enhancement factor for the lifetime of the 5.27 mev assuming iii@xmath73 to ii@xmath73 transition , should be much larger than 70 .
but it is quite difficult to understand such a large enhancement and it is much more reasonable to assume that this group is due to a iii@xmath73 to iii@xmath73 transition . considering a iii@xmath73 to iii@xmath73 transition
, it is seen in table 2 that the measured half - life is enhanced by a factor of 20 if @xmath30@xmath65 = 0 , and it is retarded by a factor of 1.6 120 if @xmath30@xmath65 and also @xmath30@xmath67 are included in the calculations . since it is easier to understand retardation of @xmath0-decay rather than enhancement , in transitions within the same potential well , it seems that a quadrupole deformation alone is not enough to describe the hyperdeformed states in the daughter nuclei around @xmath79bk .
6 shows @xcite that in the second minimum of @xmath78es the outer barrier is quite low , while for the third minimum the inner barrier is very large , and in fact the third minimum is predicted to be the true ground state of this nucleus , being 0.61 mev below the normal , low deformed , ground state @xcite .
unfortunately there are no predictions in this case for the potential at even larger deformations , beyond the third minimum .
( in the case of @xmath72th @xcite the outer barrier in the third minimum is quite high ) . to the best of our knowledge
there are no predictions for the second and third minima for lr - no nuclei .
therefore the predictions @xcite for neutron - deficient fm isotopes were used in deducing the various @xmath0-energies given in table 1 .
it is seen that such a low energy as 5.53 mev may only be due to an isomeric state in the third well of the potential . from the energy point of view
it may be either due to a iii@xmath73 to iii@xmath73 or a iii@xmath73 to ii@xmath73 transition . by extrapolating according to the @xmath0-energies from @xmath77es and @xmath80fm
, one finds the parent isotope around @xmath81no or @xmath82lr .
it is seen in table 2 that the measured half - life of 26 days is enhanced by a factor of about 160 as compared to the prediction assuming a ii@xmath73 to ii@xmath73 transition .
based on the arguments given above one may conclude that the 5.53 mev group is due to a iii@xmath73 to iii@xmath73 and not due to a iii@xmath73 to ii@xmath73 transition . regarding a iii@xmath73 to iii@xmath73 transition
, it is seen in table 2 that the measured half - life is still enhanced by a factor of about 85 if only @xmath30@xmath63 deformation is taken into account in the calculations .
however , a consistency is obtained when @xmath30@xmath65 or @xmath30@xmath65 plus @xmath30@xmath67 are included .
6 shows that the features mentioned above in @xmath78es occur also in @xmath80fm .
the outer barrier in the second minimum is low and the inner barrier in the third minimum is high , and the third minimum is in fact predicted @xcite to be the true ground state of this nucleus , being 1.76 mev below the ground state of the first minimum .
one may assume that the potential - energy curves of the nuclei around @xmath81no
@xmath82lr are similar , explaining the observation of the 5.53 mev group in the lr - no source . and [ 19 ] and which was first proposed by p. mller and j.r .
nix ( proc . int . conf . on physics and chemistry of fission , rochester , 1073 ( iaea , vienna 1974 ) vol .
i , p. 103 ) , experimental evidences were also obtained in th and u nuclei by j. blons et al .
a477 * , 231 ( 1988 ) ) , and a. krasznahorkay et al .
( aph n.s . , heavy ion physics * 7 * , 35 ( 1998 ) ) . ]
table 3 summarizes the half - lives of the isomeric states as compared to the half - lives of their corresponding normal - deformed ground states .
it is assumed in this comparison that the 5.14 , 5.27 and 5.53 mev groups are from @xmath60am , @xmath78es and @xmath81no nuclei , respectively ( see table 1 ) . for completeness the data on the isomeric states in @xmath8am and @xmath8bk @xcite
are also included in this table .
it is seen that the lifetimes of the isomeric states are 10@xmath2 to 10@xmath11 longer as compared to their corresponding ground states of normal deformations . as mentioned in the introduction
the production yield in the secondary reactions depends on two cross sections : a ) the cross section for production of suitable fragments with high enough kinetic energies , which may interact with other w nuclei in the target , and b ) the heavy - ion reaction cross section between such fragments and other w nuclei in the target . in the following
we will try to deduce the cross sections for the second reaction , based on reasonable estimates about the first step of the reaction , the production of the high - energy fragment .
table 4 summarizes the process of deducing the heavy - ion reaction cross sections .
the data regarding the production of the isomeric states in @xmath8am and @xmath8bk @xcite are also included in this table .
the numbers of produced atoms , which were deduced from the measured intensities and half - lives of the @xmath0-lines , are given in column 2 .
one can see that 3x10@xmath7 to 1.4x10@xmath2 atoms were produced in the isomeric states of the various isotopes .
column 3 gives the total cross sections which are in the 10@xmath83 to 10@xmath84 b region .
these cross sections were deduced from the number of produced nuclei , taking into account the integrated number of protons in the beam and the thickness of the target . in column 4
the most suitable possible reactions are listed assuming that the bombarding energies of the relevant secondary fragments are around the coulomb barriers between the projectile and target nuclei , at 4.2 to 4.8 mev / nucleon .
it is seen in column 5 that in general , at e@xmath15 equal to the coulomb barrier , the reactions are quite hot with excitation energies of the compound nuclei around 30 40 mev .
the yield from higher bombarding energies is presumably smaller since first , the production of high energy fragments is going down as their outgoing energy increases @xcite , and secondly , producing the compound nucleus at higher excitation energy results in a lower evaporation residue cross section , because of the competition with fission during the cooling process . only reactions with evaporation of a proton ,
an @xmath0-particle , or a proton + a neutron and an @xmath0-particle + a neutron were considered .
( the yields from the ( hi , xn ) reactions , in particular from the ( hi,3n ) reactions , are expected to be less important : first , higher neutron - excess fragments which act as projectiles are needed , and these are produced less easily ; and secondly , one loses more yield , due to competition with fission , in three step evaporation processes as compared to one or two steps = 154 mev ( e@xmath85(c.n . )
= 51.9 mev ) , cross sections of about 70 and 37 @xmath12b were obtained for the @xmath86lu(@xmath87s , p2n)@xmath88rn and the @xmath86lu(@xmath87s,@xmath02n)@xmath89at reactions respectively , as compared to about 7 @xmath12b for the @xmath86lu(@xmath87s,4n)@xmath90fr and @xmath86lu(@xmath87s,3n)@xmath88fr reactions ( s. gelberg , ph .
d. thesis , 1991 ) . ] ) .
the same cross sections were assumed for the various heavy - ion reactions which contribute to the production of a particular isotope .
the range of the total production cross sections of the various fragments listed in column 4 , are given in column 6 .
these cross sections were estimated using first the measured cross sections for the production of various sc isotopes in the interaction of 18.2 gev protons on a ta target @xcite .
( these values are very similar to those obtained in the interaction of 28.5 gev protons on pb @xcite ) . secondly , by using exponential extrapolation , values for heavier sc isotopes were obtained .
the cross sections for the various fragments considered in table 4 were then obtained using a / z scaling .
column 7 in table 4 gives the values of `` f '' , the ratio of the production cross sections for high energy fragments , above or near the coulomb barrier , to the total cross sections of the corresponding fragments .
it was assumed to be 1% for the fragments which produced the @xmath8am isotope , and reduced exponentially to 0.1% for those which produced the @xmath81no isotope .
( a cross section of about 20 @xmath12b is deduced for ar ions of @xmath64.5 mev / nucleon produced in the interaction of 5.5-gev protons with u ( fig .
14 of ref .
@xcite ) . therefore ,
cross sections of about 8 @xmath12b or 0.9 @xmath12b for the production of high energy @xmath91ti or @xmath92cu ions respectively , in the interaction of 24-gev protons with w , are reasonable ) . the last column in table 4
gives the deduced evaporation residue cross sections which are from 0.2 to 13 mb .
the results depend somewhat on the assumptions made above .
however , it seems that first , the cross sections do not decrease when the z - value of the evaporated residual nucleus increases from 95 to 102 , and secondly , the cross sections are very large as compared to known heavy - ion reactions , in particular when @xmath81no is considered , where the typical cross sections are below 1 @xmath12b @xcite .
shows calculated fusion cross sections using the coupled - channel deformed fusion code of ref .
@xcite for a typical @xmath93zn + @xmath94w reaction .
curve 7a was obtained assuming that the projectile and the target nuclei are spherical . in curve
7b the known @xcite quadrupole deformation of the target nucleus was included in the calculations .
curve 7c displays the results when coupling to the 2@xmath31 and 3@xmath95 vibrational states of the projectile nucleus were also taken into account . in the secondary reaction experiments , the projectile is a fragment which had been produced only within about 2x10@xmath84 s before interacting with another w nucleus in the target . during this short time
it is still at high excitation energy and quite deformed . in particular , since it is produced as a deformed fragment , it may have a deformation which is typical to the second or third well of the potential , rather than to normal deformation .
curve 7d shows the results of calculations taking into account the known quadrupole deformation of the target nucleus , and assuming for the projectile fragment ( @xmath93zn ) a @xmath30@xmath63 value of 0.6 , a somewhat higher value than typical for the second minimum of the potential in this region of nuclei @xcite . according to curve 7a ,
a cross section of about 10 mb is predicted at bombarding energies around the coulomb barrier of 232 mev .
this result can not explain the experimental values , since one would fall considerably below the measured evaporation residue cross sections , allowing for competition with fission in producing the final evaporation residual nuclei .
( one gets larger fusion cross sections at higher bombarding energies , but than the number of evaporated particles increases , which will decrease the evaporation residue production cross section ) . in curves 7b and 7c ,
fusion cross sections of 60 70 mb are predicted at bombarding energies around the coulomb barrier . an even larger fusion cross section of about 160 mb is predicted ( curve 7d ) when a large deformation for the fragment projectile is assumed .
these predictions are consistent with the experimental results if @xmath96@xmath97/@xmath96@xmath98 , @xmath96@xmath99/@xmath96@xmath98 and @xmath96@xmath17/@xmath96@xmath98 are not too small .
it is also seen in curve 7d that the very cold fusion reaction , i.e. the radiative capture reaction at e@xmath100 @xmath5 215 mev , below the one neutron separation energy where the competition with fission is very much reduced , is perhaps possible with a fusion cross section around 30 mb .
( if one would have assumed only this reaction as compared to the four reactions ( p , pn , @xmath0 and @xmath0n ) considered in table 4 , a four times larger cross sections than those given in the last column of table 4 are needed in order to be consistent with the experimental results ) .
thus , the unusually large evaporation residue cross sections deduced from the secondary reaction experiments , can consistently be understood , provided that very little extra - push energy @xcite is needed in order to produce the compound nucleus .
an extra - push energy means that one has to shift the plots of fig .
7 to the right by its value .
it is seen in curves 7b and 7c that around 232 mev , an extra - push energy of 5 mev reduces the fusion cross sections down to 25 35 mb .
this is pretty low , remembering that one or two particles have to be evaporated , and still leave around 10 mb for the production cross section of the residual nucleus . for curve 7d
zero extra - push energy is consistent with the data for bombarding energies around 215 mev , while around 232 mev , an extra - push energy around 10 mev is tolerable .
it should be mentioned that extra - push energies of about 300 @xcite , 50 @xcite and 20 mev @xcite were predicted for the @xmath93zn + @xmath94w reaction .
the small experimentally deduced extra - push values , of 0 10 mev , are consistent with the previous conclusion that the compound nucleus was formed in the second or third well of the potential , and not in the innermost well .
8 shows several calculated shapes of the compound nuclei , using eq .
( 1 ) for the deformed radii and various values for the deformation parameters
. the calculated shapes of the target and corresponding projectile nuclei are also given in the figure . in fig .
8 ( top ) two calculated shapes for the compound nucleus with a = 239 are displayed .
the first one ( left ) shows the calculated shape of the compound nucleus with deformation parameters @xcite of the normal ground state , while in the second one ( right ) the deformation parameters @xcite of the second minimum were used in the calculations .
three calculated shapes for a = 253 are seen in fig .
8 ( bottom ) . in the left side
the parameters of the normal ground state were used in the calculation , while in the middle only the parameter @xmath30@xmath63 of ref .
@xcite for the third minimum of the potential was used . in the right ( bottom )
side of fig .
8 the deformation parameters for the hyperdeformed minimum in @xmath72th @xcite were taken into account in our calculations .
it is qualitatively seen that much less rearrangement and matter penetration accompanied by dissipation , and hence much less extra - push energy , is needed to form the compound nucleus in the second and particularly in the third well of the potential , compared to formation in the normal states .
the existence of the long - lived isomeric states in the second and third wells of the potential can explain , in a very consistent manner @xcite , the discovery @xcite of the superheavy element with z = 112 .
the experimental evidence for the existence of this element was based mainly on the observation of fission fragments in hg sources which were separated from the cern w target @xcite , and on the mass measurements of the fissioning nuclei , which were interpreted as due to five different molecules of the isotope with z = 112 and n @xmath101 160 @xcite .
while it was impossible to disprove the data it seemed very difficult to understand them @xcite , in particular the long lifetime of the observed activities of several weeks , and the large deduced heavy ion reaction cross section of a few mb .
it was pointed out @xcite that the radiative capture processes like @xmath102w(@xmath103sr,@xmath18)@xmath104112 or @xmath94w(@xmath105sr,@xmath18)@xmath104112 are possible at bombarding energies around the coulomb barrier .
such reactions in principle could yield large evaporation residue cross sections .
however , it was thought that the large extra - push energy expected for such reactions will drastically reduce the fusion cross section .
the existence of the long - lived isomeric states in the second and third wells of the potential , and the relatively very little extra - push energy required for their production , provide a consistent interpretation to the experimental observations .
the long lifetime of several weeks shows that the @xmath106112 nucleus was probably produced in a long - lived isomeric state rather than in the normal ground state where a half - life of about 240 @xmath12s was observed for the @xmath107112 isotope @xcite .
the large deduced fusion cross section of a few mb ( @xcite and the discussion below ) , as compared to the evaporation residue cross section of about 1 pb for the production of @xmath107112 @xcite , strongly indicates that the isomeric state is in the second or third well of the potential , rather than in the first normal - deformed well .
( the observation of the isomeric states in the third well , which may actually be the true ground states in the very heavy es and lr - no nuclei , may imply a similar situation also in the superheavy nuclei region around z = 112 ) . in table 4 a cross section of 3.8 mb for the second step of the reaction
is deduced from the data , assuming a contribution from the radiative capture process alone .
( adding contributions from other reactions like ( @xmath108sr , n ) , ( @xmath108y , p ) or ( @xmath108zr,@xmath0 ) , will decrease the deduced cross section ) .
this value is about the same as those obtained in the actinides .
however , in the actinides , relatively hot reactions took place , while in the sr + w reactions an extremely cold type of reaction , the radiative capture with excitation energies of the compound nuclei of 2.0 - 10.0 mev ( table 4 ) , is possible . the lower fusion cross section in the superheavy element region of about 4 mb as compared to the predicted 60 160 mb in the actinides ( see above ) , may be due to a larger extra - push energy needed in the superheavies as compared to the actinides .
it was shown that extra - push energies around 30 40 mev are consistent with the measured fusion cross section of about 4 mb , assuming various deformations ( figs . 4 and 5 of ref .
@xcite ) , or deformations which are typical to the second minimum of the potential ( fig . 3 of ref .
@xcite ) , for the secondary sr fragment . an extra - push energy of 30 40 mev is small compared to the predictions of @xmath109300 @xcite and @xmath10950 @xcite made for the production of the compound nucleus in the normal states .
( it is compatible with an extra - push value of @xmath540 mev predicted for this reaction in ref .
it is in accord with the values of 0 10 mev obtained above in the actinides for producing the compound nucleus in the second or third well of the potential - energy surfaces .
thus , the combined effect of first , producing the compound nucleus in the third or second well of the potential , which requires much lower extra - push energy as compared to its production in the normal states , and secondly , the reduced coulomb repulsion between a w target nucleus and a deformed fragment ( produced in the secondary reaction within about 2x10@xmath84 s before interacting with the w target ) , as compared to the coulomb repulsion caused by the interaction with a normal projectile , enables one to understand the large fusion cross section , even if an extra - push energy of 30 40 mev were required in the process .
the problems of the low energies , large lifetime enhancements and high production cross sections of the @xmath0-particle groups of 5.14 , 5.27 and 5.53 mev , seen in the bk , es and lr - no sources , respectively , have consistently been interpreted as due to formation of long - lived isomeric states in the second and third wells of the potential - energy surfaces .
the low energies fit well with theoretical predictions @xcite for ii@xmath73 to ii@xmath73 ( or iii@xmath73 to iii@xmath73 ) transition in the first case , and to iii@xmath73 to iii@xmath73 transitions in the second and third cases .
the large enhancements of the @xmath0-particle decays are due to the increased penetrabilities through the barriers of superdeformed or hyperdeformed nuclei .
the large production cross sections are due to two effects : a ) the rather little , if any , extra - push energy needed to produce the compound nucleus in the second or third well of the potential - energy surfaces .
b ) the increased fusion cross section due to the deformation of the projectile ( fragment ) in the secondary reactions , as compared to normal projectiles .
this increased fusion cross section can explain the observed production of the evaporation residue nuclei , even for an extra - push energy up to 10 mev . according to theoretical predictions @xcite , the isomeric states in the third well in the es and lr - no cases ,
may actually be the true ( or nearby the true ) ground states of the particular nuclei .
the existence of long - lived isomeric states in the second and third wells of the potential - energy surfaces seems to be a general phenomenon in heavy nuclei .
they have been seen in neutron - deficient nuclei from the os
hg region @xcite , through fr @xcite , and up to the heavy actinide nuclei , like am , bk , es and lr - no .
in addition to their very long lifetimes , much longer than that of their corresponding normal - deformed ground states , they may decay by very enhanced @xmath0-particle emission , in ii@xmath73 to ii@xmath73 @xcite or iii@xmath73 to iii@xmath73 ( the present work ) transitions , or by very retarded @xmath0-particle decay , in ii@xmath73 to i@xmath73 or in iii@xmath73 to ii@xmath73 transitions @xcite , and also by long - lived proton radioactivity @xcite .
it was shown that the existence of the isomeric states in the second and third well of the potential - energy surfaces enables one to understand , in a fully consistent manner , the production of the long - lived superheavy element with z = 112 @xcite , much longer than seen in ref .
in general the discovery of such isomeric states with their very unusual decay properties , the very much reduced extra - push energy needed for their production in interactions between very heavy ions , and the possibility that the third minimum is actually the true ground state of very heavy and perhaps superheavy nuclei , provide completely new considerations in the study of heavy and superheavy elements .
we are grateful to c. j. batty , a. i. kilvington , g. w. a. newton , j. d. hemingway and s. eshhar for taking part in the first stages of the experiments .
the late a. m. friedman and v. j. robinson also participated at the beginning of the experiments .
we appreciate very much the valuable discussions with n. zeldes and the technical assistance of s. gorni , o. skala and the electronic team of the hebrew university .
d. k. acknowledges the financial support of the dfg .
a. marinov , s. eshhar , and j. l. weil , _ proc .
symp . on superheavy elements , lubbock , texas , 1978 _ , edited by m. h. k. lodhi ( pergamon , new york , 1978 ) , p. 72 .
a. marinov , s. eshhar , and d. kolb , phys .
b191 * , 36 ( 1987 ) .
a. marinov , c. j. batty , a. i. kilvington , g. w. a. newton , v. j. robinson , and j. d. hemingway , nature * 229 * , 464 ( 1971 ) .
a. marinov , c. j. batty , a. i. kilvington , j. l. weil , a. m. friedman , g. w. a. newton , v. j. robinson , j. d. hemingway , and d. s. mather , nature * 234 * , 212 ( 1971 ) .
a. marinov , s. eshhar , j. l. weil , and d. kolb , phys .
. lett . * 52 * , 2209 ( 1984 ) ; * 53 * , 1120(e ) ( 1984 ) .
a. marinov , s. eshhar , and d. kolb , fizika * 19 * , supplement 1 , 67 ( 1987 ) . v. e. viola jr . and g. t. seaborg , j. inorg . nucl . chem . * 28 * , 741 ( 1966 ) . s. hofmann , vi int .
school - seminar on _ heavy ion physics _ , dubna , russia , 1997 , eds .
oganessian and r. kalpakchieva , world scientific , p. 385 .
a. marinov , s. gelberg , and d. kolb , mod .
a11 * , 861 ( 1996 ) .
a. marinov , s. gelberg , and d. kolb , mod .
* a11*. 949 ( 1996 ) .
a. marinov , s. gelberg , and d. kolb , preprint , arxiv : nucl - ex/0006012 .
w. m. howard and p. mller , at .
data nucl .
data tables * 25 * , 219 ( 1980 ) .
g. h. higgins , _ the radiochemistry of the transcurium elements , _ nas - ns 3031 ( 1960 )
. g. w. a. newton , v. j. robinson , m. skarestad , and j. d. hemingway , j. inorg . nucl .
chem . * 35 * , 1435 ( 1973 ) .
j. r. grover , phys . rev . *
126 * , 1540 ( 1962 ) .
g. rudstam and g. s@xmath110renson , j. inorg . nucl .
, * 28 * , 771 ( 1966 ) .
a. m. poskanzer , g. w. butler , and e. k. hyde , phys .
c * 3 * , 882 ( 1971 ) .
b. neidhart and k. bchmann , j. inorg .
, * 33 * , 3227 ( 1971 ) .
s. j. krieger , p. bonche , m. s. weiss , j. meyer , h. flocard , and p .- h .
heenen , nucl . phys . *
a542 * , 43 ( 1992 ) .
a. ata et al . , z. phys .
a , * 355 * , 343 ( 1996 ) ) .
g. audi , o. bersillon , j. blachot , and a. h. wapstra , nucl .
* a624 * , 1 ( 1997 ) . g. igo , phys .
* 1 * , 72 ( 1958 ) . h. enge , _ introduction to nuclear physics _
( addison - wesley , 1971 ) .
w. nazarewicz and i. ragnarsson , _ handbook of nuclear properties _ ,
d. poenaru and w. greiner , clarendon press , oxford , p. 80
s. wiok , w. nazarewicz , j. x. saladin , w. plciennik , and a. johnson , phys . lett .
* b 322 * , 304 ( 1994 ) .
e. hageb@xmath110 and h. ravn , j. inorg .
nucl . chem . *
31 * , 897 ( 1969 ) .
y. y. chu , g. friedlander , and l. husain , phys .
c15 * , 352 ( 1977 ) . j. fernndez - neillo , c. h. dasso , and s. landowne , code ccdef , comp .
. comm . * 55 * , 409 ( 1985 ) .
s. raman , c. h. malarkay , w. t. miller , c. w. nestor , jr . , and
p. h. stelson , at .
data nucl .
data tables * 36 * , 1 ( 1987 ) .
m. girod , j. p. delaroche , d. gogny , and j. p. berger , phys .
* 62 * , 2452 ( 1989 ) .
j. p. blocki , h. feldmeier , and w. j. swiatecki , nucl . phys .
* a459 * , 145 ( 1986 ) . j. r. nix and a. j. sierk , proc .
6@xmath111 adriatic int .
_ frontiers of heavy - ion physics _ , eds . n. cindro , w. greiner and r. aplar , ( world scientific , 1987 ) , p. 333 .
frbrich , phys . lett . *
b215 * , 36 ( 1988 ) .
a. marinov , s. gelberg , and d. kolb , in _
6@xmath111 int .
conf . on nuclei
far from stability and 9@xmath111 int . conf . on atomic masses and fundamental constants ,
_ bernkastel - kues , germany , ins .
132 , p. 437
r. c. barber , n. n. greenwood , a. z. hrynkiewicz , y. p. jeannin , m. lefort , m. sakai , i. ulehla , a. h. wapstra , and d. h. wilkinson , proc .
* 29 * , 453 ( 1992 ) .
s. hofmann , v. ninov , f. p. he@xmath112berger , p. armbruster , h. folger , g. mnzenberg , h. j. schtt , a. g. popeko , a. v. yeremin , s. saro , r. janik , and m. leino , z. phys .
a * 354 * , 229 ( 1996 ) .
a. marinov , s. gelberg , and d. kolb , _ proc .
int . sym . on exotic nuclear shapes ,
debrecen , hungary , ( 1997 ) _ , eds .
dombrdi , z. gcsi , and a. krasznahorkay , aph n.s . , heavy ion physics * 7 * , 47 ( 1998 ) .
.@xmath0-particle transition energies for various actinide isotopes .
ground state to ground state transition energies were taken from audi et al .
calculated values for transitions from the second minima of the potential in the parent nuclei to the ground states and to the second and third minima in the daughters , and from the third minima in the parents to the third and second minima in the daughters , were taken from the predictions of howard and mller [ 12 ] .
numbers in * bold * are consistent with the experimental results ( see text ) . [ cols= " < ,
< , < , < , < , < , < " , ] @xmath113 excitation energies of the compound nuclei produced by the @xmath48w(x , p ) ( left ) and the @xmath94w(x , p ) ( right ) reactions are given , assuming bombarding energies which are equal to the coulomb barriers between the corresponding projectile and target nuclei ( r@xmath62 = 1.4 fm ) .
+ @xmath114 fragment cross sections were estimated according to the data of ref .
[ 26 ] ( see text ) .
+ @xmath115 the value of `` f '' , the ratio of the production cross section for a high energy fragment ( e@xmath15 @xmath116 4.2
mev / nucleon ) to the total production cross section of the particular fragment , was assumed to be 1% for the fragments which produced the @xmath8am isotope , and reduced exponentially to 0.1% for those which produced the @xmath81no isotope ( see text ) .
+ @xmath117 the same cross section was assumed for the various heavy ion reactions which contribute to the production of a particular isotope .
+ @xmath118 ref .
+ @xmath119 as `` a '' above but for the @xmath48w(@xmath120sr,@xmath18 ) ( left ) and @xmath94w(@xmath105sr,@xmath18 ) ( right ) reactions .
the mass of the @xmath104112 nucleus was taken from s. liran and n. zeldes , _ at .
data nucl .
data tables * 17 * , 431 ( 1976)_. + @xmath121 taken from u. trabitzsch and k. bchmann , radiochim .
acta * 16 * , 15 ( 1971 ) . | relatively low energy and very enhanced @xmath0-particle groups have been observed in various actinide fractions produced via secondary reactions in a cern w target which had been irradiated with 24-gev protons . in particular , 5.14 , 5.27 and 5.53 mev @xmath0-particle groups with corresponding half - lives of [email protected] y , 625@xmath184 d and 26@xmath17 d ,
have been seen in bk , es and lr - no sources , respectively .
the measured energies are a few mev lower than the known ground state to ground state @xmath0-decays in the corresponding neutron- deficient actinide nuclei .
the half - lives are 10@xmath2 to 10@xmath3 shorter than expected from the systematics of @xmath0-particle decay in this region of nuclei .
the deduced evaporation residue cross sections are in the mb region , about 10@xmath2 times higher than expected .
a consistent interpretation of the data is given in terms of production of long - lived isomeric states in the second and third wells of the potential - energy surfaces of the parent nuclei , which decay to the corresponding wells in the daughters .
the possibility that the isomeric states in the third minimum are actually the true or very near the true ground states of the nuclei , and consequences regarding the production of the long - lived superheavy elements , are discussed .
pacs numbers : 23.60.+e , 21.10.tg , 25.60.pj , 27.90.+b |
predictions of neutrino cross - sections at high energies have sizeable uncertainties which derive from the uncertainties on the parton distribution functions ( pdfs ) of the nucleon . in the framework of the quark - parton model , high energy neutrino deep inelastic scattering ( dis ) accesses large values of @xmath1 , the invariant mass of the exchanged vector boson , and small values of bjorken @xmath2 , the fraction of the momentum of the incoming nucleon taken by the struck quark .
thus in evaluating uncertainties on high energy neutrino dis cross - sections it is important to use the most up - to - date information from the experiments at hera , which have accessed the lowest @xmath2 and highest @xmath1 scales to date .
h1 and zeus have now combined the data collected in the years 19942000 to give very accurate inclusive cross - sections in the range @xmath3 and @xmath4 gev@xmath5 @xcite .
we consider modern pdf sets which include these data in order to provide the best benchmark cross - section for experiments searching for high energy cosmic neutrinos such as anita @xcite , icecube @xcite , the pierre auger observatory @xcite and rice @xcite , as well as forthcoming experiments such as ara @xcite , arianna @xcite , jem - euso @xcite and lunaska @xcite .
conventional pdf fits use the next - to - leading - order ( nlo ) dglap formalism @xcite of qcd to make predictions for dis cross - sections of leptons on hadrons . at low @xmath2 it may be necessary to go beyond the dglap formalism in order to sum @xmath0 diagrams , as in the bfkl formalism @xcite ( for recent work see refs .
@xcite ) , or to even consider non - linear terms as in the colour glass condensate model @xcite .
the present paper is concerned with estimating the uncertainties on high energy neutrino cross - sections in the conventional nlo dglap formalism .
however , the input assumptions of some of the pdf fits are arguably beyond this formalism , and such cases will be highlighted .
pioneering estimates of high energy neutrino cross - sections were obtained at leading - order @xcite using a pdf set ( cteq4 m ) which is now well out of date , yet these values continue to be used in analysing current data ( e.g. from anita @xcite ) .
two of the present authors ( css ) evaluated @xcite the cross - sections at nlo using the zeus pdfs which included more modern hera data @xcite and using a systematic procedure for estimating pdf uncertainties .
more recently the cross - sections have been calculated in ref .
@xcite ( ctw ) using the mstw2008nlo pdfs , which use hera data from the same vintage as the zeus pdfs and also account for pdf uncertainties .
however , neither of these more recent works use the newly combined hera results which have increased the accuracy of low @xmath2 data by up to a factor of 3 ( see fig.3 of ref .
it is the purpose of the present paper to re - evaluate the high energy cross - sections using the most up - to - date pdf sets , with particular emphasis on those which do use these precise , combined hera data .
the calculation is made using pdfs which were evaluated in nlo dglap fits , and our calculation of the neutrino structure functions and cross - sections is also made consistently at nlo .
the input pdfs we consider treat heavy quarks by using general - mass - variable - flavour number schemes @xcite , and we have also used such a scheme in our calculation of the structure functions .
however , the difference between the use of a general - mass or a zero - mass scheme in the latter part of this procedure is negligible since the neutrino cross - sections are dominated by scattering at high @xmath1 .
we consider carefully the source of uncertainty on the input pdfs ; these derive not only from the experimental uncertainty on the input data but also from model assumptions , and the form of the pdf parameterization .
it is important to quantify these uncertainties carefully in order to be able to probe non - perturbative qcd effects at low @xmath2 @xcite and/or new physics beyond the standard model @xcite through a determination of the dis cross - section using cosmic neutrinos which have energies extending up at least to @xmath6 gev @xcite .
the kinematics of lepton hadron scattering is described in terms of the variables @xmath1 , bjorken @xmath2 , and @xmath7 which measures the energy transfer between the lepton and hadron systems .
the double differential charged current ( cc ) cross - section for neutrino and antineutrino production on isoscalar nucleon targets is given by @xcite @xmath8 where the reduced cross - sections @xmath9 are @xmath10 , \label{eqn : nu}\ ] ] @xmath11 , \label{eqn : nubar}\ ] ] and @xmath12 , @xmath13 and @xmath14 are related directly to quark momentum distributions , with @xmath15 .
the qcd predictions for these structure functions are obtained by solving the dglap evolution equations at nlo in the scheme with the renormalisation and factorization scales both chosen to be @xmath1 .
these equations yield the pdfs at all values of @xmath1 provided these distributions have been input as functions of @xmath2 at some input scale @xmath16 . in qcd at leading order , the structure function @xmath14 is identically zero , and the structure functions @xmath12 and @xmath13 for charged current neutrino interactions on isoscalar targets can be identified with quark distributions as follows : @xmath17 and for antineutrino interactions , @xmath18 at nlo these expressions must be convoluted with appropriate co - efficient functions in order to obtain the structure functions ( and @xmath14 is no longer zero ) but these expressions still give us a good idea of the dominant contributions .
the contribution of the @xmath19 quark will be suppressed until scales @xmath20 , since the ckm element @xmath21 .
although the dominant contributions to the cc cross - sections come from @xmath22 , this does not mean that the @xmath19 contribution is always suppressed , because the relevant scale for @xmath23 production is the virtual boson - nucleon centre - of - mass energy , @xmath24 , and the high energy cross - sections are dominated by contributions from very small @xmath25 ( see fig .
[ fig : kinreg ] ) .
this point had been overlooked in earlier work , including our own @xcite .
the neutral current cross - sections on isoscalar targets are given by expressions similar to eqs .
( [ eqn : sigma ] , [ eqn : nu ] , [ eqn : nubar ] ) , with the @xmath26 propagator replaced by the @xmath27 propagator , while the leading order expressions for the structure functions given by @xmath28,\\ \nonumber xf_3^{\nu,\bar{\nu } } & = & x [ ( u -\bar{u } + d-\bar{d})(v_u a_u + v_d a_d)],\end{aligned}\ ] ] where @xmath29 , @xmath30 , @xmath31 , @xmath32 are the neutral current vector and axial - vector couplings for @xmath33 and @xmath34type quarks .
the pdf4lhc group has recently benchmarked modern parton density functions @xcite .
since our concern is with high energy neutrino cross - sections , rather than with lhc physics , we focus on pdf sets which make use of the newly combined accurate hera data @xcite .
of all the pdfs considered by the pdf4lhc only herapdf1.0 @xcite ) and nnpdf2.0 @xcite used these data .
however there has been a subsequent update of the cteq6.6 @xcite pdfs to ct10 @xcite which does use these data , while herapdf1.0 has recently updated to herapdf1.5 @xcite using an preliminary combination of hera data from 20032007 as well as the published combined data .
we will utilise the ct10 and herapdf1.5 pdfs for the present study ; we also consider the mstw2008 pdfs in order to compare with other recent calculations of high energy neutrino cross - sections @xcite , although we caution that these have _ not _ included the most accurate hera low @xmath2 data relevant to the present study .
pdfs are generally determined by assuming a parameterization in @xmath2 which is valid at a starting value of @xmath35 , where the value of @xmath16 is chosen to be sufficiently large that perturbative qcd calculations can be applied .
the form of the parameterization is usually @xmath36 , where @xmath37 is some smooth function of @xmath2 .
such forms are assumed for the light quarks and the gluon , whereas heavy quarks are generated dyamically from boson - gluon fusion .
the pdfs at all other scales @xmath38 are calculated by using the dglap equations to evolve the parameterized forms in @xmath1 .
the evolved pdfs are then convoluted with nlo matrix elements to calculate scattering cross - sections for processes of interest .
these cross - section predictions are then fitted to deep inelastic data over a broad range of the @xmath39 plane ( hera data cover 5 decades in both @xmath2 and @xmath1 ) in order to determine the parameters of the starting pdf parameterization . and
@xmath40 valence quarks are fixed by the momentum and number sum rules respectively while some other parameters are fixed by model assumptions ; the individual publications of the pdf groups should be consulted for details @xcite . ] in this way a small number of parameters ( 1025 , depending on the pdf set ) are fitted to a large number ( 10002000 ) of data points . in this approach
one can predict pdfs for @xmath2 values below those for which data exists because the fitted functional form may be applied for all @xmath2 , although uncertainties will naturally increase outside the fitted region . as we will illustrate
, the uncertainty depends mainly on the theoretical prejudice underlying the parameterization at low @xmath2 .
the hera data form the back - bone of pdf fits and are the _ only _ data which extend to the low @xmath2 region .
herapdfs use exclusively these data , while mstw and ct also use older , fixed target data , drell - yan data including @xmath26 and @xmath27 production , and tevatron jet data . moreover herapdfs are unique in using only proton data , so they are free of any assumptions concerning heavy target corrections and deuterium binding corrections @xcite .
pdfs are presented accounting for the correlated systematic errors of the data as well as the statistical and uncorrelated sources , however , each pdf group has its own approach to the estimation of confidence limits on the uncertainties . a general discussion of approaches to estimating pdf uncertainties is given in refs .
@xcite and the approaches used in the pdf sets considered here are reviewed in the pdf4lhc document @xcite .
pdf uncertainties can also arise from input assumptions made in the pdf fitting .
these include the form of the input pdf parameterization at the starting scale for evolution , @xmath16 , the value of @xmath16 itself , the kinematic cuts made on the data entering the fit , the value of @xmath41 , the values of the heavy quark masses or even the scheme used to account for heavy quark production within the dglap fits .
the pdfs considered here all account for the heavy quark production using general mass variable flavour number schemes , although the specific schemes differ : mstw2008 and herapdf1.5 use the thorne - roberts scheme @xcite , whereas ct10/cteq6.6 use the acot scheme @xcite .
the experimental uncertainties on the pdfs are presented as eigenvector error sets .
these eigenvectors represent linear combinations of the pdf parameters which are uncorrelated with each other since they are obtained by diagonalisation of the error matrix of the fitted parameters .
thus uncertainties on the pdfs , and quantities derived from them , can be calculated simply from adding in quadrature the difference between the independent eigenvector sets and the central pdf set .
however the confidence limits represented by these pdf uncertainties are not always set using the conventional tolerances of @xmath42 for a 68% c.l . , and @xmath43 for a 90% c.l . instead both ct(eq ) and mstw _ increase _ the tolerances to account for the marginal inconsistencies of some of the input data sets , and for possible parameterization bias .
the exact value of the tolerance is different for each eigenvector ( see the individual pdf publications for details ) . the average tolerance for @xmath44 c.l .
is @xmath45 for mstw2008 and @xmath46 for cteq6.6 and ct10 .
note that mstw supply both @xmath47 and @xmath44 c.l .
sets , whereas ct10/cteq6.6 provide only @xmath44 c.l .
sets ; since we wish to consider @xmath47 c.l .
uncertainties , we simply scale the ct10 error sets by @xmath48 .
the herapdfs use @xmath42 to set the size of their @xmath47 c.l . experimental errors .
the combination of the hera data results in a consistent data set with small , well understood systematic errors @xcite .
the herapdf1.0 used nc and cc @xmath49 and @xmath50 cross - section data from the first phase of hera running 19922000 ( hera - i ) , while the herapdf1.5 update uses the preliminary combination of both hera - i and hera - ii ( 20032007 ) running .
both of these herapdf sets provide not only pdf eigenvector error sets for experimental errors but also pdf variation sets which account for variation of input assumptions : namely variation of the charm and beauty masses , variation of the strangeness fraction in the sea , variation of the @xmath1 cuts applied to the data , variation of the parametrization at @xmath16 and of the value of @xmath16 itself .
not all of these sources of uncertainty are specifically considered by the other pdfs .
a potentially important uncertainty is the variation of the minimum @xmath1 cut for data entering the fit .
this is set at @xmath51 gev@xmath5 for herapdf1.5 and at @xmath52 gev@xmath5 for ct10 and mstw2008 .
although this cut should be high enough to avoid the non - perturbative region , it may include low @xmath2 data for which bfkl , or other beyond - dglap effects , are already important .
hence we vary this cut to investigate the possible bias thus introduced .
more importantly , herapdfs consider a variant of the input parameterization of the gluon at the starting scale which allows the gluon to become negative at low @xmath2 , @xmath1 . at nlo
the gluon pdf does not have to be positive , although one might consider that it going negative signals a breakdown of the dglap formalism .
this form of the gluon is in fact standard in the mstw2008 parameterization ; by contrast , the ct(eq ) analyses do not allow such negative gluon variants .
however measurable quantities such as the longitudinal structure function @xmath53 , which is closely related to the gluon at small @xmath2 , _ must _ be positive
. therefore we check whether this is the case when a negative gluon pdf is encountered .
a further uncertainty concerns the choice of the value of @xmath41 because the shape of the low @xmath2 gluon and the value of @xmath41 are correlated through the dglap equations .
all the pdf sets considered have been determined with a fixed value of @xmath41 0.120 for mstw2008 , 0.118 for ct / cteq and 0.1176 for herapdf1.5 .
however pdf sets with a range of fixed values of @xmath54 are also supplied for all these pdfs .
the uncertainty on the pdfs due to @xmath55 can be obtained by varying the value of @xmath55 by @xmath56 at @xmath47c.l .
@xcite .
the pdfs discussed above are calculated at nlo in the dglap formalism and when they are used to make predictions of the neutrino dis cross - sections the calculations should naturally be performed at a consistent order as we have done .
we have also checked the impact of the use of nnlo pdfs and evolution for herapdf1.5 and found it to be negligible .
the calculation of the cc and nc cross - sections in nlo has been performed using ` dispred ` @xcite . `
qcdnum ` @xcite as included in lhapdf is used for the evaluation of the structure functions @xmath12 , @xmath53 and @xmath13 .
the numerical integration of the differential cross - section is performed using the ` vegas ` monte carlo integration routine from the ` gsl ` library @xcite , reducing the integration error to less than 1% .
the most up - to - date pdfs in the lhapdf @xcite format are given in form of grids with a limited kinematic range .
for example , the grid extends down to @xmath57 in @xmath2 for the ct10 and herapdf1.5 sets and down to only @xmath58 for the mstw2008 set . below this range the pdfs `
freeze ' , i.e. they take the value at the lower grid boundary .
however , for very high energies the total cross - section has significant contributions from lower @xmath2 ( see also fig . [
fig : kinreg ] ) such that the cross - section calculated using the lhapdf sets will be too _
small_. therefore , we have used implementations for the pdfs provided by the cteq @xcite and mstw @xcite groups which allow an extrapolation beyond the grid .
of course , such an extrapolation using e.g. polynomials is not necessarily physical but in the absence of a grid that extends to low enough values of @xmath2 or the parametric form of the pdfs , this is the only viable alternative . we have checked the agreement between the cross - sections calculated with ` freezing ' and with the extrapolations and the difference is more than 50% for mstw2008 but only 3% for ct10 at @xmath59 gev .
figure [ fig : glu ] shows the predicted sea and gluon distributions from the herapdf1.5 fit and their fractional uncertainties , at various @xmath1 values . this shows the importance of the low @xmath2 contribution and illustrates that the pdf uncertainties are largest at low @xmath1 and at low @xmath2 .
pdf uncertainties are also large at very high @xmath2 but this kinematic region is not important for scattering at high neutrino energies . . ]
this is seen from figure [ fig : kinreg ] which indicates the kinematic regions in the @xmath39 plane which contribute to the neutrino cross - sections for two representative values of the neutrino energy : @xmath60 gev@xmath5 ( @xmath61 gev ) in the left panel and @xmath62 gev@xmath5 ( @xmath63 gev ) in the right panel .
( we do not show the antineutrino cross - sections separately because these are very close to the neutrino cross - sections at high energy . )
one can see that the dominant contributions come from @xmath64gev@xmath5 and @xmath65 for the lower neutrino energy , and @xmath66 , for the higher neutrino energy .
thus although pdf uncertainties are large at low @xmath2 and low @xmath1 , we see that the dominant contributions to the cross - section do not come from the kinematic regions of greatest uncertainty for the pdfs . plane and their contribution to the total neutrino cross - section using herapdf1.5 for two different values of @xmath67 .
the labels show the relative contribution to the total cross - section contained within each contour . ]
figure [ fig : glucomp ] compares the gluon pdf and its uncertainty at @xmath68 gev@xmath5 for the three pdfs which we consider .
this value of @xmath1 is in the middle of the range which contributes significantly to the neutrino cross - sections .
we see that the central values of the gluon pdfs are all very similar , whereas the uncertainty estimates differ .
the ct10 and herapdf1.5 uncertainties are actually very similar if we leave out member 52 from the ct10 error set .
this error set was introduced into the ct10 analysis to allow for a larger uncertainty at low @xmath2 @xcite .
previous cteq analyses such as cteq6.6 @xcite do not have such an extreme error set see left panel of fig .
[ fig : glucompnegative ] .
the problem with such an _ ad hoc _ introduction of a steeply increasing gluon pdf is that at low @xmath2 it leads to a very strong rise of the neutrino cross - section which seems unphysical ( see later discussion ) .
for the three pdf sets used .
* right panel . * the relative deviations and uncertainties ( at 68% c.l . ) with respect to the central value of herapdf1.5 .
the uncertainty bands are shown _ with _ member 9 for herapdf1.5 and member 52 for ct10 . ]
the larger error band of mstw2008 is partly due to the fact that it does not include the most up to date hera data , which have significantly reduced errors at low @xmath2 .
however the more striking difference between mstw2008 and both herapdf1.5 and ct10 is the downward divergence of its error band which is due to the gluon becoming negative at low @xmath69 .
this is best understood by reference to the right panel of fig.[fig : glucompnegative ] which shows herapdf1.5 both with and without member 9 ( the variant which allows the gluon to become negative at low @xmath69 ) .
however fig .
[ fig : flneg ] shows that while the longitudinal structure function @xmath53 is always positive for herapdf1.5 , it becomes _ negative _ , hence unphysical , for some of the mstw2008 pdf error sets . .
* left panel : * ct10 with and without member 52 ( cteq6.6 shown for comparison ) .
* right panel : * herapdf1.5 with and without member 9 . ] calculated from member 18 of mstw2008nlo68cl for @xmath70 and different values of @xmath1 ( labelled in units of @xmath71 ) .
* right panel : * the same for member 9 ( var set ) of herapdf1.5.,title="fig:",scaledwidth=49.0% ] calculated from member 18 of mstw2008nlo68cl for @xmath70 and different values of @xmath1 ( labelled in units of @xmath71 ) .
* right panel : * the same for member 9 ( var set ) of herapdf1.5.,title="fig:",scaledwidth=49.0% ] the total neutrino cross - sections are now obtained by integrating the predicted double differential cross - section @xmath72 with no cuts on either kinematic variable .
[ fig : xsechera ] and [ fig : xsecct10 ] show the nc and cc neutrino cross - sections as a function of @xmath73 as evaluated from the herapdf1.5 and ct10 pdfs respectively .
the pdf uncertainties of these predictions are shown relative to the central values underneath each plot .
cut from 2.5 to 5 gev@xmath5 is also shown ( member 8) . ]
it is clear already that the variation of input assumptions considered by the herapdf1.5 apart from the negative gluon variation do not increase the error band very significantly .
thus the other parameterization variations and the input values of the charm and beauty mass and the fraction of strangeness in the sea are not important for the predictions of neutrino cross - sections .
however it is interesting to consider the variation due to the choice of the minimum @xmath1 for data entering the fit . decreasing this cut to @xmath74 gev@xmath5 has negligible effect but the effect of increasing the cut to @xmath75 gev@xmath5 is just about visible in fig .
[ fig : xsechera ] .
this larger @xmath1 cut also cuts out data at @xmath76 the kinematic region where there are already some hints of beyond - dglap behaviour such as bfkl or non - linear effects .
this results in a steeper low @xmath2 gluon and we can see this in the marginally enhanced neutrino cross - section .
another potentially important effect comes from the variation of @xmath41 which is correlated to the gluon pdf such that lower values of @xmath41 result in a steeper low @xmath2 gluon .
we evaluate this by considering a 68% c.l .
variation of @xmath77 from its central value .
the slight enhancement in the neutrino dis cross - section is so small that it is not noticeable on the scale of figs .
[ fig : xsechera ] and [ fig : xsecct10 ] hence we have not attempted to show it . in figs [ fig : xsecallwith ] ( top panels ) we compare the nc and cc cross - sections , along with their total uncertainties ( including that coming from the variation of @xmath55 ) , as predicted by herapdf1.5 and ct10 . the mstw2008 central prediction is also included for comparison . in fig [ fig : xsecallwith ] ( bottom panels ) we emphasize the small differences in the central values of the pdfs and their relative uncertainties . in order to highlight the effect of the extreme members of herapdf1.5 and ct10 in figs [ fig : xsecallwithout ] ,
we show these plots without member 9 of the herapdf15 variations ( which allows for the gluon to become negative at low @xmath2 and @xmath1 ) and without member 52 for ct10 ( the cross - section for which rises @xmath78 whereas for the central member it rises @xmath79 ) .
however _ any _ power - law rise in the cross - section will eventually violate the froissart bound , which requires the rise to be no faster than @xmath80 @xcite .
this should result in a reduction of the cross - section at high energies , by a factor of @xmath81 at @xmath82 gev @xcite and perhaps even more @xcite . , but excluding member 9 of the herapdf1.5 set and member 52 of the ct10 set.,title="fig : " ] , but excluding member 9 of the herapdf1.5 set and member 52 of the ct10 set.,title="fig : " ] there are small differences from the previous work of css @xcite and ctw @xcite . in fig .
[ fig : ctw ] we compare our calculation using mstw2008 pdfs to that of ctw .
we find that we can reproduce their results well only if we use the mstw2008 nlo pdfs together with an _ leading - order _ treatment of the coefficient functions , rather than with a consistent nlo approach . in fig [ fig : css ]
we compare our calculation using herapdf1.5 to that of css @xcite .
the predictions for both the central values and the uncertainties of the neutrino nc cross - section are quite close .
this is also true for the cc cross - section at low @xmath73 , however above @xmath83 gev the difference increases since in the present work we have included the contribution of the @xmath19 quark which was missed out in our earlier work .
accordingly the bounds derived on the cosmogenic neutrino flux using the css cross - sections @xcite are conservative , being @xmath84 too high .
-quark contribution .
for nc scattering ( right ) the @xmath19-quark was included already by css @xcite and the agreement is excellent . ] in fig .
[ fig : anis ] we compare our results for herapdf1.5 to the cross - sections used in the neutrino event generator ` anis ` @xcite which is based on cteq5d .
note that at energies below a tev ( which is the most important energy range for neutrino telescopes like icecube
@xcite ) there is a @xmath85 discrepancy .
we also compare the cc cross - section for herapdf1.5 to its value in the genie low energy neutrino event generator @xcite at around 100 gev , finding the match to be consistent within errors .
we find that the predictions of high energy neutrino dis cross - sections from the central values of herapdf1.5 , ct10 and mstw2008 pdfs are very similar .
however the predictions for the uncertainties ( deriving from the uncertainties on the input pdfs ) differ quite strongly .
in fact pdf uncertainties derive from the input assumptions as well as from the input experimental data .
if we exclude error sets which either lead to too steep a rise in the cross - section , or allow the low @xmath2 gluon to be negative at low @xmath1 , then we find that the uncertainty estimates of herapdf1.5 and ct10 both of which use the most up - to - date , accurate hera data are remarkably consistent . our results for the high energy neutrino and antineutrino cc and
nc dis cross - sections and their uncertainties using herapdf1.5 at nlo are shown in fig . [ fig : nuandnubarhera ] .
the general trend of the uncertainties can be understood by noting that as one moves to higher neutrino energy one also moves to lower @xmath2 where the pdf uncertainties are increasing .
the pdf uncertainties are smallest at @xmath86 , corresponding to @xmath87 gev@xmath5 . moving to smaller neutrino energies brings us into the high @xmath2 region where pdf uncertainties increase again .
this effect is greater for the herapdf1.5 because the hera data have less statistics at high @xmath2 than the fixed target data which are included in ct10 ; however these data have further uncertainties that are not fully accounted for in ct10 , e.g. heavy target corrections , deuterium corrections and assumptions regarding higher twist effects .
when the high @xmath2 region becomes important the neutrino and antineutrino cross - sections are different because the valence contribution to @xmath13 is now significant .
this is seen in fig .
[ fig : nuandnubarhera ] , as is the onset of the linear dependence of the cross - sections for @xmath88 .
note that our predictions are made for @xmath89 gev@xmath5 since perturbative qcd can not sensibly be used at lower values .
moreover for @xmath67 below @xmath90 gev@xmath5 , there can be contributions to the cross - section of @xmath91 from even lower values of @xmath1 which are not accounted for here ; hence we do not show results for @xmath73 below 50 gev where there are other contributions to the neutrino cross - section and the use of a code such as genie @xcite is appropriate . for higher energies , we intend to upgrade anis @xcite to use the herapdf1.5 ( differential ) cross - sections .
meanwhile we have provided the total dis cross - sections for cc and nc scattering of neutrinos and antineutrinos on isoscalar targets in tables [ tbl : nuhera ] and [ tbl : nubarhera ] and recommend these as a benchmark for use by experimentalists .
these cross - sections as well as those for isoscalar targets are available from a webpage @xcite ; differential cross sections are available upon request . any measured deviation from these values would signal the need for new physics beyond the dglap formalism .
we are grateful to james ferrando for providing us with an up - to - date version of ` dispred ` and for discussions .
we also thank mike whalley and voica radescu for the speedy implementation of herapdf1.5 in lhapdf .
pm and ss thank their colleagues in the auger and icecube collaborations for stimulating exchanges and acknowledge partial support from the eu research & training network `` unification in the lhc era '' ( pitn - ga-2009 - 237920 ) .
.neutrino cc and nc cross - sections on isoscalar targets , along with their uncertainties , in the perturbative dglap formalism at nlo , using herapdf1.5 ( both with and without member 9 ) .
[ cols= " > , > , < , > , < , > , < , > , < , > , < , > , < , > , < , > , > " , ] f. d. aaron _ et al . _
[ h1 & zeus collaboration ] , jhep * 1001 * ( 2010 ) 109 .
p. w. gorham _ et al .
_ [ anita collaboration ] , phys .
* 103 * ( 2009 ) 051103 .
d * 82 * ( 2010 ) 022004 .
r. abbasi _ et al . _
[ icecube collaboration ] , phys .
d * 82 * ( 2010 ) 072003 ; phys .
d * 83 * ( 2011 ) 092003 .
j. abraham _ et al .
_ [ the pierre auger collaboration ] , phys . rev .
* 100 * ( 2008 ) 211101 ; phys .
d * 79 * ( 2009 ) 102001 .
i. kravchenko _ et al .
_ , [ arxiv:1106.1164 [ astro-ph.he ] ] . c. w. james _ et al .
_ , phys .
d * 81 * ( 2010 ) 042003 .
g. altarelli , g. parisi , nucl .
b * 126 * ( 1977 ) 298 .
v. n. gribov , l. n. lipatov , sov .
j. nucl .
phys . * 15 * ( 1972 ) 438 .
l. n. lipatov , sov .
j. nucl .
* 20 * ( 1975 ) 94 .
y. l. dokshitzer , sov .
jetp * 46 * ( 1977 ) 641 .
e. a. kuraev , l. n. lipatov , v. s. fadin , sov .
jetp * 45 * ( 1977 ) 199 .
i. i. balitsky , l. n. lipatov , sov .
j. nucl .
* 28 * ( 1978 ) 822 .
l. n. lipatov , sov .
jetp * 63 * ( 1986 ) 904 .
g. altarelli , r. d. ball , s. forte , nucl .
b * 742 * ( 2006 ) 1 .
m. ciafaloni , d. colferai , g. p. salam , a. m. stasto , phys .
b * 635 * ( 2006 ) 320 . c. d. white , r. s. thorne , phys .
d * 74 * ( 2006 ) 014002 .
j. rojo , g. altarelli , r. d. ball , s. forte , [ arxiv:0907.0443 [ hep - ph ] ] .
f. caola , s. forte , j. rojo , phys . lett .
* b686 * ( 2010 ) 127 ; nucl .
* a854 * ( 2011 ) 32 . f. gelis , e. iancu , j. jalilian - marian , r. venugopalan , ann .
nucl . part .
* 60 * ( 2010 ) 463 .
v. p. goncalves , p. hepp , phys .
d * 83 * ( 2011 ) 014014 r. gandhi , c. quigg , m. h. reno , i. sarcevic , phys .
d * 58 * ( 1998 ) 093009
. s. chekanov _ et al .
_ [ zeus collaboration ] , phys . rev .
d * 67 * ( 2003 ) 012007 .
a. cooper - sarkar , s. sarkar , jhep * 0801 * ( 2008 ) 075 .
a. connolly , r. s. thorne , d. waters , arxiv:1102.0691 [ hep - ph ] .
r. s. thorne , r. g. roberts , phys .
d * 57 * ( 1998 ) 6871 .
r. s. thorne , phys .
d * 73 * ( 2006 ) 054019 .
l. a. anchordoqui , a. m. cooper - sarkar , d. hooper , s. sarkar , phys .
d * 74 * ( 2006 ) 043008 .
a. kusenko , t. j. weiler , phys .
* 88 * ( 2002 ) 161101 .
l. anchordoqui , t. han , d. hooper , s. sarkar , astropart .
* 25 * ( 2006 ) 14 .
s. palomares - ruiz , a. irimia , t. j. weiler , phys .
d * 73 * ( 2006 ) 083003 . l. a. anchordoqui , h. goldberg , d. gora , t. paul , m. roth , s. sarkar , l. l. winders , phys . rev .
d * 82 * ( 2010 ) 043001 .
l. a. anchordoqui , t. montaruli , ann .
nucl . part .
* 60 * ( 2010 ) 129 .
r. devenish , a. cooper - sarkar , `` deep inelastic scattering '' , oxford university press ( 2004 ) .
w. k. tung , h. l. lai , a. belyaev , j. pumplin , d. stump , c. p. yuan , jhep * 0702 * ( 2007 ) 053 .
h. -l .
lai , m. guzzi , j. huston , z. li , p. m. nadolsky , j. pumplin , c. -p .
yuan , phys . rev .
* d82 * ( 2010 ) 074024 .
a. cooper - sarkar [ on behalf of the h1 & zeus collaborations ] , arxiv:1012.1438 [ hep - ph ] .
a. accardi , w. melnitchouk , j. f. owens , m. e. christy , c. e. keppel , l. zhu , j. g. morfin , arxiv:1102.3686 [ hep - ph ] .
a. m. cooper - sarkar , j. phys .
g * 28 * ( 2002 ) 2669 .
m. kramer , 1 , f. i. olness , d. e. soper , phys .
* d62 * ( 2000 ) 096007 . j. m. campbell , j. w. huston , w. j. stirling , rept .
phys . * 70 * ( 2007 ) 89
. m. botje _ et al .
_ , arxiv:1101.0538 [ hep - ph ]
. j. ferrando , [ arxiv:1007.5489 [ hep - ph ] ] ; j. ferrando , p. mertsch , in preparation .
r. fiore , l. l. jenkovszky , a. v. kotikov , f. paccanoni , a. papa , phys .
d * 73 * ( 2006 ) 053012 .
m. m. block , p. ha , d. w. mckay , phys .
* d82 * ( 2010 ) 077302 .
a. y. illarionov , b. a. kniehl , a. v. kotikov , [ arxiv:1105.2829 [ hep - ph ] ] .
a. gazizov , m. p. kowalski , comput .
commun .
* 172 * ( 2005 ) 203 .
r. abbasi _ et al . _
[ icecube collaboration ] , phys .
* d83 * ( 2011 ) 012001 . c. andreopoulos
_ et al . _ ,
instrum .
meth . * a614 * ( 2010 ) 87 . | updated predictions are presented for high energy neutrino and antineutrino charged and neutral current cross - sections within the conventional dglap formalism of nlo qcd using modern pdf fits .
pdf uncertainties from model assumptions and parametrization bias are considered in addition to the experimental uncertainties .
particular attention is paid to assumptions and biases which could signal the need for extension of the conventional formalism to include effects such as @xmath0 resummation or non - linear effects of high gluon density . |
the wronski map is a finite morphism from a grassmannian to a projective space of the same dimension , which is remarkably well - behaved . in this paper , we consider the problem of counting points in a fibre of the wronski map that are fixed by the action of a cyclic group .
our motivation comes from questions in real algebraic geometry and combinatorics , which shall be discussed later in the introduction .
we begin by defining the wronski map and recalling some of the of properties that make it an interesting object of study in enumerative geometry . for any non - negative integer @xmath9 , let @xmath10}$ ] denote the vector space of polynomials of degree at most @xmath9 .
fix integers @xmath11 , and let @xmath12})$ ] denote the grassmannian of @xmath13-planes in the @xmath14-dimensional vector space @xmath15}$ ] .
the dimension of @xmath16 is @xmath17 .
if @xmath18 is spanned by polynomials @xmath19}$ ] , we consider the wronskian @xmath20 it is easy to show that @xmath21 is a non - zero polynomial of degree at most @xmath22 , and different choices of basis for @xmath23 give the same wronskian up to a scalar multiple . thus @xmath21 determines a well - defined element of projective space @xmath24})}$ ] .
we adopt the convention that equations involving @xmath21 hold only up to a scalar multiple , and should be properly interpreted in projective space .
the morphism @xmath25})}$ ] , @xmath26 is called the * wronski map*. there are many equivalent formulations . for example , the map @xmath27 $ ] defines a rational curve @xmath28 ; the roots @xmath21 are the _ ramification points _ of @xmath29 .
we refer the reader to the survey article @xcite for a detailed background .
eisenbud and harris showed that the wronski map is a finite morphism @xcite , of degree is equal to the number of standard young tableaux of rectangular shape @xmath30 . by studying the asymptotic behaviour of the fibres , eremenko and gabrielov refined this result @xcite .
they described a bijection between the set @xmath31 of all standard young tableaux of shape @xmath32 , and points in the fibre @xmath33 for certain polynomials @xmath34})}$ ] .
more generally , one can consider the restriction of the wronski map to a richardson variety .
let @xmath35 denote the set of all partitions @xmath36 , with at most @xmath13 parts and @xmath37 , and consider the complete flags @xmath38 } \end{gathered } \qquad\qquad \text{and } \qquad\qquad \begin{gathered } { \widetilde f}_\bullet : { \widetilde f}_0 \subset { \widetilde f}_1 \subset \dots \subset { \widetilde f}_n \\ { \widetilde f}_i = { { \mathbb{c}}_{i-1}[z]}\ , . \end{gathered}\ ] ] for partitions @xmath39 in @xmath35 , the * richardson variety * @xmath40 is the intersection of a schubert variety ( corresponding to @xmath41 ) and and opposite schubert variety ( corresponding to @xmath42 ) .
we will say that @xmath43 is * compatible * with a polynomial @xmath44 if @xmath45 and @xmath46 .
this definition is motivated by the following fact .
[ prop : richardson ] every point in @xmath47 lies in a unique @xmath44-compatible richardson variety .
if we restrict the wronski map to any richardson variety @xmath43 , we obtain a finite morphism over the closure of of the space of compatible polynomials .
these restrictions interact nicely with the combinatorics of tableaux .
the fibres are generically reduced , and the degree of this map is equal to the number of standard young tableaux of skew shape @xmath48 .
the correspondence of eremenko and gabrielov extends to this situation , giving a bijection for specially chosen polynomials .
one of the most remarkable facts about the wronski map is the reality theorem of mukhin , tarasov and varchenko ( formerly the shapiro - shapiro conjecture ) .
we say that a point @xmath18 is * real * if the subspace of @xmath15}$ ] represented by @xmath23 has a basis @xmath49}$ ] .
[ thm : mtv ] let @xmath50}$ ] be a polynomial of degree @xmath22 or @xmath51 whose roots are distinct real numbers .
then the fibre @xmath47 is reduced and every point in the fibre is real .
apart from being a very striking and surprising statement one does not normally expect an algebraic system with real parameters to have only real solutions this theorem has interesting combinatorial consequences . for example , it can be used to show that the correspondence of eremenko and gabrielov can be unambiguously extended to polynomials with real roots .
this extension is not continuous , however it is discontinuous in an interesting way . in @xcite
we showed that the discontinuities encode the jeu de taquin theory on tableaux .
many of the results and methods from this previous work will be used in our present paper .
the mukhin - tarasov - varchenko theorem does not hold if @xmath50}$ ] is a real polynomial with complex roots .
computer experiments by hauenstein , hein , martn del campo and sottile show that in general , the number of real points in @xmath47 depends in a complicated way on the polynomial @xmath44 , and not just on the number of real roots @xcite . however , there are other situations where the number of real solutions is predictable .
the work in this paper was originally motivated by the following problem .
suppose @xmath50}$ ] is a real polynomial of degree at most @xmath22 with only _ pure imaginary _ roots .
how many real points are in the fibre @xmath47
? special cases of this problem were studied by eremenko and gabrielov , and used to construct instances of the pole placement problem in control theory with no real solutions @xcite .
to answer this ( and other related questions ) , it is helpful to observe that the wronski map is equivariant with respect to a @xmath0-action .
the group @xmath52 acts on each vector space @xmath10}$ ] by mbius transformations : if @xmath53 , we let @xmath54 for @xmath55}$ ] .
the action @xmath52 on @xmath15}$ ] and @xmath56}$ ] induces an action of @xmath0 on @xmath16 , and on @xmath24})}$ ] , and the wronski map intertwines these two @xmath0-actions .
the involution @xmath57 is given by @xmath58 , and ( up to a sign ) fixes every real polynomial @xmath44 with only imaginary roots .
hence for these polynomials , @xmath59 gives an involution on the fibre @xmath47 .
one can then show , using the mukhin - tarasov - varchenko theorem , that the fixed points of this involution are exactly the real points in the fibre .
we are left with the problem of counting the number of @xmath59-fixed points in the fibre over a @xmath59-fixed polynomial .
more generally , let @xmath2 be a cyclic subgroup of order @xmath3 .
we can consider the problem of counting @xmath4-fixed points in the fibre of a @xmath4-fixed polynomial in @xmath24})}$ ] .
as any two are of these cyclic subgroups are conjugate , we will generally take @xmath60 then , a polynomial @xmath44 is @xmath4-fixed as an element of @xmath24})}$ ] if and only if it is of the form @xmath61 where @xmath62 , @xmath63 and @xmath64 .
when we speak of a polynomial being @xmath4-fixed , we will always mean that it is fixed as an element of the appropriate projective space ; indeed this is the only reasonable interpretation , since @xmath4 is a subgroup of @xmath0 , not of @xmath52 .
let @xmath65 denote the @xmath4-fixed subscheme of @xmath16 .
for any @xmath44-compatible richardson variety @xmath43 , we consider @xmath66 the @xmath4-fixed points of the fibre of the wronski map , that lie in the richardson variety @xmath43 . our main result is a combinatorial formula for the number of points in @xmath67 : we show that there is a bijection between these points and the set of of _ standard @xmath3-ribbon tableaux _ of shape @xmath48 . when @xmath68 , these are just standard young tableaux , and the group is trivial ; hence we recover the fact that @xmath69 has degree @xmath70 .
the problem of counting real solutions when @xmath44 is a real polynomial with pure imaginary roots corresponds to the special case of an involution , i.e. @xmath5 . in the case where @xmath3 divides @xmath22 , @xmath71 and @xmath72
, we can determine the answer in a second way in terms of the jeu de taquin _ promotion _ operator on rectangular tableaux giving a non - trivial combinatorial identity .
these results can also be extend to dihedral group actions .
we now recall the definitions of the combinatorial objects that are relevant to the statements of our results .
we assume some basic familiarity here , and refer the reader to @xcite for a more thorough introduction .
let @xmath73 be partitions , with @xmath39 . by a * tableau * of shape @xmath48 , we will mean a filling of the boxes of the diagram of @xmath48 with positive integer entries that are weakly increasing along rows and down columns .
a tableau of shape @xmath48 is a * standard young tableau * if its entries are exactly the numbers @xmath74 .
we denote the set of all standard young tableau of shape @xmath48 by @xmath75 .
more generally , suppose @xmath76 , and consider a tableau of shape @xmath48 satisfying the following conditions : the entries are @xmath77 .
the shape determined by the entries labelled @xmath78 is a ( connected ) @xmath3-*ribbon * ; that is , a connected skew shape with @xmath3 boxes that does not contain a @xmath79 square . such a tableau
is called a * standard @xmath3-ribbon tableau * ( also called a _
standard @xmath3-rim hook tableau _ ) . when @xmath68 , this definition coincides with the definition of a standard young tableaux . when @xmath5 , the entries labelled @xmath78 must cover two adjacent boxes , and hence standard @xmath80-ribbon tableaux are often called * domino tableaux*. an example for @xmath81 is given in figure [ fig : ribbontableau ] .
we denote by @xmath82 the set of all standard @xmath3-ribbon tableaux of shape @xmath48 . , & ! 1 & ! 1 & ! 1 & ? ?
1 & ! 1 & ? 2 & ? ?
2 & ? 2 & ? 2 & ? ? 3 & ! ! ! 5 & ! ! !
5 + ? 2 & ! !
4 ribbon tableaux have been well studied .
there are several formulae for @xmath83 , including hook - length formulae and character formulae ( see @xcite and the references therein ) .
lascoux , leclerc and thibon showed that evaluations of kostka - foulkes polynomials at roots of unity can be interpreted as enumerations of ribbon tableau . in our case ,
the relevant kostka - foulkes polynomial is @xmath84 ( where @xmath85 ) , which is the generating function for @xmath75 with respect to the _ charge _ statistic .
the following is a special case of their result , which may be found in ( * ? ? ?
* section 4 ) .
[ thm : llt ] suppose @xmath76
. then @xmath86 in the case where @xmath87 , the kostka - foulkes polynomial is up to a power of @xmath88 given by the @xmath88-analogue of the hook - length formula @xmath89!_q } { \displaystyle\prod_{\alpha \in { { \mathchoice{\raisebox{.05ex}{$\sqsubset\!\!\sqsupset$ } } { \raisebox{.05ex}{$\sqsubset\!\!\sqsupset$ } } { \sqsubset\!\!\sqsupset } { \sqsubset\!\!\sqsupset } } } } [ h_\alpha]_q}\,,\ ] ] and the sign appearing in theorem [ thm : llt ] is @xmath90 .
these same numbers appear in another combinatorial context . for @xmath91 , let @xmath92 be the jeu de taquin * promotion * operator . for @xmath93
, @xmath94 is defined by performing the following steps : delete the entry @xmath95 from @xmath96 , leaving an empty box in the northwest corner .
slide the empty box through the remaining entries of @xmath96 .
( the empty box repeatedly swaps places with either its neighbour to the east or its neighbour to the south , whichever is smaller , until neither exists . )
decrement all entries by @xmath95 and place a new entry @xmath97 in the empty box .
the result is @xmath94 .
an example is given in figure [ fig : promotionex ] .
@xmath98 1 & 2 & 5 \\ 3 & 4 & 7 \\ 6 & 8 & 9 \end{young } } \ \to\ { \begin{young}[c ] ? & 2 & 5 \\ 3 & 4 & 7 \\ 6 & 8 & 9 \end{young } } \ \to\ & { \begin{young}[c ] 2 & ? & 5 \\ 3 & 4 & 7 \\ 6 & 8 & 9 \end{young } } \
\to\ { \begin{young}[c ] 2 & 4 & 5 \\ 3 & ? & 7 \\ 6 & 8 & 9 \end{young } } \ \to&\\[1ex ] \to\ & { \begin{young}[c ] 2 & 4 & 5 \\ 3 & 7 & ?
\\ 6 & 8 & 9 \end{young } } \
\to\ { \begin{young}[c ] 2 & 4 & 5 \\ 3 & 7 & 9 \\ 6 & 8 & ?
\end{young } } \ \to\ { \begin{young}[c ] 1 & 3 & 4 \\ 2 & 6 & 8 \\ 5 & 7 & 9 \end{young } } = { \textbf{\textit{j}}}(t)\,.\end{aligned}\ ] ] it is not hard to see that @xmath92 is invertible , and hence generates a cyclic group action on @xmath99 . in the case
where @xmath100 , this operator has particularly nice properties .
haimain showed that @xmath101 for all @xmath102 @xcite , hence we may think of this cyclic group as having order @xmath22 .
more generally , rhoades proved the following _ cyclic sieving theorem _ for the promotion operator on @xmath31 @xcite ( see also @xcite ) .
[ thm : cyclicsieving ] suppose @xmath3 divides @xmath22 .
then the number of @xmath103-fixed tableaux in @xmath31 is equal to @xmath104 .
when @xmath68 , the evaluation of the kostka - foulkes polynomial @xmath105 is just the ordinary kostka number @xmath106 , and hence we recover haiman s result .
more generally this theorem gives a complete description of the orbit structure of promotion on @xmath31 .
the proof of theorem [ thm : cyclicsieving ] is quite non - elementary .
the idea is to show that the long cycle acts by promotion on the kazhdan - lusztig basis of the representation of the symmetric group @xmath107 associated to the partition @xmath32 ; hence the number of fixed points is given as a character evaluation . in light of theorem [ thm : llt ] ,
rhoades result is equivalent to the following purely combinatorial theorem , for which there is no known bijective proof .
[ thm : combinatorial ] suppose @xmath3 divides @xmath22 .
the number of @xmath103-fixed tableaux in @xmath31 is equal to the number of standard @xmath3-ribbon tableaux of shape @xmath32 .
one consequence of our present work will be an alternate proof of theorem [ thm : combinatorial ] .
our proof is direct in the sense that it does not require any knowledge of representation theory , kostka - foulkes polynomials , their evaluations at roots of unity , or any other algebraic formulae for the two quantities involved . on the other hand
, the proof does not give a direct bijection between the two sets , nor does it suggest how to construct one ( except in the @xmath5 case , where a bijection is known @xcite ) . from our point of view , the two sets come from two different ( and seemingly incompatible ) ways of thinking about the problem of counting @xmath4-fixed points in the fibre of the wronski map .
there is another operator on standard young tableaux , closely related to promotion . for @xmath91 , the * evacuation * operator
@xmath108 is @xmath109 where @xmath110 is the promotion operator restricted to the entries @xmath111 . in other words , to compute @xmath112 , for @xmath93 , we first apply the promotion operator @xmath113 to @xmath96 .
we then freeze the largest entry @xmath97 in @xmath94 , and apply promotion to the subtableau of @xmath94 consisting of entries @xmath114 . at the next stage entries
@xmath97 and @xmath115 are frozen , and we continue in this way until all entries are frozen .
the result is @xmath112 .
the operator @xmath116 is an involution on @xmath99 . in the case of a rectangular shape @xmath100
, there is a very simple description of this involution : it coincides with the operation of rotating the tableau by @xmath117 and reversing the order of the entries ( replacing each entry @xmath118 by @xmath119 ) . from here
, it is easy to see that @xmath120 , and hence @xmath116 and @xmath121 generate a dihedral group @xmath122 of order @xmath123 acting on @xmath31 .
we may therefore consider the problem of counting tableaux in @xmath31 that are fixed by a dihedral subgroup @xmath124 of order @xmath125 .
if @xmath22 is odd , every dihedral subgroup of order @xmath125 is conjugate , so we may take @xmath126 to be generated by @xmath116 and @xmath103 .
if @xmath22 is even , there are up to conjugacy two dihedral subgroups of order @xmath125 : one is generated by @xmath116 and @xmath103 , the other by @xmath127 and @xmath103 . to answer this question in a manner analogous to theorem [ thm : combinatorial ] , we consider two special types of ribbon tableaux . for each @xmath91 ,
let @xmath128 denote the dual partition : @xmath129 for @xmath130 .
we will say a tableau is * rotationally - invariant * if it is fixed by the operation of rotating @xmath117 inside the rectangle @xmath32 , and replacing each entry @xmath118 by @xmath131 , where @xmath9 is the largest entry . for a tableau to be rotationally - invariant ,
its shape must be of the form @xmath132 , where @xmath133 .
suppose @xmath76 .
we define @xmath134 to be the set of all rotationally - invariant standard @xmath3-ribbon tableaux ( this is empty if @xmath135 .
if @xmath136 is even , we also define @xmath137 be the set of rotationally - invariant tableaux with entries @xmath138 , in which the entries not equal to @xmath139 form @xmath3-ribbons , and the entries equal to @xmath139 form a @xmath125-ribbon .
some examples are given in figure [ fig : rotribbon ] .
we note that @xmath137 is always empty if @xmath13 and @xmath14 are both even . !
1 & ! 1 & ! 1 & ! 1 & ! !
4 & ? 5 + !
1 & ? 2 & ? ? 3 & ! ! 4 & ! !
4 & ? 5 + ?
2 & ? 2 & ? ? 3 & ! !
4 & ? 5 & ? 5 + ?
4 & ? 5 & ! 6 + ?
2 & ? ? 3 & ! 6 & ! 6 & ! 6 & ! 6 @xmath140 ! 1 & ! 1 & ! 1 & ! 1 & ? ?
1 & ? 2 & ? 2 & ? ?
2 & ? 2 & ? ?
3 & ? 4 & ? 4
3 & ? 4 & ? 4 & !
3 & ! 5 & ! 5 & ! 5 & ! 5 [ thm : d - combinatorial ] suppose @xmath3 divides @xmath22 .
the number of tableaux in @xmath31 fixed by both @xmath116 and @xmath103 is equal to @xmath141 .
the number of tableaux in @xmath31 fixed by both @xmath127 and @xmath103 is equal to @xmath141 if @xmath8 is odd , and @xmath142 if @xmath8 is even . in the @xmath68 case we are computing the numbers of standard young tableaux of rectangular shape fixed by @xmath116 , or @xmath127 , which were previously studied in @xcite . since these are involutions , the quantities involved are related to the @xmath5 case of a cyclic group action .
the general case of theorem [ thm : d - combinatorial ] appears to be new , and follows from dihedral group analogues of our main theorems .
our main result relates the problem of counting @xmath4-fixed points in the fibre of the wronski map to enumeration of @xmath3-ribbon tableaux .
[ thm : ribbon ] let @xmath44 be a generic @xmath4-fixed polynomial of the form .
for any @xmath44-compatible richardson variety @xmath43 , @xmath67 is a reduced finite scheme .
the number of points in @xmath67 is equal to the number of standard @xmath3-ribbon tableaux of shape @xmath48 .
when @xmath143 , and @xmath144 where @xmath62 , there is only one compatible richardson variety , @xmath145 itself . in this case
, we write @xmath146 for the @xmath4-fixed points in the fibre @xmath47 .
theorem [ thm : ribbon ] then states that @xmath146 is generically reduced , of size equal to @xmath147 . in this case
, we will also show the following .
[ thm : promotion ] let @xmath44 be a generic polynomial of the form .
the number of points in @xmath146 is equal to the number of @xmath103-fixed tableaux in @xmath31 . putting theorems [ thm : ribbon ] and [ thm : promotion ] together , we deduce theorem [ thm : combinatorial ] .
these two results are proved by considering the fibres over certain special classes of polynomials . to prove theorem [ thm : ribbon ] , we change our field from the complex numbers to a field of puiseux series @xmath148 , where @xmath149 is an algebraically closed field containing @xmath150 .
since both fields are algebraically closed of characteristic zero , this does not change the number of points .
however , it allows us to make some non - algebraic assumptions about the roots : specifically , we can consider polynomials @xmath4-fixed polynomials @xmath44 , where @xmath151 have distinct valuations . in this case
, we establish a bijection between @xmath82 and @xmath67 .
these methods are related to the asymptotic approach used by eremenko and gabrielov in @xcite , and build on the theory developed in @xcite . to prove theorem [ thm : promotion ]
, we consider the case where all @xmath22 roots of @xmath44 lie on the unit circle .
as the unit circle is @xmath0-equivalent to the real line , we can apply methods based on mukhin - tarasov - varchenko theorem .
the proof is essentially a straightforward interpretation of results proved in @xcite . from this point of view of this argument
, it is clear why the tableau need to be of rectangular shape in theorem [ thm : cyclicsieving ] : points corresponding to non - rectangular shaped tableaux lie in some proper richardson variety of @xmath16 ; the image under the wronski map is a compatible polynomial , which can not have @xmath22 roots on the unit circle .
the case of an involution is rather special .
there are a number of interpretations of the @xmath152-fixed points in the fibre .
for example , if we assume that @xmath44 is a real polynomial with only pure imaginary roots , then the @xmath152-fixed points are precisely the real points in the fibre , and theorem [ thm : ribbon ] states that the number of these in any compatible richardson variety @xmath43 is equal to the number of domino tableaux of shape @xmath48 .
alternatively may assume that the roots of @xmath44 are all real , and thereby deduce theorem [ thm : ribbon ] directly from results in @xcite .
we can also describe what happens when the roots of @xmath44 are not necessarily distinct .
another consequence in real algebraic geometry concerns maximally inflected real rational curves in projective space .
consider a real rational curve @xmath153 , @xmath27 $ ] where @xmath49}$ ] are linearly independent .
we say @xmath29 is * maximally inflected * if it has only simple , real ramification points .
equivalently for the point @xmath18 spanned by @xmath154 , @xmath21 has @xmath22 or @xmath51 distinct roots , all of which are real .
consider the space @xmath155 of all maximally inflected real rational curves @xmath153 of the above form .
the results used to prove theorem [ thm : promotion ] imply that the connected components of @xmath155 are in a one - or - two - to - one correspondence with the @xmath121-orbits on @xmath31 .
[ thm : mirrc ] let @xmath156 be the group of order two if @xmath13 is even , or the trivial group if @xmath13 is odd , and let @xmath157 be the element of order @xmath158 .
the connected components of @xmath155 are smooth , and are in bijection with the set @xmath159 of orbits of @xmath160 acting on @xmath161 .
if @xmath162 is a reflection , then the automorphism @xmath163 of @xmath155 descends to the action of @xmath164 on @xmath159 .
if @xmath165 is a reflection , then the automorphism @xmath166 of @xmath155 descends to the action of @xmath167 on @xmath159 . both theorems [ thm : ribbon ] and [ thm : promotion ] have extensions to actions of a dihedral group .
let @xmath168 be the dihedral subgroup of order @xmath125 generated by @xmath4 and the reflection @xmath169 .
a polynomial @xmath44 is @xmath126-fixed if it is of the form where @xmath170 , and @xmath171 is invariant under @xmath172 .
there are three distinct generic types : @xmath136 is odd , one of the @xmath173 is equal to @xmath174 , and the rest form generic inverse pairs ;
@xmath136 is even , all of the @xmath173 form generic inverse pairs ( none are equal to @xmath174 ) ; @xmath136 is even , one of the @xmath173 is equal to @xmath95 , another is equal to @xmath175 , and the rest form generic inverse pairs .
[ thm : d - ribbon ] let @xmath44 be a @xmath126-fixed polynomial of one the three generic types above , and let @xmath43 , be a compatible richardson variety . then @xmath67 is a reduced finite scheme .
moreover : if @xmath44 is of type ( 1 ) or ( 2 ) , the number of @xmath126-fixed points in @xmath176 is equal to @xmath177 . if @xmath44 is of type ( 3 ) , the number of @xmath126-fixed points in @xmath176 is equal to @xmath178 . in the case
where @xmath179 ( hence @xmath44 is of the form and @xmath87 ) , we also prove : [ thm : d - promotion ] let @xmath44 be a @xmath126-fixed polynomial of one of the three generic types , with @xmath180 .
if @xmath44 is of type ( 1 ) or ( 2 ) , the number of @xmath126-fixed points in @xmath47 is equal to the number tableaux in @xmath31 fixed by both @xmath116 and @xmath103 .
if @xmath44 is of type ( 1 ) or ( 3 ) , the number of @xmath126-fixed points in @xmath47 is equal to the number tableaux in @xmath31 fixed by both @xmath127 and @xmath103 .
putting theorems [ thm : d - ribbon ] and [ thm : d - promotion ] together , we deduce theorem [ thm : d - combinatorial ] . the main idea behind the proof of theorem [ thm : d - ribbon ] is to show that the action of @xmath169 on the fibre models the action of @xmath117 rotation on tableaux .
this follows fairly readily from the results we use to study @xmath4-fixed points . in the type ( 3 )
case , a small amount of additional analysis is needed , because we ca nt choose the @xmath173 to have distinct valuations .
theorem [ thm : d - promotion ] uses essentially the same ideas as theorem [ thm : promotion ] .
it is worth mentioning that every virtually every result in this paper should have an analogue for the orthogonal grassmannian @xmath181 that could be proved similarly using the results of @xcite .
in particular , theorems [ thm : combinatorial ] and [ thm : d - combinatorial ] should have analogues where tableaux are replaced by shifted tableaux of staircase shape . we have chosen not to check all of the details here .
an open question is whether it is possible to use our approach to prove analogues of theorems [ thm : combinatorial ] and [ thm : d - combinatorial ] for semistandard tableaux , such as ( * ? ? ?
* theorems 1.4 and 1.5 ) .
the geometry of this question is closely related the discussion in section [ sec : schubert ] , where fibre of the wronski map is replaced by a schubert intersection .
a more ambitious problem is to generalize thereom [ thm : ribbon ] to counting @xmath4-fixed points of these schubert intersections .
the paper is effectively divided into two parts , which can be read independently of each other .
the first part comprises sections [ sec : circle ] and [ sec : involution ] . in these two sections
, we will use arguments that apply in the case where @xmath44 has @xmath22 roots that lie on a circle in @xmath182 . for such @xmath44
, one can define a correspondence between @xmath31 and points in @xmath47 .
we review the important properties of this correspondence in section [ sec : circle ] , and use it to prove theorems [ thm : promotion ] , [ thm : mirrc ] and [ thm : d - promotion ] . in section [ sec : involution ] , we consider the special case of counting points in @xmath47 that are fixed by an involution in @xmath0 .
we study this in a variety of ways , which include giving different interpretations of the problem , and giving alternate proofs of some of our main results .
more generally we explain how to count involution - fixed points in certain intersections of schubert varieties .
sections [ sec : fixedpoints ] , [ sec : ribbon ] and [ sec : dihedral ] form the second part of the paper .
the focus in these three sections is on understanding the restriction of the wronski map to the fixed point scheme @xmath65 , explicitly in terms of the plcker coordinates . in section [ sec : fixedpoints ] , we describe the components of @xmath65 , their coordinate rings in terms of plcker coordinates , and their initial ideals with respect to certain weight functions . in section [ sec : ribbon ] , we describe a very general way to associate a tableau to any point in @xmath16 , working over a valuation ring . using this association ,
we prove theorem [ thm : ribbon ] . in section [ sec : dihedral ] , we extend our results to dihedral group actions and prove theorem [ thm : d - ribbon ] .
this work began with discussions at the october 2010 aim workshop on algebraic systems with only real solutions .
i am grateful to all the organizers , staff and participants who made the workshop possible .
i particularly wish to thank alexandre eremenko , frank sottile and alex yong for the conversations that got everything started , and brendon rhoades for valuable correspondence and corrections .
this research was partially supported by an nserc discovery grant .
in the case where the roots of @xmath44 all lie on a circle in @xmath182 , points in the fibre of the wronski map @xmath47 can be identified with standard young tableaux of shape @xmath32 . moreover , any element of @xmath0 that fixes @xmath44 can be identified with an operator on @xmath31 .
we begin with a review these results , wherein our circle is just @xmath183 .
the proofs may be found in @xcite , but our presentation here more closely follows @xcite .
let @xmath184 be a multiset , where @xmath185 . following the notation of @xcite
, we will think of a polynomial in terms of its roots , and write @xmath186 , where @xmath187 suppose @xmath188 is a set ( i.e. @xmath189 for @xmath190 ) , and let @xmath191 .
we wish to associate to @xmath23 , a tableau @xmath192 . to accomplish this
, we consider the total order @xmath193 on @xmath194 , defined by @xmath195 , if either @xmath196 , @xmath197 , or @xmath198 .
since @xmath188 is a set , we may assume without loss of generality , that @xmath199 now , for @xmath200 and @xmath201 $ ] , let @xmath202 for all @xmath203 $ ] , @xmath204 is a set and hence by theorem [ thm : mtv ] , the fibres @xmath205 are reduced .
it follows that there is unique ( continuous ) lifting of the path @xmath204 , @xmath201 $ ] to a path @xmath206 , where @xmath207 .
now , since @xmath208 with multiplicity @xmath118 , and @xmath209 we have @xmath210 . by proposition [ prop : richardson ]
, @xmath211 lies in a unique richardson variety @xmath212 , where @xmath213 .
the sequence @xmath214 is a chain of partitions @xmath215 moreover no two points of @xmath216 give rise to the same chain .
we define @xmath192 to be the tableau that encodes the chain of partitions @xmath217 :
@xmath218 has entry @xmath118 in the box corresponding to @xmath219 .
informally , we are defining @xmath218 by moving the point @xmath23 so that @xmath118 smallest roots of @xmath21 are sent to zero , and seeing where @xmath23 ends up .
alternatively we can obtain @xmath218 by sending the @xmath220 largest roots to @xmath221 . for @xmath200 and @xmath201 $ ] , let @xmath222 and lift to @xmath223 where @xmath224 . in this case , since @xmath225 with multiplicity @xmath220 , we have @xmath226 , and hence @xmath227 lies in a unique richardson variety @xmath228 , where @xmath229 .
it turns out that @xmath230 is equal to the partition @xmath231 constructed above , and hence both constructions yield the same tableau .
( this can be seen from a third construction in which one considers @xmath232 the lifted path @xmath233 ( with @xmath234 ) is compatible with the previous two , in that we must have @xmath235 .
hence this richardson variety is non - empty , which implies @xmath236 .
see ( * ? ? ?
* section 6 ) for details . ) this definition of @xmath218 was based on the idea of lifting a path of subsets of @xmath194 to a path in @xmath16 . more generally , if @xmath237 , @xmath201 $ ] are paths in in @xmath194 and @xmath238 has the property that @xmath239 is reduced for all @xmath240 , we may consider lifting @xmath241 to a path @xmath242 .
we can understand such liftings in terms of the associated tableaux . in the case where @xmath243 remain in the same @xmath193-order ,
the tableau associated to @xmath244 remains constant .
[ thm : sameorder ] suppose @xmath245 for all @xmath201 $ ] . if @xmath242 is a lifting of @xmath241 to @xmath16 , then @xmath246 .
the relationship between @xmath247 and @xmath248 is more interesting if @xmath237 do not remain in the same @xmath193-order .
first note that by theorem [ thm : mtv ] , the relation between @xmath249 and @xmath250 ( and hence between @xmath247 and @xmath248 ) depends only on the homotopy class of the path @xmath241 .
it suffices , therefore , to describe this relationship in the simplest possible case : @xmath251$}\ , , \\
( a_k)_0 \prec ( a_{k+1})_0 \quad \text{and } \quad ( a_{k+1})_1 \prec ( a_k)_1\ , .
\end{gathered}\ ] ] any path is homotopic to a concatenation of paths of the form , so the answer in this simple case gives a complete description of the relationship between @xmath247 and @xmath248 in general .
[ thm : sliding ] suppose that @xmath241 is a path of the form , and @xmath242 is a lifting to @xmath16 .
if @xmath252 and @xmath253 have the same sign , then @xmath246
. if @xmath118 and @xmath254 are in the same row or column of @xmath247 then @xmath246 .
if @xmath252 and @xmath253 have opposite signs , and @xmath118 and @xmath254 are in different rows and columns of @xmath247 , then @xmath248 is obtained from @xmath247 by switching the positions of @xmath118 and @xmath254 .
we will mainly be interested in the case where @xmath252 and @xmath253 have opposite signs .
to see what is going on here , it is helpful to think of replacing the entries @xmath255 of the tableau associated to a point @xmath191 with real entries from the set @xmath188 .
specifically , we will replace @xmath118 by @xmath256 , where @xmath257 .
we refer to the resulting object as a * real valued tableau * , and write @xmath258 for the real valued tableau associated to @xmath23 . in case ( ii )
, the tableau @xmath218 does not change ; hence in the real valued tableau @xmath259 the relative @xmath193-order of entries does not change .
but since the relative order of @xmath260 and @xmath261 does change , these must switch places in @xmath259 . in case ( iii )
, the positions of @xmath118 and @xmath262 are changing along with the relative order of @xmath260 and @xmath261 .
in the real valued tableau , these two changes nullify each other , and the entries @xmath260 and @xmath261 remain where they are in @xmath259 . in both cases , this is a very small instance of a schtzenberger slide .
the box containing entry @xmath260 is sliding through the subtableau consisting of the single box @xmath261 . as the next example illustrates , the big picture emerges by considering what happens when a negative number passes by several positive numbers .
[ ex : promotion ] consider a path @xmath263 , where @xmath264 , @xmath265 , and the path @xmath266 traverses the negative real numbers .
let @xmath242 be a lifting such that @xmath249 is associated to the real valued tableau @xmath267[auto = promotion ] 0 & 3 & 19 \\ 8 & 12 & 27 \\ 21 & 34 & 39 \end{young}\,\,.\ ] ] as @xmath266 passes each of the other entries , the two entries switch places if and only if they are in the same row or column .
@xmath268[auto = promotion ] ? 0 & 3 & 19 \\ 8 & 12 & 27 \\ 21 & 34 & 39 \end{young } }
\quad \to \quad { \begin{young}[c][auto = promotion ] 3 & ?
-5 & 19 \\
8 & 12 & 27 \\ 21 & 34 & 39 \end{young } } \quad \to \quad { \begin{young}[c][auto = promotion ] 3 & ?
-10 & 19 \\ 8 & 12 & 27 \\ 21 & 34 & 39 \end{young } } \quad \to \\[2ex ] \to \quad & { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & ?
-15 & 27 \\ 21 & 34 & 39 \end{young } } \quad \to \quad { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & ?
-20 & 27 \\ 21 & 34 & 39 \end{young } } \quad \to \quad { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & ?
-25 & 27 \\ 21 & 34 & 39 \end{young } } \quad \to\\[2ex ] \to \quad & { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & 27 & ?
-30 \\ 21 & 34 & 39 \end{young } } \quad \to \quad { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & 27 & ?
-35 \\ 21 & 34 & 39 \end{young } } \quad \to \quad { \begin{young}[c][auto = promotion ] 3 & 12 & 19 \\ 8 & 27 & 39 \\ 21 & 34 & ?
\infty \end{young}}\end{aligned}\ ] ] comparing the steps here with the example in figure [ fig : promotionex ] , we see that @xmath269 , where @xmath121 is the promotion operator on tableaux
. in general , we can model both the promotion and evacuation operators on @xmath31 , in terms of lifting a suitable path @xmath241 that begins at a set of positive real numbers : @xmath270 we will describe promotion as the lifting of a loop @xmath241 , @xmath201 $ ] . let @xmath266 be a path that first decreases from @xmath271 to @xmath272 , then moves along the _ negative _ real axis to @xmath221 , and finally decreases from @xmath221 to the value @xmath273 .
meanwhile for @xmath274 , let @xmath260 move monotonically from @xmath252 to @xmath275 .
thus @xmath241 is a loop that cyclically rotates the elements @xmath276 .
( see figure [ fig : modelpromote ] . ) in terms of the @xmath193-order , @xmath266 is changing places with each of @xmath260 , @xmath277 , and doing so with opposite signs .
thus , in the real valued tableau associated to a lifting , the box containing @xmath266 begins in the northwest corner , and slides through all the other entries , ending up in the southeast corner .
hence we see that for any lifting @xmath242 of such a loop , we have @xmath269 . , drawn with @xmath272 south , @xmath221 north , @xmath95 east , and @xmath175 west . ] evacuation , on the other hand ,
is described by lifting a path @xmath241 , @xmath201 $ ] , for which @xmath278 . for @xmath279 , let @xmath260 be a path that begins at @xmath252 , decreases to @xmath272 , and continues moving monotonically until it reaches @xmath280 .
( see figure [ fig : modelevac ] . )
this describes a unique homotopy class of paths . to show this homotopy class models evacuation , it is convenient to consider a specific representative , which we will describe by dividing up the interval @xmath281 $ ] into @xmath22 subintervals @xmath282 $ ] , @xmath283 . on the first interval
@xmath284 $ ] , @xmath266 moves monotonically to @xmath285 .
meanwhile for @xmath286 , @xmath260 moves monotonically to @xmath287 .
hence @xmath266 has changed places in the @xmath193-order with each of the other @xmath260 , with opposite signs ; this models the promotion operator @xmath288 . on the next interval
@xmath289 $ ] , @xmath290 remains constant , while @xmath291 moves monotonically to @xmath292 , and for @xmath293 , @xmath260 moves monotonically to @xmath294 .
hence @xmath291 is changing places in the @xmath193-order , with each @xmath260 , @xmath295 , but does not pass @xmath266 .
this models the promotion operator restricted to the @xmath51 smallest entries of the tableau , which is @xmath296 . in general , on the @xmath78@xmath297interval
@xmath282 $ ] , @xmath298 have reached their final positions , and @xmath299 moves to @xmath300 , while @xmath260 moves to @xmath301 ; this models @xmath110 . hence the complete path models @xmath302 ; that is , for any lifting @xmath242 , we have @xmath303 . from this perspective
it is clear why @xmath116 is an involution : if we reverse the roles of the positive and negative real numbers , we see that the reverse of a path that models @xmath116 also models @xmath116 , hence @xmath304 .
the group @xmath0 acts on @xmath182 by mbius transformations . if @xmath306 and @xmath307 , then @xmath308 this action restricts to an action of @xmath305 on @xmath194 .
let @xmath184 be a subset of @xmath194 , let @xmath191 and let @xmath309 .
since the wronski map is @xmath0-equivariant , we have @xmath310 , where @xmath311 is also a subset of @xmath194 .
this defines an action on real valued tableaux
. if @xmath312 is the real valued tableau associated to @xmath23 , we define @xmath313 if @xmath59 is in the connected component of @xmath314 , we can compute this action explicitly , using theorem [ thm : sliding ] .
let @xmath315 , @xmath201 $ ] , be any path in @xmath305 from @xmath316 to @xmath317 .
then @xmath318 is a lifting of the path @xmath319 .
theorem [ thm : sliding ] therefore tells us exactly how the tableau @xmath320 is related to @xmath321 .
if @xmath59 is not in the connected component of the identity , then we can write @xmath322 , or @xmath323 , where @xmath324 are in the connected component of the identity matrix .
the next lemma tells us how to compute the actions of @xmath325 and @xmath169 .
[ lem : tworeflections ] suppose that latexmath:[\[\label{eqn : absincreasing } valued tableau with entries from the set @xmath188 .
@xmath327 , where @xmath328 is obtained from @xmath96 by replacing @xmath256 with @xmath329 , for @xmath279 .
@xmath330 , where @xmath331 is obtained from @xmath96 by rotating @xmath117 and replacing @xmath256 with @xmath332 , for @xmath279 . if we have a set @xmath188 for which @xmath333 for some @xmath190 , then lemma [ lem : tworeflections ] can not be used directly .
the problem is that if we replace by the weaker condition @xmath257 , then @xmath328 and @xmath331 may not be real valued tableaux : the operation @xmath334 is not @xmath193-order preserving , and @xmath335 is not @xmath193-order reversing . in this case
, we can still compute @xmath336 and @xmath337 , but we need to do so by perturbing the point @xmath188 , which introduces additional sliding .
we consider examples of this in section [ sec : involution ] .
let @xmath191 , and @xmath338 .
let @xmath339 .
recall from section [ sec : lifting ] that the tableaux @xmath218 is computed by lifting a path @xmath204 ( see ) to a path @xmath340 .
similarly the tableau @xmath341 is computed by lifting a path @xmath342 to a path @xmath343 .
now , since @xmath257 and @xmath344 , @xmath345 for all @xmath200 , @xmath201 $ ] , and therefore @xmath346 . finally , since @xmath325 fixes every richardson variety , @xmath347 if and only if @xmath348 , and we see that @xmath349 , which is equivalent to ( i ) . for ( ii )
, the argument is similar .
this time , let @xmath350 , and @xmath351 . since , @xmath352 , the path @xmath353 is equal to @xmath354 , where @xmath355 is the path that appears in the _ alternate _ definition of @xmath218 .
thus we obtain @xmath356 , and deduce the result from the fact that @xmath357 .
now suppose that @xmath358 for some set @xmath188 . in this case , the real valued tableau @xmath359 has the same entries as @xmath258 . by identifying real valued tableaux with entries from the set @xmath188 with @xmath31
, we can regard @xmath59 as an operator on @xmath31 .
we note that this identification is always defined with respect to a particular choice of @xmath59-fixed set @xmath188 , and that changing this set will change the identification .
we now show that if @xmath360 are positive real numbers , then the operator defined by @xmath59 belongs to the dihedral subgroup @xmath361 generated by the promotion and evacuation operators @xmath362 .
moreover , for a suitable choice of @xmath188 , we will realize @xmath121 and @xmath116 by elements of @xmath305 .
from here it will be a simple matter to prove theorems [ thm : promotion ] and [ thm : d - promotion ] .
[ lem : fixeddihedral ] let @xmath309 , and suppose that @xmath358 , where @xmath363 .
if @xmath59 is in the connected component of @xmath314 then the action of @xmath59 on @xmath31 coincides with @xmath364 for some integer @xmath118 . if @xmath59 is not in the connected component of @xmath314 , then the action of @xmath59 on @xmath31 coincides with @xmath365 , for some integer @xmath118 .
if @xmath59 is in the connected component of the identity , then @xmath366 is defined by lifting a path @xmath319 , @xmath201 $ ] .
such a path must cyclically rotate the elements of @xmath188 by some amount ; i.e. there is some @xmath118 such that @xmath367 .
then @xmath319 is homotopic to the path that models @xmath368 , and hence @xmath369 .
now suppose @xmath59 is not in the connected component of the identity .
write @xmath370 , and consider the action of @xmath371 on real valued tableau .
the path @xmath372 used to define this action preserves cyclic order , and @xmath373 .
hence @xmath372 must homotopic to a path that cyclically rotates the elements of @xmath188 by some amount , followed by a path that models evacuation .
in other words , if @xmath218 is the tableau corresponding to a point @xmath191 , then the tableau corresponding to @xmath374 is @xmath375 , for some @xmath118 . by lemma [ lem : tworeflections](i ) , this is also the tableau corresponding to @xmath376 . in order to explicitly realize @xmath121 and @xmath116 by elements of @xmath305 and relate these to the statements of theorems [ thm : promotion ] and [ thm : d - promotion ] , it will be convenient work with the unit circle @xmath377 , rather than @xmath194 .
to relate the two , we will use a mbius transformation that maps `` most '' of the unit circle to the positive real numbers .
let @xmath378 where @xmath379 , and @xmath380 then @xmath381 defines a mbius transformation which sends @xmath382 to @xmath194 , with @xmath383 , @xmath384 and @xmath385 .
we let @xmath386 denote the subgroup of @xmath0 that fixes the unit circle . for @xmath279 ,
let @xmath387 .
since @xmath360 lie on the long arc of the unit circle between @xmath388 and @xmath389 , we have @xmath390 for each point @xmath191 , we define the tableau @xmath192 associated to @xmath23 to be @xmath391 . using this identification , any @xmath392 that fixes @xmath188 defines an operator on @xmath31 . [
lem : evacpromote ] with @xmath360 as above , the @xmath393 group elements @xmath394 fix @xmath188 , and act on @xmath31 as @xmath121 and @xmath116 respectively .
together with lemma [ lem : tworeflections](ii ) , lemma [ lem : evacpromote ] shows that for @xmath102 , @xmath112 is the tableau obtained by rotating @xmath96 by @xmath117 and replacing each entry @xmath118 by @xmath119 .
this is well known , but not entirely obvious .
we argue as in the proof of lemma [ lem : fixeddihedral ] .
that both of these mbius transformations fix @xmath188 is easy to check .
in particular , @xmath395 , which implies that @xmath59 acts as @xmath396 . for the action of @xmath169 , note that @xmath169 commutes with @xmath381 , so it is enough to show that @xmath169 acts as @xmath116 on real valued tableaux with entries @xmath397 .
but now , @xmath398 , and the action of @xmath399 on @xmath397 is computed by lifting the path @xmath400 , @xmath201 $ ] , which models evacuation .
we are now in a position to prove theorems [ thm : promotion ] and [ thm : d - promotion ] .
let @xmath401 .
then @xmath44 is a @xmath4-fixed polynomial , and the fibre @xmath402 is reduced .
therefore the number of @xmath4-fixed points in @xmath216 is the generic answer .
a point @xmath191 is @xmath4-fixed , if and only if the corresponding tableau @xmath218 is @xmath4-fixed ; but
by lemma [ lem : evacpromote ] , the generator of @xmath4 acts as @xmath103 on @xmath31 . hence the @xmath4-fixed points in @xmath216 are in bijection with @xmath103-fixed tableaux in @xmath31 . case ( i ) is similar to the proof of theorem [ thm : promotion ] .
the polynomial @xmath401 is fixed by @xmath4 and also by @xmath169 .
moreover , @xmath403 , so @xmath44 is of type ( 1 ) or ( 2 ) , depending on whether @xmath8 is odd or even . by lemma
[ lem : evacpromote ] , the generators of @xmath126 act as @xmath103 and @xmath116 , and hence the @xmath126-fixed points of @xmath216 are in bijection with tableaux in @xmath31 fixed by @xmath103 and @xmath116 . for case ( ii ) , we need to consider a different polynomial .
let @xmath404 .
since @xmath405 , this is of type ( 1 ) or ( 3 ) , depending on whether @xmath8 is odd or even .
let @xmath406 then @xmath407 , so @xmath408 .
a point @xmath409 is fixed by @xmath126 if and only if @xmath410 is fixed by @xmath411 , i.e. by @xmath412 and @xmath413 by lemma [ lem : evacpromote ] , the tableaux corresponding to points @xmath414 fixed by these two group elements are those fixed by @xmath103 and @xmath127 . as an application of the results from this section , we will briefly discuss the classification of maximally inflected real rational curves .
let @xmath153 be a real parameterized curve @xmath415\,,\ ] ] with @xmath49}$ ] linearly independent .
a * ramification point * of @xmath29 is a point @xmath416 at which @xmath417 do not span @xmath418 . when @xmath419 , a ramification point is a critical point of the rational function @xmath420 .
when @xmath421 , @xmath29 is a planar curve and the simplest example of a real ramification point of @xmath29 is an inflection point . in general ,
if @xmath422 denotes the vector space spanned by @xmath154 , it is not hard to see that a ramification point of @xmath29 is a root of @xmath423 ; hence @xmath29 can have at most @xmath22 real ramification points .
the space @xmath155 of maximally inflected real rational curves is the algebraic set of all curves @xmath29 of this form with exactly @xmath22 distinct , real ramification points .
these curves and their relationship to theorem [ thm : mtv ] were first studied in @xcite .
one interpretation of theorem [ thm : mirrc ] is that the different `` types '' of curves in @xmath155 correspond to the @xmath424-orbits on @xmath31 , where @xmath424 is the dihedral group of order @xmath123 generated by @xmath121 and @xmath116 . here , two curves are of the same type if one is obtained from the other by deforming continuously within @xmath155 , and possibly composing with a reflection of @xmath194 or @xmath425 ( or both ) .
the actual statement of theorem [ thm : mirrc ] is about a finer invariant than this notion of type , in which one does not identify curves related by reflections .
on the other hand , curves of different types are fundamentally different because they degenerate in different ways when the ramification points collide ( see @xcite ) ; hence the type of a curve is the coarsest invariant one wants for any classification .
the analysis of the components of @xmath155 is slightly different depending on whether @xmath13 is odd or even .
this is because @xmath425 is orientable if @xmath13 is even , and non - orientable if @xmath13 is odd .
when @xmath13 is even , we put @xmath426 , and @xmath427
. let @xmath428 denote the right hand side of .
although the coordiates of @xmath429 are only defined up to a real scalar multiple , since @xmath13 is even , the sign @xmath428 is unchanged by any real rescaling of @xmath430 $ ] .
we can therefore define the sign of @xmath29 to be @xmath431 if @xmath432 , then @xmath433 is just the sign of @xmath434 .
when @xmath13 is odd there is no well - defined notion of the sign of a curve : we put @xmath435 , @xmath436 , and @xmath437 for every @xmath438 .
the group @xmath305 acts on @xmath155 : if @xmath309 , then @xmath439 $ ] is well - defined .
let @xmath440 denote the connected component of the identity element .
it is clear that if @xmath441 then @xmath29 and @xmath442 are in the same connected component of @xmath155 .
we also have an action of @xmath443 on @xmath155 , given by @xmath163 for @xmath162 .
if @xmath444 is in the connected component of the idenity element then @xmath29 and @xmath445 are in the same component of @xmath155 . otherwise , we have @xmath446 and @xmath447 .
we now associate to each @xmath438 a discrete invariant @xmath448 , which will be a subset of @xmath161 .
let @xmath449 , and @xmath450 we define @xmath448 to be the set of pairs @xmath451 our next goal is to show that @xmath448 has the properties needed to prove theorem [ thm : mirrc ] . [
lem : orbitinvariant1 ] for every curve @xmath438 , @xmath452 ; i.e. @xmath448 is an orbit of @xmath160 acting on @xmath161 .
let @xmath449 , and @xmath453
. then @xmath454 , where @xmath188 is a set of the form @xmath455 .
we must show that @xmath448 is equal to the @xmath160-orbit of @xmath456 . for any other @xmath457
, we can write @xmath458 where @xmath459 . since @xmath460 we must have @xmath461 for some integer @xmath118 . as in the proof of lemma [ lem : fixeddihedral ] , this implies that @xmath462 .
if @xmath13 is even , the sign of @xmath428 changes each time we pass a root ; hence we also have @xmath463 .
thus @xmath448 is contained in the @xmath160-orbit of @xmath456 . for the reverse inclusion
, choose any @xmath459 such that @xmath464 if @xmath458 , then @xmath457 and @xmath465 .
thus @xmath448 contains the @xmath160-orbit of @xmath456 .
[ lem : orbitinvariant2 ] for every orbit @xmath466 , there exists a curve @xmath29 such that @xmath467 .
choose any @xmath468 , any @xmath469 , any @xmath191 such that @xmath470 , and any @xmath29 such that @xmath471 .
if @xmath472 , then replace @xmath29 by @xmath445 for any reflection @xmath162 .
then @xmath473 , and thus by lemma [ lem : orbitinvariant1 ] , @xmath467 .
[ lem : orbitinvariant3 ] two curves @xmath29 and @xmath474 are in the same connected component of @xmath155 if and only if @xmath475 .
suppose @xmath29 , @xmath474 are in the same connected component of @xmath155 .
let @xmath476 , @xmath201 $ ] , be a path in @xmath155 from @xmath477 to @xmath478 .
let @xmath479 .
there exists a continuous path @xmath315 in @xmath480 such that @xmath481 . by theorem [ thm : sameorder ]
, @xmath482 is constant for all @xmath201 $ ] . since @xmath483 for all @xmath201 $ ] , @xmath484 is also constant .
thus we have @xmath485 , which implies @xmath475 .
conversely , suppose that @xmath475 .
let @xmath449 , @xmath486 .
choose any @xmath473 , and any @xmath453 , @xmath487 , so that @xmath488 choose a path of sets @xmath489 such that @xmath490 @xmath491 , and let @xmath242 be a lifting of @xmath241 such that @xmath492 . by theorem [ thm : sameorder ] , @xmath493 , which shows that @xmath494 .
lifting @xmath244 to a path in @xmath155 , it follows that @xmath442 is in the same component as @xmath495 for some @xmath162 .
but since @xmath496 , @xmath444 must be in the connected component of the identity element .
it follows that @xmath29 and @xmath474 are in the same connected component of @xmath155 . for ( i ) ,
lemmas [ lem : orbitinvariant1 ] , [ lem : orbitinvariant2 ] and [ lem : orbitinvariant3 ] imply that @xmath497 descends to a bijection from the connected components of @xmath155 to the set @xmath498 .
to see @xmath155 is smooth it is enough to show that @xmath499 is smooth .
but this follows from the fact that @xmath16 is smooth at @xmath449 , and the fibre @xmath500 is reduced .
for any reflection @xmath501 we have @xmath502 and @xmath446 ; statement ( ii ) follows .
it is enough to prove ( iii ) for the reflection @xmath503 .
let @xmath504 , @xmath505 . if @xmath453 , then @xmath506 .
by lemma [ lem : tworeflections](i ) , @xmath507 . to compute the sign @xmath508 , note that there is a path @xmath476 from @xmath509 to @xmath510 such that @xmath483 for all @xmath240 .
thus the sign of @xmath509 is the same as that of @xmath510 , which is @xmath511 .
thus @xmath512 , as required .
suppose @xmath513 is an involution ( other than the identity element ) and let @xmath34})}$ ] be a @xmath59-fixed polynomial .
we now consider the problem of counting the @xmath59-fixed points in @xmath47 .
this is , of course , a special case of our more general problem ; however , it is an interesting one because there are several interpretations of its various ingredients .
for example , special choices of @xmath44 and @xmath59 lead to different interpretations of the problem . also , @xmath59 generates a group of order @xmath80 , which we can think of as either the cyclic group @xmath152 , or the dihedral group @xmath514 .
hence all four of our theorems about fixed points ( theorems [ thm : ribbon ] , [ thm : promotion ] , [ thm : d - ribbon ] and [ thm : d - promotion ] ) will have something to say about this case . unlike the general case , we can explicitly see how the different combinatorial objects that arise are related to each other .
finally , our results for involutions extend to the case where the roots of @xmath44 are non - distinct .
before we begin our investigations , we should note that there will be different cases to consider .
suppose @xmath515 .
for now we will assume that @xmath184 is subset of @xmath182 such that @xmath358 , and also that @xmath216 is reduced .
there are exactly two points in @xmath182 that are fixed by @xmath59 .
the number of non - fixed points of @xmath59 in @xmath188 must be even .
thus we have three cases : @xmath22 is odd , and the set @xmath188 contains exactly one @xmath59-fixed point of @xmath182 .
@xmath22 is even , and the set @xmath188 contains no @xmath59-fixed points of @xmath182 .
@xmath22 is even , and the set @xmath188 contains both @xmath59-fixed points @xmath182 .
when @xmath516 , these three cases correspond to the three generic types for @xmath44 discussed in the introduction .
they are fundamentally different , and we should expect to see different behaviours in each one . on the other hand ,
the answer only depends on which of the three cases we are in .
[ lem : involutioncases ] for any @xmath188 and @xmath59 as above , the number of @xmath59-fixed points in @xmath216 depends only on the number of @xmath59-fixed elements of the set @xmath188 . let @xmath24})}^\phi$ ] denote the @xmath59-fixed polynomials in @xmath24})}$ ] .
let @xmath517 denote the two @xmath59-fixed points .
the @xmath59-fixed point scheme of @xmath16 is flat and finite over @xmath24})}^\phi$ ] .
hence the number of points in the reduced fibre @xmath216 depends only on the component of of @xmath24})}^\phi$ ] that @xmath44 lies in .
it is not hard to see that @xmath24})}^\phi$ ] has two exactly components .
if @xmath22 is odd , one component is the closure of the set of @xmath44 identified with @xmath518 ; the other is the closure of the set of all @xmath44 identified with @xmath519 .
these two components are isomorphic any involution @xmath381 commuting with @xmath59 such that @xmath520 gives an isomorphism so we obtain the same answer for both components
. if @xmath22 is even , the two components are non - isomorphic and correspond to cases ( 2 ) and ( 3 ) .
we now describe a few of the different interpretations of the problem .
call @xmath18 an * even - odd * point if @xmath23 has a basis @xmath154 where each @xmath521 is either an even polynomial or an odd polynomial .
[ prop : evenodd ] let @xmath44 be an even or an odd polynomial @xmath22 with distinct roots .
the even - odd points in @xmath47 are exactly the the @xmath325-fixed points .
an even - odd point in @xmath16 is clearly @xmath325-fixed .
conversely , any @xmath325-fixed subspace of @xmath15}$ ] has a basis of @xmath325-eigenvectors , and hence is an even - odd point .
[ prop : imaginary ] let @xmath44 be a real polynomial whose roots are all pure imaginary numbers .
then the real points in @xmath47 are exactly the @xmath325-fixed points .
let @xmath522 .
we claim that @xmath409 is real point if and only if @xmath523 is an even - odd point .
the result then follows from the fact that @xmath381 commutes with @xmath325 .
the polynomial @xmath524 has real roots , and hence by theorem [ thm : mtv ] every point @xmath525 is real .
suppose @xmath523 has basis @xmath154 where each @xmath526 is either an even polynomial or an odd polynomial .
then @xmath23 has basis @xmath527 , each of which is either real , or @xmath78 times a real polynomial .
hence @xmath23 is a real point .
conversely , suppose @xmath527 are real polynomials that form a basis for @xmath409 .
since @xmath154 span a real vector space @xmath525 , so do their complex conjugates .
since @xmath528 is real , @xmath529 , and so we deduce that @xmath530 span @xmath523 .
thus @xmath523 is spanned by @xmath531 , which are odd or even polynomials , i.e. @xmath523 is an even - odd point .
[ prop : onsomecircle ] let @xmath44 be a real polynomial with @xmath22 roots lying on a circle @xmath532 that is symmetrical with respect to the real axis .
there is an involution @xmath513 that restricts to complex conjugation on the circle @xmath532 .
the real points in @xmath47 are the @xmath59-fixed points .
hence , by lemma [ lem : involutioncases ] the number of real points in @xmath47 depends only the number of real roots of @xmath44 . if @xmath533 is the unit circle , then this involution is @xmath516 .
there is a mbius transformation @xmath165 that sends the imaginary line to @xmath532 .
the involution @xmath534 restricts to complex conjugation on @xmath532 . since @xmath409 is real and @xmath59-fixed if and only if @xmath535 is real and @xmath325-fixed , this result is equivalent to proposition [ prop : imaginary ] . [
prop : evenreal ] suppose @xmath22 is even .
let @xmath44 be a real , even polynomial with @xmath22 distinct non - real roots on the unit circle .
then the number of even - odd points in @xmath47 is equal to the number of real points in @xmath47 .
we consider two involutions : @xmath325 , whose fixed points in @xmath182 are @xmath272 , @xmath221 , and @xmath169 , whose fixed points in @xmath182 are @xmath174 .
if @xmath515 then the set @xmath188 is fixed by both @xmath325 and @xmath169 , and none of the elements of @xmath188 are fixed by either .
it follows from lemma [ lem : involutioncases ] that the number of @xmath325-fixed points in @xmath47 is equal to the number @xmath169-fixed points in @xmath47 .
but the former are even - odd points , and the latter are real points .
this last result has an interesting combinatorial interpretation .
by lemma [ lem : evacpromote ] the real points in @xmath47 correspond to @xmath116-fixed tableaux in @xmath31 .
note that @xmath116-fixed tableaux are rotationally - invariant tableaux , so this statement is in agreement with both theorems [ thm : d - ribbon ] and [ thm : d - promotion ] ( here we are considering the case where @xmath68 , and @xmath44 is of type ( 2 ) ) .
on the other hand , by theorem [ thm : promotion ] , the odd - even points correspond to @xmath536-fixed tableaux . hence from proposition [
prop : evenreal ] , we deduce : [ cor : combinv1 ] if @xmath22 is even , the number of @xmath116-fixed tableaux in @xmath31 is equal to the number the number of @xmath536-fixed tableaux in @xmath31 .
the @xmath5 case of theorem [ thm : ribbon ] asserts that the fixed points of an involution are also counted domino tableaux .
we can prove the @xmath5 case of theorem [ thm : ribbon ] using lemma [ lem : evacpromote ] .
[ thm : specialribbon ] let @xmath44 be an even or an odd polynomial of degree @xmath22 or @xmath51 , with distinct real roots . there is a unique @xmath44-compatible richardson variety @xmath43 , and the number of even - odd points in @xmath176 is the number of domino tableaux of shape @xmath48 . here the involution we use is @xmath325 , which is the generator for @xmath152 .
the four possibilities for @xmath48 are : @xmath32 ( if @xmath44 is even and @xmath22 is even ) , @xmath537 ( if @xmath44 is odd and @xmath22 is odd ) , @xmath538 ( if @xmath44 is even and @xmath22 is odd ) , @xmath539 ( if @xmath44 is odd and @xmath22 is even ) .
the two cases where @xmath22 is odd give the same answer ; here @xmath188 contains either @xmath272 or @xmath221 , but not both .
when @xmath22 is even and @xmath87 , we are in the case where @xmath188 contains neither @xmath272 nor @xmath221 .
when @xmath22 is even and @xmath540 , @xmath188 contains both @xmath272 and @xmath221 .
hence these cases correspond exactly to the different possibilities for the number of @xmath325-fixed elements in @xmath188 .
first , suppose that @xmath87 .
then we can write @xmath184 , where @xmath541 , and @xmath542 for all @xmath118 . let @xmath312 be the real valued tableau associated to a point @xmath191 . to compute @xmath543
, we write @xmath544 where @xmath545 is a small positive real number
. then @xmath546 .
note that @xmath547 by theorem [ thm : sliding ] , @xmath548 is obtained from @xmath96 by switching the positions of @xmath549 and @xmath542 if they are in the same row and column ; then adding @xmath545 to each of the entries .
the action of @xmath325 on the result simply negates each of the entries by lemma [ lem : tworeflections](i ) ; finally acting by @xmath550 the second time does not change the @xmath193-order of the entries , so we simply add @xmath545 to each entry .
the net result is that @xmath336 is obtained from @xmath96 by switching the positions of @xmath549 and @xmath551 , whenever these are _ not _ in the same row or column .
it follows that @xmath552 if and only if the corresponding tableau @xmath258 has the property that @xmath549 and @xmath551 are in the same row or column for all @xmath553 .
these real valued tableaux are in bijection with domino tableaux of shape @xmath32 . in the other three cases , @xmath188 is of the form @xmath554[c]{$\{0 , a_1 , \dots , a_{n-1}\}$ } & \qquad&\text{if } \lambda/\mu = { { \mathchoice{\raisebox{.05ex}{$\sqsubset\!\!\sqsupset$ } } { \raisebox{.05ex}{$\sqsubset\!\!\sqsupset$ } } { \sqsubset\!\!\sqsupset } { \sqsubset\!\!\sqsupset } } } /1\ , , \\ & & & \makebox[\wd\mysavebox][c]{$\{a_1 , \dots , a_{n-1 } , \infty\}$ } & \qquad&\text{if }
\lambda/\mu = 1^\vee\ , , \\ & \text{or}&\qquad & \usebox{\mysavebox } & \qquad&\text{if } \lambda/\mu = 1^\vee/1\,,\end{aligned}\ ] ] where @xmath555 for all @xmath118 .
a similar argument applies .
[ cor : combinv2 ] if @xmath22 is even , the number of rotationally - invariant tableaux in @xmath31 is equal to the number of domino tableaux of shape @xmath32 .
a similar argument gives the @xmath68 case of theorem [ thm : d - ribbon ] .
[ thm : speciald - ribbon ] let @xmath44 be polynomial of degree @xmath22 , whose roots are distinct real numbers invariant under @xmath172 . if @xmath556 or @xmath403 , then the the number of @xmath169-fixed points in @xmath47 is the number of rotationally - tableaux in @xmath31 . if @xmath557 , then the number of @xmath169-fixed points in @xmath47 is the number of rotationally - invariant tableaux in @xmath31 where @xmath558 and @xmath559 are in the same row or column
write @xmath184 , where @xmath257 .
let @xmath191 , with corresponding real valued tableau @xmath312 it is enough to prove this when @xmath44 is generic , so that @xmath560 unless @xmath561 . in the first case ,
when @xmath556 or @xmath403 , this implies that @xmath562 . by lemma [ lem : tworeflections](ii ) , @xmath330 .
and so @xmath563 if and only if the corresponding tableau is rotationally - invariant . in the second case , when @xmath557 , we write @xmath564 where @xmath545 is a small positive real number . arguing as in the proof of theorem [ thm : specialribbon ] , we see that @xmath337 is obtained by rotating @xmath96 by @xmath117 , replacing each entry @xmath256 by @xmath332 , and switching the positions of @xmath95 and @xmath175 if and only if they are in the same row or column .
for @xmath96 to be invariant under this operation , we require rotation - invariance of @xmath218 , and the entries @xmath565 and @xmath566 of @xmath258 must be in the same row or column .
corollaries [ cor : combinv1 ] and [ cor : combinv2 ] can also be proved bijectively .
the bijection between @xmath116-fixed and @xmath536-fixed tableaux in @xmath31 is given by the involution @xmath567 , which is the evacuation operator restricted to the entries @xmath568 .
if @xmath22 is even , we have @xmath569 hence @xmath570 gives a bijection between @xmath116-fixed and @xmath536-fixed tableaux in @xmath31 .
let @xmath571 . for @xmath102 , one can think of @xmath572 as performing `` reverse evacuation '' on the largest @xmath558 entries of @xmath96 .
it is enough to show that @xmath573 .
but as shown in figure [ fig : modelbijection1 ] the two sides of this equation are modelled by lifts of homotopic paths .
( 60,50)(0,0 ) ( 0,15 ) ( top ) and @xmath574 ( bottom).,title="fig : " ] ( 16,30 ) ( top ) and @xmath574 ( bottom).,title="fig : " ] ( 40,27 ) ( top ) and @xmath574 ( bottom).,title="fig : " ] ( 16,0 ) ( top ) and @xmath574 ( bottom).,title="fig : " ] ( 40,3 ) ( top ) and @xmath574 ( bottom).,title="fig : " ] ( 13,27.5 ) ( 10,10 ) 45@xmath575 ( 16.5,33.5)@xmath116 ( 13,12.5 ) ( 10,10 ) -45@xmath575 ( 18.25,18.75)@xmath536 ( 33,33.5 ) ( 10,10 ) -10@xmath575 ( 37,40)@xmath576 ( 33,6.5 ) ( 10,10 ) 10@xmath575 ( 36,12.75)@xmath570 the bijection for corollary [ cor : combinv2 ] is a bit more involved .
if @xmath96 is a domino tableau of shape @xmath32 , there are two entries labelled @xmath118 for each @xmath577 .
one of these is closer to the northwest corner ; colour this entry blue and the other one red . with each @xmath118 ,
starting at @xmath578 and descending to @xmath579 , slide the red @xmath118 through the blue entries @xmath580 , and then replace the red @xmath118 by @xmath581 .
the result will be a standard young tableau of shape @xmath32 .
finally apply @xmath570 to this tableau , and denote the result @xmath582 . the map @xmath583 gives a bijection between domino tableaux of shape @xmath32 and rotationally - invariant tableaux in @xmath31 .
let @xmath184 , where @xmath584 , and @xmath542 for @xmath577 . as we saw in the proof of theorem [ thm : specialribbon ] , domino tableaux are in bijection with real valued tableaux of shape @xmath32 with entries @xmath188 , where @xmath549 and @xmath551 are in the same row or column , and
these tableaux correspond to @xmath325-fixed points in @xmath216 . in this
set - up , the map @xmath585 is modelled by the path illustrated in figure [ fig : modelbijection2 ] .
let @xmath586 when @xmath194 is drawn as a circle ( as in figure [ fig : modelpromote ] ) , @xmath309 acts by rotating the circle @xmath587 clockwise .
the path illustrated in figure [ fig : modelbijection2 ] is homotopic to a path that computes the action of @xmath59 ; hence under the identifications above , the action of @xmath59 coincides with the map @xmath585 .
let @xmath312 be the real valued tableau corresponding to a point @xmath191 .
note that @xmath588 .
thus @xmath23 is an @xmath325-fixed point if and only if @xmath376 is a @xmath169-fixed point in @xmath589 . by lemma [ lem : tworeflections ]
, the tableau corresponding to @xmath590 is @xmath591 ; hence @xmath366 corresponds to a @xmath169-fixed point if and only if @xmath592 .
it follows that @xmath96 is identified with a domino tableau if and only if @xmath366 is identified with a rotationally invariant tableau , and the result follows .
( 60,30)(0,0 ) ( 0,0 ) , and rotationally invariant tableaux of shape @xmath593 , when @xmath22 is even.,title="fig : " ] ( 20,10 ) , and rotationally invariant tableaux of shape @xmath593 , when @xmath22 is even.,title="fig : " ] ( 40,0 ) , and rotationally invariant tableaux of shape @xmath593 , when @xmath22 is even.,title="fig : " ] ( 15,10 ) ( 10,10 ) 25@xmath575 ( 35,10 ) ( 10,10 ) -25@xmath575 similar arguments can be used to establish explicit bijections between domino tableaux and rotationally - invariant tableaux ( of the appropriate type ) in cases ( 1 ) and ( 3 ) .
one can also show that the procedure used to define @xmath582 gives a bijection between domino tableau of shape @xmath42 and the @xmath116-fixed tableaux in @xmath99 for any partition @xmath42 of even size .
we leave the details of these generalizations to the reader .
the correspondence between points in @xmath216 and tableaux can be extended to the case where @xmath188 is a multiset of points in @xmath194 . in this more general situation ,
the correspondence is no longer bijective .
nevertheless , we can still use it to obtain generalizations of theorems [ thm : specialribbon ] and [ thm : speciald - ribbon ] to the case where @xmath188 is a multiset fixed by an involutions @xmath513 .
although the problem of counting @xmath59-fixed points in @xmath216 still makes some sense , the fibre @xmath216 is not necessarily reduced when @xmath188 is a multiset .
we resolve this by refining our problem , and considering @xmath59-fixed points in an intersection of schubert varieties . for each @xmath306
, we define a flag in @xmath15}$ ] : @xmath594}\,.\ ] ] if @xmath595 , @xmath596 \cap { { \mathbb{c}}_{n-1}[z]}\,.\ ] ] in particular @xmath597 is the flag @xmath598 defined in the introduction .
we also define @xmath599 to be the flag @xmath600 ( defined in the section [ sec : introwronskian ] ) , which is equal to @xmath601 . for every @xmath91
, we have a * schubert variety * in @xmath16 relative to the flag @xmath602 : @xmath603 the codimension of @xmath604 in @xmath16 is @xmath97 .
[ prop : wronskischubert ] the wronskian @xmath21 is divisible by @xmath605 if and only if @xmath606 for some partition @xmath607 .
also , @xmath608 for some @xmath609 .
hence every point in @xmath216 lies in some intersection of schubert varieties relative to the flags @xmath610 .
( we will not discuss the multiplicities of such points here , but more information can be found in @xcite . )
we will are left to consider the problem of counting @xmath59-fixed points in an intersection @xmath611 where @xmath612 are distinct , and @xmath613 .
mukhin , tarasov and varchenko proved that if @xmath188 is real , such intersections of schubert varieties are always transverse @xcite .
[ thm : mtv2][mukhin - tarasov - varchenko ] if @xmath614 are distinct real points , and @xmath615 are partitions with @xmath613 , then the intersection is finite , transverse , and real .
let us assume that @xmath612 are real , and @xmath616 .
let @xmath23 be a point in the intersection . instead of associating a unique tableau to the point @xmath23 , we will associate a set of tableaux @xmath617 .
we do this by the reverse of the construction described in @xcite .
let @xmath188 denote the multiset containing @xmath612 with multiplicities @xmath618 respectively , and let @xmath191 . consider a path @xmath619 ,
@xmath201 $ ] , starting at @xmath620 , and ending at a set @xmath621 .
( essentially we want to perturb @xmath188 without changing the relative @xmath193-order of the entries . ) for any lifting to a path @xmath242 with @xmath622 , it is reasonable to think of @xmath248 as being a tableau associated to @xmath23
. however , since @xmath623 is a multiset , the fibre @xmath624 need not be reduced , and so there may be more than one such lifting .
we define @xmath625 the set of tableaux coming all possible liftings of the path @xmath241 .
the sets @xmath626 that arise in this way are characterized in terms of haiman s dual equivalence relation @xcite .
we will not review all the relevant definitions here but refer the reader to ( * ? ? ?
* section 2 ) , where they may be found along with the proof of theorem [ thm : dualequiv ] , below .
let @xmath627 denote the sequence of partitions @xmath628 .
for @xmath102 , @xmath629 , let @xmath630 $ ] be the subtableau of @xmath96 consisting of entries @xmath631 if @xmath630 $ ] is dual equivalent to @xmath632 $ ] for all @xmath633 .
this is an equivalence relation on @xmath31 .
we will say that @xmath96 has * type * @xmath627 if @xmath630 $ ] has rectification shape @xmath231 for all @xmath118 . note that if @xmath96 has type @xmath627 and @xmath634 , then @xmath635 also has type @xmath627 .
[ thm : dualequiv ] for every point @xmath191 , the associated set of tableaux @xmath626 is an equivalence classes of the relation @xmath636 .
the point @xmath23 is in the intersection if and only if @xmath96 has type @xmath627 for some ( equivalently for every ) @xmath637 . for @xmath102 , * switching * the subtableaux @xmath630 $ ] and @xmath638 $ ] is the following procedure ( see e.g. @xcite ) .
first add @xmath639 to each of the entries of @xmath630 $ ] , and subtract @xmath639 from each of the entries of @xmath638 $ ] .
then , slide each of the boxes of @xmath630 $ ] , from largest entry to smallest , through @xmath638 $ ] .
the result is a new tableau in @xmath31 .
let @xmath640 be the tableau obtained by switching @xmath630 $ ] and @xmath638 $ ] for each @xmath118 such that @xmath641 .
[ thm : tworeflectionsgeneral ] let @xmath23 be a point in the intersection .
the point @xmath642 lies in the intersection @xmath643 and the associated set of tableaux is @xmath644 .
the point @xmath645 lies in the intersection @xmath646 and the associated set of tableaux is @xmath647 . here
@xmath648 is the involution on @xmath31 which rotates a tableau by @xmath117 and replaces each entry @xmath78 by @xmath649
. we will prove ( i ) , leaving ( ii ) , which is similar , to the reader .
let @xmath650 and let @xmath338 . call a path @xmath238 * short *
if @xmath651 is in some sufficiently small neighbourhood of @xmath252 , for all @xmath201 $ ] , @xmath652 .
the definition of @xmath626 requires us to consider a `` suitable '' path @xmath241 , i.e. a path @xmath619 , @xmath201 $ ] , where @xmath620 and @xmath621 is a set .
it is possible to choose this to be a short path , in which @xmath653 for all @xmath118 , and we will assume that this is the case .
this ensures that for any lifting @xmath242 , the tableau associated to @xmath654 is just @xmath248 ( by lemma [ lem : tworeflections](i ) ) .
thus @xmath626 is also obtained from all possible liftings of the path @xmath655 , starting at @xmath656 . similarly , to obtain @xmath657 using we need a suitable path @xmath658 , @xmath201 $ ] .
again , we assume this is a short path , and that @xmath659 for all @xmath118 .
note that we have @xmath660 now , any two short paths from @xmath661 to @xmath662 are homotopy equivalent to each other .
in particular , the path @xmath663 is homotopy equivalent to the concatenation of @xmath655 with any short path of sets from @xmath664 to @xmath662 ; hence we can compute @xmath657 by lifting this concatenation of paths , starting at @xmath656 . as noted above , the first part ( lifting @xmath655 to a path starting at @xmath656 )
gives us @xmath626 .
it follows that the relationship between @xmath626 and @xmath657 is described by lifting a short path from @xmath664 to @xmath662 . from
, we see that along such a path the points close to @xmath256 must change places ( in the @xmath193-order ) with points close to @xmath665 , whenever @xmath666 . by theorem [ thm : sliding ] ,
the effect of this on a tableau @xmath637 is to switch the subtableaux @xmath630 $ ] and @xmath638 $ ] , and the result follows .
theorem [ thm : tworeflectionsgeneral ] gives us two ways of computing the number of fixed points of an involution @xmath513 in the intersection .
first note , that for there to any fixed points , the intersection itself must be @xmath59-invariant , which is to say we must have @xmath667 if we take @xmath668 , the number of fixed points is the number of equivalence classes of @xmath636 of type @xmath627 that are invariant under @xmath669 . if we take @xmath516 , the number of fixed points is the number of equivalence classes of @xmath636 of type @xmath627 that are invariant under @xmath670 .
we note that the equivalence classes of @xmath636 are in bijection with sequences of littlewood - richardson tableaux , which makes it quite manageable to compute moderate sized examples by hand , by either method .
let @xmath421 , @xmath671 , and @xmath672 .
suppose @xmath673 are distinct non - real complex numbers .
we consider the problem of counting the number of real points in the intersection @xmath674 the points @xmath675 lie on a circle in @xmath182 , so by proposition [ prop : onsomecircle ] , the real points in are the fixed points of an involution @xmath513 .
hence this problem is equivalent to counting the @xmath325-fixed points in the intersection @xmath676 where @xmath677 . by theorem [ thm : tworeflectionsgeneral](i ) , the number of these points is the number of equivalence classes of @xmath636 of type @xmath678 that are invariant under @xmath679 .
there are eight equivalence classes of @xmath636 of type @xmath627 ; representatives of these classes are shown in figure [ fig : eighttableaux ] . to compute @xmath669 , we switch @xmath680 $ ] with @xmath681 $ ] , and switch @xmath682 $ ] with @xmath683 $ ] .
the four tableaux in the top row are invariant under @xmath679 ; therefore so are the equivalence classes they represent . for each tableaux
@xmath96 in the bottom row , one can check that @xmath684 .
thus , exactly four of our equivalence classes are fixed by @xmath679 , and hence there four real points in the intersection .
this answer agrees with the experimental calculation in @xcite . we can also obtain this answer by counting @xmath169-fixed points in the intersection @xmath685 where @xmath686 . by theorem [ thm : tworeflectionsgeneral](ii ) , these are counted by equivalence classes of @xmath636 of type @xmath627 that are invariant under @xmath687 .
the four tableaux on the left side of figure [ fig : eighttableaux ] are representatives of these invariant classes . !
2 & ? 7 & ? 8 + !
5 & ? 9 & ! !
11 + ? ? 4 & ? ?
6 & ? 8 & ! !
7 & ? 9 & ! !
3 & ? 7 & ? 8 & !
6 & ? 9 & ! ! 10 & !
5 & ? 8 + !
11 + ? 7 & ? 9 & !
! 12 +
5 & ? 7 + !
4 & ? 9 & ! !
6 & ? 8 & !
5 & ? 9 + !
6 & ? 7 & !
! 10 + ? ? 4 & ?
5 & ? 9 + !
4 & ? 7 & ! !
6 & ? 8 & ! !
5 & ? 7 + !
6 & ? 9 & !
! 10 + ? ? 4 & ?
virtually all of our calculations in the remaining sections of this paper will be done using the plcker coordinates for the grassmannian .
both the richardson variety @xmath43 and the fixed point set @xmath65 are characterized by the fact that certain plcker coordinates are zero . in describing of the latter ,
we begin to see how @xmath3-ribbons and @xmath3-ribbon tableaux fit into the picture .
there are several different ways to index plcker coordinates : we will use @xmath35 as our indexing set . for @xmath91 , set @xmath688 suppose @xmath18 is the subspace spanned by polynomials @xmath154 . consider the @xmath689 matrix @xmath690f_i(z)$ ] , whose entries are the coefficients of the polynomials @xmath521
. our convention will be that the rows of @xmath691 are indexed by @xmath692 , while the columns are indexed @xmath693 .
the * plcker coordinates * of a point @xmath18 are @xmath694_{\lambda \in \lambda}$ ] , where @xmath695 is the maximal minor of @xmath696 with column set @xmath697 .
these are homogeneous coordinates : up to a scalar multiple , @xmath694_{\lambda \in \lambda}$ ] does not depend on the choice of basis .
we also define positive integer constants @xmath698 for @xmath91 .
the wronski map can be written explicitly in terms of the plcker coordinates and the constants @xmath699 .
[ prop : wrplucker ] for any point @xmath18 , the wronskian @xmath21 is given in terms of the plcker coordinates of @xmath23 by @xmath700 both sides of are only defined up to a scalar multiple , hence this should rightly be interpreted as an equation in @xmath24})}$ ] .
we note a few useful results about the relationship between plcker coordinates and richardson varieties .
the next two results are standard facts .
[ prop : richardsonplucker1 ] let @xmath18 .
there is a unique maximal @xmath91 such that @xmath701 , and a unique minimal @xmath702 , such that @xmath703 . for these partitions we have @xmath704 .
let @xmath705 denote the interval in the poset @xmath35 between @xmath41 and @xmath42 .
the following theorem can be found in @xcite .
[ thm : richardsonplucker2 ] as a subscheme of @xmath16 , @xmath43 is defined by the the equations @xmath706 . in particular , for any point @xmath704 , if @xmath707 then @xmath708 .
[ prop : richardsonplucker3 ] if @xmath709}$ ] , and @xmath43 is a compatible richardson variety , then for any point in @xmath710 we have @xmath703 and @xmath701 .
since @xmath711 , taking coefficients of @xmath712 and @xmath713 in , we have @xmath714h(z ) \qquad\text{and}\qquad \sum_{|\nu| = |\lambda| } q_\nu p_\nu(x ) = [ z^{|\lambda|}]h(z)\,.\ ] ] since @xmath43 is @xmath44-compatible , the right hand sides are non - zero . finally , for @xmath715 to be non - zero we must have @xmath716 ; hence the first sum is just @xmath717 , and the second is @xmath718 .
the result follows . from these facts , we deduce proposition [ prop : richardson ] .
let @xmath409 .
let @xmath41 be the unique minimal partition such that @xmath703 , and let @xmath42 be the unique maximal partition such that @xmath701
. then @xmath704 .
moreover from , we have @xmath719h(z ) = 0 $ ] for @xmath720 and @xmath721h(z ) = q_\lambda p_\lambda(x ) \neq 0 $ ] .
thus @xmath722 , and similarly @xmath723 .
thus @xmath23 lies in some @xmath44-compatible richardson variety .
if @xmath23 were in two @xmath44-compatible richardson varieties , @xmath23 would lie in their intersection .
this is a proper richardson subvariety @xmath724 , where @xmath725 , or @xmath726 .
but this contradicts theorem [ thm : richardsonplucker2 ] , since @xmath703 , and @xmath701 .
the key to adapting the methods of @xcite to the proof of theorem [ thm : ribbon ] is an understanding of the components of @xmath65 .
we first show that these components are naturally indexed by two sets .
the component of a point @xmath727 can be indexed by an integer vector , determined by the eigenvalues of the @xmath4-action ; it can also be indexed by an _ @xmath3-core _ in @xmath35 , determined by the non - vanishing plcker coordinates .
let @xmath728 , and let @xmath729 .
viewed as an element of @xmath0 , @xmath730 is a generator for the cyclic group @xmath4 .
however , viewed as an element of @xmath52 , it acts on the vector space @xmath15}$ ] as a semisimple operator .
the monomials are a basis of eigenvectors : the action of @xmath730 on the polynomial @xmath731}$ ] is given by @xmath732 for @xmath733 , let @xmath734 denote @xmath735-eigenspace for the action of @xmath730 on @xmath15}$ ] .
thus @xmath734 has a basis of monomials @xmath736 , where @xmath737 .
if @xmath727 , then the @xmath730 action on @xmath15}$ ] restricts to the invariant subspace @xmath23 .
let @xmath738 , where @xmath739 is the multiplicity of the eigenvalue @xmath735 for the action of @xmath730 on @xmath23 .
note that we have @xmath740 , and as the action is semisimple , @xmath741 . for any non - negative integer vector @xmath742 , with @xmath743 and @xmath744 for @xmath733 ,
let @xmath745 [ lem : grproduct ] let @xmath746 be as above .
then @xmath747 is naturally isomorphic to the product of grassmannians @xmath748 .
in particular @xmath747 is irreducible , which shows that the eigenvalues of the action of @xmath730 distinguish the components of @xmath65 .
each @xmath727 decomposes as @xmath749 now @xmath750 iff @xmath751 , @xmath733 , hence we must have @xmath744 for all @xmath118 . in this case the @xmath3-tuple @xmath752 is a point in @xmath753 .
conversely any point @xmath754 in this product of grassmannians corresponds to the point @xmath755 .
a partition is called an * @xmath3-core * if its diagram does not have any hook lengths equal to @xmath3 .
equivalently @xmath756 is an @xmath3-core if there does not exist an @xmath3-ribbon tableau of shape @xmath757 for any @xmath758 .
it is well known that if @xmath42 is any partition , there is a unique @xmath3-core , @xmath759 ( called the @xmath3-core of @xmath42 ) , such that there exists an @xmath3-ribbon tableau of shape @xmath760 .
it can be obtained by successively deleting @xmath3-ribbons from the shape @xmath42 until it is no longer possible to do so . from this description
it is not entirely obvious that @xmath761 is well defined ; this will be more evident from the alternate descriptions of @xmath3-cores that are given below .
our next goal is to show that the components of @xmath65 are in bijection with the @xmath3-cores in @xmath35 .
suppose @xmath762 , @xmath763 , and @xmath764 , and consider the component @xmath765 of @xmath65 .
the points of this component are subspaces of @xmath766}$ ] spanned by @xmath767 polynomials of the form @xmath768 , and @xmath80 polynomials of the form @xmath769 .
the matrix @xmath770f_i(z)$ ] , as in section [ sec : plucker ] , is of the form @xmath771 from this matrix , one can see that at most @xmath772 of the @xmath773 plcker coordinates of a point @xmath774 can be non - zero .
the partitions indexing these plcker coordinates all have @xmath767-core equal to @xmath775 .
[ lem : corecomponents ] let @xmath747 be a component of @xmath65 .
there is a unique @xmath3-core in @xmath776 with the following property : for every @xmath750 and every partition @xmath91 whose @xmath3-core is not equal to @xmath756 , the plcker coordinate @xmath777 is zero .
conversely , for every @xmath3-core @xmath776 , there is a unique component @xmath778 whose associated @xmath3-core is @xmath756 .
thus the partitions indexing non - zero plcker coordinates of a point @xmath727 all have the same @xmath3-core , and this @xmath3-core determines the component of @xmath23 .
we will write @xmath779 , if @xmath756 is the @xmath3-core corresponding to the component @xmath747 . to prove lemma [ lem : corecomponents ]
, we use the correspondence between partitions and abacus diagrams .
an * @xmath3-abacus * is a an arrangement of @xmath13 beads inside an array with @xmath3 columns .
the columns of this array are called * runners * , and are indexed @xmath780 .
the possible positions of the beads are indexed by non - negative integers , where @xmath781 are the positions in the first row ( from left to right ) , @xmath782 are the positions in the second row , and so on .
thus the possible positions for beads on the @xmath118@xmath297runner are congruent to @xmath118 modulo @xmath3 .
the @xmath3-abacus associated to the partition @xmath42 has beads in positions @xmath697 .
an example is given in figure [ fig : abacus ] .
& & & & _ 9 & _ 10 & _ 11 & _ 12 & , ^_13 + & & & & , ^_8 + & & & & , ^_7 + & & _ 4 & _ 5 & , ^_6 + & _ 2 & , ^_3 + _ 0 & , ^_1 ( 16,18)(0,0 ) ( 1,16)(0,1)2 ( 15,16)(0,1)2 ( 1,16)(1,0)14 ( 1,18)(1,0)14 ( 2,0)(0,1)16 ( 6,0)(0,1)16 ( 10,0)(0,1)16 ( 14,0)(0,1)16 ( 2,14 ) ( 6,14 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 10,14 ) ( 14,14 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 2,11 ) ( 6,11 ) ( 10,11 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 14,11 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 2,8 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 6,8 ) ( 10,8 ) ( 14,8 ) ( 2,5 ) ( 6,5 ) ( 0,0)(0,0 ) ( 0,0 ) ( 0,0 ) ( -2,-.4 ) ( 10,5 ) ( 14,5 ) ( 2,2 ) ( 6,2 ) ( 10,2 ) ( 14,2 ) it is not hard to show that deleting an @xmath3-ribbon from a partition corresponds to sliding a bead up one position in the corresponding @xmath3-abacus ( see @xcite ) . from this
it follows that the @xmath3-core of a partition is determined by the number of beads on each runner of the @xmath3-abacus .
a partition is an @xmath3-core if the corresponding @xmath3-abacus has its beads upwardly justified .
let @xmath727 , and let @xmath91 .
let @xmath783 be the number of beads on the @xmath118@xmath297runner of the @xmath3-abacus @xmath697 , @xmath733 .
we claim that if @xmath784 , then @xmath785 .
suppose @xmath786 , and @xmath787 .
then for some @xmath118 , @xmath788 .
let @xmath154 be is a basis of @xmath730-eigenvectors for @xmath23 , where @xmath789 are @xmath735-eigenvectors .
then @xmath777 is given as the determinant of the minor of the matrix @xmath691 as in section [ sec : plucker ] with columns @xmath697 .
consider the first @xmath790 rows of this minor .
there are at most @xmath783 non - zero entries in each row they are in columns corresponding to the beads on the @xmath118@xmath297runner of the abacus .
but since @xmath788 these rows can not be linearly independent , and so the plcker coordinate is @xmath777 zero .
it follows that if @xmath750 , then the @xmath785 for every @xmath91 that does not have @xmath783 beads on the @xmath118@xmath297runner of the associated @xmath3-abacus .
let @xmath756 be the @xmath3-core corresponding to @xmath3-abacus with @xmath783 upwardly justified beads on runner @xmath118 . if @xmath791 , then the @xmath3-abacus does not the correct number of beads on each runner , and so @xmath785 , as required .
finally , let @xmath776 be an @xmath3-core .
the argument above shows that there there is at most one component @xmath792 of @xmath65 associated to the @xmath3-core @xmath756 . to see that such a component is always non - empty
, we note that there exists a point in @xmath727 whose plcker coordinates are : @xmath793 , @xmath785 for @xmath794 .
this @xmath23 is necessarily in @xmath792 .
[ thm : fixedpointplucker ] let @xmath18 and let @xmath795
. then @xmath796 if and only if all partitions in @xmath797 have @xmath3-core equal to @xmath756 .
we have already shown one direction in lemma [ lem : corecomponents ] .
it remains to show that if every partition in @xmath797 has @xmath3-core @xmath756 , then @xmath798 .
the action of @xmath730 on @xmath16 is given in plucker coordinates by @xmath799 if all partitions in @xmath797 have the same @xmath3-core @xmath756 , then for all @xmath91 we have either @xmath800 ( if @xmath801 ) , or @xmath785 ( otherwise ) . thus
@xmath802_{\lambda \in \lambda}$ ] is a scalar multiple of @xmath694_{\lambda \in \lambda}$ ] , i.e. @xmath803 .
finally , @xmath796 , since @xmath701 for some partition with @xmath801 . without any further work
, we can already deduce a few facts about fixed points in the fibres of the wronski map .
let @xmath44 be a @xmath4-fixed polynomial , and let @xmath43 be a compatible richardson variety .
every point in @xmath67 lies in the same component of @xmath65 .
let @xmath804 . by proposition [ prop : richardsonplucker3 ] ,
@xmath703 , so @xmath23 lies in the component of @xmath65 corresponding to the @xmath3-core of @xmath41 .
the next corollary is a special case of theorem [ thm : ribbon ] .
[ cor : notsamecore ] let @xmath44 be a @xmath4-fixed polynomial , and let @xmath43 be a compatible richardson variety . if @xmath42 and @xmath41 do not have the same @xmath3-core , then the @xmath67 is empty . in this case
there are no standard @xmath3-ribbon tableaux of shape @xmath48 .
if there exists a point @xmath804 then @xmath703 and @xmath701 , so @xmath41 and @xmath42 must have the same @xmath3-core . since the @xmath3-core is obtained by deleting @xmath3-ribbons ,
if there exists an @xmath3-ribbon tableaux of shape @xmath48 then @xmath41 and @xmath42 have the same @xmath3-core .
let @xmath805 denote the set of @xmath3-tuples of partitions @xmath806 , where @xmath807 has at most @xmath783 parts and largest part at most @xmath808 .
the @xmath3-abacus @xmath697 can be encoded by a sequence of @xmath3 partitions in @xmath805 , where @xmath783 is the number of beads on the @xmath118@xmath297runner @xcite .
we do this by encoding the sequence of beads on each runner into a partition in the standard way , where the positions of beads correspond to up - steps along walking along the boundary of the partition .
it is easier to state the reverse map , which sends @xmath809 to the abacus @xmath810 the @xmath3-tuple @xmath806 is called the * @xmath3-quotient * of @xmath42 .
let @xmath756 be the unique @xmath3-core whose abacus has @xmath783 beads on runner @xmath118 .
let @xmath811 be the set of all partitions with @xmath3-core @xmath756 .
this is also the set of all partitions whose @xmath3-abacus has @xmath783 beads on runner @xmath118 .
hence the the @xmath3-quotient construction gives a bijection between @xmath812 and @xmath805 .
moreover , sliding a bead up the @xmath118@xmath297runner of the abacus corresponds to deleting a box from the partition @xmath813 and to deleting a @xmath3-ribbon from @xmath42 .
it follows that @xmath814 is an isomorphism of partially ordered sets @xmath815 .
since the latter is a distributive lattice , the former is too .
( note , however , it is not a sublattice of @xmath35 : the meet and join operations on @xmath816 do not coincide with those of @xmath35 ) . the elements of @xmath816 index the non - zero plcker coordinates on @xmath817 , which by lemma [ lem : grproduct ] is naturally identified with the product of @xmath818 .
the @xmath3-quotient construction tells us how to identify the plcker coordinates @xmath819 with coordinates of the segre embedding of this product of grassmannians .
let @xmath820 $ ] denote the polynomial ring generated by indeterminates @xmath821 .
the homogeneous coordinate ring of @xmath16 is @xmath822 = { \mathbb{c}}[{{\bf p}}]/i$ ] , where @xmath823 is the plcker ideal .
@xmath823 is generated by quadratics , and describes all relations among the plcker coordinates .
similarly , we write @xmath824 = { \mathbb{c}}[{{\bf p}}_{k}]/i_k$ ] for the homogeneous coordinate ring of @xmath825 . the tensor product @xmath826 \otimes \dots \otimes { \mathbb{c}}[{\mathrm{gr}}(s_{r-1},m_{r-1})]$ ] is therefore the multihomogeneous ( @xmath827-graded ) coordinate ring for @xmath818 . abusing notation slightly ,
we will write the multihomogeneous ideal for this product as @xmath828 \otimes \dots \otimes { \mathbb{c}}[{{\bf p}}_{r-1}]\big ) + \dots + \big({\mathbb{c}}[{{\bf p}}_0 ] \otimes \dots \otimes { \mathbb{c}}[{{\bf p}}_{r-2 } ] \otimes i_{r-1}\big)\,.\ ] ] there is a sign @xmath829 associated to each partition @xmath91 , which is easily obtained from the @xmath3-abacus .
first let @xmath830 , be the sequence obtained by listing the positions of the beads on runner @xmath272 ( in increasing order ) , followed by those on runner @xmath95 , and so on .
this list gives an ordering of the elements of @xmath831 .
we define @xmath832 to be the sign of the permutation that puts them in increasing order .
[ lem : segre ] consider the @xmath150-algebra homomorphism @xmath833 \to { \mathbb{c}}[{{\bf p}}_0 ] \otimes \dots \otimes { \mathbb{c}}[{{\bf p}}_{r-1}]$ ] defined by @xmath834 the homogeneous ideal of @xmath792 is @xmath835 $ ] .
thus @xmath836 descends to a homomorphism @xmath837 \to { \mathbb{c}}[{\mathrm{gr}}(s_0,m_0 ) ] \otimes \dots \otimes { \mathbb{c}}[{\mathrm{gr}}(s_{r-1},m_{r-1})]\ ] ] from the homogeneous coordinate ring of @xmath792 to the multihomogeneous coordinate ring of @xmath818 .
hence identifies the coordinates of @xmath838 with the coordinates of segre embedding of @xmath818 .
recall from the proof of lemma [ lem : grproduct ] that the isomorphism @xmath839 , is given by @xmath840 , where @xmath841 .
we must show that the plcker coordinates of @xmath23 are ( up to the specified sign ) products of plcker coordinates of @xmath842 .
let @xmath843 .
@xmath844 be a basis for @xmath841 .
the plcker coordinates of @xmath845 are given indexed by partitions @xmath813 with at most @xmath783 parts , and largest part @xmath846 .
let @xmath847 be the @xmath848 matrix with entries @xmath849f_{ki}(z)$ ] .
then @xmath850 is the minor of @xmath849f_{ki}(z)$ ] with columns specified by the positions of the beads on runner @xmath118 of the @xmath3-abacus @xmath697 .
on the other hand @xmath777 is computed as the minor of the matrix @xmath851 with column set @xmath697 .
suppose @xmath852 .
if we take the columns of this minor in the order @xmath830 , we see that this is a block diagonal matrix with determinant equal to @xmath853 but reordering the columns in this way changes the sign of the minor by @xmath832 , and thus we have @xmath854 consider a richardson variety @xmath43 , where @xmath42 and @xmath41 both have @xmath3-core equal to @xmath756 . then the intersection @xmath855 is non - empty . write @xmath856 $ ] for the homogeneous ideal of @xmath857 . also put @xmath858 .
[ lem : richardsonquotient ] suppose @xmath42 and @xmath41 have @xmath3-quotients @xmath859 , and @xmath860 respectively . under the isomorphism @xmath861 ,
@xmath862 is identified with @xmath863 , where @xmath864 denotes the richardson variety in @xmath865 corresponding to the skew shape @xmath866 . for @xmath733 ,
let @xmath867 $ ] denote the homogeneous ideal of the richardson variety @xmath864 . by theorem [ thm : richardsonplucker2 ] , this is generated by @xmath868 , together with @xmath869 the ideal @xmath870 is generated by @xmath871 and @xmath872 .
we must show that @xmath873 the right hand side of is generated by @xmath874 together with all @xmath875 by , @xmath876 is in this set if and only if @xmath877 or @xmath878 is in the set for some @xmath118 .
the latter occurs when @xmath879 .
thus the generators for the right hand side of are the same as the generators for the left hand side , as required .
we will need some facts about the initial ideal of @xmath871 and more generally the initial ideal of @xmath870 for certain choices of weights on the plcker coordinates .
let @xmath880 be a vector of rational numbers .
the * weight * of a monomial @xmath881 $ ] with respect to @xmath882 is @xmath883 if @xmath884 $ ] , then the * initial form * of @xmath885 with respect to @xmath882 , denoted @xmath886 , is the sum of all monomial terms in @xmath885 for which the weight of the term is minimized .
the * initial ideal * of an ideal @xmath887 $ ] with respect to @xmath882 is the ideal @xmath888 in this context , the vector @xmath882 is called a * weight vector*. the weight vectors of interest to us will be constructed in the following way .
we begin with an @xmath3-ribbon tableau @xmath96 of shape @xmath48 and a sequence of rational numbers @xmath889 , where @xmath890 .
let @xmath891 , so that @xmath892 is the maximal element in the lattice @xmath805 . for each @xmath893 , the entries labelled @xmath78 in @xmath96 form a ribbon .
this ribbon , under the @xmath3-quotient correspondence , is identified with a single box in some @xmath894 .
put the rational number @xmath895 in this box .
we require our input data to be such that the following condition holds .
[ cond : square ] in the filling of @xmath894 , the entries are weakly decreasing along rows and down columns , and no @xmath79 square has four equal entries . in every case we will consider , @xmath889 will be a weakly decreasing sequence in which at most two of the @xmath895 are equal ; this guarantees that condition [ cond : square ] will always satisfied .
if not all the boxes of all @xmath894 are filled , we fill the remaining boxes with rational numbers that are either less than @xmath896 or greater than @xmath897 , so that condition [ cond : square ] is still satisfied . for each partition @xmath898 , let @xmath899 be the sum of all entries in the filling of @xmath894 that are _ outside _ the partition @xmath900 .
the weight vector we consider will be @xmath901 suppose @xmath5 , @xmath421 , @xmath902 .
let us construct a weight vector @xmath882 for the domino tableau @xmath903 , & ? ? 2 & ! 3 & ! ! 4 & ! !
1 & ? ? 2 & !
3 & ! ! !
6 \\ ? 1 &
6 \end{young}\ .\ ] ] and the decreasing sequence @xmath904 .
we start by placing these numbers in the boxes of @xmath905 and @xmath906 corresponding to the ribbons in @xmath96 : @xmath907 9 & 8 & \end{young } \qquad\quad { { \mathchoice { \raisebox{.05ex}{\footnotesize $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.05ex}{\footnotesize $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.1ex}{\tiny $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.1ex}{\tiny $ \sqsubset\!\!\!\sqsupset$ } } } } _ 1\ : \ \begin{young}[c ] 7 & 6 \\ 5 & 4 \end{young}\ .\ ] ] for the empty box we have a choice we are allowed to fill it with any number less than @xmath908 here we will choose to fill it with a @xmath95 . to determine any particular @xmath909 , say for @xmath910 , we shade the boxes corresponding to @xmath911 , @xmath907 ? 9 & 8 & 1 \end{young } \qquad\quad { { \mathchoice { \raisebox{.05ex}{\footnotesize $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.05ex}{\footnotesize $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.1ex}{\tiny $ \sqsubset\!\!\!\sqsupset$ } } { \raisebox{.1ex}{\tiny $ \sqsubset\!\!\!\sqsupset$ } } } } _ 1\ : \ \begin{young}[c ] ? 7 & 6 \\ ?
5 & 4 \end{young}\ , \ ] ] and sum the numbers outside of the shaded boxes : @xmath912 .
we do the same procedure for the partition @xmath913 : in this case the sum is @xmath95 .
then @xmath914 is the difference of these two sums @xmath915 .
note that @xmath916 is always @xmath272 , and the choice of numbers that fill the empty boxes does affect @xmath917 if @xmath716 .
for such a weight vector , the initial ideals of @xmath918 and @xmath870 are described by the lattice structure of @xmath816 .
[ lem : initialideal ] for every weight vector @xmath882 obtained from the process above , the initial ideal @xmath919 is generated by @xmath920 and the initial ideal @xmath921 is generated by @xmath922 here @xmath923 and @xmath924 denote the meet and join operations on the lattice @xmath816 .
it is enough to prove the second statement , since @xmath925 in the case where @xmath926 and @xmath42 is the largest partition in @xmath816 . as in the proof of lemma [ lem : richardsonquotient ] ,
let @xmath867 $ ] denote the ideal of the richardson variety @xmath864 .
each of the vectors @xmath927 , @xmath733 defines a weight vector for @xmath928 $ ] .
it is well known ( see ( * ? ? ?
* theorem 14.16 ) ) that for these weight vectors , if condition [ cond : square ] holds , the initial ideal @xmath929 is generated by the binomials @xmath930 more generally , one can show ( e.g. using ( * ? ? ? * theorem 7.1 ) ) that @xmath931 is generated by @xmath932 the initial ideal of @xmath933 \otimes \dots \otimes { \mathbb{c}}[{{\bf p}}_{r-1}]$ ] with respect to the weight vector @xmath934 is then the sum of these : @xmath935 .
this is a basic fact about sums of ideals in disjoint sets of variables , which follows readily from buchburger s criterion .
now , for any @xmath884 $ ] , @xmath936 is either @xmath272 or equal to @xmath937 .
moreover , any @xmath938-initial form in the image of @xmath836 is the image of some @xmath882-initial form in @xmath820 $ ] .
it follows that @xmath939 since each of the initial ideals @xmath931 is generated in degree @xmath95 and @xmath80 , @xmath940 is generated by the @xmath836-preimage of the multihomogeneous elements of degree @xmath941 and @xmath942 in @xmath943 .
thus @xmath940 is generated by @xmath944 together with binomials of the form @xmath945 where the @xmath3-quotients of @xmath946 , @xmath947 , @xmath948 and @xmath949 satisfy @xmath950 among these are the binomials from , which arise when @xmath951 and @xmath952 for all @xmath118 .
on the other hand , each binomial , is in the ideal generated by .
indeed , implies that @xmath953 and @xmath954 , and so can be written as @xmath955 thus @xmath921 is generated by , as required .
let @xmath149 be an algebraically closed field of characteristic zero containing @xmath150 as a subfield .
let @xmath956 be the field of puiseux series over @xmath149 .
this is also algebraically closed field of characteristic zero , which contains @xmath150 as a subfield . as noted in the introduction , it is enough to prove theorem [ thm : ribbon ] working over any algebraically closed field , and we will do so working over the field @xmath957 . for most of our arguments it will suffice to consider @xmath958 , however , for others we will take @xmath959 , the field of puiseux series over @xmath150 .
from this point forward , all algebraic objects ( @xmath16 , @xmath43 , @xmath960 , etc . ) will be considered over the field @xmath957 , rather than over @xmath150 .
for example , @xmath961})$ ] will be the grassmannian of @xmath13-planes in the vector space @xmath962}$ ] over @xmath957 .
the advantage of working over @xmath957 is that it is a complete valuation ring . if @xmath963 , the * valuation * of @xmath885 is defined to be @xmath964 the * leading term * @xmath965 and * leading coefficient * @xmath966 are @xmath967 g = c_\ell\,.\ ] ] additionally , we set @xmath968 , @xmath969 and @xmath970 .
let @xmath971 .
let @xmath43 be a richardson variety in @xmath16 .
let @xmath972 be non - zero puiseux series , and put @xmath973 it will be convenient to adopt the convention that @xmath974 , and @xmath975 .
we will assume in what follows that the plcker coordinates of a point @xmath976 are normalized so that @xmath977 . in the case where @xmath978 have distinct valuations , the points in the fibre of @xmath176 can be identified with standard young tableaux of shape @xmath48 .
[ lem : valtableau ] suppose that @xmath979 for every point @xmath976 , there is a unique chain of partitions @xmath980 with the property that @xmath981 moreover , for this chain we have @xmath982 the chain of partitions @xmath983 can be identified with a standard young tableaux in @xmath984 , obtained by placing entry @xmath118 in the box @xmath219 .
@xmath218 is the tableau corresponding to the point @xmath976 . for any shape @xmath716 ,
let @xmath985 denote the set of entries in @xmath218 that are outside the shape @xmath946 .
thus @xmath986 .
the system can be therefore rewritten as @xmath987 .
in fact , when written this way , these equations hold for all plcker coordinates of @xmath43 , not just those in the chain .
[ lem : valothers ] suppose that @xmath979 if @xmath976 , let @xmath988 be the corresponding tableau , and let @xmath989 be the associated chain of partitions .
we have @xmath990 in particular , @xmath991 has the unique minimal valuation among all plcker coordinates of @xmath23 indexed by partitions of size @xmath639 .
that is , if @xmath992 is any partition of size @xmath639 but not equal to @xmath231 , we have @xmath993 .
this proof is little more than a trivial modification of the proof of ( * ? ? ?
* corollary 4.4 ) .
we sketch the main idea here , omitting several details that are the same .
since the number of points in @xmath176 is exactly @xmath70 , it is enough to show that for each tableau @xmath994 encoding a chain of partitions @xmath995 , there exists a point @xmath976 such that , and hold . to do this
, we use ( * ? ? ?
* theorem 4.2 ) , which tells us how to obtain such a point from a solution to a certain system of equations . in our case ,
the system of equations we need to solve is @xmath996 in complex variables @xmath997 .
this has the unique solution @xmath998 according to ( * ? ? ?
* theorem 4.2 ) , every solution @xmath999 of gives rise to a point @xmath976 satisfying @xmath1000 taking valuations of both sides of , we see that @xmath23 satisfies . in the case where @xmath1001 , we substitute the solution into , giving @xmath1002 which is equivalent to .
this in turn implies .
finally , we turn to the last statement of lemma [ lem : valothers ] .
let @xmath992 be any partition of size @xmath639 .
if @xmath708 , then since @xmath1003 , @xmath1004 equality occurs if and only if @xmath1005 , i.e. @xmath1001 . if @xmath1006 then by theorem [ thm : richardsonplucker2 ] @xmath1007 . for a polynomial @xmath44 of the form with coefficients in @xmath957 ,
the roots of @xmath44 do not have distinct valuations . thus lemmas [ lem : valtableau ] and [ lem : valothers ] can not be used directly in the case of interest to us .
nevertheless , some parts of these lemmas still hold when the valuations of the roots are non - distinct .
[ lem : nongenericval ] suppose that @xmath1008 there exists a chain of partitions @xmath980 ( not necessarily unique ) encoded by a tableau @xmath994 with the following properties : for all @xmath716 we have @xmath1009 equality holds if the leading coefficients of @xmath1010 are generic . if @xmath1011 , then and hold . if @xmath1011 , then @xmath991 has the unique minimal valuation among all plcker coordinates of @xmath23 indexed by partitions of size @xmath639 .
fix @xmath188 , and consider the family of finite @xmath149-schemes over @xmath1012 , whose fibre at a point @xmath1013 is @xmath1014 , where @xmath1015 for each component @xmath1016 of the total space , there is a unique vector @xmath1017 , such that @xmath1018 for all @xmath1019 , with equality on zariski - dense subset of @xmath1016 .
we claim that there is a zariski - dense subset @xmath1020 such that if @xmath1021 , then for any @xmath1022 , holds for the point @xmath1023 and some tableau @xmath994 .
this shows that the vector @xmath1017 above is given by the right hand side of , which implies ( 1 ) .
this claim will also be used in the proof of ( 2 ) . to prove the claim , we work over @xmath959 , though still assuming @xmath1024 .
consider the set @xmath1025 which is zariski - dense in @xmath1026 .
fix @xmath1027 , and let @xmath1028 be a small positive rational number
. put @xmath1029 where @xmath1030 . since
@xmath545 is assumed to be small , we have @xmath1031 , so for any point @xmath1032 , holds : in particular , there is a tableau @xmath96 such that @xmath1033 for @xmath716 . now consider the substitution @xmath1034 .
although this does not give a well defined transformation from @xmath957 to itself we do have @xmath1035 , because the valuations of the plcker coordinates of @xmath656 are uniformly bounded below for all positive @xmath545 near @xmath272 .
thus we obtain a point @xmath1036 and every point in @xmath1037 is of this form .
the plcker coordinates of @xmath1023 satisfy @xmath1038 for @xmath716 . since @xmath545 can be made arbitrarily small , we see that holds for the point @xmath1023 .
next we prove ( 2 ) .
since the right hand size of is a continuous nowhere - vanishing function of @xmath1039 , it is enough to prove this under the assumption that the leading coefficients @xmath1040 are generic .
our claim shows that in this case holds for @xmath23 and some tableau @xmath994 .
suppose @xmath96 encodes the chain of partitions @xmath980 .
as in the proof of lemma [ lem : valothers ] , it follows that when @xmath1011 , @xmath991 has the unique minimal valuation among all plcker coordinates of @xmath23 indexed by partitions of size @xmath639 .
now @xmath1041 , which we can rewrite using as @xmath1042 ( since the plcker coordinates are assumed to be normalized so that @xmath977 , we need the factor @xmath699 on the left hand side to ensure that the coefficients of @xmath713 agree on both sides . ) taking the coefficient of @xmath1043 on both sides , we see that @xmath1044 where @xmath1045 denotes the @xmath78@xmath297elementary symmetric function .
now take the leading term of both sides . in the case where @xmath1011 , the leading term of the right hand side is @xmath1046 as all other monomials in the sum have strictly larger valuation . by similar reasoning ,
the leading term of the left hand side is @xmath1047 equating and , we obtain .
taking valuations of both sides gives .
finally , we use the last argument in the proof of lemma [ lem : valothers ] once more to deduce ( 3 ) from ( 1 ) and ( 2 ) .
the chain of partitions in lemma [ lem : nongenericval ] is not necessarily uniquely determined by the point @xmath23 .
however if @xmath1048 are the indices for which @xmath1049 , then the subchain @xmath1050 is characterized by the property ( 3 ) in lemma [ lem : nongenericval ] .
hence this particular subchain of partitions is uniquely determined by @xmath23 .
we can encode this subchain into a tableau , again denoted @xmath218 , by placing entry @xmath78 in each of the boxes of @xmath1051 . in this way we can associate a tableau to every point @xmath18 .
this is not always an interesting tableau : for example if @xmath1052 $ ] , then all roots have the same valuation and so @xmath218 is the tableau in which every entry is @xmath95 . in particular , the reader should note that this definition of @xmath218 is not does _ not _ agree with the definition given in section [ sec : circle ] .
we now show that in the case of interest to us , when @xmath23 is a generic @xmath1053-fixed point , this tableau is a standard @xmath3-ribbon tableau .
[ lem : alwaysribbon ] suppose that @xmath44 is @xmath4-fixed polynomial of the form , where @xmath1054 let @xmath43 be an @xmath44-compatible richardson variety . if @xmath804 , then the corresponding tableau @xmath218 is a standard @xmath3-ribbon tableau of shape @xmath48 .
rewrite @xmath44 in the form .
since each binomial @xmath1055 has @xmath3 roots , each with valuation @xmath1056 , we may assume that @xmath1057 thus the tableau @xmath218 has entries @xmath1058 , each appearing @xmath3 times .
let @xmath1059 be the chain of partitions associated to @xmath218 .
by lemma [ lem : nongenericval](2 ) , each of plcker coordinates @xmath991 satisfies and hence is non - zero , for @xmath1060 . by lemma [ lem : corecomponents ] , this implies that the each of the partitions in the chain have the same @xmath3-core . in particular
, @xmath1061 is a partition of size @xmath3 where @xmath1062 and @xmath231 have the same @xmath3-core .
any such partition must be an @xmath3-ribbon , and the result follows .
suppose that @xmath44 is @xmath4-fixed polynomial of the form , where @xmath1063 .
we have shown that every point @xmath976 has some associated tableau @xmath218 , and that for a point in @xmath65 , this tableau is a standard @xmath3-ribbon tableau .
now , fix @xmath1064 , and let @xmath1065 be the corresponding chain of partitions . to complete the proof of theorem [ thm : ribbon ] , we need to prove that there exists a unique point @xmath804 such that @xmath470 , and moreover that @xmath23 is a reduced point of @xmath67 .
suppose that such a point @xmath23 exists
. then the following must be true : @xmath1066 ; equivalently by theorem [ thm : richardsonplucker2 ] and theorem [ thm : fixedpointplucker ] , @xmath1067 for @xmath1068 ; by lemma [ lem : nongenericval](2 ) , @xmath1069 we will first show that there exists a point in @xmath1070 satisfying these two conditions .
then we will show that the additional condition @xmath711 makes this point unique and reduced in @xmath67 .
let @xmath882 be a weight vector constructed ( as in section [ sec : ideals ] ) from the tableau @xmath96 and the decreasing sequence of rational numbers @xmath1071 .
note that @xmath1072 .
let @xmath1073 as we will see , the point @xmath1074 corresponding to @xmath96 will be in @xmath1075 .
consider a vector @xmath1076 satisfying the following conditions : @xmath1077 here @xmath1078 is the sign defined in section [ sec : ideals ] . for any such @xmath1079 , let @xmath1080 [ lem : alphachain ] there for any sequence @xmath1081 , there is a unique vector @xmath1082 satisfying , and @xmath1083 for @xmath1084 .
this follows from the fact that @xmath1085 is a distributive lattice .
condition asserts that @xmath1086 is a modular function , and so this function is uniquely by its values on the maximal chain @xmath1087 .
[ lem : inycondition ] if @xmath1066 is a point such that @xmath1088 then @xmath1089 .
let @xmath1090 p_\nu(x)$ ] .
since @xmath1067 for @xmath1091 , it is enough to show that @xmath1092 for @xmath1093 .
equivalently , we must show that @xmath1094 and @xmath1095 for @xmath1093 .
note that @xmath1096 for @xmath1091 and @xmath1097 for @xmath1084 . by lemma [ lem : nongenericval](1 ) , holds , which can be re - expressed as @xmath1098 thus @xmath1099 .
this implies that @xmath1100 is in the variety of the initial ideal @xmath1101 . by lemma [ lem : initialideal ]
we see that @xmath1102 is a vector satisfying , and @xmath1097 for @xmath1084 . by lemma
[ lem : alphachain ] , we deduce that and we deduce that @xmath1103 , as required .
[ lem : ylocalcoords ] there exist local coordinates @xmath1104 on an open subset of @xmath16 containing @xmath1075 with the following properties : for each @xmath1093 , there exists a polynomial @xmath1105 $ ] such that @xmath1106 for all @xmath1107 .
@xmath1108 for all @xmath1107 .
the map @xmath1109 gives a bijection @xmath1110 in particular @xmath1111 is non - empty , and @xmath1112 for all @xmath1107 .
[ ex : ylocalcoords ] we illustrate the construction of the local coordinates @xmath1104 , in the case where @xmath68 .
suppose @xmath421 , @xmath671 , and @xmath1113 , & 2 & 4 & 7 \\ 1 & 5 & 6 & 9 \\ 3 & 8 \end{young}\ .\ ] ] for any point @xmath1107 , we can choose a basis for @xmath23 so that the matrix @xmath691 of section [ sec : plucker ] has the form @xmath1114 in which the entries are @xmath272 or products of indeterminates : @xmath1115 where @xmath1116 denotes the entry of @xmath96 in row @xmath78 , column @xmath9 .
the coordinates @xmath1104 are defined to be @xmath1117 the reader can check that for any @xmath44 , with @xmath1063 , properties ( 1)(3 ) of lemma [ lem : ylocalcoords ] are satisfied .
when @xmath68 , the construction of @xmath1104 is illustrated in example [ ex : ylocalcoords ] .
the proof that properties ( 1)(3 ) hold for this construction is routine .
when @xmath1118 , we use the identification @xmath1119 of lemma [ lem : richardsonquotient ] , and define @xmath1104 to be the local coordinates on appropriate subsets of @xmath1120 . specifically ,
if @xmath1121 are the entries of @xmath96 that correspond to boxes of @xmath894 under the @xmath3-quotient construction , then @xmath1122 are the local coordinates ( defined as in the @xmath68 case ) on @xmath864 .
we leave it to the reader to verify that properties ( 1)(3 ) can be deduced from the @xmath68 case .
we note that lemmas [ lem : alphachain ] , [ lem : inycondition ] and [ lem : ylocalcoords ] ( and their proofs ) are still valid under the weaker condition @xmath1123 , provided that condition [ cond : square ] is satisfied in the construction of the weight vector @xmath882 .
we will need this fact in section [ sec : dihedral ] .
[ lem : ytableau ] if @xmath1107 , then @xmath470 .
for @xmath1107 we have @xmath1124 for every partition @xmath1093 such that @xmath1125 , @xmath1126 .
the inequality here follows from the definition of @xmath882 .
this shows that @xmath470 .
[ lem : existunique ] with @xmath44 as above , let @xmath43 be a compatible richardson variety . for every @xmath1064 , there exists a unique point @xmath804 whose corresponding tableau is @xmath470 .
moreover @xmath67 is reduced . by lemma [ lem : alphachain ]
there is a unique @xmath1082 satisfying , and @xmath1127 for this @xmath1079 , is equivalent to . thus if @xmath804 is a point such that @xmath470 , then by lemma [ lem : inycondition ] we must have @xmath1089 .
conversely , by lemma [ lem : ytableau ] for every point @xmath1089 we have @xmath470 .
it suffices therefore to show that there exists a unique , reduced point in @xmath1128 .
let @xmath1129 } { \ \big|\ } { \text{\rm \footnotesize lt}}\big([z^j]g(z)\big ) = { \text{\rm \footnotesize lt}}\big([z^j]h(z)\big ) \text { for } j=0 , \dots , n\big\}\,.\ ] ] we now show that @xmath1130 .
for any @xmath1089 we have @xmath1131{\mathrm{wr}}(x;z)\big ) & = \tfrac{1}{q_\lambda}\ , { \text{\rm \footnotesize lt}}\big([z^j ] \sum_{\nu \vdash j } q_\nu p_\nu(x)\big ) \\ & = { \begin{cases } \frac{q_{\lambda_k}}{q_\lambda}\ , { \text{\rm \footnotesize lt}}(p_{\lambda_k}(x ) ) & \quad\text { if $ j = |\lambda_k|$ for some $ k$ } \\ 0 & \quad \text{otherwise}\ , , \end{cases}}\end{aligned}\ ] ] since , in the first case , @xmath470 , so @xmath991 has unique minimal valuation among all plcker coordinates indexed by partitions of size @xmath1132 . on the other hand , if @xmath1132 , @xmath1133h(z ) \big ) & = \prod_{i = k+1}^\ell { \text{\rm \footnotesize lt}}(h_i ) \\ \label{eqn : hcoeffalpha } & = \tfrac{q_{\lambda_k}}{q_\lambda}\ , \alpha_{\lambda_k } u^{w_{\lambda_k } } \\ \nonumber & = \tfrac{qi{\lambda_k}}{q_\lambda}\ , { \text{\rm \footnotesize lt}}(p_{\lambda_k}(x))\,,\end{aligned}\ ] ] and @xmath1134h(z ) = 0 $ ] otherwise .
thus @xmath1135 , for all @xmath1089 .
to finish , we show that the map @xmath1136 is invertible . for this
we write @xmath711 as a system of equations in the local coordinates @xmath1104 of lemma [ lem : ylocalcoords ] , and use hensel s lemma ( see e.g. ( * ? ? ?
* exer . 7.25 ) ) , which is the analogue of the inverse function theorem for valuation rings .
comparing non - zero coefficients of both sides , @xmath711 can be written as @xmath1137 h(z ) \qquad\text{for } k=0 , \dots , \ell\,.\ ] ] by lemma [ lem : ylocalcoords](2 ) , the leading term of the left hand side of is @xmath1138 from , the leading term of the right hand side of is @xmath1139 dividing both sides by @xmath1140 , the system of equations can be written as @xmath1141 where @xmath1142 , and by lemma [ lem : ylocalcoords](1 ) @xmath1143 $ ] . to apply hensel s lemma to the system
, we need to check two things .
every @xmath1089 is a first order solution to ; that is , @xmath1144 for @xmath1089 .
this follows from lemma [ lem : ylocalcoords](3 ) .
the jacobian matrix @xmath1145 is invertible over @xmath1146 , for @xmath1089 .
this is true because the jacobian matrix is upper - triangular , and the diagonal entries are invertible by lemma [ lem : ylocalcoords](3 ) .
we deduce that the system has a unique , reduced solution @xmath1147 .
this solution gives the local coordinates of the point @xmath1148 , as required .
lemma [ lem : existunique ] is the final ingredient in the proof of theorem [ thm : ribbon ] .
since @xmath44 is assumed to be a generic polynomial of the form , we may assume that @xmath1063 .
lemmas [ lem : alwaysribbon ] and [ lem : existunique ] show that in this case , @xmath1149 gives a bijection between the reduced finite scheme @xmath67 and @xmath82 .
although @xmath67 is reduced , it is not always true that @xmath176 is reduced when @xmath44 is a generic polynomial of the form .
one can see this from the combinatorics .
for example , suppose @xmath764 , @xmath1150 and @xmath1151 .
then @xmath1152 , and @xmath1153 , which means that @xmath176 is a finite scheme of length @xmath80 , and @xmath1154 acts on it with exactly @xmath95 fixed point .
since @xmath1154 can not act on a set of size @xmath80 with only one fixed point , this shows that @xmath176 is supported on a single point , hence non - reduced .
more generally , if @xmath3 is prime and @xmath1155 , then for every @xmath4-fixed @xmath44 compatible with @xmath43 , we have that @xmath176 is non - reduced .
the action of @xmath169 on a polynomial @xmath15}$ ] is given by @xmath1160 in terms of the matrix @xmath691 used to defined the plcker coordinates ( see section [ sec : plucker ] ) , the action of @xmath169 simply reverses the order of the columns ; hence the minor @xmath1161 for @xmath23 becomes @xmath1162 for @xmath1158 .
these are the plcker coordinates @xmath1163 and @xmath1164 respectively .
the sign is a constant scalar multiple , so it can be discarded . since @xmath1167 , @xmath1168 for all @xmath992 . in particular ,
if @xmath42 is the maximal partition such that @xmath701 , then @xmath1169 is the minimal partition such that @xmath1170 . by proposition [ prop : richardsonplucker1 ]
, we must have @xmath1166 .
this establishes one easy special case of theorem [ thm : d - ribbon ] : if @xmath1171 , then there are no @xmath126-fixed points in @xmath176 , and the combinatorial sets @xmath134 and @xmath137 are both empty .
let @xmath1172)}$ ] be a @xmath126-fixed polynomial , and let @xmath43 be a compatible richardson variety .
henceforth , we will assume that @xmath1166 .
working over the field @xmath957 , we will write @xmath44 in the form , where where @xmath1173 , @xmath890 , and @xmath1174 the space of all @xmath44 that are compatible with @xmath43 always has two components .
if @xmath136 is odd , the analysis will be the same for both components , and we treat them as a single case .
if @xmath136 is even , the analysis is different for the two components .
hence will need to consider three different cases , corresponding to the three different generic types of @xmath126-fixed polynomials described in the introduction . in type ( 1 ) , @xmath136 is odd , and we can assume the @xmath1175 have distinct valuations .
since @xmath171 is invariant under @xmath172 , this implies that @xmath1176 , and @xmath1177 for @xmath1178 . similarly in type ( 2 ) , where @xmath136 is even , we may assume that @xmath1175 have distinct valuations ; hence @xmath1179 for @xmath1180 , and none are equal to @xmath174 . in type ( 3 ) , @xmath136 is even , and we are required to have @xmath1181 and @xmath1182 , which both have valuation @xmath272 . the rest of the @xmath173 can be assumed to have distinct valuations , and hence @xmath1183 , for @xmath1184 . a summary is given in table [ tab : generictypes ] .
let @xmath1172)}$ ] be polynomial compatible with @xmath43 . for @xmath1187 , let @xmath1188 denote the largest integer for which @xmath1189 divides @xmath44 .
if @xmath1190 then @xmath1191 .
it follows that @xmath1192 is even
. there are four cases .
we continue to assume that @xmath44 is a @xmath126-fixed polynomial , and @xmath43 is a compatible richardson variety with @xmath1166 . for any tableau @xmath96 of shape @xmath48 , let @xmath331 denote the tableau obtained by rotating @xmath96 by @xmath117 , and replacing each entry
@xmath118 by @xmath131 , where @xmath9 is the largest entry .
thus the action of @xmath169 on the fibre @xmath47 models the action of @xmath117 rotation on tableaux .
we note that this result does not require @xmath44 to be generic , of one of the three types , though for our applications it will be .
let @xmath1198 be the chain of partitions associated to @xmath218 .
then @xmath991 has unique minimal valuation among all plcker coordinates of @xmath23 indexed by partitions of size @xmath639 . by
, @xmath1199 has unique minimal valuation among all plcker coordinates of @xmath1158 indexed by partitions of size @xmath1200 .
thus @xmath1201 encodes @xmath1202 , i.e. @xmath1203 .
if @xmath44 is of type ( 1 ) or ( 2 ) , then @xmath151 have distinct valuations , and hence by lemma [ lem : existunique ] we have a bijective correspondence between the reduced finite scheme @xmath67 and @xmath82 . for any @xmath804 , we have @xmath1204 iff @xmath1205 , where the last equality is by lemma [ lem : reflectrotate ] .
hence @xmath169-fixed points in @xmath67 are in bijection with the set @xmath134 of rotationally - invariant @xmath3-ribbon tableaux .
if @xmath44 is of type ( 3 ) , then a bit more work is required , since @xmath1206 and @xmath1207 have the same valuation . in this case , the corresponding tableau is not a standard @xmath3-ribbon tableau , but something close .
let @xmath96 be a tableau of shape @xmath48 , where @xmath1208 and @xmath136 is even .
we say @xmath96 is an * almost - standard * @xmath3-ribbon tableau if the entries of @xmath96 not equal to @xmath139 form @xmath3-ribbons , and the entries equal to @xmath139 form a skew shape with @xmath125-boxes that supports an @xmath3-ribbon .
we will write the chain of partitions corresponding to such a tableau @xmath96 as @xmath1209 one possibility , if @xmath96 is an almost - standard @xmath3-ribbon tableau of shape @xmath48 , is that the entries equal to @xmath139 actually form a @xmath125-ribbon . in this case
we say that @xmath96 has a * long ribbon*. to prove the claim , we work over @xmath1213 .
let @xmath1214 fix @xmath1215 and a small positive rational number @xmath1028 .
put @xmath1216 arguing as in the proof of lemma [ lem : nongenericval ] , we see that for any point @xmath1032 such that @xmath1217 there is a corresponding point @xmath1218 such that @xmath1219 .
conversely , if @xmath656 does not satisfy , then @xmath1220 .
points @xmath1032 satisfying are in bijection with tableaux @xmath1064 whose associated chain of partitions is of the form @xmath1221 for some @xmath1222 ; in other words , these points correspond to extensions of the chain to a maximal chain in @xmath1085 .
if @xmath96 has a long ribbon , there is a unique way to extend this chain , hence one point in @xmath1223 corresponding to @xmath96 .
if @xmath96 does not have a long ribbon , there are two ways to extend the chain , hence two corresponding points .
this proves the claim .
now , if @xmath1224 is a point such that @xmath1219 , then by lemma [ lem : nongenericval](2 ) , we have @xmath1225 the right hand side of is a nowhere - vanishing function of @xmath1226 .
this implies that property is preserved when we take the limit as @xmath1227 .
thus in case ( i ) the limiting fibre @xmath67 contains exactly one reduced point @xmath23 such that @xmath470 , which proves ( i ) . in case ( ii )
the limiting fibre @xmath67 contains either : [ lem : twopoints ] let @xmath96 be an almost - standard @xmath3-ribbon tableau , and suppose @xmath44 is of type ( 3 ) . if @xmath96 does not have a long ribbon , then there are exactly two points @xmath1230 such that @xmath1229 , and both are reduced points of @xmath67 .
the proof uses essentially the same idea as the proof of lemma [ lem : existunique ] .
first , we identify a vector @xmath1082 , such that one of the points will lie in @xmath1111 .
then we apply hensel s lemma , to show that the point exists and is unique .
let @xmath1222 , and @xmath1231 denote the two partitions of size @xmath1232 that extend the chain to a maximal chain .
let @xmath1233 be the tableau corresponding to the extended chain of partitions @xmath1234 , and let @xmath882 be a weight vector obtained from @xmath1235 and the weakly decreasing sequence @xmath1071 .
then with @xmath1075 defined as in , we have @xmath470 for every @xmath1107 : the proof is analogous to that of lemma [ lem : ytableau ] . for @xmath1242 , and @xmath1243}$ ] , let @xmath1244g(z)\big ) & \quad \text{if } j=|\lambda_k|,\ { \mathrm{val}}\big([z^j]g(z)\big ) \leq
\sum_{i = k+1}^\ell { \mathrm{val}}(h_i ) \\ 0 & \quad \text{otherwise}\ , .
\end{cases}\ ] ] note that @xmath1245h(z)\big)\geq\sum_{i = k+1}^\ell { \mathrm{val}}(h_i)$ ] , with equality when @xmath1246 . on the other hand , there are two terms in the expansion of @xmath1247h(z)$ ] that have this minimal valuation : @xmath1248h(z )
\\ = \big[u^{\sum_{i=(\ell/2)+1}^\ell { \mathrm{val}}(h_i ) } ] \big ( \big(h_{\ell/2 } h_{(\ell/2)+2 } \dotsb h_\ell\big)\,+\ , \big(h_{(\ell/2)+1 } h_{(\ell/2)+2 } \dotsb h_\ell\big ) \big ) = 0\,,\end{gathered}\ ] ] and so we have @xmath1249h(z)\big ) & \quad \text{if } j \neq \tfrac{n}{2 } \\ 0 & \quad\text{if } j=\tfrac{n}{2}\ , .
\end{cases}\ ] ] define @xmath1129 } { \ \big|\ } l_j(g(z ) ) = l_j(h(z ) ) \text { for } j=0 , \dots , n\big\}\,.\ ] ] we now show that the @xmath1130
. let @xmath1089 . for @xmath1250
, we have @xmath1251{\mathrm{wr}}(x;z)\big ) = { \text{\rm \footnotesize lt}}\big([z^j]h(z)\big ) = l_j(h(z))\,;\ ] ] this is the same calculation as in the proof of lemma [ lem : existunique ] . for @xmath1252 , we have @xmath1253 and all other plcker coordinates of @xmath23 indexed by partitions of size @xmath1254 have strictly larger valuation .
thus we have @xmath1255 p_\nu(x ) & \qquad&\\ & = q_{\lambda_{\ell/2 } } \big[u^{w_{\lambda_{\ell/2}}}\big]p_{\lambda_{\ell/2}}(x ) + q_{\lambda'_{\ell/2 } } \big[u^{w_{\lambda'_{\ell/2}}}\big]p_{\lambda'_{\ell/2}}(x ) \\ & = q_{\lambda_{\ell/2 } } \alpha_{\lambda_{\ell/2}}(x ) + q_{\lambda'_{\ell/2 } } \alpha_{\lambda'_{\ell/2}}(x ) & & \text{(by \eqref{eqn : defyalpha})}\\ & = 0 & & \text{(by \eqref{eqn : twomiddlealphas})}\,,\end{aligned}\ ] ] and hence @xmath1135 . for @xmath1246 , leading term of the left hand side and right hand side of are given by and respectively .
when @xmath1256 , the leading term of the left hand side is @xmath1257 ( here we have used @xmath1258 , which comes from applying lemma [ lem : ylocalcoords ] with the chain of partitions @xmath1259 the local coordinates for this chain are the same as those for the chain @xmath1260 , but the roles of @xmath1261 and @xmath1262 are reversed . ) by , the right hand side of has valuation strictly greater than @xmath1263 when @xmath1264 .
thus we can write the system as : @xmath1265 where @xmath1142 , and @xmath1143 $ ] , for @xmath1266 .
let @xmath1267 we need to check two things .
every @xmath1089 is a first order solution to ; that is , for @xmath1089 , @xmath1268f_k(y_1(x ) , \dotsb , y_\ell(x ) ) = \begin{cases } \alpha_{\lambda_k } & \quad\text{if } k \neq \frac{\ell}{2 }
\\ 0 & \quad\text{if } k = \frac{\ell}{2 } \ , . \end{cases}\ ] ] for @xmath1269 , this follows from lemma [ lem : ylocalcoords](3 ) . for @xmath1256
, we calculate @xmath1270 the jacobian matrix @xmath1271 is invertible over @xmath1146 , for @xmath1089 .
this is a block upper triangular matrix in which most of the diagonal blocks are @xmath1272 blocks and are invertible by lemma [ lem : ylocalcoords](3 ) ; however , there is one @xmath79 diagonal block ( in rows @xmath139 , @xmath1273 ) , which is @xmath1274 this @xmath79 matrix is invertible over @xmath1146 if and only if @xmath1275 but by lemma [ lem : ylocalcoords](3 ) , this is equivalent to @xmath1276 which follows from , as @xmath1277 .
thus all diagonal blocks of the jacobian matrix are invertible over @xmath1146 , as required .
therefore , by hensel s lemma the system has a unique , reduced solution @xmath1147 .
this solution gives the local coordinates of a point @xmath1148 ; since @xmath1107 , we have @xmath470 .
we obtain the second point @xmath1278 , by replacing @xmath1279 by @xmath1280 in the definition of @xmath1281 .
this shows that there are at least two reduced points in @xmath67 corresponding to @xmath96 . by lemma [ lem : oneortwopoints](ii ) , these are the only such points .
suppose to the contrary that @xmath1204 but the tableau @xmath988 does not have a long ribbon .
let @xmath1282 and @xmath1079 be as in the proof of lemma [ lem : twopoints ] . by lemma [ lem : reflectrotate ] we must have @xmath1283 ,
so @xmath1284 , for @xmath1246 .
it is not hard to see that we must also have @xmath1285 . by
, @xmath1204 implies that @xmath1286 is constant for all @xmath118 .
in particular we must have @xmath1287 but from and , we have @xmath1288 which gives a contradiction . if @xmath44 is of type ( 3 ) , then lemmas [ lem : oneortwopoints ] and [ lem : twopoints ] show that @xmath67 is reduced .
. then @xmath218 must be an almost - standard ribbon tableau ( by lemma [ lem : almostribbon ] ) with a long ribbon ( by lemma [ lem : longribbon ] ) , and be rotationally - invariant ( by lemma [ lem : reflectrotate ] ) .
in other words , @xmath1289 .
conversely , if @xmath1289 , then @xmath1290 .
but by the uniqueness in lemma [ lem : oneortwopoints](i ) we have @xmath1204 .
thus @xmath169-fixed points in @xmath67 are in bijection with @xmath137 .
j. hauenstein ,
n. hein , a. martn del campo , and f. sottile , _ beyond the shapiro conjecture and eremenko - gabrielov lower bounds _ , website ( 2010 ) ` www.math.tamu.edu/\simsottile/research/pages/lower_shapiro/index.html ` .
k. purbhoo , _ the wronski map and shifted tableau theory _ , int .
res . not .
( to appear ) .
b. rhoades , _ cyclic sieving , promotion , and representation theory _ , j. combin .
theory ser .
a , * 117 * ( 2010 ) no .
1 , 3876 . | the wronski map is a finite , @xmath0-equivariant morphism from the grassmannian @xmath1 to a projective space ( the projectivization of a vector space of polynomials ) .
we consider the following problem .
if @xmath2 is a cyclic subgroup of order @xmath3 , how may @xmath4-fixed points are in the in a fibre of the wronski map over a @xmath4-fixed point in the base ? in this paper , we compute a general answer in terms of @xmath3-ribbon tableaux .
when @xmath5 , this computation gives the number of _ real _ points in the fibre of the wronski map over a real polynomial with purely imaginary roots .
more generally , we can compute the number of real points in certain intersections of schubert varieties .
when @xmath3 divides @xmath6 our main result says that the generic number of @xmath4-fixed points in the fibre is the number of standard @xmath3-ribbon tableaux rectangular shape @xmath7 .
computing by a different method , we show that the answer in this case is also given by the number of of standard young tableaux of shape @xmath7 that are invariant under @xmath8 iterations of jeu de taquin promotion .
together , these two results give a new proof of rhoades cyclic sieving theorem for promotion on rectangular tableaux .
we prove analogous results for dihedral group actions . |
the aim of this paper is to provide a method to determine the location and scale relationship between two groups of one - dimensional observations for two samples , say @xmath0 and @xmath1 , such as the responses of two different products on different subjects , the scores of people on two examinations and so on .
suppose @xmath2 are independent identity distributed according to @xmath3 and @xmath4 are independent identity distributed according to @xmath5 , where @xmath3 and @xmath5 are two unknown distribution functions . when the test for normality is not passed , nonparametric analysis methods should be applied . usually , the mean difference or midian difference of the two samples is used to determine the location difference and use mean ratio or midian ratio to obtain the scale .
they are not reliable since only a small information of the two samples are extracted and the results are not meaningful .
based on the idea of location - scale transformation , freitag , munk and vogt @xcite has developed an approach to access the structure relationship between distributions , in which the whole information of the samples are used . however , the problem we concern are the location difference and scale between two distributions rather than the model structure . that s to say , we have to determine the values of location difference or scale or both for any two distributions . using the same transformation idea ,
our approach can be described as follows .
let @xmath6 a linear function , i.e. , @xmath7 , where @xmath8 and @xmath9 .
@xmath10 is a given measure of discrepancy between two distributions .
denote the distribution function of @xmath11 by @xmath12 .
let @xmath13 .
we want to find the value @xmath14 which minimize @xmath15 .
that s to say , if we transform @xmath2 to be @xmath16 , the two groups of observations are closest and under the `` closest '' mean we can not tell there are any location difference or scale between them .
therefore , we can say @xmath4 is at least @xmath17 larger than @xmath18 times of @xmath2
. there are situations where @xmath14 is not unique .
let @xmath19 , @xmath20 and @xmath21 .
conservatively , at first we can take @xmath22 if @xmath23 , else @xmath24 ; then take @xmath17 to be the value in @xmath25 satisfying @xmath26 .
when @xmath27 is a continues region , it is easy to see that the selected @xmath14 is unique .
therefore , we should find certain @xmath10 that @xmath27 is a continues region .
besides , if we let @xmath28 in @xmath29 , the location difference between @xmath30 and @xmath31 could be determined .
if we let @xmath32 in @xmath29 , the scale between @xmath30 and @xmath31 could be determined . in practice
, we could use the empirical distributions of the two group of data for @xmath3 and @xmath5 , respectively . the discrepancy measure
@xmath10 we consider in this paper will be focused on mallows distance and expand to kolmogorov - smirnov distance .
mallows distance was presented in the formulation of statistics framework in 1972 , however , an independent physics research work had involved such a related concept a little earlier in 1940s .
the rest of the paper is unfolded as follows . in section 2
the main results under mallows distance for the location transformation , scale transformation or both are presented , showing that we can uniquely determine the location and scale relationship between two distributions and thus mallows distance is suitable discrepancy measure to use . in section 3
the similar results can be obtained under kolmogorov - smirnov distance but only for location transformation .
section 4 gives the application of this approach to determine the location and scale relationship on real data .
in this subsection , we consider the proposed approach under mallows distance . formally , the mallows @xmath33-distance ( also known as wasserstein @xmath33-distance ) between distributions @xmath3 and @xmath5 regarding to random variables @xmath34 and @xmath35 , respectively , is defined as @xmath36 where the infimum is taken over the set ( denoted by @xmath37 ) of all joint distributions of @xmath34 and @xmath35 with marginals @xmath3 and @xmath5 . here
we require that @xmath34 and @xmath35 have finite @xmath33th moment , i.e. , @xmath38 and @xmath39 . for @xmath40 , the mallows @xmath33-distance @xmath41 has the two properties .
* mallows distance @xmath41 , i.e. satisfies axioms of a metric on @xmath37 . * the convergence of distributions in mallows distance is equivalent to weak convergence plus @xmath33th moment convergence.(lavina and bickel , 2001 ) let @xmath42 be an uniform random variable , @xmath43 , and @xmath44 is the inverse of a distribution function , @xmath45 .
according to johnson and samworth ( 2005 ) we know @xmath46 equation ( [ eq - u ] ) gives an easier computation formula to calculate the distance , that is @xmath47 particularly , when @xmath48 , we have a further relationship for computation @xmath49 which is especially useful when calculating mallows 1-distance using empirical distribution for real data , in order to circumvent the unknown real distribution .
let @xmath50 , where @xmath8 and @xmath51 , and @xmath52 be its distribution function , then it is easy to obtain that @xmath53 .
the purpose of our approach is to find the optimal shift and scale values @xmath14 to minimizing the mallows @xmath33-distance between @xmath54 and @xmath55 , that is @xmath56 then the following result can be obtained . [ thr-1 ] for distribution functions @xmath3 and @xmath5 , let @xmath53 with @xmath57 .
then the mallows @xmath33-distance ( @xmath58 ) between @xmath54 and @xmath55 , denoted by @xmath29 , a function of two variables @xmath59 and @xmath60 , is a continuous and convex function on half plane , i.e. , for any @xmath61 , and @xmath62 , @xmath63 , it holds that @xmath64 it can be easily obtained that @xmath65 . from ( [ eq - uu ] )
, we know @xmath66 then the continuity of the @xmath67 is trivial . besides , using minkowski unequality we have @xmath68 according to the definition of theorem [ thr-1 ] , scaled parameter @xmath59 should be greater than zero , but we can easily give an apparent analysis of transformed distribution function if @xmath69 .
@xmath70 also , the theorem shows that under mallows @xmath33-distance ( @xmath71 ) @xmath67 is a convex function of @xmath72 , thus ( [ eq-1 ] ) is a continues region
. we can select @xmath14 according to the plan in section [ sec : intro ] . if we only consider the shifted case or scaled case , let @xmath28 or @xmath32 in @xmath67 ,
then we can obtain the following results .
for distribution functions @xmath3 and @xmath5 , let @xmath73 .
then the mallows @xmath33-distance ( @xmath58 ) between @xmath54 and @xmath55 , denoted by @xmath74 , a function of @xmath60 , is a continuous and convex function on @xmath75 , i.e. for any @xmath61 , and @xmath63 , it holds that @xmath76 for distribution functions @xmath3 and @xmath5 , let @xmath77 with @xmath57 .
the mallows distance ( @xmath58 ) between the scaled distribution @xmath54 and @xmath55 , denoted as @xmath78 , a function of @xmath59 , is a continuous and convex function on @xmath79 , i.e. for any @xmath61 , and @xmath62 , it holds that @xmath80 in order to illustrate @xmath67 may not be strictly convex , let distributions @xmath3 and @xmath5 to be @xmath81 actually , @xmath3 is the uniform distribution over two half unit intervals @xmath82 $ ] and @xmath83 $ ] , and @xmath5 is uniform over @xmath84 $ ] . from ( [ eq - uuu ] )
we know @xmath74 can be calculated via @xmath85 .
then it is easy to verify that @xmath74 reaches minimum of 0.5 in the entire interval @xmath86 $ ] .
the optimal shifted value for @xmath87 is not unique .
therefore , @xmath74 is not strictly convex on @xmath75 , nor is @xmath67 .
since our approach is successful under mallows distance , there is nothing preventing us from exploring other discrepancy measure . here , we are able to realize our approach for shifted case under kolmogorov - smirnov distance ( k - s distance ) , @xmath88 . for k - s distance and distribution functions @xmath3 and @xmath5 ,
our purpose is to find the optimal shift value @xmath17 to minimize @xmath89 .
let @xmath90 and @xmath91 for distribution @xmath30 .
define @xmath92 $ ] and @xmath93 $ ] , then we have @xmath94 .
and we denote @xmath95 and @xmath96 . due to these definitions ,
the statement @xmath97 is apparently hold and we have the following result .
[ thr-2 ] @xmath98 if @xmath99 , then the function @xmath74 decreases on @xmath100 $ ] and increases on @xmath101 .
@xmath102 if @xmath103 , then the function @xmath74 decreases on @xmath104 and increases on @xmath105 .
@xmath98 @xmath106 , if @xmath107 , then we have @xmath108 if @xmath109 , then @xmath110 holds and @xmath111 . by ( [ eq-2 ] ) , we can obtain @xmath112 . thus @xmath112 and @xmath113 .
therefore , @xmath114 decreases on @xmath100 $ ] . similarly , @xmath115 , if @xmath116 , then we have @xmath117 if @xmath112 , then @xmath118 holds and @xmath119 . by ( [ eq-3 ] ) , we can obtain @xmath109 .
thus @xmath120 . therefore ,
@xmath114 increases on @xmath121 .
@xmath102 the proof can follow the similar method as @xmath98 trivially .
theorem [ thr-2 ] shows that our approach under k - s distance can also provide a reasonable , possibly unique location difference between two distributions .
in hair study , we need to assess effets of hair care products in changing hair diameters after a period of use . there are two treatments , say @xmath122 and @xmath30 .
the experiments are conducted as follows .
there are 30 subjects and each subject use @xmath122 and @xmath30 on the left and right head , respectively .
there are two study visit , baseline and 8 weeks later . at each study visit ,
hair diameters are measured on several hundred of hairs on left and right head from a subject .
comparison between visits is to compare the distributions of hair diameters at two visit point .
the diameters from one subject often follows non - traditional distributions . for example , a subject at baseline and 8 weeks later hair diameter frequency plot for treatment @xmath122 are shown in figure [ fig1 ] and the distributions are shown in figure [ fig2 ] .
it is of importance to know holistically how much diameters have changed . for each subject ,
the optimal shift amount of the distributions of hair diameters at the two visit point for each treatment under mallows distance and k - s distance can be obtained .
for instance , the shift plots for @xmath122 and @xmath30 of a subject are displayed in figure [ fig3 ] and shift plots of all subjects for @xmath122 are displayed in figure [ fig4 ] .
the shift corresponding to the minimum distance is the difference between two distributions , for comparison analysis .
shift plot of all subject for @xmath122,width=345 ] use the optimal values for each subject each treatment as responses and perform wilcoxon signed rank test on differences in shifts between treatments for all subjects to detect a difference between the treatments .
the results are shown in table [ tab-1 ] , from which we concludes that the two treatments are difference at 0.01 level .
.[tab-1]comparison of treatments under distance shift [ cols="^ , < , < , < , < , < , < " , ]
in this paper , we demonstrated a significant theorem relating to how to measure two distribution within the probabilistic interpretation under mallows distance or k - s distance , and a well studied simulation on real data had been implemented for an illustration of this method . the solid theoretical foundation would be beneficial to others who would have a further understanding or research on mallows distance measures the discrepancy between two distributions , especially two similar distributions with inner relationship .
besides those distances , there might be possibility to use other divergence measures to be minimized after proper transformations .
comparison among various underlying divergence measures will be of interest .
this is an area of research that we continue to pursue provided available resources and interests .
in addition , a comparison of this approach versus other methods is another topic to be investigated .
freitag , g. , munk , a. , vogt , m. 2003 .
_ assessing structural relationships between distributions - a quantile process approach based on mallow s distance .
_ in : recent advances and trends in nonparametric statistics
akritas , m. g. , politis , d. n. , amsterdam : elsevier b. v. , 123 - 137 . | despite of many measures applied for determine the difference between two groups of observations , such as mean value , median value , sample standard deviation and so on , we propose a novel non parametric transformation method based on mallows distance to investigate the location and variance differences between the two groups .
the convexity theory of this method is constructed and thus it is a viable alternative for data of any distributions .
in addition , we are able to establish the similar method under other distance measures , such as kolmogorov - smirnov distance .
the application of our method in real data is performed as well .
mallows distance ; shift and scaled ; kolmogorov - smirnov distance |
a network formalism is often very useful for describing complex systems of interacting entities @xcite .
scholars in a diverse set of disciplines have studied networks for many decades , and network science has experienced particularly explosive growth during the past 20 years @xcite .
the most traditional network representation is a static graph , in which nodes represent entities and edges represent pairwise connections between nodes .
however , many networks are time - dependent @xcite or multiplex ( include multiple types of connections between nodes ) @xcite .
moreover , network structure is influenced profoundly by spatial effects @xcite . to avoid discarding potentially important information , which can lead to misleading results , it is thus crucial to develop methods that incorporate features such as time - dependence , multiplexity , and spatial embeddedness in a context - dependent manner @xcite . because of the newfound wealth of rich data , it has now become possible to validate increasingly complicated network structures and methods using empirical data . in the present paper ,
we study a mesoscale network structure known as _ community structure_. a `` community '' is a set of nodes with dense connections among themselves , and with only sparse connections to other communities in a network @xcite .
communities arise in numerous applications .
for example , social networks typically include dense sets of nodes with common interests or other characteristics @xcite , networks of legislators often contain dense sets of individuals who vote in similar ways @xcite , and protein - protein interaction networks include dense sets of nodes that constitute functional units @xcite .
the algorithmic detection of communities and the subsequent investigation of both their aggregate properties and the properties of their component members can provide novel insights into the relationship between network structure and function ( e.g. , functional groupings of newly discovered proteins @xcite ) .
myriad community detection methods have been developed @xcite .
the most popular family of methods entails the optimization of a quality function known as _
modularity _ @xcite . to optimize modularity ,
one compares the actual network structure to some _ null model _ , which quantifies what it means for a pair of nodes to be connected `` at random '' .
traditionally , most studies have randomized only network structure ( while preserving some structural properties ) and not incorporated other features ( such as spatial or other information ) . the standard null model for modularity
optimization is the `` newman - girvan '' ( ng ) null model , in which one randomizes edge weights such that the expected strength distribution is preserved @xcite .
it is thus related to the classical configuration model @xcite .
it has become very popular due to its simplicity and effectiveness , and it has been derived systematically through the consideration of laplacian dynamics on networks @xcite . however , it is also a naive choice , as it does not incorporate domain - specific information .
the choice of a null model is an important consideration because ( 1 ) it can have a significant effect on the community structure obtained via optimization of a quality function , and ( 2 ) it changes the interpretation of communities @xcite .
the best choice for a null model depends on both one s data set and scientific question . in the present paper
, we explore the issue of null model choice in detail in the context of spatially embedded and temporal networks . most existing research on community detection
does not incorporate metadata about nodes ( or edges ) or information about the timing and location of interactions between nodes .
however , with the increasing wealth of space - resolved and time - resolved data sets , it is important to develop community detection techniques that take advantage of the additional spatial and temporal information ( and of domain - specific information , such as generative models for human interactions @xcite ) .
indeed , community detection in temporal networks has become increasingly popular @xcite , but the majority of methods use networks that are constructed from either static snapshots of data or aggregations of data over time windows .
few investigations of community structure in temporal networks have used methods that exploit temporal structure ( see , e.g. , @xcite ) .
there is also starting to be more work on the influence of space on community structure @xcite , but much more research is necessary . in the present paper ,
we use modularity maximization to study communities in spatially embedded and time - dependent networks .
we compare the results of community detection using two different spatial null models a _ gravity null model _ @xcite and a new _ radiation null model _ to the standard ng null model using novel synthetic benchmark networks that incorporate spatial effects via distance decay or disease flux as well as temporal correlation networks that we constructed using time - series data of recurrent epidemic outbreaks in peru .
we also evaluate a recently - proposed _ correlation null model _ , which was developed specifically for correlation networks that are constructed from time series @xcite , on the epidemic - correlation data .
our direct analysis of disease data in the present paper provides a complementary ( e.g. , more systemic ) approach to the majority of studies using network science methodology in this field , which focus on the importance of interpersonal contact networks on the disease spread on an individual level .
these types of network methods have become increasingly prevalent in the modeling of infectious diseases @xcite .
our work also complements other approaches , such as large - scale compartmental models that incorporate transportation networks to link local populations .
such models have been used to study large - scale spatial disease spread ( e.g. , to examine the influence of features such as spatial location , climate , and facility of transportation on phenomena such as disease persistence and synchronization of disease spread ) @xcite .
the rest of the present paper is organized as follows . in section
[ section : networks ] , we give an overview of networks and community detection .
we also discuss the gravity null model and introduce a new radiation null model .
we give our results for synthetic spatial networks in section [ section : benchmarks ] , and we give our results for correlation networks that we construct from disease data in section [ section : dengue ] .
we summarize our results in section [ section : conclusions ] . in appendices ,
we include the results of additional numerical experiments from varying parameters in the benchmark networks .
we also include an additional examination of the similarity between network partitions for the benchmarks and the dengue fever correlation networks .
a network describes a set of entities ( called _ nodes _ ) that are connected by pairwise relationships ( called _ edges _ ) . in the present paper ,
we study weighted networks which are _ spatially embedded _ : each node represents a location in space .
one can represent a weighted network with @xmath0 nodes as an @xmath1 adjacency matrix @xmath2 , where an edge @xmath3 represents the strength of the relationship between nodes @xmath4 and @xmath5 .
we seek to find _ communities _ , which are sets of nodes that are densely connected to each other but sparsely connected to other dense sets in a network @xcite .
we wish to study the evolution of network structure through time .
the simplest way to represent temporal data is through an ordered set of _ static networks _
, which can arise either as snapshots at different points in time or as a sequence of aggregations over consecutive time windows ( which one can take either as overlapping or nonoverlapping ) .
static networks provide a good starting point for the development and investigation of new methods which , in our case , entails how to incorporate spatial information into null models for community detection via modularity maximization .
however , they do not take full advantage of temporal information in data that changes in time .
for example , it can be hard to track the identity of communities in temporal sequences of networks @xcite . to mitigate the community - tracking problem
, we also use a type of _ multilayer network _ @xcite known as a multislice network @xcite .
this gives an @xmath6 adjacency tensor @xmath7 that has @xmath8 layers and @xmath0 nodes in each layer , where each layer has a copy each node @xmath4 .
the intralayer edges in the network are exactly the same as they were for the sequence of static networks : the tensor element @xmath9 gives the weight of an intralayer edge between nodes @xmath4 and @xmath5 in layer @xmath10 .
additionally , each node is connected to copies of itself in consecutive layers @xmath10 and @xmath11 using interlayer edges of weight @xmath12 . in this paper , we will suppose for simplicity that @xmath13 , but one can also consider more general situations @xcite .
a multislice network can have up to ( @xmath14 ) _ multilayer nodes _
( i.e. , node - layer tuples ) , each of which corresponds to a specific ( node , time ) pair .
hence , this structure makes it possible to detect temporally evolving communities in a natural way . for our computations of community structure ,
we flatten the @xmath6 adjacency tensor into a @xmath15 adjacency matrix , such that the intralayer connections are on the main block diagonal and the interlayer connections occur on the off - block - diagonal entries .
we detect communities by maximizing modularity , which we use to describe the `` quality '' of a particular network partition into communities in terms of its departure from a null model @xcite .
the null model amounts to a prior belief regarding influences on network structure , so it is important to carefully consider the choice of null model @xcite . for a weighted static network @xmath2 , modularity is @xcite @xmath16 where @xmath17 is the total edge weight , @xmath18 denotes the community that contains node @xmath4 , the function @xmath19 is the kronecker delta , and @xmath20 is the @xmath21-th element of the null model matrix .
one can examine different scales of community structure by incorporating a resolution parameter @xmath22 @xcite .
smaller values of @xmath22 tend to yield larger communities and vice versa . for multislice networks ,
modularity is given by @xmath23 \delta ( { \mkern 1.5mu\overline{\mkern-1.5muc\mkern-1.5mu}\mkern 1.5mu}_{is } , { \mkern 1.5mu\overline{\mkern-1.5muc\mkern-1.5mu}\mkern 1.5mu}_{jr})}\,,\ ] ] where @xmath24 , the quantity @xmath25 denotes the community that contains node @xmath4 in layer @xmath10 , and @xmath26 is the @xmath21-th element of the null model tensor in layer @xmath10 @xcite . to detect communities via modularity maximization ,
one searches the possible network partitions for the one with the highest modularity score .
because exhaustive search over all possible partitions is computationally intractable @xcite , practical algorithms invariably use approximate optimization methods ( e.g. , greedy algorithms , simulated annealing , or spectral optimization ) , and different approaches offer different balances between speed and accuracy @xcite . in the present paper , we optimize modularity using a two - phase iterative procedure similar to the louvain method @xcite . however , rather than using the adjacency matrix @xmath2 , we work with the modularity matrix @xmath27 with elements @xmath28 for static networks and with the modularity tensor with elements @xmath29 for multislice networks @xcite . the employed louvain - like algorithm @xcite is stochastic , and a modularity landscape for empirical networks typically includes a very large number of nearly - optimal partitions @xcite .
for each of our numerical experiments , we thus apply the computational heuristic 100 times to obtain a _ consensus community structure _ @xcite by constructing an _ association matrix _ @xmath30 ( where the entries @xmath31 represent the fraction of times that nodes @xmath4 and @xmath5 are classified together in the 100 partitions ) and performing community detection on @xmath30 using the uniform null model @xmath32 $ ] @xcite .
we choose the uniform null model in order to detect the strongest community structure in the association matrix ( i.e. , one that is often detected by the original optimization process ) . for multislice networks , we perform community detection and then consensus clustering using the same basic procedure .
this yields an assignment of each multilayer node ( i.e. , node - layer tuple ) to a community .
we are also sometimes interested in community assignments of the original entities ( i.e. , a partition of the set of nodes regardless of what layer they are in ) .
for example , we might wish to compare the result of algorithmic community detection to known partitions , such as grouping a node ( i.e. , province ) by climate , population , administrative region , etc . to do this
, we perform what we call _ province - level community detection _ , which proceeds in two rounds : ( 1 ) we detect communities in a multislice network using any method and null model of choice ; ( 2 ) we use this partition to construct an @xmath1 province - level association matrix ( i.e. , a matrix @xmath33 where entries @xmath34 represent the fraction of times that nodes @xmath4 and @xmath5 are classified together in all layers ) , and we detect province - level communities by maximizing modularity on this association matrix using a uniform null model .
we choose the uniform null model to detect the most temporally persistent community structure in the association matrix ( i.e. , one that is often detected in multiple layers ) .
we can then follow this with consensus community detection across 100 repeat province - level structures .
the choice of null model is vital for the detection of communities using modularity maximization @xcite .
the most common choice is the newman - girvan ( ng ) null model , which randomizes a network such that the expected strength sequence of nodes is preserved @xcite .
for static , weighted networks , the ng null model is given by @xmath35 where @xmath36 is the strength ( i.e. , weighted degree ) of node @xmath4 and @xmath17 is the total edge weight in the network .
for multislice networks , the ng null model is @xcite @xmath37 where @xmath38 is the intralayer strength of node @xmath4 in layer @xmath10 and @xmath24 . despite its popularity and demonstrated effectiveness in many situations , the ng null model is naive in the sense that only takes node strengths into account and it does not incorporate problem - specific information ( such as spatial embeddedness ) .
it is often important to incorporate additional ( domain - specific or even problem - specific ) information , and what one considers to be connected `` at random '' depends fundamentally on the research question of interest .
consequently , ng null model is not suitable for all applications . in many spatially embedded networks
, proximity has a strong effect on the connections between nodes , as ( all else held equal ) neighboring nodes are more likely to be connected to each other ( and their connections are likely to have to have larger weights ) than nodes that are far away @xcite .
moreover , proximity can mask other underlying influences .
consequently , incorporating the expected influence of proximity on edge weights into null models for community detection should make it possible to discover new and important types of structures .
expert et al .
@xcite proposed a spatial null model that was inspired by the `` gravity model '' of human mobility @xcite .
a gravity model assumes that the interaction between two locations is proportional to their importance ( e.g. , population ) , but it decays with distance . in the standard gravity model , the interaction between locations @xmath4 and @xmath5 with respective populations @xmath39 and @xmath40 that are a distance @xmath41 apart is @xmath42 where the `` deterrence function '' @xmath43 describes the effect of space on node interactions .
common choices for the deterrence function include inverse proportionality to distance ( i.e. , @xmath44 ) , inverse proportionality to squared distance ( i.e. , @xmath45 ) , exponential decay ( i.e. , @xmath46 ) , and other interactions of the form @xmath47
it is common to estimate the parameters @xmath48 , @xmath49 , and @xmath50 using regression .
gravity models have been employed successfully during the past half century to model spatial interactions such as population migration @xcite , trade @xcite , and disease spread @xcite .
the simplest form of a gravity - like interaction in eq .
( [ eq : gravitymodel ] ) , with @xmath51 and @xmath52 , was incorporated into a _ gravity null model _
@xcite , to give @xmath53 where @xmath54 is the importance of node @xmath4 .
one estimates the `` deterrence function '' @xmath55 from data for all nodes at distance @xmath56 between them in a data set .
expert et al .
@xcite used @xmath39 , the population of province @xmath4 , as their measure of node importance .
after briefly experimenting with variations , such as using population density or a logarithm of the population ( i.e. , @xmath57 ) and observing no significant differences in performance , we will follow their lead .
another simple choice is node strength ( i.e. , @xmath58 ) , though the null model then becomes very similar to the usual ng null model @xcite .
moreover , if @xmath43 does not depend on distance , then the null model becomes exactly the ng null model in that case . in most data sets , distances are continuous
, so one needs to bin distance data to obtain enough nodes in each distance bin to construct a meaningful deterrence function @xmath43 in eq .
( [ eq : gravityfd ] ) .
possible binning methods include binning into equal - distance bins ( e.g. , every @xmath59 km ) and equal - sized bins ( e.g. , each bin containing @xmath60 elements ) .
after testing the choice of binning procedure on the benchmark networks and applying the null model to empirical data and observing no qualitative differences in null model performance , we selected the equal - distance method for the rest of the paper , while choosing a bin size large enough that there are always more than 5 elements in each bin . additionally , for the benchmark networks we can test the influence of bin sizes on similarity of algorithmic partitions to the planted community structure
. we will give the specific bin sizes for spatial benchmark and dengue correlation networks in their respective sections .
combining eqs .
( [ eq : gravity ] ) and ( [ eq : gravityfd ] ) allows us to write the gravity null model as @xmath61 expert et al .
used the null model ( [ eq : gravitynm ] ) to uncover a linguistic partition of a network of belgian mobile phone calls into the french and flemish speaking parts of belgium .
this partition was obscured by geographical communities when using the ng null model @xcite . in the present paper
, we generalize the gravity null model to a multislice setting by calculating a separate gravity null model for each layer @xmath10 .
the resulting multislice gravity null model is @xmath62 where we have assumed that the population stays constant across time .
if one has reliable information about changes in population with time , one can incorporate such information into the null model ( [ eq : gravitym ] ) by substituting @xmath54 with an analogous quantity @xmath63 that depends both on the node @xmath4 and on the layer @xmath10 .
gravity models include multiple parameters that one needs to either choose arbitrarily or estimate from data .
moreover , by their design , gravity models are unable to predict different fluxes between locations that are the same distance apart but which have regions with different population densities between them . for example , one would expect a higher flux of infectious disease between two locations that are separated by a space with high population density than between locations that are separated by a space with low population density ( because of the higher availability of susceptible hosts in the latter case ) @xcite .
by contrast , one would expect a smaller commuting flux between such locations in the latter case due to higher availability of nearby jobs , as this reduces peoples willingness to commute for longer distances @xcite .
the radiation model @xcite was developed to attempt to address these issues .
it was designed for population flows and has subsequently been applied successfully in several situations @xcite . because the radiation model is designed to capture human mobility between populations , and the long - distance spread of many infectious diseases including dengue is believed to be largely due to long - distance mobility @xcite , the radiation model might provide a useful but simplified description for the spread of disease across space . in this section ,
we use it to construct a new spatial null model for community detection that we believe might be well - suited for studying the long - distance spread of dengue . the mean commuting flux predicted by the radiation model for locations @xmath4 and @xmath5 with populations @xmath39 and @xmath40 is @xmath64 where @xmath65 is the population between locations @xmath4 and @xmath5 , and @xmath66 is the number of commuters in location @xmath4 . a simple way to calculate @xmath65 is to use the population @xmath67 in the circle of radius @xmath41 centered at @xmath4 and subtract the total of the populations at the origin and destination .
that is , @xmath68 .
although the radiation model is relatively recent @xcite , several modifications to it have already been proposed .
these include incorporating a normalization for finite systems @xcite and the development of a general framework that includes ideas from the radiation , gravity , and intervening - opportunities models @xcite .
we propose a novel null model for community detection based on the original formulation of the radiation model @xcite .
we use a similar formulation to eq .
( [ eq : gravitynm ] ) to incorporate both the expected distance - dependent flux and the actual network structure . to avoid creating a directed network
, we use a symmetrized predicted flux @xmath69 between nodes @xmath4 and @xmath5 . we thereby construct the _ radiation null model _ @xmath70 in section [ section : dengue ] , we will study community structure in empirical data from several years of dengue fever occurrences in peru . because we do not possess detailed data on the commuting patterns in peru ( see the description of our data in section [ data ] ) , we assume that commuters are distributed uniformly across space
. we can then simplify eq .
( [ eq : radiation ] ) by substituting @xmath71 , where @xmath72 is the fraction of the population that commutes . because the quantity @xmath72 is present in both the numerator and denominator of eq .
( [ eq : radiation - null ] ) , we can now cancel it out . however ,
if one possesses commuting data , it would be desirable to use it to improve the radiation null model .
we also extend the radiation null model to a multislice setting in an analogous manner to the gravity null model .
the multislice radiation null model is @xmath73 again , one can incorporate temporal data about population sizes and thereby replace @xmath74 with @xmath75 to improve the null model .
the incorporation of spatial information into null models for community detection is an important problem , and several other ideas have been proposed recently .
for example , cerina et al . @xcite focused on disentangling the correlation between node attributes and space , so they used a simple exponential decay : @xmath76 , where @xmath77 is the mean distance between nodes in a network .
shakarian et al .
@xcite focused on finding geographically - disperse communities , so they introduced a decay constant @xmath78 such that @xmath79 .
another recently - proposed null model was used to attempt to find geographically - proximate communities @xcite .
as the exact nature of the influence of spatial distance on interactions in the dengue fever data is unclear , we decided to focus only on null models that include a contribution from the data , rather than using null models with an arbitrarily chosen functional dependence .
thus , we do not test these null models in the present paper .
to test the performance of the spatial null models , we develop novel synthetic benchmark networks that represent idealized spatially - embedded networks with planted community structure . in what we call the _ distance benchmark _ , the probability of an edge between two nodes depends only on the geographical distance between nodes and on their community assignments .
we assign @xmath0 nodes uniformly at random to positions on the lattice @xmath80 .
we assign a population @xmath39 to each node @xmath4 ( which is an idealized `` city '' ) .
we create two versions of the distance benchmark : the `` uniform population distance benchmark '' and the `` random population distance benchmark '' .
the uniform population version corresponds to the benchmark in expert et al .
@xcite ; we assign the same population ( @xmath81 ) to each node . in the random population benchmark
, we assign an integer population uniformly at random from the set @xmath82 .
we also assign the nodes uniformly at random to one of two communities . in the distance benchmarks ,
the probability @xmath83 that an edge exists between nodes @xmath4 and @xmath5 at distance @xmath41 is inversely proportional to distance : @xmath84 where @xmath18 is the community that contains node @xmath4 and the function @xmath85 if nodes @xmath4 and @xmath5 are in the same community and @xmath86 otherwise .
the `` inter - community connectivity '' @xmath87 controls the degree of mixing between communities . when @xmath88 , only nodes in the same community are adjacent to each other ; when @xmath89 , there are no distinct communities .
the normalization constant @xmath90 ensures that @xmath91 .
we place @xmath92 edges , where there is an edge between nodes @xmath4 and @xmath5 with probability @xmath83 each , and the parameter @xmath93 determines the network s edge density . we interpret multiple edges as weights .
we normalize the weights in the network to @xmath94 $ ] by dividing each entry by the maximum weight in the network . with our _ flux benchmark _ , we aim to mimic the spread of disease on a network .
we allocate its edge weights depending on the mean flux between pairs of nodes that is predicted by the radiation model .
we place @xmath0 nodes uniformly at random on the lattice @xmath80 , and we assign populations and communities in the same manner as for the distance benchmark . again as with the distance benchmark , we consider both uniform - population and random - population versions of the flux benchmark . now , however , the edge probability @xmath95 is directly proportional to the mean predicted radiation - model flux between nodes @xmath4 and @xmath5 ( @xmath96 , which is turn is inversely proportional to distance @xmath41 ) : @xmath97 where @xmath98 is a normalization constant to ensure that @xmath99 . in table [ table - benchmarks ] , we summarize the four synthetic benchmark networks that we have just introduced .
.primary characteristics ( i.e. , population and edge probability ) for the distance and flux benchmarks for static networks .
the quantity @xmath100 signifies that select a number uniformly at random from the set @xmath101 .
additionally , @xmath85 if nodes @xmath18 and @xmath102 are in the same community and @xmath86 otherwise , @xmath41 is the distance between nodes @xmath4 and @xmath5 in space , and @xmath90 and @xmath98 are normalization constants . [
cols="<,<,<",options="header " , ] we create both static ( i.e. , single - layer ) and multilayer benchmarks networks .
the static benchmarks enable us to study the performance of modularity maximization using a chosen null model in a simple setting without the additional complications of a multilayer network . however , the multilayer benchmarks are ultimately more appropriate for disease data because they can incorporate temporal evolution .
we begin by placing nodes in space and assigning populations in the same manner as for the static benchmarks .
we then assign nodes uniformly at random into one of two communities , and we extend this structure into a multilayer planted community structure with @xmath8 layers .
for the `` temporally stable '' benchmarks , the planted community structure is the same for each layer . for the `` temporally evolving '' multilayer benchmarks , we change the community assignment of a fraction @xmath103 of the nodes . for each of these nodes
, we select a new community assignment uniformly at random , and we change the community of the node in each layer ; we start at a layer that we select uniformly at random , and we also change the assignment ( to the same new community ) in all remaining layers .
we then generate the edges for each layer independently , in the same manner as we generate a static benchmark and using identical parameter values @xmath104 for each ; see fig .
[ figure : benchmarks - multislice ] .
independent generation of each layer based on the same starting conditions represents differences between observations due to noise and experimental variation . for each of the above types of multilayer benchmarks
, we set the value of the interlayer edges between corresponding nodes in consecutive layers to be @xmath105 $ ] . thanks to normalizing the intralayer edge weights
, we yield synthetic multilayer benchmark networks in which the relative magnitudes of interlayer edges and intralayer edges are comparable to those in the disease - correlation networks .
each of the reported community detection results for these benchmarks is an average over consensus community detection ( over 50 repeats ) for 50 independently drawn instances of a benchmark with the same values of the same parameter values @xmath104 ,
@xmath106 , and ( when relevant ) @xmath103 .
nodes uniformly at random to positions on a @xmath107 lattice ( which we show in layer 1 ) and divide them into two equal - sized communities ( black and white ) whose nodes we choose uniformly at random .
node @xmath4 has a population of @xmath39 , and each slice has the same set of nodes . for each slice ,
we allocate edges uniformly at random according to a probability distribution that depends on the type of benchmark ; for details , see the text and table [ table - benchmarks ] .
we interpret multiple edges as weights , and we visualize these weights using edge thickness .
we connect copies of nodes in adjacent layers with interlayer edges of weight @xmath108 ( dashed lines).[figure : benchmarks - multislice ] ] we evaluate the performance of the ng , gravity , and radiation null models on our benchmarks by comparing algorithmic partitions with the planted community structure using normalized mutual information ( nmi ) @xcite .
nmi is an information - theoretic similarity measure that is relatively sensitive to small differences in partitions , such as the move of a single node from one community to another , compared to pair - counting measures such as the rand coefficient and z - rand scores @xcite .
this sensitivity makes it suitable for assessing performance on benchmarks that are based on well - defined , ground - truth planted partitions .
nmi is one of many normalized versions of mutual information ( mi ) @xcite .
both mi and nmi are based on the concept of _ information entropy _ , which is a measure of uncertainty .
mi measures the amount of information that one can predict about one random variable ( which in the present paper is a partition of a network into communities ) based on another one . for a partition @xmath109 with @xmath110 communities and a partition @xmath111 with @xmath112 communities , mi is defined as @xmath113\,,\ ] ] where @xmath114 and @xmath115 are the marginal probabilities of observing communities @xmath116 and @xmath117 in partitions @xmath118 and @xmath119 , respectively , and @xmath120 is the joint probability of observing communities @xmath116 and @xmath117 simultaneously in partitions @xmath118 and @xmath119 .
mi takes values between @xmath121 and @xmath122 , where @xmath123 is the entropy of @xmath118 .
normalized mutual information ( nmi ) @xcite is defined as @xmath124\,.\ ] ] the normalization to lie within the range @xmath125 $ ] facilitates interpretation and comparisons .
we use nmi in the following sections , and we obtain the same qualitative conclusions using variation of information @xcite , which is a different normalized measure of similarity
. see appendix [ appendix : vi ] for our comparisons using vi . , @xmath126 and ( right )
@xmath127 , @xmath128 , edge density parameter @xmath129 and uniform populations of @xmath130 for different bin sizes ( colored curves ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : bench - static - even - d - nmi - spacesizevsnm ] ] to emphasize the difference between the gravity and radiation null models , we take @xmath126 and @xmath131 to obtain a relatively densely filled lattice .
( see appendix [ appendix : cities ] for the results for a synthetic network with parameter values @xmath132 and @xmath133 . )
we first compare this benchmark versus a situation with parameter values @xmath134 and @xmath135 ( which are the parameter values that were used in expert et al .
we test varying bin sizes in uniformly - spaced bins using the parameter values @xmath136 ,
@xmath137 and @xmath138 , @xmath135 .
we find that bin width makes a large difference on both benchmarks : @xmath139 produces the highest nmi scores ( i.e. , it has the `` best performance '' ) and increasing bin width leads to a decrease in performance of both spatial null models ( see fig . [ figure : bench - static - even - d - nmi - spacesizevsnm ] ) .
this effect is especially pronounced for the gravity null model . in both cases ,
the best aggregate performance of the spatial null models at optimal bin sizes is similar for @xmath137 and @xmath135 , so we henceforth use the @xmath137 benchmark with @xmath139 to lower computational time and memory usage .
however , one needs to keep the strong influence of bin size on algorithm results in mind for applications .
m0.03| m0.16 m0.16 m0.16 m0.16 pop .
& uniform & uniform & random & random + & distance & flux & distance & flux + ng & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] + grav .
& , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] + rad .
& , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] & , @xmath140 , @xmath129 and ( columns 1 , 2 ) uniform populations of @xmath81 or ( columns 3 , 4 ) populations @xmath39 determined uniformly at random from the set @xmath141 .
we plot nmi for different values of the resolution parameter @xmath22 ( colored curves ) as a function of inter - community connectivity @xmath142 $ ] .
we examine both distance benchmarks ( in columns 1 , 3 ) and flux benchmarks ( in columns 2 , 4 ) .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] , title="fig : " ] + we then study the performance of the three null models using several values of the resolution parameter @xmath143 and the inter - community connectivity @xmath144 on static benchmarks with @xmath126 nodes and lattice size parameter @xmath131 .
smaller values of @xmath22 tend to yield larger communities and vice versa .
considering larger @xmath87 increases the level of mixing between the communities and makes community detection more difficult . for simplicity , we fix the density parameter @xmath129 . as we discuss in appendix [ appendix : mu ] , the value of @xmath145 has little effect on the results of community detection when it is above a certain minimum . for the uniform population distance benchmark ,
the only factor that influences edge placement is the distance between nodes . on this benchmark , the gravity null model has the best performance , as it is able to find the correct partitions for @xmath146 ( see fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] ) .
the radiation null model has the second best performance and is able to find partially meaningful partitions for @xmath147 , above which we observe a plateau of `` near - singleton '' partitions in which most nodes are placed into singleton communities .
( we use the term `` singleton partition '' to refer to a partition in which every node is assigned to its own community . ) the ng null model , which does not incorporate spatial information , does much worse than either of the spatial null models ; it suffers a sharp decline in performance at @xmath148 .
this demonstrates that , although incorporating spatial influence is beneficial for its own sake , we see that using a null model that incorporates population information to study community structure in networks whose structure does not depend on population decreases the performance of community detection .
that is , incorporating spatial information is important , but it needs to be done intelligently . on the uniform population flux benchmark in which we include the population density in the region between two nodes in the flux prediction ( so the population density influences edge structure ) the radiation null model outperforms the other null models .
the gravity null model comes in second place , and the ng null model is a distant third . for the random population distance benchmark , we observe a fast deterioration in quality of the detected communities for @xmath149 for all null models , and all null models reach a `` near - singleton '' regime by @xmath150 .
the ng null model has the best performance among the three null models for @xmath151 . for @xmath152 ,
the gravity null model has the best performance , although the partitions consist largely of singletons for @xmath153 . for the random population flux benchmark
, the radiation null model has the best performance of the three null models .
it has the slowest decrease in nmi scores with the increase in @xmath87 .
the gravity null model has the second - best performance , and ng fails even when there is no mixing between the two communities ( see fig . [ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] ) .
however , even the best performance is much worse on random population benchmarks than it is on the uniform population benchmarks .
note additionally that including population information into the edge placement probability by taking @xmath154 ( `` distance and population benchmark '' ) brings back the advantage for the gravity null model ( see appendix [ app : benchmark - distpop ] ) . among the parameter values
that we consider ( @xmath143 ) , @xmath155 appears to give the best results ( i.e. , the largest nmi scores ) . in the near - singleton regime
, @xmath156 outperforms it slightly ( see fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] ) , however this partition is vastly different from the planted partition .
we now study the influence of the resolution parameters @xmath22 and @xmath108 on the community quality of multilayer benchmarks .
we first study the performance of the ng , gravity , and radiation null models on temporally stable uniform population benchmarks ( see fig .
[ figure : benchmark : multi - gammavsnm ] ) with parameter values @xmath126 ,
@xmath131 , and @xmath157 layers using @xmath143 and @xmath158 .
we expect that for larger @xmath108 values the weight of the interlayer edges outweighs the intralayer edges , leading to each node being assigned to the same community as its copies in other layers .
however , for the temporally stable benchmarks we did not observe this effect ; here , we only show figures for @xmath159 , as different values of @xmath108 give very little difference in results ( in some plots nearly unnoticeable ) .
we also experimented with `` random population '' benchmarks ( see appendix [ appendix : multi - random ] ) and smaller and larger values of @xmath108 .
our results on multilayer benchmarks follow our findings from static benchmarks .
once again , we find that the choice of @xmath22 has a large influence on the quality of the algorithmic partitions , and ( as with our findings for static benchmarks ) @xmath155 seems to yield the best performance ( i.e. , the highest nmi scores ) in most cases , except the near - singleton regime , where @xmath156 outperforms it slightly . for all @xmath4 ) multilayer temporally stable spatial benchmarks with @xmath126 , @xmath131 , @xmath160 , and @xmath129 for @xmath159 and various values of @xmath22 ( colored curves ) as a function of @xmath87 for ( left ) the distance benchmark and ( right ) the flux benchmark .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : multi - gammavsnm ] ] we now examine the nmi between algorithmic versus planted partitions on temporally stable multilayer benchmarks while varying @xmath108 and @xmath87 for fixed @xmath155 . as we show in fig .
[ figure : benchmark : multi - omegavsnm ] , we find that the value of @xmath108 usually has little effect on our ability to detect the planted communities via modularity maximization on benchmarks with a temporally stable community structure .
this suggests that the small interlayer variation due to the independent creation of layers is not enough to observe the influence of @xmath108 on community detection . for all @xmath4 ) multilayer
temporally stable spatial benchmarks with @xmath126 , @xmath131 , @xmath160 , and @xmath129 for @xmath161 and different values of interlayer edge weights @xmath108 ( colored curves ) as a function of @xmath87 for ( left ) the distance benchmark and ( right ) the flux benchmark .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : multi - omegavsnm ] ] we then study the performance of the three null models on temporally evolving uniform population benchmarks ( see fig .
[ figure : benchmark : multi - t - gammavsnm ] ) with parameter values of @xmath126 nodes , a lattice parameter of @xmath131 , a fraction @xmath162 of nodes that change community over the whole timeline , and @xmath157 layers .
we show results for @xmath143 for @xmath159 and for @xmath158 for @xmath161 .
compare fig .
[ figure : benchmark : multi - t - gammavsnm ] to the left panels of figs .
[ figure : benchmark : multi - gammavsnm ] and [ figure : benchmark : multi - omegavsnm ] . on
temporally evolving benchmarks varying @xmath108 makes a difference , where the structures for @xmath163 for the gravity null model and @xmath164 for the radiation null model are the most similar to the planted partitions .
this is in accordance with our expectation that algorithmically detected community structure becomes overly biased towards connecting copies of nodes across layers above a critical @xmath108 value ( which depends on network structure ) .
for all i ) multilayer temporally evolving spatial distance benchmarks with @xmath126 , @xmath131 , @xmath160 , and @xmath129 for ( left ) @xmath159 and different values of the resolution parameter @xmath22 ( colored curves ) and ( right ) @xmath161 and different values of the interlayer weights @xmath165 ( colored curves ) as a function of @xmath87 .
we detect communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : multi - t - gammavsnm ] ] we also perform a `` province - level '' community detection on the multilayer benchmarks in which we seek assignments of nodes ( regardless of what layer they are in ) to communities and compare the results to benchmark networks with planted community structure .
this is analogous to trying to detect community structure in disease data that persists over time e.g. , to seek the influence of climate on disease patterns .
this is easiest to apply to temporally stable multilayer networks .
we successfully detect the underlying communities , and we obtain similar performance results as with the multilayer communities that we discussed above ( see the discussion in appendix [ appendix : bench - regions ] ) .
our results on synthetic benchmark networks suggest that using a spatial null model on a spatial network does not necessarily assure a better result for community detection .
the quality of results with different null models depends strongly on the data and the choice of parameter values .
for example , incorporating population information into a null model in a situation in which the population is not influencing connectivity structure might cause community detection to yield spurious communities ( as we discussed in the context of random population benchmarks ) .
the level of influence of different node properties or events ( such as disease flux on edge placement ) and the extent of mixing between communities is often unknown for networks that are constructed from real data . for such networks ,
we recommend to try both spatial and non - spatial null models over a wide parameter range and to study the results carefully in light of any other known information about the network . in section [ section : dengue ] , we will present an example of using such a procedure to study the community structure of correlation networks that are created from time series of dengue fever cases .
in this section , we assess the performance of the ng , gravity , radiation , and correlation . ] null models on multilayer correlation networks that we construct from disease incidence data that describe the spatiotemporal spread of dengue fever ( a tropical , mosquito - borne viral infection ) in peru from 1994 to 2008 .
disease dynamics are strongly influenced by space , as the distance between regions affects the migration of both humans and mosquitos @xcite .
disease dynamics are also affected by climate due to the temperature dependence of the mosquito life cycle @xcite , and different regions of peru have substantially different climates .
therefore it is important to examine and evaluate the performance of different spatial null models when examining communities in networks that are constructed from disease data .
dengue is a human viral infection that is prevalent in most tropical countries and is carried primarily by the _
_ mosquito @xcite .
the dengue virus has four strains ( denv-1denv-4 ) .
infection with one strain is usually mild or asymptomatic , and it gives immunity to that strain , but subsequent infection with another strain is usually associated with more severe disease @xcite .
although dengue was considered to be nearing extinction in the 1970s , increased human mobility and mosquito abundance have led to its resurgence in many countries often as recurrent epidemics with an increasing number of cases and severity of disease .
dengue is a rising threat in tropical and subtropical climates due to the introduction of new virus strains into many countries and to the rise in mosquito prevalence since the cancellation of mosquito eradication programs @xcite .
it is currently the most prevalent vector - borne disease in the americas @xcite .
peru is located on the pacific coast of south america .
its population of about 29 million people is distributed heterogeneously throughout the country .
the majority live in the western coastal plain , and there are much smaller population densities in the andes mountains in the center and the amazon jungle in the east .
the climate varies from dry along the coast to tropical in the amazon and cold in the andes .
such heterogeneities influence dengue transmission @xcite .
for example , temperature @xcite and rain @xcite affect the life cycle of the main dengue vector _ ae .
aegypti _ , and temperature affects its role in disease transmission @xcite .
the jungle forms a reservoir of endemic disease ; from there , the disease occasionally spreads across the country in an epidemic @xcite .
additionally , as _ ae .
aegypti _ typically only travels short distances @xcite , human mobility can contribute significantly to the heterogeneous transmission patterns of dengue at all spatial scales @xcite .
our dengue data set consists of 15 years of weekly measurements of the number of disease cases across 79 provinces of peru collected by the peruvian ministry of health @xcite between 1994 and 2008 .
these data have previously been analyzed by chowell et al . to study the relationship between the basic reproductive number , disease attack rate , and climate and populations of provinces @xcite . until 1995 ,
the denv-1 strain dominated peru ; it mostly caused rare and isolated outbreaks @xcite .
the denv-2 strain was first observed in 19951996 , when it caused an isolated large epidemic @xcite .
denv-3 and 4 entered peru in 1999 and led to a countrywide epidemic in 20002001 @xcite , and there was subsequent sustained yearly transmission @xcite .
the data contains a total of 86,631 dengue cases ; most of them are in jungle and coastal provinces ( 47% and 49% , respectively ) , and only 4% of the cases occur in the mountains .
the disease is present in 79 of the 195 provinces across the data set , and never in all 79 provinces at once . in this paper
, we use the definition of `` epidemic '' from the us agency for international development ( usaid ) : an _ epidemic _ occurs when the disease count is two standard deviations above the baseline ( i.e. , mean ) @xcite .
when stating countrywide epidemics , we apply this definition when considering all nodes . when stating local epidemics , we apply this definition to individual provinces ( though one could also consider particular sets of provinces ) . our data set @xmath166 consists of @xmath167 time series of weekly disease counts @xmath168 over @xmath169 weeks .
the quantity @xmath170 denotes the number of disease cases in province @xmath4 at time @xmath171 .
( see fig .
[ figure : multislice ] for a plot of the number of cases versus time . )
we create networks from this data by calculating the pearson correlation coefficient between each pair of time series .
we seek to study the temporal evolution of the correlations by constructing separate networks for different time windows we either construct a set of static networks or a multislice network .
to create these networks , we divide each of the time series into @xmath8 time windows by explicitly defining a list of the starting points @xmath172 for each time window and the time window width @xmath173 . in the present paper , we use @xmath174 unless we state otherwise .
the starting point @xmath175 and window width @xmath176 define a time window that we use to select a portion of the disease time series .
for example , for the time series of disease cases in province @xmath4 , the time - series portion @xmath177 represents the numbers of disease cases in province @xmath4 at times @xmath178 . by considering all provinces
, one can use such time series either to construct a set of static networks or a multislice network . for a static network ,
we define a set of @xmath0 nodes @xmath179 , where node @xmath4 corresponds to province @xmath4 .
the edge weight @xmath180 represents the similarity between the time series @xmath181 and @xmath182 ; the kronecker delta @xmath183 removes self - edges .
the quantity @xmath184 is the pearson correlation coefficient between the disease time series for provinces @xmath4 and @xmath5 .
that is , @xmath185 where @xmath186 indicates averaging over the time window under consideration , and @xmath187 is the standard deviation of @xmath181 .
our construction yields a fully connected ( or almost fully connected ) network @xmath2 with elements @xmath188 $ ] . when studying static networks , we use @xmath189 to form a set of @xmath190 overlapping static networks .
to construct a multislice network , we use the times @xmath191 to create nonoverlapping time windows .
the intralayer edge weights are @xmath192 for each layer @xmath10 .
we connect each node @xmath4 in the @xmath11th time window to copies of itself in an adjacent time window @xmath10 using interlayer edges of uniform weight @xmath193 .
this yields a weighted multislice correlation network .
the case @xmath194 in the multislice network corresponds to a sequence of static networks .
see fig .
[ figure : multislice ] for a schematic that shows the construction of a multislice network .
similar constructions of ( both static and multislice ) networks from time series have been employed for systems such as functional brain networks @xcite , currency exchange - rate networks @xcite , and political voting networks @xcite . and @xmath195 .
( the time points correspond to 1/1/1994 , 27/12/1997 , 22/12/2001 , and 17/12/2005 ) .
the nodes represent provinces and each intralayer edge weight is given by a pearson correlation between a pair of single - province time series in a given time window .
one set of correlations gives one temporal layer , and we connect copies of each node in neighboring layers using interlayer edges of uniform weight @xmath196 ( dashed lines ) .
the case @xmath194 yields a set of static networks .
( all other aspects of our network construction are the same . )
[ figure : multislice ] ] many features , such as the number of layers and the mean and variance of the pearson correlation values , depend on the parameters that we use in constructing our networks .
for example , it is important to consider the choice of the time window size @xmath176 .
there is a trade - off between having many layers to obtain a good temporal resolution of events and ensuring that we construct each layer using enough time points to be confident of the statistical significance of the similarity values in the adjacency - tensor layer @xcite .
larger values of @xmath176 yield smaller variations in mean correlation across the years and lessen the effects of small , regional epidemics on the number of cases and on the correlation between disease profiles in different provinces .
therefore , we want to use a sufficiently large value of @xmath176 so that we can examine long - term , repetitive disease patterns .
studies based on random matrix theory ( rmt ) suggest an additional constraint of @xmath197 , because correlation matrices generated from time series that are shorter than the number of time series being analyzed ( i.e. , than the number of nodes ) are indistinguishable from the correlations that one calculates from short , uncorrelated sequences of noise @xcite .
however , choosing a value of @xmath176 that is too large risks over - smoothing data and losing important information . unless we state otherwise , we use @xmath198 for the ( overlapping ) static networks and @xmath199 for multilayer networks ( which never have more than 59 nodes in one slice ) in order to obtain meaningful correlation matrices while preserving interesting disease patterns .
it is well - known that geographical distance has an important influence on disease spread @xcite .
additionally , climate exerts a significant influence on dengue , and it is also necessary to consider peru s particular topography ( as its mountains form a barrier to disease spread ) @xcite .
therefore , we expect the community structure in the disease - spread networks to be strongly geographical .
we also expect to observe large changes in community structure at certain time points such as when the introduction of the new disease strains around 1999 led to large epidemics and the onset of yearly countrywide epidemics @xcite . in this section
, we explore the similarity of algorithmically obtained community structures to spatial and temporal groupings of nodes across a range of parameter values .
to compare the algorithmic partitions of the correlation networks versus manual partitions , we use the z - score of the rand coefficient @xcite .
the rand coefficient is @xmath200 where @xmath201 is the number of node pairs that are in the same community in both partitions , @xmath202 is the number of node pairs that are in different communities in both partitions , and @xmath203 is the total number of node pairs .
we use z - rand scores instead of nmi because the former measure is good at detecting similarities in coarse structure @xcite but is less sensitive to minor changes such as one node changing community assignment . for the disease data
, we do not possess ground - truth partitions as we did for our synthetic benchmark examples , so we seek to evaluate broad organizational similarities in the algorithmic and manual partitions rather than attempting to conduct a fine - grained evaluation of community structure versus a planted partition .
we thereby aim to inform our understanding of the general structural influences on the spatiotemporal patterns of disease spread .
one can also examine measures of spatial autocorrelation ( e.g. , moran s @xmath204 ) @xcite . to examine the spatial community structures in the static and multilayer networks ,
we compare the results of the partitions that we obtain algorithmically to manual partitions using z - rand scores . in the `` climate partitions '' , we group nodes according to the topography of their associated provinces jungle , coastal , and mountainous provinces and then subsequently divide the coastal and mountainous communities into northern , central , and southern provinces [ see fig .
[ figure : peru - regions](a , b ) ] .
we use the detailed climate partition for the subsequent study . in the 19-community `` administrative partition '' , we assign each node to its associated administrative region [ see fig .
[ figure : peru - regions](c ) ] .
we compare each of the 700 static networks against these two manual partitions to study the spatial element of the data . for the multilayer networks ,
we compare the algorithmic partition versus a manual partition by taking the same manual partition of nodes for all layers .
\(a ) , title="fig : " ] ( b ) , title="fig : " ] ( c ) , title="fig : " ] we use the term `` spatial partitions '' to describe partitions that yield high z - rand scores in comparison to the climate or administrative manual partitions . for multilayer networks
, we also compare the algorithmic partitions to partitions that contain a planted temporal change in community structure . for these comparisons , we group the multilayer nodes into ones that occur before or after a `` critical '' time @xmath205 , and we use the term `` temporal partitions '' to describe partitions that yield high z - rand scores in this comparison .
such temporal partitions contain two communities .
we test all of the times @xmath191 that we use to create the multilayer network , and we report the time with the highest z - rand score as the critical time @xmath205
. we also test for pairs of critical times ( yielding a partition into three communities ) by examining all possible pairs of critical times in the same manner . before looking at multilayer networks , we first study the community structures of the @xmath206 overlapping static networks formed by taking @xmath207 and using @xmath198 .
we select the networks for which the algorithmic partitions score the highest against manual spatial partitions of the network for further study .
the community structures that we obtain from maximizing modularity have a strong spatial element , as suggested by the high z - rand scores when compared to topographical partitions .
as one can see in fig .
[ figure : ng - static - boxes](a ) , which shows a compact box plot of the z - rand scores versus climate partitions for resolution parameter values of @xmath208 ( each box ) across the 700 networks covering the data set ( the horizontal axis ) , the spatial element is especially evident after the year 2000 . as one can see from a plot of number of epidemic cases over time ( see fig .
[ figure : multislice ] ) , this transition seems to occur around the time of the largest countrywide epidemic in the data , and the subsequent period includes recurring yearly epidemics that have been linked to climatic patterns in prior studies @xcite .
there are two periods of significantly spatial partitions : one corresponds to the 20002001 epidemic and it contains the spatial partition with the highest z - rand score against climate [ see fig .
[ figure : ng - static - boxes](b ) ] , and the second occurs in 20022004 and it contains the spatial partition with the highest z - rand score against administrative partition [ see fig . [
figure : ng - static - boxes](c ) ] . note that the topographical z - rand scores decrease after 2004 despite the continuing yearly dengue epidemics .
\(a ) ( b ) ( c ) in figs .
[ figure : ng - static - boxes](b , c ) , we plot the partitions that have the highest z - scores with respect to the manual climate and administrative partitions on a map of peru .
we observe that the high - scoring climate partition consists of one community that is dominated by the jungle ( red community ) and another community that is dominated by the coast ( pink community ) , whereas the high - scoring administrative partition is composed of 7 smaller communities .
the jungle nodes form the largest communities in both of these spatial partitions , and it is these communities that contribute the most to the high spatial scores . there was a dengue epidemic in most of the provinces in these large communities during the time periods covered by the relevant networks .
it is possible that their proximity drove the large amount of synchrony in the epidemic spread in these provinces .
we now consider community structure in the multilayer disease network with nonoverlapping layers that we construct using the time points @xmath209 and the time window width @xmath199 . to find interesting parameter regimes , we compare the algorithmically computed community structure of the dengue fever multilayer disease - correlation network to manual partitions across a range of @xmath108 and @xmath22 parameter values between @xmath121 and @xmath210 ( see fig .
[ figure : ngog ] ) . for @xmath211 ,
all nodes are in one community . for @xmath212
$ ] and @xmath213 , the algorithmically detected partitions have a relatively high z - rand score when compared to the temporal partition [ see fig . [
figure : ngog](c ) ] . when looked at in detail
, the partitions exhibit a mixture of spatial and temporal features .
[ see fig .
[ figure : ngog](a ) for an example . ]
\(a ) ( b ) ( c ) when studying the qualitative features of the partitions for @xmath214 $ ] ( where the endpoints of this interval are approximate ) and @xmath213 , we observe that community detection repeatedly finds 2001 as the single critical time , and 2001 and 2005 as the most common pair of critical times @xmath205 ( i.e. , the strongest change points in temporal community structure ) , which agrees with the visual observations .
this finding suggests that a strong shift in the patterns of disease correlations occurred around these times .
indeed , peru experienced a large countrywide dengue epidemic in 20002001 , and this period also marks the onset of new yearly epidemic dynamics @xcite .
thus , our method recovers the most important biological event in this data set in addition to providing additional information about spatial influences on disease spread .
we also observe several other time when new communities are born , but we do not know the biological significance of these dates .
notably , in this parameter regime , our community structure does not identify the large epidemic in the jungle utcubamba province in 1996 ( see fig .
[ figure : multislice ] ) , which is the other large event in this data set .
the community structure that we detect depends heavily on parameter values . in many parameter regimes
especially when @xmath215 and @xmath216 communities appear to be predominantly spatial , and we find high z - rand scores when compared to the climate and administrative partitions [ see fig . [ figure : ngog](b ) ] .
the high influence of spatial proximity on the community structure is unsurprising , as spatial distance is an important influence on disease spread @xcite .
previous studies have also noted that the community structure of spatial networks obtained by maximizing modularity using the ng null model tends to be strongly influenced by geographical factors @xcite .
if there are other interactions that shape the dengue fever correlation network , they might be obscured by the strong influence of spatial proximity .
however , such interactions might be revealed by using a spatial null model that incorporates the expected effect of space on interactions .
we pursue this idea in section [ space ] .
we apply spatial null models to the dengue fever correlation networks .
we obtained province locations from the geonames.org website @xcite , and we obtained their populations from the peruvian instituto nacional de estadstica e informtica ( inei ) @xcite .
we were only able to obtain the 1994 and 2007 populations ; due to the limited range of data and the several changes in peruvian administrative structures between the two times , we only use the 2007 populations .
the maximum inter - province distance is about 1300 km .
we report numerical experiments using a bin size of 400 km after testing the spatial deterrence for several other sizes ( ranging between 50 and 500 km ) in the same manner as in ref .
@xcite : that is , we study the shape of the deterrence function [ see eq .
[ eq : gravitymodel ] and the nearby discussion ] with changing distance across bin sizes , and we then examine the community structures that we obtain using different bin sizes .
we find that bin sizes have an effect on the shape of the deterrent function ( with lower sizes giving smoother results ) , but all of the bin sizes that we tested produced very similar partitions for both the gravity and radiation spatial null models .
we selected the smallest bin size that guaranteed more than 5 nodes in each bin .
recall from section [ data ] that only 79 of the 195 provinces had reported cases of dengue fever in our data , so we use the location and population data only for those provinces .
we first study the community structure on static disease - correlation networks using the gravity and radiation null models .
both null models seem to remove most of the spatial element of the community structures ( including temporal variation in the spatial correlations ) , as indicated by low values and low variation of spatial z - rand scores ( not shown ) .
for both the gravity and radiation null models , we observe high similarity between layers for a variety of values of the resolution parameter @xmath22 [ see fig . [
figure : spatial - static - boxes](a ) ] .
these structures contain one dominant community with the majority of nodes present at any given time , and several singleton communities [ see fig .
[ figure : spatial - static - boxes](b ) ] . by examining the partitions directly using a map of peru , we see that the singleton communities tend to consist of the highest - populated nodes .
\(a ) ( b ) ( c ) we also examine the spatial null models for multilayer correlation networks .
the community structures again exhibit one large community containing the majority of multilayer nodes [ see fig .
[ figure : grog](a , b ) ] , and several multilayer nodes corresponding to provinces with highest populations form singleton communities across time .
this situation occurs for all of the tested parameter values .
additionally , we do not observe any clear pattern in the z - rand scores as we change @xmath22 and @xmath108 .
\(a ) across 100 repeats of community detection .
we use a resolution - parameter value of @xmath161 and consider @xmath217 .
mid - layer timepoints are plotted on the horizontal axis , and nodes are on the vertical axis .
node community membership is indicated by color .
[ figure : grog],title="fig:",height=151 ] ( b ) across 100 repeats of community detection .
we use a resolution - parameter value of @xmath161 and consider @xmath217 .
mid - layer timepoints are plotted on the horizontal axis , and nodes are on the vertical axis .
node community membership is indicated by color .
[ figure : grog],title="fig:",height=151 ] our findings suggest that the addition of space into a null model for modularity optimization might remove the majority of the variation in the correlation structure of the dengue fever correlation networks , such that the influence of population size could be the only major factor that remains .
this could relate to the concept that a minimum population size is required for sustained disease transmission ; it has been estimated that this size is between 10,000 and 500,000 for dengue @xcite .
there are only 5 provinces with populations over 500,000 , and these provinces are often assigned to singleton communities when we use a spatial null model .
this suggests that they have different disease patterns from the other provinces .
recently , macmahon et al .
@xcite proposed a new null model that they designed specifically for modularity maximization for networks that are constructed based on the pairwise pearson correlations between time series .
they used ideas from rmt @xcite to generate a null model that represents the `` random '' component of a correlation matrix and can take into account the single most strongly influencing factor on the correlation structure . in the context of financial systems , which was the focal example of ref .
@xcite , this factor is often called a `` market mode '' .
given that we often found a single large community when we used spatial null models , it is interesting to see what results we obtain using such a correlation null model . to use a correlation null model , we need to construct our network directly from pairwise correlations without subsequently shifting them to @xmath125 $ ] and removing self - edges .
we construct networks by selecting time windows and calculating pearson correlations in the same manner as in section [ sec : netcreate ] , but the here edge weights are left as raw correlations : @xmath218 ( eq . [ equation : pearson ] ) .
because of the special structure of correlation matrices , modularity using the standard ng null model assigns importance to pairs of nodes @xmath4 and @xmath5 whose pearson correlation is larger than the product of the correlations of each node with the time series of the total number of disease cases in the country over the chosen time window : @xmath219 , where @xmath220 .
by contrast , the correlation null model that we adopt from ref .
@xcite uses ideas from rmt to detect communities of nodes that are more connected than expected under the null hypothesis that all time series are independent of each other . for a given correlation matrix constructed from @xmath0 time series that each have length @xmath221 ( with @xmath222 ) , one posits based on rmt that any eigenvalues that are smaller than the eigenvalue @xmath223 are due to noise . here
, @xmath224 is the maximum eigenvalue predicted for a correlation matrix that is constructed from the same number of entirely random time series .
additionally , for many empirical correlation matrices , the largest eigenvalue @xmath225 is much larger than the others , and its corresponding eigenvector has all positive signs @xcite . in this situation , there is a common factor , which is called the `` market mode '' in financial applications , that influences all of the time series @xcite .
we can thus decompose our correlation matrix @xmath226 as follows : @xmath227 , where @xmath228 is the `` random '' component of the matrix , @xmath229 is the `` market mode '' , and the `` group mode '' @xmath230 is embodies the meaningful correlations between time series .
we write @xmath231 and @xmath232 , where @xmath233 and @xmath234 are an eigenvalue and its corresponding eigenvector , @xmath235 is the outer product of the two vectors ( a special case of the kronecker product for matrices ) , and @xmath225 is the maximum observed eigenvalue in the correlation matrix @xmath226 .
we can construct a correlation null model either by removing both the `` random '' component of the matrix and the influence of the `` market mode '' ( i.e. , by using the null model @xmath236 ) or by only removing the random component ( i.e. , by using the null model @xmath237 ) . to satisfy the @xmath222 requirement to applying the rmt approach of ref .
@xcite , we require @xmath238 . for subsequent calculations , we use @xmath198 for static networks , and @xmath199 for multilayer networks ( which have a maximum of 59 nodes per slice , as not all provinces experience disease at the same time ) , unless stated otherwise .
although our maximum eigenvalue is larger than the other eigenvalues and every component of the associated eigenvector is positive , the eigenvector does not appear to affect all nodes to the same extent .
the above construction thus yields a non - uniform null model for our data in practice , so we are unable to identify the analog of a market mode .
we thus do not incorporate such a mode into the null models that we employ for community detection .
we use the correlation null model @xmath239 where @xmath22 is the resolution parameter . for the multilayer setting , we write @xmath240 where @xmath241 and @xmath242 are an eigenvalue and its corresponding eigenvector for layer @xmath10 .
we test the performance of this correlation null model on correlation networks that we construct from dengue fever time series with @xmath198 . in most of the static networks ,
the community structures appear to be affected by spatial proximity especially for post-2000 networks , as illustrated by the high z - rand scores versus the climate partition ( particularly in 19951996 , 20002001 , 20032004 , 20052006 ) .
see fig .
[ figure : temporal - static - boxes](a ) .
these high z - rand scores result from ( 1 ) the classification of the majority of jungle provinces into one community and ( 2 ) the existence of a community that contains many of the northern coastal provinces [ see figs . [
figure : temporal - static - boxes](b , c ) ] .
\(a ) ( b ) ( c ) we also perform community detection on multilayer networks using the correlation null model @xmath243 for @xmath244 \times [ 0.1,3]$ ] .
we obtain partitions with a mixture of temporal and spatial features .
we calculate a consensus community structure across 100 realizations for each value of @xmath22 and @xmath108 . in fig .
[ figure : tempog](a ) , we show the best - looking partition , which we obtain for @xmath161 and @xmath217 .
this partition includes 9 communities .
although several communities coexist in each layer , the primary divisions appear to be largely temporal .
for example , community 2 shrinks after layer 6 ( january 2001 ) . however , the highest z - rand score versus a temporal partition ( with either one or two critical times ) for this partition is one that has a single critical time @xmath205 at the end of layer 7 ( i.e. , in july 2002 ) .
that is , this manually - constructed temporal partition has one community that contains layers 17 and a second community that contains layers 812 .
we obtain a temporal partition with one critical time for @xmath211 and @xmath245 .
the other parameter regimes have different critical times . for @xmath246 and
the allowance of a pair of critical times , we obtain the highest temporal z - rand score when the critical times occur immediately after layer 6 ( i.e. , january 2001 ) and immediately after layer 9 ( i.e. , march 2004 ) . see fig .
[ figure : tempog](d ) . for @xmath247
, we obtain the highest temporal z - rand scores with a pair of critical times when those times occur immediately after layer 5 ( i.e. , june 1999 ) and immediately after layer 9 .
( note that these z - rand scores tend to be lower than those for @xmath246 . )
see fig .
[ figure : tempog](e ) . in our sweep over the different values of @xmath22 and @xmath108
, we obtain lower climate z - rand scores compared to what we obtained using the ng null model ( compare fig .
[ figure : tempog](b ) with fig .
[ figure : ngog](b ) ) .
we do not observe any clear patterns in the spatial z - rand scores as we vary @xmath22 and @xmath108 .
( a ) ( b ) ( c ) + ( d ) ( e ) we now examine the province - level information that we can glean from the data .
the simplest approach is to construct a single static network from the entire length-@xmath221 time series , but our multilayer approach allows us to aggregate data less severely .
this , in turn , allows us to lose less information .
when we aggregate all time series to construct a single similarity network ( i.e. , we choose @xmath248 and @xmath249 ) , we find that the community structures that we obtain via modularity maximization with the spatial and correlation null models all consist of a single large homogenous community with up to three outlier nodes ( see fig .
[ figure : dengue - fullyaggregated ] in appendix [ appendix : regions ] ) . only the ng null model is able to detect meaningful - looking communities , especially for @xmath155 and @xmath250
[ see fig .
[ figure : dengue - regionlevel - map](a ) ] . for @xmath251 ,
the we find three communities ; one is a singleton , and the middle one consists almost exclusively ( 15 of 17 nodes ) of northern coastal provinces .
this partition has z - rand score versus climate of 7.3 . for @xmath250 , using the ng null model yields 28 communities , and many of them are small .
nodes grouped in the community of northern coastal provinces are the provinces of peru that were most strongly involved in the 20002001 dengue epidemic ; 15 nodes in this community experienced this epidemic , whereas only two other nodes experienced it . the data aggregation over the whole time series results in the 20002001 epidemic dominating all other events in the time series .
if we use the community structure of the temporally evolving multilayer network to create the province - level structure , we might be able to shed some more light on other interactions between provinces .
( a ) ( b ) ( c ) ( d ) we then study the structure of province - level communities that we obtain from community detection using the uniform null model on an association matrix @xmath33 . as we discussed in section [ section : networks ] , we create this matrix by counting the number of multilayer nodes that are classified together in a consensus community detection on a multislice network .
we consider the parameter values @xmath155 and @xmath252 over 100 repeats . comparing the province - level communities that we obtain using the ng and correlation null models versus the broad topographical categories of coast , mountain , and jungle reveals large - scale climatic influence on disease patterns .
the two null models yield similar results : more than 40 nodes are grouped into one large community that includes central coast , northwestern and southern jungle , and eastern jungle ; and coastal north nodes form smaller , strongly spatial communities [ see fig .
[ figure : dengue - regionlevel - map](c , d ) and fig . [ figure : dengue - regionlevel - climates ] ] . when we study the disease time series of the provinces grouped into the province - level communities , we observe distinct types of disease incidence patterns .
the ng null model finds one more pattern type ( nodes with late onset of disease , as we illustrate in fig .
[ figure : dengue - regionlevel - timeseries ] ) than the correlation null model .
\(a ) ( b ) \(a ) ( b )
in conclusion , we examined time - dependent community structure and the effect of different null models including ones that incorporate spatial information on the results of modularity maximization .
we conducted our computational experiments using novel synthetic benchmark spatial networks and correlation networks constructed from spatiotemporal dengue fever incidence data in provinces of peru ( a system that is strongly influenced by spatial effects ) .
we compared our results for the standard newman - girvan null model versus two null models that incorporate spatial information : a gravity null model @xcite and a novel radiation null model .
we also compared the ng null model on disease - correlation networks with a recently - developed correlation null model ( and a multilayer generalization of it ) that is designed specifically for studying correlation networks that are derived from time series @xcite .
our results indicate that it is very important to incorporate problem - specific information such as spatial information into the null models for community detection .
our results also illustrate that there are many nuances to consider .
that is , it is not simply a matter of incorporating spatial information in an arbitrary way but rather developing spatial null models that are motivated by application - appropriate generative models .
for example , the ng null model performs better than the spatial null models ( which both use population data ) on the random population distance benchmark where populations vary but edge weight does not depend on them .
however , when we remove the variation in population or modify the benchmark to include population in edge placement probabilities , we find that the gravity null model performs best ( as expected ) .
parameter choices can also be extremely important , as demonstrated by the large influence of bin size ( when binning distances for the spatial null models ) on community detection results , the failure to find meaningful communities with any of the null models at low edge densities , and the strong influence of resolution parameter @xmath22 on the results . to summarize
, one needs to consider seriously what variables that influence the connections in a system of interest one wants to include in a null model , be careful about including spurious variables , and test how the results change for many parameter values .
finally , not incorporating space at all can be more appropriate than incorporating it in a manner that is overly naive .
( see , for example , our results on the random population benchmarks . ) in our consideration of dengue fever data , we observed for static networks that the ng and correlation null models find structures that are strongly spatial especially after the onset of yearly epidemics in 2000 . in our study , we observed that spatial partitions are often dominated by large communities of neighboring jungle nodes that experience local epidemics during the time window . on a multilayer network , maximizing ng modularity can result in either spatial or temporal partitions ( depending on the parameter regime ) .
temporal partitions successfully find the most important time point in the history of the disease namely , the introduction of a new disease strain that caused a large epidemic in 20002001 and a subsequent shift in disease patterns and several other potentially interesting time points and periods of high spatial correlation .
when studying province - level connectivity , we illustrated that consensus province - level communities from an association matrix that we constructed from the multilayer network across time is a far preferable approach to complete data aggregation .
for the aggregation into a static network , maximizing modularity using any of the test null models except the ng null model failed to detect any meaningful communities ; the ng community structure corresponds to the large 20002001 epidemic . aggregating networks results in loss of information that is desirable to study for meaningful patterns @xcite .
when we constructed multilayer networks and computed consensus communities , the computed `` spatial '' multilayer partitions and province - level partitions highlight the importance of climate to the disease patterns of dengue , as the jungle provinces are placed into distinct communities from most mountainous and coastal provinces .
this is sensible , as the yearly epidemic patterns tend ( on average ) to exhibit an earlier epidemic onset in the jungle @xcite and the jungle climate is rather distinct from the climate in coastal and mountainous provinces .
the main climatic difference between jungle provinces and other provinces is temperature , and the influence of temperature on dengue transmission @xcite and attack rate and persistence has been documented @xcite .
the province - level communities that we detect using both the ng and the correlation null models yield distinct temporal disease incidence patterns .
the ng null model also finds one additional jungle community ( with a late disease onset ) than the correlation null model .
the assignment of different jungle nodes into separate communities hints that the variables that influence jungle epidemics may be different than those in other climates .
moreover , the variability in disease patterns between jungle provinces is high , perhaps due to the year - round disease presence ( in contrast to the existence of a summer disease season on the coast ) .
chowell et al .
@xcite reported that the coastal and mountainous provinces exhibit more spatial heterogeneity of disease incidence than the jungle provinces , and population size appears to play a larger role in disease persistence in the jungle . additionally , the jungle climate is more homogenous ( especially in the north - south direction ) than the other two climates .
when we attempt to remove the influence of space by using the gravity and radiation null models , we obtain one large community that contains all but the highest - population provinces ( which are assigned to singleton communities ) .
in contrast to the linguistic example in ref .
@xcite , this suggests for our disease networks that the incorporation of space into the null model accounts for the majority of the structure present in the network .
the spatial structure that we removed likely includes the structure that corresponds to the climate variation that causes different epidemic patterns in the jungle , coastal , and mountainous provinces .
the only variable that we were able to identify as influencing community structure when using spatial null models is province population : the highly populated ( and typically coastal ) provinces forming singleton communities .
these highly populated provinces are likely to be economic centers , with many people traveling there from the other provinces and thereby transmitting the disease @xcite .
these provinces could then be the seeds of epidemics for the other coastal and mountainous provinces , and two studies have in fact reported ( so - called ) `` hierarchical '' transmission of dengue from populous regions to those with low populations in both peru and thailand @xcite .
this situation could lead to high correlations across atypically long distances compared with the majority of the data , which could in turn cause populous provinces to be assigned to singleton communities .
additionally , it is known that population size influences dengue transmission : the basic reproductive number @xmath253 and disease persistence ( i.e. , the fraction of weeks with disease cases ) are positively correlated with population size , and the attack rates are negatively correlated with it @xcite .
the incorporation of spatial information into null models for community detection is both interesting and desirable . as we have illustrated in the present paper , however , there are many nuances that it is important to consider .
we have also demonstrated that it is important to develop null models that incorporate generative mechanisms for human mobility and flux .
we similarly expect that domain - dependent , mechanistic null models will also be crucial in many other applications .
ms was supported by the epsrc systems biology doctoral training centre .
gc acknowledges support from grant number 1r01gm100471 - 01 from the national institute of general medical sciences ( nigms ) at the national institutes of health and the multinational influenza seasonal mortality study ( misms ) led by the fogarty international center , national institutes of health @xcite .
map was supported by the european commission fet - proactive project plexmath ( grant no .
317614 ) and a grant ( ep / j001759/1 ) from the epsrc .
we thank marya bazzi , paul brodersen , andrew elliott , paul expert , lucas jeub , heather harrington , felix reed - tsochas , and jess san martn for helpful discussions .
we thank vittoria colizza and chiara poletto for several extensive discussions and comments on the manuscript , and we thank andrew elliott for reading and commenting on this manuscript and for his help with creating the figures .
additionally , we thank yulian ng and tom prescott for their contributions at early stages of this project .
the dengue fever data was collected by the peruvian ministry of health @xcite .
we obtained province locations from the geonames.org website @xcite , and we obtained their populations from the peruvian instituto nacional de estadstica e informtica ( inei ) @xcite .
99 anselin , l. ( 1995 ) local indicators of spatial association lisa .
_ , * 27*(2 ) , 93115 .
balcan , d. , colizza , v. , gonalves , b. , hu , h. , ramasco , j. j. & vespignani , a. ( 2009 ) multiscale mobility networks and the spatial spreading of infectious diseases .. _ proc .
usa _ , * 106*(51 ) , 2148421489 .
balcan , d. , hu , h. , goncalves , b. , bajardi , p. , poletto , c. , ramasco , j. j. , paolotti , d. , perra , n. , tizzoni , m. , van den broeck , w. , colizza , v. & vespignani , a. ( 2009 ) seasonal transmission potential and activity peaks of the new influenza a(h1n1 ) : a monte carlo likelihood analysis based on human mobility .. _ bmc medicine _ , * 7*(1 ) , 45 .
barrat , a. , barthelemy , m. & vespignani , a. ( 2008 ) _ dynamical processes on complex networks_. cambridge university press .
barthlemy , m. ( 2011 ) spatial networks .
_ , * 499*(1 - 3 ) , 1101 .
bassett , d. s. , owens , e. t. , porter , m. a. , manning , m. l. & daniels , k. e. ( 2014 ) practical methods for the examination of force chain network architecture in granular materials . in preparation .
bassett , d. s. , porter , m. a. , wymbs , n. f. , grafton , s. t. , carlson , j. m. & mucha , p. j. ( 2013 ) robust detection of dynamic community structure in networks .. _ chaos _ , * 23*(1 ) , 013142 . bassett , d. s. , wymbs , n. f. , porter , m. a. , mucha , p. j. , carlson , j. m. & grafton , s. t. ( 2011 ) dynamic reconfiguration of human brain networks during learning .. _ proc .
natl . acad .
, * 108*(18 ) , 76417646 .
bazzi , m. , porter , m. a. , mcdonald , m. , fenn , d. j. , williams , s. & howison , s. d. ( 2014 ) community structure in multilayer networks of financial - asset correlations . in preparation . berger - wolf , t. y. & saia , j. ( 2006 ) a framework for analysis of dynamic social networks . in _
kdd 06 : proceedings of the 12th acm sigkdd international conference on knowledge discovery and data mining _ , pages 523528 .
blondel , v. d. , guillaume , j .- l . , lambiotte , r. & lefebvre , e. ( 2008 ) fast unfolding of communities in large networks .
_ j. stat .
_ , * 2008*(10 ) , p10008 .
boccaletti , s. , bianconi , g. , criado , r. , del genio , c. i. , gmez - gardees , j. , romance , m. , sendia - nadal , i. , wang , z. & zanin , m. ( 2014 ) the structure and dynamics of multilayer networks .
arxiv:1407.0742 ( physics reports , in press , doi:10.1016/j.physrep.2014.07.001 ) .
brandes , u. , delling , d. , gaertler , m. , gorke , r. , hoefer , m. , nikoloski , z. & wagner , d. ( 2008 ) on modularity clustering .
_ ieee transactions on knowledge and data engineering _ , * 20*(2 ) , 172188 .
cerina , f. , de leo , v. , barthelemy , m. & chessa , a. ( 2012 ) spatial correlations in attribute communities .. _ plos one _ , * 7*(5 ) , e37507 .
chan , s .- y . , hui , p. & xu , k. ( 2009 ) community detection of time - varying mobile social networks . in zhou , j. , editor ,
_ complex sciences _ , volume 4 of _ lecture notes of the institute for computer sciences , social informatics and telecommunications engineering _ , pages 11541159 .
springer berlin heidelberg .
chen , y. , kawadia , v. & urgaonkar , r. ( 2013 ) detecting overlapping temporal community structure in time - evolving networks .
arxiv:1303.7226 .
chowell , g. , cazelles , b. , broutin , h. & munayco , c. v. ( 2011 ) the influence of geographic and climate factors on the timing of dengue epidemics in per , 1994 - 2008 .. _ bmc infectious diseases _ , * 11*(1 ) , 164 .
chowell , g. & sanchez , f. ( 2006 ) climate - based descriptive models of dengue fever : the 2002 epidemic in colima , mexico .. _ j. environ . health _
, * 68*(10 ) , 4044 , 55 .
chowell , g. , torre , c. a. , munayco - escate , c. , surez - ognio , l. , lpez - cruz , r. , hyman , j. m. & castillo - chavez , c. ( 2008 ) spatial and temporal dynamics of dengue fever in peru : 1994 - 2006 .
_ epidemiol .
_ , * 136*(12 ) , 16671677 .
colizza , v. , pastor - satorras , r. & vespignani , a. ( 2007 ) reaction diffusion processes and metapopulation models in heterogeneous networks . _ nature physics _ , * 3*(4 ) , 276282 .
cummings , d. a. t. , irizarry , r. a. , huang , n. e. , endy , t. p. , nisalak , a. , ungchusak , k. & burke , d. s. ( 2004 ) travelling waves in the occurrence of dengue haemorrhagic fever in thailand .. _ nature _ , * 427*(6972 ) , 344347 .
de domenico , m. , sol - ribalta , a. , cozzo , e. , kivel , m. , moreno , y. , porter , m. a. , gmez , s. & arenas , a. ( 2013 ) mathematical formulation of multilayer networks .
x _ , * 3 * , 041022 .
depradine , c. & lovell , e. ( 2004 ) climatological variables and the incidence of dengue fever in barbados .
j. environ .
health res .
_ , * 6 * , 429441 .
( 20112014 ) .
http://www.dge.gob.pe .
disdier , a .- c . & head , k. ( 2008 ) the puzzling persistence of the distance effect on bilateral trade . _ rev econ . stat .
_ , * 90 * , 3748 .
expert , p. , evans , t. s. , blondel , v. d. & lambiotte , r. ( 2011 ) uncovering space - independent communities in spatial networks .. _ proc .
usa _ , * 108*(19 ) , 76637668 .
fenn , d. j. , porter , m. a. , mcdonald , m. , williams , s. , johnson , n. f. & jones , n. s. ( 2009 ) dynamic communities in multichannel data : an application to the foreign exchange market during the 2007 - 2008 credit crisis .. _ chaos _ , * 19*(3 ) , 033119 .
fenn , d. j. , porter , m. a. , mucha , p. j. , mcdonald , m. , williams , s. , johnson , n. f. & jones , n. s. ( 2012 ) dynamical clustering of exchange rates .
_ quantitative finance _ , pages 128 .
( 20112012 ) . http://www.origem.info/misms/index.php .
fortunato , s. ( 2010 ) community detection in graphs .
_ , * 486*(3 - 5 ) , 75174 .
( 20112014 ) .
http://www.geonames.org/. goh , s. , lee , k. , park , j. s. & choi , m. y. ( 2012 ) modification of the gravity model and application to the metropolitan seoul subway system .
e _ , * 86 * , 026102 .
gonzalez , f. , maritan , m. c. , simini , a. & barabasi , a .-
( 2012 ) a universal model for mobility and migration patterns .
_ nature _ , * 484*(7392 ) , 96100 .
good , b. h. , de montjoye , y. a. & clauset , a. ( 2010 ) performance of modularity maximization in practical contexts .
e _ , * 81 * , 046106 .
guzman , m. ( 2003 ) dengue and dengue hemorrhagic fever in the americas : lessons and challenges .
_ j. clin .
_ , * 27*(1 ) , 113 .
guzman , m. g. , halstead , s. b. , artsob , h. , buchy , p. , farrar , j. , gubler , d. j. , hunsperger , e. , kroeger , a. , margolis , h. s. , martnez , e. , nathan , m. b. , pelegrino , j. , simmons , c. , yoksan , s. & peeling , r. w. ( 2010 ) dengue : a continuing global threat .. _ nat .
_ , * 8*(12 suppl ) , s716 . hannigan , j. , hernandez , g. , medina , r. m. , roos , p. & shakarian , p. ( 2013 ) mining for spatially - near communities in geo - located social networks .
arxiv:1311.1924 .
harrington , l. c. , scott , t. w. , lerdthusnee , k. , coleman , r. c. , costero , a. , clark , g. g. , jones , j. j. , kitthawee , s. , kittayapong , p. , sithiprasasna , r. et al .
( 2005 ) dispersal of the dengue vector aedes aegypti within and between rural communities .
_ , * 72*(2 ) , 209220 .
holme , p. & saramki , j. ( 2012 ) temporal networks .
_ phys . rep .
_ , * 519*(3 ) , 97125 .
holme , p. & saramki , j. , editors ( 2013 ) _ temporal networks_. springer - verlag .
( 20112014 ) .
http://www.inei.gob.pe/. jetten , t. h. & focks , d. a. ( 1997 ) potential changes in the distribution of dengue transmission under climate warming .
_ , * 57*(3 ) , 285297 .
johansson , m. a. , dominici , f. & glass , g. e. ( 2009 ) local and global effects of climate on dengue transmission in puerto rico .
_ plos negl .
_ , * 3*(2 ) , e382 .
jones , k. e. , patel , n. g. , levy , m. a. , storeygard , a. , balk , d. , gittleman , j. & daszak , p. ( 2008 ) global trends in emerging infectious diseases .
_ nature _ , * 451 * , 990993 .
jutla , i. s. , mucha , p. j. & jeub , l. ( 20112012 ) genlouvain : a generalized louvain method for community detection implemented in matlab .
http://netwiki.amath.unc.edu/genlouvain .
kawadia , v. & sreenivasan , s. ( 2012 ) sequential detection of temporal communities by estrangement confinement .
_ scientific reports _ , * 2 * , 794 .
keating , j. ( 2001 ) an investigation into the cyclical incidence of dengue fever .
_ , * 53 * , 15871597 .
kivel , m. , arenas , a. , barthelemy , m. , gleeson , j. p. , moreno , y. & porter , m. a. ( 2014 ) multilayer networks .
arxiv:1309.7233 .
kochel , t. j. , watts , d. m. , halstead , s. b. , hayes , c. g. , espinoza , a. , felices , v. , caceda , r. , bautista , c. t. , montoya , y. , douglas , s. & russell , k. l. ( 2002 ) effect of dengue-1 antibodies on american dengue-2 viral infection and dengue haemorrhagic fever .
_ lancet _ , * 360*(9329 ) , 310312 .
kraskov , a. , stgbauer , h. , andrzejak , r. g. & grassberger , p. ( 2005 ) hierarchical clustering using mutual information .
_ europhys .
_ , * 70*(2 ) , 278284 .
kuno , g. ( 1997 ) factors influencing the transmission of dengue viruses . in gubler
d j , k. g. , editor , _ dengue and dengue hemorrhagic fever .
_ , pages 6188 .
cab international , wallingford , uk .
lambiotte , r. , delvenne , j .- c . & barahona , m. ( 2009 ) laplacian dynamics and multiscale modular structure in networks .
arxiv:0812.1770 .
lancichinetti , a. & fortunato , s. ( 2012 ) consensus clustering in complex networks .
_ sci rep _ ,
* 2*(336 ) , 794 .
lee , s. h. , ffrancon , r. , abrams , d. m. , kim , b. j. & porter , m. a. ( 2014 ) matchmaker , matchmaker , make me a match : migration of populations via marriages in the past . _
x _ , * 4 * , 041009 .
lewis , a. c. f. , jones , n. s. , porter , m. a. & deane , c. m. ( 2010 ) the function of communities in protein interaction networks at multiple scales . _ bmc sys biol _ , *
4*(100 ) .
li , c. f. , lim , t. w. , han , l. l. & fang , r. ( 1985 ) rainfall , abundance of aedes aegypti and dengue infection in selangor , malaysia .
_ southeast asian j. trop .
public health _ , * 16*(4 ) , 560568 .
lloyd , l. s. ( 2003 ) best practices for dengue prevention and control in the americas .
strategic report , environmental health project office of health infectious diseases and nutrition . .
, m. & garlaschelli , d. ( 2013 ) community detection for correlation matrices .
arxiv:1311.1924 .
macon , k. t. , mucha , p. j. & porter , m. a. ( 2012 ) community structure in the united nations general assembly . _ physica a _ , * 391*(12 ) , 343361 .
martens , p. & hall , l. ( 2000 ) malaria on the move : human population movement and malaria transmission . _
emerging infect dis .
_ , page 6 .
masucci , a. p. , serras , j. , johansson , a. & batty , m. ( 2013 ) gravity versus radiation models : on the importance of scale and heterogeneity in commuting flows .
e _ , * 88 * , 022812
. mehta , m. ( 2004 ) _ random matrices_. academic press , san diego , california , third edition .
meil , m. ( 2007 ) comparing clusterings an information based distance .
_ j. multivariate anal .
_ , * 98*(5 ) , 873895 . montoya , y. , holechek , s. , caceres , o. , palacios , a. , burans , j. , guevara , c. , quintana , f. , herrera , v. , pozo , e. & anaya , e. ( 2003 ) circulation of dengue viruses in north - western peru , 20002001 .
_ dengue bulletin _ ,
* 27 * , 5262 .
mucha , p. j. & porter , m. a. ( 2010 ) communities in multislice voting networks .. _ chaos _ , * 20*(4 ) , 041108 .
mucha , p. j. , richardson , t. , macon , k. , porter , m. a. & onnela , j .-
( 2010 ) community structure in time - dependent , multiscale , and multiplex networks .. _ science _ , * 328*(5980 ) , 876878 .
newman , m. & girvan , m. ( 2004 ) finding and evaluating community structure in networks . _ phys .
e _ , * 69*(2 ) , 026113 .
newman , m. e. j. ( 2004 ) analysis of weighted networks .
e _ , * 70 * , 056131 .
newman , m. e. j. ( 2006 ) finding community structure in networks using the eigenvectors of matrices .
e _ , * 74 * , 036104
. newman , m. e. j. ( 2006 ) modularity and community structure in networks .. _ proc .
usa _ , * 103*(23 ) , 85778582 .
newman , m. e. j. ( 2010 ) _ networks : an introduction_. oxford university press .
onnela , j .-
p . , fenn , d. j. , reid , s. , porter , m. a. , mucha , p. j. , fricker , m. d. & jones , n. s. ( 2012 ) taxonomies of networks .
e _ , * 86*(036104 ) .
osorio , l. , todd , j. & bradley , d. j. ( 2004 ) travel histories as risk factors in the analysis of urban malaria in colombia .
_ , * 71*(4 ) , 380386 . porter , m. a. , onnela , j .-
mucha , p. j. ( 2009 ) communities in networks .
_ notices amer .
_ , * 56*(9 ) , 10821097 , 11641166 .
ratti , c. , sobolevsky , s. , calabrese , f. , andris , c. , reades , j. , martino , m. , claxton , r. & strogatz , s. h. ( 2010 ) redrawing the map of great britain from a network of human interactions . _
plos one _ , * 5*(12 ) , e14248 .
reichardt , j. & bornholdt , s. ( 2006 ) statistical mechanics of community detection .. _ phys .
e _ , * 74 * , 016110
. sengupta , a. m. & mitra , p. p. ( 1999 ) distributions of singular values for some random matrices .
e _ , * 60 * , 33893392 .
shakarian , p. , roos , p. , callahan , d. & kirk , c. ( 2013 ) mining for geographically disperse communities in social networks by leveraging distance modularity .
arxiv:1305.3668 .
simini , f. , maritan , a. & nda , z. ( 2013 ) human mobility in a continuum approach .
_ plos one _ , * 8*(3 ) , e60069 .
sinha , s. , chatterjee , a. , chakraborti , a. & chakrabarti , b. k. ( 2011 ) _
econophysics_. wiley - vch .
smith , s. m. , miller , k. l. , salimi - khorshidi , g. , webster , m. , beckmann , c. f. , nichols , t. e. , ramsey , j. d. & woolrich , m. w. ( 2011 ) network modelling methods for fmri .. _ neuroimage _ , * 54*(2 ) , 875891 .
spirin , v. & mirny , l. a. ( 2003 ) protein complexes and functional modules in molecular networks .
usa _ , * 100*(21 ) , 1212312128 .
stewart , j. q. ( 1947 ) empirical mathematical rules concerning the distribution and equilibrium of population .
_ , * 37 * , 461485 .
stewart , j. q. & warntz , w. ( 1958 ) physics of population distribution .
_ , * 1 * , 99121 . stoddard , s. t. , morrison , a. , vazquez - prokopec , g. m. , paz soldan , v. , kochel , t. j. , kitron , u. , elder , j. & scott , t. w. ( 2009 ) the role of human movement in the transmission of vector - borne pathogens .
_ plos negl .
_ , * 3*(7 ) , e481 .
strehl , a. , ghosh , j. & cardie , c. ( 2002 ) cluster ensembles - a knowledge reuse framework for combining multiple partitions .
_ j. mach .
_ , * 3 * , 583617 .
tatem , a. j. , rogers , d. j. & hay , s. i. ( 2006 ) global transport networks and infectious disease spread . in s
i. hay , a. g. & rogers , d. j. , editors , _ global mapping of infectious diseases : methods , examples and emerging applications _ , volume 62 of _ advances in parasitology _ , pages 293343 . academic press .
thiemann , c. , theis , f. , grady , d. , brune , r. & brockmann , d. ( 2010 ) the structure of borders in a small world .
_ plos one _ , * 5*(11 ) , e15422 .
traud , a. l. , kelsic , e. d. , mucha , p. j. & porter , m. a. ( 2011 ) comparing community structure to characteristics in online collegiate social networks .
_ siam review _ , * 53*(3 ) , 526 .
traud , a. l. , mucha , p. j. & porter , m. a. ( 2012 ) social structure of facebook networks .
_ physica a _ , * 391*(16 ) , 41654180 .
truscott , j. & ferguson , n. m. ( 2012 ) evaluating the adequacy of gravity models as a description of human mobility for epidemic modelling .
_ plos comput biol _ , * 8*(10 ) , e1002699 .
wasserman , s. & faust , k. ( 1994 ) _ social network analysis : methods and applications_. cambridge university press .
waugh , a. s. , pei , l. , fowler , j. h. , mucha , p. j. & porter , m. a. ( 2012 ) party polarization in congress : a network science approach .
arxiv:0907.3509 .
wilson , a. g. ( 1958 ) a statistical theory of spatial distribution models . _ transport .
_ , * 1*(3 ) , 253269 .
xia , y. , bjornstad , o. n. & grenfell , b. t. ( 2004 ) measles metapopulation dynamics : a gravity model for epidemiological coupling and dynamics .
_ , * 164*(2 ) , 267281 .
zipf , g. k. ( 1946 ) the p1 p2/d hypothesis : on the intercity movement of persons .
_ , * 11 * , 677686 .
normalized variation of information ( nvi ) @xcite is a viable alternative similarity measure to nmi for the spatial benchmark networks .
in contrast to nmi , variation of information ( vi ) and nvi are metrics in the mathematical sense .
both measures are related to mutual information . for a partition @xmath109 with @xmath110 communities ,
vi is defined as @xmath254 where @xmath255 is the entropy of the random variable associated to partition @xmath118 , the quantity @xmath256 $ ] is the mutual information , @xmath114 and @xmath115 are the respective marginal probabilities of observing communities @xmath116 and @xmath117 in partitions @xmath118 and @xmath119 , and @xmath120 is the joint probability of observing communities @xmath116 and @xmath117 simultaneously in partitions @xmath118 and @xmath119 ) .
vi is equal to @xmath121 if partitions @xmath118 and @xmath119 are identical , and @xmath257 , where @xmath0 is the number of nodes in the whole network . normalizing vi yields nvi , which is given by @xcite @xmath258\,.\ ] ]
see refs .
@xcite for additional discussions . as one can see in fig .
[ figure : benchmark : static - vi - gammavsnm ] , both nmi and nvi perform similarly and neither gives visibly better precision .
cities , a grid size of @xmath131 , and a density parameter of @xmath259 .
we examine the partitions for different values of the resolution parameter @xmath22 as a function of inter - community connectivity @xmath87 using the ( top ) ng null model , ( middle ) gravity null model , and ( bottom ) radiation null model .
[ figure : benchmark : static - vi - gammavsnm ] ]
we now vary the number @xmath0 of cities in benchmarks with a fixed size of @xmath137 , density parameter of @xmath129 , and a uniform population of @xmath130 people in each city . in fig .
[ figure : benchmark : static - even - d - nmi - binsizevsnm ] , we plot the nmi of algorithmic partitions versus planted partitions for several values of the resolution parameter @xmath22 using the ng null model and both spatial null models . in combination with fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] in the main text , which has @xmath140 cities , we observe no qualitative changes in nmi aside from an expected increase in variability when @xmath0 is small . , a density parameter of @xmath129 , and uniform populations of @xmath130 for different numbers of cities in an underlying space of the same size .
the number of cities is ( left ) @xmath132 , and ( right ) @xmath133 .
we use the ng ( top ) , gravity ( middle ) , and ( bottom ) radiation null models . see fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] in the main text for plots with @xmath126 .
[ figure : benchmark : static - even - d - nmi - binsizevsnm ] ]
we present the results of varying the edge density parameter @xmath145 in static benchmarks .
the edge density has a strong effect on the ability of the modularity - maximization methods to detect communities . for @xmath260
, we obtain smaller nmi scores than the maximum attained for each particular @xmath87 for larger @xmath145 values .
( see figs .
[ figure : benchmark : static - even - rovsnm ] and [ figure : benchmark : static - varypop - rovsnm ] . )
we therefore focus on using a density parameter of @xmath261 in the main text to follow the choice that was used for the benchmarks networks in ref .
@xcite . , and planted partitions for uniform population static spatial benchmarks with @xmath126 , a size parameter of @xmath131 , uniform city populations of @xmath130 , and several values of inter - community connectivity @xmath87 .
we plot the nmi scores as a function of the edge density parameter @xmath145 for ( left ) the distance benchmark and ( right ) the flux benchmark .
[ figure : benchmark : static - even - rovsnm ] ] , and planted partitions for random population static spatial benchmarks with @xmath126 , a size parameter of @xmath131 , city populations @xmath262 selected uniformly at random from @xmath263 $ ] , and several values of inter - community connectivity @xmath87 .
we plot the nmi scores as a function of the edge density parameter @xmath145 for ( left ) the distance benchmark and ( right ) the flux benchmark .
[ figure : benchmark : static - varypop - rovsnm ] ]
in this section , we construct a `` distance and population '' spatial benchmark . in fig . [ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] in the main text we observed that the gravity null model performs best on the uniform population distance benchmark , but the ng null model performs better than spatial null models on the random population distance benchmark because the edge placement in that benchmark does not include population information . here , we study the effects of incorporating population into edge probabilities for the `` distance and population '' benchmark .
we construct the new type of benchmark network in the same manner as the distance benchmark in section [ section : benchmarks ] , but we now incorporate population into the edge - placement probability by taking @xmath154 .
as expected , this brings back the advantage that the gravity null model has for the uniform population distance benchmark ( compare fig .
[ figure : benchmark : distpop ] with fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] in the main text ) .
the radiation null model has the second - best performance on this benchmark , with a better performance than on the random population distance benchmark .
however , it does not do as well as it did on the random population flux benchmark ( see fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] ) . , @xmath131 , @xmath160 , @xmath129 , and @xmath161 for various values of @xmath108 ( colored curves ) as a function of @xmath87 .
we detected communities by optimizing modularity using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models [ figure : benchmark : distpop ] ]
we now study the influence of the parameters @xmath22 and @xmath108 on the community structure for random - population multilayer temporally - stable benchmarks .
we first compare the results to our findings from static benchmarks by varying @xmath22 and @xmath87 for fixed values of @xmath108 .
we study the performance of modularity maximization with the ng , gravity , and radiation null models on random population benchmarks ( see fig .
[ figure : benchmark : multi - gammavsnm ] ) with parameter values of @xmath126 , @xmath131 , and @xmath157 layers using @xmath143 and @xmath264 .
we only show plots for @xmath159 , as the values of @xmath108 do not noticeably influence the results .
we obtain results that are similar to our results for the corresponding static benchmarks inn fig .
[ figure : benchmark : static - even - d - nmi - gammavsnm - tab ] .
once again , we find that the choice of @xmath22 has a large influence on the quality of algorithmic partitions , and ( as with our findings for static benchmarks ) that @xmath155 seems to yield the best performance ( i.e. , the largest nmi scores ) for low values of @xmath87 , whereas larger values of @xmath22 perform better for larger @xmath87 ( see fig .
[ figure : benchmark : multi - rand - gammavsnm ] ) .
the effect of varying @xmath22 is most pronounced for the radiation null model on flux benchmarks .
cities uniformly at random from the set @xmath82 .
we consider various values of the resolution parameter @xmath22 , and the other parameter values are @xmath131 , @xmath160 , @xmath129 , and @xmath159 .
we plot nmi as a function of @xmath87 for ( left ) the distance benchmark and ( right ) the flux benchmark using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : multi - rand - gammavsnm ] ] we now examine the nmi of algorithmic versus planted partitions in temporally - stable multilayer benchmarks for fixed @xmath155 while varying @xmath108 and @xmath87 . as we show in fig .
[ figure : benchmark : multi - rand - omegavsnm ] , we find that the value of @xmath108 usually has very little effect on our ability to detect the planted communities via modularity maximization the same as for the uniform population temporally stable multilayer benchmarks ( see fig .
[ figure : benchmark : multi - omegavsnm ] ) .
the parameter @xmath108 becomes important for the random - population , temporally - evolving multilayer benchmarks in the same manner as what we observed in the main text for uniform population benchmark networks ( not shown ; see fig .
[ figure : benchmark : multi - t - gammavsnm ] in the main text for the uniform population results ) .
cities uniformly at random from the set @xmath82 .
we consider various values of the parameter @xmath108 , and the other parameter values are @xmath131 , @xmath160 , @xmath129 , and @xmath155 .
we plot nmi as a function of @xmath87 for ( left ) the distance benchmark and ( right ) the flux benchmark using the ( top ) ng , ( middle ) gravity , and ( bottom ) radiation null models .
[ figure : benchmark : multi - rand - omegavsnm ] ]
in fig . [ figure : benchmark : multi - nodelevel - u - gammavsnm ] , we present our results for province - level community detection on uniform population temporally stable multilayer benchmarks . as one can see by comparing these results to those in fig . [
figure : benchmark : multi - gammavsnm ] , we obtain similar nmi scores for the performance of community detection for province - level communities as we did for the ordinary community detection in multilayer networks . for all @xmath4 ) temporally stable multilayer spatial benchmarks with @xmath160 layers .
each layer has a single - layer planted partition with @xmath126 cities , a size parameter of @xmath131 , and a density parameter of @xmath129 .
we use @xmath252 and consider various values of the resolution parameter @xmath22 , and we plot nmi as a function of the inter - community connectivity @xmath87 for ( left ) the distance benchmark and ( right ) the flux benchmark .
[ figure : benchmark : multi - nodelevel - u - gammavsnm ] ]
in fig . [ figure : dengue - fullyaggregated ] , we show additional results of community detection on fully aggregated networks ( i.e. , we use @xmath248 and @xmath249 ) from the dengue fever times series . in section [ dengue - region - ng ] of the main text , in fig . [
figure : dengue - regionlevel - map](a ) we showed the results of modularity maximization using the ng null model .
we now also show similar results for the gravity , radiation , and correlation null models .
the gravity , radiation , and correlation null models find one large community and a few small communities . because of the aggregation
, we have lost the rich set of information that we were able to study using multilayer community detection . | in the study of networks , it is often insightful to use algorithms to determine mesoscale features such as `` community structure '' , in which densely connected sets of nodes constitute `` communities '' that have sparse connections to other communities . the most popular way of detecting communities algorithmically is to optimize the quality function known as modularity . when optimizing modularity , one compares the actual connections in a ( static or time - dependent ) network to the connections obtained from a random - graph ensemble that acts as a null model .
the communities are then the sets of nodes that are connected to each other densely relative to what is expected from the null model . clearly , the process of community detection depends fundamentally on the choice of null model , so it is important to develop and analyze novel null models that take into account appropriate features of the system under study . in this paper
, we investigate the effects of using null models that incorporate spatial information , and we propose a novel null model based on the radiation model of population spread . we also develop novel synthetic spatial benchmark networks in which the connections between entities are based on distance or flux between nodes , and we compare the performance of both static and time - dependent radiation null model to the standard ( `` newman - girvan '' ) null model for modularity optimization and a recently - proposed gravity null model . in our comparisons , we use both the above synthetic benchmarks and time - dependent correlation networks that we construct using countrywide dengue fever incidence data for peru .
we also evaluate a recently - proposed correlation null model , which was developed specifically for correlation networks that are constructed from time series , on the epidemic - correlation data .
our findings underscore the need to use appropriate generative models for the development of spatial null models for community detection .
community detection , spatial null model + 87.19.xx,89.20.-a,89.75.fb,05.45.tp |
the study of linear statistics of eigenvalues of random matrices has played a major role for the applications of random matrix theory to physical problems , like quantum transport @xcite , quantum entanglement @xcite , etc , as it was emphasized in our recent papers @xcite . whereas previous work has considered _ unrestricted _ sums of the form @xmath12 , where @xmath0 is the spectrum of eigenvalues of a random matrix , the question of _ truncated _ linear statistics ( tls ) of the form @xmath13 was recently raised in our previous paper @xcite .
there , we have considered the case where the eigenvalues contributing to the partial sum are further constrained to be the @xmath3 largest ( or smallest ) ones , which corresponds to interpolating between two well studied types of problems : for @xmath14 , the statistical analysis of the linear statistics @xmath12 @xcite ( and many other references ) , and for @xmath15 , the study of the distribution of the largest ( or smallest ) eigenvalue @xcite .
the aim of the present paper is to analyse the statistical properties of tls _ in the absence of further restriction _ concerning the ordering of the @xmath2 eigenvalues ; we will see that lifting this restriction has strong implications on the analysis and the results . for convenience , we rescale the tls as @xmath16 where the exponent @xmath17 is chosen such that @xmath18 remains of order @xmath19 as @xmath20 .
the precise value of @xmath17 depends on the function @xmath21 and the matrix ensemble under consideration .
the general scenario presented in the paper is valid for any choice of function @xmath21 .
moreover it applies to different matrix ensembles , although we will focus on wishart matrices ( i.e. the laguerre ensemble of random matrices ) . in the sequel , for reasons
explained later , we will consider the case @xmath11 , leading to @xmath22 . denoting by @xmath23 the joint probability distribution function for the eigenvalues ( for the laguerre ensemble , eq .
below ) , we can write the distribution of @xmath18 as : @xmath24 here , @xmath23 is a symmetric function of its @xmath10 arguments . due to the restriction to a fraction of the eigenvalues in
, the only avalaible approaches seem to be the orthogonal polynomial technique , which provides naturally the @xmath3-point correlation functions , and the coulomb gas method .
however the former is restricted to the unitary class and does not permit a simple analysis of the large deviations .
for this reason , we will use the coulomb gas technique . because the method deals with the eigenvalue density , the restriction to a fraction of eigenvalues requires to introduce `` occupation numbers '' @xmath25 and write the tls as : @xmath26 where @xmath27 if the eigenvalue @xmath28 contributes to the sum and @xmath29 otherwise .
note that subset sums @xmath30 of eigenvalues of certain random matrices recently appeared in the spectral analysis of fermionic systems @xcite .
since the sum should contain @xmath31 eigenvalues , the @xmath32 $ ] acceptable configurations @xmath33 verify @xmath34 recently , the case where the summation is restricted to the @xmath31 largest ( or smallest ) eigenvalues was considered in ref .
this corresponds to considering the configuration @xmath35 and @xmath36 . here
, we consider the same problem in the absence of this restriction , which leads us to introduce a new ensemble described by the joint distribution for the two sets of random numbers @xmath37 where @xmath38 and @xmath39 the heaviside step function .
. is obviously normalised when integrated over all @xmath28 s and with summation over the occupations .
the distribution we aim to determine in the paper is now written as @xmath40 where @xmath41 in the subscript of @xmath42 .
plane , where @xmath41 and @xmath43 ( here for the case @xmath44 in the laguerre ensemble , thus with @xmath22 ) .
insets show the corresponding optimal `` charge '' density profiles . shaded part of the density ( insets on the right )
shows the fraction of ( largest or smallest ) @xmath31 eigenvalues contributing to @xmath18 .
dashed profiles ( phases ii and iii ) mean fluctuations in the `` occupations '' @xmath45 ( cf .
text ) . in the central part of the diagram ( phase ii )
, the distribution is dominated by entropy , while in phases i and iii it is dominated by the energy .
_ , scaledwidth=95.0% ] considering the laguerre ensemble , eq .
below , with @xmath11 , we have obtained the distribution @xmath46 in the limit of large matrix size , @xmath47 . in this limit
the multiple integral is dominated by the optimal configuration for the two sets of random numbers .
as the two parameters @xmath48 and @xmath18 are tuned , we obtain different types of optimal configurations , which are interpreted as different `` phases '' .
the phase diagram can be drawn in the @xmath49 plane , where the physical available phases are located in the stripe @xmath50\times[0,\infty[$ ] .
as one increases @xmath18 for fixed @xmath48 , one encounters three phases described below , the phase transitions taking places at @xmath51 and @xmath52 , which defines two lines in the @xmath49 plane ( see also fig . [
fig : phdiag ] ) : * phase i : the optimal configuration @xmath53 , that dominates the integrals , varies with @xmath49 while the occupations @xmath33 remain frozen
. this case is conveniently analysed within the conventional coulomb gas interpretation where the eigenvalues are considered as the positions of a one - dimensional gas of @xmath10 particles with logarithmic interactions . *
phase ii : @xmath53 are frozen while @xmath33 fluctuate .
to get some insight , we interpret the problem as a gas of @xmath3 fictitious fermions occupying @xmath10 fixed `` energy levels '' @xmath54 s .
this phase includes the typical fluctuations ( the dashed line in fig . [
fig : phdiag ] indicates where @xmath46 is maximum ) . *
phase iii : the last phase is described in terms of a mixed picture where one single eigenvalue splits off the bulk , while the density of the remaining eigenvalues is frozen and the occupation numbers fluctuate ( they freeze as @xmath55 ) .
the different phases are related to the following behaviours for the distribution of the tls must be understood as @xmath56 . ] : @xmath57 \exp \left\lbrace - n \phi_0(\kappa;s ) \right\rbrace \hspace{0.5 cm } & \text{for } s_0(\kappa)<s <
s_1(\kappa ) \\[0.2 cm ] n^{-\frac{\beta n}{2 } } \exp \left\lbrace - n\left [ \frac{\beta } { 2}\phi_+(\kappa;s ) + \phi_0(\kappa ; \tilde{s}(s ) ) \right ] \right\rbrace & \text{for } s > s_1(\kappa ) \end{cases}\end{aligned}\ ] ] where @xmath58 smoothly interpolates between @xmath59 and @xmath48 ( it will be studied in section [ sec : ldflarge ] ) .
the large deviation function in the last expression combines two different functions with different origins ( energy _ versus _ entropy ) .
a sketch of the distribution is represented in fig .
[ fig : schemadistr ] .
the different scalings with @xmath10 in the exponential arise from the fact that entropy @xmath8 dominates in phase ii , including the typical fluctuations , while energy @xmath6 dominates in phase i ( the scaling for phase iii is due to a subleading contribution to the energy ) .
the scaling in phase ii implies in particular that the fluctuations of the tls scale as @xmath60 ( and are independent of the symmetry index @xmath61 ) .
the limiting behaviours of the large deviation functions for the three regimes are : @xmath62 \displaystyle \omega_\kappa\ , ( s - s_0(\kappa))^2 & \text{as } s \to s_0(\kappa)^- \end{cases } \ : , \\ & \phi_0(\kappa;s ) \simeq \begin{cases } c_{\kappa } - \cp \sqrt{s - s_{0}(\kappa ) } & \text{as } s \to s_0(\kappa)^+ \\[0.2 cm ] \displaystyle \frac{(s-\kappa)^2}{2 \kappa(1-\kappa ) } & \text{as } s \to \kappa \\[0.2 cm ] c_{\kappa } - \cpt \sqrt { s_{1}(\kappa ) - s } & \text{as } s \to s_1(\kappa)^- \end{cases } \\ & \phi_+(\kappa;s ) + \frac{2}{\beta } \phi_0(\kappa;\tilde{s}(s ) ) \simeq \begin{cases } \displaystyle \ln(s - s_1(\kappa ) ) + \mathrm{cste } & \text{as } s \to s_1(\kappa)^+ \\[0.2 cm ] \displaystyle \ln ( s - \kappa ) + \mathrm{cste } & \text{as } s \to + \infty \end{cases}\end{aligned}\ ] ] where @xmath63 , @xmath64 , @xmath65 and @xmath66 are positive constants .
obtained in the paper , eq . .
_ , scaledwidth=80.0% ] section [ sec : motivation ] exposes the motivations which have led us to the study of the distribution of the tls for @xmath11 . in section [ sec : coulombgas ] , we will describe phase i , which will be analysed within the usual coulomb gas picture .
we will introduce the notations .
the derivation of the results has been made as short as possible , as the problem is similar to the one studied in our previous article @xcite , although the case considered here is more complicated ( in ref .
@xcite , we have made use of simplifications arising from the specific choice @xmath67 ) .
section [ sec : typfluct ] will discuss phase ii , analysed within the picture of fictitious non interacting fermions .
phase iii is studied in section [ sec : ldflarge ] .
the paper is closed by some concluding remarks .
the problem studied in this article is motivated by the statistical analysis of partial sums of proper time delays in chaotic scattering .
proper time delays are characteristic times capturing temporal aspects of quantum scattering ( for a review on time delays , cf . @xcite ; aspects related to chaotic scattering are reviewed in @xcite ) .
for the sake of concreteness , let us consider a quantum dot ( qd ) connected to external contacts playing the role of wave guides through which electrons can be injected ( fig .
[ fig : qd ] ) .
an electronic wave of energy @xmath68 injected in the conducting channel @xmath69 is scattered in channel @xmath70 with probability amplitude given by the matrix element @xmath71 of the @xmath72 on - shell scattering matrix @xmath73 .
probing the energy structure of @xmath73 allows one to define several characteristic times . ref .
@xcite has emphasized the difference between the _ partial _ time delays @xmath74 , which are derivatives of the scattering phase shifts , and the _ proper _ time delays @xmath75 , which are eigenvalues of the wigner - smith matrix @xmath76 in other terms , these two sets of characteristic times are obtained by performing derivation and diagonalisation in different orders .
they however satisfy the sum rule @xmath77 . has @xmath78 open channels.__,scaledwidth=40.0% ] the wigner - smith matrix is a central concept allowing to characterize the amount of injected charge in the qd , which can be understood from krein - friedel type formulae relating the scattering matrix to the density of states .
this observation has played a major role in bttiker s quantum transport theory for coherent devices beyond dc linear transport : non - linear transport @xcite , ac transport @xcite , pumping , frequency dependent noise , etc ( see @xcite for a review and further references ) .
an important observation is that , in order to address the out - of - equilibrium situation , the contribution of the electrons from a given contact must be identified .
this has led to the concept of _ injectance _ , @xmath79 , measuring the contribution to the density of states ( dos ) inside the device of the scattering states incoming from the channels associated with the contact @xmath80 ; hence the total dos is @xmath81 , which is related to the wigner time delay : @xmath82 the symmetric concept of _ emittance _ ,
@xmath83_{ii}$ ] , was also introduced in order to allow for a postselection of the contact , whereas the injectance @xmath84 corresponds to the preselection ( see ref .
@xcite and the review @xcite ) . in systems with complex dynamics , such as chaotic quantum dots ( fig .
[ fig : qd ] ) , the statistical approach is the most efficient , which leads to assume that the scattering matrix @xmath85 belongs to a random matrix ensemble @xcite . in chaotic qds with perfect contacts , it is quite natural to choose the uniform distribution over the unitary group , up to some additional constraints related to the symmetry ( time reversal symmetry and/or spin rotation symmetry ) : this corresponds to the so - called circular matrix ensembles .
this approach however does not provide any information on the energy structure of the scattering matrix , what is required in order to characterize the statistical properties of the matrix @xmath86 .
an `` alternative stochastic approach '' has been introduced in @xcite and applied in @xcite where the wigner - smith matrix distribution was shown to be related to the laguerre ensemble of random matrices @xcite . , with the same spectrum of eigenvalues than @xmath86 .
the precise statement of these references is that @xmath87 is a wishart matrix . because @xmath85 and @xmath86 are only independent in the unitary case
, @xmath88 is a wishart matrix only in this case , strictly speaking .
however its eigenvalues are always given by the laguerre distribution .
] the joint distribution of the inverse of the proper times @xmath89 is @xmath90 where @xmath91 is the dyson index corresponding to orthogonal , unitary and symplectic classes .
the time scale denotes the heisenberg time @xmath92 , where @xmath93 is the mean level spacing ( or mean resonance spacing ) .
the dos interpretation thus leads to @xmath94 .
combined with the sum rule , we have @xmath95 this scale is also known as the dwell time and measures the average time spent by a wave packet inside the quantum dot .
later we will set @xmath96 for simplicity .
the variance and the correlations of proper times are also known ( see updated arxiv version of @xcite ) @xcite : @xmath97 +
2}{n^2(n+1)(\beta n - 2 ) } \ : , \\
\label{eq : covariancetaua } \cov(\tau_i , \tau_j ) & = - \frac{1}{n^2(n+1 ) } \hspace{1cm}\mbox{for } i\neq j \:.\end{aligned}\ ] ] the joint distribution for the partial time delays @xmath98 s and the diagonal matrix elements @xmath99 s are still unknown .
an interesting connection between them was pointed out in @xcite ( see also @xcite ) : introducing the unitary matrix @xmath100 which diagonalizes the scattering matrix , these authors have obtained the non trivial relation @xmath101_{ii}$ ] . in the unitary case , @xmath100 is expected to be independent of @xmath86 . because @xmath86 is invariant under unitary transformations in this case , we have @xmath102 .
let us emphasize few properties of the @xmath99 s , which are important as they are involved in the injectances which should be accessible experimentally , and emphasize their difference with the proper times .
@xmath99 s and @xmath103 s can be related by writing @xmath104 ( the unitary matrices @xmath100 and @xmath105 differ in general ) , so that @xmath106 in the unitary case ( @xmath107 ) the matrix @xmath105 and the eigenvalues are uncorrelated , which allows to go further in the statistical analysis of the @xmath99 s ( this is not the case in the orthogonal and symplectic cases @xcite ) .
thus we will focus on the unitary case for a moment , and will come back to the general case later .
correlations of the @xmath105 matrix elements are controlled by weingarten functions @xcite .
some algebra gives ] where the weingarten functions @xmath108 is a function of the matrix size and the permutation of indices . in particular : @xmath109 and @xmath110 . ]
@xmath111 which matches with the covariances of the partial times given in the updated arxiv version of @xcite , @xmath112 ( see the discussion in appendices a and e of @xcite ) . both the scaling with @xmath10 and the sign of the correlations for @xmath113 s and @xmath103 s differ : @xmath114 and @xmath115 , where the factor @xmath7 can be understood as the contribution of the unitary matrix elements in .
the striking difference between the two quantities is best understood from the relation @xmath116 \:.\ ] ] if @xmath103 s were identical and idependent random variables , @xmath117 , the first term would dominate which would lead to `` @xmath118 '' . instead ,
the proper times are strongly anti - correlated , resulting in @xmath119 and therefore @xmath120 .
these relations lead to important differences bewteen partial sums of @xmath99 s and @xmath103 s . considering a contact with @xmath3 conducting channels , the two first cumulants of the related injectance are @xmath121 ( the leading order to the second expression was given in @xcite , eq .
85 of the arxiv version ; it was deduced from the results of @xcite ) . on the other hand
, the fluctuations of the partial sum of proper time delays can be deduced from ( [ eq : variancetaua],[eq : covariancetaua ] ) by writing @xmath122 , explicitly @xmath123 comparing and leads to two interesting observations : * for @xmath124 , the variance of the injectance is @xmath125 smaller than the variance of partial sum of @xmath103 s : @xmath126 . * for @xmath127 , they are both equal to @xmath128 as the two quantities satisfy the sum rule .
this is a consequence of the fact that @xmath129 drops by a factor @xmath7 as @xmath130 .
this intringuing features have led us to analyse the full distribution of these two partial sums .
the distribution of the injectance @xmath131 is obtained in appendix [ app : injectance ] for @xmath107 .
this derivation is based on a result of savin , fyodorov and sommers @xcite , stating that , for @xmath107 , the distribution of a sub - block of the matrix @xmath86 is also related to the laguerre distribution .
, introduced in the previous footnote .
the distributions of the two matrices @xmath86 and @xmath132 coincide for @xmath107 .
] this makes this study a variant of the wigner time delay distribution analysis of ref .
@xcite : the distribution of the partial sum @xmath133 is studied in appendix [ app : injectance ] . on the other hand , the question of the distribution of the truncated linear statistics @xmath134 where @xmath28 s are eigenvalues of a wishart matrix , eq
, is much more challenging and is the main object of investigation of the present article .
the discussion of the statistical properties of @xmath99 s has led us to restrict ourselves to the unitary case . however , as we will study the full distribution of @xmath18 for arbitrary symmetry class , it is useful to express the two cumulants of @xmath18 for arbitrary @xmath61 . using eqs .
( [ eq : variancetaua],[eq : covariancetaua ] ) we get @xmath135 setting @xmath136 , we check that gives the result of ref .
@xcite : @xmath137 ( the leading order was obtained earlier in @xcite for @xmath138 and @xcite ; see also @xcite ) .
eq . exhibits another interesting feature as the leading order term is independent of the symmetry class , while the subleading term surviving in the @xmath130 limit does depend on @xmath61 . the origin of this observation will be clarified in section [ sec : typfluct ] .
let us introduce the general framework used in the paper . for large @xmath10 , it is convenient to rescale the eigenvalues as @xcite @xmath140 the coulomb gas method consists in rewriting the joint distribution of eigenvalues ( [ eq : jpdfgamma ] ) as a gibbs measure @xmath141/2)$ ] , where @xmath142 = -\frac{1}{n^2 } \sum_{i \neq j } \ln { \left| x_i - x_j \right| } + \frac{1}{n } \sum_{i=1}^n ( x_i - \ln x_i ) \label{eq : ediscr}\ ] ] is the energy of a gas of particles on a line , at positions @xmath143 , interacting with logarithmic repulsion and trapped in the external potential @xmath144 .
we want to compute the distribution of @xmath145 for any given function @xmath21 .
this is given by eq .
( [ eq : defpns ] ) , which rewrites : @xmath146 } \delta \left ( s - \frac{1}{n } \sum_{i=1}^{n } \m_i f(x_i ) \right ) \ , \constn } { \displaystyle \sum _ { \lbrace \m_i \rbrace } \int_0^\infty \hspace{-0.25cm}\dd x_1 \int_0^{x_1 } \hspace{-0.25cm}\dd x_2 \cdots \int_0^{x_{n-1}}\hspace{-0.25 cm } \dd x_n \ , \e^{-\frac{\beta n^2}{2 } \mathcal{e}_\mathrm{gas}[\lbrace x_i \rbrace ] } \ , \constn } \nonumber \:.\end{aligned}\ ] ] since we are interested in partial sums of proper time delays @xmath147 , we will consider the case @xmath148 therefore , eq .
( [ eq : defsrescaled ] ) gives @xmath149 .
the limit @xmath150 corresponds to @xmath151 , for all @xmath31 eigenvalues such that @xmath152 .
the @xmath153 others which do not contribute to the sum are not constrained , so they remain of order 1 .
thus , in this case , the @xmath31 eigenvalues in the sum are the largest .
this corresponds to @xmath154 eq .
( [ eq : defsrescaled ] ) reduces to : @xmath155 this problem was analysed in detail in @xcite , for some specific choices of function @xmath21 .
the general case is discussed in the appendix of this paper . here
we will only sketch briefly the method in order to be self - contained . in the remainder of this section
, we will focus on the situation where the occupation numbers @xmath33 are frozen according to ( [ eq : nifrozen ] ) .
this will be the case as long as @xmath139 . in sections
[ sec : typfluct ] and [ sec : ldflarge ] , we will study the case @xmath156 where the @xmath33 s are allowed to fluctuate .
let us introduce the empirical density of eigenvalues @xmath157 in terms of this density , ( [ eq : slargest ] ) reads : @xmath158 where @xmath159 is a lower bound ensuring that only the @xmath31 largest eigenvalues contribute to the integral .
it is fixed by imposing @xmath160 we emphasize that it is only possible to express @xmath18 in terms of a simple integral over the density of eigenvalues @xmath161 in the case where only the largest ( or the smallest ) eigenvalues contribute to @xmath18 , corresponding to large deviations . in the limit @xmath20 , it is possible to rewrite the multiple integrals in eq .
( [ eq : pnsratiomultint ] ) as functional integrals over the density , leading to the measure @xcite : @xmath162 } \dd x_1 \cdots
\dd x_n \rightarrow \e^{- \frac{\beta n^2}{2 } \mathscr{e}[\rho ] + n ( 1 - \frac{\beta}{2 } ) \mathscr{s}[\rho ] } \d \rho \ : , \label{eq : measurepathint}\ ] ] with the energy @xmath163 = - \int \dd x \int \dd y \ : \ln { \left| x - y \right| } + \int \dd x \ : \rho(x)(x - \ln x ) \label{eq : erho}\ ] ] and the entropy @xmath164 = -\int \dd x \ : \rho(x ) \ln \rho(x ) \ : .
\label{eq : srho}\ ] ] the energy @xmath165 $ ] is obtained by rewriting ( [ eq : ediscr ] ) in terms of the density ( [ eq : defempdens ] ) .
the diagonal terms @xmath166 which are not present in ( [ eq : ediscr ] ) are removed by adding the term @xmath167 \mathscr{s}[\rho]$ ] @xcite .
the other entropic term @xmath168 $ ] comes from the loss of information when describing the set @xmath143 by the density @xmath161 .
since we are interested in the limit @xmath20 , we can neglect the subleading entropic term .
( [ eq : pnsratiomultint ] ) becomes : @xmath169 } \delta \left ( \int_c \rho - \kappa \right ) \delta \left ( \int^c
\rho - ( 1-\kappa ) \right ) \delta \left ( s - \int_c \rho(x ) f(x ) \ : \dd x \right ) } { \displaystyle \int \dd c
\int \d \rho \ : \e^{-\frac{\beta n^2}{2 } \mathscr{e}[\rho ] } \delta \left ( \int_c \rho - \kappa \right ) \delta \left ( \int^c \rho - ( 1-\kappa ) \right ) } \nonumber \:,\end{aligned}\ ] ] where the factor @xmath170 , with @xmath171 , arises from the sum over occupations in the denominator of ( cf .
discussion in section [ sec : typfluct ] ) .
this additionnal factor would be absent if one would consider the distribution of the tls restricted to the largest eigenvalue , like in ref .
@xcite : without @xmath170 would describe the full range and would be clearly normalized . here
, this form only describes the _ tail _ of the distribution of the tls without restriction on the ordering of the eigenvalues , when @xmath172 , hence the expression has no reason to be normalised .
these integrals can be estimated using a saddle point method .
the saddle point will give @xmath173 and @xmath174 as functions of @xmath48 and @xmath18 .
the numerator is dominated by the density @xmath175 which minimizes the energy @xmath165 $ ] under the constraints imposed by the dirac @xmath176-functions .
similarly , the denominator is dominated by a density @xmath177 .
the minimization takes the form similar to ( [ eq : fcoulomb],[eq : minfapp ] ) , with additional constraints , which leads to two coupled integral equations , instead of one .
the precise equations can be found in ref .
@xcite , where this procedure was carried out explicitly . here , we will simply jump to the solution of the coupled integral equations ( see below eq . ) .
finally , we obtain the distribution @xmath178 where we have introduced the large deviation function @xmath179 - \mathscr{e}[\rhomp(x ) ] \ : .
\label{eq : ldflenegy}\ ] ] in , the scaling as @xmath180 is a manifestation of energy of the coulomb gas and reflects the long range nature of the interaction between the @xmath10 particles . the density @xmath181 is given explicitly in the general case in the appendix of ref .
@xcite . for any monotonous function @xmath21
, it reads : @xmath182 where the principal value is needed only if @xmath183 $ ] .
this density is supported on two disjoint intervals @xmath184 $ ] and @xmath185 $ ] ( see fig . [
fig : rhosplit ] ) , where @xmath174 is the boundary introduced above in eqs .
( [ eq : constrscoulomb],[eq : constrkcoulomb ] ) .
the boundaries of these intervals and the parameter @xmath186 are fixed by the conditions that the density vanishes at these points : @xmath187 along with the constraints @xmath188 the parameter @xmath186 is a lagrange multiplier introduced to handle the constraint on @xmath18 .
we denote @xmath189 the solution of ( [ eq : rhocond1],[eq : rhocond2],[eq : rhocond3],[eq : rhocond4 ] ) .
its knowledge allows to compute easily the energy thanks to the thermodynamic identity @xcite : @xmath190}{\mathrm{d}s } } = - \mu_1^\star(\kappa;s ) \ : .
\label{eq : thermoidcoulomb}\ ] ] the density @xmath191 which dominates the denominator of ( [ eq : pnsratiopathint ] ) is the well known marenko - pastur distribution @xcite : @xmath192 this density can be obtained from ( [ eq : rhospl ] ) by taking the limit @xmath193 , which corresponds to @xmath194 .
( solid line ) , for @xmath195 , compared to the density @xmath191 ( dashed ) .
the shaded area correspond the the fraction @xmath48 of the largest eigenvalues .
the support of the density @xmath181 is @xmath184\cup[c , d]$ ] .
_ , scaledwidth=60.0% ] this derivation is valid when only the largest eigenvalues contribute to @xmath18 .
this is the case as long as the gap between the two intervals of the support of @xmath181 is non - zero .
the limit of validity is therefore given by @xmath196 , which corresponds to @xmath197 .
the distribution is then @xmath191 , thus eq .
( [ eq : rhocond4 ] ) gives the value of @xmath18 : @xmath198 where @xmath199 is fixed by @xmath200 solving the second equation for @xmath199 and plugging the result into the first equation gives the maximal value @xmath201 allowed for @xmath18 .
this defines a line @xmath202 in the @xmath49 plane which delimits the region where the assumption that only the largest eigenvalues contribute to @xmath18 is true .
this is the lower solid line in fig .
[ fig : phdiag ] .
exactly on this line , the density of eigenvalues is the marenko - pastur distribution @xmath191 . for @xmath195 ,
the limit @xmath150 , corresponds to push the fraction @xmath48 of the rightmost eigenvalues towards infinity .
this means to let @xmath203 in the previous equations .
expanding eqs .
( [ eq : rhocond1],[eq : rhocond2],[eq : rhocond3],[eq : rhocond4 ] ) in this limit yields : @xmath204 using the thermodynamic identity ( [ eq : thermoidcoulomb ] ) , a simple integration of this last relation gives the behaviour of the energy , therefore of the large deviation function , for @xmath150 : @xmath179 - \mathscr{e}[\rhomp(x ) ] = \frac{\kappa^2}{s } + \frac{3 \kappa(2-\kappa)}{2 } \ln s + \o(1 ) \ : .
\label{eq : ldfszero}\ ] ] this expression gives the left tail of the distribution : @xmath205 note that we recover the tail computed in ref .
@xcite simply by setting @xmath127 : @xmath206 interestingly , the leading term in ( [ eq : ldfszero ] ) can be obtained easily by a heuristic argument . for small @xmath18 ,
the energy is dominated by the potential energy of the eigenvalues pushed to infinity .
the typical value of these eigenvalues is given by @xmath207 , corresponding to @xmath208 .
the energy is estimated as @xmath209 \sim \int_c \rho^\star v \sim \kappa v(x_\mathrm{typ } ) \sim \kappa^2/s$ ] .
qed . in the limit @xmath210 , the two bulks of the density @xmath181 merge .
hence , it corresponds to @xmath194 .
the behaviour of @xmath211 for @xmath18 close to @xmath201 is obtained by expanding eqs .
( [ eq : rhospl], ... ,[eq : rhocond4 ] ) in this limit .
a straightforward but cumbersome computation gives , for any monotonous function @xmath21 : @xmath212 where @xmath213 where @xmath199 is fixed by ( [ eq : constkc0 ] ) .
the large deviation function is deduced from the thermodynamic identity ( [ eq : thermoidcoulomb ] ) : @xmath214 we have performed a numerical simulation in order to check the quadratic behaviour and the value of the coefficient @xmath63 , in the case @xmath215 .
the energy @xmath209 $ ] is computed using a monte carlo method , see ref .
@xcite for details on the procedure .
fitting the numerics must be performed with caution : for @xmath216 , eq . ( [ eq : coefquadbehav ] ) gives @xmath217 , however a fit of the numerical data with a purely quadratic behaviour leads to @xmath218 .
this apparent discrepancy is explained by the fact that is only the leading term of an expansion near @xmath201 : @xmath219 where the higher order term can not be neglected in practice as it involves a large coefficient .
indeed , fitting the numerics with both the quadratic and the cubic terms , we find @xmath220 and @xmath221 , now in excellent agreement with the prediction ( see fit in fig . [ fig : mc ] ) . we have checked that the accuracy on @xmath222 extracted from the fit is improved by increasing the order of the polynomial used to fit the data .
interestingly , the simulation shows that the energy is frozen @xmath209 = \mathscr{e}[\rho_0^\star(x)]$ ] for @xmath156 .
this will be explained in the next section
. - \mathscr{e}[\rho_0^\star(x)]$ ] ( energy of the coulomb gas ) determined by a monte carlo simulation for @xmath216 and @xmath223 ( dots ) .
the results are compared to the quadratic behavior ( [ eq : ldfphi0quadmin ] ) ( dashed line ) and a polynomial fit of degree 3 , eq . , ( solid line ) .
the dashed vertical line is @xmath201 .
_ , scaledwidth=60.0% ]
in the previous section , we have seen that when @xmath2250,s_0(\kappa)]$ ] , the density @xmath181 changes with @xmath18 while the occupation numbers @xmath33 remain fixed .
when @xmath18 reaches @xmath201 , the density is given by the marenko - pastur distribution @xmath191 , which minimizes the energy @xmath165 $ ] .
we now enter the domain @xmath156 where it is possible to vary @xmath18 by changing the occupation numbers , while keeping the density fixed .
the range of accessible values of @xmath18 is given by two extreme cases : selecting either the largest eigenvalues ( @xmath202 ) or the smallest ones ( fig .
[ fig : mpminmax ] , right ) .
this latter case corresponds to the value @xmath226 given by : @xmath227 where @xmath228 is fixed by @xmath229 this defines another line in the @xmath49 plane .
it is the upper solid line shown in fig .
[ fig : phdiag ] .
the two lines are simply related via the relation : @xmath230 in the laguerre ensemble considered here , with @xmath215 , we simply have @xmath231 .
the goal of this section is to describe what occurs for @xmath18 between @xmath201 and @xmath59 , and compute the distribution in this domain .
note that the procedure exposed in this section is valid for a monotonous function @xmath21 . in order to determine the distribution @xmath42 on the interval
@xmath232 $ ] , let us go back to the multiple integrals in eq .
( [ eq : pnsratiomultint ] ) . for large @xmath10 ,
these integrals are dominated by the set of eigenvalues @xmath233 which minimizes the energy @xmath234 $ ] given by eq .
( [ eq : ediscr ] ) , i.e. @xmath235}{\partial x_i } } \right| _ { \lbrace x_i^\star \rbrace } = 0 \ : , \hspace{0.5 cm } \forall i \in \lbrace 1 , \cdots , n \rbrace \ : .
\label{eq : minediscr}\ ] ] without changing this set of eigenvalues , it is possible to construct @xmath236 different values , given by @xmath237 however , several configurations may give a similar value of @xmath18 .
this shows that the distribution of @xmath18 comes from an entropic contribution @xmath238 : the same `` macroscopic '' value @xmath18 may be associated to many `` microscopic '' configurations @xmath33 .
we stress that this should not be confused with the entropy @xmath239 $ ] defined by eq .
( [ eq : srho ] ) . indeed , this latter measures the loss of information when describing the set of eigenvalues @xmath143 by a density @xmath161 .
the entropy @xmath238 discussed here is hidden in the sums over the configurations @xmath33 in ( [ eq : pnsratiomultint ] ) .
it is more convenient to first compute the moment generating function , which is the laplace transform of @xmath42 , using standard tools from statistical physics . in a second step
, we will come back to the distribution .
, with either the largest eigenvalues selected ( left ) , corresponding to @xmath51 , or the smallest eigenvalues ( right ) , corresponding to @xmath52 .
_ , title="fig:",scaledwidth=45.0% ] , with either the largest eigenvalues selected ( left ) , corresponding to @xmath51 , or the smallest eigenvalues ( right ) , corresponding to @xmath52 .
_ , title="fig:",scaledwidth=45.0% ] the moment generating function is given by the laplace transform of @xmath240 : @xmath241 where the sum runs over the @xmath236 configurations @xmath33 .
the problem of choosing @xmath31 eigenvalues among @xmath10 , with the constraint that @xmath18 is fixed , is equivalent to placing @xmath31 particles on @xmath10 `` energy levels '' @xmath242 , with the constraint that the total `` energy '' is fixed .
since an eigenvalue can be chosen only once , an `` energy level '' can host only one particle .
therefore these particles behave like @xmath31 fermions .
the parameters @xmath48 and @xmath18 are thus related to two fundamental properties of the fermion gas : @xmath243 the set @xmath33 coincides with the occupation numbers of the energy levels : @xmath152 if level @xmath244 is occupied and @xmath29 otherwise . with this interpretation , ( [ eq : defgenfun ] ) is simply related to the canonical partition function @xmath245 } \ , \constn \hspace{0.3cm}\mbox{where}\hspace{0.3 cm } e_\mathrm{ferm}[\lbrace \m_i \rbrace ] = \sum_{i=1}^n \m_i \varepsilon_i \:,\ ] ] via the relation @xmath246 the parameter @xmath247 now plays the role of an inverse temperature for the fermions , not to be confused with the dyson index @xmath248 .
it is natural to introduce the grand - canonical partition function , which can be calculated straightforwardly : @xmath249 where @xmath250 is the fugacity . for large @xmath10 ,
the distribution of the eigenvalues @xmath233 is the marenko - pastur law @xmath191 .
therefore in this limit we can write : @xmath251 we can go back to the canonical partition function using the relation @xmath252 where the integral runs over a contour which encloses the origin once in the counter - clockwise direction . for large @xmath10 , this integral can be estimated using a saddle point method . in the thermodynamic limit @xmath20 , this corresponds to use the equivalence between the canonical and grand - canonical ensembles through a legendre transform .
the value @xmath253 of the fugacity is fixed by imposing that the mean number of fermions is @xmath31 : @xmath254 then eq . ( [ eq : canfromgcan ] ) yields the free energy per fermion in the thermodynamic limit : @xmath255 note that we dropped the factor @xmath247 in the usual definition of the free energy for convenience .
we get : @xmath256 eq .
( [ eq : momparfun ] ) shows that the logarithm of the moment generating function , namely the cumulant generating function , is given by the free energy of the fermions : @xmath257 the cumulants of ( [ eq : deftrlinstat0 ] ) can be obtained by expanding @xmath258 as a power series : @xmath259 the procedure is as follows : first determine @xmath260 as a power series in @xmath247 from ( [ eq : constk ] ) . then plug this expression into ( [ eq : freee ] ) to get the expansion of @xmath261 in powers of @xmath247 , and read the cumulants from the coefficients of this expansion . in particular , for @xmath195 ,
the first six cumulants are given by : @xmath262 the first two cumulants coincide with those obtained before , eqs .
( [ eq : means],[eq : vars ] ) , as it should .
in addition , we have performed an orthogonal polynomial calculation in the case @xmath107 @xcite .
this provides a check of our expressions as we have obtained that the first four cumulants given by this second method perfectly match the above expressions .
we also stress the @xmath263 symmetry of the cumulants : under this change , the even order cumulants are unchanged , while the odd ones change sign .
consequently , for @xmath264 , all odd order cumulants vanish : the distribution is symmetric around its mean @xmath265 .
this can be understood as the symmetry between fermions and holes .
it is interesting to compare these cumulants with the cumulants of the wigner time delay @xcite , corresponding to @xmath127 : @xmath266 ( the variance was also found in @xcite ; ref .
@xcite has given the four first cumulants explicitly ) .
the vanishing of the cumulants ( [ eq : variance], ... ,[eq : cumul6 ] ) for @xmath127 can be straightforwardly understood from the fermionic nature of the problem : when @xmath267 , the occupation numbers are all equal to one , and therefore do not fluctuate . however , ( [ eq : variance], ... ,[eq : cumul6 ] ) correspond to the dominant terms of expansions in powers of @xmath7 .
when @xmath127 the leading terms vanish , and only subleading contributions remain . this explains why the cumulants do no present the same scaling with @xmath10 : @xmath268 the fluctuations are much larger when @xmath269 .
this observation on the scaling of the moments has a simple interpretation : the energy of the @xmath10 interacting particles scales as @xmath270 due to the long range nature of the interaction , and the entropy of the @xmath3 fictitious fermions scales as @xmath271 . in usual studies of linear statistics ,
the energy dominates , and the entropy @xmath272 is zero because all the occupation numbers are fixed to @xmath152 .
hence the fluctuations scale as @xmath273 . but here , for @xmath124 , the energy is frozen and the additional entropy dominates .
therefore the fluctuations scale as @xmath274 .
this interpretation also explains why the cumulants do not depend on the dyson index @xmath61 when @xmath124 : this index is present in the gibbs measure @xmath275 $ ] , but not in the measure of entropic nature @xmath276 $ ] ( as stressed in [ subsec : discdistentropy ] , the entropy of the fermions should not be confused with the entropy of the density of the coulomb gas ) .
we have also performed numerical simulations to check our results .
figure [ fig : histnum ] shows histograms obtained by diagonalizing @xmath277 complex hermitian matrices ( @xmath278 ) of various sizes .
these histograms are compared to the approximate distribution reconstructed using a edgeworth series , from the large @xmath10 expressions of the cumulants ( [ eq : variance], ... ,[eq : cumul4 ] ) . on purpose ,
we first consider rather small matrices , of size @xmath279 ( fig .
[ fig : histnum ] , left ) : we can see a significant deviation , which we mostly attribute to the large @xmath10 approximation of the cumulants . for still moderate size @xmath280 ( fig .
[ fig : histnum ] , right ) , we see that the agreement is already very good .
the plots of figure [ fig : histnum ] correspond to @xmath281 .
note that , increasing @xmath48 with fixed @xmath10 , we have observed that the agreement becomes less good , although cumulants are symmetric under @xmath282 at leading order in @xmath10 ; this is explained by the fact that finite @xmath10 corrections increase with @xmath48 , cf .
we have also performed the calculation up to @xmath283 where the histogram is almost indistinguishable from the simple gaussian approximation . , for @xmath195 , obtained from @xmath277 complex matrices ( @xmath107 ) of various sizes , for @xmath284 .
the lines correspond to a reconstruction of the distribution using a edgeworth series , using the first two ( dashed ) and four ( solid ) asymptotic forms of the cumulants .
_ , title="fig:",scaledwidth=45.0% ] , for @xmath195 , obtained from @xmath277 complex matrices ( @xmath107 ) of various sizes , for @xmath284 .
the lines correspond to a reconstruction of the distribution using a edgeworth series , using the first two ( dashed ) and four ( solid ) asymptotic forms of the cumulants .
_ , title="fig:",scaledwidth=45.0% ] the distribution of the truncated linear statistics ( [ eq : deftrlinstat0 ] ) can be obtained in the interval @xmath285 $ ] by laplace inversion : @xmath286 for large @xmath10 , we can estimate this integral with a saddle point method .
this corresponds to performing another legendre transform on the free energy @xmath261 . in section [ subsec : momentsgenfct ] we already performed a legendre transform to go from grand - canonical to canonical ensemble .
thus , here we perform a second one to reach the microcanonical ensemble where the energy ( i.e. @xmath18 ) and the fermion number ( i.e. @xmath48 ) are fixed . using ( [ eq : cumulgenfctasymp ] ) for the expression of @xmath287 , we obtain that the saddle point @xmath288 is given by @xmath289 using the expression of @xmath261 , this equation reads : @xmath290 where we denoted @xmath291 .
then the distribution becomes : @xmath292 } \label{eq : pnsphi0}\ ] ] where we introduced the large deviation function @xmath293 , which is given by : @xmath294 explicitly , it reads : @xmath295 where @xmath296 and @xmath297 are fixed by eq .
( [ eq : constsldf ] ) and @xmath298 the observations we made earlier discussing the cumulants are also valid for the full distribution of @xmath18 , eqs .
( [ eq : pnsphi0],[eq : ldfphi0 ] ) . for @xmath299 $ ]
the distribution @xmath42 does not depend on the dyson index @xmath61 .
moreover , the logarithm of the probability scales as @xmath10 whereas it scales as @xmath180 for @xmath139 , indicating a distribution of width @xmath60 , much broader that @xmath7 for @xmath127 .
the physical interpretation of @xmath293 is made clear by eq .
( [ eq : phi0fsbeta ] ) : @xmath18 is the energy of the fermions , @xmath261 their free energy ( up to a factor @xmath247 ) and @xmath296 is the inverse temperature .
therefore , @xmath293 is a difference of entropy : @xmath300 where @xmath301 is the entropy of the @xmath302 fermions with total energy @xmath18 .
the condition @xmath303 determines @xmath304 .
the term @xmath305 comes from the normalization of @xmath42 , namely the denominator in eq .
( [ eq : pnsratiomultint ] ) .
in particular , the number of configurations @xmath33 associated to the value @xmath18 is exactly @xmath306 $ ] .
the values of the inverse temperature @xmath296 and the fugacity @xmath297 are obtained in terms of @xmath48 and @xmath18 by eqs .
( [ eq : constsldf],[eq : constkm ] ) .
this corresponds to summing over all the eigenvalues ( equivalently , the energy levels ) , with a weight given by the fermi - dirac distribution @xmath307 in addition , the eigenvalues are distributed according to the optimal distribution in the absence of constraint , i.e. the marenko - pastur distribution @xmath191 .
note that the parameter @xmath247 , which represents the inverse temperature of the fermions , can be either positive or negative .
this quite unusual situation occurs because the spectrum @xmath308 is bounded from below and above .
the fermi - dirac distribution ( [ eq : fermidirac ] ) interpolates between two step - functions : in the limit @xmath309 , it selects only the lowest energy levels ( in our case , the largest eigenvalues since @xmath21 is decreasing , as shown in fig [ fig : mpminmaxfd ] , left ) .
the opposite limit @xmath310 corresponds to select only the highest energy levels ( smallest eigenvalues , see fig [ fig : mpminmaxfd ] , right ) . ) in the case @xmath195 , for @xmath311 ( left ) and @xmath312 ( right ) , superimposed to the eigenvalue distribution @xmath191 .
shaded areas indicate the contribution of the eigenvalues to the linear statistics .
note that the case @xmath311 corresponds to select the largest eigenvalues , hence the smallest `` energies '' @xmath313 , as usual .
_ , title="fig:",scaledwidth=45.0% ] ) in the case @xmath195 , for @xmath311 ( left ) and @xmath312 ( right ) , superimposed to the eigenvalue distribution @xmath191 .
shaded areas indicate the contribution of the eigenvalues to the linear statistics .
note that the case @xmath311 corresponds to select the largest eigenvalues , hence the smallest `` energies '' @xmath313 , as usual .
_ , title="fig:",scaledwidth=45.0% ] the derivation we followed in this section can be generalized straightforwardly to other matrix ensembles .
for example in the gaussian ensemble , one should replace the marenko - pastur distribution @xmath191 by the wigner semicircle law . because the density is given by the density obtained in the absence of constraint , the scenario described here is completely universal and valid for any monotonous function @xmath21 ( the case of @xmath21 non monotonous is discussed in section [ sec : conclusion ] ) .
the large deviation function @xmath293 being obtained from a saddle point estimate , eq .
( [ eq : saddleldf ] ) can be used to prove an identity similar to ( [ eq : thermoidcoulomb ] ) : @xmath315 from this relation , it is clear that the maximum of @xmath272 ( minimum of @xmath293 ) is given by @xmath316 , corresponding to @xmath317 .
this means an infinite temperature for the fermions : the fermi - dirac distribution ( [ eq : fermidirac ] ) is flat , and all the energy levels are occupied with probability @xmath318 .
expanding eqs .
( [ eq : constsldf],[eq : constkm ] ) for @xmath247 near @xmath319 allows to compute @xmath296 near @xmath320 . in the case
@xmath195 , this gives : @xmath321 using ( [ eq : phi0diffentropy],[eq : thermoidphi0 ] ) , we obtain : @xmath322 as we have already seen , the distribution @xmath42 is dominated by a gaussian peak located at @xmath323 , with variance given by ( [ eq : variance ] ) .
this expression of @xmath293 is singular when @xmath136 .
but as we discussed in section [ subsubsec : cumulants ] , this corresponds only to the leading term of an expansion in powers of @xmath7 of the variance .
. shows that the distribtuion is regular for @xmath130 : @xmath324 } \:.\ ] ] this expression describes correctly the limit @xmath325 for @xmath18 close to @xmath48 and account for the transition between fluctuations @xmath60 for @xmath124 and fluctuations @xmath326 for @xmath127 ( the crossover between the two regimes obviously occurs at @xmath327 ) . the limits @xmath329 and @xmath330 correspond to the inverse temperature @xmath331 and @xmath332 , respectively .
this means zero temperature for the fermions : the fermi - dirac distribution ( [ eq : fermidirac ] ) becomes a step function , selecting either the largest or the smallest eigenvalues , as shown in fig .
[ fig : mpminmaxfd ] .
therefore , eqs .
( [ eq : constsldf],[eq : constkm ] ) involving integrals of this distribution can be approximated via a standard sommerfeld expansion @xcite : for any function @xmath333 and fixed chemical potential @xmath334 , one has @xmath335 and a similar expression holds for @xmath310 . in our case , since @xmath336 , it is convenient to introduce @xmath337 , which delimits the domain of the spectrum of eigenvalues which is `` occupied '' , as shown in fig .
[ fig : mpminmax ] . the fugacity is simply related to the chemical potential @xmath334 by @xmath338 . for @xmath331 ( corresponding to @xmath329 ) , eq .
( [ eq : constkm ] ) allows to compute @xmath339 as a power series @xmath340 using the sommerfeld expansion ( [ eq : sommerfeld ] ) .
this gives @xmath341 then , following the same procedure with eq .
( [ eq : constsldf ] ) yields @xmath342 thus @xmath343 a similar computation allows to estimate the integral in eq .
( [ eq : ldfphi0 ] ) .
finally , we obtain : @xmath344 where @xmath345 .
the case @xmath332 ( corresponding to @xmath330 ) can be treated in the same way .
we obtain : @xmath346 where @xmath347 is now fixed by @xmath348 ( entropy of the fermions ) describing in particular the typical fluctuations of the truncated linear statistics . the cases @xmath216 and @xmath349 are both represented .
the dashed horizontal lines delimit the maximum @xmath350 of these functions .
_ , scaledwidth=70.0% ] note that @xmath293 reaches its maximal value @xmath351 for @xmath352 and @xmath59 .
this particular value has a very simple meaning : @xmath42 is the ( normalized ) number of configurations among the @xmath236 which give the value @xmath18 . for @xmath352 or @xmath59 , only one configuration exists , corresponding either to selecting the largest or the smallest eigenvalues .
therefore , @xmath353 . in the large @xmath10 limit , @xmath354 and similarly for @xmath59
the distribution of the truncated linear statistics ( [ eq : deftrlinstat0 ] ) for @xmath356 was determined by considering two different situations .
first , for @xmath139 , the fraction @xmath48 of the largest eigenvalues detach from the others , as we have seen in section [ sec : coulombgas ] . when @xmath18 is increased , these two bulks of eigenvalues move closer , until they merge for @xmath202 ( fig . [
fig : mpminmax ] , left ) .
then , when @xmath18 is further increased in the interval @xmath232 $ ] , the density of eigenvalues is frozen and the occupation numbers @xmath33 fluctuate , as it was shown in section [ sec : typfluct ] .
when @xmath18 reaches @xmath59 , these occupation numbers no longer fluctuate and only the smallest eigenvalues are selected ( fig .
[ fig : mpminmax ] , right ) . in the domain
@xmath357 we could naively expect a similar scenario as for the tail @xmath150 , namely to detach the fraction @xmath48 of the smallest eigenvalues .
however , in the case @xmath215 , there is another scenario more favorable energetically .
it is enough to detach one eigenvalue ( the smallest , @xmath358 ) to obtain large values of @xmath18 : a `` bulk '' of @xmath359 eigenvalues can remain frozen while only one eigenvalue contributes to the @xmath18-dependent part of the energy , which thus scales as @xmath8 instead of @xmath6 as in phase i. we write @xmath360 if the smallest eigenvalue @xmath358 scales as @xmath361 , it gives a `` macroscopic '' contribution of the same order as the other @xmath359 eigenvalues which remain of order @xmath19 . because @xmath358 is much smaller that the other eigenvalues in the limit @xmath20 , the constraint @xmath362 in the multiple integrals ( for @xmath11 ) plays no role .
we estimate the integral over @xmath363 with a saddle point method .
the saddle point denoted @xmath233 , corresponds to the minimum of the energy @xmath364 $ ] , eq .
( [ eq : minediscr ] ) .
thus , these eigenvalues are frozen and distributed according to the marenko - pastur density @xmath191 .
the corresponding energy is : @xmath365 \simeq \mathscr{e}[\rhomp ] - \frac{1}{n } \ln x_n + \frac{\mathscr{c}}{n } \:,\ ] ] where the constant @xmath366 arises from a careful treatment of @xmath7 corrections @xcite .
the same procedure can be carried out with the denominator , but this time with all the eigenvalues since they are not constrained .
we obtain : @xmath367 note that the substitution in the multiple integrals in give contributions of the entropy of the coulomb gas @xmath368 $ ] also of order @xmath369 , however the leading term of such contributions are the same in the numerator and the denominator and thus cancel .
integration over the last eigenvalue is straightforward : @xmath370 \right\rbrace \constn \:.\ ] ] in order to compute this sum , let us rewrite it as @xmath371 + \mathscr{c } ) } \sum_{\lbrace \m_i \rbrace } \delta \left ( u - \frac{1}{n } \sum_{i=1}^{n-1 } \frac{\m_i}{x_i^\star } \right ) \constn \ : .
\label{eq : pnslargesum}\ ] ] we now recognize the distribution of @xmath18 computed in section [ sec : typfluct ] : @xmath372 where @xmath293 is given by ( [ eq : ldfphi0 ] ) .
therefore , ( [ eq : pnslargesum ] ) reduces to @xmath373 + \mathscr{c } ) - n \phi_0(\kappa;u ) \right\rbrace \:.\ ] ] this last integral can be computed using a saddle point method .
the saddle point @xmath58 is given by : @xmath374 the quantity @xmath375 represents the contribution of the frozen bulk to the tls .
its limiting behaviours are found by using and , where the latter is rewritten @xmath376 .
some elementary algebra gives @xmath377
\kappa + \frac{\beta\kappa(1-\kappa)}{2(s-\kappa ) } + \mathcal{o}(s^{-3 } ) & \mbox{for } s\to\infty \end{cases}\ ] ] finally , the probability is given by : @xmath378 \right\rbrace \hspace{0.25 cm } \text{for } \hspace{0.25 cm } s > s_1(\kappa ) } \ ] ] where the large deviation function is @xmath379 - \mathcal{e}[\rhomp(x ) ] ) -\ln n = \ln ( s - \tilde{s}(s ) ) + \mathscr{c } \:,\ ] ] where @xmath380 .
there are two different contributions to the distribution of @xmath18 : * the energy of the isolated eigenvalue @xmath358 , encoded in @xmath381 ; * the entropy associated to the possible configurations @xmath33 associated to the @xmath359 frozen eigenvalues , encoded in @xmath293 .
the latter was not present in the case @xmath127 studied in ref .
@xcite because all the occupation numbers were fixed to @xmath152 and do not bring any additional entropy .
using , @xmath382 , gives @xmath383 and @xmath384 where @xmath385 was defined in eq . .
therefore , the leading order term @xmath386 is independent of the dyson index @xmath61 , which only appears in the constant .
the limiting behaviour , @xmath388 , shows that the entropy no longer plays a role and the contribution of the energy reads : @xmath389 which is again independent of @xmath61 .
this gives the right tail of the distribution : @xmath390 we have recovered the tail obtained in ref .
@xcite for @xmath127 : @xmath391 for @xmath392 , as expected since the this limit is described within the same scenario ( one isolated charge and a frozen bulk ) . from the physical point of view , this power law tail is interpreted as the manifestation of a very narrow resonance which dominate the sum of proper times .
as briefly reviewed in the introduction , the study of linear statistics has played a very important role in random matrix theory . in this article
we have introduced a new type of question by considering the statistical analysis of partial sums of eigenvalues @xmath393 , so - called `` _ _ truncated linear statistics _ _ '' ( tls ) , with @xmath394 , where @xmath10 is the size of the matrices .
this question has been set within the laguerre ensemble for @xmath11 .
the study of this particular model was motivated by the analysis of partial sums of proper time delays in chaotic scattering , which are characteristic times of the scattering process playing an important role in electronic transport .
the specificity of the problem has led us to introduce two sets of random variables , the ordered eigenvalues @xmath395 and the `` occupation numbers '' @xmath25 , with @xmath396 or @xmath397 , leading to express the tls as @xmath398 .
correspondingly , we have introduced two different physicial interpretations : * the eigenvalues @xmath399 can be interpreted as the positions of a one - dimensional gas of particles with long range ( logarithmic ) interactions , as usual in such problems . *
the variables @xmath45 can be viewed as occupation numbers for @xmath3 fictitious non - interacting fermions in @xmath10 `` energy levels '' @xmath400 .
we have determined the large @xmath10 behaviour of the distribution @xmath46 of the tls , has shown that the large @xmath10 result describes very well the distribution already for @xmath401 . ] which was shown to be controlled by an optimal configuration ( solution of saddle point equations ) , characterised by the density @xmath175 of eigenvalues ( with @xmath402 for the laguerre ensemble ) , while the optimal occupations are given by the fermi - dirac distribution for an effective temperature , eq . .
we have shown that varying the two parameters @xmath49 drives phase transitions in this optimal configuration .
[ [ phase - ii . ] ] phase ii. + + + + + + + + + + in the absence of any constraint on the eigenvalue density , the ( most probable ) density is the marenko - pastur distribution ( fig .
[ fig : mpminmax ] ) . for a fixed @xmath48
, this density can be associated to a whole spectrum of values of the tls @xmath403 $ ] , corresponding to different choices for the occupation numbers @xmath45 .
the two extreme situations correspond to choosing the @xmath3 largest or the @xmath3 smallest eigenvalues : cf .
[ fig : mpminmax ] .
this phase exists in a domain of the phase diagram bounded by two lines @xmath59 and @xmath201 ( fig .
[ fig : phdiag ] ) .
the energy of the coulomb gas is frozen and the structure of the distribution is entirely due to the _ entropy _ of the @xmath3 fictitious non - interacting fermions , which scales as @xmath8 and is independent of the symmetry index @xmath61 , thus @xmath404 , where the large deviation function @xmath405 has been determined in section [ sec : typfluct ] . as a consequence of this scaling ,
the relative fluctuations of the tls are of order @xmath60 ( whereas full linear statistics usually present relative fluctuations @xmath326 as a consequence of the strong correlations in the coulomb gas ) . as the limits @xmath406 and @xmath407 correspond to select the contributions of the largest or the smallest eigenvalues ( fig .
[ fig : mpminmax ] ) , they are associated to effective ( absolute ) fermionic temperature @xmath408 positive or negative , respectively ( see also fig . [
fig : mpminmaxfd ] ) .
this new scenario is completly universal ( independent of the function @xmath21 and the specific matrix ensemble ) .
[ [ phase - i . ] ] phase i. + + + + + + + + + when @xmath18 reaches the lower boundary @xmath201 of the central domain of the phase diagram [ fig : phdiag ] , the effective temperature is fixed to @xmath409 and only the largest eigenvalues contribute to the tls ( fig .
[ fig : mpminmax ] ) .
smaller value @xmath172 can only be realised by a deformation of the optimal density @xmath175 , which is split in two bulks , while occupations are frozen ( @xmath410 ) .
the weight of the optimal configuration is now controlled by the _ energy _ of the coulomb gas . in the @xmath411 limit
, we have @xmath412 , hence the energy is dominated by the confinment energy : @xmath413 .
this corresponds to the behaviour @xmath414 $ ] with @xmath415 ( see section [ sec : coulombgas ] ) .
[ [ phase - iii . ] ] phase iii. + + + + + + + + + + + the case where @xmath355 was analysed with a mixed scenario . as for @xmath127 @xcite ,
one remarks that it is sufficient to split off a single eigenvalue from the bulk , the smallest one @xmath416 , in order to generate large values of the tls @xmath417 : i.e. it is more favorable energetically to send one charge towards @xmath319 as @xmath18 grows , which generates an energy cost @xmath418 , than move the full bulk , which would generate a cost @xmath419 .
when the charge is split off the bulk , there is also an entropic contribution @xmath8 due to the fluctuating @xmath359 occupation numbers related to the remaining @xmath359 eigenvalues , of the same order as the interaction energy between the isolated eigenvalue and the frozen @xmath359 ones .
this part of the scenario is specific to the function @xmath11 that we have considered . for the case considered in the paper ,
we have obtained three phases ( fig .
[ fig : phdiag ] ) , however one can imagine that for another function @xmath21 , the phase diagram would involve more than three phases : the exhaustive study of ref .
@xcite for several linear statistics illustrates than the number of different phases depend on the function @xmath21 .
as the two phases i and ii involve two different scalings with @xmath10 , see , it is interesting to discuss further the transition at @xmath420 , in the vicinity of which the distribution presents the limiting behaviours @xmath421 } \simeq \e^{+n \cp \sqrt{s - s_0(\kappa ) } } & \text { for } s >
s_0(\kappa ) \end{array } \right .
\label{eq : pnktr } \:,\ ] ] where @xmath422 , cf .
eq . , and @xmath65 is defined in eq . .
from the point of view of the coulomb gas , @xmath51 corresponds to a _ freezing transition _ : below @xmath201 , the energy scales as @xmath423 , whereas the energy is frozen above the transition .
this corresponds to a _
second order _
phase transition ( a second order freezing transition of different nature was identified in @xcite , and also in @xcite although the transition was not interpreted along these lines in this latter reference ) .
note that the energy of the coulomb gas , eq . , is @xmath424 everywhere above the line @xmath201 . in the thermodynamic limit @xmath47 ,
@xmath425 becomes flat above this line and all derivatives of the energy are trivially continuous across @xmath59 .
a finer description of the transition ( beyond the thermodynamic limit ) requires to identify the non trivial scaling with @xmath10 for the crossover , given by @xmath426 this leads to introduce a scaling variable @xmath250 and rewrite the crossover between the two behaviours in terms of a unique function @xmath427 the scaling function @xmath428 describes the detail of the crossover for finite @xmath10 , at a scale @xmath429 , and thus presents the limiting behaviours @xmath430 similar considerations for the distribution of the largest eigenvalue in various matrix ensembles ( see the review @xcite ) have shown that the corresponding function is universal ( tracy widom distribution given by painlev transcendents )
. an interesting challenging open question would be to determine whether the function @xmath428 has also a universal character and find its precise nature .
the discussion concerning the crossover around @xmath201 can not be extended to the other phase boundary at @xmath59 , as it is not clear how the limiting behaviours obtained above , @xmath431 for @xmath432 and @xmath433 $ ] for @xmath434 , can be matched . due to the @xmath10 in the argument of the logarithm ,
it does not seem that the crossover can be described by a universal function as for @xmath435 .
a precise description of the crossover seems therefore even more challenging in this case .
the new universal scenario introduced in the paper was illustrated for a particular ensemble ( laguerre ) and a specific _ monotonous _ function @xmath21 .
we argue now that the same main features would also occur for a _ non monotonous _ function .
consider for concreteness the jacobi case , @xmath436 $ ] and the function @xmath437 ( when @xmath438 is the shot noise power of chaotic quantum dots @xcite ) . in phase ii , due to the fact that the `` energy levels '' @xmath54 are non monotonous as a function of the eigenvalues of the random matrix , the occupations controlled by the fermi - dirac distribution @xmath439 would also be represented by a non monotonous function of @xmath440 (
a similar feature occurs when representing the occupation of plane waves by one - dimensional non - interacting fermions : occupation is a monotonous function of the energy @xmath441 but a non monotonous function of the wave vector @xmath442 ) .
the limit @xmath443 ( lower boundary of phase ii ) should correspond to `` cold fermions '' with @xmath444 where the occupation function @xmath439 selects both the smallest and the largest eigenvalues .
the large deviations for @xmath445 ( phase i ) should correspond then to the opening of two gaps .
the upper boundary of phase ii , @xmath446 , corresponds to `` hot fermions '' with negative temperature @xmath447 involving the eigenvalues in the middle of the interval @xmath50 $ ] ; the large deviations for @xmath434 should also correspond to the opening of two gaps .
such a situation for a non monotonous @xmath21 could certainly be worth studying in details .
coming back to the initial motivation within chaotic scattering , the paper has mostly focused on the question of the partial sums of proper time delays .
we have also emphasized the role of the wigner - smith matrix diagonal elements @xmath99 , which are important for applications in quantum transport : the statistical properties of the partial sum of these matrix elements were analysed in the appendix for @xmath107 , based on the relation between @xmath99 and partial time delays @xmath98 @xcite . because the joint distribution of the @xmath99 s is still unknown ,
a general analysis of their partial sums in the general case , beyond the two first moments , remains a challenging question .
we are indebted to dmitry savin for many useful discussions on chaotic scattering .
in this section , we derive the distribution of the injectance of a quantum dot defined in section [ sec : quantscatt ] : @xmath448 we will restrict our study to the situation where @xmath107 because the joint distribution of the @xmath99 s is known only in this case .
we will derive the distribution of @xmath84 using a coulomb gas approach .
the computation is very similar to the one described in @xcite where the wigner time delay ( case where all contacts contribute ) is studied .
we start from the distribution of a subblock @xmath449 of size @xmath450 on the diagonal of @xmath86 given in @xcite for @xmath107 : @xmath451 if we suppose that the contact @xmath80 has @xmath31 open channels , we can rewrite the injectance as : @xmath452 denoting @xmath53 the eigenvalues of @xmath453 , we can write @xmath454 where the joint probability density function of the @xmath53 is given by @xmath455 rescale the eigenvalues as @xmath456 and introduce the empirical density @xmath457 the distribution of @xmath458 is given by @xmath459 } \ : \delta \left(\int \rho(x ) \dd x - 1 \right ) \delta \left(\int \frac{\rho(x)}{x } \dd x - s \right ) } { \displaystyle \int \mathcal{d}\rho \ : \e^{- \kappa^2 n^2 \mathscr{e}[\rho ] } \ : \delta \left(\int \rho(x ) \dd x - 1 \right ) } \ : , \label{eq : distinject}\ ] ] where we denoted @xmath460 and introduced the energy ( see section [ sec : coulombgas ] ) @xmath163 = - \int \dd x \int \dd y \ : \rho(x ) \rho(y ) \ln { \left| x - y \right| } + \int \dd x \left(x - \frac{1}{\kappa } \ln
x \right ) \rho(x ) \:.\ ] ] the integrals in ( [ eq : distinject ] ) are dominated by the minimum of the energy under the constraints imposed by the @xmath176-functions .
therefore , we introduce the functional @xmath461 = \mathscr{e}[\rho ] + \mu_0 \left ( \int \rho - 1 \right ) + \mu_1 \left ( \int \frac{\rho(x)}{x } \dd x - s
\right ) \:,\ ] ] where @xmath462 and @xmath186 are lagrange multipliers .
denote @xmath175 the density which dominates the numerator of ( [ eq : distinject ] ) .
it is given by @xmath463 , which reads : @xmath464 it is more convenient to derive this relation with respect to @xmath465 : @xmath466 the values @xmath467 and @xmath468 taken by the lagrange multipliers are fixed by imposing the constraints @xmath469 similarly , denote @xmath470 the density which dominates the denominator .
the distribution of @xmath18 is then given by : @xmath471-\mathscr{e}[\rho_0^\star(x ) ] \right ) \right\rbrace \:.\ ] ] the thermodynamic identity ( [ eq : thermoidcoulomb ] ) mentioned in the body of the paper allows a direct computation of the energy via the relation : @xmath472}{\mathrm{d}s } } = - \mu_1^\star(\kappa;s ) \ : .
\label{eq : thermoidapp}\ ] ] our aim is now to compute the density @xmath181 .
as in the body of the paper , depending on the values of the parameters @xmath48 and @xmath18 , we will find different types of densities @xmath181 , which we interpret as different `` phases '' .
let us assume that the solution @xmath181 has one compact support @xmath184 $ ] . then , eq .
( [ eq : minfappder ] ) can be solved using tricomi s theorem @xcite .
we obtain : @xmath473 it is convenient to parametrize this solution in terms of @xmath474 .
then , imposing @xmath475 , along with the constraints ( [ eq : appconst ] ) yields : @xmath476 @xmath477 @xmath478 @xmath479 given @xmath48 and @xmath18 , the last equation allows to compute @xmath480 , from which all the other parameters are deduced .
one can check that in the limit @xmath130 , one recovers the equations given in @xcite .
this solution exists as long as @xmath181 is positive , which corresponds to @xmath483 .
this gives the condition @xmath484 which can be rewritten as @xmath485 .
this corresponds to the lower domain delimited by the upper solid line on figure [ fig : appdiag ] ( left ) .
the typical fluctuations are controlled by the minimum of the energy , which is given by @xmath486 .
the density is then @xmath487 where @xmath488 the corresponding value of @xmath18 is @xmath48 , and expanding eqs .
( [ eq : appmu],[eq : apps ] ) for @xmath18 close to @xmath48 yields : @xmath489 the energy is obtained by simple integration , via eq .
( [ eq : thermoidapp ] ) : @xmath490-\mathscr{e}[\rho_0^\star(x ) ] = \frac{(s-\kappa)^2}{2\kappa^3(\kappa+1 ) } + \o((s-\kappa)^3 ) \:.\ ] ] thus , the distribution of @xmath18 near @xmath48 is given by : @xmath491 we recover the leading term of the variance given in the introduction , eq .
( [ eq : variancesumqii ] ) : @xmath492 the limit @xmath150 corresponds to @xmath493 . expanding eqs .
( [ eq : appv],[eq : appmu],[eq : appc],[eq : apps ] ) in this limit gives : @xmath494 using again eq .
( [ eq : thermoidapp ] ) , we obtain : @xmath490 = \frac{1}{s } + \frac{\kappa+2}{2\kappa } \ln s + \o(1 ) \:,\ ] ] thus : @xmath495 as for the the wigner time delay @xcite ( @xmath127 ) and in section [ sec : ldflarge ] , we look for a solution with an isolated eigenvalue : @xmath496 with now @xmath497 .
the minimization of @xmath498 with respect to @xmath499 reads : @xmath500 and minimization with respect to @xmath501 gives @xmath502 in addition , the constraint becomes : @xmath503 for @xmath501 to give a `` macroscopic '' contribution to @xmath18 , we must have @xmath504 .
then equation imposes @xmath505 .
finally , at leading order in @xmath31 , we get : @xmath506 @xmath507 @xmath508 this solution remains valid as long as the separate eigenvalue is away from the bulk , namely @xmath509 .
this gives the condition @xmath510 this corresponds to the upper domain represented on figure [ fig : appdiag ] ( left ) .
note that @xmath511 as @xmath20 .
the energy can be computed analytically at leading order : @xmath512 - \mathcal{e}[\rho_0^\star ] = \frac{1}{\kappa \nt } \ln[\nt ( s - \kappa ) ] + \frac{\mathscr{c}}{\nt } + \o(\nt^{-2 } ) .
\label{eq : e_ph_sep_ch}\ ] ] where @xmath513 was introduced above .
the constant term is the same as the one appearing in section [ sec : ldflarge ] and ref .
it arises from corrections of order @xmath514 to the density @xmath181 @xcite . from the expression of the energy , we deduce the expression of the distribution of @xmath18 , for @xmath515 : @xmath516 the simplification @xmath517 has thus produced the same exponent as for the tail of the distribution of the sum ( truncated or not ) of proper times .
this is explained from the interpretation of ref .
@xcite , where it was shown that @xmath99 coincides with the partial time delay for @xmath107 .
we obtained the following scalings for the distribution : @xmath518 ( \kappa n)^{-n } \exp \left\lbrace - n \psi_+(\kappa;s ) \right\rbrace \hspace{0.5 cm } & \text{for } s > s_{ii}(\kappa ) \end{cases}\ ] ] where the large deviation @xmath519 is given by : @xmath520 and @xmath521 has the following limiting behaviours : @xmath522 \displaystyle \frac{(s-\kappa)^2}{2\kappa^3(\kappa+1 ) } + \o((s-\kappa)^3 ) \hspace{0.5 cm } & \text{as } s \to \kappa \end{cases}\ ] ] the precise point @xmath523 where the transition between the two phases occurs can be obtained by matching the two large deviation functions .
one can show that , for @xmath20 , @xmath524 . therefore for large @xmath10
we have , for @xmath18 close to @xmath48 : @xmath525 0 & \text { for } s >
\kappa \end{array } \right.\ ] ] this corresponds to a _
second order _
phase transition .
this was already the case in ref .
@xcite in the study of the full linear statistics ( @xmath127 ) . .
the upper solid line corresponds to @xmath526 and delimits the domain of existence of phase i. the lower solid line is @xmath527 . phase ii exists above this line . as @xmath20
, it goes to the dashed line @xmath317 , where occurs the transition in the thermodynamic limit .
right : phase diagram for the coulomb gas.,title="fig:",scaledwidth=45.0% ] .
the upper solid line corresponds to @xmath526 and delimits the domain of existence of phase i. the lower solid line is @xmath527 . phase ii exists above this line .
as @xmath20 , it goes to the dashed line @xmath317 , where occurs the transition in the thermodynamic limit .
right : phase diagram for the coulomb gas.,title="fig:",scaledwidth=45.0% ] a. de pasquale , p. facchi , g. parisi , s. pascazio , and a. scardicchio , phase transitions and metastability in the distribution of the bipartite entanglement of a large quantum system , phys .
a * 81 * , 052324 ( 2010 ) .
a. grabsch and c. texier , distribution of spectral linear statistics on random matrices beyond the large deviation function wigner time delay in multichannel disordered wires , arxiv:1602.03370 ( 2016 ) .
a. grabsch and c. texier , capacitance and charge relaxation resistance of chaotic cavities joint distribution of two linear statistics in the laguerre ensemble of random matrices , europhys .
* 109*(5 ) , 50004 ( 2015 ) .
d. v. savin , y. v. fyodorov , and h .- j .
sommers , reducing nonideal to ideal coupling in random matrix description of chaotic scattering : application to the time - delay problem , phys .
e * 63 * , 035202 ( 2001 ) . | invariant ensembles of random matrices are characterized by the distribution of their eigenvalues @xmath0 .
we study the distribution of truncated linear statistics of the form @xmath1 with @xmath2 .
this problem has been considered by us in a previous paper when the @xmath3 eigenvalues are further constrained to be the largest ones ( or the smallest ) . in this second paper
we consider the same problem without this restriction which leads to a rather different analysis .
this question is motivated by the study of partial sums of proper time delays in chaotic quantum dots , which are characteristic times of the scattering process .
using the coulomb gas technique , we derive the large deviation function for @xmath4 .
large deviations of linear statistics @xmath5 are usually dominated by the energy of the coulomb gas , which scales as @xmath6 , implying that the relative fluctuations are of order @xmath7 . for the truncated linear statistics considered here ,
there is a whole region ( including the typical fluctuations region ) , where the energy of the coulomb gas is frozen and the large deviation function is purely controlled by an entropic effect . because the entropy scales as @xmath8 , the relative fluctuations are of order @xmath9 .
our analysis relies on the mapping on a problem of @xmath3 fictitious non - interacting fermions in @xmath10 energy levels , which can exhibit both positive and negative effective ( absolute ) temperatures .
we determine the large deviation function characterizing the distribution of the truncated linear statistics , and show that , for the case considered here ( @xmath11 ) , the corresponding phase diagram is separated in three different phases .
: 05.60.gg ; 03.65.nk ; 05.45.mt |
in cold atomic gases experiments of bose - einstein condensates(becs ) with blue - detuned laser beams , the critical velocity of superfluidity has been observed in simply - connected@xcite and multiply - connected@xcite systems . in both cases , nucleations of the quantized vortices
have been observed above the critical velocity .
the observed critical velocity in the presence of obstacles is smaller than that of landau s prediction@xcite .
our goal is to reveal a mechanism of the breakdown of superfluidity due to the vortex nucleation . as a step toward this aim , we seek for signals of instability in spectral properties and fluctuations in the present paper . theoretical studies of breakdown of superfluidity have been done by the gross - pitaevskii(gp ) equation@xcite in various setups@xcite .
these studies have reported many nontrivial phenomena such as nonlinear dynamics of solitons or vortices etc .
however , the properties of excitation spectrum related to the vortex nucleation in systems without translational or rotational symmetry have not been studied so far .
an exception is a study on soliton nucleation in infinite 1d system with a point - like defect or higher dimensional system with a sheet - like potential@xcite .
this work showed that dynamical local density fluctuations are enhanced near the critical velocity , as a precursor effect of breakdown of superfluidity .
since both solitons and quantized vortices are topological defects that kill superfluidity , we can expect that dynamical density fluctuations are enhanced near the critical velocity as a precursor effect of nucleation of vortices . in this paper , we study the stability of a bec confined in a two - dimensional torus with a moving gaussian potential by the gp and the bogoliubov equation@xcite .
our setup corresponds to a simplified model of the recent experiment of a bec in a ring trap with a rotating weak link@xcite .
since we focus on the effect of the gaussian potential , we neglect the surface and the centrifugal effects .
we consider an @xmath4-boson system confined in a two - dimensional torus @xmath5 with a moving gaussian potential @xmath6 , where the velocity of the potential @xmath7 with @xmath8 is anti - parallel to the unit vector @xmath9 of @xmath10-direction . in the mean - field approximation ,
the gp equation@xcite in the moving frame with the potential is given by @xmath11 where @xmath12 is the condensate wave function(the order parameter of bec ) , @xmath13}$ ] is the gaussian potential , @xmath14 is the strength of the repulsive interaction , and @xmath15 denotes the chemical potential that is determined by the condition @xmath16 .
the twisted periodic boundary condition in the moving frame is given by @xmath17 where @xmath18 denotes the unit vector of @xmath19-direction .
we note that the boundary condition ( [ eq : twisted_periodic_boundary_condition ] ) does not change under the transformation @xmath20 , where @xmath21 is an arbitrary integer . in the present work ,
we focus on the branch of the stationary flow states that reduces to the ground state when @xmath22 .
the bogoliubov equation@xcite is given by @xmath23 ^ 2 \\
g[\psi^{\ast}(\bm{r})]^2 & -\mathcal{l } \end{bmatrix } \begin{bmatrix } u_i(\bm{r } ) \\ v_i(\bm{r } ) \end{bmatrix } = \epsilon_i \begin{bmatrix } u_i(\bm{r } ) \\ v_i(\bm{r } ) \end{bmatrix } , \\ \mathcal{l}\equiv -\frac{\hbar^2}{2m}\nabla^2+u(\bm{r})-\mu+2g|\psi(\bm{r})|^2,\end{aligned}\ ] ] where @xmath24 and @xmath25 are the wave functions of the excited state @xmath26 and @xmath27 is the excitation energy of the state @xmath26 .
the wave functions of the excited states satisfy the following orthonormal conditions @xmath28=\delta_{i j}$ ] and @xmath29=0 $ ] and the boundary conditions @xmath30 , and @xmath31 .
we take the unit of the length , velocity , and energy , respectively , as the healing length @xmath32 , the sound velocity @xmath33 , and @xmath34 , where @xmath35 represents a mean - particle density . here
, we set @xmath36 .
our numerical calculations are performed in the @xmath37-space , that is , we expand @xmath38 , and @xmath39 with the plane wave @xmath40 , where @xmath41 and @xmath42 and @xmath43 are integers . rewriting the gp and the bogoliubov equations in terms of the expansion coefficients , we solve them numerically .
the solutions of the gp equation ( [ eq : gp_equation_in_the_moving_frame ] ) are obtained by the imaginary time evolution method .
the bogoliubov equation is diagonalized by the same technique as that used in ref .
we have found that the dependence of the present results on the number of the basis is negligibly small@xcite .
first , we present the stable stationary solution of the gp equation near the critical velocity in fig . [
fig : density_and_phase ] . here
, the critical velocity is defined as the modulus of the velocity far away from the obstacle , above which the stationary flow state becomes unstable . in the presence of the strong potential , we find that a bounded pair of vortices appears in the low density region .
this vortex - pair is called ghost vortex pair(gvp)@xcite .
although gvp exists , the excitation spectra ( shown later ) do not have anomaly and thus the stationary solution is stable or metastable .
other physical quantities except local phase spectral function ( discussed later ) do not have singularities either .
the reason why the stationary solution is metastable can be understood that the gvp is pinned to the potential and free vortex motions that cause the energy dissipation do not occur . in other parameter regions , particularly in weak potential strength cases(for example , @xmath44 and @xmath45 ) , the gvp does not appear in our numerical calculations .
we typically investigate the range of the velocity for @xmath46 , where @xmath47 is the critical velocity .
we conclude that the gvp does not exist in the weak potential case within this range of the velocity . for @xmath48 , @xmath49 , @xmath45 , @xmath50 .
white and black circles represent the positions of the gvp.,width=434 ] now we focus on the minimum eigenenergy @xmath51 ( which we call the energy gap ) of the bogoliubov equation in finite size systems .
we show the velocity dependence of the energy gap in fig .
[ fig : energy_gap](a ) , where we can see two features .
one is the linearly decreasing of @xmath0 as a function of @xmath1 in the small @xmath1 region .
this behavior reflects the property of the energy gap @xmath52 $ ] in the uniform system with the shape of two - dimensional torus .
another feature we find in fig .
[ fig : energy_gap](a ) is the steep decrease near @xmath53 .
we assume a form @xmath54^c$ ] with adjustable parameters @xmath55 , and @xmath56 to fit the four sets of data shown in fig .
[ fig : energy_gap](a ) near the critical velocity . figure [ fig : energy_gap](b ) shows that the data points near the critical velocity collapse onto a single power law as a function of @xmath57 with the exponent @xmath58 , which is nearly independent of the potential height , width , and the system size within our calculations(see table .
[ tab : exponent ] ) . and @xmath45 .
( b ) fitting result of ( a ) .
solid black line in ( b ) represents the line @xmath59^{1/4}$].,width=415 ] our results in finite size systems indicate that the energy gap vanishes at @xmath53 . from the viewpoint of non - linear physics , the origin of this behavior can be understood by the bifurcation of the solution of the non - linear differential equation . according to the bifurcation theory@xcite , eigenvalues of a linearized equation of an original non - linear differential equation become zero at the bifurcation point . here , the original equation , the linearized equation , and the bifurcation point correspond to the gp equation , the bogoliubov equation , and the critical velocity , respectively .
additionally , the eigenvalues obey a scaling law near the bifurcation point , which reflects the type of bifurcation .
the power law behavior(@xmath58 ) in the present system is consistent with a hamiltonian saddle node bifurcation reported in ref .
physically , the existence of the dynamical scaling law of the energy gap implies that the characteristic time scale diverges toward the critical velocity .
so far , we revealed the characteristic properties of the excitation spectra in a non - uniform finite size system . in order to investigate further the instability
, we calculate fluctuations from the wave functions of the excited states . in the case of the soliton nucleation , the authors of @xcite introduced a local density spectral function @xmath60 . in the bogoliubov approximation , @xmath60 is given by @xmath61 according to ref .
@xcite , the low - energy density fluctuations increase toward the critical velocity around the obstacle . in our system ,
it is difficult to calculate the spectral function because of the discretized eigenvalues in finite size systems . instead of calculating the spectral function , we plot the matrix element ( [ eq : matrix_element_density_fluctuation ] ) in fig . [
fig : density_flctuation_48_10_25 ] . in the small velocity regime ,
the low - energy density fluctuation is small because of the existence of the energy gap(see fig .
[ fig : density_flctuation_48_10_25](a ) ) .
the low - energy local density fluctuations grow with increasing the velocity shown in figs .
[ fig : density_flctuation_48_10_25](b ) and ( c ) . combining the scaling law of the energy gap and the behavior of the dynamical density fluctuations
, we can expect that the zero - energy eigenstate that contributes the density fluctuations exists at @xmath53 . in the previous work for the soliton nucleation in infinite 1d system@xcite ,
the zero - energy eigenstate related to the density fluctuations was obtained by analytical calculations and plays a crucial role of the breakdown of superfluidity .
therefore , in the case of the vortex nucleation , the density fluctuation is also expected to be a key to understand the breakdown of superfluidity .
besides the local density spectral function ( eq . ( [ eq : local_spectral_function_definition ] ) ) , the authors of @xcite also introduced the local phase spectral function . in the bogoliubov approximation , this is given by @xmath62 they showed that the local phase spectral function is not useful to detect the critical velocity .
the local phase fluctuation is nearly independent of the velocity@xcite because the zero mode related to the phase fluctuations always exists@xcite as a result of the @xmath63 symmetry breaking .
furthermore , there is no phase singularity in 1d system . in our system , however , the matrix element of the phase fluctuation ( [ eq : matrix_element_phase_fluctuation ] ) diverges locally because of the presence of the gvp .
although we can detect the existence of the gvp from the divergence of the matrix element of the local phase fluctuation , the critical velocity can not be detected from it because the gvp appears for @xmath64 .
finally , we make a remark on dynamics .
we have checked that the vortex pair is always nucleated above the critical velocity in the earlier stage of a real - time dynamics in our system , regardless of the presence / absence of gvp in the initial states . on the other hand ,
the low energy density fluctuations are enhanced regardless of existence / absence of the gvp .
therefore , the enhancement of low energy density fluctuations can be regarded as a precursor phenomenon of nucleations of quantized vortices . .fitting results .
[ cols="^,^,^,^,^,^,^,^,^",options="header " , ] [ tab : exponent ]
in summary , we studied the excitation spectra and the fluctuations of a bec in a 2d torus with a uniformly moving gaussian potential by solving the gp and the bogoliubov equation .
we found that the first excited energy(or the energy gap ) obeys a scaling law , which implies a diverging time scale near the critical velocity .
further we found enhancement of dynamical low - energy local density fluctuation near the critical velocity .
these two behaviors can be regarded as precursor effects of the vortex nucleation .
a future work is to elucidate the relation between enhancement of dynamical density fluctuations and the breakdown of superfluidity(or energy dissipation ) .
it is difficult to measure the local density spectral function experimentally .
we thus must find the relation between the local density spectral function and other physical quantities such as transport coefficients . in the case of the landau instability , the authors of ref .
@xcite showed that the energy dissipation and the density fluctuation are related through the drag force by using the perturbative approach .
although their approach is not applicable for the vortex and soliton the nucleation as it stands , their results give us a clue to the relation between the vortex nucleation and the energy dissipation .
we would like to thank s. watabe and i. danshita for fruitful discussions and d. yamamoto for useful comments .
m. k. acknowledges the support of a grant - in - aid for jsps fellows ( 239376 ) .
this work is supported by kakenhi ( 21540352 ) and ( 24543061 ) from jsps and ( 20029007 ) from mext in japan .
99 c. raman , _ et al_. , _ phys .
rev . lett_. * 83 * , 2502 ( 1999 ) .
r. onofrio , _ et al_. , _ phys .
lett_. * 85 * , 2228 ( 2000 ) .
s. inouye , _
et al_. , _ phys .
rev . lett_. * 87 * , 080402 ( 2001 ) .
t. w. neely , _
et al_. , _ phys .
rev . lett_. * 104 * , 160401 ( 2010 ) .
a. ramanathan , _
et al_. , _ phys .
rev . lett_. * 106 * , 130401 ( 2011 ) .
s. moulder , _
et al_. , _ phys .
a _ , * 86 * , 013629 ( 2012 ) .
k. c. wright , _ et al_. , _ phys .
rev . lett_.
* 110 * , 025302 ( 2013 ) .
l. d. landau , _ j. phys .
( ussr ) _ * 5 * , 71 ( 1941 ) .
e. p. gross , _
nuovo cimento _ * 20 * , 454 ( 1961 ) , l. p. pitaevskii , _ sov
jetp _ 13 , 451 ( 1961 ) .
t. frisch , y. pomeau , and s. rica , _ phys .
rev . lett_. * 69 * , 1644 ( 1992 ) .
v. hakim , _ phys .
e _ , * 55 * , 2835 ( 1997 ) .
b. jackson , j. f. mccann , and c. s. adams , _ phys .
rev . lett_. * 80 * , 3903 ( 1998 ) . c. josserand , y. pomeau , and s. rica , _ physica d _
, * 134 * , 111 ( 1999 ) .
t. winiecki , j. f. mccann , and c. s. adams , _ phys . rev . lett_.
* 82 * , 5186 ( 1999 ) .
b. jackson , j. f. mccann , and c. s. adams , _ phys .
a _ , * 61 * , 051603 ( 2000 ) . j. s. stieberger and w. zwerger , _ phys .
a _ , * 62 * , 061601(r ) ( 2000 ) . c. huepe and m - e .
brachet , _ physica d _ , * 140 * , 126 ( 2000 ) .
n. pavloff , _ phys .
a _ , * 66 * , 013610 ( 2002 ) .
c. t. pham and m. brachet , _ physica d _ , * 163 * , 127 ( 2002 ) .
a. aftalion , q. du , and y. pomeau , _ phys .
rev . lett_. * 91 * , 090497 ( 2003 ) .
g. e. astrakharchik and l. p. pitaevskii , _ phys . rev .
a _ , * 70 * , 013608 ( 2004 ) . g. a. el , a. gammal , and a. m. kamchatnov , _ phys . rev .
lett_. * 97 * , 180405 ( 2006 ) .
f. piazza , l. a. collins , and a. smerzi , _ phys .
a _ , * 80 * , 021601(r ) ( 2009 ) . k. sasaki , n. suzuki , and h. saito , _ phys . rev . lett_. * 104 * , 150404 ( 2010 ) . y. kato and s. watabe , _ phys
. rev . lett_. * 105 * , 035302 ( 2010 ) .
k. fujimoto and m. tsubota , _ phys .
a _ , * 83 * , 053609 ( 2011 ) .
t. aioi , t. kadokura , t. kishimoto , and h. saito , _ phys .
x _ , * 1 * , 021003 ( 2011 ) .
r. dubessy , t. liennard , p. pedri , and h. perrin , _ phys . rev .
a _ , * 86 * , 011602 ( 2012 ) . f. piazza , l. a. collins , and a. smerzi , _ j. phys .
b : at . mol .
opt . phys_. * 46 * , 095302 ( 2013 ) .
n. n. bogoliubov , _ j. phys
* 11 * , 23 ( 1947 ) .
m. kunimi and y. kato , _ phys .
b _ , * 86 * , 060510(r ) ( 2012 ) . for example , we use @xmath65 and @xmath66 basis for @xmath67 and @xmath68 and @xmath69 basis for @xmath48 and @xmath70 .
k. kasamatsu , m. tsubota , and m. ueda , _ phys . rev .
a _ , * 67 * , 033610 ( 2003 ) .
j. guckenheimer and p. holmes , nonlinear , _ oscillations , dynamical systems and bifurcations of vector fields _ ( springer , berlin , 1983 ) .
d. takahashi and y. kato , _ j. phys .
jpn . _ * 78 * , 023001 ( 2009 ) .
s. watabe and y. kato , arxiv:1305.6984 .
s. watabe(private communications ) . @xmath71 and @xmath72 are always solutions of the bogoliubov equation with a zero eigenvalue . | we investigate excitations and fluctuations of bose - einstein condensates in a two - dimensional torus with a uniformly moving gaussian potential by solving the gross - pitaevskii equation and the bogoliubov equation .
the energy gap @xmath0 between the current - flowing metastable state ( that reduces to the ground state for sufficiently slowly - moving potential ) and the first excited state vanishes when the moving velocity @xmath1 of the potential approaches a critical velocity @xmath2 .
we find a scaling law @xmath3 , which implies that a characteristic time scale diverges toward the critical velocity . near the critical velocity
, we show that low - energy local density fluctuations are enhanced .
these behaviors can be regarded as precursor phenomena of the vortex nucleation .
pacs numbers : 03.75.kk , 67.85.-d , 67.25.dg , 67.85.de |
let @xmath2 be a compact @xmath12-dimensional manifold , @xmath13 , with a fixed conformal class @xmath0 $ ] and a fixed spin structure @xmath1 .
let @xmath14 be conformal to @xmath15 , i.e. @xmath16 $ ] . the classical dirac operator @xmath17 on @xmath6 has discrete real spectrum with finite multiplicities .
the eigenvalues tend to @xmath10 and @xmath11 .
the dimension of the kernel of @xmath17 is a spin - conformal invariant , i.e. @xmath18 for all @xmath3 $ ] .
we denote the first ( = smallest ) positive eigenvalue of @xmath17 by @xmath19 , and the first ( = largest ) negative eigenvalue by @xmath20 .
finding bounds for this eigenvalue has attracted much interest during the last decades . among several estimates @xcite in terms of a positive scalar curvature bound ,
let us mention the following estimate due to friedrich @xcite . if the minimum of the scalar curvature of @xmath21 is at least @xmath22 , then @xmath23 .
an improvement of this inequality that is important in conformal geometry was derived by hijazi @xcite in terms of the first eigenvalue @xmath24 of the conformal laplacian @xmath25 , namely @xmath26 if @xmath27 . on the other hand ,
if the _ yamabe invariant _ @xmath28)=\inf_{g\in [ g_0]}{\int_m \scal_g\dvol_g\over \vol(m , g)^{(n-2)/n}}\in ( -\infty , n(n-1)]\ ] ] is non - negative , one can use the conformal transformation formula for the scalar curvature ( * ? ? ?
* theorem 1.159 ) , and one obtains that @xmath29).\ ] ] together with , we get @xmath30)}.\ ] ] that this inequality even holds in dimension @xmath31 , was proved by c. br @xcite . in this special case ,
the gauss - bonnet theorem tells us that @xmath32 , hence we obtain @xmath33 if @xmath2 is diffeomorphic to @xmath34 , but we do not get a bound for other surfaces . equality in and hold for the round spheres . hence , we have obtained an explicit positive lower bound @xmath35 that is uniform on the conformal class @xmath0 $ ] . from now on
, we will restrict to the first _ positive _ eigenvalue , in order to simplify the presentation ( see the remark below ) .
the _ existence _ of a positive lower bound for @xmath36 had already been derived before by j. lott .
his estimate does not require that @xmath37)$ ] is positive , but needs the weaker assumption that the dirac operator is invertible , i.e. has @xmath38-dimensional kernel .
unfortunately , his bound is not explicit , and for most spin - conformal manifold the determination of the value of the associated infimum @xmath39,\si ) : = \inf_{g\in [ g_0 ] } \la_1^+(g)\vol(m , g)^{1/n}\ ] ] is still a challenging open problem .
a key idea of j. lott s article is to derive a lower bound for the first dirac eigenvalue in terms of the supremum of a conformally invariant functional , a version of this functional will be explained in section [ sec.varprin ] . in our article @xcite , we started to study this functional in more detail .
in particular , we showed that lott s result extends to the case that the dirac operator has non - trivial kernel , namely @xmath40,\si)>0\ ] ] for any compact riemannian spin manifold @xmath41,\si)$ ] .
furthermore , it was shown that @xmath40,\si ) \leq \lammin(\ms^n),\ ] ] where @xmath42 denotes the sphere with its standard riemannian metric of constant sectional curvature @xmath43 .
this bound has been proven in @xcite unless @xmath44 and @xmath31 .
the remaining case @xmath44 and @xmath31 was given in @xcite . in the present article
, we discuss whether the associated infimum in is attained . for having a well - behaved minimization problem ,
it is reasonable to replace the conformal class @xmath0 $ ] in by its @xmath45-completion @xmath46}$ ] .
here , by definition a metric @xmath47 is in @xmath46}$ ] if @xmath48 is a real - valued @xmath45 function .
the first positive eigenvalue of the dirac operator extends naturally to this completion ( see section [ sec.spd ] ) .
we show that if we have the strict inequality @xmath49 then the infimum is obtained by a generalized metric @xmath50}$ ] .
this minimizer has the form @xmath51 where @xmath52 is a spinor of regularity @xmath53 .
the set @xmath54 is called the _ nodal set of @xmath52 _ or the _ set of degeneration of @xmath14_. [ theo.main ] let @xmath2 be a compact manifold of dimension @xmath13 with a fixed conformal class @xmath0 $ ] and a spin structure @xmath1 .
assume that holds .
let @xmath55 if @xmath56 , and let @xmath57 if @xmath58 . 1 .
then there is a spinor field @xmath59 on @xmath60 such that @xmath61 2 .
there is a @xmath62 such that @xmath63 the metric has the form @xmath64 where @xmath52 is a spinor as in ( a ) .
3 . if @xmath65 , then the metric @xmath14 is smooth and the set of degeneration of @xmath14 , denoted @xmath66 , is finite .
furthermore @xmath67 in particular , if @xmath2 is diffeomorphic to a @xmath68-torus , then the set of degeneration @xmath66 is empty .
inequality has been proven for several classes of manifolds .
it is known that it holds for non - conformally - flat manifolds of dimension @xmath69 @xcite .
inequality has also been shown if @xmath2 is conformally flat , if @xmath5 is invertible , and if the mass endomorphism is not identically zero ( after a possible change of orientation if @xmath70 ) @xcite .
the mass endomorphism is a section of @xmath71 defined as the zero order term of the development of the green function for the dirac operator at the diagonal with respect to a conformal coordinate map .
furthermore , is known for many riemann surfaces ( i.e. @xmath31 ) , e.g. all rectangular tori have a spin structure such that holds .
the euler - lagrange equation of the above minimization problem has a particularly nice interpretation in dimension @xmath31 .
locally the equation can be translated into a conformal constant mean curvature immersion into @xmath72 .
this translation is a spinorial extension of the weierstrass representation ( see section [ sec.weierrep ] ) . by pasting together these local surfaces
one obtains a `` periodic branched conformal cmc immersion '' .
more exactly , let @xmath21 be a compact riemann surface together with its universal covering @xmath73
periodic branched conformal cmc immersion based on @xmath21 _ is by definition a smooth map @xmath74 together with finitely many points @xmath75 , the so - called _ branching points _ , such that the following properties hold : 1 . _
periodicity : _ there is a homomorphism @xmath76 , the _ periodicity map _ , such that for any @xmath77 , and @xmath78 one has @xmath79 here @xmath80 denotes the action of @xmath81 on @xmath82 via deck transformation .
conformality : _ the restriction of @xmath83 to @xmath84 is a conformal immersion .
branching points : _ we have @xmath85 for any @xmath86 .
the order of the first non - vanishing term in the taylor development of @xmath87 in @xmath88 is called the _ branching index _ of @xmath83 at @xmath88 .
cmc : _ the image @xmath89 is an immersed surface with constant mean curvature .
the principle yields the existence of many periodic branched conformal cmc immersions .
the set of periodic branched conformal cmc immersions , @xmath90 , is essentially ( see section [ sec.app.cmc ] for a precise statement ) in bijection with the stationary points of the variational problem associated to our minimization problem .
in particular , all minimizers of give rise to periodic branched conformal cmc immersions .
we obtain assume that the riemann spin surface @xmath6 carries a metric @xmath14 such that the first positive eigenvalue of the dirac operator is smaller than @xmath91 .
then there is a periodic branched conformal cmc immersion @xmath83 based on @xmath21 . the regular homotopy class of @xmath83 is determined by the spin - structure @xmath1 .
the indices of all branching points are even , and the sum of these indices is smaller than @xmath92 . in particular , if @xmath2 is a torus , there are no branching points . some examples of branched conformal cmc immersions that may arise by this principle are given in section [ sec.app.cmc ] .
the problem that we are discussing in this article has many relations to the yamabe problem . for a given compact conformal manifold @xmath41)$ ]
the yamabe problem is the problem to find a metric of constant scalar curvature @xmath14 in @xmath0 $ ] .
the problem has been affirmatively solved by trudinger , aubin , schoen and yau , see @xcite for a good overview . at first , due to the conformally invariant character of our problem and the yamabe problem , certain bounded , but non - compact sobolev embeddings play an important role . in the solution of both problems it is useful to break the conformal invariance by perturbing a parameter . for this perturbed parameter
the embeddings are compact , and standard methods yield the existence of a minimizer .
it then has to be checked whether the perturbed minimizers converge to a minimizer of the unperturbed problem , or whether they concentrate in some points . in the yamabe problem
the minimizers converge if @xmath93)<y(\ms^n),\ ] ] holds whereas in our problem , the minimizers converge if holds .
secondly , implies that any manifold satisfying satisfies as well .
hence , proving for a given spin - conformal manifold @xmath41,\si)$ ] solves the yamabe problem on this manifold .
the third relationship is that some proofs of inequality in special cases resemble to proofs of inequality , e.g. @xcite resembles to @xcite .
the structure of the article is as follows . in section [ sec.varprin ]
we reformulate our problem as a variational problem .
we will see that it is natural to admit in the infimum certain singular metrics namely `` metrics '' conformal to @xmath15 whose conformal factor might vanish somewhere . such metrics called _ generalized metrics _
are the subject of section [ sec.spd ] . in the following section
, we discuss the round sphere .
this example is helpful to obtain a deeper understanding for the analytical difficulties .
however , it can be skipped if the reader is only interested in the main results of the article .
section [ sec.reg.theo ] is devoted to certain regularity issues that will become important in section [ sec.solution ] , where the variational principle is finally solved .
we then show in section [ sec.proofmain ] how this implies the main theorem .
the singularities of the minimizers are discussed in section [ sec.degen ] . in section [ sec.weierrep ]
we recall the spinorial weierstrass representation .
section [ sec.app.cmc ] uses the spinorial weierstrass representation to derive the application to constant mean curvature surfaces .
several examples are included . in appendix [ sec.regularity ]
we summarize ( without proofs ) some analytical tools .
this appendix shall also serve as a reference for fixing the notations for sobolev spaces and hlder spaces . in appendix [ sec.scha.power ]
we proof a proposition about hlder spaces , as we could not find a proof in the literature .
the infimum is also attained in if we replace @xmath94 by @xmath95 . in the case
@xmath96 this is obvious : there exists an automorphism of the spinor bundle anticommuting with @xmath5 , hence @xmath97 . in the case
@xmath98 , the proof for @xmath99 is up to the obvious sign changes completely identical . in almost all references cited in the introduction ,
all statements for @xmath94 also hold for @xmath95 and vice versa .
the only exception is @xcite .
* acknowledgements .
* i am very much indebted to christian br for various support .
thank you also to emmanuel humbert for his continuing interest in conformal spin geometry and his deep analytical ideas .
many electronic and personal discussions with robert kusner and karsten grosse - brauckmann were also very helpful . i also want to thank b. booss - bavnbek , oussama hijazi , sergiu moroianu , victor nistor , reiner schtzle and guofang wang for several stimulating discussions related to this article .
the goal of this section is to reformulate the problem of minimizing the first dirac eigenvalue in a conformal class as a variational problem .
the choice of a good functional is not very easy , as we would like to find a functional which is both bounded and conformally invariant .
the dirac operator @xmath5 has a simple behavior under conformal change of the metric , its square @xmath100 transforms in a more complicated way .
hence , it is desirable to use a functional that contains only terms in @xmath101 and @xmath52 and no term depending on @xmath102 .
the @xmath103-functional defined in , @xmath104 , is such a conformally invariant , bounded functional , and it will be turn out that by working with this functional we obtain the desired results . in case , that the reader of our article is already familiar with the analytical problems of the yamabe problem , he might find it enlightening to compare this problem to our problem .
many techniques from the resolution of the yamabe problem can be carried over to our setting . however , several arguments from the resolution of the yamabe problem fail .
two main problems arise : on the one hand the spectrum of @xmath5 is neither bounded from below nor from above , on the other hand , we are working with sections of a vector bundle instead of functions , hence the standard maximum principle is not available .
such arguments will have to be replaced by other approaches .
at first , we recall some basic definitions and facts from spin geometry . for details
we refer to textbooks as for example @xcite or to the beautifully written self - contained introduction @xcite .
let @xmath2 be a compact manifold equipped with a riemannian metric @xmath15 and a spin structure @xmath1 .
assume that @xmath14 is a metric conformal to @xmath15 .
one associates to @xmath6 a natural complex vector bundle over @xmath2 called the spinor bundle @xmath105 .
sections of this bundle are called _ spinor fields _ or simply _ spinors_. the bundle carries a hermitian metric , a metric connection , and a clifford multiplication .
these additional structures are used to define the dirac operator @xmath106 , which is a first order elliptic differential operator .
the dirac operator @xmath17 is essentially self - adjoint , and hence it has a self - adjoint extension . as a consequence of standard elliptic theory ,
the spectrum is real and discrete , and all multiplicities are finite . in the following ,
the spin structure @xmath1 will be fixed ( and often suppressed in the notation ) , whereas the metric @xmath14 varies inside the conformal class @xmath0 $ ] .
some objects will be noted with an index @xmath14 , which means that they are defined with respect to @xmath14 whereas the same object without the index @xmath14 indicates that it is defined with respect to the fixed background metric @xmath15 ( with some exceptions that are apparent from the context ) .
for example , @xmath107 is the volume element associated to @xmath14 and @xmath108 is the one associated to @xmath15 .
we will frequently use several norms , as for example @xmath109 and @xmath110 that are defined with respect to @xmath15 unless otherwise stated .
we summarize their definition and some important analytical tools in appendix [ sec.regularity ] .
the spectrum of the dirac operator @xmath17 will be denoted as @xmath111 where each eigenvalue appears with its multiplicity .
note that @xmath38 may be an eigenvalue or not . in the former case @xmath17 is not invertible , otherwise it is .
elliptic theory shows @xmath112 .
the following transformation formula will be of central importance .
to our knowledge , the formula was found by hitchin @xcite .
another reference written up in a more self - contained manner is @xcite .
[ prop.conf.change ] let @xmath113 , @xmath114 smooth and positive .
there is an isomorphism of vector bundles @xmath115 which is a fiberwise isometry such that @xmath116 it is convenient to define @xmath117 we will use this isomorphism @xmath118 to identify spinors associated to conformal metrics . with this identification
the conformal transformation formula reads as @xmath119 and @xmath120 .
in particular , with this identification the kernels of @xmath17 and @xmath121 coincide .
it is easy to verify that with this identification , the expression @xmath122 is conformally invariant .
the @xmath123-norm @xmath124 is conformally invariant if and only if @xmath125 has the value @xmath126 .
similarly , @xmath127 is conformally invariant if and only if @xmath88 has the value @xmath128 .
note that @xmath129 . for any @xmath130
$ ] and for any @xmath131-spinor @xmath52 that is not in the kernel of the dirac operator we define @xmath132 and @xmath133 , @xmath134 .
the well - definedness of @xmath135 and some basic properties are given by the following lemma .
[ lem.f.elem ] let @xmath136 .
let @xmath52 be a spinor field of regularity @xmath131 , @xmath137 .
then , @xmath138 is well - defined and real .
furthermore @xmath139 is frchet differentiable with derivation given by @xmath140 where @xmath141 .
the supremum @xmath142 is positive and finite . from the above considerations it is evident that the functional @xmath103 is conformally invariant if and only if @xmath143 .
let @xmath52 be a spinor field of regularity @xmath131 , @xmath137 .
take @xmath125 with @xmath144 , @xmath145 . because of @xmath146 , we see with hlder s inequality that @xmath147 is integrable . thus , the numerator of @xmath103 is well - defined , and hence @xmath103 is well - defined .
the self - adjointness of @xmath148 implies that @xmath138 is real .
moreover , because @xmath149 is bounded we see that @xmath150 is bounded from above by a multiple of @xmath151 . using theorem [ theo.global.lp ]
we obtain @xmath152 , and we see that @xmath103 is bounded on @xmath153 . an arbitrary spinor field @xmath154 is written as the sum of @xmath155 and a non - zero @xmath156 . as @xmath157 , we see that @xmath142 is finite . because of @xmath154 , we have @xmath158 . hence the right hand side
of defines a continuous functional on @xmath123 and thus on @xmath131 .
we denote the functional by @xmath159 .
similarly , one sees that @xmath160 hence @xmath103 is frchet differential with derivative @xmath159 . if @xmath52 is an eigenspinor to a positive eigenvalue , then @xmath161 .
[ prop.func ] the function @xmath162 is continuous from the right , and @xmath163 furthermore , if @xmath164 , then @xmath165 is non - increasing in @xmath88 .
we assume @xmath164 , the statements for @xmath166 then follow by rescaling . that @xmath165 is non - increasing follows easily from the hlder inequality .
in order to show the continuity from the right , let @xmath167 be given . we take a smooth spinor field @xmath52 such that @xmath168 . observe that @xmath169 the function @xmath170 is continuous , hence if @xmath171 is sufficiently close to @xmath88 , then @xmath172 because @xmath173 is non - increasing
, the continuity from the right follows .
the formula @xmath174 follows directly if one writes @xmath52 as a sum of eigenspinors and evaluates @xmath175 .
( -1,-)(10.5,6 ) ( 0,0)(9.5,5 ) ( 9.9,0)@xmath88 ( 0,5.4)@xmath142 ( 2.4,-)(2.4 , ) ( 3.5,-)(3.5 , ) ( 2.4,-)@xmath176 ( 3.5,-)@xmath68 ( -,2.8)(,2.8 ) ( -,3.7)(,3.7 ) ( -,2.8)@xmath177 ( -,3.7)@xmath178 ( 2.4,0)(2.4,3.7 ) ( 3.5,0)(3.5,2.8 ) ( 0,3.7)(2.4,3.7 ) ( 0,2.8)(3.5,2.8 ) ( 2.2,3.74)(2.4,3.7)(3.5,2.8)(4,2.0)(5.1,1.6)(7.2,1.4)(7.5,1.2 ) in proposition [ prop.sol.subcrit ] we will see that the supremum defining @xmath165 is attained for @xmath179 by a @xmath180-spinor .
this implies that the function @xmath181,\quad q\mapsto \mu_q^g\ ] ] is also continuous from the left .
[ prop.lamu ] @xmath182 we have already seen that @xmath183 is conformally invariant , for @xmath184 $ ] @xmath185 the previous proposition states that @xmath186 if @xmath187 , and it follows @xmath188 it remains to show the inverse inequality which amounts to showing @xmath189 } \la_1^+(g)^{-1}\vol(m , g)^{-1/n}\geq \sup \cf_\qd.\ ] ] for any @xmath190 we take a smooth spinor @xmath191 with @xmath192 and @xmath193 .
after a small perturbation of @xmath191 we can assume that @xmath194 has no zeros .
we set @xmath195 then we have @xmath196 , and @xmath197 has constant length @xmath43 .
hence , @xmath198 this implies the proposition .
now , as we have understood the relationship between the supremum of @xmath199 and the infimum of @xmath200 , we want to establish a relationship of the maximizers of @xmath199 and the minimizers of @xmath200 .
this will require some knowledge about the euler - lagrange - equation .
one easily sees that @xmath201 for any @xmath202 and @xmath203 .
hence maximizers of @xmath103 appear in families and it will be convenient to choose a good representative for each family of maximizers .
[ lem.euler.lagrange ] let @xmath130 $ ] , and choose @xmath125 with @xmath204 .
suppose that @xmath103 has a maximizing spinor @xmath205 .
then there is a maximizing spinor @xmath206 , @xmath207 , such that @xmath208 we normalize @xmath209 such that @xmath210 , thus @xmath211 . as @xmath209 is a maximizer ,
@xmath212 is identically zero .
because of we see that @xmath213 is a weak solution of @xmath214 , hence it is smooth and in the kernel of @xmath148
. then , @xmath215 satisfies @xmath216 and hence @xmath217 taking norms we obtain @xmath218 @xmath219 hence @xmath220 satisfies .
[ theo.var.corresp ] + 1 .
let @xmath221 be a maximizing spinor of @xmath199 , and suppose that @xmath221 is smooth and that @xmath222 vanishes nowhere .
then @xmath223 is a smooth metric minimizing @xmath200 in the conformal class @xmath0 $ ] .
2 . let @xmath3 $ ] be a ( smooth ) metric minimizing @xmath200 , and let @xmath221 be an eigenspinor of @xmath17 to the eigenvalue @xmath19 , then the length of @xmath224 is constant and @xmath221 maximizes @xmath199 .
a. let @xmath221 be a maximizing spinor .
according to the previous lemma , there is an @xmath225 and a @xmath226 such that @xmath227 satisfies qith @xmath143 and @xmath228 . in particular
, @xmath229 vanishes nowhere .
the metric @xmath230 satisfies @xmath231 and @xmath232 .
then @xmath233 as @xmath234 we see that @xmath235 , hence @xmath236 minimizes @xmath200 . by a simple rescaling argument
one sees that @xmath237 minimizes @xmath200 as well .
b. by rescaling we can assume that @xmath164 . unless otherwise indicated all volume measures , norms , scalar products and dirac operators in this proof are with respect to @xmath14 . in order to show that @xmath224 is constant , we define @xmath238
one calculates @xmath239 .
all metrics @xmath240 have volume @xmath43 . hence , as the infimum of @xmath200 is attained in @xmath14 , we have @xmath241 , and hence @xmath242 for @xmath243 equality is attained . hence @xmath244 using @xmath245 and deriving with respect to @xmath246 yields @xmath247 the right hand side is equal to @xmath248 , hence @xmath224 is constant .
+ this implies that @xmath249 is independent of @xmath88 , thus @xmath250 and hence @xmath199 attains it supremum in @xmath221 .
later on , we will see that maximizers of @xmath199 that vanish nowhere are always smooth .
unfortunately , we can not exclude that maximizers of @xmath103 vanish somewhere .
maximizers with zeros correspond to metrics with certain singularities , more precisely to metrics whose conformal factor has zeros .
these metrics are the main object of this section .
we will summarize some facts about the size of the zero set of maximizers in section [ sec.degen ] . roughly speaking ,
generalized metrics are metrics of the form @xmath47 where @xmath251 . however ,
for technical and formal reasons it is better to use the following definition .
let @xmath15 be a smooth metric on a compact manifold @xmath2 .
generalized metric _ is a tuple @xmath252 where @xmath3 $ ] and @xmath253 , @xmath254 .
if @xmath255 is smooth , we identify @xmath256 with @xmath257 .
furthermore we identify @xmath258 $ ] with @xmath259 .
generalized metrics having a representative of the form @xmath260 are called _ conformal _ to @xmath15 , and those having a representative @xmath259 , @xmath3 $ ] are called _ regular metrics _ they correspond to metrics in the ordinary sense .
the set of all generalized metrics conformal to @xmath15 is called the _ @xmath45-completion
_ @xmath261}$ ] of the conformal class @xmath0 $ ] .
the _ volume of @xmath262}$ ] _ is defined as @xmath263 . for a generalized metric
we say that @xmath264 is the _ set of degeneration_. the reader should pay some attention to the following technical difficulty : if @xmath48 vanishes on an open set , then the @xmath265-tensor @xmath47 does not determine ( the equivalence class ) @xmath260 .
however , despite of this remark and slightly abusing the notation , we will write @xmath47 instead of @xmath260 .
formally @xmath47 is a generalized metric in the above sense , not an @xmath265-tensor .
let us now assume that @xmath2 carries a fixed spin structure @xmath1 .
for any generalized metric @xmath113 we want to define a spinor bundle and a dirac operator on @xmath6 in such a way that the results of the previous section carry over to this generalization .
in particular , the functional @xmath103 has to be defined and has to be conformally invariant for @xmath143 . as a vector bundle
the spinor bundle @xmath266 is defined to be @xmath267 , and due to our identification of spinors for different metrics in a fixed conformal class this construction does only depend on the conformal class @xmath0 $ ] and not on the metric @xmath15 itself . for any @xmath268 and
any spinor @xmath52 in the fiber of @xmath266 over @xmath269 we define the pointwise norm @xmath270 again , this norm depends only on the conformal class . for smooth sections @xmath52 of @xmath266 such that @xmath271 almost everywhere we define the @xmath272-norm as @xmath273 where we used the conventions @xmath274 for @xmath275 and @xmath276 .
the sobolev space @xmath272 is the associated completion . there is a natural inclusion @xmath277 cauchy sequences with respect to the norm @xmath272 are also cauchy sequences with respect to @xmath278 .
thus we obtain a bounded map @xmath279 . in order to prove injectivity of this map
one shows that if @xmath280 is a cauchy sequence in @xmath272 converging to @xmath38 with respect to @xmath278 , then it converges to @xmath38 with respect to the @xmath272 as well .
we define the dirac operator @xmath281 , @xmath282 .
the spinor @xmath283 is well - defined almost everywhere , as @xmath284 has measure zero .
it is easy to verify that all these definitions only depend on the conformal class of @xmath15 and not on @xmath15 itself .
furthermore , we can reformulate the definition of the above sobolev space as @xmath285 we now extend the definitions @xmath286 and @xmath19 to the @xmath45-completion of the conformal class . for @xmath287}$ ] and any @xmath288 we define @xmath289 because of @xmath290 the functional is well - defined and bounded on @xmath291 .
it is also not hard to see that the supremum is attained .
in fact , let @xmath292 be a sequence of spinors in @xmath291 with @xmath293 , normalized such that @xmath294 .
then a subsequence @xmath295 converges weakly in @xmath272 , weakly in @xmath278 and strongly in @xmath296 towards a @xmath297 .
hence , @xmath298 .
thus , the supremum is attained in @xmath299 . in analogy
to lemma [ lem.f.elem ] one gets for any smooth test spinor @xmath221 @xmath300 which implies @xmath301 and finally for @xmath302 @xmath303 the _ first positive dirac eigenvalue _ of @xmath6 , @xmath287}$ ] is @xmath304 a non - trivial spinor with @xmath305 is an _ eigenspinor to the eigenvalue @xmath19_. as we have already seen , this definition coincides with the definition of the first positive dirac eigenvalue and a corresponding eigenspinor if @xmath14 is regular . most of the statements of the previous section still hold in a modified version for generalized metrics , the proofs are nearly identical .
for example one can extend proposition [ prop.lamu ] to [ prop.bar.nonbar.gen ] @xmath306)}\la^+_1(g)\vol(m , g)^{1/n } = \inf_{g\in \ol{[g_0]}}\la^+_1(g)\vol(m , g)^{1/n } = \mu_{\qd}(m , g,\si)^{-1}.\ ] ] the equation @xmath307}\la^+_1(g)\vol(m , g)^{1/n}=\mu_{\qd}(m , g,\si)^{-1}$ ] is exactly the statement of proposition [ prop.lamu ] and the equation @xmath308}}\la^+_1(g)\vol(m , g)^{1/n } = \mu_{\qd}(m , g,\si)^{-1}$ ] can be proven with exactly the same proof . similarly , we obtain an analogue of theorem [ theo.var.corresp ] .
[ theo.var.corresp.gen ] let @xmath309 be a compact riemannian spin manifold .
let @xmath221 be a maximizing spinor of @xmath199 ( with regularity @xmath310 ) .
then @xmath223 is a generalized metric minimizing @xmath200 in @xmath261}$ ] , the @xmath45-completion of the conformal class @xmath0 $ ] .
2 . let @xmath287}$ ] be a generalized metric minimizing @xmath200 , and let @xmath221 be an eigenspinor of @xmath17 to the eigenvalue @xmath19 , then the length of @xmath224 is constant on @xmath311 and @xmath221 maximizes @xmath199 .
the proof of this theorem is essentially the same as the proof of theorem [ theo.var.corresp ] .
later , we will see that any maximizing spinor has regularity @xmath180 , hence @xmath312 is always a generalized metric .
the sphere @xmath313 with the round sphere is an important example . on the one hand the invariant @xmath314 of the round sphere is contained in many equations and inequalities . on the other hand
the round sphere has a large conformal group , and hence studying the minimizers on the sphere helps to understand the analytical difficulties .
in particular , we will see why the conclusion in theorem [ theo.anydim ] does not hold if @xmath315 .
( recall @xmath316 . )
the invariant @xmath317 is not hard to calculate .
recall that the yamabe invariant @xmath318 is given by @xmath319}{\int \scal_g\,\dvol_g\over \vol(m , g)^{(n-2)/n}}.\ ] ] it is attained for @xmath320 , and hence @xmath321 , where @xmath322 . the hijazi inequality ( [ ineq.hij],[ineq.hij.der ] )
tells us , that @xmath323 recall that the sphere of constant sectional curvature @xmath43 carries a killing spinor @xmath221 to the constant @xmath324 , i.e. it satisfies @xmath325 note that this condition implies that the length of @xmath221 is constant .
because of @xmath326 we obtain @xmath327 and then @xmath328 the above killing spinor satisfies @xmath329 hence @xmath330 attains its supremum in such killing spinors .
let @xmath331 be an orientation preserving conformal diffeomorphism .
then the pullback of any spinor @xmath52 with respect to @xmath332 is a section of @xmath333 , and as before we identify @xmath334 by using the map @xmath335 described after proposition [ prop.conf.change ] .
if @xmath52 is a solution of , then this pullback , denoted by @xmath336 , is a solution of this equation as well . furthermore if @xmath52 maximizes @xmath330 , then @xmath336 is also a maximizer . as a consequence ,
all conformal images of killing spinors to the constant @xmath324 are maximizers of @xmath330 .
the following proposition shows that there are no other maximizers on @xmath337 .
if @xmath221 is a spinor of regularity @xmath53 that attains the supremum of @xmath330 , then there is a killing spinor @xmath52 to the killing constant @xmath324 and an orientation preserving conformal diffeomorphism @xmath331 with @xmath338 . later on
we will see that any maximizer of regularity @xmath131 with @xmath179 is even @xmath53 .
hence the statement also holds under this weaker assumption .
[ lem.scal.bound ] let @xmath6 be an arbitrary riemannian spin manifold ( not necessarily complete or compact ) .
assume that there is a spinor @xmath221 of constant length @xmath43 and with @xmath339 .
then @xmath340 where @xmath341 denotes the _ friedrich connection _ on spinors .
the friedrich connection is a metric connection , hence @xmath342 the twisted version of the schrdinger - lichnerowicz formula yields @xmath343 hence @xmath344 we obtain @xmath345 we have to show that the supremum is not attained by any other spinor . for proving this ,
assume that @xmath221 is a maximizer , @xmath346 .
as the dirac operator on @xmath337 has kernel @xmath347 , the spinor satisfies the euler - lagrange equation . on the open subset @xmath348
we define the metric @xmath349 in this metric transforms into a solution of @xmath350 on the one hand , one calculates @xmath351 and integration yields @xmath352 where the last term arises by partial integration . in order to make this step precise
one has to exhaust @xmath353 by smooth manifolds with boundary and partially integrate over these exhausting manifolds .
the boundary terms vanish in the limit , as @xmath354 on the boundaries and @xmath355 is bounded .
it is a standard fact from the resolution of the yamabe problem ( see e.g.@xcite ) that any @xmath356 function @xmath48 satisfies @xmath357 setting @xmath358 , we obtain @xmath359 , and hence the right side of is bounded from below by @xmath360 .
one the other hand , the previous lemma provides @xmath361 and as @xmath362 , we see that we must have equality in all inequalities involved , in particular in .
an application of the maximum principle @xcite yields that @xmath48 does not vanish , and hence @xmath221 has no zeros .
furthermore , @xmath363 is a metric of constant scalar curvature conformal to @xmath337 , and such a metric is necessarily of the form @xmath364 for an orientation preserving conformal diffeomorphism @xmath331 .
with respect to @xmath363 we obtain @xmath365 , hence @xmath221 is a killing spinor on @xmath366 .
this implies that @xmath367 is a killing spinor on @xmath337
. there are solutions to that do not maximize the functional .
an easy construction of such a solution is as follows .
the map @xmath368 , @xmath369 , @xmath370 is conformal with branching points @xmath38 and @xmath371 .
if @xmath52 is a solution of @xmath372 , then the pullback @xmath373 is a solution of @xmath374 on @xmath375 . here
@xmath221 is a section of the pull - backed spinor bundle , which is defined using the pull - backed spin structure .
the pull - backed spin structure on @xmath375 coincides with the standard spin structure iff @xmath376 is odd .
if @xmath376 is odd , one can show that the extension of @xmath221 by setting @xmath377 and @xmath378 is a solution of @xmath374 on @xmath379 .
however , @xmath380 .
this implies @xmath381 .
hence , if @xmath52 is a killing spinor , and @xmath382 , @xmath376 odd , then @xmath221 is a non - maximizing solution .
in this section we want to study solutions of the euler - lagrange equations . the first subsection `` removal of singularities '' will be used in the following section to extend a solution of @xmath383 to a solution on @xmath337 . the second subsection states that solutions of the euler - lagrange equations are @xmath384 .
[ theo.sing.rem ] let @xmath385 .
let @xmath386 be a ( not necessarily complete ) riemannian manifold equipped with a spin structure , let @xmath387 .
assume that @xmath388 satisfies weakly on @xmath389 the equation @xmath390 then this equation even holds weakly on @xmath391 . in particular , the distribution @xmath101 does not have singular support in @xmath269 and is contained in @xmath392 .
let @xmath221 be a smooth spinor compactly supported in @xmath391 .
we have to show @xmath393 for any small @xmath190 we choose a smooth cut - off function @xmath394 $ ] with @xmath395 on @xmath396 , with @xmath397 and with support in @xmath398 .
we rewrite the left hand side as @xmath399 as @xmath52 is a weak solution of on @xmath389 , the first term equals to @xmath400 and for @xmath401 it tends to the right hand side of .
let @xmath88 be related to @xmath125 via @xmath402 .
the absolute value of the second term is bounded by @xmath403 which tends to @xmath38 for @xmath401 . finally , the absolute value of the third term is bounded by @xmath404 our condition @xmath405 yields @xmath406 , and hence the third term also tends to @xmath38 for @xmath401 .
[ theo.reg ] suppose that @xmath154 , @xmath130 $ ] is a solution of equations .
suppose that there is an @xmath407 such that @xmath408 .
then @xmath52 is @xmath409 for any @xmath410 .
+ furthermore , we obtain a uniform bound of the @xmath409-norm in the following sense .
let us choose @xmath411 such that @xmath412 and @xmath413 .
then for any @xmath57 there is a constant @xmath414 depending only on @xmath6 , @xmath125 , @xmath415 , @xmath416 , @xmath376 and @xmath417 with @xmath418 the following example shows that the theorem can not hold without the @xmath419-bound .
let @xmath420 and @xmath228 .
let @xmath221 be a killing spinor to the killing constant @xmath324 .
suppose @xmath421 .
let @xmath331 be a mbius transformation , such that the differential in @xmath422 satisfies @xmath423 .
then @xmath424 is a family of solutions of , maximizing @xmath199 .
however , for any @xmath407 one can show that @xmath425 for @xmath426 .
hence , the @xmath419-bound is necessary for the theorem to hold .
the theorem will be applied in several versions . at first
, we will apply it when @xmath427 . in this case
the sobolev embedding already provides the required @xmath419-bound on @xmath52 .
however , the @xmath415 given by the sobolev embedding depends on @xmath125 .
it will be of central importance to obtain a bound that is uniform for @xmath428 . after having proved theorem [ theo.anydim ] the uniformity statement in the above theorem
will be used to obtain a bound that is uniform for @xmath428 .
finally , the regularity theorem will be applied in the case @xmath228 . in this case
an additional @xmath419-bound is required as well .
the proof uses the following `` bootstrap argument '' . at first
, we assume @xmath429 .
as @xmath52 is @xmath419 , the right hand side of , i.e. @xmath430 , is @xmath431 with @xmath432 .
we apply the global @xmath123-estimates [ theo.global.lp ] and get @xmath433 . using the sobolev embedding i , theorem [ theo.sobo ]
( a ) one obtains @xmath434 with @xmath435 . using @xmath436 one sees that @xmath437 ,
hence we have obtained stronger regularity for @xmath52 .
we iterate this argument and get @xmath438-bounds for arbitrarily large @xmath439 . for any @xmath440 , we obtain @xmath441 with @xmath442 .
we apply the sobolev embedding theorem ii , theorem [ theo.sobo ] ( c ) and obtain @xmath443 for any @xmath57 .
hence @xmath430 is @xmath444 as well , and applying schauder estimates [ theo.schauder ] we get @xmath445 for arbitrary @xmath417 .
the uniformity of the upper bound is clear from the construction .
the bootstrap can be continued and we obtain better regularity . [ prop.imp ] we assume the assumptions of the previous theorem . 1 .
. then @xmath447 .
2 . if @xmath448 , then @xmath449 for any @xmath450 $ ] . furthermore ,
@xmath451 where @xmath414 depends only on @xmath6 , @xmath125 , @xmath415 , @xmath416 , @xmath376 and @xmath417 .
3 . if @xmath31 and @xmath452 , then @xmath453 . furthermore , @xmath454 where @xmath414 depends only on @xmath6 , @xmath125 , @xmath415 , @xmath416 , @xmath376 and @xmath455 .
\(1 ) on @xmath391 we can continue the bootstrap argument and apply inductively the schauder estimates theorem [ theo.schauder ] .
we conclude that @xmath52 is smooth on @xmath391 .
we obtain ( 1 )
. however , uniform bounds will be difficult to obtain .
\(2 ) the case @xmath456 is trivial .
let @xmath457 .
we know that @xmath52 is @xmath458 for any @xmath459 .
hence , using appendix [ sec.scha.power ] one sees that @xmath430 is @xmath458 .
the schauder estimates imply that @xmath52 is @xmath180 .
\(3 ) similarly , if @xmath31 and @xmath460 , then @xmath461 is also smooth in @xmath38 . hence the bootstrap can go on with higher order schauder estimates , and we inductively get @xmath462 for any @xmath455 .
the construction obviously provides uniform bounds .
we obtain ( 3 ) .
[ prop.sol.subcrit ] let @xmath88 and @xmath125 be related via @xmath463 . for @xmath464 the supremum @xmath142 of @xmath103
is attained by a spinor field @xmath465 , @xmath450 $ ] , @xmath463 .
the spinor @xmath52 can be chosen such that @xmath52 is a solution of .
let @xmath280 be a maximizing sequence for @xmath103 , @xmath466 .
we may assume @xmath467 , and that @xmath280 is orthogonal to @xmath468 .
after taking a subsequence there is a @xmath469 such that @xmath280 converges weakly to @xmath299 in @xmath131 .
thus @xmath470 .
the compactness of the embedding @xmath471 provides a subsequence , that converges strongly to @xmath299 in @xmath123 .
this implies @xmath472 hence , @xmath473 as a consequence , we have equality in all inequalities , in particular @xmath474 . according to lemma [ lem.euler.lagrange ] one can find @xmath225 and @xmath475 such that @xmath476 solves .
proposition [ prop.imp ] ( improved regularity ) tells us that @xmath52 is @xmath180 .
[ theo.anydim ] let @xmath52 be a solution of with @xmath477 and @xmath478 , @xmath190 .
then there is a constant @xmath479 such that @xmath480 [ rem.sphereconc ] the conclusion of this theorem does not hold any longer , if we drop the condition @xmath481 .
in fact , if @xmath143 and @xmath482 , then there is the following counterexample : let @xmath332 be a non - isometric , but conformal map @xmath483 fixing north and south pole .
let @xmath52 be a killing spinor . as seen in the previous section @xmath52
is a maximizer of @xmath199 and a solution of .
the images @xmath484 under the conformal maps @xmath485 are also maximizers of @xmath199 and solutions of . however ,
as easily seen , @xmath486 tends to @xmath371 for @xmath487 .
assume that such a constant does not exist .
then we find a sequence of solutions @xmath488 of @xmath489 @xmath490 , @xmath491 and @xmath492 let us assume for a moment that @xmath493 . in this case , we can choose a subsequence with @xmath494 .
we have @xmath495 .
we conclude that @xmath488 is bounded in @xmath496 for a @xmath497 , and hence in @xmath419 for an @xmath407 .
then the regularity theorem ( theorem [ theo.reg ] ) says that @xmath486 is bounded , in contradiction to .
hence , the case @xmath498 can not occur , i.e. @xmath499 .
there is a sequence of points @xmath500 with @xmath501 the idea is to blow up suitably the metric such that we obtain in the limit a solution on euclidean @xmath502 .
we define @xmath503 one easily verifies @xmath504 and obviously transforms into @xmath505 we calculate @xmath506 we can assume that @xmath507 . as @xmath508 implies @xmath509 , we obtain @xmath510 the injectivity radius of @xmath511 tends to infinity , i.e. for any @xmath512 there is a @xmath513 such that for all @xmath514 the exponential map @xmath515 with respect to @xmath516 and based in @xmath517 is a diffeomorphism on the ball of radius @xmath518 around @xmath38 .
now , we identify @xmath519 with @xmath520 .
then @xmath521 is riemannian metric on @xmath522 that coincides with @xmath523 in @xmath38 . in the limit @xmath487
the metrics @xmath524 converge to @xmath523 in the @xmath525-topology . as already said in previous sections , the construction of the spinor bundle , its scalar product and its connection depends on the riemannian metric ( and the spin structure ) . in order to work out
the blowup construction one has to define a pull - back of the spinors @xmath526 via normal coordinates and then obtain a spinor in @xmath527 . in the literature
two such pull - backs are used , a pullback construction carried out in @xcite and inspired by @xcite , or via radial parallel transport ( see e.g. the solution of the index problem in @xcite using getzler rescaling ) .
both pullbacks can be used here .
the pullback in @xcite has better approximation properties , and is an important tool if one wants to study fine asymptotics of the blowup .
however , the pullback via radial parallel transport is technically simpler to introduce , hence it will be used here . for @xmath512 let @xmath528 ( resp .
@xmath529 ) be the fiber of @xmath530 over @xmath38 ( resp . @xmath531 over @xmath517 ) .
let us define the radial vector field @xmath532 on @xmath533 .
its length is the distance from @xmath38 . for sufficiently large @xmath376 ,
the exponential map of @xmath534 based in @xmath517 , denoted by @xmath535 is a diffeomorphism from @xmath522 onto its image .
one chooses a ( complex ) linear isometry @xmath536 .
this map extends uniquely to a fiber preserving map @xmath537 , such that @xmath538 commutes and such that @xmath539 in a neighborhood of @xmath517 , this condition can be equivalently characterized by saying that @xmath540 maps parallel sections of @xmath530 to radially parallel sections of @xmath541 .
one easily sees that @xmath332 is a fiberwise isometry , but the connection on the spinor bundle is not preserved .
however , this will not matter , as in the limit @xmath487 , @xmath518 fixed , the connections converge in the @xmath525-topology .
let @xmath542 .
on @xmath543 we define the spinor @xmath544 and the operator @xmath545 , @xmath546 where @xmath547 is the dirac operator on @xmath548 .
we obtain @xmath549 note that @xmath550 hence , we may apply the interior @xmath123- and the schauder - estimates [ theo.int.lp ] and [ theo.schauder ] to conclude that @xmath551 with constants @xmath552 and @xmath553 .
the constant @xmath552 does not depend on @xmath376 if @xmath524 is sufficiently @xmath525-close to @xmath523 . in particular , the @xmath525-convergence @xmath554 says that @xmath552 does not depend on @xmath376 for @xmath555 .
compare @xmath556 with the dirac operator @xmath557 on euclidean @xmath502 .
( for similar and more explicit calculations the reader might consider e.g. @xcite for pullback via radial parallel transport , and @xcite for the other pullback method ) .
we have @xmath558 with @xmath559 .
the convergence is not uniform in @xmath518 , but this will not matter in the following .
we choose a sequence of radii @xmath560 . for each @xmath561 ,
the arcela - ascoli theorem ( theorem [ theo.arc.asc ] ) allows us to choose a subsequence of @xmath562 converging in @xmath563 . after passing to a diagonal sequence
, we see that there is a spinor @xmath564 on @xmath502 , such that @xmath565 converges to @xmath566 for all @xmath512 .
then @xmath567 is a solution of @xmath568 on @xmath502 .
the estimate says that @xmath569 .
hence , for any @xmath190 and @xmath512 there is @xmath570 such that @xmath571 for all @xmath572 . because of the @xmath573-convergence @xmath574 , fatou s lemma yields @xmath575 for any @xmath576 , and finally for @xmath577 .
we identify @xmath567 via stereographic projection with an @xmath578-spinor @xmath579 on @xmath580 with the identification provided by the application @xmath335 directly after proposition [ prop.conf.change ] .
we obtain @xmath581 and @xmath582
. the removal of singularities theorem , i.e. theorem [ theo.sing.rem ] , says that holds on the whole sphere @xmath337 .
@xmath583 which is apparently a contradiction to our assumption @xmath584 .
if there is a @xmath585 and an @xmath407 such that for all @xmath586 there is a solution @xmath587 of equation with @xmath588 , @xmath402 and such that @xmath589 is bounded by a constant @xmath414 independent from @xmath246 , then there is a sequence @xmath590 such that @xmath591 converges in the @xmath573-topology to a solution of equation with @xmath228 . for @xmath125 sufficiently close to @xmath592 , we know because of proposition [ prop.func ] that @xmath593 is bounded from below by a positive constant .
thus , we can apply the regularity theorem ( theorem [ theo.reg ] ) which tells us that @xmath594 is uniformly bounded in @xmath458 .
hence , for a sequence @xmath595 with @xmath596 , converging to @xmath592 , the spinor fields @xmath591 converge in the @xmath573-topology to a @xmath573-spinor field @xmath597 which is a solution of equation with @xmath228 .
in this section we want to prove the main result of this publication , namely theorem [ theo.main ] .
proposition [ prop.sol.subcrit ] tells us that for any @xmath598 , @xmath128 , the functional @xmath103 is attained by a maximizer denoted @xmath599 satisfying @xmath600 where @xmath125 and @xmath88 are related via @xmath144 .
we have assumed that @xmath601 which is equivalent to @xmath602 . as the function @xmath603 is continuous from the right ( proposition [ prop.func ] ) ,
we see that there is an @xmath190 such that @xmath604 for @xmath88 close to @xmath176 . for such @xmath88 ,
theorem [ theo.anydim ] implies that @xmath599 are uniformly bounded in the @xmath605-norm , and then we can use theorem [ theo.reg ] to conclude that these @xmath599 are even uniformly bounded in @xmath458 .
the theorem of arcela - ascoli ( theorem [ theo.arc.asc ] ) implies that there is there is a sequence @xmath606 such that @xmath607 converges in the @xmath573-norm to a solution @xmath52 of @xmath608 theorem [ theo.reg ] and proposition [ prop.imp ] then show that @xmath52 has the desired regularity , and statement ( a ) is proven .
we will now show that statement ( b ) follows from statement ( a ) .
if we have a solution as in ( a ) , then we set @xmath609 with @xmath610 .
note that @xmath611 .
the transformation formula for the dirac operator under conformal changes ( proposition [ prop.conf.change ] ) implies that there is a spinor @xmath209 on @xmath612 such that @xmath613 then obviously , @xmath614 and ( b ) follows . statement ( c ) of theorem [ theo.main ]
will follow from proposition [ prop.zero.est ] , proven in the next section .
in this section we want to study that the zero set of solutions of the euler - lagrange equation @xmath615 note that very often the zero set of a function @xmath52 solving an equation of the above type is denoted as the _ nodal set of @xmath52_. the following theorem is due to c. br @xcite for smooth @xmath616 .
let @xmath386 be a riemannian manifold and let @xmath52 be a solution of @xmath617 where @xmath616 is a smooth function .
then the nodal set of @xmath52 has hausdorff dimension at most @xmath618 .
unfortunately , has not the desired form as @xmath619 is not smooth for non - integer @xmath125 , and br s proof does not extend to such a @xmath616 .
nevertheless , we conjecture that the theorem also holds for @xmath619 , @xmath457 .
the nodal set of any solution of has hausdorff dimension at most @xmath618 . if @xmath31 and @xmath452 , then we have better regularity . in this case solutions of equation and
the corresponding @xmath620 are smooth .
hence , using ( * ? ? ?
* main theorem ) , one sees that in this case the nodal set of a solution is a discrete subset .
the following proposition controls its cardinality .
[ prop.zero.est ] on a compact spin surface @xmath6 of genus @xmath621 let @xmath52 be a solution of equation @xmath622 then the number of zeros of @xmath52 is at most @xmath623 in particular , this implies part ( c ) of theorem [ theo.main ] .
we set @xmath624 .
then outside the zero set we know by lemma [ lem.scal.bound ] that the gauss curvature of @xmath363 is at most @xmath625 . furthermore @xmath187
let @xmath626 .
the integral of the geodesic curvature with respect to @xmath363 over small simply closed loop around @xmath125 is close to @xmath627 , where @xmath628 is the order of the first non - vanishing term in the taylor expansion of @xmath52 in @xmath125 .
we remove small open disks around the zeros of @xmath52 from @xmath2 , and we obtain a surface with boundary @xmath629 . with the gauss - bonnet theorem
we obtain @xmath630 and hence @xmath631 , which implies the proposition .
the aim of this section is to recall the spinorial weierstrass representation .
weierstrass published a representation of minimal surfaces in @xmath72 in terms of holomorphic functions @xcite .
his article deals only with local questions , everything is described in a fixed conformal chart of the surface . from a modern ( chart free ) point of view , it is clear that these holomorphic sections should be interpreted as a section of the spinor bundle , and `` holomorphy '' translates into a `` harmonicity '' , i.e. the surface is minimal iff the corresponding spinor @xmath52 satisfies @xmath632 . during the 20th century
several attempts were undertaken to globalize the weierstrass representation and to adapt it to arbitrary surfaces .
unfortunately , most approaches replaced weierstrass original approach by a formulation in terms of a holomorphic @xmath43-form and a holomorphic function .
the corresponding formula were quite involved , and hence not very suitable for applications .
an amazing breakthrough was achieved by work of d. sullivan , r. kusner , and n. schmitt around 1990 , and independently by u. abresch . in early 1989 ,
dennis sullivan put together some unpublished notes explaining the spinorial character of the weierstrass representation for minimal surfaces . in spring 1989
, robert kusner realized that the spinor formalism is not limited to minimal surfaces , but extends to conformal immersions of arbitrary surfaces , as described below .
these techniques were presented in sullivan s cuny seminar in 1992 , and spread around among the experts rapidly .
kusner s results found their continuation in the phd thesis of nick schmitt @xcite .
schmitt found many interesting applications of the spinorial weierstrass .
the results of kusner and schmitt led to the publications @xcite and @xcite .
independently , abresch developed a spinorial weierstrass for constant mean curvature surfaces .
unfortunately , the resulting document of abresch , some handwritten lecture notes from a conference in luminy , were never published .
we also want to mention an earlier result of pinkall @xcite . in the special class of oriented surfaces ,
pinkall s result establishes a bijection between regular homotopy classes of immersions of oriented surfaces @xmath2 and @xmath633-valued quadratic forms on @xmath634 .
these quadratic forms are in bijection with the spin structures obtained in the spinorial weierstrass representation .
more recent literature concerning this representation can be found for example in @xcite and in articles by pinkall , taimanov , m.u .
schmidt , morel , voss and their collaborators .
however , this list is far from being exhaustive .
in our exposition we roughly follow @xcite . the setting for the spinorial weierstrass representation is as follows .
let @xmath2 be a compact riemann surface of genus @xmath621 .
the vector bundles @xmath635 and @xmath636 are defined as the complex linear part and the complex anti - linear part of @xmath637 .
the compositions @xmath638 @xmath639 of the complexification and the projection on @xmath640 resp .
@xmath641 define vector space homomorphisms @xmath642 and @xmath643 .
these map @xmath644and @xmath645 preserve the natural connections .
however , one should pay attention to the fact , that these maps do not preserve lengths , but @xmath646 .
the maps @xmath647 resp .
@xmath648 is denoted as the _ real part _ , namely @xmath649 for @xmath650 and the same notation
is used in the @xmath651 case .
_ complex conjugation _
maps @xmath652 to @xmath653 and vice versa .
as the second stiefel - whitney - class of @xmath2 is the @xmath654 reduction of the euler - class of @xmath655 , one sees that @xmath2 is spin .
however , the space of spin structures on @xmath2 is not unique : it is an affine space for the group @xmath656 , where @xmath621 denotes the genus of @xmath2 . hence , there are @xmath657 spin structures on @xmath2 .
if a spin structure is fixed , then the associated vector bundle with respect to the standard representation of @xmath658 on @xmath659 is a complex line bundle @xmath660 satisfying @xmath661 .
if we equip @xmath660 with the natural hermitian metric , we can choose this map such that the hermitian metric ( res .
the connection ) on @xmath652 is the tensor product metric ( resp .
tensor product connection ) .
we define @xmath662 to be @xmath660 with the conjugated complex structure .
in particular , there is a natural anti - linear conjugation map @xmath663 , the hermitian product defines a complex bilinear metric contraction @xmath664 , and @xmath665 . in particular , @xmath666 , and
hence the dolbeault operator is a map @xmath667 , and similarly @xmath668 .
we define @xmath669 and @xmath670 as the compositions @xmath671 @xmath672 composing @xmath669 with complex conjugation yields @xmath669 and vice versa .
one calculates @xmath673 as a consequence , the map @xmath674 satisfies the clifford relations .
one sees , that the sum @xmath675 can be identified with the standard spinor bundle on @xmath2 in such a way that the above map is the clifford multiplication , and such that @xmath660 resp .
@xmath662 are the positive resp .
negative half - spinors . the dirac operator can be written in this notation as @xmath676 now let us assume that @xmath677 is a solution of @xmath678 , where @xmath679 is a real - valued function on @xmath2 .
this means @xmath680 @xmath681 we define @xmath682 let @xmath82 denote the universal covering of @xmath2 , and @xmath683 the group of deck transformations .
as @xmath684 is imaginary , we can find a function @xmath74 , such that @xmath685 , and there is a homomorphism @xmath686 , such that @xmath687 one calculates that @xmath83 is a conformal map with possible branching points , @xmath688 and that @xmath689 has mean curvature @xmath679 .
hence , the map @xmath83 satisfies properties ( 1 ) to ( 3 ) from the introduction , i.e. it is a _ periodic branched conformal immersion @xmath83 based on @xmath21 _ with mean curvature @xmath679 .
in any zero of of the spinor @xmath52 , the map @xmath83 has a branching point .
if @xmath83 vanishes of order @xmath376 , then @xmath417 vanishes of order @xmath690 .
hence , all branching points of @xmath83 are necessarily of even order . summarizing the above statement
, we obtain for any solution of @xmath691 a periodic branched conformal immersion of @xmath82 into @xmath72 which is uniquely determined up to translation .
if @xmath52 solves @xmath691 , then @xmath692 as well , and the corresponding @xmath83 is the same .
hence , we obtain a well - defined map @xmath693 \{}/@xmath694 and one can show that this map is even a bijection .
the inverse of this map is given by restricting a parallel spinor on @xmath72 to @xmath689 and by performing a conformal change @xcite .
if @xmath679 is constant , then there is also another version of the spinorial weierstrass representation , where the target space is @xmath695 instead of @xmath72 .
we view @xmath695 as @xmath696 with a bi - invariant metric of constant curvature @xmath43 , the multiplication in @xmath696 is denoted with @xmath697 .
the periodicity condition ( 1 ) has to be replaced by 1 .
_ left periodicity : _ there is a homomorphism @xmath698 , the _ periodicity map _ , such that for any @xmath77 , and @xmath78 one has @xmath699 here @xmath80 denotes the action of @xmath81 on @xmath82 via deck transformation .
one obtains a bijection @xmath700 \{}/@xmath694 where @xmath701 . for details
see @xcite .
if the dimension of @xmath2 is @xmath68 , then reads as @xmath702 and according to the regularity theory @xmath52 is even smooth . the spinorial weierstrass representation explained in the previous section tells us , that such a solution can be used to construct certain immersions with constant mean curvature . combining the previous results we obtain the following application that is a stronger version of the `` principle for construction of cmc - surfaces '' mentioned in the introduction .
assume that the riemann spin surface @xmath6 satisfies @xmath703,\si)<2\pi.\ ] ] then there is a periodic branched conformal cmc immersion @xmath74 based on @xmath21 .
the mean curvature is equal to @xmath704,\si)$ ] and the area of a fundamental domain is @xmath43 . the regular homotopy class of @xmath83 is determined by the spin - structure @xmath1 .
the indices of all branching points are even , and the sum of these indices is smaller than @xmath92 . in particular , if @xmath2 is a torus , there are no branching points .
the proof is a direct consequence of theorem [ theo.main ] and the previous section .
there are many examples of stationary points of the functional . however , it is still open whether they are the maximizers or not .
note , that by changing the orientation of a surfaces , a maximizers @xmath221 of @xmath199 on @xmath2 turns into a minimizer @xmath705 of @xmath199 on the surface with reversed orientation @xmath629 , with @xmath706 .
let us study some examples . 1 .
let @xmath21 be a @xmath68-dimensional torus . via a conformal change
we can achieve that @xmath14 is flat , i.e. @xmath707 , equipped with the euclidean metric .
we assume that the lattice @xmath708 is generated by @xmath709 and @xmath710 , with @xmath711 .
the spinor bundle of a flat manifold is flat as well , hence the holonomy is a map @xmath712 .
indeed , the image of this map is contained in @xmath713 .
we obtain a homomorphism @xmath714 .
this homomorphisms characterizes the spin structure @xmath1 in the sense that two spin structures on @xmath21 are isomorphic iff the homomorphisms @xmath715 coincide , and to each such homomorphism there is a spin structure .
the case @xmath716 corresponds to the so - called _ trivial _ spin structure @xmath717 , the other cases correspond to _ non - trivial _ spin structures . that defines a _ non - trivial _ element in the bordism class .
] + at first , we deal with the case @xmath718 . in this case , after a possible rotation and a possible homothety , we can achieve @xmath719 on @xmath720 the kernel of @xmath148 has complex dimension @xmath68 , and consists of parallel spinors .
if one carries out the constructions from the last sections for a parallel spinor , then one obtains an affine conformal map @xmath83 .
such an @xmath83 is trivially a periodic conformal immersion @xmath83 based on @xmath21 with _ vanishing _ mean curvature .
however , if @xmath721 then one easily sees that @xmath722 .
the proposition yields the existence of a periodic conformal immersion @xmath83 based on @xmath21 with constant mean curvature @xmath723 , and the area of a fundamental domain is @xmath43 .
however , the proposition only provides the existence of the solution , but we can not characterize the maximizer . in our example , a family of solutions can be explicitly written , namely it is the family of parameterized cylinders @xmath724 for any @xmath725 , @xmath726 .
we conjecture that these solutions are exactly those that correspond to the maximizers and the minimizers of @xmath199 , when we normalize such that all spinors have @xmath727-norm @xmath43 .
+ in the case @xmath728 we can achieve that @xmath729 @xmath730 the dirac operator is always invertible .
+ one easily sees @xmath731 .
hence , the proposition yields solutions for @xmath732 .
once again , solutions can be explicitly written , namely the parameterized cylinder @xmath733 for any @xmath725 , @xmath726 . however , in some cases , e.g. if @xmath734 and @xmath735 , these solutions no longer correspond to maximizers and minimizers , but to saddle points of the functional .
we conjecture that in the case @xmath734 , @xmath736 the maximizers and minimizers correspond to the unduloid immersions ( see figure [ fig.undu ] ) .
an unduloid is a surface of revolution of constant mean curvature .
2 . if @xmath2 has genus @xmath68 , then as in the case of the torus , the dimension of the kernel is independent of the metric , however it depends on the spin structure .
if @xmath1 is a spin structure such that @xmath737 is spin - cobordant @xmath38 , then the dirac operator is invertible for any metric . again , as in the torus case , one can find for any @xmath190 a conformal classes @xmath738 on @xmath2 with @xmath704,\si)<\ep$ ] .
@xcite 3 .
if the genus is larger than @xmath68 , then the kernel of the dirac operator on a riemannian spin manifold @xmath737 depends on the metric .
for example if @xmath2 is a surface of genus @xmath739 equipped with the spin structure @xmath1 and the conformal structure @xmath740 associated to the periodic conformal immersion with vanishing mean curvature indicated in figure [ fig.kgb.min ] . ]
this immersion induces a harmonic spinor on @xmath6 .
however , as @xmath737 is spin - cobordant @xmath38 , there is a perturbation @xmath741 $ ] of the conformal structure such that the dirac operator on @xmath742 has a trivial kernel for small @xmath743 @xcite . in this case
@xmath744,\si)=0,\ ] ] hence there exist solutions of .
such a solution is visualized in figure [ fig.kgb.cmc ] .
constant mean curvature immersions of @xmath745 into @xmath72 , in particular wente tori and twisty tori ( figure [ fig.twisty ] ) also correspond to stationary points of @xmath199 .
however , as they are not embedded @xcite tells us that @xmath746 . on the other hand maximizers and minimizers of @xmath199 satisfy @xmath747 , hence these tori do _ not _ correspond to maximizers or minimizers .
similar propositions also hold for immersions into @xmath695 and into hyperbolic space @xmath748 , see @xcite .
we will only specify the case of @xmath749 .
assume that the riemann spin surface @xmath6 satisfies @xmath750,\si)<2\pi.\ ] ] let @xmath751,\si)$ ] be given .
then there is a left periodic branched conformal cmc immersion @xmath752 based on @xmath21 .
the mean curvature is equal to @xmath753,\si)/a)^2 - 1}$ ] , and the area of a fundamental domain is @xmath754 .
the regular homotopy class of @xmath83 is determined by the spin - structure @xmath1 .
the indices of all branching points are even , and the sum of these indices is smaller than @xmath92 . in particular , if @xmath2 is a torus , there are no branching points .
an example where the image of the periodicity map has a finite image in @xmath696 is given in figure [ fig.undu.s ] .
if @xmath21 is an analytic riemannian manifold of dimension @xmath739 , then there is an analogue of the spinorial weierstrass representation .
this implies that solutions @xmath52 of can be geometrically interpreted as a conformal cmc embedding of @xmath755 into a ( non - complete ) @xmath756-dimensional riemannian manifold @xmath757 .
this manifold @xmath757 carries a parallel spinor whose restriction to @xmath21 is again the solution of .
the manifold @xmath757 depends on @xmath52 and is unique up to restriction to subsets and coverings .
a more detailed exposition of this @xmath739-dimensional version is work in progress .
in this section we want to collect some facts about elliptic regularity for dirac operators .
the proofs of these statements are analogous to proofs of the corresponding statements for the laplace operators as done e.g. in @xcite and @xcite .
details on how to prove most statements of this section are given also in @xcite .
either a proof is provided there or it is sketched how the statements are reduced to standard theorems .
let @xmath758 be a riemannian manifold , possibly with boundary , with a spin structure @xmath1 .
the interior of @xmath759 is denoted as @xmath760 .
the spinor bundle @xmath761 is a complex vector bundle carrying a natural connection and a natural hermitian metric .
we now define the sobolev norms of spinors fields . for any spinor @xmath221 smooth on @xmath759 , and @xmath762 , @xmath763 ,
we define the @xmath764-norm of @xmath221 as @xmath765 when the domain of integration is clear , we simply write @xmath766 instead of @xmath767 .
the closure of the space of smooth spinors with respect to this norms is denoted as @xmath768 .
if the underlying structure of riemannian spin manifold is clear from the context , we often write shortly @xmath764 , or in other situations where we want to emphasize the metric we write @xmath769 .
we will also write @xmath392 for @xmath770 .
[ theo.int.lp ] let @xmath771 be a compact riemannian spin manifold , possibly _
with boundary_. let @xmath416 be a compact subset of @xmath760 .
let @xmath221 be in @xmath772 and let @xmath773 be a weak solution ( i.e. in the sense of distributions ) of @xmath774 on @xmath760 .
then @xmath775 and @xmath776 where @xmath777 , i.e. @xmath414 depends only on @xmath778 and @xmath88 .
furthermore , if a sequence of metrics @xmath779 converges in the @xmath525-topology to a riemannian metric @xmath14 , then the constants @xmath414 can be chosen such that @xmath780 in order to prove the theorem , it is sufficient to prove it in the case that @xmath759 and @xmath416 are small concentric geodesic balls , the radius being controlled in terms of curvature bounds , and bounds on derivatives of the curvature .
as explained in @xcite this can be done in a way analogous to the corresponding statement for the laplacian in @xcite .
the general statement then can be reduced to the special case by covering ( the general ) @xmath416 by finitely many small open balls and an associated partition of unity . in the special case that the boundary of @xmath759 is the empty set , and that @xmath781 , then there is a stronger version , that will be used as well .
[ theo.global.lp ] let @xmath771 be a compact riemannian spin manifold _ without boundary _ and @xmath782 .
then any weak solution @xmath773 of @xmath774 satisfies @xmath783 , and there is a constant @xmath784 such that @xmath785 where @xmath786 is the @xmath787-orthogonal projection to the kernel of @xmath5 .
the proof of this theorem is not difficult if one uses some facts about pseudodifferential operators , explained for example in @xcite .
we abbreviate @xmath788 .
the spectrum of the elliptic operator @xmath789 is bounded away from @xmath38 , hence it is invertible and its inverse is a pseudodifferential operator of degree @xmath790 .
pseudodifferential operators of degree @xmath790 are continuous from @xmath392 to @xmath131 .
hence , @xmath791 this is the statement of the theorem in the case @xmath792 , and the statement for @xmath793 follows for example by using theorem [ theo.int.lp ] .
however , a uniformity statement for converging @xmath794 as in theorem [ theo.int.lp ] does not hold . if @xmath795 , then one easily shows that there are eigenspinors @xmath280 for @xmath796 to eigenvalues @xmath797 , @xmath798 with @xmath799 which would contradict a uniform version of theorem [ theo.global.lp ] .
we also needs some facts about hlder norms .
let @xmath800 $ ] .
on @xmath801 we define the _ hlder norms _ @xmath802
if the shortest geodesic is not unique , then we use the convention that the supremum runs over all possible choices of shortest geodesics .
the completions of @xmath801 with respect to the hlder norms @xmath444 and @xmath458 define the _ hlder spaces _ @xmath803 and @xmath804 .
hlder norms are important as they admit schauder estimates .
[ theo.schauder ] let @xmath771 be a compact riemannian spin manifold , possibly _ with boundary _ , and let @xmath416 be a compact subset of @xmath760 .
suppose @xmath805 , @xmath763 .
then for any weak solution @xmath806 of @xmath774 we have @xmath807 and @xmath808 where @xmath809 . furthermore ,
if a sequence of metrics @xmath779 converges in the @xmath525-topology to a riemannian metric @xmath14 , then the constants @xmath414 can be chosen such that @xmath780 as before , one important special case is that @xmath759 has empty boundary and @xmath781 .
the proof can be done in a way analogous to the proof of the corresponding statements for the laplacian in @xcite .
again it is sufficient to prove it for small concentric geodesic balls , and to glue them together .
we omit the details . for a proof for concentric geodesic balls
is provided for example in @xcite . in local charts one
easily reduces the following theorem to the standard arcela - ascoli theorem .
[ theo.arc.asc ] let @xmath771 be a riemannian spin manifold , possibly _
with boundary_. for @xmath810 , the inclusion @xmath811 is compact , i.e. a bounded sequence in @xmath812 has a subsequence convergent in @xmath813 .
sobolev and hlder spaces are related by a several embedding theorems , some of them are compact , others are only bounded .
we summarize the embeddings that are needed in the article .
[ theo.sobo]let @xmath814 , @xmath815 and @xmath816 .
let @xmath6 be a compact riemannian spin manifold _ without boundary_. all spaces of functions
are defined on sections of @xmath266 . 1 .
( sobolev embedding theorem i ) . if @xmath817 then @xmath818 is continuously embedded into @xmath819 .
( rellich - kondrakov theorem ) .
if strict inequality holds in ( a.1 ) and if @xmath820 , then the inclusion @xmath821 is a compact map .
3 . ( sobolev embedding theorem ii ) . suppose @xmath822 , @xmath823 and @xmath824 then @xmath818 is continuously embedded into @xmath825 .
in this section we want to include some proofs of probably well - known statements about hlder spaces .
the section @xmath52 is obviously lipschitz , hence @xmath845 is @xmath846 .
we have to show that @xmath847 is @xmath848 .
the first summand is a product of @xmath846 and @xmath830 , hence @xmath849 .
according to the previous lemma , @xmath850 is @xmath846 , hence the second summand is a product of @xmath830 and @xmath846 , hence also @xmath848 .
to3em , _ spectral properties of the dirac operator and geometrical structures . _ ,
ocampo , hernan ( ed . ) et al .
, geometric methods for quantum field theory . proceedings of the summer school , villa de leyva , colombia , july 12 - 30 , 1999 .
singapore : world scientific .
116 - 169 , 2001 . | let us fix a conformal class @xmath0 $ ] and a spin structure @xmath1 on a compact manifold @xmath2 .
for any @xmath3 $ ] , let @xmath4 be the smallest positive eigenvalue of the dirac operator @xmath5 on @xmath6 .
in a previous article we have shown that @xmath7 } \la_1^+(g)\vol(m , g)^{1/n}>0.\ ] ] in the present article , we enlarge the conformal class by adding certain singular metrics .
we will show that if @xmath8 , then the infimum is attained on the enlarged conformal class . for proving this , we solve a system of semi - linear partial differential equations involving a nonlinearity with critical exponent
: @xmath9 the solution of this problem has many analogies to the solution of the yamabe problem .
however , our reasoning is more involved than in the yamabe problem as the eigenvalues of the dirac operator tend to @xmath10 and @xmath11 . using the spinorial weierstra representation
, the solution of this equation in dimension 2 shows the existence of many periodic constant mean curvature surfaces .
* keywords : * dirac operator , eigenvalues , conformal geometry , critical sobolev exponents * mathematics classification : * 58j50 , 53c27 ( primary ) , 58c40 , 35p15 , 35p30 , 35b33 ( secondary ) |
an effective tool to study the space - time structure of the production amplitude is to measure the bose - einstein correlations ( bec ) between two identical particles produced in the inclusive hadron interaction , see , for example , hbt@xmath5@xcite .
consider the situation where we have one pion with momentum @xmath6 emitted at point @xmath7 and another identical pion with @xmath8 and @xmath9 .
the inclusive cross section for the two identical particles takes the form = |m|^22 + 2e^irq = |m|^2 1+e^irq , where @xmath10 is the production amplitude , and where 4-vectors @xmath11 and @xmath12 .
the @xmath13 denote the averaging over @xmath7 and @xmath9 , the @xmath14 term is due to the permutation of the identical pions ; that is , it allows for the pion with @xmath8 to be emitted from the point @xmath7 and simultaneously for @xmath6 from @xmath9 . as a rule
the @xmath15 dependence of the amplitude @xmath10 is relatively flat in comparison with the @xmath15 dependence of @xmath14 .
thus we are able to evaluate the size of the pion production domain by studying the @xmath15 dependence of the whole cross section @xmath16 . to extract the effect we compare the measured @xmath15 spectrum with a similar one but without bec . to be precise
we form the ratio r(q ) = [ eq : rq ] where @xmath17 is the two pion distribution integrated over all the variables except @xmath15 , and @xmath18 is the distribution expected in a world without bec .
there are different ways to choose @xmath19 .
we may measure the @xmath20 distribution for non - identical pions ; or we may change the sign of the three momentum of the second pion @xmath21 ; and so on . none of these approaches compensates for the @xmath15 dependence of @xmath10 completely ; but for the conventional ` one - radius ' fit r(q ) = e^-|r q [ eq : rq1 ] the different values of the mean radius , is such that @xmath22 approximates the value of @xmath14 averaged over @xmath7 and @xmath9 . ] , @xmath23 , extracted from the data are close to each other .
such analyses of high energy proton - proton interactions at the lhc have been performed by atlas @xcite , cms @xcite and alice @xcite .
the problem is that the value of @xmath24 turns out to be less than 1 . in particular ,
cms claim the @xmath25 @xcite . on the other hand from ( [ eq : rq ] ) we expect @xmath26 .
moreover , it is clear from fig . 1 of @xcite , and the analogous plots of other groups , that the fit does not describe the very low q data points .
this indicates that there should be another component of @xmath27 with a larger radius populating the region of small @xmath15 .
in the present paper we argue that the expected structure of the pion emission domain is highly inhomogeneous .
we should consider _ small _ size pion sources distributed in the much larger area of the proton - proton interaction .
that is , we are led to parametrize @xmath27 by two different mean radii .
we explain the physical origin of this situation below .
it was shown long ago that to describe a high energy ( say , proton - proton ) interaction it is convenient to first select the subset of diagrams which provides the interaction across a large rapidity gap , and whose contribution does not decrease when the rapidity separation increases @xcite .
the resulting ladder - like set of diagrams forms reggeon exchange . in terms of the hadronic degrees of freedom
the corresponding subset of diagrams called multiperipheral ladder diagrams ( fig .
1(a ) ) was studied first by amati el al . @xcite . in terms of qcd
they form the bfkl pomeron @xcite . in a general purpose monte carlo
these are the diagrams for the dglap evolution amplitude ( fig .
recall that it is not sufficient to consider only one ladder . to describe multiparticle production we have to consider the possibility of the exchange of a few ladders ( reggeons ) , see fig .
1(c ) . in a monte carlo
this is called the multiple interaction ( mi ) option . already at this stage
we observe two quite different scales .
the slope , @xmath28 , of elastic proton - proton scattering is usually parametrized in the form b_el(s ) = b_0 + 2_pln(s / s_0 ) , [ eq : slope ] where the constant @xmath29 is driven by the size of an incoming proton . on the other hand ,
the value of @xmath30 reflects the internal transverse size of the ladder .
based on the pre - lhc data , typical numbers are @xmath31 gev@xmath32 and @xmath33 @xcite . strictly speaking , ( [ eq : slope ] ) corresponds to one - ladder ( reggeon ) exchange . at very high energies the speed of the shrinkage of the diffraction cone increases due to the stronger screening of the amplitude in the centre of the disc , at small impact parameters
asymptotically , in the black disc limit , where the total cross section grows as ln@xmath34 , the slope @xmath35 also increases as ln@xmath36 .
it was shown @xcite that indeed the high energy lhc data indicate the presence of a ln@xmath34 component in the elastic slope which is consistent with the growth of the total cross section .
however if we consider an individual ladder then the effective value of @xmath30 in the ladder even decreases with energy due to the larger available @xmath37 space for the intermediate partons .
it is known that @xmath38 in the bfkl case .
another example is the monte carlo description of multiparticle production . in order to tune the generator to
the high energy data one has to introduce an infrared cutoff , @xmath39 whose value grows as @xmath40 @xcite that is , the transverse size of the ladder decreases .
thus the interaction of two high energy protons should be described by a diagram like fig.1(c ) in which the size of each individual ladder is rather small ( as seen from the small value of @xmath30 ) . yet
the separation between the ladders is of the order of the radius of the interaction amplitude which should be correlated with the total value of @xmath35 .
with increasing energy we expect the transverse size of an individual ladder ( measured in the central rapidity interval ) will decrease ( as indicated by the behaviour of @xmath41 ) . on the other hand the separation between the ladders
is expected to increase , as indicated by the behaviour of @xmath35 .
having the above picture in mind , we expect in bec to observe a new object a small - size pion source . in other words ,
bec should be described by two different radii .
one radius corresponds to the case when both pions are emitted from the same ladder this will measure the size of an individual ` pomeron ' . since the pion is not a point - like object the radius will be smeared out by fluctuations in the process of the formation of the pions .
the second radius will correspond to the pions being emitted from two different ladders
it is a measure of the separation between the ladders .
therefore we propose to fit the observed correlation @xmath27 by a formula with two different mean radii ) , other parametric forms may be considered for each term .
for example , the second term may be a gaussian , @xmath42 .
the choice should be based on statistical criteria or on the relative strength of the two terms . ]
r(q ) = e^-|r_1 q + ( 1- ) e^-|r_2 q , [ eq : rq2 ] which better reflects the complicated structure of the pion emission domain . in the ideal case
we expect the low multiplicity events to be produced via a diagram with only one ` cut ' pomeron exchange ( fig .
in general there may be more pomerons in the whole amplitude , but only one ladder radiates the secondaries . as the multiplicity becomes larger the secondaries are mainly emitted from a few different ladders , and the probability to find the two identical pions originating from the same pomeron decreases .
that is , we expect the relative contribution @xmath24 of the large component ( described by , say , @xmath43 ) to increase with @xmath44 , while on the other hand , the strength of the small size component @xmath45 decreases .
unfortunately we can not predict that @xmath46 as @xmath47 , and that @xmath48 for very large @xmath44 .
the situation is complicated by the strong fluctuations of the number of secondaries in each ladder ( or pomeron ) .
recall that actually we do not measure the total charged multiplicity of an event , but rather the number of secondaries in a limited central rapidity interval ( like @xmath49 in the case of atlas and cms ) .
then the multiplicity corresponding to one pomeron is relatively small ( @xmath50 ) and the fluctuations strongly wash out the relation between the measured values and the number of cut pomerons .
moreover for very low @xmath44 we may sample contributions from diffractive dissociation which have a qualitatively different structure . nevertheless at large
@xmath44 the multipomeron contribution dominates ; that is bec are driven mainly by the component with the largest radius , @xmath51 .
indeed it is seen in fig .
3(b ) of @xcite that , in the _ one _ radius fit , ( [ eq : rq1 ] ) , the radius increases with multiplicity reaching _ saturation _ of @xmath52fm for @xmath53 .
such saturation was predicted in @xcite , where it was explained that the radius measured in a one - radius fit , is driven , not by the initial energy , but mainly by the number of cut pomerons , @xmath54 .
indeed , the radius of the individual pomeron depends weakly on energy ( @xmath55 const . ) , while the number of cut pomerons observed in the event ( that is the separated pion sources ) is proportional to @xmath56 .
so the probability to have two identical pions from two different sources increases with @xmath44 .
when @xmath57 ( at low @xmath44 ) we observe one pomeron and measure its radius .
on the other hand , for large @xmath44 we study the separation between the pomerons and the value of @xmath58 is saturated at the radius of the interaction amplitude @xmath59 .
it was shown in @xcite that @xmath60 gev@xmath32 @xcite corresponds in a ` one - radius ' fit to @xmath61fm , which is in good agreement with fig .
3(b ) of the atlas fit @xcite .
we emphasize that the separation between the pomerons ( the pion sources ) is not equal to the incoming hadron ( proton ) radius .
first note , the radius corresponding to the interaction amplitude is larger .
it increases with energy .
recall that in each successive step of the ladder in fig . 1 the impact parameter changes by @xmath62 .
this leads to a diffusion in @xmath63 , which results in the second ( or @xmath64 ) term in the equation for the elastic slope , ( [ eq : slope ] ) .
next , the pion is not a point - like particle and its formation also occupies some volume .
finally , there may be an interaction between the final state secondaries . strictly speaking the picture that we describe above corresponds to the initial stage of the interaction , and
does not allow for possible final state rescattering .
if there are final state interactions ( either in terms of a hadron gas or a quark - gluon plasma ) then bec will measure the radius given by the point of the last interaction that is , the domain occupied by secondaries is extended up to the stage where the particle density becomes so low that further interaction is very unlikely , and the size of the domain at which the last interaction occurs will be seen in bec . the probability of final state rescattering increases with particle density @xmath44 .
if this contribution were to dominate , then it is natural to expect @xmath65 .
however , the data of fig .
3(b ) of @xcite do not show any evidence of such a behaviour for large @xmath44 .
so what are the expectations of the ` two - radii ' fit ?
the contribution of small raduis dominates at low @xmath44 and decreases with increasing @xmath44 @xcite .
the value of the small radius , @xmath66 , is almost independent of energy
there is a small tendency to decrease due to a larger @xmath37 in the ladder ( see , for example @xcite ) . on the other hand , the strength of the @xmath51 component increases with @xmath44 the value of the radius , @xmath51 , correlated with @xmath28 ( that is , the radius of interaction of the incoming protons ) slowly increases with energy . at asymptotic energies we expect @xmath67 .
a tendency already seen in lhc data @xcite .
therefore for @xmath68 we expect @xmath69ln@xmath70 .
however , the coefficient @xmath71 is numerically quite small . for very large @xmath44 the value or @xmath51 may additionally increase due to the final state rescattering , as was discussed above .
however it is not seen in the 7 tev @xmath72 collision data .
here we discuss the dependence of the radii , @xmath73 , on the transverse momentum of the identical pair , @xmath74 , first , we make a trivial remark for larger @xmath37 we have better space resolution in implementing the bec method . at low @xmath37
it may be hard to distinguish between @xmath51 and @xmath66 components , since the radii @xmath51 and @xmath66 will be larger than that measured at large @xmath37 due to uncertainty principle smearing , and thus closer to each other .
recall that the probability to produce two large @xmath37 pions from differents ladders ( pomerons ) decreases with increasing @xmath37 as the single particle inclusive cross section decreases steeply with @xmath37 .
that is , two identical pions , each with large @xmath37 , should be produced from the same large @xmath37 jet .
so with increasing @xmath37 of the pair we expect a large contribution from the @xmath66 component with the value of @xmath66 decreasing , reaching saturation corresponding to the jet size . to be more precise , we mean the size of pion formation due to the hadronization of large @xmath37 jets .
this tendency of @xmath58 to decrease with increasing @xmath37 was indeed observed in a ` one - radii ' fit of lhc data , see figs .
5,6 of @xcite .
however , the value of @xmath37 was not sufficiently large to see saturation in these plots .
another effect at very large @xmath37 , which may give an important contribution to large @xmath44 events , concerns the multiplicity of jets which increase as exp@xmath75 , see , for example , @xcite .
however the high @xmath76 jet cross section is too small to identify this effect in the present data .
we emphasize that the dynamics of high energy hadron interactions is based on subamplitudes of small transverse size which are distributed over the whole domain occupied by the full interaction amplitude .
thus , in bec we have to observe a small size object corresponding to the emission of both pions ( or both kaons etc . ) from a single subamplitude and a larger radius caused by events where the pions are produced from different subamplitudes . at large @xmath44 the relative contribution of the large radius component increase . enlarging the @xmath37 of the identical pion pair
we improve the space resolution of the bec analyzer .
this allows a better separation of the contributions from the small and large radii components .
when @xmath37 becomes too large ( say , @xmath77 gev ) the probability to produce such large @xmath37 pions from two different subamplitudes becomes small . in this case bec measure the radius of the ` jet ' which emits this high @xmath37 pair of identical pions .
that is , we expect the radius to decrease with increasing @xmath37 , reaching _ saturation _ for @xmath78 gev .
mgr was supported by the rscf grant 14 - 22 - 00281 .
vak thanks the leverhulme trust for an emeritus fellowship .
mgr thanks the ippp at durham university for hospitality .
r. hanbury - brown and r.w .
twiss , phil . mag . * 45 * , 663 ( 1954 ) ; proc .
242a * , 300 ( 1957 ) ; _ ibid _ * 243a * , 291 ( 1957 ) .
g. goldhaber , w.b .
fowler , s. goldhaber et al .
, phys . rev .
lett . * 3 * , 181 ( 1959 ) .
kopylov and m.i .
podgoretskii , sov .
phys . * 15 * , 219 ( 1972 ) ; * 18 * 336 ( 1973 ) .
g. alexander , rep .
phys . * 66 * , 481 ( 2003 ) .
atlas collaboration , g. aad et al .
j. * c75 * ( 2015 ) 466 . | we argue that the secondaries produced in high energy hadron collisions are emitted by small size sources distributed over a much larger area in impact parameter space occupied by the interaction amplitude .
that is , bose - einstein correlation of two emitted identical particles should be described by a ` two - radii ' parametrization ansatz .
we discuss the expected energy , charged multiplicity and transverse momentum of the pair ( that is , @xmath0 ) behaviour of both the small and large size components .
ippp/16/08 + dcpt/16/16 + + * two scales in bose - einstein correlations * + v.a .
khoze@xmath1 a.d .
martin@xmath2 , m.g .
ryskin@xmath1 and v.a .
schegelsky@xmath3 + @xmath2 institute for particle physics phenomenology , university of durham , durham , dh1 3le + @xmath4 petersburg nuclear physics institute , nrc ` kurchatov institute ' , gatchina , st .
petersburg , 188300 , russia + |
the process of hydrogen reionisation is thought to have started with ionised ( hii ) regions around the first galaxies , which later grew to surround groups of galaxies .
reionisation completed once these hii regions overlapped and occupied most of the volume between galaxies .
much recent theoretical attention has focused on the power spectrum ( ps ) of 21-cm emission from neutral hydrogen ( hi ) during the reionisation era ( zaldarriaga , furlanetto & hernquist 2004 ; furlanetto et al . 2004 ; morales et al . 2005 ; bowman et al .
in particular , the ionisation structure of the intergalactic medium ( igm ) owing to uv emission associated with star formation has been studied in detail using analytic ( furlanetto et al .
2004 ; barkana 2007 ) , numerical ( mcquinn et al . 2006 ; iliev et al .
2008 ) , and more recently , semi - numerical models ( zahn et al . 2007 ; mesinger & furlanetto 2007 ; geil & wyithe 2008 ) .
these studies describe a scenario in which very large h ii regions form around clustered sources within overdense regions of the igm .
the formation of these hii regions has a significant effect on the shape of the 21-cm ps because information on the small scale features in the density field is erased from the signal originating within the ionised regions .
conversely , the creation of large ionised regions imprints large scale features on the distribution of 21-cm intensity .
the sum of these effects is to move power from small to large scales , leaving a shoulder shaped feature on the ps at the characteristic scale of the hii regions ( furlanetto et al .
the detailed morphology of the ionisation structure will therefore yield information about both the ionising sources and the structure of absorbers in the igm on small scales ( mcquinn et al .
as reionisation leaves a strong imprint on the 21-cm ps , measuring the latter has become a key goal for learning about the reionisation epoch ( furlanetto , oh & briggs 2006 , and references therein ) .
in addition , since reionisation is driven by galaxy formation , whose statistics reflect those of the underlying density field , the detection of the redshifted 21-cm signal will not only probe the astrophysics of reionisation , but also the matter ps during the epoch of reionisation ( mcquinn et al .
2006 ; bowman , morales & hewitt 2007 ) .
finally , it has been recognised that cross - correlating galaxy surveys with 21-cm maps could yield powerful new insights into the morphology of reionisation , as well as eliminate some of the difficulties related to foreground removal ( furlanetto & lidz 2007 ; wyithe & loeb 2007 ; lidz et al 2008 ) . with these ideas as motivation ,
several experiments are currently under development that aim to detect the 21-cm signal during reionisation , including the low frequency array ( lofar ) , the murchison widefield array ( mwa ) and the precision array to probe epoch of reionizationdbacker / eor/ ] ( paper ) , and more ambitious designs are being planned such as the square kilometer array ( ska ) . up until now
only the component of hi residing in the igm has been considered in relation to forecasts of the statistical 21-cm signal . however , after the completion of reionisation there is known to be a residual hi fraction of a few percent in high density clumps which are believed to reside within galaxies ( e.g. , prochaska et al 2005 ) .
this high density contribution to the hi content of the universe is also present during the reionisation era ( we refer to this high density hi as _ galactic hi _ throughout the paper ) .
moreover , because galaxies at high redshift are biased relative to the density field , this galactic hi contribution could provide a significant perturbation to the predicted statistics of 21-cm fluctuations .
wyithe & loeb ( 2007 ) have modelled the density dependent reionisation process using a semi - analytic model that incorporates the important physical processes associated with galaxy bias and radiative feedback . in agreement with numerical simulations ( mcquinn et al .
2006 ; iliev et al .
2008 ) , this model demonstrates that galaxy bias leads to enhanced reionisation in overdense regions . in this paper
we use this model and its semi - numerical extension to explore the effect that the hi content of the biased galactic sources of reionisation would have on the fluctuations in redshifted 21-cm emission .
we begin by describing our density dependent model of reionisation in [ model ] .
we then summarise the observed evolution of the hi density in
[ hi ] , and our results for the evolution of the 21-cm signal in [ modelev ] .
we next describe the effect of galactic hi on the 21-cm signal using analytic and semi - numerical models in [ model1]-[seminum ] , and discuss prospects for detection in
[ noise ] .
we conclude in [ conclusion ] . throughout this paper
we adopt a concordance cosmology for a flat @xmath3cdm universe , @xmath4 , consistent with the constraints from komatsu et al .
all distances are in comoving units unless stated otherwise .
in regions of the igm that are overdense , galaxies will be over - abundant for two reasons : first because there is more material per unit volume to make galaxies , and second because small scale fluctuations need to be of lower amplitude to form a galaxy when embedded in a larger scale overdensity ( the so - called _ galaxy bias _ ; see mo & white 1996 ) .
regarding reionisation of the igm , the first effect will result in a larger density of ionising sources
. however this larger density will be compensated by the increased density of gas to be ionised , which also increases the recombination rate .
the process of reionisation also contains several layers of feedback .
radiative feedback heats the igm and results in the suppression of low - mass galaxy formation ( efstathiou , 1992 ; thoul & weinberg 1996 ; quinn et al .
1996 ; dijkstra et al .
such feedback effects can potentially be more intense in overdense regions , leading to weaker galaxy formation bias than might be expected from simple linear bias models ( kramer et al 2006 ) .
probing the morphology of reionisation could potentially lead to constraints on such feedback effects . to compute the relation between the local overdensity and the brightness temperature of redshifted 21-cm emission we use the model described in wyithe & loeb ( 2007 ) . here
we summarise the main features of the model , and describe the additions made for the purpose of including the possible contribution from galaxies .
the evolution of the ionisation fraction by mass @xmath5 of a particular region of igm with scale @xmath6 and overdensity @xmath7 ( at observed redshift @xmath8 ) may be written as @xmath9\\ & -&\alpha_{\rm b}cn_{\rm h}^0\left(1+\delta\frac{d(z)}{d(z_{\rm obs})}\right ) \left(1+z\right)^3q_{\delta , r},\end{aligned}\ ] ] where @xmath10 is the number of photons entering the igm per baryon in galaxies , @xmath11 is the case - b recombination coefficient , @xmath12 is the clumping factor ( which we assume , for simplicity , to be constant ) , and @xmath13 is the growth factor between redshift @xmath14 and the present time .
the production rate of ionising photons in neutral regions is assumed to be proportional to the collapsed fraction @xmath15 of mass in halos above the minimum thresholds in neutral ( @xmath16 ) , and in ionised ( @xmath17 ) regions .
we assume @xmath16 to correspond to a virial temperature of @xmath18k , representing the hydrogen cooling threshold , and @xmath19 to correspond to a virial temperature of @xmath20k , representing the mass below which infall is suppressed from an ionised igm ( dijkstra et al .
2004 ) . in a region of co - moving radius @xmath6 and
mean overdensity @xmath21 [ specified at redshift @xmath14 instead of the usual @xmath22 , the relevant collapsed fraction is obtained from the extended press - schechter ( 1974 ) model ( bond et al .
1991 ) as @xmath23 ^ 2-\left[\sigma(r)\right]^2\right)}}\right)},\ ] ] where @xmath24 is the complementary error function , @xmath25 is the variance of the density field smoothed on a scale @xmath6 , and @xmath26 is the variance of the density field smoothed on a scale @xmath27 , corresponding to a mass scale of @xmath16 or @xmath17 ( both evaluated at redshift @xmath14 rather than at @xmath28 ) . in this expression , the critical linear overdensity for the collapse of a spherical top - hat density perturbation is @xmath29 .
the model assumes that on large ( linear regime ) scales most ionising photons are absorbed locally , so that the ionisation of a region is caused by nearby ionisation sources .
this assumption is certainly justified during the early stages of reionisation , when the mean free path for ionising photons is short . however , even later in the reionisation process , the mean free path always remains smaller than the characteristic hii bubble size ( it could be smaller if mini - halos or pockets of residual hi block ionising photons between the sources and the edge of the hii region . )
our local ionisation assumption is therefore valid as long as the characteristic bubble size is smaller than the spatial scale of the correlations we consider .
this requirement is met in regimes where the fraction of regions at a particular scale are fully ionised is very low .
this point is discussed quantitatively in wyithe & morales ( 2007 ) .
following the completion of reionisation , around 2% of the hydrogen content of the universe is observed to be in hi within galaxies . moreover , this two percent fraction is observed to be roughly constant from @xmath1 to @xmath2 ( e.g. prochaska et al .
however , to estimate the effect of galactic h i on 21-cm fluctuation statistics during the reionisation epoch we must extrapolate to higher redshift .
there are two competing factors . on the one hand the collapsed fraction of mass rises as reionisation progresses , which would imply a galactic h i fraction that decreases towards high redshift . on the other hand
, the fact that the hi fraction is constant below @xmath2 suggests that the fraction of hydrogen within galaxies that is in hi increases with redshift . in this paper
we conservatively assume that the fraction of gas within galaxies that is hi increases at a rate that preserves the 2% galactic hi mass fraction ( relative to the total hydrogen content of the universe ) into the reionisation epoch . at redshifts
where there is not a sufficient collapsed fraction to maintain the 2% value , the galactic hi fraction is allowed to drop below 2% .
this is obviously a fairly simplistic model .
the actual physics of the hi content in galaxies depends both on star formation and feedback processes within the galaxy , as well as on the intergalactic uv background .
for instance , as the uv background falls toward high redshift , the hi fraction within galaxies could increase , and perhaps conspires to cancel with the decreasing collapse fraction to maintain the observed roughly constant hi fraction of @xmath30 , as seen from @xmath31 .
indeed , the fraction in protogalactic hi may be even larger than the @xmath32 we assume in the pre - reionisation epoch , particularly if there is a significant population of minihalos with @xmath33k .
modelling this quantity is difficult and beyond the scope of this paper . for
now , we note that our empirical extrapolation becomes increasingly uncertain at high redshift ; it is probably reasonable in the redshift range @xmath34 , where some of the most interesting effects ( such as the change of sign of the cross - correlation between galaxies and 21-cm emission ) take place .
to illustrate the contribution of galactic hi , we use the following example . we first find the value of @xmath10 that yields overlap of ionised regions at the mean density igm by @xmath35 ( fan et al .
2006 ; gnedin & fan 2006 ; white et al . 2003 ) , and then integrate equation ( [ history ] ) as a function of @xmath7 and @xmath6 . at a specified redshift ,
this yields the filling fraction of ionised regions within the igm on various scales @xmath6 as a function of overdensity .
we may then calculate the corresponding 21-cm brightness temperature contrast owing to hi in both the igm and galaxies @xmath36 where the pre - factor of 4/3 on the overdensity refers to the spherically averaged enhancement of the brightness temperature due to peculiar velocities in overdense regions ( bharadwaj & ali 2005 ; barkana & loeb 2005 ) , and @xmath37 is the mass averaged fraction of hi in galaxies relative to the total hydrogen content of the universe .
note that we only account for brightness temperature fluctuations due to variation in the density and ionisation fraction of the igm .
we ignore variations in the spin temperature ( which we assume to be always much greater than the cmb temperature ) due to fluctuations in heating and radiative ly@xmath38 coupling ; these are important primarily early in the reionisation process , when galaxies are in any case rare .
since the underlying probability distribution of overdensities @xmath39 is known ( a gaussian of variance @xmath40 ) , we may compute the observed probability distribution for @xmath41 , @xmath42 the second moment of this distribution corresponds to the auto - correlation function of brightness temperature smoothed on an angular radius @xmath43 : @xmath44,\end{aligned}\ ] ] where @xmath45 and @xmath46 is the variance of the density field ( at redshift @xmath14 ) smoothed on a scale @xmath6 .
the inclusion of galactic hi in the calculation of the 21-cm signal will also modify predictions for the cross - correlation of 21-cm emission with galaxies .
the cross - correlation function of galaxy overdensity and brightness temperature smoothed on an angular radius @xmath43 is @xmath47,\end{aligned}\ ] ] where @xmath48 , and @xmath49 is the galaxy bias for a halo of mass @xmath50 at redshift @xmath14 ( sheth , mo & tormen 2001 ) . to calculate the cross - correlation function we assume @xmath51 for the host mass of observed galaxies throughout this paper .
this host mass yields a density of galaxies that is comparable to the density observed in the subaru deep fields ( kashikawa et al .
2006 ) at @xmath52 ( furlanetto & lidz 2007 ) .
in this section we use our analytic model to estimate the effect of galactic hi on 21-cm fluctuations .
we assume that the star forming galaxies responsible for reionisation are the harbourers of galactic hi , and calculate the fraction of galactic hydrogen that is in hi form @xmath53 where @xmath15 and @xmath54 are computed for the mean universe ( @xmath55 , @xmath56 ) .
this enforces the condition that the fraction of hi in galaxies is either @xmath32 , or equal to the ( lower ) collapse fraction in galaxies .
we can then compute the fraction of cosmic hydrogen that is galactic hi in regions of overdensity @xmath7 and radius @xmath6 @xmath57 . \end{aligned}\ ] ] this contribution to the hi fraction of the universe can be used with equation ( [ eqt ] ) to estimate the effect of galactic hi on the statistics of 21-cm emission .
the evolution of mean brightness temperature , together with the mean hi fraction @xmath58 are plotted in figure [ fig1 ] .
the data points show the observed fraction from damped ly@xmath38 absorbers for comparison ( prochaska et al 2005 ) .
the solid dark lines show the case including the galactic hi contribution .
the thick grey line shows the igm - only model for comparison .
the residuals plotted above each panel show the magnitude of the effect when the galaxy contribution is ignored .
the inclusion of galactic hi increases the global 21-cm signal slightly as expected .
examples of auto - correlation functions are shown in figures [ fig1 ] and [ fig2 ] .
the auto - correlation function is plotted as a function of redshift at @xmath59 ( figure [ fig1 ] ) , and as a function of @xmath43 at @xmath60 ( figure [ fig2 ] ) . in figure [ fig1 ]
the residuals are plotted above each panel to show the magnitude of the error introduced when the galaxy contribution is ignored .
early in the reionisation process , before the appearance of hii regions begins to dominate the fluctuation amplitude , the inclusion of a galactic fraction enhances the 21-cm fluctuations .
however , in contrast to the mean signal , the size of fluctuations are reduced by the presence of galaxies late in the reionisation era .
this reduction can be traced to the fact that galaxies are biased towards overdense regions , while neutral igm is biased towards underdense regions which are reionised last . as a result ,
the inclusion of a galactic hi fraction reduces the intensity contrast between overdense and underdense regions .
the size of the residuals shows that within this model , the 2% galactic hi fraction reduces the fluctuation amplitude over a large fraction of the reionisation epoch . at @xmath61
, it reduces the fluctuation amplitude by around 10% at all angles ( which makes sense , since galactic
hi constitutes about @xmath62 of the total hi content of the universe at that point ) .
examples of the cross - correlation functions are shown in figure [ fig1 ] as a function of redshift at fixed angle ( @xmath63 ) , and in figure [ fig2 ] as a function of angle at fixed redshift ( @xmath60 ) .
the inclusion of galactic hi reduces the amplitude of the cross - correlation function by about 10% ( relative to the case where galactic hi is ignored ) late in the reionisation era for the reasons discussed in [ cc_bn_gand21 ] .
in addition , the cross - correlation changes sign at overlap because the hi content of the universe shifts from being dominated by the underdense regions of igm to hi in galaxies ( which reside in overdense regions ) .
this sign change will provide an unambiguous pointer to the redshift at which reionisation ends .
moreover , because the sign change will occur over a narrow frequency interval , it should provide an important check on possible sources of systematic error that could be present in measurements of the ps of intensity fluctuations which is always positive by construction .
the calculations presented in figures [ fig1 ] and [ fig2 ] assume that the galactic hi contributing to the 21-cm emission is found in the star forming systems .
however it is possible that the connection between galactic hi and star formation is not direct , either because there is hi remaining in older low mass galaxies , or because the fraction of galactic gas that is hi is host mass dependent .
as presented , our model is not able to address either of these issues .
therefore , to ascertain the possible range of influence that galactic hi might have on the 21-cm fluctuation statistics , we have assumed a characteristic mass @xmath64 for the hosts of galactic hi , and replaced the expression for @xmath65 ( equation [ eqt ] ) with @xmath66,\end{aligned}\ ] ] where we have defined the mass averaged fraction of cosmic hydrogen that is galactic hi ( @xmath67 ) .
we note that in order to maintain self consistency , the host mass of the galactic hi must correspond to a collapsed fraction [ @xmath68 $ ] that is in excess of @xmath69 . at a particular redshift
the maximum possible host mass ( @xmath70 ) corresponding to a collapsed fraction that equals @xmath69 is given by evaluation of @xmath71 the host mass at which the galaxy bias ( @xmath72 ) in equation ( [ tigm ] ) is evaluated is therefore @xmath73 with these modifications for the galactic hi distribution , we may compute the 21-cm fluctuation statistics as before using equation ( [ tvar ] ) .
the resulting auto - correlation functions are shown in figure [ fig3 ] as a function of redshift at fixed angle ( @xmath63 ) .
three values for @xmath64 are shown , @xmath74 , @xmath75 and @xmath76 .
late in the reionisation era , the amplitude of the auto - correlation function is decreased by @xmath77 ( relative to the case where galactic hi is ignored ) . below @xmath35
larger , more biased hosts of galactic hi result in a more significant increase of the fluctuation amplitude .
however at @xmath78 @xmath79 , and hence our model predicts the same evolution for different values of @xmath64 ( see equation [ minmass ] ) . the cross - correlation function is also shown in figure [ fig3 ] .
the cross - correlation changes sign at overlap as before .
more massive hosts of the galactic hi result in a cross - correlation with a larger amplitude following reionisation .
the analytic calculations described thus far provide a qualitative description of the effect of galactic hi on the statistics of 21-cm fluctuations and their association with galaxies . however , this analytic model is unable to describe fluctuations on scales comparable to the characteristic bubble scale once reionisation becomes established ( wyithe & morales 2007 ) . in the next section of this paper
we therefore investigate the statistics of 21-cm fluctuations including galactic hi in a semi - numerical model for the reionisation of a three - dimensional volume of the igm ( geil & wyithe 2008 ; zahn et al . 2007 ; mesinger & furlanetto 2007 ) .
our semi - numerical simulations are based on model - ii with respect to the galactic hi contribution .
our modelling follows the procedure outlined in geil & wyithe ( 2008 ) , and we refer the reader to that paper for details of the model .
the model employs a semi - analytic prescription for the reionisation process , which is combined with a realisation of the density field .
we construct an ionisation field based on a gaussian random field for the overdensity of mass , combined with the value of the ionised fraction @xmath5 ( equation [ history ] ) as a function of overdensity @xmath7 and smoothing scale @xmath6 .
we repeatedly filter the linear density field at logarithmic intervals on scales comparable to the box size down to the grid scale size . for all filter scales , the ionisation state of each grid position
is determined using @xmath5 and deemed to be fully ionised if @xmath80 .
all voxels within a sphere of radius @xmath6 centred on these positions are flagged and assigned @xmath81 , while the remaining non - ionised voxels are assigned an ionised fraction of @xmath82 , where @xmath83 corresponds to the smallest smoothing scale .
a voxel forms part of an hii region if @xmath84 on any scale @xmath6 . in this paper we present simulations corresponding to a linear density field of resolution @xmath85 , with a comoving side length of 512mpc .
this procedure yields an ionisation map @xmath86 as a function of position @xmath87 .
having computed the ionisation field we then find the distribution of the galactic hi component . rather than try to assign individual galaxies to the simulation we make the approximation that poisson noise will be negligible and assign a smooth density of hi @xmath88,\ ] ] where @xmath89 and @xmath90 are the mass averaged densities of hi that is collapsed in galaxies , and the total hydrogen density respectively .
the position dependent brightness temperature of the simulation box becomes @xmath91.\end{aligned}\ ] ] note that our semi - numerical model does not compute peculiar velocities , and so equation ( [ eqt ] ) does not include a peculiar velocity induced enhancement of the brightness temperature in overdense regions . as in earlier sections
we consider a model in which the mean igm is reionised at @xmath92 .
we again assume that star formation proceeds in halos above the hydrogen cooling threshold in neutral regions of igm . in ionised regions of the igm star formation
is assumed to be suppressed by radiative feedback ( see [ model ] ) . since our semi - numerical model computes the three dimensional ionisation structure in the igm , including the effect of hii regions , we can use it to compute the evolution of the variance .
the resulting auto - correlation functions are shown in the left hand panel of figure [ fig4 ] as a function of redshift at fixed angle ( @xmath63 ) .
as in the analytic model presented in figure [ fig3 ] , three values for @xmath64 are shown , @xmath93 , @xmath94 and @xmath95 .
these curves can be compared directly with the analytic approximation .
this comparison shows that while the analytic model yields the correct qualitative behaviour , it is not quantitatively correct , both in terms of the redshift and amplitude of maximum fluctuations .
however , in agreement with the analytic calculation , our semi - numerical calculations of the auto - correlation function show that the fluctuation amplitude is modified at the level of 10 - 20% ( relative to the case where galactic hi is ignored ) by the presence of a 2% galactic hi fraction throughout the reionisation era .
the semi - numerical calculations of the cross - correlation function are also shown in the right hand panel of figure [ fig4 ] as a function of redshift at fixed angle ( @xmath63 ) .
comparison with figure [ fig3 ] also shows that the analytic calculation of the cross - correlation is qualitatively correct , but does not predict the correct quantitative evolution .
figure [ fig5 ] shows 21-cm ps computed from the semi - numerical simulations at @xmath60 ( _ left _ ) and @xmath96 ( _ right _ ) .
we plot the dimensionless ps @xmath97 , where @xmath98 is the ps of 21-cm fluctuations .
both the case of the igm alone ( @xmath99 ) , and the case assuming a host mass of @xmath100 for the galactic hi ( @xmath101 ) are shown ( lower panels ) .
we also show the relative fluctuation @xmath102 of the 21-cm ps owing to galactic hi ( upper panels ) .
these figures illustrate the shoulder in the ps that corresponds to the typical bubble scale and which is due to the movement of power from small to large scales that accompanies the formation of hii regions . because the galactic hi is biased towards hii regions , this movement of power is lessened when the contribution of galactic hi is considered .
hence the ps at large scales evaluated from simulations which include hi in galaxies is lower than the ps computed using the igm alone .
the effect of galactic hi is therefore to change the shape of the 21-cm ps rather than just the amplitude .
the central and upper panels of figure [ fig6 ] show the modulus of the dimensionless 21cm - galaxy cross ps with ( @xmath103 ) and without ( @xmath104 ) galactic hi as well as the galactic hi induced fluctuation [ @xmath105 , computed from the semi - numerical simulations at @xmath60 ( _ left _ ) and @xmath96 ( _ right _ ) . again a host mass of @xmath100 is considered for the galactic hi .
an observed galaxy mass of @xmath106 is assumed . in the lower panels of figure [ fig6 ] we show the corresponding coefficient of the 21cm - galaxy cross ps .
since overdense regions , where galaxies are concentrated , are reionised first these figures show an anti - correlation on large scales , which drops in strength to zero on small scales after the formation of hii regions .
we find that the inclusion of galactic hi lessens the amplitude of the _ _ _ _ anti - correlation , since a fraction of hi is now co - located with the galaxies inside the hii regions . in addition , the cross ps changes sign on small scales which reflects the correlation of galaxies with the galactic hi inside the hii regions , where no power is contributed in 21-cm fluctuations of the igm .
thus , with the caveat that galaxies can be selected without bias from igm absorption ( which may not be true if , for instance , the galaxies are selected in ly@xmath38 emission
see lidz et al 2008 for discussion for the latter case ) , the scale at which the 21cm - galaxy cross ps changes sign could be used to probe the scale of hii regions late in the reionisation era .
before concluding , we compute the sensitivity with which the effect of galactic hi could be detected using forthcoming low - frequency arrays . to compute the sensitivity @xmath107 of a radio - interferometer to the 21-cm ps , we follow the procedure outlined by mcquinn et al .
( 2006 ) and bowman , morales & hewitt ( 2007 ) [ see also wyithe , loeb & geil ( 2008 ) ] .
the important issues are discussed below , but the reader is referred to these papers for further details . the sensitivity to the ps comprises components due to the thermal noise , and due to sample variance within the finite volume of the observations .
we consider observational parameters corresponding to the design specifications of the mwa , and of a hypothetical followup to the mwa ( termed the mwa5000 ) .
in particular the mwa is assumed to comprise a phased array of 500 tiles .
each tile contains 16 cross - dipoles to yield an effective collecting area of @xmath108 ( the area is capped for @xmath109 m ) .
the physical area of a tile is @xmath110m@xmath111 .
the tiles are distributed according to a radial antenna density of @xmath112 , within a diameter of 1.5 km and outside of a flat density core of radius 18 m .
the mwa5000 is assumed to follow the basic design of the mwa .
the quantitative differences are that the telescope is assumed to have 5000 tiles within a diameter of 2 km , with a flat density core of 80 m . in each case
we assume 1 field is observed for 1000 hr . following the work of mcquinn et al .
( 2006 ) we assume that foregrounds can be removed over @xmath113mhz bins , within a bandpass of @xmath114mhz [ foreground removal therefore imposes a minimum on the wave - number accessible of @xmath115^{-1}$]mpc@xmath116 .
the sensitivity to the 21-cm ps per mode may be written @xmath117 + p_{21}(k,\theta),\ ] ] where @xmath118 is the co - moving distance to the centre of the survey volume which has a co - moving depth @xmath119 . here
@xmath120 is the density of baselines which observe a wave vector with transverse component @xmath121 , and @xmath43 is the angle between the mode @xmath122 and the line of sight .
the thermal noise component ( first term ) is proportional to the sky temperature , where @xmath123k at the frequencies of interest .
the second term corresponds to sample variance .
the overall sensitivity is @xmath124 where @xmath125 denotes the number of modes observed in a @xmath126-space volume @xmath127 ( only modes whose line - of - sight components fit within the observed band - pass are included ) . in terms of the @xmath126-vector components @xmath126 and @xmath43 , @xmath128 where @xmath129 is the observed volume . taking the spherical average over bins of @xmath43 ,
the sensitivity to the 21-cm ps is @xmath130 ^ 2}=\sum_{\theta}\frac{1}{[\delta p_{21}(k,\theta)]^2}.\ ] ] the spherically averaged sensitivity curves for the mwa ( within bins of @xmath131 ) are plotted as the dotted lines in the lower panels of figure [ fig5 ] .
the sensitivity as a ratio of the ps computed without a galactic hi contribution ( @xmath132 ) is plotted in the upper panels of figure [ fig5 ] ( again as dotted lines ) .
these estimates illustrate that late in reionisation , the effect of the galactic hi will be at a level above the sensitivity of the first generation low - frequency arrays .
however , the low resolution of the mwa would mean that the shape - change of the ps owing to the galactic hi would not be detected until very close to overlap .
the corresponding sensitivity curves for the mwa5000 are plotted as the thin grey lines in figure [ fig5 ] .
the larger collecting area of the mwa5000 would allow the effect of galactic hi to be detected out to higher redshifts and at smaller scales . to compute the sensitivity of the mwa to the 21cm - galaxy cross ps ( @xmath133 ) we follow the discussion of furlanetto & lidz ( 2007 ) .
the sensitivity to a particular mode is @xmath134 ^ 2 = [ p_{\rm 21,g}(k,\theta)]^2 + \delta p_{21}(k,\theta)\delta p_{\rm g}(k,\theta),\ ] ] where @xmath135 is the uncertainty in the galaxy ps , and @xmath136 is the underlying mass ps at redshift @xmath14 . in the second term @xmath137
is the density of galaxies ( which we approximate as @xmath138 , where @xmath139 is the press - schechter 1976 ) mass function and @xmath50 is the halo mass of galaxies in the survey . after calculation of the total sensitivity @xmath140 and taking the spherical average over bins of @xmath43 as before , the sensitivity to the 21cm - galaxy cross ps is @xmath141 ^ 2}=\sum_{\theta}\frac{1}{[\delta p_{21,g}(k,\theta)]^2}.\ ] ] the spherically averaged sensitivity curves for the mwa ( within bins of @xmath131 ) are plotted as the dotted lines in the central panels of figure [ fig6 ] .
the sensitivity as a ratio of the fiducial igm only ps ( @xmath142 ) is also plotted in the upper panels of figure [ fig6 ] .
here we have assumed that galaxies have been detected down to a resolution limit corresponding to a host mass of @xmath51 over the full mwa field .
these sensitivity estimates illustrate that the effect of the galactic hi on the cross - correlation between galaxies and 21-cm emission could be detected with the first generation low - frequency arrays .
moreover the resolution of the mwa would be sufficient to detect the sign - change of the 21cm - galaxy cross ps at the scale of the hii regions late in reionisation .
the corresponding sensitivity curves for the mwa5000 are plotted as the thin grey lines in figure [ fig6 ] .
the larger collecting area of the mwa5000 would greatly increase the significance with which the effect of galactic hi could be measured .
in this paper we have investigated the impact of hi in galaxies on the statistics of 21-cm fluctuations using analytic and semi - numerical models .
our models are unable to self - consistently compute the galactic hi content of galaxies prior to the end of reionisation . as an input to our model
we have therefore assumed that during the reionisation era 2% of hydrogen is in the form of hi and located within galaxies .
this number is motivated by observations of the mass weighted fraction of cosmic hydrogen in hi after the end of reionisation , which is dominated by damped absorption systems and which has a constant value of @xmath0 between @xmath1 and @xmath2 .
our modelling shows that this assumption results in a reduction of 10 - 20% in the amplitude of 21-cm fluctuations over a range of spatial scales , and over a large fraction of the reionisation era .
in addition to the amplitude of 21-cm fluctuations we have also modelled the cross - correlation between galaxies and 21-cm emission .
we find that the inclusion of galaxies decreases the amplitude of the cross - correlation by a few tens of percent .
in addition , the cross - correlation between galaxies and 21-cm emission will change sign at the end of the reionisation era , providing an alternative avenue to pinpoint the end of reionisation .
we find that our analytic estimates become less applicable once hii regions become a dominant feature of the igm .
we have therefore supplemented our analytic estimates with semi - numerical modelling of the three dimensional ionisation structure of the igm .
we find that the qualitative conclusions from our analytic calculations are vindicated by this modelling .
in addition to the above estimates of auto and cross - correlation , we have used our semi - numerical model to compute 21-cm power spectra , and 21cm - galaxy cross power spectra .
the effect of galactic hi is to change the shape of the 21-cm ps rather than just the amplitude . in particular , because the galactic hi is biased towards hii regions , we find that the hii region induced _ shoulder _ in the ps is less prominent when the contribution of galactic hi is considered .
the amplitude of the 21-cm ps is modified in a scale dependent way by up to 20% when a 2% galactic hi fraction is included .
of course the galactic hi fraction is very uncertain at high redshift .
however we find that the fractional modification of the 21-cm fluctuation statistics is approximately proportional to the density of galactic hi .
therefore , if we assume a value lower(higher ) than the 2% observed at @xmath143 , then our results for the error introduced through neglect of galactic hi will be over(under)estimated . on the other hand , the change of sign in the cross - correlation between galaxies and
the 21-cm signal is robust to our assumptions .
we find that the inclusion of galactic hi lessens the amplitude of the anti - correlation between galaxies and 21-cm emission since a fraction of hi is now co - located with the galaxies inside the hii regions .
in addition , the cross ps changes sign on small scales , which reflects the correlation of galaxies with the galactic hi inside the hii regions , where no power is contributed in 21-cm fluctuations of the igm .
thus , the scale at which the 21cm - galaxy cross ps changes sign could be used to probe the scale of hii regions late in the reionisation era . we have estimated the sensitivity of the mwa to the spherically averaged 21-cm ps and 21cm - galaxy cross ps .
our calculations illustrate that the effect of the galactic hi on 21-cm fluctuations is at a level that would be significant with respect to the sensitivity of the mwa late in reionisation .
however the low resolution of the mwa would mean that the shape change of the ps owing to the galactic hi could not be detected until very close to overlap .
on the other hand , when combined with a suitable galaxy redshift survey , the resolution of the mwa would be sufficient to detect the sign change of the 21cm - galaxy cross ps at the scale of the hii regions late in reionisation . a followup telescope comprising 10 times the collecting area of the mwa would measure the effects of galactic hi with high significance . in summary
, our modelling shows that the hi content of the galaxies that reionise the universe provides a significant contribution to the statistics of 21-cm fluctuations . the galactic hi contribution to the 21-cm intensity
will therefore need to be considered in detailed modelling of the 21-cm intensity ps in order to correctly interpret measurements from the next generation of low - frequency arrays .
* acknowledgements * the research was supported by the australian research council ( jsbw ) .
lw and pmg acknowledge the support of australian postgraduate awards .
spo acknowledges support from nasa grant nng06gh95 g . | we investigate the impact of neutral hydrogen ( hi ) in galaxies on the statistics of 21-cm fluctuations using analytic and semi - numerical modelling . following the reionisation of hydrogen the hi content of the universe
is dominated by damped absorption systems ( dlas ) , with a cosmic density in h i that is observed to be constant at a level equal to @xmath0 of the cosmic baryon density from @xmath1 to @xmath2 .
we show that extrapolation of this constant fraction into the reionisation epoch results in a reduction of 10 - 20% in the amplitude of 21-cm fluctuations over a range of spatial scales .
the assumption of a different percentage during the reionisation era results in a proportional change in the 21-cm fluctuation amplitude .
we find that consideration of hi in galaxies / dlas reduces the prominence of the hii region induced _
shoulder _ in the 21-cm power spectrum ( ps ) , and hence modifies the scale dependence of 21-cm fluctuations .
we also estimate the 21cm - galaxy cross ps , and show that the cross ps changes sign on scales corresponding to the hii regions . from consideration of the sensitivity for forthcoming low - frequency arrays
we find that the effects of hi in galaxies / dlas on the statistics of 21-cm fluctuations will be significant with respect to the precision of a ps or cross ps measurement .
in addition , since overdense regions are reionised first we demonstrate that the cross - correlation between galaxies and 21-cm emission changes sign at the end of the reionisation era , providing an alternative avenue to pinpoint the end of reionisation .
the sum of our analysis indicates that the hi content of the galaxies that reionise the universe will need to be considered in detailed modelling of the 21-cm intensity ps in order to correctly interpret measurements from forthcoming low - frequency arrays .
[ firstpage ] cosmology : diffuse radiation , large scale structure , theory galaxies : high redshift , inter - galactic medium |
extending work of demazure @xcite and of bernstein , gelfand , and gelfand @xcite , in 1982 lascoux and schtzenberger @xcite defined remarkable polynomial representatives for schubert classes in the cohomology of a flag manifold , which they called schubert polynomials .
for each permutation @xmath0 in @xmath1 , there is a schubert polynomial @xmath2 $ ] . the collection of all schubert polynomials forms an additive homogeneous basis for this polynomial ring .
thus the identity @xmath3 defines integral _ structure constants _ @xmath4 for the ring of polynomials with respect to its schubert basis .
littlewood - richardson coefficients are a special case of the @xmath4 as every schur symmetric polynomial is a schubert polynomial .
the @xmath4 are positive integers : evaluating a schubert polynomial at certain chern classes gives a schubert class in the cohomology of the flag manifold .
hence , @xmath4 enumerates the flags in a suitable triple intersection of schubert varieties .
this evaluation exhibits the cohomology of the flag manifold as the quotient : @xmath5/{\langle { \frak s}_w\,|\ , w\not\in { \cal s}_n \rangle}.\ ] ] these constants , @xmath4 , are readily computed : the maple libraries ace @xcite include routines for manipulating schubert polynomials , grbner basis methods are applied in @xcite , and a new approach , using orbit values of kostant polynomials , is developed in @xcite .
however , it remains an open problem to understand these constants combinatorially . by this we mean a combinatorial interpretation or a bijective formula for these constants .
we expect such a formula will have the form @xmath6{3.6 in } ( saturated ) chains in the bruhat order on $ { \cal s}_\infty$ from \mbox{\ $ u$ to $ \,w$ } \,satisfying { \em \,some } \,condition \,imposed \,by $ \,v$ \end{minipage } } \right\}.\ ] ] the littlewood - richardson rule @xcite , which this would generalize , may be expressed in this form ( _ cf . _ [ sec : chain_description ] ) , as standard young tableaux are chains in young s lattice , a suborder of the bruhat order . a relation between chains in the bruhat order and the multiplication of schubert polynomials
has previously been noted @xcite . a new proof of the classical pieri s formula for grassmannians @xcite suggests a geometric rationale for such ` chain - theoretic ' formulas .
lastly , known formulas for these constants , particularly monk s formula @xcite , pieri - type formulas ( first stated by lascoux and schtzenberger @xcite but only recently given proofs using geometry @xcite and algebra @xcite ) , and other formulas of @xcite , are all of this form . recently ,
ciocan - fontanine @xcite has generalized these chain - theoretic pieri - type formulas to the quantum cohomology rings of manifolds of partial flags .
this has also been announced @xcite in the special case of quantum schubert polynomials @xcite . in [ sec : chains_and_orders ] , we give a refinement of ( [ eq : expected_rule ] ) .
we establish several new identities for the @xmath7 , including theorems [ thm : b ] ( _ ii _ ) and [ thm : c ] ( _ ii _ ) . one , theorem [ thm : a ] ( _ i_)(_b _ ) , gives a recursion for @xmath7 when one of the permutations @xmath8 , or @xmath9 has a fixed point and a condition on its inversions holds .
when @xmath10 is a schur polynomial , we give a chain - theoretic interpretation ( theorem [ thm : skew_shape ] ) for some @xmath7 , determine many more ( theorem [ thm : skew_permutation ] ) in terms of the classical littlewood - richardson coefficients , and show how a map that takes certain chains in the bruhat order to standard young tableaux and satisfies some additional properties would give a combinatorial interpretation for these @xmath7 ( theorem [ thm : combinatorial ] ) .
most of these identities have an order - theoretic companion which could imply them , were a description such as ( [ eq : expected_rule ] ) known .
the one identity ( theorem [ thm : d ] ) lacking such a companion yields a new result about the enumeration of chains in the bruhat order ( corollary [ cor : equal_chains ] ) . in [ sec : orders ] , we study a suborder called the @xmath11-bruhat order , which is relevant in ( [ eq : expected_rule ] ) when @xmath10 is schur symmetric polynomial in @xmath12 .
this leads to a new graded partial order on @xmath1 containing every interval in young s lattice as an induced suborder for which many group homomorphisms are order preserving ( theorem [ thm : new_order ] ) .
some of these identities require the computation of maps on the cohomology of flag manifolds induced by certain embeddings , including theorem [ thm : projection ] , lemma [ lemma : fixed_pts ] , and theorem [ thm : many_identities ] .
we use these to determine the effect of some homomorphisms of @xmath13 $ ] on its schubert basis .
for example , let @xmath14 and list the elements of @xmath15 and @xmath16 in order : @xmath17 define @xmath18\rightarrow { \bbb z}[y_1,y_2,\ldots , z_1,z_2,\ldots]$ ] by : @xmath19 then there exist integers @xmath20 such that @xmath21 we show ( theorem [ thm : substitution ] ) there exist @xmath22 ( depending upon @xmath0 and @xmath15 ) such that @xmath23 .
in particular , the coefficients @xmath20 are nonnegative .
this generalizes 1.5 of @xcite , where it is shown that the @xmath20 are non - negative when @xmath24 .
algebraic structures in the cohomology of a flag manifold also yield identities among the @xmath7 such as @xmath25 ( imposed by commutativity ) or @xmath26 , where @xmath27 , ( imposed by poincar duality among the schubert classes ) .
such ` algebraic ' identities for the classical littlewood - richardson coefficients were studied combinatorially in @xcite .
we expect the identities established here will similarly lead to some beautiful combinatorics , once a combinatorial interpretation for the @xmath7 is known .
these identities impose stringent conditions on the form of any combinatorial interpretation and should be useful in guiding the search for such an interpretation .
suppose the schubert polynomial @xmath10 in ( [ eq : structure ] ) is replaced by the schur polynomial @xmath28 .
the resulting identity @xmath29defines integer constants @xmath30 , which we call _ littlewood - richardson coefficients for schubert polynomials _ ,
as we show they share many properties with the classical littlewood - richardson coefficients .
they are related to chains in a suborder of the bruhat order called the _
@xmath11-bruhat order _ , @xmath31 .
its covers coincide with the index of summation in monk s formula @xcite : @xmath32 the sum over all @xmath33 where @xmath34 .
the set of permutations comparable to the identity in the @xmath11-bruhat order is isomorphic to young s lattice of partitions with at most @xmath11 parts .
this is the set of grassmannian permutations with descent @xmath11 , those permutations whose schubert polynomials are schur symmetric polynomials in @xmath12 . if @xmath35 is the number of standard young tableaux of shape @xmath36 , then ( * ? ? ?
* i.5 , example 2 ) , @xmath37 the sum over all @xmath36 which partition the integer @xmath38 . considering the coefficient of @xmath39 in the product @xmath40 and the definition ( 1.1.1 ) of @xmath41 , we obtain : [ prop : chains ] the number of chains in the @xmath11-bruhat order from @xmath42 to @xmath0 is @xmath43 in particular , @xmath44 unless @xmath45 . a chain - theoretic description of the constants @xmath41 should provide a bijective proof of proposition [ prop : chains ] . by this
we mean a function @xmath46 from the set of chains in @xmath47_k$ ] to the set of standard young tableaux @xmath48 whose shape is a partition of @xmath49 with the further condition that whenever @xmath48 has shape @xmath36 , then @xmath50 . for the classical littlewood - richardson coefficients ,
schensted insertion @xcite furnishes a proof @xcite ( _ cf .
_ [ sec : chain_description ] ) , as does schtzenberger s _ jeu de taquin _ @xcite . in [ sec : further ] we show ( theorem [ thm : combinatorial ] ) that if @xmath46 is a function where @xmath51 depends only upon the shape of @xmath48 and satisfies a further condition , then @xmath50 .
such a function @xmath46 would be a generalization of schensted insertion to this setting .
the @xmath11-bruhat order has a more intrinsic formulation , which we establish in [ sec : k - bruhat ] : [ thm : k - length ] let @xmath52 .
then @xmath53 if and only if 1 .
@xmath54 implies @xmath55 and @xmath56 .
2 . if @xmath57 , @xmath58 , and @xmath59 , then @xmath60 .
the @xmath11-bruhat order and its connection to the littlewood - richardson coefficients @xmath41 may be generalized , which leads to a refinement of ( [ eq : expected_rule ] ) .
a _ parabolic subgroup _
@xmath15 of @xmath1 @xcite is a subgroup generated by some adjacent transpositions , @xmath61 .
given a parabolic subgroup @xmath15 of @xmath1 , define the _ p - bruhat order _ by its covers .
a cover @xmath62 in the @xmath15-bruhat order is a cover in the bruhat order where @xmath63 .
when @xmath15 is generated by all adjacent transpositions except @xmath64 , this is the @xmath11-bruhat order .
let @xmath65 index the adjacent transpositions _ not _ in @xmath15 .
coloured chain _ in the @xmath15-bruhat order is a chain together with an element of @xmath66 for each cover @xmath67 in the chain .
this notion of colouring the bruhat order was introduced in @xcite .
iterating monk s rule , we obtain : @xmath68 where @xmath69 counts the coloured chains in the @xmath15-bruhat order from @xmath70 to @xmath71 .
this number , @xmath69 , is nonzero only for those @xmath71 which are minimal in their coset @xmath72 .
more generally , let @xmath73 count the coloured chains in the @xmath15-bruhat order from @xmath42 to @xmath0 .
multiplying ( 1.1.2 ) by @xmath74 and equating coefficients of @xmath39 , gives a generalization of proposition [ prop : chains ] : [ thm : chains ] let @xmath75 and @xmath15 be any parabolic subgroup of @xmath1 .
then @xmath76 this also shows @xmath77 unless @xmath78 , whenever @xmath71 is minimal in @xmath72 .
theorem [ thm : chains ] suggests a refinement of ( [ eq : expected_rule ] ) : let @xmath79 , and let @xmath15 be any parabolic subgroup such that @xmath71 is minimal in @xmath72 .
( there always is such a @xmath15 . ) then , for every coloured chain @xmath80 in the @xmath15-bruhat order from @xmath70 to @xmath71 , we expect that @xmath81{3.6 in } coloured chains in the $ p$-bruhat order on $ { \cal s}_\infty$ from $ u$ to $ w$ which satisfy { \em some } condition imposed by $ \gamma$ \end{minipage } } \right\}. \eqno(1.1.3)\ ] ] moreover , this rule should give a bijective proof of theorem [ thm : chains ] .
this @xmath15-bruhat order may be defined for every parabolic subgroup of every coxeter group .
likewise , the problem of finding the structure constants for a schubert basis also generalizes . for weyl groups ,
the basis is the schubert classes in the cohomology of a generalized flag manifold @xmath82 or the analogues of schubert polynomials in this case @xcite . for finite coxeter groups ,
the basis is the ` schubert classes ' of hiller @xcite in the coinvariant algebra .
likewise , theorem [ thm : chains ] and the expectation ( 1.1.3 ) have analogues in this more general setting . of the known formulas in this
setting @xcite ( see also the survey @xcite ) , few @xcite have been expressed in such a chain - theoretic manner . in [ sec : endomorphism ] and [ sec : fixed_point_identities ] , we study the @xmath4 when @xmath83 for some @xmath84 .
this leads to a formula for the substitution of 0 for @xmath85 in terms of the schubert basis , a recursion for some @xmath4 , and new identities . for @xmath86 and @xmath87 ,
let @xmath88 be defined by deleting the @xmath84th row and @xmath89th column from the permutation matrix of @xmath0 . if @xmath90 and @xmath91 , then @xmath92 is the permutation such that @xmath93 and @xmath94 .
the index of summation in a particular case of the pieri - type formula @xcite , @xmath95 defines the relation @xmath96 , which is described in more detail before theorem [ thm : projection ] .
define @xmath97 \rightarrow { \bbb z}[x_1,x_2,\ldots]$ ] by @xmath98 [ thm : a ] let @xmath52 and @xmath99 . 1 .
suppose @xmath83 and @xmath100 .
then 1 . @xmath101 \stackrel{\sim}{\longrightarrow } [ u , w]$ ] .
2 . for every @xmath102 ,
@xmath103 2 .
for every @xmath102 , @xmath104 the first statement ( lemma [ lem : expanding_bruhat ] ( _ ii _ ) ) is proven using combinatorial arguments , while the second ( theorem [ thm : coeff_sum ] ) and third ( theorem [ thm : theorem_a_iii ] ) are proven by computing certain maps on cohomology . since @xmath105 , theorem [ thm :
a ] ( _ i_)(_b _ ) gives a recursion for @xmath4 when one of @xmath8 , or @xmath9 has a fixed point and the condition on lengths is satisfied .
we also compute the effect of other substitutions of the variables in terms of the schubert basis : let @xmath14 and list the elements of @xmath15 and @xmath16 in order : @xmath17 define @xmath18\rightarrow { \bbb z}[y_1,y_2,\ldots , z_1,z_2,\ldots]$ ] by : @xmath19 in remark [ rem : i_p ] , we define an infinite set @xmath106 of permutations with the following property : [ thm : substitution ] for every @xmath107 , there exists an integer @xmath108 such that if @xmath109 and @xmath110 , then @xmath111 a precise version of theorem [ thm : substitution ] ( theorem [ thm : substitution_constants ] ) is proven in [ sec : substitution ] .
theorem [ thm : substitution ] gives infinitely many identities of the form @xmath112 for @xmath113 .
moreover , for these @xmath114 , and @xmath115 with @xmath116 , we have @xmath117\simeq [ e , u]\times[e,\,v]$ ] , which is suggestive of a chain - theoretic basis for these identities .
theorem [ thm : substitution ] extends 1.5 of @xcite , where it is shown that the @xmath118)$ ] are non - negative . a combinatorial proof of the non - negativity of these coefficients @xmath119 and of theorem [ thm : a ] ( _ ii _ ) using , perhaps , one of the combinatorial constructions of schubert polynomials @xcite may provide insight into the problem of determining the @xmath7 . theorems [ thm : a ] ( _ ii _ ) and [ thm : substitution ] enable the computation of rather general substitutions : let @xmath120 be any ( finite or infinite ) partition of @xmath121 . for @xmath122 , let @xmath123 be a set of variables in bijection with @xmath124 . define @xmath125\rightarrow { \bbb z}[\underline{x}^{(1)},\underline{x}^{(2)},\ldots]$ ] by @xmath126 [ cor : general_substitution ] for every partition @xmath127 of @xmath121 and @xmath107 , @xmath128 where each @xmath129 is a(n explicit ) sum of products of the @xmath130 , hence non - negative . a _ ballot sequence _
@xcite @xmath131 is a sequence of non - negative integers where , for each @xmath132 , @xmath133 ( traditionally , the @xmath134 .
one should consider @xmath135 as a vote for ` none of the above ' . ) given a ballot sequence @xmath136 , define @xmath137\rightarrow { \bbb z}[x_1,x_2,\ldots]$ ] by @xmath138 for every ballot sequence @xmath136 and @xmath139 , there exist non - negative integers @xmath140 for @xmath52 such that @xmath141 moreover , each @xmath140 is a(n explicit ) sum of products of the @xmath130 .
* proof . * if @xmath142 and for @xmath143 @xmath144 then @xmath145 , where @xmath146 is the diagonal map , @xmath147 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' if @xmath148 , and @xmath149 are partitions with at most @xmath11 parts then the classical littlewood - richardson coefficients @xmath150 are defined by the identity @xmath151 the @xmath152 depend only upon @xmath36 and the skew partition @xmath153 .
that is , if @xmath154 and @xmath155 are partitions with at most @xmath156 parts , and @xmath157 , then for all partitions @xmath36 , @xmath158 moreover , @xmath159 is the coefficient of @xmath160 when @xmath161 is expressed as a sum of schur polynomials .
the order type of the interval in young s lattice from @xmath162 to @xmath149 is determined by @xmath153 .
these facts generalize to the littlewood - richardson coefficients @xmath41 .
if @xmath45 , let @xmath47_k$ ] be the interval between @xmath42 and @xmath0 in the @xmath11-bruhat order , a graded poset . permutations @xmath163 and @xmath164 are _ shape equivalent _ if there exist sets of integers @xmath165 and @xmath166 , where @xmath163 ( respectively @xmath164 ) acts as the identity on @xmath167 ( respectively @xmath168 ) , and @xmath169 [ thm : b ] suppose @xmath170 and @xmath171 where @xmath172 is shape equivalent to @xmath173
. then 1 .
@xmath47_k\simeq[x , z]_l$ ] .
when @xmath174 , this isomorphism is given by @xmath175 .
2 . for all partitions @xmath36 , @xmath176 .
theorem [ thm : b ] ( _ i _ ) is a consequence of theorems [ thm : k - order ] and [ thm : new_order ] , which are proven using combinatorial arguments .
theorem [ thm : b ] ( _ ii _ ) is proven in
[ sec : proof_thm_b ] using geometric arguments . by theorem [ thm : b ] , we may define the constant @xmath177 for @xmath178 and @xmath36 a partition by @xmath179 and also define @xmath180 for any @xmath181 with @xmath182 . in [ sec : new : order ] , this analysis leads to a graded partial order @xmath183 on @xmath1 with rank function @xmath184 which has the defining property : let @xmath185_\preceq$ ] be the interval in the @xmath183-order from the identity to @xmath163 .
if @xmath182 , then the map @xmath185_\preceq \rightarrow [ u,\zeta u]_k$ ] defined by @xmath186 is an order isomorphism .
each interval in young s lattice is an induced suborder in @xmath187 rooted at the identity , as is the lattice of partitions with at most @xmath156 parts ( these are embedded differently for different @xmath156 ) .
proposition [ prop : chains ] may be stated in terms of this order : @xmath188 counts the chains in @xmath185_\preceq$ ] .
this order is studied further in @xcite , where an upper bound is given for @xmath177 . for @xmath189 ,
the map @xmath190 induces an order isomorphism @xmath185_\preceq \stackrel{\sim}{\rightarrow } [ e,\overline{\zeta}]_\preceq$ ] .
the involution @xmath191 shows @xmath192 , where @xmath193 is the conjugate or transpose of @xmath36 .
thus @xmath185_\preceq \simeq [ e,\eta]_\preceq$ ] is not sufficient to guarantee @xmath194 .
we express some of the littlewood - richardson coefficients in terms of chains in the bruhat order . if @xmath195 is a cover in the @xmath11-bruhat order , label that edge of the hasse diagram with the integer @xmath196 .
word _ of a chain in the @xmath11-bruhat order is the sequence of labels of edges in the chain .
[ thm : skew_shape ] suppose @xmath45 and @xmath172 is shape equivalent to @xmath197 , for some @xmath156 and partitions @xmath198 . then , for all partitions @xmath36 and standard young tableaux @xmath48 of shape @xmath36 , @xmath199 theorem [ thm : skew_shape ] gives a combinatorial proof of proposition [ prop : chains ] for some @xmath200 . we prove this in [ sec : chain_description ] using theorem [ thm : b ] and combinatorial arguments .
if a skew partition @xmath201 is the union of incomparable skew partitions @xmath155 and @xmath202 , then @xmath203 as graded posets . the skew schur function @xmath204 is defined ( * ? ? ?
* i.5 ) to be @xmath205 and @xmath206 ( * ? ? ?
* i.5.7 ) . thus @xmath207 permutations @xmath163 and @xmath164 are _ disjoint _ if @xmath163 and @xmath164 have disjoint supports and @xmath208 .
[ thm : c ] let @xmath163 and @xmath164 be disjoint permutations
. then 1 .
the map @xmath209 induces an isomorphism of graded posets @xmath210_\preceq\times[e,\eta]_\preceq \stackrel{\sim}{\longrightarrow } [ e,\zeta\eta]_\preceq.\ ] ] 2 . for every partition @xmath36 , @xmath211 the first statement
is proven in
[ sec : disjoint_permutations ] ( theorem [ thm : disjoint_iso ] ) , using a characterization of disjointness related to non - crossing partitions @xcite , and the second in [ sec : proof_c ] using geometry .
our last identity has no analogy with the classical littlewood - richardson coefficients .
let @xmath212 be the permutation which cyclicly permutes @xmath213 $ ] .
[ thm : d ] suppose @xmath214 and @xmath215 .
then , for every partition @xmath36 , @xmath194 .
this is proven in
[ sec : thmd ] using geometry . combined with proposition [ prop : chains ] ,
we obtain : [ cor : equal_chains ] if @xmath45 and @xmath216 with @xmath217 and @xmath218 , then the two intervals @xmath47_k$ ] and @xmath219_k$ ] each have the same number of chains .
the two intervals @xmath47_k$ ] and @xmath219_k$ ] of corollary [ cor : equal_chains ] are typically non - isomorphic : for example , in @xmath220 let @xmath221 , @xmath222 , and @xmath223 .
if @xmath224 , @xmath225 , and @xmath226 , then @xmath227 here are the intervals @xmath228_2 $ ] , @xmath229_2 $ ] , and @xmath230_2 $ ] .
@xmath231 the theorems of this section , together with the ` algebraic ' identities @xmath232 , greatly reduce the number of distinct littlewood - richardson coefficients @xmath233 which need to be determined .
we make this precise .
for @xmath36 a partition , let @xmath193 be its conjugate , or transpose .
note @xmath234 .
let @xmath36 be a partition and @xmath235 .
then @xmath236 let @xmath237 be the dihedral group with @xmath238 elements .
these identities show that the action of @xmath239 on these coefficients leaves their values invariant .
let @xmath240 be the group of permutations of @xmath213 : = \{1,2,\ldots , n\}$ ] .
let @xmath241 be the transposition interchanging @xmath242 .
the _ length _ @xmath243 of a permutation @xmath244 counts the _ inversions _ , @xmath245 , of @xmath0 .
the bruhat order @xmath246 on @xmath240 is the partial order whose cover relation is @xmath247 if @xmath248 and whenever @xmath249 , either @xmath250 , or @xmath251 .
thus @xmath252 , so the bruhat order is graded by length with minimal element the identity , @xmath70 .
if @xmath253 , let @xmath47:=\{v\,|\ , u\leq v\leq w\}$ ] be the interval between @xmath42 and @xmath0 in @xmath240 , a poset graded by @xmath254 . let @xmath255 ( or simply @xmath256 ) be defined by @xmath257 .
a permutation @xmath244 acts on @xmath258 $ ] , fixing @xmath259 .
thus @xmath260 .
define @xmath261 , the permutations of the positive integers @xmath121 fixing all but finitely many integers . for @xmath262 , define @xmath263 by requiring that @xmath264 act as the identity on @xmath16 and @xmath265 .
this injective homomorphism is a map of posets , but not of graded posets , as @xmath264 typically does not preserve length . if @xmath266 , then @xmath264 does preserve length . for this @xmath15 ,
set @xmath267 . if there exist permutations @xmath268 , and @xmath164 and sets of positive integers @xmath269 such that @xmath270 and @xmath271 , then @xmath163 and @xmath164 are _ shape equivalent_. lascoux and schtzenberger invented and then developed the elementary theory of schubert polynomials in a series of papers @xcite .
a self - contained exposition of some of this elegant theory is found in @xcite . for an interesting historical survey ,
see @xcite .
@xmath240 acts on polynomials in @xmath272 by permuting the variables . for a polynomial @xmath273 , @xmath274 is antisymmetric in @xmath275 and @xmath276 ,
hence divisible by @xmath277 .
define the _ divided difference _
operator @xmath278 if @xmath279 is a factorization of @xmath0 into adjacent transpositions with minimal length ( @xmath280 ) , then @xmath281 depends only upon @xmath0 , defining an operator @xmath282 for each @xmath244 . for @xmath283 , lascoux and schtzenberger
@xcite defined the _
_ @xmath284 by @xmath285 the degree of @xmath286 is @xmath287 , so @xmath284 is homogeneous of degree @xmath288 . since @xmath289 and @xmath290 , @xmath39 is independent of @xmath291 ( when @xmath244 ) .
this defines polynomials @xmath2 $ ] for @xmath292 . a _ partition _ @xmath36 is a decreasing sequence @xmath293 of integers .
each @xmath294 is a _ part _ of @xmath36 . for partitions @xmath36 and @xmath162 , write @xmath295 if @xmath296 for all @xmath297 . _
young s lattice _ is the set of partitions ordered by @xmath298 .
the partition with @xmath291 parts each equal to @xmath38 is written @xmath299 . for a partition @xmath36 with @xmath300 ,
the schur polynomial @xmath28 is @xmath301 @xmath28 is symmetric in @xmath12 and homogeneous of degree @xmath302 .
a permutation @xmath0 is _ grassmannian of descent @xmath11 _ if @xmath303 . a grassmannian permutation @xmath0 with descent @xmath11 defines , and is defined by , a partition @xmath36 with @xmath300 : @xmath304 ( the condition @xmath305 determines the remaining values of @xmath0 . ) in this case , write @xmath306 .
raison detre _ for this definition is that @xmath307 .
thus the schubert polynomials form a basis for @xmath13 $ ] which contains all schur symmetric polynomials @xmath28 for all @xmath36 and @xmath11 .
let @xmath308 .
@xmath309 in @xmath310 is a sequence @xmath311 of subspaces with @xmath312 .
flags @xmath309 and @xmath313 are _ opposite _ if @xmath314 for all @xmath315 .
the set of all flags is an @xmath316-dimensional complex manifold , @xmath317 ( or @xmath318 ) , called the _
flag manifold_. there is a tautological flag @xmath319 of bundles over @xmath317 whose fibre at @xmath309 is @xmath309 .
let @xmath320 be the first chern class of the line bundle @xmath321 .
borel @xcite showed the cohomology ring of @xmath317 to be @xmath322/{\langle e_i(x_1,\ldots , x_n)\,|\,i=1,\ldots , n \rangle},\ ] ] where @xmath323 is the @xmath297th elementary symmetric polynomial in @xmath324 .
given a subset @xmath325 , let @xmath326 be its linear span . for subspaces @xmath327 ,
let @xmath328 be their set - theoretic difference .
an ordered basis @xmath329 for @xmath310 determines a flag @xmath330 , where @xmath331 for @xmath332 .
a fixed flag @xmath309 gives a decomposition due to ehresmann @xcite of @xmath317 into affine cells indexed by permutations @xmath0 of @xmath333 .
the cell determined by @xmath0 has two equivalent descriptions : @xmath334 its closure is the schubert subvariety @xmath335 , which has complex codimension @xmath243 .
also , @xmath336 .
schubert class _ @xmath337 $ ] is the cohomology class poincar dual to the fundamental cycle of @xmath335 .
these schubert classes form a basis for the cohomology .
schubert polynomials were defined so that @xmath338 $ ] .
we write @xmath39 for @xmath337 $ ] .
if @xmath309 and @xmath313 are opposite flags , then @xmath339 is an irreducible , generically transverse intersection , a consequence of @xcite ( _ cf . _
@xcite ) . thus its codimension is @xmath340 , and the fundamental cycle of @xmath339 is poincar dual to @xmath341 .
since @xmath322\ \longrightarrow \ { \bbb z}[x_1,\ldots , x_{n+m}]/{\langle e_i(x_1,\ldots , x_{n+m } ) \,|\,i=1,\ldots , n+m \rangle},\ ] ] is an isomorphism on @xmath342 , identities of schubert polynomials follow from product formulas for schubert classes .
the schubert basis is self - dual for the intersection pairing : if @xmath343 , then @xmath344 let @xmath345 be the grassmannian of @xmath11-dimensional subspaces of @xmath310 , a @xmath346-dimensional manifold .
a flag @xmath309 induces a cellular decomposition indexed by partitions @xmath347 .
the closure of the cell indexed by @xmath36 is the schubert variety @xmath348 : @xmath349 the cohomology class poincar dual to the fundamental cycle of @xmath348 is @xmath28 , where @xmath12 are negative chern roots of the tautological @xmath11-plane bundle on @xmath345 .
write @xmath350 for @xmath28 , if @xmath11 is understood . as with the flag manifold , these _ schubert classes _ form a basis for cohomology , @xmath351 , and if @xmath352 are opposite flags , then @xmath353\ = \ [ \omega_\mu{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}}]\ \cdot\ [ \omega_\nu{{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}'}}]\ =
\ \sum_{\lambda\subset ( n - k)^k } c^\lambda_{\mu\,\nu}\ , s_\lambda,\ ] ] where the @xmath354 are the littlewood - richardson coefficients @xcite .
this schubert basis is self - dual : if @xmath355 , then let @xmath356 , the _ complement _ of @xmath36 , be the partition @xmath357 .
suppose @xmath358 , then @xmath359 we suppress the dependence of @xmath356 on @xmath291 and @xmath11 , which may be determined by context .
a map @xmath360 between manifolds induces a homomorphism @xmath361 of abelian groups via the functorial map on homology and the poincar duality isomorphism between homology and cohomology .
while @xmath362 is not a map of graded rings , it does satisfy the projection formula ( _ cf .
_ @xcite ) : let @xmath363 and @xmath364 , then @xmath365 for a(n oriented ) manifold @xmath366 of dimension @xmath367 , @xmath368 $ ] is generated by the class of a point .
let @xmath369 be the map which selects the coefficient of @xmath370 $ ] then @xmath371 .
let @xmath372 be defined by @xmath373 .
then @xmath374 and @xmath375 is generically one - to - one .
thus on cohomology , @xmath376 by the knneth formula , the cohomology of @xmath377 ( @xmath378 ) has an integral basis of classes @xmath379 for @xmath380 and @xmath381 .
likewise the cohomology of @xmath382 has a basis @xmath383 for @xmath384 and @xmath385 .
while we use the cohomology rings of complex varieties , our results and methods are valid for the chow rings @xcite and @xmath156-adic ( tale ) cohomology @xcite of these same varieties over any field .
the @xmath11-bruhat order , @xmath31 , is a suborder of the bruhat order on @xmath1 which is linked to the littlewood - richardson coefficients @xmath41 .
it appeared in @xcite , where it was called the @xmath11-coloured ehresmanodre .
its covers are given by the index of summation in monk s formula @xcite : @xmath386 thus @xmath0 covers @xmath42 in the @xmath11-bruhat order if @xmath0 covers @xmath42 in the bruhat order , so that @xmath387 and @xmath388 , with the additional requirement that @xmath33 .
the @xmath11-bruhat order has the following non - recursive characterization
. * theorem [ thm : k - length ] .
* _ let @xmath52 .
then @xmath45 if and only if _ 1 .
@xmath54 implies @xmath55 and @xmath56 .
2 . if @xmath242 , @xmath58 , and @xmath59 , then @xmath33 . *
* the idea is to show that the transitive relation @xmath389 defined by the conditions i and ii coincides with the @xmath11-bruhat order . if @xmath390 is a cover , then @xmath391 .
thus @xmath53 implies @xmath389 .
algorithm [ alg : chain ] , which , given @xmath389 produces a chain in the @xmath11-bruhat order from @xmath42 to @xmath0 , completes the proof .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ alg : chain ] input : permutations @xmath52 with @xmath389 .
output : a chain in the @xmath11-bruhat order from @xmath0 to @xmath42 .
output @xmath0 . while @xmath392 , do 1 .
choose @xmath393 with @xmath394 minimal subject to @xmath395 .
2 . choose @xmath396 with @xmath196 maximal subject to @xmath397 .
@xmath398 , output @xmath0 . at every iteration of 1 , @xmath389 . moreover , this algorithm terminates in @xmath49 iterations and the sequence of permutations produced is a chain in the @xmath11-bruhat order from @xmath0 to @xmath42 . *
it suffices to consider a single iteration .
we first show it is possible to choose @xmath399 and @xmath400 , then @xmath401 , and lastly @xmath402 is a cover in the @xmath11-bruhat order . in 1 , @xmath392 , so one may always choose @xmath399 .
suppose @xmath403 and it is not possible to choose @xmath400 .
in that case , if @xmath404 and @xmath405 , then also @xmath406 . similarly , if @xmath407 and @xmath405 , then @xmath408 .
thus @xmath409 , which contradicts @xmath410 .
let @xmath411 .
note that @xmath412 implies condition i for @xmath413 .
suppose @xmath414 .
. then @xmath416 and the minimality of @xmath394 shows that @xmath417 and @xmath418 .
similarly , if @xmath419 , then @xmath420 and @xmath421 .
continuing , we obtain a sequence @xmath422 with @xmath423 , a contradiction .
we show @xmath413 satisfies condition ii .
suppose @xmath424 and @xmath425 .
if @xmath407 , then @xmath426 . to show @xmath427 , it suffices to consider the case @xmath428 .
but then @xmath429 , and thus @xmath430 , by the minimality of @xmath394 . then @xmath431 .
similarly , if @xmath432 , then @xmath427 .
finally , suppose @xmath0 does not cover @xmath433 in the @xmath11-bruhat order . since @xmath59 , there exists a @xmath434 with @xmath249 and @xmath435 .
if @xmath436 , then condition ii implies @xmath437 and then the maximality of @xmath196 implies @xmath438 , a contradiction .
the case @xmath439 similarly leads to a contradiction .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' _ algorithm [ alg : chain ] depends only upon @xmath440 . _
input : a permutation @xmath178 .
output : permutations @xmath441 such that if @xmath182 , then @xmath442 is a saturated chain in the @xmath11-bruhat order .
output @xmath163 . while @xmath443 , do 1 . choose @xmath444 minimal subject to @xmath445 .
2 . choose @xmath446 maximal subject to @xmath447 .
@xmath448 , output @xmath163 .
_ to see this is equivalent to algorithm [ alg : chain ] , set @xmath449 and @xmath450 so that @xmath451 and @xmath452
. thus @xmath453 .
_ more is true , the full interval @xmath47_k$ ] depends only upon @xmath172 : [ thm : k - order ] if @xmath170 and @xmath454 with @xmath174 , then the map @xmath455 induces an isomorphism of graded posets @xmath47_k \stackrel{\sim}{\longrightarrow}[x , z]_k$ ] .
this is a consequence of the following lemma .
[ lem : k - order ] let @xmath170 and @xmath216 with @xmath174 . if @xmath456 is a cover with @xmath457 , then @xmath458 is a cover with @xmath459 * proof .
* let @xmath460 . by the position of @xmath80 in @xmath42 , we mean @xmath461 .
suppose @xmath462 does not cover @xmath463 in the @xmath11-bruhat order , so there is a @xmath80 with @xmath464 and @xmath465 .
then , in one line notation , @xmath463 and @xmath466 are as illustrated : @xmath467 since @xmath456 is a cover in the @xmath11-bruhat order , either @xmath468 or else @xmath469 .
we illustrate @xmath42 , @xmath470 , and @xmath0 for each possibility : @xmath471 assume @xmath468
. then theorem [ thm : k - length ] and @xmath472 imply @xmath473 and @xmath474 , since @xmath475 and both have positions greater than @xmath11 in @xmath476 .
let @xmath477 .
if @xmath439 , then @xmath216 implies @xmath478 so @xmath479 .
also , @xmath475 implies @xmath480 and thus @xmath481 , a contradiction .
similarly , if @xmath482 , then @xmath483 implies @xmath484 , another contradiction .
the other possibility , @xmath485 , leads to a similar contradiction .
thus @xmath458 is a cover in the @xmath11-bruhat order . to show @xmath486 , first note that the pair @xmath487 satisfy condition i of theorem [ thm : k - length ] , because @xmath472 . for condition
ii , we need only show : 1 . if @xmath464 and @xmath488 , so that @xmath489 , then @xmath490 , and 2 . if @xmath464 and @xmath491 , so that @xmath492 , then @xmath493 . if @xmath464 , then one of these two possibilities does occur , as @xmath494 is a cover in the @xmath11-bruhat order .
suppose @xmath488 , as the other case is similar .
since @xmath495 and @xmath216 , we have @xmath496 , by condition i. if @xmath485 , then @xmath497 . if @xmath498 , then @xmath479 , and so @xmath499 .
since @xmath500 , we can not have @xmath501 . ( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' define @xmath502 and @xmath503 .
[ thm : lengthenings_exist ] let @xmath178 . 1 . for @xmath504 , @xmath182 if and only if the following conditions are satisfied . 1 .
@xmath505 , 2 .
@xmath506 , and 3 . for all @xmath507 ( respectively @xmath508 ) , @xmath509 and @xmath510 together imply @xmath511 .
2 . if @xmath512 , then there is a permutation @xmath42 such that @xmath182 . * proof .
* statement ( _ i _ ) is a consequence of theorem [ thm : k - length ] . for ( _ ii _ ) , let @xmath513 contain @xmath514 and possibly some fixed points of @xmath163 , and let @xmath515 be the complementary set in @xmath121 .
suppose these sets are indexed so that @xmath516 for @xmath517 .
define @xmath504 by @xmath518 then @xmath519 is grassmannian with descent @xmath11 , and theorem [ thm : k - length ] implies @xmath520 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' for @xmath178 , define @xmath184 to be the quantity : @xmath521 if @xmath182 , then @xmath522 .
* by theorem [ thm : k - order ] , @xmath523 depends only upon @xmath163 .
computing this for the permutation @xmath42 defined in the proof of theorem [ thm : lengthenings_exist ] , shows it equals @xmath184 : if @xmath524 , then the number of inversions involving @xmath434 in @xmath42 equals the number involving @xmath434 in @xmath519 . the first term of the expression for @xmath184 counts the remaining inversions in @xmath519 and the last three terms the remaining inversions in @xmath42 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' by theorem [ thm : k - order ] , the interval @xmath228_k$ ] depends only upon @xmath163 if @xmath182 .
a closer examination of our arguments shows it is independent of @xmath11 , as well . that is , if @xmath525 , then the map @xmath526 defines an isomorphism @xmath527_k\stackrel{\sim}{\longrightarrow}[x,\,\zeta x]_l$ ]
this motivates the following definition . _ for @xmath528 , let @xmath529 if there exists @xmath504 and a positive integer @xmath11 such that @xmath530 .
if @xmath42 is chosen as in the proof of theorem [ thm : lengthenings_exist ] , then we see that @xmath531 if _ 1 . if @xmath532 , then @xmath533 , 2 .
if @xmath534 , then @xmath535 , and 3 .
if @xmath536 ( respectively , @xmath537 ) with @xmath509 and @xmath511 , then @xmath538 .
figure [ fig : new_order.s_4 ] illustrates @xmath183 on @xmath220 .
@xmath539 [ thm : new_order ] suppose @xmath540 . 1 .
@xmath187 is a graded poset with rank function @xmath184 .
2 . the map @xmath541 exhibits young s lattice of partitions with at most @xmath11 parts as an induced suborder of @xmath187 .
3 . if @xmath182 , then the map @xmath542 induces an isomorphism @xmath543_\preceq \stackrel{\sim}{\longrightarrow } [ u,\,\zeta u]_k$ ] .
if @xmath531 , then the map @xmath544 induces an isomorphism @xmath545_\preceq \stackrel{\sim}{\longrightarrow } [ e,\zeta\eta^{-1}]_\preceq$ ] .
5 . for every infinite
set @xmath14 , @xmath546 is an injection of graded posets .
thus , if @xmath547 are shape equivalent , then @xmath543_\preceq \simeq [ e,\,\eta]_\preceq$ ] . 6 .
the map @xmath548 induces an order reversing isomorphism between @xmath185_\preceq$ ] and @xmath549_\preceq$ ] .
the homomorphism @xmath550 on @xmath240 induces an automorphism of @xmath551 . theorem
[ thm : b ] ( _ i _ ) is an immediate consequence of the definition of @xmath183 and ( _ v _ ) .
* statements ( _ i_)(_v _ ) follow from the definitions
. suppose @xmath552 with @xmath553 . if @xmath554 , then @xmath555 , which proves ( _ vi _ ) .
similarly , @xmath556 implies ( _ vii _ ) .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ ex : neworder ] _ let @xmath557 and @xmath558 . then @xmath559 and @xmath560 .
figure [ fig : new_order ] illustrates the intervals @xmath561_2 $ ] , @xmath562_3 $ ] , and @xmath185_\preceq$ ] .
_ @xmath563 let @xmath564 and @xmath565 be the vertices of a convex planar @xmath291-gon numbered consecutively .
define @xmath566 , a directed geometric graph to be the union of directed chords @xmath567 for @xmath444 in the support , @xmath568 , of @xmath163 .
permutations @xmath163 and @xmath164 are _ disjoint _ if the edge sets of their geometric graphs @xmath566 and @xmath569 ( drawn on the same @xmath291-gon ) are disjoint as _ subsets of the plane_. this implies ( but is not equivalent to ) supp@xmath570supp@xmath571 .
disjointness may be rephrased in terms of partitions @xcite of @xmath213 $ ] .
suppose for simplicity , that supp@xmath572 and supp@xmath573 partition @xmath213 $ ]
. then @xmath163 and @xmath164 are disjoint if and only if there is a non - crossing partition @xcite @xmath115 of @xmath213 $ ] refining the partition ( supp@xmath572 , supp@xmath573 ) , and which is itself refined by the partition given by the cycles of @xmath163 and @xmath164 .
we compare the graphs of the pairs of permutations @xmath574 and @xmath575 .
@xmath576 the first pair is disjoint and the second is not .
we relate this definition to that given in
[ sec : schur_identities ] .
let @xmath577 .
then the edges of @xmath578 are disjoint from the edges of @xmath579 if and only if @xmath580 and @xmath164 have disjoint support and @xmath581 .
* proof . *
suppose @xmath163 and @xmath164 have disjoint support and let @xmath582 be an edge of @xmath566 and @xmath583 be an edge of @xmath569 .
consider the contribution of the endpoints of these edges to @xmath584 .
this contribution is zero if the edges do not cross , which proves the forward implication .
for the reverse , suppose these edges cross .
then the contribution is 1 if @xmath585 and @xmath586 ( or vice - versa ) , and 0 otherwise .
because each edge is part of a directed cycle in the graph , if one edge of @xmath566 crosses an edge of @xmath569 , then there are at least four crossings , one of each type illustrated .
@xmath587 here , the numbers increase in a clockwise direction , with the least number in the northeast ( @xmath588 ) .
thus @xmath589 . ( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ lem : disjoint_one ] let @xmath509 and @xmath590 and suppose @xmath591 is a cover
. then 1 .
@xmath444 and @xmath446 are connected in the geometric graph @xmath592 .
if @xmath593 is any chord of the @xmath291-gon meeting @xmath566 , then @xmath593 meets @xmath592 .
if @xmath84 and @xmath594 are connnected in @xmath566 , then they are connected in @xmath592 .
4 . if @xmath163 and @xmath164 are disjoint permutations and @xmath595 , then @xmath596 and @xmath164 are disjoint . *
suppose @xmath504 with @xmath597 .
define @xmath297 and @xmath315 by @xmath598 and @xmath599 , and set @xmath600 and @xmath601 . since @xmath602 is a cover , @xmath603 , and thus @xmath604 , since @xmath182 .
thus the edges @xmath605 and @xmath606 of @xmath592 meet , proving ( _ i _ ) . for ( _ ii _ ) , note that @xmath592 differs from @xmath566 only by the ( possible ) deletion of edges @xmath607 and @xmath608 and the addition of the edges @xmath605 and @xmath606 .
checking all possibilities for the chords @xmath593 , @xmath607 , and @xmath608 shows ( _ ii _ ) .
statement ( _ iii _ ) follows from ( _ ii _ ) by considering edges of @xmath609 .
the contrapositive of ( _ iv _ ) is also a consequence of ( _ ii _ ) ; if @xmath596 and @xmath164 are not disjoint and @xmath595 , then @xmath163 and @xmath164 are not disjoint .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ lem : disjoint_two ] suppose @xmath163 and @xmath164 are disjoint permutations .
for every @xmath504 , @xmath610 * proof .
* suppose @xmath611 .
let @xmath612 so that @xmath613 .
since @xmath163 and @xmath164 have disjoint supports , @xmath614 .
similarly , if @xmath615 , then @xmath616 , showing condition i of theorem [ thm : k - length ] holds for the pair @xmath617 . for condition
ii , suppose @xmath424 , with @xmath425 and @xmath618 . if @xmath407 , this implies @xmath619 .
since @xmath611 , and @xmath620 have disjoint supports , we have @xmath621 , which implies @xmath622 and so @xmath623 but then the edge @xmath624 of @xmath566 meets the edge @xmath625 of @xmath569 , a contradiction . the assumption that @xmath432 leads similarly to a contradiction .
thus @xmath182 and similarly , @xmath626 .
suppose now that @xmath182 and @xmath626 .
condition i of theorem [ thm : k - length ] for @xmath627 holds as @xmath163 and @xmath164 have disjoint support . for condition
ii , let @xmath424 with @xmath425 and suppose that @xmath407 .
if the set @xmath628 meets at most one of supp@xmath572 or supp@xmath573 , say supp@xmath572 , then @xmath182 implies @xmath629 .
suppose now that @xmath630 is in the support of @xmath163 and @xmath631 is in the support of @xmath164 .
since @xmath182 , we have @xmath632 . but
@xmath626 implies @xmath633 .
thus @xmath634 .
similar arguments suffice when @xmath432 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ thm : disjoint_iso ] suppose @xmath163 and @xmath164 are disjoint .
then the map @xmath185_\preceq\times [ e,\eta]_\preceq \rightarrow [ e,\zeta\eta]_\preceq$ ] defined by @xmath635 is an isomorphism of graded posets .
* by lemmas [ lem : disjoint_one ] and [ lem : disjoint_two ] , this map is an injection of graded posets . for surjectivity ,
let @xmath636 . by lemma [ lem : disjoint_one ]
( _ iii _ ) and downward induction from @xmath637 to @xmath638 , @xmath639 has no edges connecting supp@xmath572 to supp@xmath573 .
set @xmath640 , and @xmath641
. then @xmath642 and @xmath643 and @xmath644 are disjoint .
surjectivity will follow by showing @xmath645 and @xmath646 .
it suffices to consider the case @xmath647 is a cover . by lemma [ lem : disjoint_one ]
( _ i _ ) , @xmath444 and @xmath446 are connected in @xmath648 , so we may assume that @xmath649 are both in the support of @xmath163 . then @xmath650 and @xmath651 .
we show that @xmath652 is a cover , which will complete the proof .
choose @xmath504 with @xmath653 .
let @xmath654 and @xmath655 .
since @xmath643 and @xmath164 are disjoint , @xmath656 and so @xmath657 .
thus @xmath658 , showing @xmath659 , as @xmath660 .
since @xmath643 and @xmath164 are disjoint and @xmath661 , lemma [ lem : disjoint_two ] implies @xmath662 .
thus @xmath663 .
but since @xmath643 and @xmath164 are disjoint and @xmath664 is a cover , we have @xmath665 so @xmath666 .
since @xmath33 and @xmath667 , this implies @xmath668 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' _ let @xmath669 and @xmath670 , which are disjoint .
let @xmath671 .
then @xmath672 the intervals @xmath228_3 $ ] , @xmath673_3 $ ] , and @xmath674_3 $ ] are illustrated in figure [ fig : disjoint_interval ] .
_ @xmath675
for positive integers @xmath676 and @xmath107 , define @xmath677 by @xmath678 note that @xmath679 . however , if @xmath680 , then this injection , @xmath681 , is not a group homomorphism .
the map @xmath682 has a left inverse @xmath683 : for @xmath684 , define @xmath685 by @xmath686 representing permutations as matrices , the effect of @xmath687 on @xmath463 is to erase the @xmath84th row and @xmath688th column .
the effect of @xmath682 is to expand the matrix by adding a new @xmath84th row and @xmath594th column consisting mostly of zeroes , but with a 1 in the @xmath689th position .
for example , @xmath690 these maps have some order - theoretic properties . [
lem : expanding_bruhat ] suppose @xmath691 and @xmath676 are positive integers
. then 1
. @xmath692 .
2 . if @xmath693 , then @xmath694 \ \stackrel{\sim}{\longrightarrow } [ \varepsilon_{p , q}(x),\varepsilon_{p , q}(z)].\ ] ] 3 . if @xmath695 and either of @xmath84 or @xmath594 is equal to either 1 or @xmath259 , then @xmath693 .
4 . if @xmath216 and @xmath696 , then @xmath697 and @xmath219_k\simeq[x/_p , z/_p]_{k'}$ ] , where @xmath698 is equal to @xmath11 if @xmath699 and @xmath700 otherwise .
furthermore , @xmath701 .
* proof . *
suppose @xmath702 is a cover .
then @xmath703 is a cover if either @xmath704 or @xmath705 , or else @xmath706 and either @xmath707 or @xmath708 .
if however , @xmath706 and @xmath709 , then there is a chain of length 3 from @xmath710 to @xmath711 : @xmath712 the lemma follows from this observation .
for example , under the hypothesis of ( _ ii _ ) , the number of inversions in @xmath713 involving @xmath594 equals the number of inversion in @xmath710 involving @xmath594 .
thus , if @xmath714 , then @xmath715 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath716 with @xmath717 and @xmath718 .
suppose @xmath719 so that @xmath720 . for @xmath721 $ ]
define the injection @xmath722 by @xmath723 [ prop : embedding ] let @xmath724 and @xmath244 .
then , for every @xmath725 $ ] , @xmath726 recall that @xmath70 is the identity permutation .
[ cor : geometric_pushforward ] let @xmath244 and @xmath727 be opposite flags .
then @xmath728 and @xmath729 are opposite flags in @xmath317 and @xmath730 * proof .
* since @xmath731 , proposition [ prop : embedding ] with @xmath732 or @xmath259 implies @xmath733 is a subset of either intersection : @xmath734 since @xmath735 and @xmath736 are opposite flags , @xmath737 and @xmath738 are opposite flags , so both intersections are generically transverse and irreducible . since @xmath739
both intersections have the same dimension as @xmath733 , proving equality .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' since @xmath740 , where @xmath741 is the partition of @xmath741 into a single part , we see that @xmath742 , the complete symmetric polynomial of degree @xmath741 in @xmath743 .
similarly , @xmath744 , as @xmath745 , where @xmath746 is the partition of @xmath747 into @xmath747 equal parts , each of size 1 .
[ cor : equal_products ] let @xmath244 . in @xmath748 , @xmath749 and
these products are equal to @xmath750 .
we use this to compute @xmath751 .
the pieri - type formulas of @xcite show that if @xmath752 and @xmath753 positive integers , then @xmath754{\qquad\qquad } \hspace{1.35in}(4.2.1)\\ { \frak s}_u\cdot{\frak s}_{w_0w } \cdot h_{n+1-m}(x_1,\ldots , x_k ) & = & \left\{\begin{array}{ll}1&u\stackrel{r_{k , m}}{{\relbar\joinrel\longrightarrow } } w\\ 0&\mbox{otherwise } \end{array}\right . , \hspace{1.38in}(4.2.2 ) \ ] ] where @xmath755 if there is a ( saturated ) chain in the @xmath11-bruhat order from @xmath42 to @xmath0 : @xmath756 such that @xmath757 .
when @xmath758 , it follows that @xmath759 . when @xmath760 , we write @xmath761 for this relation .
similarly , @xmath762 if there is a chain in the @xmath11-bruhat order : @xmath763 such that @xmath764 .
[ thm : projection ] let @xmath765 . in @xmath766 , 1 .
2 . @xmath768 . * proof . * in @xmath766 , @xmath769 by the projection formula ( 2.3.1 ) and corollary [ cor : equal_products ]
, we have @xmath770 note that @xmath771 . by ( 4.2.1 ) , the triple product @xmath772 is zero unless @xmath773 , and in this case it equals @xmath774 .
this establishes the first equality of ( _ i _ ) . for the second , use the other formula for @xmath775 from corollary [ cor : equal_products ] and ( 4.2.2 ) . for ( _ ii _ ) , let @xmath776 be the tautological flag on @xmath777 , @xmath778 the tautological flag on @xmath318 , and 1 the trivial line bundle .
then @xmath779 since @xmath320 is the chern class of both @xmath321 and @xmath780 , we are done .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' for @xmath99 and @xmath684 , define @xmath782 [ lem : index_sets ] if @xmath783 and @xmath784 , then @xmath785 .
* proof . * if @xmath786 , @xmath784 , and @xmath787 , then @xmath86 , so @xmath788 .
but then @xmath789 and @xmath790 index the two equal sums in theorem [ thm : projection](_i _ ) .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath97\rightarrow{\bbb z}[x_1,x_2,\ldots]$ ] be defined by @xmath791 [ thm : theorem_a_iii ] for @xmath684 , and @xmath99 , @xmath792 .
* for @xmath793 , the homomorphism @xmath794 induces the map @xmath795 , by theorem [ thm : projection ] ( _ ii _ ) .
choosing @xmath291 large enough completes the proof .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' ''
[ cor : more_identities ] for @xmath796 and @xmath99 , @xmath797 * proof . *
apply @xmath794 to the identity @xmath798 to obtain : @xmath799 expanding the product @xmath800 and equating the coefficients of @xmath39 proves the identity .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' _ we illustrate the effect of @xmath801 with an example .
since @xmath802 we have @xmath803 however , @xmath804 which shows @xmath805 to see this agrees with theorem [ thm : theorem_a_iii ] , compute the permutations @xmath0 such that @xmath806 : @xmath807 of these , only the two underlined permutations are of the form @xmath808 : @xmath809 _ [ lem : restriction ] let @xmath36 be a partition and @xmath810 positive integers . then @xmath811 , where @xmath812 if @xmath813 and @xmath11 otherwise . *
* by the combinatorial definition of schur functions @xcite , @xmath814 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' lemma [ lem : restriction ] implies that @xmath815 is the only solution @xmath463 to the equation @xmath816 , a statement about chains in the bruhat order .
[ lemma : fixed_pts ] let @xmath818 with @xmath696 for some @xmath819 $ ] and suppose @xmath820 . in @xmath821 , @xmath822 * proof .
* let @xmath823 be opposite flags in @xmath824 . by proposition [ prop : embedding ]
, @xmath825 note that @xmath826 . since @xmath696 , the flags @xmath827 and @xmath828
are opposite in @xmath310 . moreover , as @xmath820 , both sides of ( 4.4.1 ) have the same dimension , so they are equal , proving the lemma .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ thm : coeff_sum ] let @xmath829 with @xmath696 and suppose that @xmath820 .
then , for every @xmath830 and positive integer @xmath84 , @xmath831 * proof
. * it suffices to compute this in @xmath821 , for @xmath291 such that @xmath784 , @xmath832 and @xmath833 . by lemma [ lemma : fixed_pts ] , @xmath834 since @xmath835 for @xmath836 and @xmath837 , @xmath838 by corollary [ cor : equal_products ] .
thus @xmath839 when @xmath840 , this has the following consequence : [ cor:1_restrict ] if @xmath841 , then @xmath842 unless @xmath843 . in that case ,
@xmath844 .
let @xmath845\choose n}$ ] , that is , @xmath846 $ ] and each has order @xmath291 .
index the sets @xmath269 and their complements @xmath847 as follows : @xmath848-p\ = \ p^c_1<\cdots
< p^c_m\\ q\ = \
q_1<\cdots < q_n & & q^c\ : = \ [ n+m]-q\ = \ q^c_1<\cdots < q^c_m \end{array}\ ] ] define a function @xmath849 by : @xmath850 as permutation matrices , @xmath851 is obtained from @xmath71 and @xmath0 by placing the entries of @xmath71 in the blocks @xmath852 and those of @xmath0 in the blocks @xmath853 . if @xmath854- \{p\}$ ] and @xmath855-\{q\}$ ] , then @xmath856 .
suppose @xmath857 , @xmath858 , and @xmath859\choose n}$ ] .
define a map @xmath860 by @xmath861 .
equivalently , if @xmath862 is a basis for @xmath310 and @xmath863 a basis for @xmath824 , then @xmath864 , where @xmath865 and @xmath866 . from this
, it follows that if @xmath867 and @xmath868 are pairs of opposite flags , then @xmath869 and @xmath870 are opposite flags in @xmath871 .
[ lem : product_subset ] let @xmath872\choose n}$ ] , @xmath873 , and @xmath874 . then , for @xmath875 and @xmath876 , @xmath877 * proof .
* for a flag @xmath878 , define @xmath879 . by the definition of @xmath880
, we have @xmath881 and @xmath882 . since @xmath883 @xmath884 , and @xmath885 , the lemma is a consequence of the definitions of schubert varieties and @xmath886 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ cor : geometry_product ] let @xmath867 and @xmath868 be pairs of opposite flags and let @xmath859\choose n}$ ] .
set @xmath887 .
then , for every @xmath873 and @xmath874 , @xmath888}(v , w)}\psi_{q}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } } ) \bigcap x_{\varepsilon_{p , q}(e , e ) } \psi_{[n]}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}\!'},{{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}'}})\\ & = & x_{\varepsilon_{p,[n]}(v , e ) } \psi_{q}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } ' } } ) \bigcap x_{\varepsilon_{p , q}(e , w ) } \psi_{[n]}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}\!'},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}})\\ & = & x_{\varepsilon_{p,[n]}(e , w)}\psi_{q}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}\!'},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } } ) \bigcap x_{\varepsilon_{p , q}(v , e ) }
\psi_{[n]}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}'}})\\ & = & x_{\varepsilon_{p,[n]}(e , e ) } \psi_{q}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}\!'},{{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } ' } } ) \bigcap x_{\varepsilon_{p , q}(v , w ) } \psi_{[n]}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}}).\end{aligned}\ ] ] * proof . *
since @xmath889 = q$ ] , @xmath890 , and @xmath891 , lemma [ lem : product_subset ] shows that @xmath892 is a subset of any of the four intersections .
equality follows as they have the same dimension .
indeed , for @xmath695 and @xmath893 , @xmath894}(x , y))&= & \ell(x)+\ell(y)+\#\{i\in[n],j\in[m]\,|\,p_i > p^c_j\}\\ \ell(\varepsilon_{p , q}(z , u))&= & \ell(z)+\ell(u)+\#\{i\in[n],j\in[m]\,|\,p^c_j > p_i\}.\end{aligned}\ ] ] thus @xmath895}(x , y ) ) + \ell(\varepsilon_{p , q}(z , u ) ) = \ell(x)+\ell(y)+\ell(z)+\ell(u ) + n\cdot m$ ] and so @xmath896}(x , y ) ) - \ell(\varepsilon_{p , q}(z , u))\ = \ { n\choose 2}+{m\choose 2}-\ell(x)-\ell(y)-\ell(z)-\ell(u).\ ] ] if @xmath897 is one of @xmath898 , then these are , respectively , the the dimension of one of the intersections and the dimension of @xmath899 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ cor : product_push_forward ] let @xmath900 $ ] .
for every @xmath901 , @xmath902 , and @xmath859\choose n}$ ] , the following identities hold in @xmath903 : @xmath904}(v , w)}\cdot { \frak s}_{\varepsilon_{p , q}(e , e ) } \quad=\quad { \frak s}_{\varepsilon_{p,[n]}(v , e ) } \cdot { \frak s}_{\varepsilon_{p , q}(e , w ) } \quad=\quad$ ] @xmath905}(e , w)}\cdot { \frak s}_{\varepsilon_{p , q}(v , e ) } \quad=\quad { \frak s}_{\varepsilon_{p,[n]}(e , e ) } \cdot { \frak s}_{\varepsilon_{p , q}(v , w)},\hspace{.8 in } $ ] and this common cohomology class is @xmath906 .
[ thm : many_identities ] let @xmath907 and @xmath859\choose n}$ ]
. then 1 .
@xmath908}(v , w)}_{\varepsilon_{p,[n]}(e , e ) \ x } \ { \frak s}_v\otimes{\frak s}_w\\ & = & \sum_{v\in{\cal s}_n,\ w\in{\cal s}_m } c^{\varepsilon_{p,[n]}(v , w_0^{(m)})}_{\varepsilon_{p,[n]}(e , w_0^{(m)}w)\ x } \ { \frak s}_v\otimes{\frak s}_w\\ & = & \sum_{v\in{\cal s}_n,\ w\in{\cal s}_m } c^{\varepsilon_{p,[n]}(w_0^{(n)},w)}_{\varepsilon_{p,[n]}(w_0^{(n)}v , e ) \ x } \ { \frak s}_v\otimes{\frak s}_w\\ & = & \sum_{v\in{\cal s}_n,\ w\in{\cal s}_m } c^{\varepsilon_{p,[n]}(w_0^{(n)},w_0^{(m)})}_{\varepsilon_{p,[n]}(w_0^{(n)}v , w_0^{(m)}w)\ x } \ { \frak s}_v\otimes{\frak s}_w\end{aligned}\ ] ] 2 .
let @xmath887 .
for every @xmath901 and @xmath874 , we have @xmath909}(v , w)}_{\varepsilon_{p,[n]}(e , e ) \ x } & = & c^{\varepsilon_{p,[n]}(v , w_0^{(m)})}_{\varepsilon_{p,[n]}(e , w_0^{(m)}w)\ x } & = & c^{\varepsilon_{p,[n]}(w_0^{(n)},w)}_{\varepsilon_{p,[n]}(w_0^{(n)}v , e ) \ x } & = & c^{\varepsilon_{p,[n]}(w_0^{(n)},w_0^{(m)})}_{\varepsilon_{p,[n]}(w_0^{(n)}v , w_0^{(m)}w)\ x } \\ c^{\varepsilon_{p , q}(v , w)}_{\varepsilon_{p , q}(e , e ) \ x } & = & c^{\varepsilon_{p , q}(v , w_0^{(m)})}_{\varepsilon_{p , q}(e , w_0^{(m)}w)\ x } & = & c^{\varepsilon_{p , q}(w_0^{(n)},w)}_{\varepsilon_{p , q}(w_0^{(n)}v , e ) \ x } & = & c^{\varepsilon_{p , q}(w_0^{(n)},w_0^{(m)})}_{\varepsilon_{p , q}(w_0^{(n)}v , w_0^{(m)}w)\ x } \end{array}\ ] ] _ each structure constant in ( _ ii _ ) is of the form @xmath910 , where @xmath163 is , respectively , @xmath911 , and @xmath912 .
each interval @xmath913 $ ] is isomorphic to @xmath914\times[e , w]$ ] .
this is consistent with the expectation that the @xmath915 should only depend upon @xmath916 $ ] and @xmath463 . _
* proof . * in ( _ ii _ ) , the second row is a consequence of the first as @xmath917 , for @xmath918 .
the first row of equalities is a consequence of the identities in ( _ i _ )
. for ( _ i _ ) , there exist integral constants @xmath919 defined by the identity @xmath920 since the schubert basis is self - dual with respect to the intersection pairing , we have @xmath921 each expression for @xmath922 of corollary [ cor : product_push_forward ] yields one of the sums in ( _ i _ ) .
for example , the last expression in corollary [ cor : product_push_forward ] yields @xmath923}(e , e ) } \cdot { \frak s}_{w_0^{(n+m)}\varepsilon_{p,[n]}(v , w)}\right)\\ & = & c^{\varepsilon_{p,[n]}(v , w)}_{\varepsilon_{p,[n]}(e , e ) \ x},\end{aligned}\ ] ] since @xmath924}(v , w ) = \varepsilon_{p , w_0^{(n+m)}[n]}\left(w_0^{(n)}v , w_0^{(m)}w\right)$ ] .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' [ cor : v_times_w ] let @xmath836 and @xmath925
. then @xmath926 . *
* choose @xmath38 so that @xmath927 . since @xmath928,[n]}(u , x)=u\times x$ ] , the first identity of theorem [ thm : many_identities ] ( _ i _ ) implies @xmath929 } { \frak s}_{u\times x } = { \frak s}_u \otimes { \frak s}_x$ ] . then @xmath930}({\frak s}_{u\times x}\cdot { \frak s}_{v\times y})\cdot ( { \frak s}_{w_0^{(n)}w } \otimes { \frak s}_{w_0^{(m)}z})\right)\\ & = & \deg\left ( { \frak s}_{u\times x}\cdot { \frak s}_{v\times y}\cdot { \frak s}_{w_0^{(n+m)}(w\times z)}\right)\\ & = & c^ { w\times z}_{u\times x\ \,v\times y},\end{aligned}\ ] ] as @xmath931})_*{\frak s}_{w_0^{(n+m)}(w\times z ) } = { \frak s}_{w_0^{(n)}w } \otimes { \frak s}_{w_0^{(m)}z}$ ] , by corollary [ cor : product_push_forward ] .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' let @xmath14 , define @xmath932 , and suppose @xmath933 is infinite .
enumerate @xmath15 and @xmath933 as follows : @xmath934 define @xmath935\rightarrow { \bbb z}[y_1,y_2,\ldots , z_1,z_2,\ldots]$ ] by @xmath936 then there exist integer constants @xmath937 for @xmath938 defined by the identity : @xmath939 for @xmath940 and @xmath941 with @xmath942 , define @xmath943)\bigcup r$ ] .
[ thm : substitution_constants ] let @xmath944 and @xmath107 . for any integers @xmath945 and @xmath367 exceeding the last descent of @xmath0 and any subset @xmath946 of @xmath947 of cardinality @xmath156 , set @xmath948 , @xmath949 , and @xmath950}(e , e)$ ] .
then @xmath951 unless @xmath752 and @xmath952 , and in that case , @xmath953 moreover , @xmath954 implies that @xmath955 $ ] exceeds the last descent of @xmath42 and @xmath956 exceeds the last descent of @xmath71 .
[ rem : i_p ] _ theorem [ thm : substitution_constants ] generalizes @xcite ( see also @xcite ) where it is shown that @xmath957)\geq 0 $ ] .
define @xmath106 to be @xmath958}(e , e)\;|\ ; l\in { \bbb n},\ n = l+\#(p\bigcap [ l]),\mbox { and } r\subset\{l+1,\ldots,3l\},\ # r = l\}.\ ] ] for @xmath244 , let @xmath108 be an integer such that @xmath959 exceeds both the last descent the length of @xmath0 . if @xmath109 with @xmath110 , then @xmath960}(e , e)$ ] for @xmath961 satisfying the conditions of theorem [ thm : substitution_constants ] and so @xmath962 for every @xmath963 , which establishes theorem [ thm : substitution ] .
_ apply the ring homomorphism @xmath964 to both sides of the product : @xmath965 if we expand this in terms of @xmath966 and equate the coefficients , we get a corollary .
[ cor : brutally_complicated ] let @xmath967 , and @xmath944 .
then there exists an integer @xmath968 such that if @xmath969 , then @xmath970 * proof of theorem [ thm : substitution_constants ] . *
first , a schubert polynomial @xmath971 $ ] if and only if @xmath972 exceeds the last descent of @xmath115 @xcite ( see also @xcite ) .
thus , @xmath973 $ ] , and if @xmath974 , then @xmath975 $ ] and @xmath976 $ ] , hence @xmath399 , respectively , @xmath400 , exceeds the last descent of @xmath42 , respectively @xmath71 . since @xmath977 , both @xmath978 and @xmath979 are at most @xmath156 .
consider the commutative diagram @xmath980 $ } \put(110,50){${\bbb z}[x_1,\ldots , x_{n+m}]$ } \put(130,5){$h^ * { \bbb f}\ell_{n+m}$ } \put(258,50){${\bbb z}[y_1,\ldots , y_n , z_1,\ldots , z_m]$ } \put(280,5){$h^*{\bbb f}\ell_n\otimes h^*{\bbb f}\ell_m$ } \put(72,54){\oval(4,4)[l]}\put(72,52){\vector(1,0){33 } } \put(86,57){$\iota$ } \put(195,52){\vector(1,0){58}}\put(215,57){$\overline{\psi_p}$ } \put(185,7){\vector(1,0){90}}\put(215,14){$\psi^*_{\overline{p}}$ } \put(150,43){\vector(0,-1){20}}\put(150,43){\vector(0,-1){25 } } \put(320,43){\vector(0,-1){20}}\put(320,43){\vector(0,-1){25 } } \end{picture}\ ] ] here , @xmath981 is the restriction of @xmath982 to @xmath983 $ ] .
the vertical arrows are injective on the module @xmath984 and its image @xmath985.\ ] ] moreover , since @xmath986=\overline{p}\bigcap [ d]$ ] , the composition , @xmath987 , of the top row coincides with @xmath988 .
since @xmath989 , the cohomological formula for @xmath990 in theorem [ thm : many_identities ] computes @xmath991 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' in the statement of the theorem [ thm : substitution_constants ] , @xmath243 could be replaced by @xmath992 .
let @xmath993 and @xmath994 be integers , @xmath308 , and @xmath995 .
define @xmath996 by @xmath997 [ thm : grassmann_product ] 1 . for every schubert class @xmath998 , @xmath999 2 . if @xmath1000 , then @xmath1001 where @xmath1002 , and @xmath1003 are defined by @xmath1004 , @xmath1005 , and @xmath1006 . _
if @xmath1007 are the chern roots of the tautological @xmath11-plane bundle over @xmath1008 , and @xmath1009 those of the tautological @xmath156-plane bundle over @xmath1010 , and @xmath1011 ( which is a symmetric polynomial in the negative chern roots of the tautological bundle over @xmath1012 ) . then @xmath1013 let @xmath1014 be the ring of symmetric functions , which is the inverse limit ( in the category of graded rings ) of the rings of symmetric polynomials in the variables @xmath1015 .
fixing @xmath36 and choosing @xmath1016 , and @xmath38 large enough gives a new proof of ( * ? ? ?
* i.5.9 ) : _ let @xmath36 be a partition and @xmath463 , @xmath1017 be infinite sets of variables .
then @xmath1018 where @xmath1019 denotes the schur function basis of the ring @xmath1020 of symmetric functions .
if we define a linear map @xmath1021 by @xmath1022 , then @xmath146 is induced by the maps @xmath1023 .
moreover , the obvious commutative diagrams of spaces give a new proof of ( * ? ? ?
* i.5.24 ) , that @xmath1020 is a cocommutative hopf algebra with comultiplication @xmath146 . *
proof of theorem [ thm : grassmann_product ] .
* the first statement is a consequence of the second : schubert classes form a basis for the cohomology ring , so there exist integral constants @xmath1024 such that @xmath1025 since the schubert basis diagonalizes the intersection pairing , @xmath1026 apply @xmath1027 and use the second assertion to obtain @xmath1028 the second assertion is a consequence of the following lemma .
[ lem : grassmann_computation ] suppose @xmath198 are partitions with @xmath1029 and @xmath1030 .
let @xmath1031 and @xmath876 and let @xmath1032 be any flag opposite to @xmath1033}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}})$ ] with @xmath1034
. then @xmath1035}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } } ) \bigcap \omega_{(m - l)^k}{{{g_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } ' } } , \eqno(4.7.1)\ ] ] where @xmath155 is the partition @xmath1036 we finish the proof of theorem [ thm : grassmann_product ] .
lemma [ lem : grassmann_computation ] implies @xmath1037}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture } } } } } ) \bigcap \omega_{(n - k)^l}{{{g_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}'}}\right]\\ & = & \sum_\lambda c^{\lambda^c}_{\rho^c\ : ( n - k)^l}\ ; s_{\lambda^c}.\end{aligned}\ ] ] since @xmath1038 , we see that @xmath1039 here , @xmath1040 is a skew partition with two components @xmath162 and @xmath149 and the last equality is a special case of ( 1.3.1 ) in [ sec : schur_identities ] .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' * proof of lemma [ lem : grassmann_computation ] .
* since @xmath1041 and @xmath1034 , we see that @xmath1042 .
it is an exercise in the definition of the schubert varieties involved and of @xmath1033}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}})$ ] to see that @xmath1043}({{e_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}},{{f\!_{{\setlength{\unitlength}{1pt}\begin{picture}(2.5,2)(1,1 ) \put(1,2){\circle*{2}}\end{picture}}}}}),\ ] ] which shows the inclusion @xmath298 in ( 4.7.1 ) .
equality follows as they have the same dimension : the intersection has dimension @xmath1044 , the dimension of @xmath1045 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' ''
combining lemma [ lem : restriction ] with theorem [ thm : coeff_sum ] , we deduce : [ lem : shape_reduction ] suppose @xmath216 and @xmath696 .
let @xmath812 if @xmath1046 and @xmath1047 otherwise .
then for all partitions @xmath36 , we have @xmath1048 note that @xmath173 and @xmath1049 are shape - equivalent , by lemma [ lem : expanding_bruhat ] ( _ iv _ ) . [
lem : skew_critical ] let @xmath1050 .
suppose @xmath1051 , @xmath170 , and @xmath1052 .
further suppose that @xmath0 is grassmannian with descent @xmath11 , the permutation @xmath172 has no fixed points , and , for @xmath1053 , @xmath1054 .
then , for all partitions @xmath36 with at most @xmath11 parts , @xmath1055 * proof of theorem [ thm : b ] ( _ ii _ ) using lemma [ lem : skew_critical ] . *
we reduce theorem [ thm : b ] ( _ ii _ ) to the special case of lemma [ lem : skew_critical ] . first
, by lemma [ lem : shape_reduction ] , it suffices to prove theorem [ thm : b ] ( _ ii _ ) when @xmath1050 , @xmath1056 , with @xmath174 and the permutation @xmath172 has no fixed points .
define @xmath1057 by @xmath1058 and set @xmath1059 .
then @xmath1060 and @xmath1061 it suffices to show separately that @xmath1062 and @xmath1063 each equal @xmath1064 .
thus we may further assume @xmath1054 for @xmath1065 or @xmath1054 for @xmath1053 .
suppose that @xmath1054 for @xmath1065 . if for @xmath873 , @xmath1066 , @xmath1067 set @xmath1068 and @xmath193 the partition conjugate to @xmath36
. then @xmath1069 , @xmath1070 , @xmath1071 , @xmath1072 , and @xmath1073 for @xmath1074 .
thus we may assume @xmath1075 for @xmath1065 .
finally , there is a grassmannian permutation @xmath1076 with descent @xmath11 and a permutation @xmath1077 such that @xmath1078 .
thus it suffices to further assume that @xmath0 is grassmannian with descent @xmath11 , the situation of lemma [ lem : skew_critical ] .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' we prove lemma [ lem : skew_critical ] by studying two intersections of schubert varieties and their image under the projection @xmath1079 .
let @xmath862 be a basis for @xmath310 and set @xmath1080 .
let @xmath1081 be the set of matrices satisfying the conditions : 1 .
@xmath1082 2 .
@xmath1083 if either @xmath1084 or else @xmath1085 .
then @xmath1086 as the only unconstrained entries of @xmath1087 are @xmath1088 when @xmath1089 and @xmath1090 , and there are @xmath243 such entries .
_ let @xmath1091 , a grassmannian permutation with descent 2 .
then @xmath1087 is the set of matrices @xmath1092 \ \raisebox{-30pt}{\rule{.5pt}{68pt}}\ , \mbox { for every } ( a , b , c , d)\in{\bbb c}\,^4\right\}\ ] ] _ fix a basis @xmath862 for @xmath310 . for @xmath1093 , and @xmath1094 , define the vector @xmath1095 . then @xmath1096 are the ` row vectors ' of the matrix @xmath444 and they form a basis for @xmath310 as @xmath444 has determinant @xmath1097 . set @xmath1098
since @xmath1099 , we see that @xmath1100 .
moreover , given @xmath1101 , restricted row reduction on a basis for @xmath735 shows there is a unique @xmath1093 with @xmath1102 . in the case that @xmath0 is grassmannian with descent @xmath11 , matrices in @xmath1087 have a simple form : if @xmath432 , then @xmath1103 . for opposite flags
@xmath352 , @xmath1104 is the class poincar dual to the fundamental cycle of @xmath1105 .
we use the projection formula ( 2.3.1 ) to compute the coefficient @xmath41 : @xmath1106 thus lemma [ lem : skew_critical ] is a consequence of the following calculation : [ lem : skew_calculation ] let @xmath1107 satisfy the hypotheses of lemma [ lem : skew_critical ] . then , if @xmath309 and @xmath313 are opposite flags in @xmath310 , @xmath1108 * proof .
* let @xmath862 be a basis for @xmath310 such that @xmath1080 and @xmath1109 , and define @xmath1087 as before .
let @xmath1110 consist of those matrices @xmath444 such that @xmath1111 .
if @xmath404 , set @xmath1112 . for @xmath407
construct @xmath1113 inductively , setting @xmath1113 to be the intersection of @xmath1114 and the affine space @xmath1115 .
since @xmath1116 and @xmath1117 , this intersection consists of a single , non - zero vector , @xmath1113 . then the algebraic map @xmath1118 gives a parameterized basis of @xmath310
moreover , @xmath1119 for all @xmath1120 , and if @xmath1121 , then @xmath1122 . note that for @xmath1120 , @xmath1123 and thus @xmath136 also parameterizes a subset of @xmath1124 .
indeed , for @xmath1121 , @xmath1125 . also for @xmath404 , we have @xmath1126 , thus @xmath1127 and @xmath1128 .
the definition of schubert cells for the flag manifold in [ sec : flag ] then implies ( 5.1.1 ) .
both cycles @xmath1105 and @xmath1129 are irreducible and have the same dimension , @xmath1130 .
since @xmath1131 if and only if @xmath1132 , the loci of flags @xmath1133 is dense in @xmath1134 . finally , for @xmath1120 , we have @xmath1135 , as @xmath1136 permutes @xmath1137 , which completes the proof .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' we show that if @xmath163 and @xmath164 are disjoint permutations and @xmath36 any partition , then @xmath1138 [ lem : long ] let @xmath1139 be disjoint permutations .
suppose @xmath1140 , @xmath1141 , @xmath1142 , and @xmath1143 .
let @xmath1144 be a permutation such that @xmath1145 .
let @xmath1146 be any element of @xmath1147-\mbox{supp}_\eta\choose n}$ ] which contains @xmath568 for which @xmath1148 $ ] .
set @xmath1149-q$ ] .
define @xmath1150 and @xmath1151 by @xmath1152 and @xmath1153 .
set @xmath1154 , @xmath1155 , and define @xmath901 and @xmath902 by @xmath1156 and @xmath1157 , where @xmath1158 then 1 .
@xmath1159 and @xmath1160 , 2 .
@xmath1161 and @xmath1162 , and 3 . for all pairs of opposite flags
@xmath1163 and @xmath1164 , + @xmath1165\ = } $ ] + @xmath1166 . * proof . *
since @xmath1167 , ( _ i _ ) follows from theorem [ thm : k - length ] and the definitions .
the second statement is also immediate . for ( _ iii _ ) , lemma [ lem : product_subset ] shows the inclusion @xmath298 .
since @xmath596 is shape equivalent to @xmath163 , @xmath1168 to @xmath164 , and @xmath163 and @xmath164 are disjoint , @xmath1169 , showing both cycles have the same dimension , and hence are equal , as @xmath1170 and @xmath1171 are opposite flags .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' note that if @xmath182 , then @xmath1172 thus the skew littlewood - richardson coefficients @xmath177 are defined by the identity in @xmath1173 : @xmath1174 * proof of theorem [ thm : c ] ( _ ii _ ) .
* we use the notation of lemma [ lem : long ] .
the following diagram commutes since @xmath1175=\{p_1,\ldots , p_k , p^c_1,\ldots , p^c_l\}$ ] .
@xmath1176 from this and lemma [ lem : long ] , we see that @xmath1177 is equal to @xmath1178 thus @xmath1179 is equal to @xmath1180 this , together with ( 5.2.1 ) , gives @xmath1181 this completes the proof , as @xmath1182 and @xmath1183 are shape equivalent pairs .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' _ let @xmath1185 with @xmath45 and @xmath171 .
suppose @xmath1186 and @xmath173 is shape equivalent to @xmath1187 , for some @xmath1188 .
for every partition @xmath36 , @xmath1189 _ * proof .
* by theorem [ thm : b ] ( _ ii _ ) , it suffices to prove a restricted case .
suppose @xmath1190 , @xmath45 , and @xmath0 is grassmannian with descent @xmath11 .
the idea is to construct permutations @xmath1191 with @xmath216 and @xmath1192 for which @xmath1193 where @xmath862 be a basis for @xmath310 and the flags @xmath1194 , and @xmath1032 are @xmath1195 then ( 5.3.1 ) implies the identity @xmath1196 , which completes the proof .
if @xmath1197 , then @xmath1198 , which is shape equivalent to @xmath172 , and the result follows by theorem [ thm : b ] ( _ ii _ ) .
assume @xmath1199 .
then @xmath1200 and @xmath1201 , as @xmath0 is grassmannian with descent @xmath11 .
set @xmath1202 , @xmath1203 , and @xmath1204 .
define @xmath783 by : @xmath1205 then @xmath1206 where @xmath1207 to show ( 5.3.1 ) , let @xmath1208 for @xmath1120 be the parameterized basis for flags @xmath1209 constructed in the proof of lemma [ lem : skew_calculation ] . since @xmath1210 , @xmath1211 , and @xmath1200
, there exist regular functions @xmath1212 on @xmath136 such that @xmath1213 since @xmath1214 and @xmath1215 , there exist regular functions @xmath1216 on @xmath136 with @xmath1217 nowhere vanishing such that @xmath1218 as @xmath1219 is the only vector among the @xmath1113 in which @xmath1220 has a non - zero coefficient .
thus @xmath1221 is a vector in @xmath1222 .
define a basis @xmath1223 for @xmath310 by @xmath1224 we claim @xmath1225 is a flag in @xmath1226 , which implies ( 5.3.1 ) : since @xmath1227 and @xmath1228 for @xmath1229 , we have @xmath1230 thus if @xmath1231 , then @xmath1232 and so @xmath1233 is a subset of the intersection @xmath1226 of dimension equal to @xmath1234 , the dimension of @xmath1226 .
thus @xmath1233 is dense , and so @xmath1235 implies ( 5.3.1 ) . for notational convenience ,
set @xmath1236 , and similarly for @xmath1237 .
to establish this claim , we first show that @xmath1238 for @xmath1239 , which shows @xmath1223 is a parameterized basis for @xmath310 and @xmath1240 .
then , for a fixed @xmath1120 , we construct @xmath1241 which satisfy @xmath1242 and @xmath1243 for @xmath1239 , showing @xmath1244 .
note that if @xmath1245 , then @xmath1246 .
thus @xmath1247 for @xmath1248 and @xmath1249 , and if @xmath1250 , then @xmath1251 . then ,
since @xmath1252 , we see that @xmath1253 . finally , since @xmath0 is grassmannian of descent @xmath11 , if @xmath1254 , then @xmath1255 , which shows @xmath1256 .
thus @xmath1257 .
we now show that @xmath1244 .
note that if @xmath1258 , then @xmath1259 .
thus if @xmath1260 , @xmath1261 . since @xmath1262
, we see that @xmath1263 .
fix @xmath1120 and set @xmath1264 for @xmath1121 .
define @xmath1265 since @xmath1266 , we see that @xmath1267 . finally , since @xmath1268 , if @xmath615 there exists a vector @xmath1269 such that @xmath1270 . for @xmath1250 ,
set @xmath1271 as as @xmath1219 is the only vector among @xmath1272 which is not in the span of @xmath1273 .
if @xmath1249 , set @xmath1274
. then @xmath1275 , completing the proof .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' ''
we give a chain - theoretic interpretation for some littlewood - richardson coefficients @xmath177 in terms chains in either the @xmath11-bruhat order or the @xmath183-order , similar to the main results of @xcite . if either @xmath1276 or @xmath591 is a cover , label that edge in the hasse diagram with the integer @xmath1277 .
given a saturated chain in the @xmath11-bruhat order from @xmath42 to @xmath519 , equivalently , a saturated @xmath183-chain from @xmath70 to @xmath163 , the _ word _ of that chain is its sequence of edge labels . given a word @xmath1278 , schensted insertion @xcite or @xcite of @xmath1279 into the empty tableau gives a pair @xmath1280 of young tableaux , where @xmath1281 is the _ insertion tableau _ and @xmath48 the _ recording tableau _ of @xmath1279 .
let @xmath295 be partitions .
a permutation @xmath163 is _ shape - equivalent _ to a skew young diagram @xmath1282 if there is a @xmath11 such that @xmath163 is shape - equivalent to @xmath1283 .
it follows that @xmath163 is shape equivalent to some skew partition @xmath1282 if and only if whenever @xmath1284 or @xmath1285 , @xmath1286 we prove a stronger version of theorem [ thm : skew_shape ] : [ thm : skew_shape_prime ] let @xmath1287 be partitions and suppose @xmath178 is shape equivalent to @xmath1282 .
then , for every partition @xmath149 1
. @xmath1288 , and 2 . for every standard young tableau @xmath48 of shape @xmath149 , @xmath1289 equivalently ,
if @xmath45 and @xmath1290 , then @xmath1291 _ theorem [ thm : skew_shape_prime ] ( _ ii _ ) gives a combinatorial proof of proposition [ prop : chains ] , when @xmath172 is shape equivalent to a skew partition .
theorem [ thm : skew_shape_prime ] ( _ ii _ ) is similar in form to theorem 8 of @xcite : _ * theorem 8 @xcite .
* suppose @xmath1292 , a partition of ` hook ' shape .
then for every @xmath52 and @xmath1293 , the constant @xmath1294 counts either set 1 .
@xmath1295[c]{chains in $ k$-bruhat order from $ u$ to $ w$ with}\\\makebox[3.1in][c]{\ word
\ $ a_1<\cdots < a_p > a_{p+1}>\cdots > a_{p+q-1}$. } \end{array}\right\ } } $ ] .
2 . @xmath1295[c]{chains in $ k$-bruhat order from $ u$ to $ w$ with}\\\makebox[3.1in][c]{word \
$ a_1>\cdots > a_q < a_{q+1}<\cdots < a_{p+q-1}$. } \end{array}\right\ } } $ ] . _ the recording tableaux of words in ( _ i _ ) each have the integers @xmath1296 in the first row and @xmath1297 in the first column
. furthermore , these are the only words with this recording tableau .
similarly , the recording tableaux of words in ( _ ii _ ) all have the integers @xmath1298 in the first column and @xmath1299 in the first row .
however , theorem [ thm : skew_shape ] is _ not _ a generalization of this result : the permutation @xmath1300 is not shape equivalent to any skew partition as @xmath1301 but @xmath1302 .
nevertheless , @xmath1303 .
interestingly , @xmath163 satisfies the conclusions of theorem [ thm : skew_shape ] . _ while the hypothesis of theorem [ thm : skew_shape ] is not necessary for the conclusion to hold , some hypotheses are necessary : let @xmath1304 , a product of two disjoint 3-cycles .
then @xmath1305 . hence , by theorem
[ thm : d ] , we have : @xmath1306 ( this may also be seen as a consequence of theorem [ thm : c ] and the form of the pieri - type formula in @xcite , or of the main result , theorem 5 , of @xcite . ) if @xmath1307 , then @xmath1308 and the labeled hasse diagram of @xmath228_2 $ ] is : @xmath1309 the labels of the six chains are : @xmath1310 and these have ( respective ) recording tableaux : @xmath1311 this list omits the tableau ( 15,15)(0,2 ) ( 0,0)(0,1)15 ( 0,0)(1,0)15(7.5,0)(0,1)15 ( 0,7.5)(1,0)15(15,0)(0,1)15 ( 0,15)(1,0)15 ( 1.5,1.5)1 ( 9,1.5)2 ( 1.5,9)3 ( 9,9)4 , and the third and fourth tableaux are identical . * proof of theorem [ thm : skew_shape ] .
* suppose first that @xmath1312 . then @xmath210_\preceq\quad \simeq \quad [ v(\mu , k),\:v(\lambda , k)]_k
\quad \simeq \quad [ \mu,\lambda]_\subset.\ ] ] the first isomorphism preserves the edge labeling of the hasse diagrams , and in the second the labels of the @xmath11-bruhat order correspond to diagonals in a young diagram : if @xmath1313 is a cover in young s lattice , there is a unique @xmath297 such that @xmath1314 . in that case @xmath1315 and the label of the corresponding edge in the @xmath11-bruhat order is @xmath1316 , the diagonal on which the new box in @xmath1317 lies .
a chain in young s lattice from @xmath162 to @xmath36 is a standard skew tableau @xmath946 of shape @xmath1282 .
consider the word , @xmath1318 , of that chain as a two - rowed array : @xmath1319 then the entry @xmath297 of @xmath946 is in the @xmath1320th diagonal .
let @xmath1281 and @xmath48 be , respectively , the insertion and recording tableaux for that two - rowed array .
consider the two - rowed array consisting of the columns @xmath1321 arranged in lexicographic order : that is , @xmath1321 is to the left of @xmath1322 if either @xmath1323 or @xmath1324 and @xmath424 . then the insertion and recording tableaux of this new array are @xmath48 and @xmath1281 , respectively @xcite .
the second row of this new array , the word inserted to obtain @xmath48 , is the ` diagonal ' word of the skew tableau @xmath946 .
that is , the entries of @xmath946 read lexicographically by diagonal . by lemma [ lem : diagonal_word ] ( proven below ) , the diagonal word is knuth - equivalent to the original word .
thus @xmath48 is the unique tableau of partition shape knuth - equivalent to @xmath946 .
this gives a combinatorial bijection @xmath1325 proving the theorem in this case , as it is well - known that ( see , for example @xcite ) , @xmath1326 now suppose @xmath163 is shape - equivalent to @xmath1283 . by theorem [ thm : b ]
( _ ii _ ) , @xmath1327 , proving ( _ i _ ) .
assume @xmath148 , and @xmath11 have been chosen so that @xmath1328 , for some @xmath15 . by theorem [ thm : new_order ] ( _ iii _ )
, @xmath264 induces an isomorphism @xmath1329_\preceq \
[ e , \,\zeta]_\preceq.\ ] ] moreover , if @xmath1330 is a cover in @xmath1331_\preceq$ ] , then @xmath1332 is a cover in @xmath185_\preceq$ ] which has label @xmath1333 , where @xmath1334 .
thus , if @xmath80 is a chain in @xmath1335_\preceq$ ] whose word @xmath1336 has recording tableau @xmath48 , then @xmath1337 is a chain in @xmath1338_\preceq$ ] with word @xmath1339 , which also has recording tableau @xmath48 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' order the diagonals of a skew young tableau @xmath946 beginning with the diagonal incident to the the end of the first column of @xmath946 .
the _ diagonal word _ of @xmath946 is the entries of @xmath946 listed in lexicographic order by diagonal , with magnitude breaking ties .
the tableau on the left below has diagonal word @xmath1340 .
if we apply schensted insertion to the initial segment @xmath1341 , ( those diagonals incident upon the first column ) , we obtain the tableau on the right , whose row word equals this initial segment .
@xmath1342 this observation is the key to the proof of the following lemma .
[ lem : diagonal_word ] the diagonal word of a skew tableau is knuth - equivalent to its column word . * proof .
* for a skew tableau @xmath946 , let @xmath1343 be its diagonal word . we show that @xmath1343 is knuth equivalent to the word @xmath1344 , the concatenation of the first column @xmath434 of @xmath946 and @xmath1345 , where @xmath1346 is obtained from @xmath946 by the removal of its first column .
an induction completes the proof .
suppose the first column of @xmath946 has length @xmath400 and @xmath946 has @xmath1347 diagonals .
for @xmath1348 let @xmath1349 be the subword of @xmath1343 consisting of the @xmath315th diagonal .
then @xmath1350 , @xmath1351 , and if @xmath1352 , then @xmath1353 , as these are consecutive entries in the @xmath11th column of @xmath946 .
consider the insertion tableau @xmath1354 of the word @xmath1355 for @xmath1356 .
then the @xmath11th column of @xmath1354 is @xmath1357 , where @xmath1358 .
hence @xmath1359 , where @xmath1360 is the row word of the tableau obtained from @xmath1361 by removing its first column , which is @xmath434 . since the column word of a tableau is knuth - equivalent to its row word , we are done . ( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' '' define the set of _ skew permutations _ to be the smallest set of permutations containing all permutations @xmath1283 which is closed under : 1 . shape equivalence . if @xmath164 is shape equivalent to a skew permutation @xmath163 , then @xmath164 is skew .
2 . cyclic shift .
if @xmath235 is skew , then so is @xmath1362 .
products of disjoint permutations . if @xmath620 are disjoint and skew , then @xmath637 is skew .
a _ shape _ of a skew permutation @xmath163 is a ( non - unique ! ) skew partition @xmath1363 which is defined inductively . if @xmath163 is shape equivalent to @xmath1282 , then @xmath163 has shape @xmath1282 . if @xmath564 is a skew permutation with shape @xmath1363 , then @xmath1362 has shape @xmath1363 . if @xmath163 and @xmath164 are disjoint skew permutations with respective shapes @xmath155 and @xmath202 , then @xmath637 has skew shape @xmath1364 . [ thm : skew_permutation ] let @xmath163 be a skew permutation with shape @xmath1363 , then 1 . for all partitions
, @xmath1365 2 .
the number of chains in the interval @xmath543_\preceq$ ] is equal to the number of standard young tableaux of shape @xmath1363 .
* proof . *
the number of standard skew tableaux of shape @xmath1363 is @xmath1366 , hence ( _ ii _ ) is consequence of ( _ i _ ) and proposition [ prop : chains ] . to show ( _ i _ ) , we need only consider the last part ( 3 . ) of the recursive definition of skew permutations , by theorems [ thm : b ] ( _ ii _ ) and [ thm : d ] .
suppose @xmath163 and @xmath164 are disjoint skew permutations with respective shapes @xmath155 and @xmath202 , and for all partitions @xmath149 , @xmath1367 and @xmath1368 .
then by theorem [ thm : c ] ( _ ii _ ) , @xmath1369{\hspace{1 in } { \setlength{\unitlength}{1.0pt}\begin{picture}(7.5,10 ) \put(0,-5){\rule{2.5pt}{5pt } } \put(0,0){\rule{5pt}{2.5pt } } \put(0,2.5){\rule{7.5pt}{2.5pt } } \end{picture}}}\end{aligned}\ ] ] [ example : disjoint ] _ consider the geometric graph of the permutation @xmath1370 : @xmath1371 thus the two cycles @xmath1372 and @xmath1373 are disjoint . _ note that @xmath163 is shape equivalent to @xmath1374 and @xmath1375 .
similarly , @xmath164 is shape equivalent to @xmath1376 and @xmath1377 .
both of these cycles , @xmath1374 and @xmath1376 , are skew partitions : let @xmath1378 ( 3,6 ) ( 0,0)(1,0)3(3,0)(0,1)3 ( 3,3)(-1,0)3(0,3)(0,-1)3 , @xmath1379 ( 6,6 ) ( 0,0)(1,0)6(6,0)(0,1)6 ( 6,6)(-1,0)6(0,6)(0,-1)6 ( 3,0)(0,1)6(0,3)(1,0)6 , @xmath1380 ( 9,6 ) ( 0,0)(0,1)6(0,0)(1,0)9 ( 3,0)(0,1)6(0,6)(1,0)6 ( 6,0)(0,1)6(0,3)(1,0)9 ( 9,0)(0,1)3
. then @xmath1381 and @xmath1382 hence , for every partition @xmath154 , @xmath1383 and @xmath1384 .
thus it follows that @xmath1385 , where @xmath155 is any of the four skew partitions : @xmath1386 for small symmetric groups , it is instructive to examine all permutations and determine to which class they belong . here , we enumerate each class in @xmath1387 , @xmath1388 , and @xmath1389 : [ cols="^ " , ] + @xmath220 & 14 & 21 & 24 + @xmath1390 & 42 & 79 & 120 + @xmath1391 & 132 & 311 & 678
+ if @xmath163 is one of the 42 permutations in @xmath1391 which are not skew permutations , and @xmath163 is not among @xmath1392 then there is a skew partition @xmath1363 such that @xmath1393 for all partitions @xmath149 .
it would be interesting to understand why this occurs for all but these 6 permutations .
is there a wider class of permutations @xmath163 such that there exists a skew partition @xmath1363 with @xmath1393 for all partitions @xmath149 ?
for the six ` exceptional ' permutations @xmath163 , there is a skew partition @xmath1363 for which @xmath1393 for all @xmath1394 , where @xmath1395up@xmath572 and @xmath1396down@xmath572 . for these , @xmath1397 . for example , let @xmath1398 . if @xmath1399 , then @xmath1400 and there are 42 chains in @xmath527_3 $ ] .
also @xmath1401 which verifies proposition [ prop : chains ] as @xmath1402 , @xmath1403 , and @xmath1402 . in this case , @xmath1404 .
since @xmath1405 and @xmath1406 , we see that @xmath1407 , however @xmath1408
. a combinatorial interpretation of the littlewood - richardson coefficients @xmath41 should also give a bijective proof of proposition [ prop : chains ] .
we show a partial converse to this , that a function @xmath46 from chains to standard young tableaux satisfying some extra conditions will provide a combinatorial interpretation of the littlewood - richardson coefficients @xmath41 .
let @xmath1409_k$ ] denote the set of ( saturated ) chains in the interval @xmath47_k$ ] .
for a partition @xmath162 and integer @xmath38 , let @xmath1410 be the set of partitions @xmath36 with @xmath1411 a horizontal strip of length @xmath38 .
these partitions arise in the classical pieri s formula : @xmath1412 if @xmath48 is a standard tableau of shape @xmath162 and @xmath38 and integer , let @xmath1413 be the set of tableaux @xmath1414 which contain @xmath48 as an initial segment such that @xmath1415 is a horizontal strip whose entries increase from left to right .
[ thm : combinatorial ] suppose that for every @xmath45 , there is a map @xmath1416_k & \longrightarrow & \left\ { \mbox{\begin{minipage}[c]{2.5 in } \begin{center } standard young tableau $ t$ whose shape is a partition of $ \ell(w)-\ell(u)$\end{center}\end{minipage}}\right\}\\ \gamma&\longmapsto & \tau(\gamma)\end{aligned}\ ] ] such that 1 .
@xmath1417_k\:|\ : \tau(\gamma)=t\}$ ] depends only upon the shape @xmath36 of the standard tableau @xmath48 .
if @xmath1418 is the concatenation of two chains @xmath1419 and @xmath1420 , then @xmath1421 is a subtableau of @xmath1422 .
( this means that @xmath1422 is a recording tableau . )
suppose @xmath1418 with @xmath1423_k$ ] , and hence @xmath1424_k$ ] .
then @xmath1425 only if @xmath1426 , and @xmath1427_k$ ] is unique for this to occur .
then , for every standard tableau @xmath48 of shape @xmath36 and @xmath45 , @xmath1428 such a map @xmath46 is a generalization of schensted insertion . in that respect , the existence of such a map would generalize theorem [ thm : skew_shape_prime ] .
* we induct on @xmath36 .
assume the theorem holds for all @xmath200 , and partitions @xmath115 either with fewer rows than @xmath36 , or if @xmath36 and @xmath115 have the same number of rows , then the last row of @xmath115 is shorter than the last row of @xmath36 .
the form of the pieri - type formulas expressed in @xcite ( also [ sec : emdedd ] ) and condition ( 3 ) prove the theorem when @xmath36 consists of a single row .
assume that @xmath36 has more than one row and set @xmath162 to be @xmath36 with its last row removed .
let @xmath38 be the length of the last row of @xmath36 and @xmath48 be any tableau of shape @xmath162 . recall that @xmath1429 gives a one - to - one correspondence between @xmath1413 and @xmath1410 . by the definition of @xmath1430
, we have @xmath1431 by the pieri formula for schubert polynomials , @xmath1432 by the classical pieri formula , this also equals @xmath1433 hence @xmath1434 we exhibit a bijection between the two sets @xmath1435_k\ : |\ : \tau(\delta)=t\}\ ] ] and @xmath1436 , where @xmath1437_k\ : |\ ; \tau(\gamma)\in
t*m \mbox { and $ \tau(\gamma)$ has shape $ \pi$}\}.\ ] ] this will complete the proof .
indeed , by the induction hypothesis @xmath1438 and for @xmath1439 with @xmath1440 , @xmath1441 thus the bijection shows @xmath1442 which is @xmath1443 , for any @xmath1414 of shape @xmath36 . to construct the desired bijection ,
consider first the map @xmath1444 defined by @xmath1445_k \mapsto \delta.\varepsilon(\delta)$ ] . by property 3 , @xmath1446
, so this injective map has the stated range . to see it is surjective ,
let @xmath1439 and @xmath1447 .
let @xmath1419 be the first @xmath1448 steps in the chain @xmath80 , so that @xmath1418 and suppose @xmath1445_k$ ] .
then @xmath1449 so @xmath1425 .
by 3 , this implies @xmath1450 , and hence @xmath1451 .
( 7.5,10 ) ( 0,-5 ) ' '' '' ( 0,0 ) ' '' '' ( 0,2.5 ) ' '' ''
these appendices are intended for informal distribution with this manuscript and will not appear in the published version .
they contain no results , only examples which we hope may illustrate some of the main results of this manuscript .
this appendix is intended to illustrate the geometric results in the previous sections , particularly of section [ sec : geometry ] .
we hope this may help others think about schubert varieties and intersections of schubert varieties . throughout ,
let @xmath862 be a fixed , ordered basis for the vector space @xmath1452 .
we use this basis to obtain a parameterization for schubert cells and their intersections .
flags are represented by @xmath1453 matrices @xmath1454 : let @xmath1455 be the ordered basis given by the ` change of basis ' matrix @xmath1454 .
the @xmath297th row of @xmath1454 gives the components of @xmath1456 .
then @xmath1454 represents the flag @xmath1457 .
we adopt some conventions for the entries of @xmath1454 : a dot ( @xmath1458 ) will denote an entry of zero and an asterix ( ) an entry which may assume any value in @xmath1459 .
one last convention is that the flags @xmath1460 , etc . will always be defined to be @xmath1461 and the ` primed ' flags @xmath1462 , etc
. which are opposite to their unprimed cousins will be defined by @xmath1463 .
we refer to these as the _ standard flags_. in theorem [ thm : b ] ( _ ii _ ) , we had @xmath45 , @xmath216 , and @xmath1464 and we studied @xmath1465 and @xmath1466 .
the main result was that , in @xmath1467 , @xmath1468 the general case of theorem [ thm : b ] ( _ ii _ ) was reduced to lemma [ lem : skew_critical ] , where @xmath0 was grassmannian of descent @xmath11 , and @xmath1469 ( and hence also @xmath1470 ) .
the first example illustrates this case .
let @xmath1471 , @xmath1472 , and @xmath1473 the following matrices respectively represent general flags in the schubert cells @xmath1474 , @xmath1475 , @xmath1476 , and @xmath1477 : @xmath1478 @xmath0 is chosen to be grassmannian so that the cell @xmath1474 has a particularly simple form , which gave an easy parameterization for the intersection of the two cells , @xmath1479 . in the proof of lemma [ lem : skew_calculation ]
we describe how to find bases parameterized by @xmath1480 . in practice , this method may be used to determine the subvariety @xmath136 of @xmath1087 .
first , let @xmath1481 be the rows of the following matrix , where @xmath1482 , and @xmath46 are arbitrary elements of @xmath1459 with @xmath1483 : @xmath1484 these parameters were chosen so that for each @xmath1485 , @xmath1486 and does not lie in either of @xmath1487 or @xmath1488 , hence the 1 s , the condition on @xmath1489 , and the 0 s ( @xmath1458 ) in their initial columns .
this matrix determines a flag @xmath1490 , since it is in @xmath1087 .
also , since @xmath1491 for @xmath407 , at least @xmath1492 satisfy the conditions for the flag @xmath878 to be in @xmath1475 .
the remaining conditions for @xmath1493 , @xmath1494 impose additional restrictions on the parameters . in practice
this means we seek conditions to ensure that @xmath1495 is non - empty .
for instance , for @xmath1496 , since @xmath1497 and @xmath1498 , some cancellation must occur .
indeed , since @xmath1499 we must have @xmath1500 in order that @xmath1501 . in the general situation
, more complicated determinantal conditions may arise . from these considerations ,
we arrive at a parameterization for @xmath1502 .
@xmath1503 for @xmath1504 , both matrices represent the same flag in the intersection . to see this ,
let @xmath1481 be the basis determined by the first matrix , and @xmath1505 the basis determined by the second matrix .
then , by the definition ( [ eq : schubert_cell_definition ] ) of schubert cells in [ sec : flag ] , the flags @xmath1506 and @xmath1507 . since @xmath1508 for @xmath1509 , @xmath1510,\end{aligned}\ ] ] showing @xmath1511 .
lastly , since @xmath1512 , and @xmath823 are opposite flags , we see that @xmath1513 is irreducible of dimension 7 .
this identifies a 7-parameter family of flags in this intersection , which must be dense .
similarly , ( with the same restrictions on parameters ) , the two matrices below both represent the same flag in @xmath1514 : @xmath1515 if @xmath1516 is the basis determined by the first matrix , then @xmath1517 , @xmath1518 , @xmath1519 , and @xmath1520 .
thus @xmath1521 , which proves @xmath1522 this is true even when @xmath1107 do not satisfy the extra hypotheses of lemma [ lem : skew_critical ] ( one may construct a proof using the geometric analogs of the arguments that reduce theorem [ thm : b ] ( _ ii _ ) to lemma [ lem : skew_critical ] ) . we illustrate this on another example . here
, let @xmath1471 , @xmath1523 , and @xmath1524 then the following four matrices represent , respectively , the schubert cells @xmath1474 , @xmath1475 , @xmath1476 , and @xmath1477 : @xmath1525 consider the ( equivalent pairs of ) parameterizations for flags in the intersections of the cells , @xmath1479 ( the left - hand pair ) , and @xmath1514 ( the right - hand pair ) : @xmath1526 to see that each pair of matrices does indeed give the same flag , let @xmath1481 , @xmath1505 , @xmath1516 , and @xmath1527 be the bases given by the four matrices ( read left - to - right ) . then , for @xmath1528 , @xmath1508 and @xmath1529 . also , @xmath1530 and @xmath1531 thus , @xmath1511 and @xmath1532 .
as before , these parameterized bases give dense subsets of flags in each of @xmath1502 and @xmath1466 .
moreover , since @xmath1533 , we see that @xmath1534 we complete example [ example : disjoint ] , giving the geometric side of the story .
the permutation @xmath1370 is the disjoint product of @xmath1372 and @xmath1373 .
note that @xmath1535 .
let @xmath878 and @xmath1032 be the standard flags in @xmath1536 .
the following matrices parameterize the schubert cells @xmath1537 and @xmath1538 : @xmath1539 as before , here are two parameterized matrices , each of which give bases defining the same flag in the intersection of the two cells : @xmath1540 it is clearer to display two matrices ` on top of each other ' : @xmath1541 the vertical lines in the last 5 rows illustrate that the left- and right - sides of those rows come from different ( equivalent ) bases .
the shading accentuates its ` block form ' : let @xmath1542 and @xmath1543 .
the @xmath1544 , where @xmath1545 .
then the shaded regions are @xmath1546 .
we see that @xmath1547 and @xmath1548 are uniquely defined by @xmath1549 and @xmath1550 .
moreover , we may define permutations @xmath71 and @xmath0 as in lemma [ lem : long ] ; let @xmath1551 and @xmath1552
. then 1 .
@xmath1553 and @xmath1554 .
2 . @xmath1555 and @xmath1556 . for the last part of lemma [ lem : long ] ,
let @xmath1557 be the standard flags in @xmath1558 , and @xmath1559 the standard flags in @xmath1560 .
then the following four matrices parameterize the schubert cells @xmath1561 , @xmath1562 , @xmath1563 , and @xmath1564 : @xmath1565 then the following two matrices parameterize the two intersections .
again , we have drawn two matrices on top of each other .
@xmath1566 next , note that @xmath1567 and @xmath1568 . finally , verifying that @xmath1569\ ] ] is equal to @xmath1570 may be done by comparing these parameterized matrices . in the final part of the proof of theorem [ thm : c ] ( _ ii _ ) , we compare images of these intersections under projections to grassmannians .
the row spans of the next three parameterized matrices represent @xmath1571 , @xmath1572 , and @xmath1573 in each of @xmath1574 , @xmath1575 , and @xmath1576 , respectively .
@xmath1577 thus @xmath1578 is equal to @xmath1579 consider the images of @xmath1580 and @xmath1581 under @xmath886 in @xmath1582 : @xmath1583 this should be compared with the first figure of this section , which shows the cells @xmath1537 and @xmath1538 . here
, the circles ( ( 6,6)(3,3 ) ) indicate the ` surprise ' entries of 0 ; those which are not zero in that first figure .
this illustrates the two inclusions , and serves to illustrate lemma [ lem : product_subset ] : @xmath1584 we illustrate ` cyclic shift ' .
let @xmath1585 and @xmath1586 so that @xmath1587 .
define @xmath1588 as in the proof of theorem [ thm : d]@xmath1184 ( [ sec : thmd ] ) to be @xmath1589 and @xmath1590 . then @xmath1591 . here ,
@xmath1592 , @xmath1593 , and @xmath1594 .
let @xmath1557 be the standard flags for @xmath1560 .
we define @xmath1595 and @xmath1596 .
then , with respect to these flags , the schubert cells @xmath1597 , @xmath1598 , @xmath1599 , and @xmath1600 are : @xmath1601 here , since the flags are different , the columns of the matrices on the right correspond to different elements of our fixed basis , @xmath1602 , as indicated . here
are two matrices giving ( equivalent ) parameterized bases for flags in the intersection of the cells @xmath1603 : @xmath1604 here @xmath1605 , showing that @xmath1606 parameterizes the set @xmath136 of the intersection of cells .
let @xmath1607 be the basis given by the left matrix and @xmath1608 the basis given by the right matrix . since @xmath1609 @xmath1610 are regular functions on @xmath136 .
also , since @xmath1611 @xmath1612 , @xmath1613 , and @xmath1614 are regular functions on @xmath136 with @xmath1615 nowhere vanishing .
the bases @xmath1616 and @xmath1617 defined in the proof of theorem [ thm : d]@xmath1184 , are parameterized by the following two matrices : @xmath1618 these matrices give equivalent bases , and hence define the same flag .
also , comparing them to the rightmost two matrices in the first figure of this subsection , shows this flag is in the intersection @xmath1619 .
since the first two rows of each matrix have the same span , @xmath1620 which is the main geometric result needed to deduce theorem [ thm : d]@xmath1184 .
the bruhat order is one of our main objects of study in this paper .
here is a picture of the ( full ) bruhat order and the 2-bruhat order on @xmath220 .
@xmath1621 for comparison , here is the @xmath183-order on @xmath220 ( reproduced from [ sec : new : order ] ) . @xmath1622
theorem [ thm : chains ] describes the relation between chains in the @xmath15-bruhat order and the structure constants @xmath7 , when @xmath71 is a minimal coset representative in @xmath72 .
we consider an instance of this .
let @xmath1623 .
then @xmath1624 and this is the interval @xmath1625_p$ ] : @xmath1626 the multiple edges are those with two possible colourings .
one may verify that @xmath1627 . to check theorem [ thm : chains ] , we first compute @xmath1628 for those @xmath1629 of length 4 which are minimal in their @xmath15-coset .
@xmath1630 the first two are grassmannian of descent 2 , and the last two are grassmannian of descent 3 .
since @xmath1631 , we have @xmath1632 let @xmath1633 . then @xmath1634 and @xmath1635 .
since @xmath1636 and @xmath1637 , theorem [ thm : d ] implies @xmath1638 next , let @xmath1639 be in general position . if @xmath1640 , then @xmath1641 and @xmath1642 , contradicting @xmath309 and @xmath313 in general position .
thus @xmath1643 to compute @xmath1644 , note that @xmath1645 .
two applications of monk s formula show @xmath1646 .
( the other computations could also have proceeded via monk s formula . ) to compute @xmath69 for these minimal coset representatives , consider the part of the @xmath15-bruhat order rooted at @xmath70 and restricted to permutations of length at most 4 : @xmath1647 the small numbers adjacent to each permutation @xmath71 are @xmath69 . thus @xmath1648 which equals @xmath1649 .
using the definition of schubert polynomials in [ sec : schubert ] , one may check @xmath1655 the schubert polynomials @xmath39 for @xmath1656 are indicated in figure [ fig : schubert_s_4 ] .
the schubert polynomial @xmath39 is written below the permutation @xmath0 , and these data are displayed at the vertices of the permutahedron ( cayley graph of @xmath220 ) .
the divided difference operators are displayed on the edges of this figure .
we see that @xmath1658 is equal to @xmath1659\ + \ { \frak s}_{54123}(y){\frak s}_{2413}(z)\ ] ] @xmath1660\left [ { \frak s}_{4213}(z)+ { \frak s}_{3412}(z)\right]\ + \ { \frak s}_{51324}(y ) { \frak s}_{4312}(z).\ ] ] the definition of the @xmath11-bruhat orders imply that if @xmath1661 , and @xmath1662 , then the following are equivalent : @xmath1663 these induce the following isomorphisms ( which were stated in theorem [ thm : new_order ] ) of intervals in the @xmath183-order on @xmath1664 .
suppose @xmath564 and @xmath1665 .
then @xmath210_\preceq\quad \simeq\quad [ e,\overline{\zeta}]_\preceq\quad \simeq\quad [ e,\zeta^{-1}]^{\mbox{\scriptsize op}}_\preceq\quad \simeq\quad [ e,\overline{\zeta}^{-1}]^{\mbox{\scriptsize op}}_\preceq.\ ] ] these are illustrated in the posets displayed in
[ sec : schensted_example ] .
besides algorithm [ alg : chain ] , there are three other ` canonical ' algorithms for finding a chain between @xmath42 and @xmath0 when @xmath45 , each induced from algorithm [ alg : chain ] by one of the automorphisms of the previous section .
for example , here is one .
. then @xmath1674 . in figure
[ fig:125634 ] , we display the labeled hasse diagram of @xmath1675_4 $ ] and beside it a table of the words of the 14 chains in this interval , each displayed above its insertion and recording tableau .
note that @xmath1677 and @xmath1678 .
we continue this example , and illustrate theorem [ thm : d ] . in figure [ fig:145236 ]
are the labeled hasse diagram of @xmath1679_4 $ ] , and the insertion and recording tableaux for all 14 chains in this interval . for these last two intervals , it is interesting to view them with the permutation @xmath1681_k$ ] replaced by the geometric graph of @xmath1682 , as illustrated in figure [ fig : cyclic_geometric ] .
this gives an idea of the effect of a ` cyclic shift ' on the @xmath183-order .
in the theory of partially ordered sets , one often constructs a simplicial complex @xmath1684 from a poset , @xmath15 .
we compute one such for an interval in the @xmath11-bruhat order , which shows these intervals are not in general shellable .
we illustrate this with one example drawn from this paper . in example
[ ex : neworder ] , we considered the interval @xmath561_2 $ ] .
we display that interval below , together with the hasse diagram of an isomorphic poset : @xmath1685 the simplicial complex @xmath1684 associated to a poset @xmath15 has as simplices all chains , including the non - maximal ones . in our case above , the maximal simplices are @xmath1686 while @xmath1687 and @xmath1688 are attached along facets ( @xmath1689 and @xmath1690 , respectively ) , the pairs @xmath1691 and @xmath1692 are not .
they are attached along codimension 2 faces , @xmath1693 and @xmath1694 , respectively .
thus this simplicial complex is not shelable .
below , we display a geometric realization of this simplicial complex : @xmath1695 , _ sur les dcompositions cellulaires des espaces @xmath82 _ , in algebraic groups and their generalizations : classical methods , american mathematical society , 1994 , pp .
proceedings and symposia in pure mathematics , vol .
56 , part 1 . ,
_ yang - baxter equation , symmetric functions and schubert polynomials _ , discrete math .
, 153 ( 1996 ) , pp .
proceedings of the 5th conference on formal power series and algebraic combinatorics ( florence , 1993 ) . height 2pt depth -1.6pt width 23pt , _ symmetric polynomials and divided differences in formulas of intersection theory _ , in parameter spaces , vol .
36 of banach center publications , banach center workshop , 1994 , institute of mathematics , polish academy of sciences , 1996 , pp .
125177 .
height 2pt depth -1.6pt width 23pt , _ la correspondence de robinson _ , in combinatoire et reprsentation du groupe symtrique , d. foata , ed .
579 of lecture notes in math .
, springer - verlag , 1977 , pp . | we illuminate the relation between the bruhat order and structure constants for the polynomial ring in terms of its basis of schubert polynomials .
we use combinatorial , algebraic , and geometric methods , notably a study of intersections of schubert varieties and maps between flag manifolds . we establish a number of new identities among these structure constants .
this leads to formulas for some of these constants and new results on the enumeration of chains in the bruhat order . a new graded partial order on the symmetric group which contains young s lattice arises from these investigations .
we also derive formulas for certain specializations of schubert polynomials .
[ subsection ] [ lem]theorem [ lem]proposition [ lem]corollary [ lem]example [ lem]conjecture [ lem]algorithm [ lem]remark [ lem]definition to the memory of marcel paul schtzenberger |
there are many possible approaches to the selection of ccqe candidates in the minerva detector .
we expect , eventually , to have at least three strategies for selection , each of which will select different ccqe - rich final states and admit different backgrounds in the analysis .
first , we can perform a topological track based analysis with a one - track ( muon ) low @xmath0 component and a two - track ( muon and nucleon ) component at higher @xmath0 where reconstruction of the nucleon , at least when it is a proton , is feasible .
this style of analysis can be most easily compared to results using similar techniques from sciboone @xcite and nomad @xcite .
second , we can perform an inclusive analysis that selects only a muon and removes background based on a veto on michel electrons from the chain @xmath23 which is most similar to the analysis from the miniboone @xcite experiment .
finally , one can identify tracks as in the first analysis but remove events with additional visible recoil energy in the detector which is far from the vertex .
the latter stipulation is important for keeping events with low energy nucleons in the final state from the remaining nucleus after the ccqe interaction . the analysis presented here as our first preliminary work towards this reaction is in the final of these three styles .
as we progress in this analysis , we plan to add the other techniques . .
the track originates in the scintillator region and is momentum analyzed in the minos near detector .
the @xmath24 mev energy deposit away from the track is consistent with the direction of the recoil neutron .
( right ) for simulated events , the recoil energy away from the vertex vs. @xmath0 derived from the muon energy and angle .
the region selected in this space keeps very high efficiency but lets in significant background at high @xmath0.,title="fig:",scaledwidth=40.0% ] .
the track originates in the scintillator region and is momentum analyzed in the minos near detector .
the @xmath24 mev energy deposit away from the track is consistent with the direction of the recoil neutron .
( right ) for simulated events , the recoil energy away from the vertex vs. @xmath0 derived from the muon energy and angle .
the region selected in this space keeps very high efficiency but lets in significant background at high @xmath0.,title="fig:",scaledwidth=40.0% ] this analysis first selects events with a @xmath25 originating from the scintillator tracker whose momentum and charge are measured in the minos near detector @xcite . from the @xmath25 reconstructed momentum and angle with respect to the neutrino beam direction we can reconstruct the neutrino energy and @xmath0 , @xmath26 where @xmath27 , @xmath28 and @xmath29 are the neutron , proton and muon masses , @xmath30 and @xmath31 are the muon energy and angle , and @xmath32 is the target proton binding energy , @xmath33 mev in our estimation of the neutrino kinematics . a quasi - elastic candidate event passing this selection is shown in fig .
[ fig : candidate ] .
the only other visible energy in the event is the single bar with a @xmath24 mev deposit consistent with the expected direction of recoil based on the direction and energy of the muon .
the expected recoil is a neutron with kinetic energy of @xmath34 mev .
typically only a small fraction of the neutron energy is observed calorimetrically in the minerva detector . originates .
( left and middle ) the transverse directions , @xmath35 and @xmath36 within the hexagonal fiducial volume .
( right ) the direction along the beam , including regions outside the fiducial volume selected by the analysis such as the more dense electromagnetic calorimeter ( ecal ) region downstream of the fiducial volume and the very dense hadron calorimeter ( hcal ) farther downstream.,title="fig:",scaledwidth=33.0% ] originates .
( left and middle ) the transverse directions , @xmath35 and @xmath36 within the hexagonal fiducial volume .
( right ) the direction along the beam , including regions outside the fiducial volume selected by the analysis such as the more dense electromagnetic calorimeter ( ecal ) region downstream of the fiducial volume and the very dense hadron calorimeter ( hcal ) farther downstream.,title="fig:",scaledwidth=33.0% ] originates .
( left and middle ) the transverse directions , @xmath35 and @xmath36 within the hexagonal fiducial volume .
( right ) the direction along the beam , including regions outside the fiducial volume selected by the analysis such as the more dense electromagnetic calorimeter ( ecal ) region downstream of the fiducial volume and the very dense hadron calorimeter ( hcal ) farther downstream.,title="fig:",scaledwidth=33.0% ] for this first analysis , we chose a very conservative cut on the recoil energy in order to keep the efficiency of the analysis high .
this approach , however , lets in significant backgrounds , particularly at high @xmath0 as shown in fig .
[ fig : candidate ] . for events originating within the fiducial volume ,
the asymptotic efficiency at high energies is @xmath37 .
the efficiency is highest at low @xmath0 and falls slowly to @xmath38 at high @xmath0 not because of the low recoil selection , but because of the decrease in muon acceptance to enter the minos near detector .
a hit - level geant4-based simulation of the detector is used to evaluate these efficiencies and correct the neutrino interaction simulation for detector efficiencies and resolutions .
we gain confidence in this simulation by comparison of distributions of candidate events in the data with the simulation in quantities that are largely independent of the neutrino flux and the interactions . for example , fig .
[ fig : positions ] shows that the distribution of candidate vertices in the transverse and longitudinal dimensions of the detector agrees with a relatively normalized simulation .
we present the results as comparisons to the simulation , including background events generated as described above .
the simulation also includes our prediction of the untuned flux from a beamline simulation which has large uncertainties that we expect to reduce in the future @xcite .
we evaluate these uncertainties in the _ a priori _ neutrino flux to be lowest , about @xmath39 , at the peak energy of the flux shown in fig .
[ fig : spectrumrates ] and significantly larger above this `` focusing peak '' energy with an asymptotic uncertainty at high energy of @xmath40 .
these large flux uncertainties must be included in the comparisons between data candidates and the simulation .
these uncertainties allow a significant distortion in shape of the neutrino flux with neutrino energy , @xmath41 .
however , since the @xmath0 spectrum as a function of @xmath41 is approximately independent of @xmath41 , at least far below the kinematic limit , @xmath42 , we do not expect the flux uncertainties to significantly distort the predicted @xmath0 distribution .
( left ) and @xmath0 ( right ) reconstructed from muon kinematics by eqns .
[ eqn : enu ] and [ eqn : qsq ] , respectively , for @xmath43 candidates .
different reactions in the simulated event sample are shown stacked in the solid histogram .
the flux uncertainty is shown as a band around the simulation prediction.,title="fig:",scaledwidth=49.0% ] ( left ) and @xmath0 ( right ) reconstructed from muon kinematics by eqns .
[ eqn : enu ] and [ eqn : qsq ] , respectively , for @xmath43 candidates .
different reactions in the simulated event sample are shown stacked in the solid histogram .
the flux uncertainty is shown as a band around the simulation prediction.,title="fig:",scaledwidth=49.0% ] distributions for @xmath44 gev ( left ) and @xmath45 gev ( right ) .
the uncertainties shown are statistical only ; flux uncertainties approximately cancel in the analysis of the shape of this distribution , detector systematics and background uncertainties which are not included in the shown uncertainties may not .
the change in shape expected if @xmath46 gev is also shown.,scaledwidth=90.0% ] figure [ fig : enu - qsq ] shows the reconstructed neutrino energy and @xmath0 of the reactions for candidate events .
the comparison shows a disagreement outside the uncertainty band in the distribution of neutrino energies just above the focusing peak , @xmath47 gev@xmath48 gev .
however , the shape of the @xmath0 distribution , at least in the region where ccqe events make up the majority of the signal , is in reasonable agreement with our simulation .
note that the simulation uses an axial mass of @xmath49 gev .
miniboone and sciboone , measuring quasi - elastic neutrino and anti - neutrino scattering for @xmath50 gev , prefer significantly higher axial masses , consistent with @xmath46 gev @xcite , whereas the nomad experiment , measuring quasi - elastic neutrino scattering , @xmath51 gev , prefers @xmath52 consistent with the value of @xmath49 gev assumed in our simulation .
figure [ fig : qsq - by - energy ] shows the ratio of data to simulation for the minerva analysis in the low and high neutrino energy portions of the sample .
both appear more consistent with the low @xmath52 assumed rather than the high @xmath52 of miniboone and sciboone ; however , caution is warranted in drawing this conclusion because the minerva result only includes systematic uncertainties from the neutrino flux , and does not yet include uncertainties on the detector reconstruction or the backgrounds to the quasi - elastic interactions .
the minerva collaboration has performed its first comparison of a quasi - elastic rich anti - neutrino sample with expectations for a range of assumed axial masses .
the selection requires a muon from a neutrino interaction in the scintillator target and a low observed recoil away from the vertex in the final state to enhance elastic events .
minerva possesses an anti - neutrino sample approximately eight times as large as that in this analysis on tape , and will be able to improve the background rejection of this analysis in the future through a variety of techniques that the capabilities of the detector make possible .
this work was supported by the fermi national accelerator laboratory , which is operated by the fermi research alliance , llc , under contract no . de - ac02 - 07ch11359 , including the minerva construction project , with the united states department of energy .
construction support also was granted by the united states national science foundation under nsf award phy-0619727 and by the university of rochester .
support for participating scientists was provided by nsf and doe ( usa ) by capes and cnpq ( brazil ) , by conacyt ( mexico ) , by conicyt ( chile ) , by concytec , dgi - pucp and idi - uni ( peru ) , and by latin american center for physics ( claf ) .
additional support came from jeffress memorial trust ( mk ) , and research corporation ( em ) .
finally , the authors are grateful to the staff of fermilab for their contributions to this effort and to the minos collaboration for their efforts to operate the minos near detector and willingness to share its data . | determination of the quasi - elastic scattering cross - section over a broad range of neutrino energies , nuclear targets and @xmath0 is a primary goal of the minerva experiment .
we present preliminary comparisons of data and simulation in a sample rich in @xmath1 events from approximately one eighth of the total @xmath2 events collected by minerva to date .
we discuss future plans for quasi - elastic analyses in minerva .
address = department of physics and astronomy , university of rochester , rochester , ny 14627 usa + and fermi national accelerator laboratory , batavia , il 60510 usa + _ _ on behalf of the minerva collaboration _
_ the minerva experiment was designed , in part , to study charged - current quasi - elastic ( ccqe ) neutrino ( @xmath3 ) and anti - neutrino ( @xmath1 ) scattering of neutrinos with energy @xmath4 to @xmath5 or more gev over a broad range of @xmath0 .
minerva is currently running in the `` low energy '' ( le ) numi beam designed for the minos experiment in which it plans to integrate @xmath6 protons on target ( pot ) in neutrino mode and @xmath7 pot in anti - neutrino mode .
the anti - neutrino exposure is a mixture of data taken with a partially constructed detector and the full minerva detector . after a shutdown beginning in 2012
, the numi beamline will operate in the `` medium energy '' ( me ) tune .
figure [ fig : spectrumrates ] shows the predicted neutrino flux in both the le and me beams .
the active elements of the minerva detector @xcite are polystyrene scintillator bars formed into detector planes . a continuous series of these planes forms the `` tracker '' region of the minerva detector . the analysis reported in this
paper used events exclusively from the tracker ; however , the minerva detector also contains a region where the planes of scintillator are interspersed with passive nuclear targets
. these will allow study of @xmath8-dependence in the ccqe rates at different energy or @xmath0 .
figure [ fig : spectrumrates ] also shows the expected rate of ccqe interactions in the scintillator tracker and in the passive nuclear targets .
of course , not all of these events will be accepted in the analysis , but the combination of the significant mass of the minerva targets and the high intensity statistics available in the numi beam are very promising for this measurement .
the planned results for ccqe in minerva include measurements of the ccqe total cross - section , @xmath9 , and @xmath10 on the scintillator target . at low to moderate @xmath0 , @xmath11 gev@xmath12 ,
the primary goal is comparison with results from k2k @xcite , miniboone @xcite , sciboone @xcite and nomad @xcite .
in particular , there are open questions about the apparent cross - section enhancement at low energy and moderate @xmath0 and suppression at low @xmath0 in the existing data .
minerva will repeat these measurements on its multiple nuclear targets to look for changes with increasing @xmath8 and nuclear density . at higher
@xmath0 , minerva will be able to make the first measurements of the axial form factor and compare the high @xmath0 behavior with vector form factors .
minerva also plans to compare @xmath13 ccqe with the same @xmath14 process using the @xmath15 of its events from @xmath13 contamination in the beam . finally , minerva will search for the analogous neutral current process @xmath16 which is experimentally more challenging due to backgrounds from non - leptonic processes .
the study described in this paper is a preliminary comparison of muon kinematics in ccqe candidates between data and simulation with a sample of @xmath17 pot in the le anti - neutrino beam .
these data were taken during the construction of the minerva detector when only approximately one half of the detector had been installed .
the scintillator tracker fiducial mass for this sample is @xmath18 of the @xmath19 metric ton fiducial mass available in the full detector .
this sample is @xmath20 of the total exposure measured as the product of pot and fiducial mass .
the reference simulation is genie 2.6.2 @xcite .
the ccqe cross - section at nucleon level comes from the derivation of llewellyn - smith @xcite with vector form factors from the bbba2005 parametrization @xcite , the pseudo - scalar form factor from pcac , and the axial form factor in the dipole form with @xmath21 gev / c@xmath12 .
the nuclear effects when scattering from carbon come from the fermi gas model of bodek and ritchie @xcite , and pauli blocking is implemented by a requirement that the outgoing nucleon has momentum of above the fermi momentum of @xmath22 mev / c .
final state interactions of outgoing hadrons from the hard scattering process is simulated by intranuke @xcite . |
critical phenomena are observed in vicinity of phase transition points in a variety of systems ( e. g. , solids
ferromagnets , ferroelectrics ; fluids superfluid @xmath9-transition ; liquid gas transition , etc . ) , which manifest themselves in power like singularities of physical observables , described by critical exponents . in many systems , like isotropic ferromagnets (
e. g. , polycrystalline ni ) , certain singularity is observed not only at the critical temperature @xmath10 , but also at vanishing external field @xmath11 for any temperature @xmath12 .
this phenomenon , known as the goldstone mode singularity , is also described by power like divergences and certain exponents . in this sense
, it can be viewed as a critical phenomenon , which takes place in vicinity of the critical line @xmath13 in the @xmath14@xmath15 plane .
the vicinity of critical point ( or line ) is not the natural domain of validity of any perturbation theory , therefore one should resort the non perturbative methods such as * exact and rigorous analytical solution methods ( transfer matrix methods , combinatorial methods , bethe - ansatz ) , * conformal field theory analysis , * non perturbative renormalization group ( rg ) analysis , * numerical transfer matrix calculations , * molecular dynamics simulations , * monte carlo ( mc ) simulations .
the exact solutions @xcite and conformal field analysis @xcite are powerful tools to determine the critical exponents of two dimensional models
. however , these approaches are not helpful in most of the three dimensional cases .
the mc method can be used here .
the exact non perturbative rg equations are known for various models on fractal lattices ( see , e. g. @xcite and references therein ) , and also in four dimensions a rigorous rg analysis has been made ( see e. g. , @xcite ) . in most of other cases , the non perturbative rg equations provide only approximations , the accuracy of which can not be well controlled ( see , e. g. , @xcite ) . to the contrary , the mc method allows , in principle , to reach any desired accuracy .
besides , it allows to treat much larger systems than the other numerical methods .
it is extremely important in studying critical phenomena .
therefore , the mc method has no real alternative in non perturbative determination of critical exponents of the three dimensional systems like lattice spin ( @xmath0 ) models .
the general framework of perturbative methods includes the study of critical point phenomena at @xmath16 , as well as the goldstone mode effects at @xmath12 .
our aim is to verify nonperturbatively the validity of different theoretical perturbative approaches .
hence , the mc test of the goldstone mode singularities is one of the important points here .
we start with the goldstone mode effects , since this case is simpler in the sense that it does not require the determination of the critical temperature @xmath17 , i. e. , there are less ambiguities as compared to the @xmath16 case .
here we consider such lattice spin models , where the spin is an @xmath18component vector with @xmath19 .
these are called @xmath0 models due to the @xmath0 global rotational symmetry exhibited by the @xmath18vector model in absence of the external field .
the hamiltonian @xmath20 of such a model reads @xmath21 where @xmath14 is temperature , measured in energy units , @xmath22 is the spin variable ( @xmath18component vector of unit length ) of the @xmath23th lattice site , @xmath24 is the coupling constant , and @xmath25 is the external field with magnitude @xmath26 .
the summation takes place over all pairs @xmath27 of the nearest neighbors in the lattice . in the thermodynamic limit @xmath28 below the critical point , i. e. , at @xmath29 , the magnetization @xmath30 and the fourier
transformed transverse ( @xmath31 ) and longitudinal ( @xmath32 ) two point correlation functions exhibit goldstone mode power
law singularities : @xmath33 according to the standard theory @xcite , @xmath34 and @xmath35 hold for @xmath36 , and @xmath37 is true in three dimensions .
more nontrivial universal values are expected according to @xcite , such that @xmath38 hold for @xmath36 .
these relations have been obtained in @xcite by analyzing self - consistent diagram equations for the correlation functions without cutting the perturbation series .
this approach is based on certain grouping of feynman diagrams introduced in @xcite , and therefore is called the gfd theory .
mc simulations have been performed @xcite to test these relations for the @xmath39 and @xmath2 models on the simple cubic lattice .
the theoretical predictions ( [ eq : pred1 ] ) and ( [ eq : pred2 ] ) are confirmed by mc simulation results in the 3d @xmath2 model @xcite , where an estimate @xmath40 has been found for the transverse correlation function . moreover , it has been stated that the behavior of the longitudinal correlation function is well consistent with @xmath41 being about @xmath42 , in agreement with ( [ eq : pred2 ] ) at @xmath43 being about @xmath44 , but it is not consistent with the standard theoretical prediction of @xmath45 . according to ( [ eq : pred3 ] )
, we have @xmath46 in three dimensions .
a reasonable numerical evidence for this relation has been obtained in @xcite from the susceptibility data , where the mc estimate @xmath47 has been reported for the 3d @xmath1 ( @xmath39 ) model .
it corresponds to @xmath48 according to ( [ eq : pred3 ] ) .
in vicinity of the phase transition point , various quantities have often power law singularities , which are described by the critical exponents . for three dimensional systems ,
exact results are difficult to obtain , and one usually relies on approximate methods .
a review of standard perturbative rg methods applied here can be found , e. g. , in @xcite .
the results of the borel resummation of the perturbation series for the critical exponents , obtained within this approach , are reported in @xcite .
we will further focus on the monte carlo testing of the theoretical predictions for the 3d ising model , which is a particular case of @xmath49 .
the critical exponent @xmath4 , describing the @xmath50 singularity of the critical correlation function @xmath51 , is of particular interest here .
according to @xcite , the most accurate theoretical value @xmath52 is obtained from the series at fixed dimension @xmath53 .
the results of the resummation of the @xmath54expansion in @xcite are @xmath55 and @xmath56 .
if all these estimates are correct within the error bars , then we have @xmath57 .
this estimation fairly well agrees with the value @xmath58 extracted from the finite size scaling analysis of the mc data within @xmath59 $ ] , reported in @xcite . we will present the mc results for much larger lattices up to @xmath3 to test the agreement more precisely .
the other relevant here exponents are the correlation length exponent @xmath60 and correction to scaling exponent @xmath61 .
the widely accepted values for the 3d ising model are @xmath62 and @xmath63 .
they are in agreement , e. g. , with the estimates @xmath64 and @xmath65 reported in @xcite . as discussed in sec .
[ sec : goldstone ] , the alternative theoretical approach , called the gfd theory , provides promising results concerning the goldstone mode singularities . therefore , it is interesting to verify the predictions of this theory also for the critical point singularities . in @xcite , a set of possible values of the exact critical exponents for the @xmath66 model in two ( @xmath67 ) and three ( @xmath53 ) dimensions has been proposed : @xmath68 where @xmath69 is the susceptibility exponent , @xmath70 may have a natural value starting with 1 , and @xmath71 is an integer equal or larger than @xmath72 . other critical exponents can be calculated from these ones , using the known scaling relations .
these values agree with the known exact solutions of the two dimensional ising model ( @xmath73 , @xmath74 ) and of the spherical model ( @xmath75 ) .
a prediction has been made also for the 3d ising model , i. e. , @xmath76 and @xmath8 , corresponding to @xmath73 and @xmath74 , as in the two dimensional case .
this value of @xmath60 is consistent with the logarithmic singularity of specific heat ( according to @xmath77 ) proposed earlier by tseskis @xcite .
the exponents @xmath76 and @xmath8 have been later conjectured for the 3d ising model by zhang @xcite .
the critical exponents @xmath78 and @xmath79 , calculated for the liquid gas system by bondarev @xcite , also correspond to ( [ eq : result ] ) . in this case @xmath80 and @xmath81 hold .
as explained in @xcite , the equations ( [ eq : result ] ) are meaningful for positive integer @xmath18 , and we can have @xmath82 and @xmath83 in the case where the order parameter is an @xmath18component vector . the spatial dimensionality @xmath84 might be considered as a continuous parameter in ( [ eq : result ] ) within @xmath85 . at @xmath49 , the second order phase transition at a finite temperature
can be expected also for @xmath86 . however , according to the discussion in @xcite , an analytic extension of ( [ eq : result ] ) to this region , probably , is only formal and does not correspond to any real system ( fractal lattice ) . according to @xcite
, corrections to scaling can be represented by an expansion of correction factor ( amplitude ) in integer powers of @xmath87 and @xmath88 at @xmath89 , where @xmath90 is the reduced temperature .
this expansion is simplified , since @xmath91 is an integer number according to ( [ eq : result ] ) , and @xmath92 holds according to the known scaling relation .
hence , we obtain the expansion in powers of @xmath93 , where @xmath94 . in other words , the correction to scaling exponent in the expansions at @xmath89 is @xmath95 , and the corresponding correction to scaling exponent in the finite size scaling analysis is @xmath96 , if the first expansion coefficient is nonzero . allowing that some of the expansion coefficients are zero ,
we can have @xmath97 , where @xmath98 is a positive integer . however , our actual mc data support the most natural choice of @xmath99 for the 3d ising model .
it implies that the expansion coefficients do not vanish in general , but the nontrivial correction terms vanish only in some special cases , as the 2d ising model or the spherical model , where @xmath100 and/or @xmath101 .
according to the numerical transfer matrix calculations in @xcite , a nontrivial correction to finite
size scaling with the exponent @xmath102 , probably , exists in the two point correlation function even in the 2d ising model , although its amplitude is very small here .
the discussed here gfd critical exponents @xmath103 and @xmath8 for @xmath49 seem to be quite incompatible with the mc data of the 3d ising model , if we assume the usual correction to scaling exponent @xmath63 .
however , it turns out that our data for large enough sizes are very well consistent even with the set of exponents @xmath7 and @xmath8 . the disagreement of these exponents with those of the perturbative rg method can also be understood based on a critical analysis @xcite .
we have simulated the 3d ising model on a simple cubic lattice ( at @xmath13 ) , using the iterative method introduced in @xcite to adjust the coupling @xmath24 to the pseudo - critical coupling @xmath104 , which corresponds to certain value @xmath105 of the quantity @xmath106 , where @xmath70 is the magnetization per spin .
the pseudo - critical coupling @xmath104 tends to the true critical coupling @xmath107 at @xmath28 . by this method
, we estimate the moments of energy @xmath108 and magnetization @xmath70 per spin , i. e. , @xmath109 for a set of @xmath24 values fluctuating around @xmath104 , and then calculate these mean values and related quantities at any given @xmath24 near @xmath104 by using the taylor series expansion for @xmath110 .
we have evaluated @xmath111 for various lattice sizes @xmath112 and then estimated @xmath107 by fitting these @xmath111 data . besides , we have calculated the susceptibility @xmath113 and the derivative @xmath114 , where @xmath115 , at @xmath116 , as well as at certain estimated critical couplings @xmath107 , in order to make a finite size scaling analysis of the critical exponents .
this method has been tested and discussed in detail in @xcite , giving also some data for @xmath117 .
here we have extended the simulation results up to @xmath3 .
a parallel ( openmp ) implementation of the wolff single cluster algorithm has been used in our simulations , combining it with a sophisticated shuffling scheme for generation of a high quality pseudo random numbers , as described in @xcite .
several tests have been made in @xcite to verify the quality of the pseudo random numbers .
here we have tested and used also the lagged fibonacci pseudo random number generator ( prng ) with multiplication operation and the lags @xmath118 , @xmath119 ( see @xcite ) .
we have improved it with the standard shuffling scheme @xcite , using the shuffling box of the size @xmath120 in this case .
although the lagged fibonacci generator with additive operation produces certain correlations in the sequence of pseudo random numbers , which can be well detected by the directed random walk test @xcite , this defect is practically not observed in the actual multiplicative case .
like in @xcite , we have verified it by performing such test with @xmath121 trajectories .
good results have been obtained in this test both for the original prng and for the one improved by the shuffling .
it indicates that the multiplicative lagged fibonacci generator works fine in applications with cluster algorithms , as it also has been mentioned in @xcite .
the simulations of the two largest lattices with @xmath122 and @xmath3 have been performed with the shuffling scheme of @xcite and partly ( 24 usable iterations from 56 or 54 ones at @xmath122 and @xmath3 , respectively ) also with the improved lagged fibonacci generator . for complete confidence
, we have verified that the simulation results of the two generators agree within the statistical error bars , whereas the final simulated values have been obtained by summarizing the data from both of them .
note that one iteration included @xmath123 mc ( wolff cluster algorithm ) steps at @xmath122 and @xmath124 mc steps at @xmath3 , the mc measurements being performed after each @xmath125 and @xmath126 steps , respectively .
our simulation results for the pseudo - critical coupling @xmath104 , as well as for the corresponding values of @xmath127 and @xmath114 depending on the lattice size @xmath112 are given in tab .
[ tab1 ] . in tab .
[ tab2 ] , the latter two quantities are given , calculated at @xmath128 and @xmath129 , corresponding to two different estimates of the critical coupling .
besides , we have calculated these quantities also at @xmath130 for @xmath131 and have verified that the values agree within the error bars with those reported for this @xmath24 in @xcite . ' '' '' l & @xmath132 & @xmath127 & @xmath133 + 1536 & 0.2216546081(114 ) & 1.1900(22 ) & 96.01(73 ) + 1280 & 0.2216546524(136 ) & 1.2009(25 ) & 72.04(53 ) + 1024 & 0.221654625(22 ) & 1.2046(28 ) & 50.57(45 ) + 864 & 0.221654635(25 ) & 1.2165(21 ) & 38.71(25 ) + 768 & 0.221654672(27 ) & 1.2212(20 ) & 31.89(20 ) + 640 & 0.221654615(31 ) & 1.2281(17 ) & 23.95(12 ) + 512 & 0.221654662(45 ) & 1.2367(16 ) & 16.785(77 ) + 432 & 0.221654637(58 ) & 1.2450(18 ) & 12.907(59 ) + 384 & 0.221654567(65 ) & 1.2480(16 ) & 10.627(50 ) + 320 & 0.221654716(75 ) & 1.2578(16 ) & 7.967(36 ) + 256 & 0.22165460(11 ) & 1.2656(15 ) & 5.577(24 ) + 216 & 0.22165460(13 ) & 1.2726(12 ) & 4.288(14 ) + 192 & 0.22165425(16 ) & 1.2734(14 ) & 3.533(14 ) + 160 & 0.22165414(18 ) & 1.2818(11 ) & 2.6495(87 ) + 128 & 0.22165430(20 ) & 1.2913(10 ) & 1.8643(49 ) + 108 & 0.22165376(26 ) & 1.2969(10 ) & 1.4170(36 ) + 96 & 0.22165369(32 ) & 1.3012(10 ) & 1.1796(28 ) + 80 & 0.22165278(32 ) & 1.30659(74 ) & 0.8822(18 ) + 64 & 0.22165159(52 ) & 1.31466(78 ) & 0.6192(11 ) + 54 & 0.22164968(56 ) & 1.31916(76 ) & 0.47334(81 ) + 48 & 0.22164790(69 ) & 1.32164(66 ) & 0.39331(63 ) + 40 & 0.22164383(80 ) & 1.32562(64 ) & 0.29424(40 ) + 32 & 0.22163444(98 ) & 1.32835(59 ) & 0.20586(25 ) + 27 & 0.2216212(11 ) & 1.32829(52 ) & 0.15703(16 ) + 24 & 0.2216125(12 ) & 1.33027(47 ) & 0.13076(13 ) + 20 & 0.2215821(17 ) & 1.32860(42 ) & 0.097717(87 ) + 16 & 0.2215235(18 ) & 1.32510(34 ) & 0.068538(48 ) + & ' '' '' @xmath127 & @xmath134 & @xmath127 & @xmath134 + 1536 & 1.1882(38 ) & 95.94(69 ) & 1.1930(38 ) & 96.12(69 ) + 1280 & 1.1849(30 ) & 71.62(45 ) & 1.1885(30 ) & 71.72(45 ) + 1024 & 1.1998(37 ) & 50.48(40 ) & 1.2023(37 ) & 50.53(40 ) + 864 & 1.2110(32 ) & 38.63(22 ) & 1.2130(32 ) & 38.66(22 ) + 768 & 1.2110(31 ) & 31.77(18 ) & 1.2126(32 ) & 31.79(18 ) + 640 & 1.2268(27 ) & 23.94(10 ) & 1.2280(27 ) & 23.95(10 ) + 512 & 1.2321(27 ) & 16.756(69 ) & 1.2330(27 ) & 16.762(69 ) + 432 & 1.2429(24 ) & 12.898(52 ) & 1.2436(24 ) & 12.901(52 ) + 384 & 1.2499(24 ) & 10.634(44 ) & 1.2505(24 ) & 10.636(44 ) + 320 & 1.2534(22 ) & 7.955(33 ) & 1.2538(22 ) & 7.956(33 ) + 256 & 1.2656(23 ) & 5.577(22 ) & 1.2659(23 ) & 5.577(22 ) + 216 & 1.2727(21 ) & 4.288(13 ) & 1.2729(21 ) & 4.288(13 ) + 192 & 1.2795(21 ) & 3.540(12 ) & 1.2797(21 ) & 3.541(12 ) + 160 & 1.2879(17 ) & 2.6550(78 ) & 1.2880(17 ) & 2.6551(78 ) + 128 & 1.2941(13 ) & 1.8660(42 ) & 1.2942(13 ) & 1.8661(42 ) + 108 & 1.3028(13 ) & 1.4198(32 ) & 1.3029(13 ) & 1.4198(32 ) + 96 & 1.3066(14 ) & 1.1817(25 ) & 1.3066(14 ) & 1.1817(25 ) + 80 & 1.3147(11 ) & 0.8845(16 ) & 1.3147(11 ) & 0.8845(16 ) + 64 & 1.3241(12 ) & 0.62115(96 ) & 1.3241(12 ) & 0.62116(96 ) + 54 & 1.33098(85 ) & 0.47518(69 ) & 1.33101(85 ) & 0.47518(69 ) + 48 & 1.33504(93 ) & 0.39503(54 ) & 1.33506(93 ) & 0.39503(54 ) + 40 & 1.34177(77 ) & 0.29577(34 ) & 1.34179(77 ) & 0.29577(34 ) + 32 & 1.34958(71 ) & 0.20728(21 ) & 1.34960(71 ) & 0.20728(21 ) + 27 & 1.35514(60 ) & 0.15838(14 ) & 1.35515(60 ) & 0.15838(14 ) + 24 & 1.35836(53 ) & 0.13192(11 ) & 1.35836(53 ) & 0.13192(11 ) + 20 & 1.36471(61 ) & 0.098821(75 ) & 1.36471(61 ) & 0.098821(75 ) + 16 & 1.37072(45 ) & 0.069508(41 ) & 1.37072(45 ) & 0.069508(41 ) +
we have evaluated the critical coupling @xmath107 by fitting the @xmath135 data to the finite size scaling ansatz @xmath136 neglecting higher order corrections . the fits with the usual ( rg ) exponents @xmath137 and @xmath138 within the range @xmath139 $ ] are shown in fig .
[ betac1 ] . vs @xmath140 .
the dashed line shows the fit to ( [ betac ] ) , including only the leading correction term to @xmath141 with coefficient @xmath142 and exponents @xmath137 and @xmath138 , whereas the solid line is the fit with both correction terms included . ]
the dashed line shows the fit with only the leading correction term ( the term with coefficient @xmath142 ) to @xmath143 included , whereas the solid line the fit with both correction terms .
these fits give @xmath144 and @xmath145 , respectively .
the @xmath146 of both these fits is @xmath147 .
the data for @xmath148 are omitted here , since inclusion of these relatively small sizes only increases the systematic errors without an essential reduction of the statistical errors . surprisingly , the @xmath132 data within @xmath139
$ ] can be even better fit with the exponents @xmath8 and @xmath149 of the gfd theory .
the fit , with only the leading correction to @xmath141 included , is shown in fig .
[ betac2 ] .
it yields @xmath150 . in this case
@xmath146 has the value @xmath151 , which is even smaller than for the fits with @xmath137 and @xmath138 .
vs @xmath152 .
the solid curve shows the fit to ( [ betac ] ) , including only the leading correction term to @xmath141 with coefficient @xmath142 and exponents @xmath8 and @xmath149 . ]
inclusion of the second correction term changes the result only slightly , i. e. , it gives @xmath153 with @xmath154 , and the fit curve lies almost on the top of that one shown in fig .
[ betac2 ] . the statistical error ,
however , is strongly increased in this case .
alternatively , the critical coupling @xmath107 can be determined from the susceptibility data , requiring the consistency with the finite size scaling ansatz @xmath155 or with the corresponding ansatz for the effective exponent @xmath156 at @xmath157 . here
the effective exponent @xmath158 is the local slope of the @xmath159 vs @xmath160 plot , evaluated from the linear fit within @xmath161 $ ] .
we observe that the plot of the effective exponent within @xmath162 $ ] ( evaluated from the data within @xmath163 $ ] ) is optimally described by ( [ etaeff ] ) with @xmath7 at @xmath164 , as shown in fig .
[ etam ] .
vs @xmath165 at @xmath128 ( solid circles ) and @xmath129 ( x ) .
the solid line is the fit of circles to ( [ etaeff ] ) with @xmath7 . ] from this we conclude that , if the exponents @xmath7 are correct , then @xmath166 .
this is the estimate of @xmath24 at which the fit to ( [ etaeff ] ) with @xmath149 and @xmath4 as a fit parameter yields @xmath103 within the statistical error bars .
the solid line fit at @xmath128 is really good , and the effective exponents can be quite well fit with @xmath7 also at @xmath129 ( the data shown by `` x '' ) , which is the best @xmath107 value provided by the already considered estimation with the rg exponents .
the value @xmath166 is our best estimate of the critical coupling in the case where we use the exponents of the gfd theory , since two corrections to scaling are included , giving smaller statistical error bars than for the other estimate @xmath153 with two corrections included .
our estimated value @xmath166 is sufficiently reasonable and plausible , since it agrees well within the error bars with all other estimations considered here , as well as with the most accurate values of @xmath107 provided by other authors , i. e. , @xmath167 @xcite and @xmath168 @xcite . concerning the standard rg exponents , the method of fitting @xmath169 data is less convenient , since it is sensitive to the precise value of @xmath4 , which is not known accurately enough in this approach .
in order to test the results of the perturbative rg theory , we have estimated the critical exponent @xmath4 by fitting the susceptibility data to ( [ chi ] ) and also by fitting the @xmath169 data to ( [ etaeff ] ) at fixed rg exponent @xmath138 . note that ( [ chi ] ) and ( [ etaeff ] ) are valid at @xmath170 , as well as at @xmath171 .
the fits have been made at @xmath171 and at the estimated in sec .
[ betacest ] values of the critical coupling , obtained by using the rg exponents .
we start with the fitting of the effective exponent , since this method can be better controlled to see which fits are most appropriate .
the plot of @xmath169 vs @xmath172 with @xmath138 , evaluated at @xmath171 , is shown in fig .
[ etap ] vs @xmath173 at @xmath171 .
the solid curve is the fit to ( [ etaeff ] ) with fixed @xmath138 .
the dotted lines indicate the range @xmath174 $ ] of @xmath4 values consistent with the set of perturbative rg estimations in @xcite . ]
we see that the data are well described by a quadratic curve for @xmath175 , therefore one has to include both correction terms in ( [ etaeff ] ) .
the fit yields @xmath176 .
this value is slightly larger than the upper limit @xmath177 of the interval @xmath178 for the values , which are consistent with the set of perturbative rg estimations of @xcite discussed in sec .
however , the discrepancy is within the error bars .
somewhat larger deviations above @xmath177 are provided by the fits within @xmath179 , shown in fig .
[ eta ] , at approximately estimated values @xmath180 and @xmath181 of the critical coupling .
vs @xmath173 at @xmath180 ( diamonds ) and @xmath181 ( circles ) .
the curves are the fits to ( [ etaeff ] ) with fixed @xmath138 .
the dotted lines indicate the range @xmath174 $ ] of @xmath4 values consistent with the set of perturbative rg estimations in @xcite . ]
these fits yield @xmath182 and @xmath183 , respectively .
the indicated here standard errors are calculated by the jackknife method , taking into account the statistical fluctuations in the estimated @xmath107 value .
namely , the standard error @xmath184 is evaluated as @xmath185 , where @xmath186 is the shift in the estimated @xmath4 value , taking into account also the shift in the fitted @xmath107 value , when the @xmath71th simulation block ( iteration ) for the @xmath23th lattice size is omitted .
the statistical correlations are such that the total standard error is smaller than in the case where there are no correlations between the fluctuations in the estimated @xmath107 value and in the @xmath4 value obtained at a given @xmath24 .
we have also evaluated @xmath4 from the fits to ( [ chi ] ) at the pseudo - critical coupling @xmath111 , as well as at the two estimates @xmath180 and @xmath181 of the critical coupling .
the results of fits within @xmath187 $ ] , assuming @xmath138 , are collected in tab .
[ tab3 ] . ' '' '' @xmath188 & @xmath189 & @xmath190 & @xmath191 + 32 & 0.0345(12 ) & 0.0363(14 ) & 0.0379(25 ) + 40 & 0.0353(14 ) & 0.0364(15 ) & 0.0384(27 ) + 48 & 0.0359(17 ) & 0.0369(17 ) & 0.0394(29 ) + 54 & 0.0369(19 ) & 0.0376(20 ) & 0.0402(28 ) + 64 & 0.0380(23 ) & 0.0376(30 ) & 0.0405(25 ) + 80 & 0.0377(37 ) & 0.0366(38 ) & 0.0398(32 ) + the best estimates in these three cases are assumed to be @xmath192 , @xmath193 and @xmath194 , since they perfectly agree with the corresponding estimates at a larger @xmath188 , but have smaller statistical errors .
note that the decrease of the standard error for @xmath191 in tab .
[ tab3 ] at @xmath195 as compared to @xmath196 is the effect of the already mentioned statistical correlations .
these best @xmath4 values are very similar to the ones obtained before from the effective exponents .
remarkable is the fact that a better estimation of the critical coupling , including two corrections to scaling instead of only one correction , leads to a worse agreement with the results of the perturbative rg theory . in particular , the value @xmath194 deviates above @xmath177 ( the upper limit of the best rg estimate @xmath197 of @xcite ) by @xmath198 .
the observed here deviations suggest that @xmath4 , probably , is larger than normally expected from the perturbative rg theory .
the statistical errors , however , do not allow to make a strict conclusion . as already discussed in sec .
[ betacest ] , the pseudo - critical coupling @xmath111 can be even better fit with the exponents @xmath8 and @xmath149 of the gfd theory than with those of the perturbative rg theory .
we have also verified that the effective exponent @xmath158 is very well described by the ansatz ( [ etaeff ] ) with @xmath7 ( see fig .
[ etam ] ) at a certain @xmath24 corresponding to a reasonable estimate of the critical coupling @xmath166 .
also at the pseudo - critical coupling @xmath171 , the effective exponent @xmath158 within @xmath199 $ ] ( extracted from the susceptibility data within @xmath200 $ ] ) can be well fit to ( [ etaeff ] ) with fixed gfd exponents @xmath7 , as shown in fig .
[ etapm ] .
vs @xmath165 at @xmath171 .
the solid line is the fit with @xmath7 . ]
if the exponent @xmath4 is considered as a fit parameter at fixed @xmath149 , then this fit yields @xmath201 in agreement with the expected value @xmath202 , although the statistical error is rather large in this case .
sufficiently accurate data for even larger @xmath112 values would be very helpful here to perform a more precise testing of the agreement .
we have verified that not only the effective exponent @xmath169 , but also the susceptibility data can be well described by the exponents @xmath7 .
in particular , @xmath203 of the fit with @xmath7 is @xmath204 for the susceptibility data at @xmath171 within @xmath139 $ ] . at @xmath205
, the data can be well fit with @xmath206 over a wider range of sizes @xmath207 $ ] .
we have fitted also the @xmath208 data to the finite size scaling ansatz @xmath209 these data can be well fit with the rg exponents , as well as with those of the gfd theory . in the latter case ( @xmath210 , @xmath149 ) , the data within @xmath163 $ ] are fit with @xmath211 at @xmath171 and with @xmath212 at @xmath205 .
in fact , reasonable fits with the exponents @xmath103 and @xmath8 are possible because of the value @xmath202 of the correction to scaling exponent @xmath61 . a remarkably larger value , such as @xmath63 , would make such fits not good . according to ( [ etaeff ] ) , the plot of the effective exponent @xmath169 vs @xmath172 has to be almost linear at @xmath28 .
if @xmath61 is as large as @xmath213 , then @xmath214 is already quite small for @xmath175 , so that a good linearity of the fit curve in fig .
[ etap ] can be expected .
however , the plot looks rather nonlinear .
moreover , it can be much better approximated by a straight line ( within @xmath199 $ ] ) in the @xmath165 scale than in the @xmath173 scale , as it is clearly seen comparing the plots in figs .
[ etap ] and [ etapm ] .
according to this , it seems , in fact , very plausible that @xmath61 could be about @xmath202 .
the analysis of the mc data for the @xmath39 and @xmath2 models below the critical point supports the recently published theoretical results @xcite , predicting that the goldstone mode singularities in the @xmath0 models are described by nontrivial exponents , as discussed in sec .
[ sec : goldstone ] .
therefore , it has been important to verify the earlier predictions of this approach , called the gfd theory , concerning the critical point singularities . for this purpose
, we have performed mc simulations of the 3d ising model for very large lattices with linear size up to @xmath3 , using a parallel implementation of the wolff single cluster algorithm .
the finite size scaling analysis shows that the actual data can reasonably well interpreted with the usual critical exponents @xmath215 , @xmath62 and @xmath63 of the perturbative rg theory , and can also be well described by a different set of critical exponents , @xmath7 and @xmath8 , found within the gfd theory .
the validity of the fits with the latter set of exponents depends on whether the correction to scaling exponent @xmath61 can , indeed , be as small as @xmath202 .
we have demonstrated in sec .
[ sec : our ] that this is a very plausible scenario .
we have performed a high accuracy estimation of the critical coupling @xmath107 , combined with a sophisticated fitting of the data , including the leading , as well as sub - leading , corrections to scaling , to distinguish between the two sets of the critical exponents .
we have found that a self - consistent estimation of @xmath107 and @xmath4 , assuming the rg exponents @xmath62 and @xmath138 , gives somewhat larger values of @xmath4 ( e. g. , @xmath183 and @xmath194 when two corrections to scaling are included ) than the value @xmath6 expected from the most accurate resummation of the rg perturbative series @xcite .
these discrepancies are not large enough to make a strict conclusion about an inconsistency . on the other hand ,
the observed gradual increase in the effective exponent @xmath158 with growing system size @xmath112 is just expected if @xmath7 and @xmath8 are the correct asymptotic exponents .
therefore , it would be very interesting and important to see whether such an inrease to even larger @xmath169 values is supported by the data for even larger lattice sizes .
this work was made possible by the facilities of the shared hierarchical academic research computing network ( sharcnet : www.sharcnet.ca ) .
r. m. acknowledges the support from the nserc and crc program .
we thank also the deisa consortium ( www.deisa.eu ) , funded through the eu fp7 project ri-222919 , for support within the deisa extreme computing initiative . | critical phenomena and goldstone mode effects in spin models with @xmath0 rotational symmetry are considered .
starting with the goldstone mode singularities in the @xmath1 and @xmath2 models , we briefly review different theoretical concepts as well as state - of - the art monte carlo simulation results .
they support recent results of the gfd ( grouping of feynman diagrams ) theory , stating that these singularities are described by certain nontrivial exponents , which differ from those predicted earlier by perturbative treatments .
furthermore , we present the recent monte carlo simulation results of the three - dimensional ising model for very large lattices with linear sizes up to @xmath3 .
these results are obtained , using a parallel openmp implementation of the wolff single cluster algorithm .
the finite size scaling analysis of the critical exponent @xmath4 , assuming the usually accepted correction - to - scaling exponent @xmath5 , shows that @xmath4 is likely to be somewhat larger than the value @xmath6 of the perturbative renormalization group ( rg ) theory .
moreover , we have found that the actual data can be well described by different critical exponents : @xmath7 and @xmath8 , found within the gfd theory . * pacs * 05.10.ln ; 75.10.hk ; 05.50.+q |
the @xmath1 meson is the ground @xmath7-wave spin - triplet bottomonium ( bound state of @xmath8 ) with the well - established quantum number of @xmath9 @xmath10 @xmath11 @xcite .
its mass , @xmath12 @xmath10 @xmath13 mev @xcite , is less than the kinematic open - bottom threshold .
phenomenologically , the dominated @xmath1 hadronic decay through the @xmath8 pairs annihilation into three gluons , with branching ratio @xmath14 @xmath10 @xmath15 @xcite , is suppressed by the okubo - zweig - iizuka rule @xcite .
the partial width of the @xmath1 electromagnetic decay through the @xmath8 pairs annihilation into a virtual photon , @xmath16 , is proportional to @xmath17 , where @xmath18 @xmath10 @xmath19 is the electric charge of the bottom quark in the unit of @xmath20 , @xmath21 is the ratio of the inclusive production cross section of hadrons to the @xmath22 pair production cross section , and @xmath23 is the partial width of the pure leptonic @xmath1 @xmath2 @xmath24 decay .
besides , the @xmath1 meson can also decay via the weak interactions within the standard model , although the branching ratio is very small , about @xmath25 @xmath26 @xmath27 @xcite , where @xmath28 and @xmath29 are the lifetime of the @xmath30 meson and the total width of the @xmath1 meson , respectively . in this paper , we will study the @xmath1 @xmath2 @xmath3 weak decays with the perturbative qcd ( pqcd ) approach @xcite .
the motivation is listed as follows . from the experimental point of view , ( 1 ) over @xmath31 @xmath1 data samples were accumulated by the belle detector at the kekb @xmath32 asymmetric energy collider @xcite .
it is hopefully expected that more and more upsilon data samples will be collected with great precision at the forthcoming superkekb and the running upgraded lhc .
a large amount of @xmath1 data samples offer a realistic possibility to search for the @xmath1 weak decays which in some cases might be detectable .
theoretical studies on the @xmath1 weak decays are necessary to give a ready reference .
( 2 ) for the @xmath1 @xmath2 @xmath3 weak decay , the back - to - back final states with opposite electric charges have definite momentums and energies in the center - of - mass frame of the @xmath1 meson .
in addition , identification of either a single flavored @xmath33 or @xmath34 meson is free from the low double - tagging efficiency @xcite , and can provide an unambiguous evidence of the @xmath1 weak decay . of course , it should be noticed that small branching ratios for the @xmath1 weak decays make the observation extremely challenging , and any evidences of an abnormally large production rate of either a single @xmath33 or @xmath34 meson might be a hint of new physics @xcite . from the theoretical point of view
, the @xmath1 weak decays permit one to crosscheck parameters obtained from the @xmath35-flavored hadron decays , to further explore the underlying dynamical mechanism of the heavy quark weak decay , and to test various phenomenological approaches . in recent several years , many attractive methods have been developed to evaluate hadronic matrix elements ( hme ) where the local quark - level operators are sandwiched between the initial and final hadron states , such as pqcd @xcite , the qcd factorization @xcite and the soft and collinear effective theory @xcite , which could give reasonable explanation for many measurements on the nonleptonic @xmath36 decays .
the @xmath1 @xmath2 @xmath3 weak decay is favored by the color factor due to the external @xmath37 emission topological structure , and by the cabibbo - kobayashi - maskawa ( ckm ) factors @xmath38 , so it should have a large branching ratio . however , as far as we know , there is no theoretical investigation on the @xmath1 @xmath2 @xmath3 weak decay at the moment . in this paper
, we will predict the branching ratio and direct @xmath0-violating asymmetry of the @xmath1 @xmath2 @xmath3 weak decay with the pqcd approach to confirm whether it is possible to search for this process at the future experiments .
this paper is organized as follows . in section [ sec02 ] , we present the theoretical framework and the amplitude for the @xmath1 @xmath2 @xmath3 decay . section [ sec03 ] is devoted to numerical results and discussion .
finally , we conclude with a summary in the last section .
using the operator product expansion and renormalization group equation , the effective hamiltonian responsible for the @xmath1 @xmath2 @xmath3 weak decay is written as @xcite @xmath39 where @xmath40 @xmath10 @xmath41 @xcite is the fermi coupling constant ; the ckm factors are expressed as a power series in the wolfenstein parameter @xmath42 @xmath26 @xmath43 @xcite , @xmath44 the wilson coefficients @xmath45 summarize the physical contributions above the scale of @xmath46 , and have been reliably evaluated to the next - to - leading logarithmic order .
the local operators are defined as follows .
@xmath47 [ \bar{s}_{\beta } { \gamma}^{\mu}(1-{\gamma}_{5})c_{\beta } ] \label{q1 } , \\
% ----------------------------------------------------- q_{2 } & = & [ \bar{c}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\beta } ] [ \bar{s}_{\beta}{\gamma}^{\mu}(1-{\gamma}_{5})c_{\alpha } ] \label{q2 } , \end{aligned}\ ] ] @xmath48 [ \bar{q}_{\beta } { \gamma}^{\mu}(1-{\gamma}_{5})q_{\beta } ] \label{q3 } , \\
% ----------------------------------------------------- q_{4 } & = & \sum\limits_{q } [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\beta } ] [ \bar{q}_{\beta}{\gamma}^{\mu}(1-{\gamma}_{5})q_{\alpha } ] \label{q4 } , \\
% ----------------------------------------------------- q_{5 } & = & \sum\limits_{q } [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\alpha } ] [ \bar{q}_{\beta } { \gamma}^{\mu}(1+{\gamma}_{5})q_{\beta } ] \label{q5 } , \\
% ----------------------------------------------------- q_{6 } & = & \sum\limits_{q } [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\beta } ] [ \bar{q}_{\beta}{\gamma}^{\mu}(1+{\gamma}_{5})q_{\alpha } ] \label{q6 } , \end{aligned}\ ] ] @xmath49 [ \bar{q}_{\beta } { \gamma}^{\mu}(1+{\gamma}_{5})q_{\beta } ] \label{q7 } , \\
% ----------------------------------------------------- q_{8 } & = & \sum\limits_{q } \frac{3}{2}q_{q}\ , [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\beta } ] [ \bar{q}_{\beta}{\gamma}^{\mu}(1+{\gamma}_{5})q_{\alpha } ] \label{q8 } , \\ % ----------------------------------------------------- q_{9 } & = & \sum\limits_{q } \frac{3}{2}q_{q}\ , [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\alpha } ] [ \bar{q}_{\beta } { \gamma}^{\mu}(1-{\gamma}_{5})q_{\beta } ] \label{q9 } , \\ % ----------------------------------------------------- q_{10 } & = & \sum\limits_{q } \frac{3}{2}q_{q}\ , [ \bar{s}_{\alpha}{\gamma}_{\mu}(1-{\gamma}_{5})b_{\beta } ] [ \bar{q}_{\beta}{\gamma}^{\mu}(1-{\gamma}_{5})q_{\alpha } ] \label{q10 } , \end{aligned}\ ] ] where @xmath50 , @xmath51 , and @xmath52 are usually called as the tree operators , qcd penguin operators , and electroweak penguin operators , respectively ; @xmath53 and @xmath54 are color indices ; @xmath55 denotes all the active quarks at the scale of @xmath46 @xmath26 @xmath56 , i.e. , @xmath55 @xmath10 @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath35 . to obtain the decay amplitudes , the remaining works are to calculate the hadronic matrix elements of local operators as accurately as possible .
based on the @xmath61 factorization theorem @xcite and the lepage - brodsky approach for exclusive processes @xcite , hme can be written as the convolution of hard scattering subamplitudes containing perturbative contributions with the universal wave functions reflecting the nonperturbative contributions with the pqcd approach , where the transverse momentums of quarks are retained and the sudakov factors are introduced , in order to regulate the endpoint singularities and provide a naturally dynamical cutoff on nonperturbative contributions .
usually , the decay amplitude can be factorized into three parts : the hard effects incorporated into the wilson coefficients @xmath62 , the process - dependent scattering amplitudes @xmath63 , and the universal wave functions @xmath64 , i.e. , @xmath65 where @xmath66 is a typical scale , @xmath67 is the longitudinal momentum fraction of the valence quark , @xmath35 is the conjugate variable of the transverse momentum , and @xmath68 is the sudakov factor . the light cone kinematic variables in the @xmath1 rest frame are defined as follows .
@xmath69 where @xmath70 and @xmath71 are the longitudinal momentum fraction and transverse momentum of the valence quark , respectively ; @xmath72 is the longitudinal polarization vector of the @xmath1 meson .
the notation of momentum is showed in fig.[fig1](a ) .
there are some relations among these kinematic variables .
@xmath73 @xmath74\,[m_{1}^{2}-(m_{2}-m_{3})^{2 } ] } } { 2\,m_{1 } } \label{kine - pcm},\ ] ] where @xmath75 is the common momentum of the final @xmath34 and @xmath33 states ; @xmath76 @xmath10 @xmath12 , @xmath77 @xmath10 @xmath78 and @xmath79 @xmath10 @xmath80 denote the masses of the @xmath1 , @xmath34 and @xmath33 mesons , respectively . the hme of diquark operators squeezed between the vacuum and @xmath1 , @xmath34 , @xmath33 mesons are defined as follows .
@xmath81 \big\}_{ji } \label{wave - bbl},\ ] ] @xmath82 { \phi}_{b_{c}}(k_{2 } ) \big\}_{ji } \label{wave - bcp},\ ] ] @xmath83 { \phi}_{d_{s}}(k_{3 } ) \big\}_{ji } \label{wave - ds},\ ] ] where @xmath84 , @xmath85 , @xmath86 are decay constants .
there are several phenomenological models for the @xmath33 meson wave functions ( for example , eq.(30 ) in ref.@xcite ) . in this paper
, we will take the model favored by ref.@xcite via fitting with measurements on the @xmath87 @xmath2 @xmath88 decays .
@xmath89 where @xmath90 @xmath10 @xmath91 @xmath92 @xmath67 ; @xmath67 and @xmath35 are the longitudinal momentum fraction and the conjugate variable of the transverse momentum @xmath61 of the strange quark in the @xmath33 meson , respectively ; the exponential term represents the @xmath61 distribution ; @xmath93 @xmath10 @xmath94 and @xmath95 @xmath10 @xmath43 gev @xcite . due to
@xmath12 @xmath96 @xmath97 and @xmath78 @xmath96 @xmath98 @xmath99 @xmath100 , nonrelativistic quantum chromodynamics @xcite and schrdinger equation can be used to describe both @xmath1 and @xmath34 mesons .
the wave functions of an isotropic harmonic oscillator potential are given in ref .
@xcite , @xmath101 @xmath102 @xmath103 where @xmath104 @xmath10 @xmath105 with @xmath106 @xmath10 @xmath107 ; parameters @xmath108 , @xmath87 , @xmath109 are the normalization coefficients satisfying the following conditions @xmath110 the feynman diagrams for the @xmath1 @xmath2 @xmath3 decay are shown in fig.[fig1 ] .
there are two types : the emission and annihilation topologies , where diagrams containing gluon exchanges between the quarks in the same ( different ) mesons are entitled ( non)factorizable diagrams . @xmath2 @xmath3 decay with the pqcd approach , including the factorizable emission diagrams ( a , b ) , the nonfactorizable emission diagrams ( c , d ) , the nonfactorizable annihilation diagrams ( e , f ) , and the factorizable annihilation diagrams ( g , h).,scaledwidth=95.0% ] by calculating these diagrams with the pqcd master formula eq.([hadronic ] ) , the decay amplitudes of @xmath1 @xmath2 @xmath3 decay can be expressed as : @xmath111 - v_{tb } v_{ts}^{\ast}\,\big [ { \cal a}_{a+b}^{ll}\,(a_{4}+a_{10 } ) \nonumber \\ & & + { \cal a}_{a+b}^{sp}\ , ( a_{6}+a_{8 } ) + { \cal a}_{c+d}^{ll}\ , ( c_{3}+c_{9 } ) + { \cal a}_{c+d}^{sp}\ , ( c_{5}+c_{7 } ) \nonumber \\ & & + { \cal a}_{e+f}^{ll}\ , ( c_{3}+c_{4}-\frac{1}{2}c_{9}-\frac{1}{2}c_{10 } ) + { \cal a}_{e+f}^{lr}\ , ( c_{6}-\frac{1}{2}c_{8 } ) \nonumber \\ & & + { \cal a}_{g+h}^{ll}\ , ( a_{3}+a_{4}-\frac{1}{2}a_{9}-\frac{1}{2}a_{10 } ) + { \cal a}_{g+h}^{lr}\ , ( a_{5}-\frac{1}{2}a_{7 } ) \nonumber \\ & & + { \cal a}_{e+f}^{sp}\ , ( c_{5}-\frac{1}{2}c_{7 } ) \big ] \big\ } \label{amp - all } , \end{aligned}\ ] ] where @xmath112 @xmath10 @xmath113 and the color number @xmath114 @xmath10 @xmath115 . the parameters @xmath116 are defined as follows .
@xmath117 the building blocks @xmath118 , @xmath119 , @xmath120 , @xmath121 denote the contributions of the factorizable emission diagrams fig.[fig1](a , b ) , the nonfactorizable emission diagrams fig.[fig1](c , d ) , the nonfactorizable annihilation diagrams fig.[fig1](e , f ) , the factorizable annihilation diagrams fig.[fig1](g , h ) , respectively .
they are defined as @xmath122 where the subscripts @xmath123 and @xmath124 correspond to the indices of fig.[fig1 ] ; the superscript @xmath125 refers to one of the three possible dirac structures , namely @xmath125 @xmath10 @xmath126 for @xmath127 , @xmath125 @xmath10 @xmath128 for @xmath129 , and @xmath125 @xmath10 @xmath130 for @xmath131 .
the explicit expressions of these building blocks are collected in the appendix [ blocks ] .
in the rest frame of the @xmath1 meson , the @xmath0-averaged branching ratio and direct @xmath0-violating asymmetry for the @xmath1 @xmath2 @xmath3 weak decay are written as @xmath132 @xmath133 where the decay width @xmath134 @xmath10 @xmath135 kev @xcite .
the numerical values of other input parameters are listed as follows .
\(1 ) the wolfenstein parameters @xcite : @xmath108 @xmath10 @xmath136 , @xmath42 @xmath10 @xmath137 , @xmath138 @xmath10 @xmath139 , and @xmath140 @xmath10 @xmath141 , where @xmath142 @xmath10 @xmath143 .
\(2 ) masses of quarks @xcite : @xmath100 @xmath10 @xmath144 gev and @xmath98 @xmath10 @xmath145 gev .
\(3 ) decay constants : @xmath146 @xmath10 @xmath147 mev @xcite , @xmath85 @xmath10 @xmath148 mev @xcite , and @xmath86 @xmath10 @xmath149 mev @xcite . finally , we get @xmath150 @xmath151 where the central values are obtained with the central values of input parameters ; the first uncertainties come from the ckm parameters ; the second uncertainties are due to the variation of mass @xmath98 and @xmath100 ; the third uncertainties arise from the typical scale @xmath46 @xmath10 @xmath152 , where the expressions of @xmath153 for different topologies are given in eqs.([tab]-[tgh ] ) ; and the fourth uncertainties correspond to the variation of decay constants @xmath84 , @xmath85 , @xmath86 and shape parameter @xmath93 in eq.([da - ds ] )
. there are some comments .
\(1 ) it is seen from eq.([vbr ] ) that branching ratio for the @xmath1 @xmath2 @xmath3 decay can reach up to @xmath6 , which might be accessible at the running lhc and forthcoming superkekb .
for example , the @xmath1 production cross section in p - pb collision is a few @xmath154 with the lhcb @xcite and alice @xcite detectors at lhc . over @xmath155 @xmath1 mesons per @xmath156 data collected at lhcb and alice
are in principle available , corresponding to a few hundreds of the @xmath1 @xmath2 @xmath3 events .
\(2 ) compared the @xmath1 @xmath2 @xmath3 decay with the @xmath1 @xmath2 @xmath157 decay @xcite , they are both the color - favored and ckm - favored .
there is only the emission topologies and only the tree operators contributing to the @xmath1 @xmath2 @xmath157 decay . besides the emission topologies and tree operators , there are other contributions from the annihilation topologies and penguin operators for the @xmath1 @xmath2 @xmath3 decay .
in addition , there is another important factor , the decay constant @xmath86 @xmath158 @xmath159 .
this might explain the fact that although the final phase spaces for the @xmath1 @xmath2 @xmath3 decay is more compact than those for the @xmath1 @xmath2 @xmath157 decay , there is still the relation , @xmath4 @xmath158 @xmath160 with the pqcd approach .
\(3 ) it is shown from eq.([vcp ] ) that the direct @xmath0 asymmetry for the @xmath1 @xmath2 @xmath3 decay is close to zero .
the fact should be so .
as it is well known , the magnitude of direct @xmath0 asymmetry is proportional to the sine of weak phase difference .
first and foremost , the weak phase difference between the ckm factors @xmath161 and @xmath162 are suppressed by the factor of @xmath163 .
secondly , compared with the tree contributions appearing with @xmath161 , the penguin and annihilation contributions always accompanied with @xmath162 are suppressed by the small wilson coefficients .
\(4 ) as it is well known , due to mass @xmath78 @xmath158 @xmath164 , the momentum transition in the @xmath1 @xmath2 @xmath3 decay may be not large enough .
one might question whether the pqcd approach is applicable and whether the perturbative calculation is reliable .
therefore , it is necessary to check what percentage of the contributions comes from the perturbative region .
the contributions to branching ratio from different region of @xmath165 are showed in fig.([fig : br - as ] ) .
one can clearly see from fig.([fig : br - as ] ) that more than 90% contributions to branching ratio come from the @xmath165 @xmath166 @xmath167 region , and the contributions from nonperturbative region with large @xmath165 are highly suppressed .
one important reason is that assisting with the typical scale in eqs.([tab]-[tgh ] ) , the quark transverse momentum is retained and the sudakov factor is introduced to effectively suppress the nonperturbative contributions within the pqcd approach @xcite .
( horizontal axises ) , where the numbers over histogram denote the percentage of the corresponding contributions.,scaledwidth=50.0% ] \(5 ) there are many uncertainties on our results .
other factors , such as the contributions of higher order corrections to hme , relativistic effects and so on , which are not considered here , deserve the dedicated study . our results just provide an order of magnitude estimation .
the @xmath1 weak decay is legal within the standard model . with the potential prospects of the @xmath1 at high - luminosity dedicated heavy - flavor factories , the @xmath1 @xmath2 @xmath3 ,
weak decays are studied with the pqcd approach .
it is found that with the nonrelativistic wave functions for @xmath1 and @xmath34 mesons , branching ratios @xmath4 @xmath5 @xmath6 , which might be measurable in future experiments .
the direct @xmath0-violating asymmetry for the @xmath1 @xmath2 @xmath3 decay is close to zero because of the tiny weak phase difference .
we thank professor dongsheng du ( ihep@cas ) and professor yadong yang ( ccnu ) for helpful discussion .
we thank the referees for their constructive suggestions .
for the sake of simplicity , we decompose the decay amplitude eq.([amp - all ] ) into some building blocks @xmath168 , where the subscript @xmath123 on @xmath168 corresponds to the indices of fig.[fig1 ] ; the superscript @xmath125 on @xmath168 refers to one of the three possible dirac structures @xmath169 of the four - quark operator @xmath170 , namely @xmath125 @xmath10 @xmath126 for @xmath127 , @xmath125 @xmath10 @xmath128 for @xmath129 , and @xmath125 @xmath10 @xmath130 for @xmath131
. the explicit expressions of @xmath168 are written as follows .
@xmath171 @xmath172 \nonumber \\ & & + { \phi}_{\upsilon}^{t}(x_{1 } ) \big [ 2\,r_{2}\,x_{1}-r_{c } \big ] \big\ } \label{amp - figb-01 } , \\ % ----------------------------------------------------- { \cal a}_{b}^{sp } & = & -2\,r_{3}\ , { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{\infty}b_{1 } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2}\ , { \phi}_{b_{c}}(x_{2})\ , e_{b}(t_{b})\ , { \alpha}_{s}(t_{b } ) \nonumber \\ & & h_{ab}({\alpha}_{e},{\beta}_{b},b_{2},b_{1})\ , \big\ { { \phi}_{\upsilon}^{v}(x_{1})\ , ( 2\,r_{2}-r_{c } ) -{\phi}_{\upsilon}^{t}(x_{1})\,\bar{x}_{1}\big\ } \label{amp - figb-03 } , \end{aligned}\ ] ] @xmath173 + { \phi}_{\upsilon}^{t}(x_{1})\ , r_{2}\,(x_{2}-x_{1 } ) \big\ } \label{amp - figc-01 } , \\ % ----------------------------------------------------- { \cal a}_{c}^{sp } & = & -\frac{1}{n}\,r_{3 } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty}b_{3 } db_{3}\,{\alpha}_{s}(t_{c } ) \nonumber \\ & & { \delta}(b_{1}-b_{2})\ , { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{c}(t_{c})\ , h_{cd}({\alpha}_{e},{\beta}_{c},b_{2},b_{3 } ) \nonumber \\ & & \big\ { { \phi}_{\upsilon}^{v}(x_{1})\,r_{2}\ , ( \bar{x}_{3}-x_{2 } ) + { \phi}_{\upsilon}^{t}(x_{1})\ , ( x_{1}-\bar{x}_{3 } ) \big\ } \label{amp - figc-03 } , \end{aligned}\ ] ] @xmath174 + { \phi}_{\upsilon}^{t}(x_{1})\,r_{2}\,(x_{2}-x_{1 } ) \big\ } \label{amp - figd-01 } , \\ % ----------------------------------------------------- { \cal a}_{d}^{sp } & = & -\frac{1}{n}\,r_{3 } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty}b_{3 } db_{3}\,{\alpha}_{s}(t_{d } ) \nonumber \\ & & { \delta}(b_{1}-b_{2})\ , { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{d}(t_{d})\ , h_{cd}({\alpha}_{e},{\beta}_{d},b_{2},b_{3 } ) \nonumber \\ & & \big\ { { \phi}_{\upsilon}^{v}(x_{1})\ , r_{2}\ , ( r_{c}/r_{3}+x_{2}-x_{3 } ) + { \phi}_{\upsilon}^{t}(x_{1})\,(x_{3}-x_{1}-r_{c}/r_{3 } ) \big\ } \label{amp - figd-03 } , \end{aligned}\ ] ] @xmath175 -r_{b}\,{\phi}_{\upsilon}^{t}(x_{1 } ) \big\ } \label{amp - fige-01 } , \\ % ----------------------------------------------------- { \cal a}_{e}^{lr } & = & \frac{1}{n } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty}b_{1 } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty } db_{3}\ , { \delta}(b_{2}-b_{3 } ) \nonumber \\ & & { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{e}(t_{e})\,{\alpha}_{s}(t_{e})\ , h_{ef}({\alpha}_{a},{\beta}_{e},b_{1},b_{2})\ , \big\ { { \phi}_{\upsilon}^{v}(x_{1 } ) \nonumber \\ & & \big [ \frac{s\,(x_{2}-x_{1})}{m_{1}^{2 } } + 2\,r_{3}^{2}\ , ( \bar{x}_{3}-x_{1 } ) + r_{2}\,r_{3}\ , ( x_{2}-\bar{x}_{3 } ) \big ] + r_{b}\,{\phi}_{\upsilon}^{t}(x_{1 } ) \big\ } \label{amp - fige-02 } , \\ % ----------------------------------------------------- { \cal a}_{e}^{sp } & = & \frac{1}{n } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty}b_{1 } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty } db_{3}\ , { \delta}(b_{2}-b_{3 } ) \nonumber \\ & & { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{e}(t_{e})\,{\alpha}_{s}(t_{e})\ , h_{ef}({\alpha}_{a},{\beta}_{e},b_{1},b_{2 } ) \nonumber \\ & & \big\ { { \phi}_{\upsilon}^{v}(x_{1})\ , r_{b}\,(r_{2}+r_{3 } ) + { \phi}_{\upsilon}^{t}(x_{1 } ) \big [ r_{2}\,(x_{2}-x_{1 } ) + r_{3}\,(\bar{x}_{3}-x_{1 } ) \big ] \big\ } \label{amp - fige-03 } , \end{aligned}\ ] ] @xmath176 -r_{b}\,{\phi}_{\upsilon}^{t}(x_{1 } ) \big\ } \label{amp - figf-01 } , \\ % ----------------------------------------------------- { \cal a}_{f}^{lr } & = & \frac{1}{n } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty}b_{1 } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty } db_{3}\ , { \delta}(b_{2}-b_{3 } ) \nonumber \\ & & { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{f}(t_{f})\,{\alpha}_{s}(t_{f})\ , h_{ef}({\alpha}_{a},{\beta}_{e},b_{1},b_{2})\ , \big\ { { \phi}_{\upsilon}^{v}(x_{1 } ) \nonumber \\ & & \big [ \frac{s\,(x_{1}-x_{3})}{m_{1}^{2 } } + 2\,r_{2}^{2}\,(x_{2}-\bar{x}_{1 } ) + r_{2}\,r_{3}\,(\bar{x}_{3}-x_{2 } ) \big ] + r_{b}\,{\phi}_{\upsilon}^{t}(x_{1 } ) \big\ } \label{amp - figf-02 } , \\ % ----------------------------------------------------- { \cal a}_{f}^{sp } & = & \frac{1}{n } { \int}_{0}^{1}dx_{1 } { \int}_{0}^{1}dx_{2 } { \int}_{0}^{1}dx_{3 } { \int}_{0}^{\infty}b_{1 } db_{1 } { \int}_{0}^{\infty}b_{2 } db_{2 } { \int}_{0}^{\infty } db_{3}\ , { \delta}(b_{2}-b_{3 } ) \nonumber \\ & & { \phi}_{b_{c}}(x_{2})\ , { \phi}_{d_{s}}(x_{3},b_{3})\ , e_{f}(t_{f})\,{\alpha}_{s}(t_{f})\ , h_{ef}({\alpha}_{a},{\beta}_{e},b_{1},b_{2 } ) \nonumber \\ & & \big\ { { \phi}_{\upsilon}^{v}(x_{1})\ , r_{b}\,(r_{2}+r_{3 } ) + { \phi}_{\upsilon}^{t}(x_{1 } ) \big [ r_{2}\,(x_{2}-\bar{x}_{1 } ) + r_{3}\,(x_{1}-x_{3 } ) \big ] \big\ } \label{amp - figf-03 } , \end{aligned}\ ] ] @xmath177 where the mass ratio @xmath178 @xmath10 @xmath179 ; @xmath180 @xmath10 @xmath91 @xmath92 @xmath70 ; variable @xmath70 is the longitudinal momentum fraction of the valence quark ; @xmath181 is the conjugate variable of the transverse momentum @xmath182 ; and @xmath183 is the qcd coupling at the scale of @xmath66 .
the function @xmath184 are defined as follows .
\big\ } \nonumber \\ & { \times } & \big\ { { \theta}(b_{2}-b_{3 } ) k_{0}(\sqrt{-{\alpha}_{e}}b_{2 } ) i_{0}(\sqrt{-{\alpha}_{e}}b_{3 } ) + ( b_{2}{\leftrightarrow}b_{3 } ) \big\ } \label{hcd } , \\
% ----------------------------------------------------- h_{ef}({\alpha}_{a},{\beta},b_{1},b_{2 } ) & = & \big\ { { \theta}(-{\beta } ) k_{0}(\sqrt{-{\beta}}b_{1 } ) + \frac{{\pi}}{2 } { \theta}({\beta } ) \big [ ij_{0}(\sqrt{{\beta}}b_{1 } ) -y_{0}(\sqrt{{\beta}}b_{1 } ) \big ] \big\ } \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! { \times } \frac{{\pi}}{2 } \big\ { { \theta}(b_{1}-b_{2 } ) \big [ ij_{0}(\sqrt{{\alpha}_{a}}b_{1 } ) -y_{0}(\sqrt{{\alpha}_{a}}b_{1 } ) \big ] j_{0}(\sqrt{{\alpha}_{a}}b_{2 } ) + ( b_{1}{\leftrightarrow}b_{2 } ) \big\ } \label{hef } , \\
% ----------------------------------------------------- h_{hg}({\alpha}_{a},{\beta},b_{i},b_{j } ) & = & \frac{{\pi}^{2}}{4 } \big\ { ij_{0}(\sqrt{{\alpha}_{a}}b_{j } ) -y_{0}(\sqrt{{\alpha}_{a}}b_{j } ) \big\ } \nonumber \\ & { \times } & \big\ { { \theta}(b_{i}-b_{j } ) \big [ ij_{0}(\sqrt{{\beta}}b_{i } ) -y_{0}(\sqrt{{\beta}}b_{i } ) \big ] j_{0}(\sqrt{{\beta}}b_{j } ) + ( b_{i}{\leftrightarrow}b_{j } ) \big\ } \label{hgh } , \end{aligned}\ ] ] where @xmath186 and @xmath187 ( @xmath188 and @xmath189 ) are the ( modified ) bessel function of the first and second kind , respectively ; @xmath190 ( @xmath191 ) is the gluon virtuality of the emission ( annihilation ) diagrams ; the subscript of the quark virtuality @xmath104 corresponds to the indices of fig.[fig1 ] .
the definition of the particle virtuality is listed as follows . @xmath192
the typical scale @xmath153 and the sudakov factor @xmath193 are defined as follows , where the subscript @xmath123 corresponds to the indices of fig.[fig1 ] .
@xmath194 @xmath195 @xmath196 where @xmath197 @xmath10 @xmath198 is the quark anomalous dimension ; the explicit expression of @xmath199 can be found in the appendix of ref.@xcite .
99 k. olive _
_ ( particle data group ) , chin
c 38 , 090001 ( 2014 ) .
s. okubo , phys .
lett . 5 , 165 ( 1963 ) . g. zweig , cern - th-401 , 402 , 412 ( 1964 ) .
j. iizuka , prog .
phys . suppl .
37 - 38 , 21 ( 1966 ) . c. patrignani , t. pedlar , and j. rosner , annu .
nucl . part .
63 , 21 ( 2013 ) .
h. li , phys .
d 52 , 3958 ( 1995 ) .
c. chang , h. li , phys .
d 55 , 5577 ( 1997 ) .
t. yeh , h. li , phys .
d 56 , 1615 ( 1997 ) .
j. brodzicka _
et al . _ ( belle collaboration ) , prog .
2012 , 04d001
. m. sanchis - lozano , z. phys .
c 62 , 271 ( 1994 ) .
m. beneke _
b 591 , 313 ( 2000 ) . c. bauer _ et al . _ ,
phys . rev .
d 63 , 114020 ( 2001 ) . c. bauer , d. pirjol , i. stewart , phys .
d 65 , 054022 ( 2002 ) . c. bauer _
et al . _ ,
d 66 , 014017 ( 2002 ) .
m. beneke _
et al . _ ,
b 643 , 431 ( 2002 ) .
g. buchalla , a. buras , m. lautenbacher , rev .
68 , 1125 , ( 1996 ) .
s. catani , m. ciafaloni , and f. hautmann , nucl .
b 366 , 135 ( 1991 ) .
g. lepage , s. brodsky , phys .
d 22 , 2157 ( 1980 ) .
r. li , c. l , h. zou , phys . rev .
d 78 , 014018 ( 2008 ) .
g. lepage _ et al .
d 46 , 4052 ( 1992 ) .
g. bodwin , e. braaten , g. lepage , phys . rev .
d 51 , 1125 ( 1995 ) .
n. brambilla _
et al . _ ,
77 , 1423 ( 2005 ) . j. sun _ et al .
_ , phys . rev .
d 92 , 074028 ( 2015 ) .
t. chiu , t. hsieh , c. huang , k. ogawa , phys .
b 651 , 171 ( 2007 ) .
r. aaij _ et al . _
( lhcb collaboration ) , jhep 1407 , 094 ( 2014 ) .
b. abelev _
( alice collaboration ) , phys .
b 740 , 105 ( 2015 ) . | the branching ratio and direct @xmath0 asymmetry of the @xmath1 @xmath2 @xmath3 weak decay are estimated with the perturbative qcd approach firstly .
it is found that ( 1 ) the direct @xmath0-violating asymmetry is close to zero .
( 2 ) the branching ratio @xmath4 @xmath5 @xmath6 might be measurable at the future experiments . |
the main aim of this article is an attempt to derive a new theory for dark matter and dark energy , and to derive a new set of gravitational field equations .
the primary motivation of this study is the great mystery of the dark matter and dark energy .
the natural starting point for this study is to fundamentally examine the einstein field equations , given as follows : @xmath7 where @xmath8 is the ricci curvature tensor , @xmath9 is the scalar curvature , @xmath1 is the riemannian metric of the space - time , and @xmath5 is the energy - momentum tensor of matter ; see among many others @xcite .
the einstein equations can also be derived using the principle of lagrangian dynamics to the einstein - hilbert functional : @xmath10 whose euler - lagrangian is exactly @xmath11 , which is the left hand side of the einstein field equations ( [ 1.1 ] ) .
it is postulated that this euler - lagrangian is balanced by the symmetric energy - momentum tensor of matter , @xmath5 , leading to the einstein field equations ( [ 1.1 ] ) .
the bianchi identity implies that the left hand side of the einstein equations is divergence - free , and it is then postulated and widely accepted that the energy - momentum tensor of matter @xmath5 is divergence - free as well .
however , there are a number of difficulties for the einstein field equations : first , the einstein field equations failed to explain the dark matter and dark energy , and the equations are inconsistent with the accelerating expansion of the galaxies . in spite of many attempts to modify the einstein gravitational field equation to derive a consistent theory for the dark energy ,
the mystery remains .
second , we can prove that there is no solution for the einstein field equations for the spherically symmetric case with cosmic microwave background ( cmb ) .
one needs clearly to resolve this inconsistency caused by the non - existence of solutions .
third , from the einstein equations ( [ 1.1 ] ) , it is clear that @xmath12 where @xmath13 is the energy - momentum density .
a direct consequence of this formula is that the discontinuities of @xmath14 give rise to the same discontinuities of the curvature and the discontinuities of space - time .
this is certainly an inconsistency which needs to be resolved .
fourth , it has been observed that the universe is highly non - homogeneous as indicated by e.g. the `` great walls '' , filaments and voids .
however , the einstein equations do not appear to offer a good explanation of this inhomogeneity .
these observations strongly suggest that further fundamental level examinations of the einstein equations are inevitably necessary .
it is clear that any modification of the einstein field equations should obey three basic principles : * the principle of equivalence , * the principle of general relativity , and * the principle of lagrangian dynamics .
the first two principles tell us that the spatial and temporal world is a 4-dimensional riemannian manifold @xmath15 , where the metric @xmath16 represents gravitational potential , and the third principle determines that the riemannian metric @xmath16 is an extremum point of the lagrangian action .
there is no doubt that the most natural lagrangian in this case is the einstein - hilbert functional as explained in many classical texts of general relativity .
the key observation for our study is a well - known fact that the riemannian metric @xmath1 is divergence - free .
this suggests two important postulates for deriving a new set of gravitational field equations : * the energy - momentum tensor @xmath5 of matter need not to be divergence - free due to the presence of dark energy and dark matter ; and * the field equations obey the euler - lagrange equation of the einstein - hilbert functional under the natural divergence - free constraint , with divergence defined at the extremum riemannian metric @xmath17 : @xmath18 here @xmath19 is the contra - variant derivative with respect to the extremum point @xmath17 , and @xmath20 are the variational elements .
namely , for any @xmath21 with @xmath22 , @xmath23=(\delta f(g_{ij } ) , x)=0.\ ] ] for this purpose , an important part of this article is to drive an orthogonal decomposition theorem of tensors on riemannian manifolds , which we shall explain further in the last part of this introduction . under these two postulates , using the orthogonal decomposition theorem of tensors , we derive the following new set of gravitational field equations with scalar potential : @xmath24 where the scalar function @xmath25 is called the scalar potential .
the corresponding conservations of mass , energy and momentum are then replaced by @xmath26 and the energy - momentum density @xmath27 and the scalar potential energy density @xmath28 satisfy @xmath29 the scalar potential energy density @xmath30 has a number of important physical properties : * gravitation is now described by the riemannian metric @xmath1 , the scalar potential @xmath0 and their interactions , unified by the new gravitational field equations ( [ 1.3 ] ) .
* this scalar potential energy density @xmath30 represents a new type of energy / force caused by the non - uniform distribution of matter in the universe .
this scalar potential energy density varies as the galaxies move and matter of the universe redistributes . like gravity
, it affects every part of the universe as a field . * this scalar potential energy density @xmath30 consists of both positive and negative energies .
the negative part of this potential energy density produces attraction , and the positive part produces repelling force
. the conservation law ( [ 1.6 ] ) amounts to saying that the the total scalar potential energy density is conserved . *
the sum of this new potential energy density @xmath4 and the coupling energy between the energy - momentum tensor @xmath5 and the scalar potential field @xmath0 , as described e.g. by the second term in the right - hand side of ( [ 1.9 - 1 ] ) , gives rise to a new unified theory for dark matter and dark energy : the negative part of @xmath31 represents the dark matter , which produces attraction , and the positive part represents the dark energy , which drives the acceleration of expanding galaxies . * the scalar curvature of space - time obeys ( [ 1.5 ] ) .
consequently , when there is no normal matter present ( with @xmath32 ) , the curvature @xmath9 of space - time is balanced by @xmath33 .
therefore , there is no real vacuum in the universe . *
the universe with uniform distributed matter leads to identically zero scalar potential energy , and is unstable .
it is this instability that leads to the existence of the dark matter and dark energy , and consequently the high non - homogeneity of the universe .
hereafter , we further explore a few direct consequences of the above new gravitational field equations .
first , the new field equations are consistent with the spherically symmetric case with cosmic microwave background ( cmb ) .
namely , the existence of solutions in this case can be proved .
second , our new theory suggests that the curvature @xmath9 is always balanced by @xmath34 in the entire space - time by ( [ 1.5 ] ) , and the space - time is no longer flat .
namely the entire space - time is also curved and is filled with dark energy and dark matter . in particular , the discontinuities of @xmath9 induced by the discontinuities of the energy - momentum density @xmath14 , dictated by the einstein field equations , are no longer present thanks to the balance of @xmath35 .
third , this scalar potential energy density should be viewed as the main cause for the non - homogeneous distribution of the matter / galaxies in the universe , as the dark matter ( negative scalar potential energy ) attracts and dark energy ( positive scalar potential energy ) repels different galaxies ; see ( [ 1.9 - 1 ] ) below .
fourth , to further explain the dark matter and dark energy phenomena , we consider a central matter field with total mass @xmath36 and radius @xmath37 and spherical symmetry .
with spherical coordinates , the corresponding riemannian metric must be of the following form : @xmath38 where @xmath39 and @xmath40 are functions of the radial distance . with the new field equations , the force exerted on an object with mass @xmath41
is given by @xmath42 , \qquad r = \phi \qquad \text { for } r > r_0.\la{1.9 - 1}\ ] ] where @xmath43 , @xmath9 is the scalar curvature , and @xmath0 is the scalar potential . the first term is the classical newton gravitation , the second term is the coupling interaction between matter and the scalar potential @xmath0 , and the third term is the interaction generated by the scalar potential energy density @xmath30 as indicated in ( [ 1.5 ] ) ( @xmath33 for @xmath44 ) . in this formula ,
the negative and positive values of each term represent respectively the attracting and repelling forces .
it is then clear that the combined effect of the second and third terms in the above formula represent the dark matter , dark energy and their interactions with normal matter . also , importantly
, this formula is a direct representation of the einstein s equivalence principle .
namely , the curvature of space - time induces interaction forces between matter .
in addition , one can derive a more detailed version of the above formula : @xmath45,\la{f0}\ ] ] where @xmath46 .
the conservation law ( [ 1.6 ] ) of @xmath34 suggests that @xmath9 behaviors as @xmath47 for @xmath48 sufficiently large .
consequently the second term in the right hand side of ( [ f0 ] ) must dominate and be positive , indicating the existence of dark energy .
in fact , the above formula can be further simplified to derive the following approximate formula : @xmath49 , \la{1.9}\\ & k_0=4 \times 10^{-18 } km^{-1 } , \qquad k_1=10^{-57 } km^{-3}.\la{1.10}\end{aligned}\ ] ] again , in ( [ 1.9 ] ) , the first term represents the newton gravitation , the attracting second term stands for dark matter and the repelling third term is the dark energy .
the mathematical part of this article is devoted to a rigorous derivation of the new gravitational field equations .
first , as mentioned earlier , the field equations obey the euler - lagrange equation of the einstein - hilbert functional under the natural divergence - free constraint , with divergence defined at the extremum riemannian metric @xmath17 : @xmath50 as the variational elements @xmath51 are divergence - free , ( [ el ] ) does not imply @xmath52 , which is the classical einstein equations .
in fact , ( [ el ] ) amounts to saying that @xmath53 is orthogonal to all divergence - free tensor fields @xmath51 .
hence we need to decompose general tensor fields on riemannian manifolds into divergence - free and gradient parts . for this purpose ,
an orthogonal decomposition theorem is derived in theorem [ t3.1 ] . in particular , given an @xmath54 tensor field @xmath55 , we have @xmath56 the gradient part is acting on an @xmath57 tensor field @xmath58 .
second , restricting to a @xmath59 symmetric tensor field @xmath60 , the gradient part in the above decomposition is given by @xmath61 then using symmetry , we show in theorem [ t3.4 ] that this @xmath62 tensor @xmath58 can be uniquely determined , up to addition to constants , by the gradient of a scalar field @xmath0
: @xmath63 and consequently we obtain the following decomposition for general symmetric @xmath59 tensor fields : @xmath64 finally , for the symmetric and divergence free @xmath59 field @xmath53 , which is the euler - lagrangian of the einstein - hilbert functional in ( [ el ] ) and is orthogonal to all divergence - free fields , there is a scalar field @xmath65 such that @xmath66 which , by adding the energy - momentum tensor @xmath5 , leads to the new gravitational field equations ( [ 1.3 ] ) .
we remark here that the orthogonal decompositions ( [ rsd ] ) and ( [ 02d ] ) are reminiscent of the orthogonal decomposition of vectors fields into gradient and divergence parts , which are crucial for studying incompressible fluid flows ; see among many others @xcite .
this article is divided into two parts .
the physically inclined readers can go directly to the physics part after reading this introduction .
let @xmath15 be an @xmath67-dimensional riemannian manifold with metric @xmath68 , and @xmath69 be an @xmath70-tensor bundle on @xmath36 .
a mapping @xmath71 is called a section of the tensor - bundle @xmath72 or a tensor field . in a local coordinate system @xmath73 , a tensor field @xmath60
can be expressed component - wise as follows : @xmath74 where @xmath75 are functions of @xmath76 . the section @xmath60 is called @xmath77-tensor field or @xmath77-section if its components are @xmath77-functions . for any real number @xmath78 ,
let @xmath79 be the space of all @xmath80-integrable sections of @xmath72 : @xmath81 equipped with the norm @xmath82^{{1}/{p}}=\left[\int_m\sum|u^{j_1\cdots j_r}_{i_1\cdots i_s}|^pdx\right]^{1/p}.\ ] ] for @xmath83 , @xmath84 is a hilbert space equipped with the inner product @xmath85 where @xmath68 is riemannian metric , @xmath86 , @xmath87 , and @xmath88 is the volume element . for any positive integer @xmath89 and any real number @xmath90 , we can also define the sobolev spaces @xmath91 to be the subspace of @xmath79 such that all covariant derivatives of u up to order @xmath89 are in @xmath79 .
the norm of @xmath91 is always denoted by @xmath92 . as @xmath83 ,
the spaces @xmath91 are hilbert spaces , and are usually denoted by @xmath93 equipped with inner product @xmath94 and norm @xmath95 .
let @xmath96 be an @xmath70-tensor field , with the local expression @xmath97 then the gradient of @xmath60 is defined as @xmath98 where @xmath99 is the covariant derivative .
it is clear that the gradient @xmath100 defined by ( [ ( 2.6 ) ] ) is an @xmath101-tensor field : @xmath102 we define @xmath103 as @xmath104 for an @xmath105-tensor field @xmath106 the divergence of @xmath60 is defined by @xmath107 therefore , the divergence @xmath108 defined by ( [ ( 2.8 ) ] ) is an @xmath70-tensor field .
likewise , for an @xmath101-tensor field @xmath109 the following operator is also called the divergence of @xmath60 , @xmath110 where @xmath111 , which is an @xmath70-tensor field . for the gradient operators ( [ ( 2.6)])-([(2.7 ) ] ) and the divergent operators ( [ ( 2.8)])-([(2.9 ) ] ) , it is well known that the following integral formulas hold true ; see among others @xcite .
let @xmath15 be a closed riemannian manifold .
if @xmath60 is an @xmath112-tensor and @xmath113 is an @xmath70 tensor , then we have @xmath114 where @xmath103 is as in ( [ ( 2.7 ) ] ) and @xmath115 is as in ( [ ( 2.8 ) ] ) , the inner product @xmath116 is as defined by ( [ ( 2.1 ) ] ) . if @xmath60 is an @xmath117-tensor and @xmath113 is an @xmath70 tensor , then @xmath118 where @xmath100 is as in ( [ ( 2.6 ) ] ) and @xmath119 is as in ( [ ( 2.9 ) ] ) .
let @xmath120 be a hilbert space equipped with inner product @xmath121 and norm @xmath122 , and @xmath123 be a mapping .
we say that @xmath124 is weakly continuous if for any sequence @xmath125 weakly converging to @xmath126 , i.e. @xmath127 we have @xmath128 if the operator @xmath124 is linear and bounded , then @xmath124 is weakly continuous .
the following theorem is called the acute - angle principle @xcite .
if a mapping @xmath123 is weakly continuous , and satisfies @xmath129 for some constants @xmath130 , then for any @xmath131 there is a @xmath132 such that @xmath133
the aim of this section is to derive an orthogonal decomposition for @xmath70-tensor fields with @xmath134 into divergence - free and gradient parts .
this decomposition plays a crucial role for the theory of gravitational field , dark matter and dark energy developed later in this article .
let @xmath36 be a closed riemannian manifold , and @xmath135 @xmath136 .
we say that @xmath113 is divergence - free , i.e. , @xmath137 , if for all @xmath138 , @xmath139 here @xmath140 or @xmath141 , and @xmath116 is the @xmath142-inner product defined by ( [ ( 2.1 ) ] ) .
we remark that if @xmath143 satisfies ( [ ( 3.1 ) ] ) , then @xmath113 is weakly differentiable , and @xmath144 in the @xmath142-sense . if @xmath145 is not differential , then ( [ ( 3.1 ) ] ) means that @xmath113 is divergence - free in the distribution sense .
let @xmath36 be a closed riemannian manifold , and @xmath146 with @xmath147 .
the following assertions hold true : * the tensor field @xmath60 has the following orthogonal decomposition : @xmath148 where @xmath149 or @xmath150 , and @xmath151 . *
if @xmath36 is compact , then @xmath60 can be orthogonally decomposed into @xmath152 where @xmath0 and @xmath113 are as in ( [ ( 3.2 ) ] ) , and @xmath153 is a harmonic field , i.e. @xmath154 in particular the subspace of all harmonic tensor fields in @xmath155 is of finite dimensional : @xmath156 we proceed in several steps as follows .
step 1 proof of assertion ( 1 ) .
let @xmath157 , @xmath69 @xmath136 .
consider the equation @xmath158 where @xmath159 is the laplace operator defined by @xmath160 without loss of generality , we only consider the case where @xmath161 .
it is clear that if the equation ( [ ( 3.6 ) ] ) has a solution @xmath162 , then by ( [ ( 3.7 ) ] ) , the following vector field must be divergence - free @xmath163 moreover , by ( [ ( 3.1 ) ] ) we have @xmath164 namely @xmath113 and @xmath165 are orthogonal .
therefore , the orthogonal decomposition @xmath166 follows from ( [ ( 3.8 ) ] ) and ( [ ( 3.9 ) ] ) .
it suffices then to prove that ( [ ( 3.6 ) ] ) has a weak solution @xmath162 : @xmath167 to this end , let @xmath168 then we define a linear operator @xmath169 by @xmath170 it is clear that the linear operator @xmath171 is bounded , weakly continuous , and @xmath172 based on theorem [ t2.3 ] , for any @xmath131 , the equation @xmath173 has a weak solution @xmath174 .
hence for @xmath175 the equation ( [ ( 3.6 ) ] ) has a solution , and assertion ( 1 ) is proved .
in fact the solution of ( [ ( 3.6 ) ] ) is unique .
we remark that by the poincar inequality , for the space @xmath176 , ( [ ( 3.12 ) ] ) is an equivalent norm of @xmath120 .
in addition , by theorem [ t2.1 ] , the weak formulation ( [ ( 3.10 ) ] ) for ( [ ( 3.6 ) ] ) is well - defined .
step 2 proof of assertion ( 2 ) . based on assertion ( 1 )
, we have @xmath177 where @xmath178 define an operator @xmath179 by @xmath180 where @xmath181 is the canonical orthogonal projection . we know that the laplace operator @xmath159 can be expressed as @xmath182 where @xmath183 is the lower order derivative linear operator . since @xmath36 is compact ,
the sobolev embeddings @xmath184 are compact , which implies that the lower order derivative operator @xmath185 where the integer @xmath186 is the dimension of the tensor bundle @xmath72 . according to the elliptic operator theory , the elliptic operator in ( [ ( 3.14 ) ] )
@xmath187 is a linear homeomorphism .
therefore the operator in ( [ ( 3.14 ) ] ) is a linear completely continuous field @xmath188 which implies that the operator of ( [ ( 3.13 ) ] ) is also a linear completely continuous field : @xmath189 by the the spectral theorem of completely continuous fields @xcite , the space @xmath190 is finite dimensional , and is the eigenspace of the eigenvalue @xmath191 . by theorem [ t2.1 ] , for @xmath192 @xmath193 it follows that @xmath194 where @xmath120 is the harmonic space as in ( [ ( 3.5 ) ] ) .
thus we have @xmath195 the proof of theorem [ t3.1 ] is complete . in theorem [ t3.1 ] ,
a tensor field @xmath146 with @xmath134 can be orthogonally decomposed into @xmath196 now we address the uniqueness problem of the decomposition ( [ ( 3.15 ) ] ) .
in fact , if @xmath60 is a vector field or a co - vector field , i.e. @xmath197 then the decomposition of ( [ ( 3.15 ) ] ) is unique .
we can see that if @xmath146 with @xmath198 , then there are different types of the decompositions of ( [ ( 3.15 ) ] ) .
for example , for @xmath199 , the local expression of @xmath60 is given by @xmath200 in this case , @xmath60 has two types of decompositions : @xmath201 it is easy to see that if @xmath202 then both ( [ ( 3.16 ) ] ) and ( [ ( 3.17 ) ] ) are two different decompositions of @xmath203 .
namely @xmath204
if @xmath205 is symmetric , @xmath60 can be orthogonally decomposed into the following two forms : @xmath206 and @xmath0 and @xmath58 satisfy @xmath207 by @xmath205 we have @xmath208 .
hence , ( [ ( 3.18 ) ] ) and ( [ ( 3.19 ) ] ) are the same , and @xmath209 .
therefore , the symmetric tensors @xmath203 can be written as @xmath210 from ( [ ( 3.20)])-([(3.21 ) ] ) we can deduce the following theorem .
let @xmath199 be symmetric , i.e. @xmath205 , and the first betti number @xmath211 for @xmath36 .
then the following assertions hold true : * @xmath60 has a unique orthogonal decomposition if and only if there is a scalar function @xmath212 such that @xmath60 can be expressed as @xmath213 * @xmath60 can be orthogonally decomposed in the form of ( [ ( 3.22 ) ] ) if and only if @xmath203 satisfy @xmath214 where @xmath215 and @xmath8 are the ricci curvature tensors . *
if @xmath216 in ( [ ( 3.20 ) ] ) is symmetric : @xmath217 , then @xmath60 can be expressed by ( [ ( 3.22 ) ] ) .
we only need to prove assertions ( 2 ) and ( 3 ) . we first prove assertion ( 2 ) .
it follows from ( [ ( 3.20 ) ] ) that @xmath218 where @xmath219 . by the weitzenbck formula @xcite , @xmath220 and @xmath221
is the laplace - beltrami operator . we know that for @xmath222 , @xmath223 where @xmath224 is the gradient operator , and @xmath225 namely @xmath226 hence , we infer from ( [ ( 3.25 ) ] ) that @xmath227 thus , by ( [ ( 3.24 ) ] ) and ( [ ( 3.26 ) ] ) , we obtain that @xmath228 holds true if and only if the tensor @xmath229 in ( [ ( 3.20 ) ] ) is a gradient @xmath230 .
assertion ( 2 ) is proven .
now we verify assertion ( 3 ) .
since @xmath216 in ( [ ( 3.20 ) ] ) is symmetric , we have @xmath231 note that @xmath232 where @xmath233 is the levi - civita connection , and @xmath234 .
we infer then from ( [ ( 3.27 ) ] ) that @xmath235 by assumption , the 1-dimensional homology of @xmath36 is zero , @xmath236 and it follows from the de rham theorem and ( [ ( 3.29 ) ] ) that @xmath237 for some scalar function @xmath0 .
thus assertion ( 3 ) follows and the proof is complete .
hereafter we always assume that @xmath36 is a closed manifold .
a riemannian metric @xmath124 on @xmath36 is a mapping @xmath238 which is symmetric and nondegenerate , i.e. , in a local coordinate @xmath239 can be expressed as @xmath240 and the matrix @xmath68 is invertible on @xmath36 : @xmath241 if we regard a riemannian metric @xmath242 as a tensor field on manifold @xmath36 , then the set of all metrics @xmath242 on @xmath36 constitutes a topological space , called the space of riemannian metrics on @xmath36 .
we denote @xmath243 the space for riemannian metrics on @xmath36 is denied by @xmath244 which is a metric space , but not a banach space . however , it is a subspace of the direct sum of two sobolev spaces @xmath245 and @xmath246 : @xmath247 a functional defined on @xmath248 @xmath249 is called the functional of riemannian metric .
usually , the functional ( [ ( 4.6 ) ] ) can be expressed as @xmath250 since @xmath251 is the inverse of @xmath68 , we have @xmath252 therefore , @xmath253 in ( [ ( 4.7 ) ] ) also depends on @xmath254 , i.e. putting ( [ ( 4.8 ) ] ) in ( [ ( 4.7 ) ] ) we get @xmath255 we note that although @xmath256 is not a linear space , for a given element @xmath257 and any symmetric tensor @xmath20 and @xmath258 , there is a number @xmath259 such that @xmath260 due to ( [ ( 4.10 ) ] ) , we can define the derivative operators of the functional @xmath261 , which are also called the euler - lagrange operators of @xmath261 , as follows @xmath262 where @xmath263 is the dual space of @xmath264 , and @xmath265 are given by @xmath266 for any given metric @xmath267 , the value of @xmath268 and @xmath269 at @xmath1 are second - order contra - variant and covariant tensor fields respectively , i.e. @xmath270 moreover , the equations @xmath271 are called the euler - lagrange equations of @xmath261 , and the solutions of ( [ ( 4.15 ) ] ) and ( [ ( 4.16 ) ] ) are called the extremum points or critical points of @xmath261 .
let @xmath261 be the functionals defined by ( [ ( 4.6 ) ] ) and ( [ ( 4.9 ) ] ) .
then the following assertions hold true : * for any @xmath272 and @xmath273 are symmetric tensor fields . *
if @xmath274 is the extremum point of @xmath261 , then @xmath275 is also an extremum point of @xmath261 , i.e. @xmath16 satisfies ( [ ( 4.15 ) ] ) and ( [ ( 4.16 ) ] ) if and only if @xmath275 satisfies ( [ ( 4.15 ) ] ) and ( [ ( 4.16 ) ] ) . *
@xmath268 and @xmath269 have the following relation @xmath276 where @xmath277 and @xmath278 are the components of @xmath269 and @xmath268 respectively .
we only need to verify assertion ( 3 ) .
noting that @xmath279 we have the variational relation @xmath280 it implies that @xmath281 in addition , in ( [ ( 4.13 ) ] ) and ( [ ( 4.14 ) ] ) , @xmath282 therefore , by ( [ ( 4.17 ) ] ) we get @xmath283 hence @xmath284 thus assertion ( 3 ) follows and the proof is complete .
we know that the critical points of the functional @xmath261 in ( [ ( 4.6 ) ] ) are the solution @xmath285 in the following sense @xmath286 where @xmath287 . hence , the critical points of functionals of riemannian metrics are not solutions of ( [ ( 4.18 ) ] ) in the usual sense .
it is easy to see that @xmath288 can be orthogonally decomposed into the direct sum of the symmetric and contra - symmetric spaces , i.e. @xmath289 since @xmath290 is symmetric , by ( [ ( 4.20 ) ] ) the extremum points @xmath16 of @xmath261 satisfy more general equality @xmath291 thus , we can say that the extremum points of functionals of the riemannian metrics are solutions of ( [ ( 4.18 ) ] ) in the usual sense of ( [ ( 4.21 ) ] ) , or are zero points of the variational operators @xmath292 now we consider the variations of @xmath261 under the divergence - free constraint . in this case , the euler - lagrangian equations with symmetric divergence - free constraints are equivalent to the euler - lagrangian equations with general divergence - free constraints . hence we have the following definition .
let @xmath293 be a functional of riemannian metric .
a metric tensor @xmath294 is called an extremum point of @xmath261 with divergence - free constraint , if @xmath16 satisfies @xmath295 where @xmath296 is the space of all divergence - free tensors : @xmath297 it is clear that an extremum point satisfying ( [ ( 4.22 ) ] ) is not a solution of ( [ ( 4.18 ) ] ) .
instead , we have the scalar potential theorem for the extremum points of divergence free constraint ( [ ( 4.22 ) ] ) , which is based on the orthogonal decomposition theorems .
this result is also crucial for the gravitational field equations and the theory of dark matter and dark energy developed later .
assume that the first betti number of @xmath36 is zero , i.e. @xmath211 . let @xmath261 be a functional of the riemannian metric .
then there is a @xmath298 such that the extremum points @xmath16 of @xmath261 with divergence - free constraint satisfy @xmath299 let @xmath16 be an extremum point of @xmath261 under the constraint ( [ ( 4.22 ) ] ) .
namely , @xmath53 satisfies @xmath300 by theorem [ t3.1 ] , @xmath301 can be orthogonally decomposed as @xmath302 by theorem [ t2.1 ] , for any @xmath303 , @xmath304 therefore it follows from ( [ ( 4.24)])-([(4.26 ) ] ) that @xmath305 let @xmath306 .
since @xmath307 we have @xmath308 thanks to @xmath309 .
inserting @xmath306 into ( [ ( 4.27 ) ] ) leads to @xmath310 which implies that @xmath311 .
thus , ( [ ( 4.25 ) ] ) becomes @xmath312 by theorem [ t4.1 ] , @xmath290 is symmetric .
hence we have @xmath313 it follows from ( [ ( 3.28 ) ] ) that @xmath314 by assumption , the first betti number of @xmath36 is zero , i.e. the 1-dimensional homology of @xmath36 is zero : @xmath315 .
it follows from the de rham theorem that if @xmath316 then there exists a scalar function @xmath0 such that @xmath317 thus , we infer from ( [ ( 4.29 ) ] ) that @xmath318 therefore we get ( [ ( 4.23 ) ] ) from ( [ ( 4.28 ) ] ) .
the theorem is proved . if the first betti number @xmath319 , then there are @xmath320 number of 1-forms : @xmath321 which constitute a basis of the 1-dimensional de rham homology @xmath322 . we know that the components of @xmath323 are co - vector fields : @xmath324 which possess the following properties : @xmath325 or equivalently , @xmath326 namely , @xmath327 are symmetric second - order contra - variant tensors . hence theorem [ t4.3 ] can be extended to the non - vanishing first betti number case as follows .
let the first betti number @xmath319 for @xmath36 .
then for the functional @xmath261 of riemannian metrics , the extremum points @xmath16 of @xmath261 with the constraint ( [ ( 4.22 ) ] ) satisfy the equations @xmath328 where @xmath329 are constants , @xmath212 , and the tensors @xmath330 are as given by ( [ ( 4.31 ) ] ) .
the proof of theorem [ t4.4 ] is similar to theorem [ t4.3 ] , and is omitted here .
the general theory of relativity is based on three basic principles : the principle of equivalence , the principle of general relativity , and the principle of lagrangian dynamics .
the first two principles tell us that the spatial and temporal world is a 4-dimensional riemannian manifold @xmath15 , where the metric @xmath16 represents gravitational potential , and the third principle determines that the riemannian metric @xmath16 is an extremum point of the lagrangian action , which is the einstein - hilbert functional .
let @xmath15 be an @xmath67-dimensional riemannian manifold .
the einstein - hilbert functional @xmath331 is defined by @xmath332 where @xmath333 is defined by ( [ ( 4.5 ) ] ) , @xmath334 and @xmath335 are the scalar and the ricci curvatures , @xmath336 is the stress tensor , @xmath124 is the gravitational constant , and @xmath337 is the speed of light .
the euler - lagrangian of the einstein - hilbert functional @xmath261 is given by @xmath338 where @xmath5 is the energy - momentum tensor given by @xmath339 and the ricci curvature tensor @xmath8 is given by @xmath340 by ( [ ( 5.3)])-([(5.5 ) ] ) , the euler - lagrangian @xmath53 of the einstein - hilbert functional is a second order differential operator on @xmath16 , and @xmath341 is symmetric .
the general theory of relativity consists of two main conclusions : 1 .
the space - time of our world is a 4-dimensional riemannian manifold @xmath342 , and the metric @xmath16 represents gravitational potential .
the metric @xmath16 is the extremum point of the einstein - hilbert functional ( [ ( 5.2 ) ] ) . in other words ,
gravitational field theory obeys the principle of lagrange dynamics .
the principle of lagrange dynamics is a universal principle , stated as : * principle of lagrange dynamics . * _ for any physical system , there are a set of state functions @xmath343 which describe the state of this system , and there exists a functional @xmath344 of @xmath60 , called the lagrange action : @xmath345 such that the state @xmath60 is an extremum point of @xmath344 .
usually the function @xmath346 in ( [ ( 5.6 ) ] ) is called the lagrangian density .
_ based on this principle , the gravitational field equations are the euler - lagrange equations of the einstein - hilbert functional : @xmath347 which are the classical einstein field equations : @xmath348 by the bianchi identities , the left hand side of ( [ ( 5.8 ) ] ) is divergence - free , i.e. @xmath349 therefore it is required in the general theory of relativity that the energy - momentum tensor @xmath350 in ( [ ( 5.8 ) ] ) satisfies the following energy - momentum conservation law : @xmath351 motivated by the mystery of dark energy and dark matter and the other difficulties encountered by the einstein field equations as mentioned in introduction , we introduce in this section a new set of field equations , still obeying the three basic principles of the general theory of relativity . our key observation is a well - known fact that the riemannian metric @xmath1 is divergence - free .
this suggests us two important postulates for deriving the new gravitational field equations : * the energy - momentum tensor of matter need not to be divergence - free due to the presence of dark energy and dark matter ; and * the field equation obeys the euler - lagrange equation of the einstein - hilbert functional under the natural divergence - free constraint . under these two postulates , by the scalar potential theorem , theorem [ t4.3 ] , if the riemannian metric @xmath16 is an extremum point of the einstein - hilbert functional ( [ ( 5.2 ) ] ) with the divergence - free constraint ( [ ( 4.22 ) ] ) , then the gravitational field equations are taken in the following form : @xmath352 where @xmath212 is called the scalar potential .
we infer from ( [ ( 5.9 ) ] ) that the conservation laws for ( [ ( 5.13 ) ] ) are as follows @xmath353 using the contraction with @xmath254 in ( [ ( 5.13 ) ] ) , we have @xmath354 where @xmath355 represent respectively the energy - momentum density and the scalar potential density .
physically this scalar potential density @xmath34 represents potential energy caused by the non - uniform distribution of matter in the universe .
one important property of this scalar potential is @xmath356 which is due to the integration by parts formula in theorem [ t2.1 ] .
this formula demonstrates clearly that the negative part of this quantity @xmath34 represents the dark matter , which produces attraction , and the positive part represents the dark energy , which drives the acceleration of expanding galaxies .
we shall address this important issue in the next section .
the topological structure of closed universe is given by @xmath357 where @xmath358 is the time circle and @xmath359 is the 3-dimensional sphere representing the space .
we note that the radius @xmath9 of @xmath359 depends on time @xmath360 , @xmath361 and the minimum time @xmath362 , @xmath363 is the initial time of the big bang . for a closed universe as ( [ 4.3 ] ) , by theorem [ t4.3 ] ,
the gravitational field equations are in the form @xmath364 where @xmath365 the scalar potential , @xmath366 the vector potential , and @xmath367 is a constant . the conservation laws of ( [ 4.4 ] ) are as follows @xmath368
we know that the metric of a central gravitational field is in a diagonal form @xcite : @xmath369 and physically @xmath370 is given by @xmath371 where @xmath58 is the newton gravitational potential ; see among others @xcite . if the central matter field has total mass @xmath36 and radius @xmath37 , then for @xmath44 , the metric ( [ 5.1 ] ) is the well known schwarzschild solution for the einstein field equations ( [ ( 5.8 ) ] ) , and is given by @xmath372 we derive from ( [ 5.2 ] ) and ( [ 5.3 ] ) the classical newton gravitational potential @xmath373 we now consider the metric determined by the new field equations ( [ ( 5.13 ) ] ) , from which we derive a gravitational potential formula replacing ( [ 5.4 ] ) .
equations ( [ ( 5.13 ) ] ) can be equivalently expressed as @xmath374 where @xmath375 for the central matter field with total mass @xmath36 and radius @xmath37 , by the schwarzschild assumption , for @xmath44 , there exists no matter , i.e. @xmath376 therefore the conservation laws of ( [ 5.5 ] ) are @xmath377 the tensors @xmath1 in ( [ 5.1 ] ) are written as @xmath378 noting that the central field is spherically symmetric , we assume that @xmath379 for the metric ( [ 5.8 ] ) , the non - zero components of the levi - civita connection are as follows @xmath380 hence the ricci tensor @xmath381 are given by @xmath382,\\ & r_{11}=\frac{u^{\prime\prime}}{2}-\frac{v^{\prime}}{r}+\frac{u^{\prime}}{4}(u^{\prime}-v^{\prime}),\\ & r_{22}=e^{-v}\left[1-e^v+\frac{r}{2}(u^{\prime}-v^{\prime})\right]\\ & r_{33}=r_{22}\sin^2\theta , \\
& r_{ij}=0\qquad \forall i\neq j. \end{aligned}\label{5.12}\ ] ] furthermore , we infer from ( [ 5.8 ] ) , ( [ 5.9 ] ) and ( [ 5.11 ] ) that @xmath383,\\ & d_1d_1\varphi -\frac{1}{2}g_{11}\phi = \frac{1}{2}\left[\varphi^{\prime\prime}-\frac{1}{2}(v^{\prime}+u^{\prime}+\frac{4}{r})\varphi^{\prime}\right],\\ & d_2d_2\varphi -\frac{1}{2}g_{22}\phi = -\frac{r^2}{2}e^{-v}\left[\varphi^{\prime\prime}+\frac{1}{2}(u^{\prime}-v^{\prime})\varphi^{\prime}\right],\\ & d_3d_3\varphi -\frac{1}{2}g_{33}\phi = \sin^2\theta\left(d_2d_2\varphi -\frac{1}{2}g_{22}\phi\right ) .
\end{aligned}\label{5.13}\ ] ] thus , by ( [ 5.12 ] ) and ( [ 5.13 ] ) , the equations ( [ 5.5 ] ) are as follows @xmath384 we need to consider the existence and uniqueness of solutions of the equations ( [ 5.14])-([5.16 ] ) .
first , in the vacuum case , the classical einstein equations are in the form @xmath385 two of which are independent .
the system contains two unknown functions , and therefore for a given initial value ( as @xmath386 is basic in ( [ 6.1a ] ) ) : @xmath387 the problem ( [ 6.1a ] ) with ( [ 6.2a ] ) has a unique solution , which is the schwarzschild solution @xmath388 where @xmath389 now if we consider the influence of the cosmic microwave background ( cmb ) for the central fields , then we should add a constant energy density in equations ( [ 6.1a ] ) : @xmath390 namely , @xmath391 where @xmath392 is the density of the microwave background , whose value is @xmath393 .
then it is readily to see that the problem ( [ 6.4a ] ) with ( [ 6.2a ] ) has no solution .
in fact , the divergence - free equation @xmath394 yields that @xmath395 which implies that @xmath396 .
hence @xmath397 , a contradiction to ( [ 6.4a ] ) .
furthermore , if we regard @xmath392 as an unknown function , then the equations ( [ 6.4a ] ) still have no solutions . on the other hand ,
the new gravitational field equations ( [ 5.14])-([5.16 ] ) are solvable for the microwave background as the number of unknowns are the same as the number of independent equations
. equations ( [ 5.14])-([5.16 ] ) have the following equivalent form : @xmath398 equipped with the following initial values : @xmath399 it is classical that ( [ 6.5a ] ) with ( [ 6.6a ] ) possesses a unique local solution . in fact , we can prove that the solution exists for all @xmath400 .
we now derive the gravitational interaction formula from the basic model ( [ 5.14])-([5.16 ] ) .
first , we infer from ( [ 5.14])-([5.16 ] ) that @xmath401 . \end{aligned}\label{6.8a}\ ] ] consequently , @xmath402 by ( [ 5.2 ] ) and ( [ 5.8 ] ) , we have @xmath403 as the interaction force @xmath261 is given by @xmath404 it follows from ( [ 6.8a ] ) and ( [ 6.9a ] ) that @xmath405 , \la{6.10a } \\ & \varphi^{\prime\prime}= - e^v r + \frac12 ( u ' - v ' + \frac4r ) \varphi^\prime . \la{6.10b}\end{aligned}\ ] ] of course , the following energy balance and conservation law hold true as well : @xmath406 where @xmath9 is the scalar curvature and @xmath407 . equation ( [ 6.10b ] )
is derived by solving @xmath408 using ( [ r - conservation ] ) .
namely , for @xmath44 where @xmath32 , @xmath409.\ ] ] we now consider the region : @xmath410 .
physically , we have @xmath411 hence @xmath60 and @xmath113 in ( [ 6.10a ] ) can be replaced by the schwarzschild solution : @xmath412 as @xmath413 is small for @xmath48 large , by ( [ 6.11a ] ) , the formula ( [ 6.10a ] ) can be expressed as @xmath414.\la{6.13a}\ ] ] this is the interactive force in a central symmetric field .
the first term in the parenthesis is the newton gravity term , and the added second term @xmath415 is the scalar potential energy density , representing the dark matter and dark energy .
in addition , replacing @xmath60 and @xmath113 in ( [ 6.10b ] ) by the schwarzschild solution ( [ 6.12a ] ) , we derive the following approximate formula : @xmath416 consequently we infer from ( [ 6.13a ] ) that @xmath42 , \qquad r=\phi \quad \text { for } r > r_0 .
\la{6.16a}\ ] ] the first term is the classical newton gravitation , the second term is the coupling interaction between matter and the scalar potential @xmath0 , and the third term is the interaction generated by the scalar potential energy density @xmath34 . in this formula , the negative and positive values of each term represent respectively the attracting and repelling forces . integrating ( [ 6.15a ] ) yields ( omitting @xmath417 ) @xmath418 where @xmath419 is a free parameter .
hence the interaction force @xmath261 is approximated by @xmath45,\la{f2}\ ] ] where @xmath420 , @xmath33 for @xmath44 , and @xmath421 we note that based on ( [ r - conservation ] ) , for @xmath422 , @xmath9 is balanced by @xmath34 , and the conservation of @xmath34 suggests that @xmath9 behaviors like @xmath47 as @xmath48 sufficiently large .
hence for @xmath48 large , the second term in the right hand side of ( [ f2 ] ) must be dominate and positive , indicating the existence of dark energy .
we note that the scalar curvature is infinite at @xmath423 : @xmath424 also @xmath9 contains two free parameters determined by @xmath386 and @xmath113 respectively .
hence if we take a first order approximation as @xmath425 where @xmath426 and @xmath427 are free yet to be determined parameters
. then we deduce from ( [ 6.1b ] ) and ( [ 6.2b ] ) that @xmath428 therefore , @xmath429.\ ] ] physically it is natural to choose @xmath430 also , @xmath426 and @xmath431 are much smaller than @xmath432 for @xmath433 . hence @xmath434 .
\la{f3}\ ] ] where @xmath435 and @xmath436 can be estimated using the rubin law of rotating galaxy and the acceleration of the expanding galaxies : @xmath437 we emphasize here that the formula ( [ f3 ] ) is only a simple approximation for illustrating some features of both dark matter and dark energy .
dark matter and dark energy are two of most remarkable discoveries in astronomy in recent years , and they are introduced to explain the acceleration of the expanding galaxies . in spite of many attempts and theories , the mystery remains . as mentioned earlier , this article is an attempt to develop a unified theory for the dark matter and dark energy .
a strong support to the existence of dark matter is the rubin law for galactic rotational velocity , which amounts to saying that most stars in spiral galaxies orbit at roughly the same speed .
namely , the orbital velocity @xmath438 of the stars located at radius @xmath48 from the center of galaxies is almost a constant : @xmath439 typical galactic rotation curves @xcite are illustrated by figure [ f6.1](a ) , where the vertical axis represents the velocity ( km / s ) , and the horizontal axis is the distance from the galaxy center ( extending to the galaxy radius ) .
however , observational evidence shows discrepancies between the mass of large astronomical objects determined from their gravitational effects , and the mass calculated from the visible matter they contain , and figure [ f6.1 ] ( b ) gives a calculated curve . the missing mass suggests the presence of dark matter in the universe . in astronomy and cosmology ,
dark energy is a hypothetical form of energy , which spherically symmetrically permeates all of space and tends to accelerate the expansion of the galaxies .
the high - z supernova search team in 1998 and the supernova cosmology project in 1998 published their observations which reveal that the expansion of the galaxies is accelerating . in 2011 ,
a survey of more than @xmath440 galaxies from austrian astronomers confirmed the fact .
thus , the existence of dark energy is accepted by most astrophysicist . with the new gravitational field equation with the scalar potential energy , and we are now in position to derive the nature of the dark matter and dark energy .
more precisely , using the revised newton formula derived from the new field equations : @xmath441 we determine an approximation of the constants @xmath442 , based on the rubin law and the acceleration of expanding galaxies .
first , let @xmath443 be the total mass in the ball with radius @xmath48 of the galaxy , and @xmath444 be the constant galactic rotation velocity . by the force equilibrium
, we infer from ( [ 6.2 ] ) that @xmath445 which implies that @xmath446 this matches the observed mass distribution formula of the galaxy , which can explain the rubin law ( [ 6.3 ] ) .
second , if we use the classical newton formula @xmath447 to calculate the galactic rotational velocity @xmath448 , then we have @xmath449 inserting ( [ 6.4 ] ) into ( [ 6.5 ] ) implies @xmath450 as @xmath451 , ( [ 6.6 ] ) can approximatively written as @xmath452 which is consistent with the theoretic rotational curve as illustrated by figure [ f6.1](b ) .
it implies that the distribution formula ( [ 6.4 ] ) can be used as a test for the revised gravitational field equations .
third , we now determine the constants @xmath435 and @xmath436 in ( [ 6.2 ] ) . according to astronomic data ,
the average mass @xmath453 and radius @xmath454 of galaxies is about @xmath455 where @xmath456 is the mass of the sun .
taking @xmath457 , then we have @xmath458 based on physical considerations , @xmath459 by ( [ 6.7])-([6.9 ] ) , we deduce from ( [ 6.4 ] ) that @xmath460 now we consider the constant @xmath436 . due to
the accelerating expansion of galaxies , the interaction force between two clusters of galaxies is repelling , i.e. for ( [ 6.2 ] ) , @xmath461 where @xmath462 is the average distance between two galactic clusters .
it is estimated that @xmath463 we take @xmath464 as the distance at which @xmath465 .
namely , @xmath466 hence we derive from ( [ 6.10 ] ) and ( [ 6.11 ] ) that @xmath467 thus , the constants @xmath435 and @xmath436 are estimated by @xmath468 in summary , for the formula ( [ 6.2 ] ) with ( [ 6.12 ] ) , if the matter distribution @xmath443 is in the form @xmath469 then the rubin law holds true .
in particular , the mass @xmath470 generated by the revised gravitation is @xmath471 where @xmath472 is the theoretic value of total mass . hence @xmath473
namely , the revised gravitational mass @xmath470 is four times of the visible matter @xmath474 .
thus , it gives an alternative explanation for the dark matter .
in addition , the formula ( [ 6.2 ] ) with ( [ 6.12 ] ) also shows that for a central field with mass @xmath36 , an object at @xmath475 ( @xmath462 as in ( [ 6.11 ] ) ) will be exerted a repelling force , resulting the acceleration of expanding galaxies at @xmath476 . thus the new gravitational formula ( [ 6.2 ] ) provides a unified explanation of dark matter and dark energy . in this section
, we prove that if the matter is homogeneously distributed in the universe , then the scalar potential @xmath0 is a constant , and consequently the scalar potential energy density is identically zero : @xmath477 .
it is known that the metric for an isotropic and homogeneous universe is given by @xmath478,\la{7.1}\ ] ] which is called the friedmann - lematre - robertson - walker ( flrw ) metric , where the scale factor @xmath479 represents the cosmological radius , and @xmath89 takes one of the three numbers : @xmath480 . for the flrw metric ( [ 7.1 ] )
, the energy - momentum tensor @xmath350 is given by @xmath481 where @xmath392 is the mass density , @xmath482 is pressure , and @xmath483 by ( [ 7.1 ] ) and ( [ 7.2 ] ) , the einstein field equations ( [ 1.1 ] ) are reduced to two equations : @xmath484 and the conservation law @xmath485 gives @xmath486 then only two of the above three equations ( [ 7.3])-([7.5 ] ) are independent , and are called the friedman equations . on the other hand , for the metric ( [ 7.1 ] ) with ( [ 7.2 ] ) , the new gravitational field equations ( [ 1.3 ] ) with scalar potential are reduced to @xmath487 and the conservation equation @xmath488 again two of the above three equations ( [ 7.6])-([7.8 ] ) are independent .
it follows from ( [ 7.6 ] ) and ( [ 7.7 ] ) that @xmath489 in the following we shall prove that @xmath490 namely , @xmath491constant .
in fact , let @xmath408 and @xmath392 have the form : @xmath492 inserting ( [ 7.11 ] ) in ( [ 7.6 ] ) , ( [ 7.8 ] ) and ( [ 7.9 ] ) , we arrive at @xmath493 multiplying both sides of ( [ 7.12 ] ) by @xmath494 we obtain @xmath495 + \frac{4\pi g}{3 } \frac{\theta'}{a } - \frac16 \frac{\psi'}{a } = - \frac{4\pi g}{c^2 } p a a ' - \frac12 ( a')^2 \varphi ' . \la{7.15}\ ] ] it follows then from ( [ 7.13])-([7.15 ] ) that @xmath496 which implies that ( [ 7.10 ] ) holds true .
the conclusion ( [ 7.10 ] ) indicates that if the universe is in the homogeneous state , then the scalar potential energy density @xmath30 is identically zero : @xmath477 .
this fact again demonstrates that @xmath0 characterizes the non - uniform distribution of matter in the universe .
we have discovered new gravitational field equations ( [ 1.3 ] ) with scalar potential under the postulate that the energy momentum tensor @xmath5 needs not to be divergence - free due to the presence of dark energy and dark matter : @xmath497 second , associated with the scalar potential @xmath0 is the scalar potential energy density @xmath30 , which represents a new type of energy / force caused by the non - uniform distribution of matter in the universe .
this scalar potential energy density varies as the galaxies move and matter of the universe redistributes . like gravity
, it affects every part of the universe as a field . this scalar potential energy density @xmath30 consists of both positive and negative energies .
the negative part of this potential energy density produces attraction , and the positive part produces repelling force .
also , this scalar energy density is conserved with mean zero : @xmath498 third , using the new field equations , for a spherically symmetric central field with mass @xmath36 and radius @xmath37 , the force exerted on an object of mass @xmath41 at distance @xmath48 is given by ( see ( [ 6.16a ] ) ) : @xmath42 , \qquad r=\phi \quad \text { for } r > r_0.\ ] ] where @xmath499 .
fourth , the sum @xmath500 of this new potential energy density @xmath501 and the coupling energy between the energy - momentum tensor @xmath5 and the scalar potential field @xmath0 @xmath502 gives rise to a new unified theory for dark matter and dark energy : the negative part of @xmath31 represents the dark matter , which produces attraction , and the positive part represents the dark energy , which drives the acceleration of expanding galaxies .
fifth , the scalar curvature @xmath9 of space - time obeys : @xmath503 consequently , when there is no normal matter present ( with @xmath32 ) , the curvature @xmath9 of space - time is balanced by @xmath33 . therefore , there is no real vacuum in the universe . height 2pt depth -1.6pt width 23pt , _ geometric theory of incompressible flows with applications to fluid dynamics _ , vol . 119 of mathematical surveys and monographs , american mathematical society , providence , ri , 2005 . | the main objective of this article is to derive a new set of gravitational field equations and to establish a new unified theory for dark energy and dark matter .
the new gravitational field equations with scalar potential @xmath0 are derived using the einstein - hilbert functional , and the scalar potential @xmath0 is a natural outcome of the divergence - free constraint of the variational elements .
gravitation is now described by the riemannian metric @xmath1 , the scalar potential @xmath0 and their interactions , unified by the new gravitational field equations .
associated with the scalar potential @xmath0 is the scalar potential energy density @xmath2 , which represents a new type of energy caused by the non - uniform distribution of matter in the universe .
the negative part of this potential energy density produces attraction , and the positive part produces repelling force .
this potential energy density is conserved with mean zero : @xmath3 .
the sum of this new potential energy density @xmath4 and the coupling energy between the energy - momentum tensor @xmath5 and the scalar potential field @xmath0 gives rise to a new unified theory for dark matter and dark energy : the negative part of this sum represents the dark matter , which produces attraction , and the positive part represents the dark energy , which drives the acceleration of expanding galaxies .
in addition , the scalar curvature of space - time obeys @xmath6 .
furthermore , the new field equations resolve a few difficulties encountered by the classical einstein field equations . |
the spin glass ( sg ) phase@xcite is not stable at finite temperature in two dimensions ( 2d ) . when the energies are not quantized , the behavior at @xmath13 can be understood in terms of a scaling theory,@xcite which is usually called the droplet model .
this theory describes the scaling of the average domain - wall ( dw ) energy @xmath14 with length scale in terms of the stiffness exponent @xmath15 , i.e. @xmath16 this exponent has a value of about @xmath17 for 2d , almost independent of the detailed nature of the bond distribution @xcite and describes the scaling of different kinds of excitations like domain walls and droplets , at least if large enough system sizes are studied.@xcite the possibility that quantization of the energies might lead to the special behavior @xmath18 was first pointed out by bray and moore.@xcite it has become clear over the last few years that there actually is a fundamental difference in the behavior of the 2d ising spin glass at zero temperature , between those cases where the energies are quantized and those where it is not.@xcite nevertheless , recent results @xcite indicate that in the low - temperature critical scaling regime , the behavior for quantized and non - quantized models might be very similar .
the hamiltonian of the edwards - anderson model for ising spins is @xmath19 where each spin @xmath20 is a dynamical variable which has two allowed states , @xmath21 and @xmath22 .
the @xmath23 indicates a sum over nearest neighbors on a simple square lattice of size @xmath24 .
the standard model for quantized energies is the @xmath25 model , where we choose each bond @xmath26 to be an independent identically distributed ( iid ) quenched random variable , with the probability distribution @xmath27 thus we actually set @xmath28 , as usual .
the concentration of antiferromagnetic bonds is @xmath0 , and @xmath29 is the concentration of ferromagnetic bonds . with the @xmath30 of eqn .
( [ eq : pj ] ) , the ea hamiltonian is equivalent to the @xmath31 gauge glass model.@xcite wang , harrington and preskill@xcite have argued that the anomalous behavior of the @xmath25 model is caused by topological long - range order , as a consequence of the gauge symmetry . along the @xmath32 axis
this model exhibits a phase transition from a ferromagneticly ordered phase @xcite for small concentrations @xmath33 of the antiferromagnetic bonds to a sg critical line at large values @xmath34 ( and @xmath35 due to the bipartite symmetry of the square lattice ) .
recently , this phase transition was characterized by high - precision ground - state calculations,@xcite which have in particular yielded @xmath36 .
the ferromagnetic phase persists at finite temperatures @xmath37 for @xmath38 , while the spin - glass correlations become long - range only at zero temperature.@xcite interestingly , @xmath39,@xcite _ i.e. _ the ferromagnetic phase is reentrant .
the first hint of the remarkable behavior of the @xmath25 model in the sg regime was observed by wang and swendsen.@xcite they found that , although for periodic boundary conditions there is an energy gap of @xmath40 between the ground states ( gs ) and the first excited states , the specific heat when @xmath41 appeared to be proportional to @xmath42 .
this result was questioned by saul and kardar.@xcite the behavior of the specific heat was finally demonstrated in a convincing fashion by lukic _
et al._@xcite saul and kardar also found that the scaling of the dw entropy with size did not appear to agree with the prediction of the droplet model.@xcite this issue has recently been clarified,@xcite and the dw entropy scaling anomaly has been associated with zero - energy domain walls . in this work
we explore the behavior of the @xmath25 model inside the full sg phase @xmath43 ( the behavior @xmath44 is equivalent due to the symmetry of the model ) .
in particular we study the finite - size scaling behavior of the gs energy , of the fluctuations of the gs energy , and of the domain - wall energy .
we also employ aspect - ratio ( ar ) scaling , @xcite _ i.e. _ we study rectangular lattices of width @xmath45 and height @xmath46 , the aspect ratio being @xmath47 .
carter , bray and moore@xcite have extended ar scaling to the ising sg , and demonstrated that studying the scaling as a function of @xmath10 is an effective method for calculating the exponent @xmath15 . here , we look at the limit @xmath12 and show that ar indeed works for the @xmath25 model as well , in contrast to previous attempts.@xcite the entire probability distribution @xmath48 has a simple gaussian form for small @xmath10 , although , as we will show , the underlying assumption of independent contributions to the dw energy is not strictly valid .
we define domain walls for the sg as it was done in the seminal work of mcmillan.@xcite we look at differences in the gs energy between two samples with the same set of bonds , and the same boundary conditions in one direction , but different boundary conditions in the other direction .
in particular we use periodic ( p ) or antiperiodic ( ap ) boundary conditions along the * x*-axis and free boundary conditions along the * y*-axis . for an example of a dw created in such a way , cf .
[ fig : normaldestroyed ] . for each set of bonds ,
the dw energy @xmath14 is then defined to be the difference in the gs energies of the two different boundary conditions : @xmath49 we use a so - called _ matching algorithm _ to calculate the gs.@xcite let us now sketch just the basic ideas of the matching algorithm
. for the details , see refs . .
the algorithm allows us to find ground states for lattices which are planar graphs .
this is the reason why we apply ( anti-)periodic boundary conditions only in one direction , * x * , while the other direction , * y * , has free boundary conditions . in the left part of fig .
[ fig : matching ] a small 2d system with ( for simplicity ) free boundary conditions in both directions is shown .
all spins are assumed to be `` up '' , hence all antiferromagnetic bonds are not satisfied .
if one draws a dashed line perpendicular to each broken bond , one ends up with the situation shown in the figure : all dashed lines start or end at frustrated plaquettes and each frustrated plaquette is connected to exactly one other frustrated plaquette by a dashed line .
each pair of plaquettes is then said to be _
matched_. in general , closed loops of broken bonds unrelated to frustrated plaquettes can also appear , but this is possible only for excited states .
now , one can consider the frustrated plaquettes as the vertices and all possible pairs of connections as the edges of a ( dual ) graph .
the dashed lines are selected from the edges connecting the vertices and called a _ perfect _ matching , since _ all _ plaquettes are matched .
one can assign weights to the edges in the dual graph , the weights are equal to the sum of the absolute values of the bonds crossed by the dashed lines .
the weight @xmath50 of the matching is defined as the sum of the weights of the edges contained in the matching .
as we have seen , @xmath50 counts the broken bonds , hence , the energy of the configuration is given by @xmath51 with @xmath52 from eqn .
( [ eq : pj ] ) .
note that this holds for _ any _ configuration of the spins , if one also includes closed loops in @xmath50 , since a corresponding matching always exists . obtaining
a gs means minimizing the total weight of the broken bonds ( see right panel of fig . [
fig : matching ] ) .
this automatically forbids closed loops of broken bonds , so one is looking for a _ minimum - weight perfect matching_. this problem is solvable in polynomial time .
the algorithms for minimum - weight perfect matchings@xcite are among the most complicated algorithms for polynomial problems .
fortunately the leda library offers a very efficient implementation,@xcite which we have applied here . with free boundary conditions in at least one direction , @xmath14 is a multiple of 2 ( since @xmath28 ) .
when periodic or antiperiodic boundary conditions are applied in both the * x * and * y * directions , however , @xmath14 becomes a multiple of 4 if the number of @xmath22 bonds is even .
if the number of @xmath22 bonds is odd , @xmath14 takes on values @xmath53 with these boundary conditions .
when @xmath46 is odd , changing the boundary conditions in the * x * direction from periodic to antiperiodic changes the number of @xmath22 bonds from odd to even , or vice versa .
the behavior of the dw energy for the @xmath25 model with @xmath54 was discussed previously.@xcite for the boundary conditions we are using , the average @xmath55 goes to zero exponentially as @xmath10 becomes much greater than one . here
@xmath56 denotes an average over the ensemble of random bond distributions for an @xmath6 lattice .
we can understand this result by thinking of the system as consisting of blocks of size @xmath57 pasted together along the * x * direction .
the probability of having a zero - energy domain wall in each block is almost independent of the other blocks .
the same thing happens when @xmath46 is even if the boundary conditions in the * y * direction are periodic or antiperiodic .
however , if @xmath46 is odd and the boundary conditions in the * y * direction are periodic or antiperiodic , @xmath55 goes to 2 at large @xmath10 , because then @xmath58 is not allowed .
since a critical exponent should be independent of boundary conditions , this is a(nother ) demonstration that @xmath18 for the @xmath25 model in 2d . as pointed out earlier@xcite , these results for large @xmath10 do not agree with the scaling law prediction of carter , bray and moore,@xcite due to the special role of zero - energy domain walls . on the other hand , in the limit @xmath12 with our boundary conditions ,
the prediction of carter , bray and moore is @xmath59 this limit has not been studied before for the @xmath25 model . when @xmath18 , @xmath55 should scale as @xmath60 in 2d . for small @xmath10 we may think of the lattice as consisting of @xmath61-sized subsystems stacked in the * y * direction , with @xmath14 being the sum of the dw energies of the subsystems .
therefore , by the central limit theorem , we anticipate that the probability distribution of @xmath14 should approach a gaussian distribution in the limit of small @xmath10 . in the spin - glass region of the phase diagram ,
the center of this limiting gaussian will approach zero as @xmath45 increases .
we will show that our numerical results for the small @xmath10 limit are indeed in good agreement with these expectations , even though the subsystems are not fully independent .
we begin the study of the @xmath32 behavior of the @xmath25 random bond model by studying the gs energy for different concentrations @xmath0 of the antiferromagnetic bonds [ @xmath62 ( ferromagnetic phase ) , @xmath63 and @xmath64 and different system sizes @xmath65 for square samples @xmath66 , _ i.e. _ aspect ratio @xmath67 .
all results are averages over many different realizations of the disorder . the minimum number of independent samples used varies with size , ranging between typically 100000 ( @xmath68 ) to typically 10000 ( @xmath69 ) .
( color online ) the symbols show the average ground state energy @xmath70 for the square systems ( @xmath67 ) as function of the system size @xmath45 for selected values of @xmath0 .
the lines give the results to the fits ( @xmath71 ) ( see text ) .
the data for @xmath62 look similar , but is outside the frame ( @xmath72 ) .
the inset shows the same data , rescaled as @xmath73 so that the expected behavior yields straight lines when plotted as function of @xmath74 .
the data for @xmath75 have been shifted downwards by 0.2 for better visibility . ] in fig .
[ fig : e0 ] , the gs energy per spin @xmath70 is shown as a function of the system size for selected values of @xmath0 .
this finite - size scaling behavior yields particular insight into the physics and seems to be related @xcite to the stiffness exponent @xmath15 .
it has been argued @xcite that for the case of open boundary conditions in exactly one direction and periodic boundary conditions in the other direction the gs energy per spin follows to lowest orders the form @xmath76 note that the arguments used in ref .
should apply for the ferromagnetic phase @xmath77 as well . since we expect @xmath18 everywhere in the sg phase ( see below ) and @xmath78 for the ferromagnetic phase
, we can restrict ourselves to the contributions @xmath79 , @xmath80 and @xmath81 .
when fitting@xcite the data to this @xmath70 , fits with very high quality result , as shown by lines in fig .
[ fig : e0 ] .
this is confirmed when plotting the data and the fitting functions in the form @xmath82 as a function of @xmath83 .
this should give , according to eqn .
( [ eq : e0l ] ) , a linear behavior with slope @xmath84 and ordinate intersection @xmath85 .
we see that indeed the data follows the functional form well , although even higher order corrections seem to be present .
they become significant at very small sizes , but we can not quantify them within the statistical accuracy of the data .
we only observe that this correction term seems to have an opposite sign for small ( @xmath86 ) compared to larger values of @xmath0 .
furthermore , the behavior of the @xmath81 term is quite similar inside the sg phase for different values of @xmath0 , while the behavior in the ferromagnetic phase @xmath87 is very different .
the value of the ordinate intercept @xmath85 is monotonic in @xmath0 , as expected.@xcite ( color online ) the symbols show the width @xmath88 of the distributions of the gs energy @xmath79 for square systems ( @xmath67 ) as function of system size @xmath45 for selected values of @xmath89 .
the data is rescaled with the average gs energy @xmath70 and with the leading behavior @xmath45 .
additionally , the dashed line displays the result for @xmath62 .
the full lines show fits to constants at larger system sizes . ]
next , we consider the width @xmath88 of the distribution of gs energies per spin . for short - range @xmath90-dimensional spin glasses , it has been proven @xcite that @xmath91 , i.e. @xmath92 in this case . to remove finite - size effects caused by our special choice of boundary conditions
, we normalize @xmath88 by the finite - size gs energy .
thus , we plot @xmath93 . as seen in fig .
[ fig : widthpe0 ] , this is indeed the case . only corrections for very small system - sizes become visible .
we have also included the data for @xmath62 , which behave in the same way .
hence , here the ferromagnetic phase and the sg phase can not be distinguished just from the asymptotic behavior , in contrast to the behavior of the mean alone .
the symbols show the limiting value @xmath94 for the width @xmath88 of the distributions of the gs energy ( as shown in fig.[fig : widthpe0 ] ) as a function of @xmath0 .
the solid line represents a fit according to eqn .
( [ eq : fitcsigma ] ) , see text . ] on the other hand , as visible in fig .
[ fig : widthpe0 ] , the value of @xmath95 increases when approaching the phase transition @xmath96 . for a more detailed analysis ,
we have plotted @xmath95 as function of @xmath0 inside the sg phase in fig .
[ fig : widthconst ] , also for values of @xmath0 closer to @xmath9 than those shown in fig .
[ fig : widthpe0 ] .
one can see a strong increase when approaching the critical concentration .
fitting a power law ( with @xmath97 ) @xmath98 in the range @xmath99 $ ] yields values @xmath100 ,
@xmath101 and @xmath102 , _ i.e. _ a cusp at the phase transition .
note that the value of the scaling exponent is very close to the exponent @xmath103 , which describes the finite - size scaling of the dw energy @xmath14 ; see below . nevertheless , the basic observation of the ( almost ) universal behavior inside the sg region , which we find when looking at the gs energy , becomes even more apparent , when studying the behavior of the dw energy , which we do in the next section . for all samples considered in the previous section
, we have also calculated the dw energy as defined in eqn .
( [ eq : edw ] ) . in fig .
[ fig : emean ] the average value @xmath104 of the dw energy is shown as a function of system size for selected values of @xmath0 .
close to the phase transition @xmath97 , the systems exhibit a high degree of ferromagnetic order for small sizes , leading to relatively large values of the average dw energy .
this explains why very large system sizes were needed in ref . to determined the location @xmath97 of the phase transition
precisely . for larger values @xmath105 , the ferromagnetic correlation length is small . the average value of the dw energy as function of system size for selected values of @xmath0 .
error bars are very small for values @xmath106 but relatively large where the dw energy is small .
thus , they are omitted for readability .
note the double - logarithmic axes .
lines are guides to the eye . ] in fig .
[ fig : estiff ] the average absolute value latexmath:[$\langle size for selected values of @xmath0 . for very small values of @xmath0 , close to the phase transition
, this dw energy increases with system size for small system sizes . hence ,
if only slow algorithm were available , the system would look like exhibiting an ordered phase ( _ i.e. _ @xmath108 ) .
when going to larger system sizes , it becomes apparent that this is only a finite - size effect . for intermediate values of @xmath0 ,
the dw energy decreases with growing system size , but no saturation of @xmath55 is visible on the accessible length scales , hence the data look similar to the results for gaussian systems @xcite with @xmath109 . when looking at the results for @xmath110 , it becomes apparent that @xmath55 converges to a finite value .
this is due to the quantized nature @xcite of the possible values for @xmath14 .
therefore , it is clear that also for intermediate values of @xmath0 , the dw energy converges to a finite plateau - value as well , but for much larger system sizes .
hence , the model provides a striking example that finite - size corrections can persist up to huge length scales .
note that the convergence to a non - zero plateau value is compatible with the results from the previous section , where we have found that inside the sg phase the scaling function eqn .
( [ eq : e0l ] ) with @xmath18 describes the data well everywhere . the average absolute value of the dw energy as a function of system size for selected values of @xmath0 .
note the double - logarithmic axes .
lines are guides to the eye . ] due to the quantization of the possible dw energies , a value @xmath55 close to zero means that many system exhibit actual zero dw energy @xmath58 .
we have also studied the finite - size dependence of the probability @xmath111 for a zero - energy dw as a function of the system size , as shown in fig . [
fig : pstiff0 ] .
one observes that zero - energy domain walls are very common .
their probability of occurrence increases with the concentration @xmath0 of the antiferromagnetic bonds ( until @xmath112 due to the symmetry of the model in @xmath113 ) and with the system size leading to a convergence to a limiting value for @xmath114 .
the probability @xmath115 for a zero - energy domain wall as function of system size for selected values of @xmath0 .
all error bars are at most of the symbol size .
lines are guides to the eye . ]
the simplest assumption for the convergence of @xmath116 and @xmath111 , at given values of @xmath0 , to their limiting values for @xmath117 is a power law @xmath118 the exponent @xmath119 describes the rate of convergence .
it depends on the fraction @xmath0 of the antiferromagnetic bonds and on the aspect ratio @xmath10 , as well as the constants @xmath120 and @xmath121 .
we have included the dependence on @xmath10 because below we also study values of @xmath122 , but for the moment we stick to @xmath67 .
the results of fitting eqn .
( [ eq : simplefit ] ) to the data ( @xmath67 ) : exponent @xmath119 ( main plot ) and limiting value @xmath120 ( inset ) . ]
when fitting eqn .
( [ eq : simplefit ] ) to the dw energy data for @xmath123 , _
i.e. _ far enough away from @xmath9 , and for sizes @xmath124 , we obtain results as shown in fig .
[ fig : edwsimple ] . for small concentrations @xmath0 , the exponent @xmath119 is less negative than for @xmath0 close to @xmath125 , corresponding to needing larger sizes to reach the asymptotic value @xmath120 . on
the other hand @xmath120 itself seems to depend only weakly on @xmath0 , confirming the notion that everywhere inside the sg region the dw energy reaches a non - zero plateau value .
the dependence of @xmath119 appears like a strong violation of universality inside the sg region . to investigate this assumption , whether it is a true non - universality or just a finite - size scaling effect , we have performed fits which also include corrections to scaling .
we do this by fitting the data to a form @xmath126 @xmath127 represents any function of @xmath128 . two examples that we chose to study
were @xmath55 and the probability @xmath129 .
the number of correction - to - scaling terms included , @xmath130 , was chosen according to the amount of data to be fit .
here we have tried 2 and 3 .
of course , the values of the coefficients @xmath131 and @xmath132 which are found by the fits depend somewhat on the choice of @xmath130 .
the computed statistical errors do not include any allowance for systematic errors due to the choice of the form of the scaling function .
fortunately , it turned out that the value of @xmath131 is relatively insensitive to the choice of @xmath130 . in general , whether we chose @xmath130 to be 2 or 3 , the computed @xmath132 coefficients did not all have the same sign .
this effect is caused by the behavior at small @xmath45 .
for this reason , there is no well defined prescription for deciding what the best choice of the exponent @xmath133 is .
we have not included @xmath133 in the set of free fitting parameters . since it is the purpose of this section to show that the behavior inside the full sg region is universal , we have selected a value for @xmath133 , such that the data for all values of @xmath0 can be fit .
thus the strong finite - size effects close to @xmath9 can be explained also .
we have chosen the value @xmath134 , which is similar to the values used previously in other work on this model,@xcite empirically . for a critical point with @xmath135 ,
the general theory of finite - size scaling@xcite requires that @xmath136 , where @xmath137 is the correlation length exponent .
that result does not apply here , however , since @xmath138 .
( color online ) finite - size scaling fits for @xmath139 and @xmath140 : ( a ) @xmath55 vs. @xmath141 ; ( b ) probability of @xmath58 vs. @xmath141 .
the error bars show one standard deviation.,title="fig:",width=326](color online ) finite - size scaling fits for @xmath139 and @xmath140 : ( a ) @xmath55 vs. @xmath141 ; ( b ) probability of @xmath58 vs. @xmath141 .
the error bars show one standard deviation.,title="fig:",width=326 ] as an example , our finite - size scaling fits ( using @xmath142 ) to the data using eqn .
( [ eq : fitfull ] ) for the case @xmath139 and @xmath143 are shown in fig .
[ fig.3 ] , the data for all system sizes was included in the fits . in the figures , we show @xmath144 as a function of @xmath141 .
one observes a straight line for @xmath145 , showing that the choice of the exponent is compatible with the data .
note that the same @xmath140 data was originally@xcite fit ( only for the case @xmath112 ) by making different assumptions which are not consistent with eqn .
( [ eq : fitfull ] ) .
( color online ) fit coefficients ( defined by eqn .
[ eq : fitfull ] ) for @xmath140 : ( a ) @xmath146 vs. @xmath0 ; ( b ) @xmath129 vs. @xmath0.,title="fig:",width=326 ] ( color online ) fit coefficients ( defined by eqn .
[ eq : fitfull ] ) for @xmath140 : ( a ) @xmath146 vs. @xmath0 ; ( b ) @xmath129 vs. @xmath0.,title="fig:",width=326 ] fig
. [ fig.7](a ) shows the fit coefficients for @xmath147 , as defined by eqn .
( [ eq : fitfull ] ) , as a function of @xmath0 for @xmath140 .
the quality of fit was good everywhere , proving that the apparent non - universality visible in the dependence of the effective exponent @xmath119 is due to the presence of corrections to scaling .
we again see that the value of @xmath131 , the estimate for @xmath114 , changes very little over this range of @xmath0 ; the values are for @xmath148 very similar to the fit according to eqn .
( [ eq : simplefit ] ) , but they are larger in the present fit for small concentrations , @xmath149 .
we also note that there are substantial variations in the @xmath132 coefficients .
in particular , @xmath150 alternates in sign for the smaller values of @xmath0 . in fig . [
fig.7](b ) the results of fitting @xmath129 , which shows the same type of behavior , are given .
in this case the signs of @xmath151 and @xmath150 alternate .
domain walls of subsystems are not independent : comparison of a minimum - energy dw for a system of size @xmath152 ( left ) having @xmath153 and of the same system cut into two systems of size @xmath154 ( right ) , having @xmath155 .
note that in the left figure , the dw wraps around the system in the @xmath156-direction .
note also that the dws are highly degenerate and the algorithm does not find dws in a controlled way .
a `` jump '' in the dw at the right is visible .
this figure illustrates what a typical dw looks like .
the important point here is the difference in the dw energies . ]
the main assumption when using ar scaling eqn .
( [ eq : ar ] ) in the limit @xmath157 is that different subsystems of size @xmath158 are independent of each other , which leads to a gaussian distribution of the dw energies @xmath14 in the limit @xmath157 .
nevertheless , one can easily imagine that the dws in different parts of the system are not truly independent of each other .
this is demonstrated for a sample system in fig .
[ fig : normaldestroyed ] , where the dw energy of each subsystem ( 1)/(2 ) of size @xmath154 is @xmath159 , but the dw energy of the full system is @xmath153 .
also , in a previous study @xcite , the validity of ar scaling could not be established for the @xmath25 model . here
, we will show that ar scaling indeed works also for this case , _
i.e. _ the distribution of the dw energies becomes indeed gaussian , despite the non - independencies of the domain walls inside the different subsystems .
because the shape of the @xmath128 distribution is changing with @xmath10 , we expect that the ar scaling will not be perfect in the range of our data .
as we shall see , however , the deviations from the gaussian fit become smaller than our statistical errors for @xmath160 .
thus we are able to verify that we are approaching the predicted scaling limit .
first , we concentrate on @xmath161 where the distribution @xmath162 of eqn .
( [ eq : pj ] ) is symmetric about zero .
thus in this case the distribution @xmath128 is ( ignoring statistical fluctuations ) also symmetric around zero , for any values of @xmath45 and @xmath46 .
therefore , we used @xmath161 to collect data for sets of random lattices with sequences of @xmath45 and @xmath46 having the aspect ratios @xmath10 = 1 , 1/2 , 1/4 , 1/8 , 1/16 and 1/32 . for each value of @xmath10
, we then used finite - size scaling to extrapolate to the large lattice limit .
the symbols display the distribution of dw energies for the case @xmath163 ( _ i.e. _ @xmath67 ) and @xmath112 , the lines shows the result of a fit to a gaussian .
strong deviations from a gaussian distribution are visible at @xmath115 ( main plot ) and in the `` tail '' of the distribution ( inset ) . ]
we start by studying @xmath5 samples ( @xmath67 ) , to establish that in this case the distribution of the dw energies is indeed not gaussian .
( otherwise , it would not be surprising if it is also gaussian for @xmath164 . )
the distribution for @xmath165 is shown in fig .
[ fig : gaussfita ] .
this distribution is obtained from an average over 39000 realizations .
we have fit@xcite a gaussian to the data .
( note that the fitting procedure ignores the fact that the energies are quantized . )
strong deviations from gaussian behavior are visible at @xmath115 ( main plot ) and in the `` tail '' of the distribution ( inset ) .
correspondingly the quality of the fit , _
i.e. _ @xmath166 per degree of freedom , as given by our fitting program@xcite in this case , is very high : @xmath167ndf@xmath168 .
even larger deviations are expected in the tail . to observe these ,
more sophisticated techniques would be needed @xcite , which is beyond the scope of this work .
the strong deviation from gaussian behavior is not a finite - size effect , as may be seen from fig .
[ fig : ndf ] , where the circles display @xmath167ndf as a function of @xmath45 for the @xmath67 case .
quality of the gaussian fit for the dw energy distribution measured by @xmath167ndf as a function of the block size @xmath45 for @xmath112 ( circles : @xmath67 , squares : @xmath169 ) . ]
the reader should also note that we find no evidence for any `` sawtooth '' structure in the data .
this is in marked contrast to the case of periodic or antiperiodic boundary conditions in the * y * direction , which causes alternating values of @xmath14 to have zero probability .
therefore @xmath128 does not become independent of the boundary conditions even in the limit of large lattices .
this is an aspect of the topological long - range order@xcite which exists in this model when the boundary conditions are periodic along both * x * and * y*. the symbols show the distribution of dw energies for the case @xmath170 ( _ i.e. _ @xmath169 ) and @xmath112 , the lines shows the result of fitting to a gaussian .
main plot : linear ordinate , inset : logarithmic ordinate scale .
the data match the gaussian very well . ] in fig .
[ fig : gaussfitb ] we show the probability distribution @xmath171 for a set of lattices with @xmath161 , @xmath172 and @xmath173 . in this case
, there are 90,000 random samples in the data set .
again , the distribution was fit to a gaussian @xcite .
this fit is quite good , as demonstrated by the fact that the value of @xmath167ndf are is close to one .
note that the precise value of @xmath167ndf depends on the number of points in the tail of the distribution that are included in the fit , which is somewhat arbitrary .
the gaussian behavior here is not a finite - size effect either .
this can be seen from fig .
[ fig : ndf ] , where the square symbols display @xmath167ndf as function of @xmath45 for the @xmath172 case .
( color online ) kurtosis vs. @xmath45 ( @xmath174 . ) the * x * axis is scaled logarithmically.,width=326 ] when @xmath139 the configuration averages of all the odd moments of @xmath128 must vanish .
then the dominant contribution to non - gaussian behavior will be the kurtosis , which simplifies to @xmath175 , where @xmath176 is the @xmath177 moment of @xmath128 , under these conditions .
[ fig.4 ] shows the behavior of the kurtosis of the @xmath128 distributions at @xmath139 as a function of @xmath45 and @xmath10 .
it shows that for @xmath160 the kurtosis is negligible , except when @xmath178 , quantifying the convergence towards a gaussian distribution for small @xmath10 .
this convergence can be also seen when fitting @xmath116 to eqn .
( [ eq : fitfull ] ) , now for aspect ratios @xmath179 . in table
i we show the results of the finite - size scaling fits at @xmath180 . according to eqn .
( [ eq : ar ] ) , the prediction of ar scaling theory is that the ratio @xmath181 should approach @xmath182 for small @xmath10 .
considering the statistical errors and the corrections to scaling , the agreement is quite satisfactory .
here we observe that for @xmath183 the scaling assumption is reasonably accurate .
.[tab : table1]@xmath55 for @xmath139 , extrapolated to @xmath184 , as a function of the aspect ratio , @xmath10 .
the @xmath130 and @xmath131 are defined by eqn .
( [ eq : fitfull ] ) , and the @xmath185 are statistical error estimates . [ cols="<,^,^,^,^",options="header " , ] we have also studied the ar approach for @xmath186 and found similar results ( not shown ) . for aspect ratios @xmath183 the assumptions of a gaussian distribution of the dw energies is again valid .
hence , we believe that inside the entire sg region the behavior is consistent with the assumptions of the ar approach . at the end of section ii we argued that for small @xmath10 the system may be thought of as a set of @xmath5 subsystems stacked in the * y * direction .
this implies that the behavior of @xmath14 in this limit should obey eqn .
( [ eq : ar ] ) with @xmath18 .
as we have seen , our numerical results are indeed consistent with this argument .
however , the hand - waving argument does not specify how the subsystems are to be connected to each other . before actually seeing the numerical results
, we were far from confident that this argument would correctly predict what was found .
there is another apparently reasonable argument , which leads one to expect that the results for small @xmath10 should not give this answer . as @xmath12 , we know that the width of the @xmath14 distribution will diverge .
this means that for most of our sample lattices @xmath187 in this limit .
one might have thought , therefore , that the fact that @xmath14 is quantized would no longer matter .
thus one might have guessed that the system would behave in this limit as if @xmath188 , the value for an unquantized distribution .
therefore , our results indicate that a bohr correspondence principle , that in the limit of large quantum numbers the results should be indistinguishable from those of an unquantized system , does not apply .
how can we understand this ?
in our opinion , what we learn from this is that @xmath18 is directly connected to the existence of topological long - range order@xcite at @xmath189 .
one might try to argue that @xmath190 is not small enough , and that for even smaller @xmath10 one would indeed find a crossover to @xmath188 .
such a viewpoint is suggested by the recent work of jrg _
et al._@xcite on the other hand a more detailed analysis@xcite of the low temperature behavior of the specific heat of the @xmath25 model does not support this interpretation .
we have studied the statistics and finite - size scaling behavior of gs and dw energies for the 2d ising random - bond spin glass with an mixture of @xmath21 and @xmath22 bonds , where @xmath0 is the concentration of antiferromagnetic bonds . by using sophisticated matching algorithms
, we can calculate the gs energy and dw energies for quite large systems exactly , which allows us to study the scaling behavior very precisely .
we find that the scaling behavior of the gs energy can be described everywhere inside the sg region by the same simple scaling function eqn .
( [ eq : e0l ] ) , as predicted by ref .. furthermore , when looking at the fluctuations of the gs energy , we find , after carefully taking into account finite - size corrections , that they follow the predicted simple @xmath191-scaling with an amplitude which has a cusp singularity at the ferromagnetic - sg transition @xmath9 .
the singularity is described by an exponent which is close to @xmath192 .
the behavior of the dw energy is characterized close to @xmath9 , by huge finite - size effects .
nevertheless , the results are compatible with a convergence of @xmath193 to a finite plateau value everywhere inside the sg phase .
this convergence can be described , again universally inside the sg phase , by a single exponent @xmath194 , just by taking into account higher - order corrections to scaling .
finally , we have studied @xmath24 rectangular lattices with aspect ratio @xmath10 between 1/32 and 1 .
we have demonstrated by an example that one assumption underlying ar scaling , i.e. the assumed independence of the dw energies of different blocks , is not strictly valid .
nevertheless , we find that for large lattices the probability distribution of @xmath14 in the sg region of the phase diagram approaches for @xmath195 a gaussian centered at @xmath58 .
hence , in the small @xmath10 limit the behavior obeys the ar scaling predictions of carter , bray and moore.@xcite rf is grateful to s. l. sondhi , f. d. m. haldane and d. a. huse for helpful discussions , and to princeton university for providing use of facilities .
akh acknowledges financial support from the _ volkswagenstiftung _ ( germany ) within the program `` nachwuchsgruppen an universitten '' and from the dyglagemen program funded by the eu .
reviews on spin glasses can be found in : k. binder and a. p. young , rev .
phys . * 58 * , 801 ( 1986 ) ; m. mezard , g. parisi , m. a. virasoro , _ spin glass theory and beyond _ , ( world scientific , singapore 1987 ) ; k. h. fischer and j. a. hertz , _ spin glasses _ , ( cambridge university press , cambridge 1991 ) ; a. p. young ( ed . ) , _ spin glasses and random fields _ , ( world scientific , singapore 1998 ) .
a. j. bray and m. a. moore , phys .
b * 31 * , 631 ( 1985 ) .
d. s. fisher and d. a. huse , phys .
lett . * 56 * , 1601 ( 1986 ) . c. amoruso , e. marinari , o. c. martin and a. pagnani , phys .
lett . * 91 * , 087201 ( 2003 ) .
a. k. hartmann and m. a. moore , phys .
lett . * 90 * , 127201 ( 2003 ) a. j. bray and m. a. moore , _ heidelberg colloquium on glassy dynamics _ , j. l. van hemmen and i. morgenstern , ( ed . ) , ( springer , berlin , 1986 ) , pp .
121 - 153 . | the statistics of the ground - state and domain - wall energies for the two - dimensional random - bond ising model on square lattices with independent , identically distributed bonds of probability @xmath0 of @xmath1 and @xmath2 of @xmath3 are studied .
we are able to consider large samples of up to @xmath4 spins by using sophisticated matching algorithms .
we study @xmath5 systems , but we also consider @xmath6 samples , for different aspect ratios @xmath7 .
we find that the scaling behavior of the ground - state energy and its sample - to - sample fluctuations inside the spin - glass region ( @xmath8 ) are characterized by simple scaling functions .
in particular , the fluctuations exhibit a cusp - like singularity at @xmath9 .
inside the spin - glass region the average domain - wall energy converges to a finite nonzero value as the sample size becomes infinite , holding @xmath10 fixed . here ,
large finite - size effects are visible , which can be explained for all @xmath0 by a single exponent @xmath11 , provided higher - order corrections to scaling are included .
finally , we confirm the validity of aspect - ratio scaling for @xmath12 : the distribution of the domain - wall energies converges to a gaussian for @xmath12 , although the domain walls of neighboring subsystems of size @xmath5 are not independent . |
a detailed analysis of @xmath2 and @xmath3 based on the itep sum rules @xcite has been performed several years ago @xcite . during the past years
new and more precise data for @xmath4 have become available in the low energy region , in particular for the parameters of the charmonium and bottomonium resonances .
furthermore , the error in the strong coupling constant @xmath5 which enters this analysis has been reduced .
last not least , the vacuum polarization induced by massive quarks has been computed in four - loop approximation ; more precisely : its first derivative at @xmath6 , which corresponds to the lowest moment of the familiar @xmath7-ratio has been evaluated in @xcite . based on these developments a new determination of the quark masses has been performed in ref .
more recently also the second moment has been calculated @xcite .
the extraction of @xmath8 from low moments of the cross section @xmath9 exploits its sharp rise close to the threshold for open charm and bottom production . by evaluating the moments @xmath10 with low values of @xmath11 ,
the long distance contributions are averaged out and @xmath12 involves short distance physics only , with a characteristic scale of order @xmath13 . through dispersion relations the moments
are directly related to derivatives of the vacuum polarization function at @xmath6 , @xmath14 which can be evaluated in perturbative qcd ( @xmath15 is the @xmath16 mass at the scale @xmath17 ) .
the perturbative series for the coefficients @xmath18 in order @xmath19 was originally evaluated up to @xmath20 in ref .
@xcite , and to `` arbitrary '' high order in @xcite .
the four - loop contributions to @xmath21 and @xmath22 were evaluated in refs .
@xcite , those to @xmath23 in @xcite .
combining eqs .
( [ eq::mexp ] ) and ( [ eq::mexp1 ] ) the quark mass can be extracted . at this point
it should be emphasized that the relative weight of resonances and continuum is quite different in the experimental moments .
furthermore , low moments are less sensitive to non - perturbative contributions from condensates , to the coulombic higher order effects , the variation of @xmath17 and the parametric @xmath24 dependence . for @xmath25 :
@xmath26 the moment with @xmath27 is less sensitive to data for @xmath28 from the continuum region above @xmath29 , where experimental results are scarce and the aforementioned theory uncertainties are still relatively small .
the agreement between @xmath25 and @xmath27 ( @xmath30 ) , together with the nice convergence with increasing order in @xmath24 can be considered as additional confirmation of this approach .
instead of measuring the moments @xmath31 in @xmath32 annihilation they can also be determined in lattice simulations .
this approach has recently been pioneered in @xcite using the highly improved staggered quarks ( hisq ) discretization of the quark action in combination with four - loop perturbative results @xcite .
the final result , @xmath33 corresponds to a scale - invariant mass @xmath34 and is in excellent agreement with the determinations based on @xmath32 data . and
@xmath35.[fig::mb_compare],title="fig : " ] and @xmath35.[fig::mb_compare],title="fig : " ] the approach based on @xmath32 data is also applicable to the determination of @xmath3 .
the three results based on @xmath36 and 3 are of comparable precision .
the relative size of the contributions from the threshold and the continuum region decreases for the moments @xmath27 and 3 . on the other hand , the theory uncertainty is still small .
therefore the result from @xmath27 was taken as the final answer @xcite , despite the fact that @xmath23 was not yet known .
the result , @xmath37 , corresponds to @xmath38 .
the recent evaluation @xcite of @xmath23 has lead to a decrease of the central value by @xmath39 and a reduction of the error from @xmath40 to @xmath41 .
a comparison of a few selected @xmath3- and @xmath2-determinations is shown in fig .
[ fig::mb_compare ] .
one of the most precise and theoretically safe determination of @xmath24 is based on measurements of the cross section for electron - positron annihilation into hadrons @xcite .
these have been performed in the low - energy region between 2 gev and 10 gev and , in particular , at and around the @xmath1 resonance at 91.2 gev .
conceptually closely related is the measurement of the semileptonic decay rate of the @xmath0-lepton , leading to a determination of @xmath24 at a scale below 2 gev @xcite .
the perturbative expansion for the ratio @xmath42 in numerical form is given by @xmath43 here @xmath44 and the normalization scale @xmath45 .
the @xmath46 corrections are conveniently classified according to their power of @xmath47 , with @xmath47 denoting the number of light quarks .
the @xmath48 term is part of the `` renormalon chain '' , the next term of order @xmath49 was evaluated in @xcite , the complete five - loop calculation has been performed in @xcite .
let us now move to the analysis of present data for @xmath32 annihilation and @xmath0 decays .
measurements of @xmath28 at lower energies , with their correspondingly larger values of @xmath24 , are in principle more sensitive to @xmath5 if the same relative precision could be obtained . at present , however , the systematic experimental error of 2% is a limiting element for a competitive measurement .
the final result of a recent analysis @xcite @xmath50 represents the combined information on the strong coupling from @xmath7 measurements in the region between @xmath51 and the bottom threshold .
in contrast to the situation below 10 gev the extraction of @xmath24 from @xmath1-decays is affected by the @xmath52 terms .
the @xmath53 analysis of the electroweak working group @xcite is shifted @xcite by @xmath54 and one finds @xmath55 the theory error can now safely be neglected .
higher orders are of particular relevance in the low - energy region , for example in @xmath0 decays .
the correction from perturbative qcd to the ratio @xmath56 is given by the factor @xmath57 which can be evaluated in fixed order perturbation theory or with `` contour improvement '' as proposed in @xcite : @xmath58 starting from @xmath59 @xcite one obtains @xcite @xmath60 and @xmath61 , respectively .
the second uncertainty corresponds to a change in the renormalization scale @xmath62 between @xmath63 and @xmath64 .
as stated above the theory error for @xmath24 from @xmath1 decays is small compared to the experimental uncertainties . for @xmath0 decays the difference between fort and cipt must be considered as irreducible uncertainty @xcite : @xmath65 applying four - loop running and matching @xcite ( with negligible error from the evolution from @xmath66 to @xmath67 ) : @xmath68 the shifts in @xmath5 from z- and @xmath0-decays , are opposite in sign and move the values in the _ proper _ direction , decreasing , thus , the current slight mismatch between two independent determinations of @xmath24 .
the two results are in remarkable agreement and can be combined to one of the most precise and presently only @xmath69 result : @xmath70 as discussed above , @xmath24 from @xmath0-decays is strongly affected by theory uncertainties .
this is reflected in three recent publications which all are based on the same set of @xmath0 data and the new five - loop results , which , however , arrive at significantly different estimates of higher order terms and hence of @xmath24 . in @xcite
it is argued that fopt exhibits a poorly `` convergent '' series , in contrast to cipt , where a subset of higher order contributions ( so - called @xmath71-terms ) is automatically summed .
this analysis leads to @xmath72 .
the small theory uncertainty is a consequence of the ( artificially ) reduced @xmath17-dependence and the restrictive assumptions about higher orders in perturbation theory .
the opposite viewpoint has been advocated in @xcite where an explicit and plausible behaviour of the perturbative series is modelled .
strong cancellations are observed between the aforementioned @xmath71-terms and those terms which can only be obtained from a concrete higher order calculation .
all these papers are based on similar assumptions about non - perturbative power - suppressed contributions .
these assumptions have been questioned in @xcite , employing different weight functions in order to suppress these poorly determined terms with dimension @xmath73 .
a significantly smaller result , @xmath74 , is obtained within cipt . in total the spread among the different results is covered reasonably well by eq .
( [ als_tau ] ) .
last not least it is instructive to compare eqs .
( [ alpha_mz_new ] ) and ( [ eq : asres_mz ] ) with @xmath24 determinations based on lattice simulations .
the most recent results from a precision simultaion by the hpqcd collaboration @xcite @xmath75 and from a study @xcite based on essentially the same lattice data and an alternative perturbative analysis , @xmath76 are in mutual agreement , as should be expected because they are based on a similar set of data . remarkably enough , a completely different approach @xcite , which exploits the lattice simulation of the pseudoscalar correlator and which has been mentioned already above leads to @xmath77 again well consistent with the other determinations .
v. a. novikov , l. b. okun , m. a. shifman , a. i. vainshtein , m. b. voloshin and v. i. zakharov , phys .
* 41 * ( 1978 ) 1 .
j. h. khn and m. steinhauser , nucl .
b * 619 * ( 2001 ) 588 [ erratum - ibid .
b * 640 * ( 2002 ) 415 ] [ arxiv : hep - ph/0109084 ] .
k. g. chetyrkin , j. h. khn and c. sturm , eur .
j. c * 48 * ( 2006 ) 107 [ arxiv : hep - ph/0604234 ] .
r. boughezal , m. czakon and t. schutzmeier , phys .
d * 74 * ( 2006 ) 074006 [ arxiv : hep - ph/0605023 ] .
, , , * * , ( ) , .
k. g. chetyrkin , j. h. khn and m. steinhauser , phys .
b * 371 * ( 1996 ) 93 [ arxiv : hep - ph/9511430 ] ; nucl .
b * 505 * ( 1997 ) 40 [ arxiv : hep - ph/9705254 ] .
r. boughezal , m. czakon and t. schutzmeier , nucl .
* 160 * ( 2006 ) 160 [ arxiv : hep - ph/0607141 ] .
a. maier , p. maierhofer and p. marquard , nucl .
b * 797 * ( 2008 ) 218 [ arxiv:0711.2636 [ hep - ph ] ] .
a. maier , p. maierhfer , p. marquard , [ arxiv:0806.3405[hep - ph ] ] and these proceedings .
m. davier , s. descotes - genon , a. hocker , b. malaescu and z. zhang , arxiv:0803.0979 [ hep - ph ] . m. beneke and m. jamin , arxiv:0806.3156 [ hep - ph ] .
k. maltman and t. yavin , arxiv:0807.0650 [ hep - ph ] .
c. t. h. davies _ et al _ [ hpqcd collaboration ] , arxiv:0807.1687 [ hep - lat ] . k. maltman , d. leinweber , p. moran and a. sternbeck , arxiv:0807.2020 [ hep - lat ] . | the impact of recent multi - loop calculations on precise determinations of charm- and bottom - quark masses and the strong coupling constant is discussed .
qcd has developed from a qualitative model for strong interactions into a quantitative theory with precise predictions for a multitude of observables .
this development , which has taken place in particular during the past two decades , is due to significant improvements in our theoretical understanding , in calculational techniques and in improved experimental results , ranging from low energy studies in @xmath0-lepton decays through precise cross section measurements in electron - positron annihilation up to measurements of the @xmath1-boson decay rate into hadrons with permille accuracy .
many of these inclusive observables are related to essentially the same object in quantum field theory , namely the absorptive part of the current - current correlator . restricting the discussion to measurements at high energies or to properly chosen integrals
, these observables can be evaluated in perturbation theory and lead to precise determinations of the fundamental parameters of qcd , quark masses and the strong coupling constant . |
the ongoing search for the direct detection of gravitational waves using earth based experiments involves the analysis of observations made at a variety of different detectors .
these observations are time series samples of the detector state that are then processed by various means to identify gravitational wave candidates . broadly speaking the searches for gravitational waves
may be broken up into two categories , those searches that are based upon a model , such as the search for gravitational waves from inspiraling binaries , and those searches that endeavor to find gravitational waves without using a model .
the latter category of searches are often referred to as `` burst '' searches .
these searches typically seek to identify portions of data that are , for a short period , anomalously `` loud '' in comparison to the surrounding data . identifying a gravitational wave burst in the absence of a source model
is an involved and potentially computationally expensive process .
this is especially true when the ratio of signal power to noise power is low .
a convenient and natural approach to mitigating the computational expense of identifying such bursts divides the problem of detection into two parts . in the first part
, an inexpensive procedure is used to identify candidate sections of data that trigger the second part of the analysis .
the second part of the analysis focuses on the subintervals of data identified by the first .
it is a more computationally complex and expensive analysis that either discards the candidate or identifies it as a gravitational wave burst . in this way the first part of the analysis carried out by a so - called `` event - trigger generator '' , performs triage on the data that must be analyzed by the more complex second stage of the analysis .
blocknormal is an event trigger generator that analyzes data in the time domain and searches for moments in time where the statistical character of the time series data changes . in particular , blocknormal characterizes the time series between change - points by the mean(@xmath0 ) and variance(@xmath1 ) of the samples .
change - points are thus demarcation points , separating `` blocks '' of data that are consistent with a distribution having a given mean and variance , which differs from the mean and/or variance that best characterizes the data in an adjacent block .
the onset of a signal in the data will , because it is uncorrelated with the detector noise , increase the variance of the time series for as long as the signal is present with significant power . in this way blocks with variance greater than a `` background '' variance mark candidate gravitational wave bursts .
since candidate gravitational wave bursts are identified with changes in the detector noise character it is best if the detector noise is itself stationary and white .
the blocknormal analysis thus starts by identifying long segments of data , epochs , which are relatively stationary .
this process involves comparing adjacent stretches of data of fixed a duration long relative to the expected duration of a gravitational wave burst and asking whether adjacent stretches have consistent means and variances . in order to avoid any bias that might come from analyzing outliers in the tails of the distribution ( where one might expect any true signal to be located ) , the mean and variance
are computed on only those samples that are within the 2.5th and 97.5th percentiles of the sample values .
if two consecutive stretches are inconsistent at the @xmath2 confidence level , the begining of the first stretch is used to define a new stationary segment .
segments thus defined are split into a set of frequency bands whose lower band edge is heterodyned to zero frequency .
this base - banding allows for a crude determination of the frequency of any identified triggers .
line and other spectral features are removed , either by kalman filtering or by regression against diagnostic channels , and the final data whitened with a linear filter .
once the data has undergone the base - banding and whitening process , the search for change - points begins in earnest .
the method employed is similar to that described in @xcite and relies on a bayesian analysis of the relative probability of two different hypotheses : * @xmath3 , that the time series segment @xmath4 is drawn from a distribution characterized by a single mean and variance ; and * @xmath5 , that the time series segment @xmath4 consists of two continuous and adjacent subsegments each drawn from a distribution characterized by a different mean and/or variance . given the time series segment @xmath4 , consisting of @xmath6 samples , we write the probability of hypothesis @xmath3 as @xmath7 and the probability of hypothesis @xmath5 as @xmath8 .
the odds of @xmath5 compared to @xmath3 is thus : @xmath9 applying bayes theorem and simplifying this becomes : @xmath10 where , @xmath11 is independent of the data @xmath4 itself , although it does depend on the number of samples @xmath6 .
if @xmath5 is true , then @xmath3 will be a good hypothesis for two subsets of the data , one from sample 1 to @xmath12 , denoted @xmath13 and another from @xmath14 to @xmath6 , @xmath15
. then we can write : @xmath16 to calculate @xmath17 we thus need only be able to calculate @xmath18 for arbitrary time series @xmath4 the probability that a given data set is drawn from a normal distribution with unknown mean and variance is equal to : @xmath19-\mu)^{2}}{2\sigma^{2}}}\end{aligned}\ ] ] where @xmath20 is the a priori probability that the mean takes on a value @xmath0 and the variance a value @xmath21 . with the usual uninformative priors for @xmath0 and @xmath22 ( @xmath23 and @xmath24 )
the integral for @xmath18 can be evaluated in closed form : @xmath25^{-\frac{(n-1)}{2}}i_{n-2 } \\
i_{n } \equiv ( n-1 ) ! ! \left\{\begin{array}{ll}1
& \mbox{$n$ odd } \\ \sqrt{\frac{\pi}{2 } } & \mbox{$n$ even } \end{array } \right.\end{aligned}\ ] ] where @xmath26 an @xmath27 .
the value of @xmath17 is therefore the odds that there is a change - point in @xmath28 to there not being any change - points , and the value @xmath29 is related to the odds that there is a change - point at sample @xmath14 to there not being one anywhere in @xmath4 .
the calculated value of @xmath17 is compared to a threshold , @xmath30 , and if greater than this threshold , a change - point is considered to be at the sample with the largest value of @xmath29 .
figure [ signal ] shows some simulated data along with @xmath29 for that data .
the two peaks in the value of @xmath29 corresponds to where the mean of the simulated noise changes .
[ signal ] as a function of the hypothetical change point time.,title="fig : " ] this process either leaves @xmath4 free of change - points , or it divides the data into two subsets . in the latter case , the blocknormal pipeline repeats the change - point analysis on these subsets and all subsequent subsets , until either no more change - points are found , or the subset is less than 4 data points long . by this iteration method
blocknormal breaks the data at these change - points into a set of blocks , each of which is free of any change point .
a final refinement step is taken where successive pairs of blocks are analyzed to check that the change - point would still be considered significant over the subset of the data that is contained within the two blocks .
blocknormal identifies blocks time series segments between successive change - points are well characterized by a mean , a variance , a frequency band , a start time , and a duration . to identify unusual blocks , the means and variances
are compared to the mean and variance , @xmath31 and @xmath32 of all the data in their band from the epoch in which they were found .
blocknormal defines `` events '' to be blocks in which the following condition holds true for a value , @xmath33 , that is used to characterize the block : @xmath34 here , @xmath35 , is called the event threshold and is a free parameter in the algorithm which adjusts the sensitivity in defining what `` unusual '' means .
there are a limited number of reasonable other possible threshold requirements based on the three characteristics of a block , @xmath0,@xmath22,@xmath36 , however , at this time these have not been explored .
once events have been identified , immediately adjacent events in the same frequency band are clustered together , with a peak - time for the cluster defined by the central time of the block with the largest value of @xmath37 .
a cluster `` energy '' is a sum over the @xmath38 blocks that comprise it : @xmath39 \end{aligned}\ ] ] where @xmath40 is the duration in samples .
a key factor in building confidence in any identification of gravitational waves is the presence of a signal in different detectors . using this `` coincidence '' as the basis for further reducing the number of periods of interest , blocknormal requires that there be coincidence in time between events in the same band but different detectors before a `` trigger ''
is formed . triggers from different bands
are merged if they overlap in time into a single trigger .
figure [ coinc ] illustrates how this coincidence and merging step works using the three ligo interferometers .
the blocknormal event trigger generator is a time domain analysis in base - banded data that identifies blocks in time which are well characterized by a mean and variance .
blocks with means and variances that are unusually large compared to the mean and variance of the much longer data segment containing them are marked for consideration as being unusual events .
several coincident events in different detectors together form a trigger which can be used to define periods of interest that a more computationally intense analysis can use to reduce the total computing time needed to search for unmodeled gravitational wave events .
this work was supported by the center for gravitational wave physics , the international virtual data grid laboratory , and the national science foundation under award phy 00 - 99559 .
the international virtual data grid laboratory is supported by the national science foundation under cooperative agreement phy-0122557 ; the center for gravitational wave physics is supported by the national science foundation under cooperative agreement phy 01- 14375 .
1 smith a f m 1975 _ biometrika _ * 62*,2 p. 407416 | in the search for unmodeled gravitational wave bursts , there are a variety of methods that have been proposed to generate candidate events from time series data .
block normal is a method of identifying candidate events by searching for places in the data stream where the characteristic statistics of the data change .
these change - points divide the data into blocks in which the characteristics of the block are stationary .
blocks in which these characteristics are inconsistent with the long term characteristic statistics are marked as event - triggers which can then be investigated by a more computationally demanding multi - detector analysis . |
the phobos detector at rhic was designed to offer an overview of particle production in the collisions of a variety of systems ( p+p , d+au , au+au ) over a wide range of control parameters : beam energy , pseudorapidity , and centrality .
the phobos apparatus was upgraded substantially for the 2003 d+au rhic run .
two calorimeters ( pcal ) were added to provide information on spectator protons from the d and au projectiles .
another major upgrade involved the phobos particle - identification system .
the two time - of - flight ( tof ) walls were moved back from their design position to improve the mass resolution .
time resolution was improved by the addition of two sets of 10 cerenkov timing start counters ( t0 ) .
most importantly , two small hodoscopes were added just behind the phobos spectrometer to provide the means to trigger on high - momentum particles which also hit the tof walls . this spectrometer trigger ( spectrig )
was installed to enrich the sample of identified particles at high transverse momentum ( @xmath17 gev / c ) .
( integrated over @xmath18 ) divided by @xmath19 vs. @xmath0 .
error bars are 90% c.l . while the grey band expresses the systematic error on @xmath0 in au+au .
[ dauntot ] , width=264 ] ( integrated over @xmath18 ) divided by @xmath19 vs. @xmath0 .
error bars are 90% c.l . while the grey band expresses the systematic error on @xmath0 in au+au .
[ dauntot ] , width=302 ] in 2003 , phobos took data at @xmath20 gev for both p+p and d+au collisions .
minimum - bias d+au data has already been released @xcite . as with the d+au data ,
the p+p data were corrected for trigger bias in an attempt to estimate the pseudorapidity distribution for the total inelastic cross section . using the signals in the phobos ring detectors ( @xmath21 )
, the d+au data has been divided into centrality bins , which are shown in fig .
[ daudndeta ] . as with au+au collisions , both limiting fragmentation relative to lower - energy proton - nucleus data @xcite , and participant scaling of the total multiplicity @xcite
are observed , the latter being shown in fig .
[ dauntot ] .
these results are somewhat surprising since different dynamical effects could be expected to dominate in the deuteron and gold hemispheres , such as energy stopping and secondary cascading .
the robustness of the @xmath0 scaling perhaps suggests some sort of long - range correlation among the produced particles .
the different constant of proportionality between the p+p / d+au data and the au+au data also suggests a fundamental difference between the two , once moderate centralities are reached in au+au .
these results , along with the detailed energy dependence of pseudorapidity distributions in p+a and d+a , are discussed in more detail in ref .
as a function of @xmath22 for au+au collisions at three energies .
statistical errors are shown as bars .
systematic errors are suppressed in this figure .
[ v1eprime ] , height=226 ] as a function of @xmath22 for au+au collisions at three energies .
statistical errors are shown as bars .
systematic errors are suppressed in this figure .
[ v1eprime ] , height=226 ] phobos results on @xmath23 as a function of pseudorapidity have been shown previously for data integrated over all available centralities @xcite .
preliminary results have extended this measurement by a division of the data into three centrality bins , shown in fig .
[ v2cent ] .
these results show that the shape is manifestly not boost invariant over all @xmath4 in peripheral collisions ( although one must not ignore kinematic effects near midrapidity , as pointed out by kolb @xcite ) . moreover ,
the shape does not change dramatically as a function of centrality , apart from an overall scale factor .
new at qm2004 are results on @xmath24 , the first harmonic of the azimuthal angular distribution , as a function of pseudorapidity and beam energy .
a simple picture , which considers the effect of the net baryons , would suggest that @xmath24 should develop opposite signs for baryons and mesons , with the baryons defined to have a positive directed flow signal near projectile rapidities .
our results on @xmath24 show a striking energy dependence , which leads to a large range in @xmath4 with @xmath25 for the higher energies and then a small but systematic negative @xmath24 that is approximately independent of energy when viewed relative to beam rapidity , shown in fig .
[ v1eprime ] .
this may be apparently another manifestation of `` limiting fragmentation '' , already seen in the pseudorapidity distributions of charged particles @xcite .
studies such as these should be useful in understanding the global dynamics of au+au collisions .
these results , along with comparisons to other experimental data , are explained in more detail in ref .
vs. @xmath4 for a bin width of @xmath26 .
[ sigmac ] , width=264 ] vs. @xmath4 for a bin width of @xmath26 .
[ sigmac ] , height=264 ] another approach to understanding the role of correlations in rapidity space is the use of multi - particle correlation measures like two - particle hbt correlations and forward - backward correlations .
phobos has performed the first measurement of the rapidity dependence of hbt correlations at rhic . using the ykp ( yano - koonin - podgoretskii ) parametrization , we have also been able to correlate the rapidity of the local source as a function of the average rapidities of the two particles .
shown in fig .
[ yykp ] , one sees that these two rapidities are tightly correlated , suggesting that particles produced at a certain rapidity were produced by a source moving collectively at the same rapidity .
this result is similar that found by the na49 experiment at the sps .
these results are explained in more detail in ref .
@xcite .
another indication that multiparticle correlations reflect physics local in rapidity comes from the measurement of forward - backward correlations .
one would expect a strong correlation between particles produced in rapidity bins in opposite hemispheres , simply from the fact that particle production is controlled by the number of participants .
the quantitative analysis of relative fluctuations can be used to assess the level of short range correlations , which should lead to a non - binomial distribution of multiplicities between two identical bins in the forward ( @xmath27 ) and backward ( @xmath28 ) directions . by analyzing the variance of a variable @xmath29 , @xmath30 , as a function of the total multiplicity @xmath31 , compared to a detailed monte carlo simulation of our apparatus
, we can study non - binomial contributions to the fluctuations . in fig .
[ sigmac ] , we show data on @xmath30 for a bin width @xmath32 vs. @xmath4 as open squares
. results from hijing events are shown for primary charged particles ( closed circles ) and full simulations ( open circles ) .
the deviations of @xmath30 for the primary charged particles from unity indicate that short - range correlations exist with a mild pseudorapidity dependence for @xmath33 .
the agreement of the data with the fully - simulated hijing suggests similar correlations are present in our data set .
these results are shown for a different values of @xmath34 and discussed in detail in ref .
@xcite . and
@xmath35 ratios vs. @xmath36 in 200 gev d+au collisions .
statistical errors are shown as bars .
systematic errors are indicated by brackets .
[ ratiosvspt ] , width=302 ] and @xmath35 ratios vs. @xmath36 in 200 gev d+au collisions .
statistical errors are shown as bars .
systematic errors are indicated by brackets .
[ ratiosvspt ] , width=226 ]
one of the major upgrades to the phobos detector in 2003 was the rearrangement of the tof walls and the addition of a trigger to increase our statistics of events with a high - momentum track in the spectrometer . with the enriched data set , we were able to make several new measurements of identified particles at high @xmath36 in d+au collisions .
these results are explained in more detail in ref .
. ratios of identified particles and anti - particles are a tool to study baryon transport .
various models predict a substantial transport of the initial baryon number of the deuteron all the way to mid - rapidity , leading to a progressive decrease in the @xmath37 ratio as a function of the number of collisions of each deuteron participant ( characterized by @xmath38 ) @xcite .
our results for this ratio using @xmath39 pid , shown in fig .
[ ratiosvsnu ] , show surprisingly little centrality dependence for particle momenta below 1 gev / c and essentially the same result is found in p+p and d+au . using the tof to extend our particle identification from 1.75 gev / c to 3 gev / c
, this ratio shows a similarly weak centrality dependence .
moreover , this ratio is also weakly dependent on @xmath36 , as seen in fig .
[ ratiosvspt ] .
this suggests that proton and anti - proton spectra are similar , except for a difference in overall yield .
the increased statistics in the spectrometer has enabled phobos to measure spectra of identified pions , kaons , and protons of both charge signs up to 3.5 gev .
the presence of a `` cronin effect '' in d+au collisions at rhic energies has raised the question of whether any particular species is responsible for this effect .
we have found that the relative contributions to the inclusive spectra vary substantially with @xmath36 , especially the growing fraction of protons and anti - protons , as shown by the invariant yields plotted vs. @xmath36 in fig .
[ pt_mt_pos ] . however , when the yields are shown as a function of @xmath2 , the relative contributions appear to have little @xmath2-dependence . in other words ,
an apparent @xmath2 scaling is evident in the @xmath40 data .
this is also seen in @xmath41 data , but is dramatically violated at low momentum in au+au @xcite .
( left ) and @xmath2 ( right ) .
the open and closed symbols represent two different bending directions .
the data are not corrected for contributions from weak decays .
statistical errors are shown as bars .
systematic errors are indicated by brackets .
[ pt_mt_pos],height=226 ]
one of the more striking results from the 2003 data is the observation of a significant @xmath4 dependence in the modification of @xmath40 spectra relative to @xmath41 data @xcite .
this has been interpreted as an indication of the onset of parton saturation @xcite in the forward region . to address this topic ,
phobos has repeated its measurement of @xmath3 , originally integrated over our full acceptance ( @xmath42 ) @xcite , but now in three bins ( @xmath43 , @xmath44 , and @xmath45 ) . even within the acceptance of the phobos spectrometer , @xmath46 units , a dramatic drop in this ratio is seen as a function of @xmath4 . when compared with other results from rhic in fig .
[ rdauvseta ] , it is found that the suppression seen at forward rapidities is a continuation of a trend started already at midrapidity .
measurements in the backward hemisphere by phenix show a clear enhancement of pion production with @xmath47 gev / c @xcite , suggesting that this is not an effect that `` turns on '' in the forward region , but rather part of a difference between @xmath41 and @xmath40 data that extends over a large range in @xmath4 . at fixed values of @xmath36 shown as a function of @xmath4 , compared with results from other rhic experiments .
error bars are statistical only .
[ rdauvseta ] , height=264 ]
recent p+p , d+au , and au+au results at 200 gev from the phobos experiment at rhic have been presented .
the systematic measurements shown here should serve as a useful baseline for understanding the differences between heavy ion collisions and more elementary systems .
this work was partially supported by u.s .
doe grants de - ac02 - 98ch10886 , de - fg02 - 93er40802 , de - fc02 - 94er40818 , de - fg02 - 94er40865 , de - fg02 - 99er41099 , and w-31 - 109-eng-38 , by u.s .
nsf grants 9603486 , 0072204 , and 0245011 , by polish kbn grant 2-p03b-10323 , and by nsc of taiwan under contract nsc 89 - 2112-m-008 - 024 .
b b back , arxiv : nucl - ex/0311009 . w. busza , _ acta phys .
b _ * 8 * , 333 ( 1977 ) .
j. e. elias , d * 22 * , 13 ( 1980 ) .
b. b. back , arxiv : nucl - ex/0301017 .
r. nouicer , these proceedings .
b. b. back , , 222301 ( 2002 ) .
s. manly , a * 715 * , 611 ( 2003 ) .
p. f. kolb , _ acta phys . hung .
. heavy ion phys . _ * 15 * , 279 ( 2002 ) .
b. b. back , , 052303 ( 2003 ) .
m. belt tonjes , these proceedings .
b. holzman , these proceedings .
k. wozniak , these proceedings .
g. i. veres , these proceedings .
b. b. back , arxiv : nucl - ex/0309013 . b. b. back , arxiv : nucl - ex/0401006 . i. arsene , arxiv : nucl - ex/0403005 . b. b. back , , 072302 ( 2003 ) .
j.jalilian-marian , these proceedings .
liu , these proceedings . | recent results from the phobos experiment at rhic are presented , both from au+au collisions from the 2001 run and p+p and d+au collisions from 2003 . the centrality dependence of the total charged particle multiplicity in p+p and d+au show features , such as @xmath0-scaling and limiting fragmentation , similar to @xmath1 collisions at lower energies .
multiparticle physics in au+au is found to be local in ( pseudo)rapidity , both when observed by hbt correlations and by forward - backward pseudorapidity correlations .
the shape of elliptic flow in au+au , measured over the full range of pseudorapidity , appears to have a very weak centrality dependence .
identified particle ratios in d+au reactions show little difference between the shape of proton and anti - proton spectra , while the absolute yields show an approximate @xmath2 scaling .
finally , results on @xmath3 as a function of pseudorapidity , show that this ratio decreases monotonically with @xmath4 , even between @xmath5 .
b b back@xmath6 , m d baker@xmath7 , m ballintijn@xmath8 , d s barton@xmath7 , b becker@xmath7 , r r betts@xmath9 , a a bickley@xmath10 , r bindel@xmath10 , a budzanowski@xmath11 , w busza@xmath8 , a carroll@xmath7 , m p decowski@xmath8 , e garca@xmath9 , t gburek@xmath11 , n george@xmath12 , k gulbrandsen@xmath8 , s gushue@xmath7 , c halliwell@xmath9 , j hamblen@xmath13 , a s harrington@xmath13 , g a heintzelman@xmath7 , c henderson@xmath8 , d j hofman@xmath9 , r s hollis@xmath9 , r hoyski@xmath11 , b holzman@xmath7 , a iordanova@xmath9 , e johnson@xmath13 , j l kane@xmath8 , j katzy@xmath14 , n khan@xmath13 , w kucewicz@xmath9 , p kulinich@xmath8 , c m kuo@xmath15 , j w lee@xmath8 , w t lin@xmath15 , s manly@xmath13 , d mcleod@xmath9 , a c mignerey@xmath10 , r nouicer@xmath16 , a olszewski@xmath11 , r pak@xmath7 , i c park@xmath13 , h pernegger@xmath8 , c reed@xmath8 , l p remsberg@xmath7 , m reuter@xmath9 , c roland@xmath8 , g roland@xmath8 , l rosenberg@xmath8 , j sagerer@xmath9 , p sarin@xmath8 , p sawicki@xmath11 , i sedykh@xmath7 , w skulski@xmath13 , c e smith@xmath9 , g s f stephans@xmath8 , a sukhanov@xmath7 , j -l tang@xmath15 , m b tonjes@xmath10 , a trzupek@xmath11 , c vale@xmath8 , g j van nieuwenhuizen@xmath8 , r verdier@xmath8 , g i veres@xmath8 , f l h wolfs@xmath13 , b wosiek@xmath11 , k woniak@xmath11 , a h wuosmaa@xmath6 , b wysouch@xmath8 and j zhang@xmath8 + |
discrete element method ( dem ) simulations are used extensively to study the flow and segregation of granular materials in many situations as a predictive tool and access to data that are otherwise difficult to obtain experimentally .
one of the key aspects of any simulation approach is the implementation of boundary conditions at walls .
two types of wall boundary conditions can be implemented in dem simulations : 1 ) geometrically smooth surfaces @xcite , which are assumed to have infinite mass and a specified radius of curvature ( infinite for planar walls ) ; and 2 ) a geometrically rough surface made up of a closely packed monolayer of fixed particles conforming to the geometry of the wall surface ( for example , see @xcite ) .
however , a recent study of monodisperse flow in a spherical tumbler suggests that the results using the latter approach , often called a `` rough wall , '' differ from those using a smooth wall , not only locally at the particle scale but also globally across the entire flowing layer @xcite . in this paper
, we explore the impact of rough and smooth walls on the axial segregation of bidisperse particles in a partially - filled spherical tumbler rotating with angular velocity @xmath0 about a horizontal axis .
we consider the situation where the free surface is essentially flat and continuously flowing . in this regime ,
the surface of the flowing layer maintains a dynamic angle of repose @xmath1 with respect to horizontal that depends on the frictional properties , the diameter @xmath2 of the particles , and the rotational speed of the tumbler @xcite . in experiments with
spherical tumblers approximately half filled with a 50%-50% size bidisperse mixture of particles and with smooth walls , large particles accumulate near the poles of the tumbler with a band of small particles at the equator @xcite .
this pattern is inverted for lower fill fractions so that small particles accumulate near the poles with a band of large particles at the equator @xcite .
we note that multiple bands of small and large particles occur for bidisperse mixtures in long rotating cylindrical tumblers , which are used in applications for materials ranging from foodstuffs to mining to cement , typically after o(10 ) to o(100 ) rotations and having a wavelength of about one tumbler diameter @xcite under a wide range of conditions @xcite . in the cylindrical tumbler case , however , large particles segregate near the flat end walls of the tumbler @xcite , as a consequence of radial segregation combined with the nonuniform axial distribution of velocity in the flowing layer due to friction at the endwall @xcite .
this mechanism can not occur in the spherical tumblers studied here .
although dem simulations of spherical tumblers with smooth walls readily reproduce the segregation experiments in an acrylic spherical tumbler @xcite , the mechanism for the inversion of the segregation bands from large - small - large ( lsl ) for higher fill fractions to small - large - small ( sls ) at lower fill fractions remains unresolved . in an initial effort to understand the segregation mechanism
, we attempted further dem simulations at different fill fractions in spherical tumblers .
however , to simplify the implementation of the simulations , we performed the simulations with a rough wall boundary conditions .
the results are dramatically different from the simulations with smooth walls , as shown in fig .
[ roughsmooth30p ] . for
otherwise identical systems ( same particle sizes , rotational speed , tumbler diameter , and fill fraction ) , the surface segregation pattern changes from sls for smooth walls ( fig .
[ roughsmooth30p](a ) ) to lsl for rough walls made of 2 mm diameter particles ( fig .
[ roughsmooth30p](c ) ) . using an intermediate wall particle size of 1.5 mm results in no significant surface band formation at all ( fig .
[ roughsmooth30p](b ) ) . the unexpectedly strong influence of wall roughness on band formation led us to first study monodisperse flows in a spherical tumbler @xcite .
for monodisperse flows , the wall roughness strongly affects the particles trajectories , even far from walls .
particle trajectories at the free surface curve further toward the poles for smooth walls than for rough walls .
however , the particle trajectories curve back toward the tumbler equator more in the smooth case as well , resulting in a smaller net poleward drift at the surface for smooth walls than for rough walls . in this paper
, we examine through both dem simulations and qualitative experiments the impact of wall boundary roughness on band formation in bidisperse flows . 200 s.
the tumbler walls are ( a ) smooth , ( b ) rough wall of 1.5 mm particles and ( c ) rough wall of 2 mm particles .
the rotation axis is horizontal and arrows show the direction of the surface flow . ]
since the different results for smooth and rough walls were initially obtained via dem simulations , it was crucial to confirm that the predicted segregation patterns did in fact occur experimentally .
qualitative experiments were performed using clear acrylic spheres consisting of two mating hemispheres of diameter @xmath3 cm rotated by an electric motor at 14.7 rpm about a horizontal axis .
the tumbler was filled to 30% by volume with equal volumes of @xmath4 mm and @xmath5 mm diameter spherical glass particles .
for the rough wall case , the small particles were bonded to the wall of a tumbler using epoxy , thereby reducing the tumbler inner diameter to 13.6 cm .
the tumbler was rotated for approximately 100 rotations and stopped so that all the particles were in one hemisphere .
then the upper hemisphere of the spherical tumbler was removed to obtain an image of the surface segregation pattern .
the sls segregation pattern occurs for the smooth tumbler wall ( fig .
[ exproughsmooth](a ) ) , while the lsl segregation pattern occurs for the rough wall having 2 mm particles bonded to it ( fig .
[ exproughsmooth](b ) ) , confirming the validity of the dem results and the surprising effect of the wall roughness on the segregation pattern .
for the dem simulations , a standard linear - spring and viscous damper force model @xcite was used to calculate the normal force between two contacting particles : @xmath6\hat r_{ij}$ ] , where @xmath7 and @xmath8 are the particle overlap and the relative velocity @xmath9 of contacting particles @xmath10 and @xmath11 respectively ; @xmath12 is the unit vector in the direction between particles @xmath10 and @xmath11 ; @xmath13 is the reduced mass of the two particles ; @xmath14 $ ] is the normal stiffness and @xmath15 is the normal damping , where @xmath16 is the collision time and @xmath17 is the restitution coefficient @xcite .
a standard tangential force model @xcite with elasticity was implemented : @xmath18 , where @xmath19 is the relative tangential velocity of two particles @xcite , @xmath20 is the tangential stiffness , @xmath21 the coulomb friction coefficient and @xmath22 is the net tangential displacement after contact is first established at time @xmath23 .
the velocity - verlet algorithm @xcite was used to update the position , orientation , and linear and angular velocity of each particle .
tumbler walls were modeled as both smooth surfaces ( smooth walls ) and as a monolayer of bonded particles ( rough walls ) .
both wall conditions had infinite mass for calculation of the collision force between the tumbling particles and the wall .
the spherical tumbler of radius @xmath24 cm was filled to volume fraction @xmath25 with equal volumes of small and large particles of diameter 2 and 4 mm ; gravitational acceleration was @xmath26 = 9.81 m s@xmath27 ; particle properties correspond to cellulose acetate : density @xmath28 1308 kg m@xmath29 , restitution coefficient @xmath30 @xcite .
the two species were initially randomly distributed in the tumbler with a total of about @xmath31 particles in a typical simulation . to avoid a close - packed structure , the particles had a uniform size distribution ranging from 0.95@xmath2 to 1.05@xmath2 . unless otherwise indicated , the friction coefficients between particles and between particles and walls was set to @xmath32 .
the collision time was @xmath16 = 10@xmath33 s , consistent with previous simulations @xcite and sufficient for modeling hard spheres @xcite .
these parameters correspond to a stiffness coefficient @xmath34 ( n m@xmath35 ) @xcite and a damping coefficient @xmath36 kg s@xmath35 .
the integration time step was @xmath37 s to meet the requirement of numerical stability @xcite .
: ( a ) smooth , ( b ) rough 1.5 mm particles and ( c ) rough 2 mm particles . ] the steady - state concentration profiles corresponding to the segregation patterns in fig . [ roughsmooth30p ]
are shown in fig .
[ proroughsmooth ] .
a relative concentration of 1.0 corresponds to pure particles of one size .
the profiles were obtained by determining the size of the particles intersecting a series of planes perpendicular to the axis of rotation ( with @xmath38 at the equator of the tumbler ) and extracting the volume concentration for each species in that plane .
this approach allows a much higher axial resolution for the concentration measurements than can be achieved with simple binning . for the sls pattern in the smooth wall tumbler ( fig .
[ proroughsmooth](a ) ) , the three bands are nearly pure ( one particle size or the other for most of the width of each band ) . on the other hand , for the lsl pattern in the 2 mm rough wall tumbler ( fig . [ proroughsmooth](c ) ) , the bands are less pure and the transition between bands is not as sharp .
the 1.5 mm rough wall ( fig . [ proroughsmooth](b ) ) results in a configuration intermediate between the two other cases .
the underlying reason for the nature of the concentration profiles is evident when viewing a cross - section in a vertical plane and containing the axis of rotation ( fig .
[ sliceroughsmooth30p ] ) . for the sls segregation pattern ,
the small particles and large particles form more distinct bands through the depth of the particle bed ( fig .
[ sliceroughsmooth30p](a ) ) , while for the lsl surface pattern , the bands are much less sharp ( fig .
[ sliceroughsmooth30p](d ) ) .
more interesting are the 1.0 and 1.5 mm rough wall cases ( fig .
[ sliceroughsmooth30p](b ) and [ sliceroughsmooth30p](c ) ) . in both cases
, there is a core of small particles across the entire width of the bed of particles with large particles surrounding the core .
[ sliceroughsmooth30p](b ) for 1 mm rough walls shows a configuration where the surface pattern is still sls , but the band of large particles is less pure . in fig
[ sliceroughsmooth30p](c ) , particles have segregated radially but not axially , except right at the poles leading to no visible surface bands . and concentration profiles in fig . [ proroughsmooth ] .
the tumbler walls are ( a ) smooth , ( b ) rough with 1 mm particles , ( c ) rough with 1.5 mm particles and ( d ) rough with 2 mm particles . ]
the segregation patterns develop over time .
a convenient means to quantify the degree of segregation and its time evolution is an axial segregation index , defined as : @xmath39 where @xmath40 is the position along the rotation axis with the equator at @xmath38 , @xmath41 is the volume of the particle species @xmath10 , @xmath42 is the radius of the sphere , and the summations are over the large and small particles .
the axial segregation index is positive for lsl and negative for sls .
the limit for perfect segregation is always less than one , but depends on the fill fraction . for a 50% full tumbler
, perfect segregation would result in a segregation index of approximately @xmath430.4 based on the tumbler s spherical shape .
for equal volumes of 2 mm and 4 mm particles rotated at 15 rpm for several tumbler fill fractions ( volume percentages indicated in the figure ) and wall roughnesses : ( a ) smooth , ( b ) rough with 1.5 mm particles and ( c ) rough with 2 mm particles . ] figure [ ios0 - 2 mm ] shows the time evolution of the axial segregation index for the three wall roughness cases in fig .
[ roughsmooth30p ] ( with increasing wall roughness from top to bottom ) for a range of fill fractions .
note that simulations were conducted until the segregation index reached its asymtotic value .
only the time evolution up to 400 s is presented here , but in some cases , simulations reached 1000 s. regardless of the wall roughness , the segregation pattern is usually lsl ( @xmath44 ) for larger fill fractions and sls ( @xmath45 ) for smaller fill fractions , consistent with experimental results for smooth walls @xcite . further note that lsl segregation ( @xmath44 ) is achieved more slowly than sls segregation ( @xmath46 ) for a smooth wall ,
also consistent with experimental results @xcite .
however , this is to be expected because the time that particles spend in the flowing layer depends directly on the fill fraction . for low fill fractions , particles pass through the flowing layer more frequently than at high fill fractions for the same elapsed time @xcite .
since only particles in the flowing layer have the opportunity to rearrange themselves ( unlike particles below the flowing layer , which are locked into place in the bed of particles in solid body rotation ) , one could reasonably expect the segregation pattern to appear more quickly for low fill fractions than for high fill fractions .
similar results occur for rough walls .
the transition between lsl and sls occurs at different fill fractions depending on the wall roughness ( fig .
[ charnin ] ) .
the fill fraction for transition decreases from 43% for the smooth wall case to 12% for the 4 mm rough wall case . at high fill fractions
, the segregation gets close to nearly perfect lsl segregation , regardless of the wall roughness . increasing the wall roughness favors lsl segregation . at low fill fractions , the degree of sls segregation depends on the roughness of the wall , with the smooth wall having the greatest segregation .
however the magnitude of the segregation index for sls segregation is not as large as with lsl segregation , except for smooth walls . note that fill fractions lower than those shown in fig .
[ charnin ] for roughnesses of 2 mm or less result in slip of the particle bed with respect to the tumbler , so they are not included in the figure .
the transition from sls to lsl at @xmath47 is steepest for smooth walls .
it is helpful to consider the axial segregation index in fig .
[ charnin ] in the context of the segregation patterns in fig .
[ sliceroughsmooth30p ] . moving upward along a vertical line in fig .
[ charnin ] at a fill fraction of 30% starts with strong sls segregation corresponding to fig .
[ sliceroughsmooth30p](a ) for a smooth wall .
increasing the wall roughness to 1 mm particles corresponds to the development of a core of small particles at the equator , shown in fig .
[ sliceroughsmooth30p](b ) , reducing axial segregation index , though the poles retain nearly pure small particles . for a 1.5 mm rough wall
the segregation index is approximately zero , corresponding to a core of small particles surrounded by large particles extending nearly to the poles , shown in fig .
[ sliceroughsmooth30p](c ) . a rougher wall of 2 mm particles corresponds to strong lsl segregation in which large particles dominate near the poles , and small particles reach the flowing layer surface at the equator , shown in fig .
[ sliceroughsmooth30p](d ) .
this sequence is quite different from what occurs for a 50% fill level .
there is almost no difference in the axial segregation index at a 50% fill level in fig .
[ charnin ] for different wall roughness values .
the lsl segregation patterns shown in fig .
[ bidi50p ] at a 50% fill volume for smooth and 2 mm rough walls bear out the similarity in the axial segregation index .
previous results for the effect of wall roughness and fill level on particle trajectories in monodisperse flows provide some insights .
the trajectories in the flowing layer for both smooth and rough walls are curved @xcite .
this curvature is negligible at the equator ( at @xmath38 cm , which is a line of symmetry ) and increases moving toward the pole . for particles near the surface of the flowing layer ,
the trajectory curvature for smooth walls is greater than that for rough walls .
however , the net poleward drift at the surface with each pass through the flowing layer for rough walls is larger than for smooth walls . since surface particles drift poleward , particles deeper in the flowing layer drift toward the equator to conserve mass .
as the fill level is reduced , the curvature of the particle trajectories increases for all roughnesses , and the poleward drift decreases only for smooth walls but does not change significantly for rough walls .
these monodisperse results can be used to explain the bidisperse segregation patterns .
two effects compete to select the segregation pattern .
first , particles are subject to a depth dependent poleward drift .
since drift is larger at the surface @xcite , large particles that have segregated to the surface axially drift further poleward than small particles below the surface .
large particles remain at the surface due to radial segregation , so they accumulate near the poles .
small particles below the surface in the radially segregated core are transported by the global convection cells : from equator to pole nearer the surface and from pole to equator deeper in the flowing layer @xcite . the second effect , which counteracts the drift , is the high curvature of the particle trajectories at the surface of the flowing layer , which is typical of smooth walls @xcite .
it results in both small and large particles being carried further poleward in the upstream portion of the flowing layer .
small particles tend to fall out of the flowing layer sooner than larger particles due to percolation , thus depositing in the fixed bed when they are closer to the poleward extrema of the trajectory than large particles , which tend to stay near the surface to curve back more toward the equator to deposit at an axial position near where they started .
it is likely that for smooth walls , this trajectory curvature effect dominates , leading to the sls pattern ; for rough walls , where the trajectory curvature is smaller , the drift is very efficient and dominates @xcite leading to the lsl pattern . to confirm this mechanism , we plot pairs of trajectories for the two species of particles ( 2 and 4 mm ) starting from the same initial positions in the flowing zone near the surface ( specifically , when the center of a particle starts within a sphere of radius 1 mm that is 3 mm below the free surface ) , for tumblers with smooth or with 2 mm rough walls in figs .
[ comparetraj30ptop ] and [ comparetraj30pside ] .
these trajectories are obtained by averaging thousands of individual particle trajectories starting from the same initial coordinate during the first few seconds ( between @xmath48 s and @xmath49 s ) of the flow , before radial segregation is achieved . like the monodisperse case @xcite , the trajectories in a tumbler with a smooth wall have a larger curvature than those for the rough case ( fig .
[ comparetraj30ptop ] ) .
however , for the bidisperse case , large particles remain at surface while small particles sink deeper into the flowing layer ( fig .
[ comparetraj30pside ] ) , regardless of whether the walls are smooth or rough . ) in the flowing zone , at various @xmath50 coordinates , in a tumbler made of a ) smooth walls or b ) rough 2 mm walls and filled at 30% . ] ) in the flowing zone , but in various @xmath50 positions , in a tumbler made of a ) smooth walls or b ) rough 2 mm walls and filled at 30% . ] the trajectory curvature results in the large particles tending toward the equator and small particles toward the poles .
this is most evident comparing the trajectories starting at @xmath51 m , in the smooth wall tumbler ( fig .
[ comparetraj30ptop](a ) ) .
this process dominates for smooth walls , leading to the sls segregation pattern .
the process is quite rapid , and thus is always dominant in the first moments of the flow .
this is evident in fig .
[ ios0 - 2 mm ] where the segregation indices in all cases are negative during the few first seconds , even in the cases of asymptotic lsl segregated systems .
it is also evident that fully developed segregation is reached more quickly for sls than for lsl ( also observed experimentally @xcite ) . on the other hand , for rough walls , the curvature of the trajectories is less , but the net drift of both particles is toward the pole . due to the radial segregation ,
large particles stay at the surface , whereas small particles are deeper in the flowing layer where there is a return flow toward the equator .
the drift effect tends toward an lsl segregation pattern and dominates when the trajectory curvature is small .
analogous results occur when varying the fill level . at low fill levels ,
particle trajectories are more curved @xcite , which favors sls segregation . at high fill levels , trajectories are nearly straight @xcite , so the axial drift , as the particles segregate to different depths , dominates leading to the lsl segregation pattern . at steady state for either sls or lsl patterns ,
the particle trajectories appear to stabilize the pattern .
this is shown in fig .
[ trajbidicombi ] where we compare the trajectories of large and small particles by plotting pairs of trajectories for the two particle species at the boundary between the large and small particles . in the case of smooth walls ,
the trajectories for the small and large particles diverge substantially even after only one pass through the flowing layer , resulting in displacement toward the pole for small particles , and displacement toward the equator for large particles , reinforcing the sls pattern .
this is not the case for the lsl segregation pattern for rough walls where displacements are very small , and both toward the pole ( see inset of fig .
[ trajbidicombi](b ) ) .
both species have nearly the same axial position after one pass through the flowing layer .
the axial segregation occurs indirectly as a result of the radial segregation in the flowing layer , which keeps large particles near the surface .
consequently , the return flow toward the equator deep in the flowing layer consists of only small particles .
thus small particles reach the surface only near the equator where the core current emerges .
the consequence is a very sharp boundary at the surface between small and large particles for the sls segregation pattern and a more diffuse boundary for the lsl case ( see fig .
[ trajbidicombi ] ) .
these differences are also evident in the experimental results in fig .
[ exproughsmooth ] and @xcite .
a question that naturally arises is if wall coulomb friction could play a similar role to wall roughness in the evolution of segregation patterns based on the implicit assumption that a rough wall should have a similar effect to a high coefficient of friction for a smooth wall . to consider this
, we use a higher coefficient of friction for wall - particle interactions than for particle - particle interactions in simulations for smooth walls . as shown in fig .
[ coulomb ] , the segregation evolution for a wall coefficient of friction of 1.5 is nearly identical to that for a wall coefficient of friction of 0.7 , which is the particle - particle coefficient of friction . of course , increasing the coefficient of friction too much results in a non - physical situation in which slip will not occur at all .
hence , it is difficult to explain the differences between smooth and rough walls based on a argument that coulomb friction equivalent to wall roughness , at least within the constraints of the simulations .
= 0.7 or 1.5 for fill fractions of 30% , 40% , and 50% . ]
next , we consider the effect of variation of the roughness of the wall .
the evolution of the axial segregation index for the different wall roughnesses and a fill level of 30% is shown in fig .
[ segregvsrough5 mm ] . as observed previously ( figs .
[ ios0 - 2 mm ] and [ charnin ] ) the segregation index increases with wall roughness . when the roughness is much larger than the size of the flowing particles ,
there is little influence on the axial segregation index .
this is likely because the smaller flowing particles fill the gaps between the larger wall particles , which alters particle trajectories @xcite .
however , when the wall particles are smaller than the flowing particles , small changes can have significant impact .
this is even more evident when considering the asymptotic value of the segregation index , shown in fig .
[ indexvsrough ] .
the greatest impact on the segregation index for a mixture of 2 and 4 mm particles occurs when the wall particle size is between 1 and 2 mm .
it is also interesting to note that the perfectly smooth wall , modeled as spherical smooth surface with infinite mass , behaves similarly to a rough wall of 0.25 mm particles , suggesting that modeling a wall as fixed particles can be effective so long as the wall particles are much smaller than the flowing particles . to further consider the relative size effects of the tumbler and particles
, we performed a limited number of simulations with 1 and 2 mm particles in a 14 cm tumbler with both smooth and 1 mm rough walls .
because many more particles are simulated in this situation than with 2 and 4 mm particles , only low fill levels ( 30% ) and shorter runs were feasible . the results in fig .
[ chasize](a ) demonstrate that relative particle and tumbler sizes make a difference for both smooth and rough walls .
the decrease in particle size induces an increase in the axial segregation index .
for example , in the smooth wall case , the 1 and 2 mm particles form an lsl pattern , while 2 and 4 mm particles form an sls pattern , consistent with experimental results @xcite .
this is because particle trajectory curvature increases with particles size , but net axial drift remains almost independent of particles size @xcite , and large curvature trajectories favor sls segregation . with trajectory curvature being smaller for small particles ,
a 1 + 2 mm particle system will adopt the lsl segregation pattern more easily than a 2 + 4 mm system , as is observed experimentally @xcite and shown in fig .
[ chasize](a ) .
consider now a 7 cm tumbler ( 30% full ) in which the entire system ( particle sizes , wall roughness , and tumbler diameter ) is scaled to half the size of the previous simulations , compared to a 30% full 2 + 4 mm system at the same rotation speed .
the differences between the large and small systems are relatively small as shown in fig .
[ chasize](b ) .
results for the same froude number , @xmath52 , for the 7 cm tumbler at 21 rpm and the 14 cm tumbler 15 rpm are also shown in fig .
[ chasize](b ) . for
now , we simply note that conserving the rotation speed while reducing the system size results in similar axial segregation index curves while conserving the froude number results in different degrees of axial segregation for different system sizes . changing
the drum size suggests that the key mechanism is probably the differences in the trajectories for large and small particles .
bidisperse particle segregation in a spherical tumbler provides an ideal test for evaluating the impact of wall boundary conditions because of its sensitivity to the wall roughness and easily visualized results .
it is clear that surface boundary conditions have a strong influence on the flow and subsequent segregation patterns .
the bands form due to a combination of curved particle trajectories and the axial drift in the flowing layer .
while the roughness of the walls determines the curvature and drift , these quantities along with radial size segregation determine the nature of the axial segregation pattern .
for adequate fill levels , smooth walls result in more curved trajectories with little drift and consequently sls patterns ; rough walls result in less curved trajectories with more drift and consequently lsl patterns . at large fill levels
, the axial lsl patterns always occur regardless of wall roughness . at lower fill levels ,
axial sls patterns are more likely to occur .
many questions remain including that of why the curvature and drift are so dependent upon the size of the system and rotation speed .
nevertheless , the non - locality of granular flow ( wall roughness modifies the trajectories and band formation far from the wall ) is evident , as is the case in many other situations for granular pattern formation .
99 m. moakher , t. shinbrot , and f. j. muzzio , powder technol . * 109 * , 58 ( 2000 ) .
p. chen , j. m. ottino , and r. m. lueptow , new j. phys .
* 13 * , 055021 ( 2011 ) .
n. taberlet , w. losert , and p. richard , europhys . lett .
* 68 * , 522 ( 2004 ) .
d. c. rapaport , phys .
e * 65 * , 061306 ( 2002 ) .
n. taberlet , m. newey , p. richard , and w. losert , j. stat .
p07013 ( 2006 ) . f. da cruz , s. emam , m. prochnow , j .-
roux , and f. chevoir , phys .
e * 72 * , 021309 ( 2005 ) .
j. j. mccarthy , and j. m. ottino , powder technol . *
97 * , 91 ( 1998 ) .
t. pschel and v. buchholtz , chaos , solitons and fract .
* 5 * , 1901 ( 1995 ) .
g. juarez , p. chen , and r. m. lueptow , new j. phys .
* 13 * , 053055 ( 2011 ) .
f. bertrand , l. a. leclaire , and g. levecque , chem .
sci . * 60 * , 2517 ( 2005 ) .
u. dortona , n. thomas , z. zaman , and r. m. lueptow , influence of rough and smooth walls on macroscale flows in tumblers , submitted to phys .
e n. taberlet , p. richard , a. valance , w. losert , j. m. pasini , j. t. jenkins , and r. delannay , phys . rev . lett . * 91 * , 264301 ( 2003 ) . s. w. meier , r. m. lueptow , and j. m. ottino , adv . phys .
* 56 * , 757 ( 2007 ) .
s. courrech du pont , p. gondret , b. perrin , and m. rabaud , europhys .
* 61 * , 492 ( 2003 ) .
f. pignatel , c. asselin , l. krieger , i. c. christov , j. m. ottino , and r. m. lueptow , phys .
e * 86 * , 011304 ( 2012 ) .
j. f. gilchrist and j. m. ottino , phys .
e * 68 * , 061303 ( 2003 ) .
p. chen , b. j. lochman , j. m. ottino , and r. m. lueptow , phys .
lett . , * 102 * , 148001 ( 2009 ) .
g. juarez , j. m. ottino , and r. m. lueptow , phys .
e * 78 * , 031306 ( 2008 ) .
y. oyama , bull .
( tokyo ) * 18 * , 600 ( 1939 ) .
m. b. donald and b. roseman , br .
* 7 * , 749 ( 1962 ) .
m. nakagawa , chem .
sci . * 49 * , 2540 ( 1994 ) . o. zik , d. levine , s. g. lipson , s. shtrikman , and j. stavans , phys .
lett . * 73 * , 644 ( 1994 ) .
k. m. hill and j. kakalios , phys .
e * 49 * , r3610 ( 1994 ) . k. m. hill and j. kakalios , phys .
e * 52 * , 4393 ( 1995 ) .
n. jain , d. v. khakhar , r. m. lueptow , and j. m. ottino , phys .
* 86 * , 3771 ( 2001 ) .
s. j. fiedor and j. m. ottino , phys .
lett . * 91 * , 244301 ( 2003 ) .
p. chen , j. m. ottino , and r. m. lueptow , phys .
lett . * 104 * , 188002 ( 2010 ) .
p. chen , j. m. ottino , and r. m. lueptow , phys .
e * 78 * , 021303 ( 2008 ) .
j. schfer , s. dippel , and d. e. wolf , j. phys .
i ( france ) * 6 * , 5 ( 1996 ) .
g. h. ristow , pattern formation in granular materials ( springer - verlag , berlin , 2000 ) .
p. a. cundall and o. d. l. strack , geotechnique
* 29 * , 47 ( 1979 ) .
m. p. allen and d. j. tildesley , computer simulation of liquids ( oxford university press , new york , 2002 ) . t. g. drake and r. l. shreve , j. rheol .
* 30 * , 981 ( 1986 ) . s. f. foerster , m. y. louge , h. chang , and k. allia , phys .
fluids * 6 * , 1108 ( 1994 ) .
z. zaman , u. dortona , p. b. umbanhowar , j. m. ottino , and r. m. lueptow , phys .
e * 88 * , 012208 ( 2013 ) . c. s. campbell , j. fluid mech . *
465 * , 261 ( 2002 ) .
l. e. silbert , g. s. grest , r. brewster , and a. j. levine , phys .
lett . * 99 * , 068002 ( 2007 ) . | size bidisperse granular materials in a spherical tumbler segregate into two different patterns of three bands with either small particles at the equator and large particles at the poles or vice versa , depending upon the fill level in the tumbler .
here we use discrete element method ( dem ) simulations with supporting qualitative experiments to explore the effect of the tumbler wall roughness on the segregation pattern , modeling the tumbler walls as either a closely packed monolayer of fixed particles resulting in a rough wall , or as a geometrically smooth wall .
even though the tumbler wall is in contact with the flowing layer only at its periphery , the impact of wall roughness is profound .
smooth walls tend toward a small - large - small ( sls ) band pattern at the pole - equator - pole at all but the highest fill fractions ; rough walls tend toward a large - small - large ( lsl ) band pattern at all but the lowest fill fractions .
this comes about because smooth walls induce poleward axial drift of small particles and an equator - directed drift for large particles , resulting in an sls band pattern . on the other hand
, rough walls result in both sizes of particles moving poleward at the surface of the flow , but due to radial segregation , small particles percolate lower in the flowing layer where there is a return drift toward the equator while large particles remain at the surface near the pole , resulting in an lsl band pattern .
the tendency toward either of the two band patterns depends on the fill level in the tumbler and the roughness of the tumbler s bounding wall . |
we would like to thank g.j .
parker for making his llg solver available to us . | the phenomenon of stochastic resonance ( sr ) has been mainly studied in one - dimensional systems with additive noise .
we show that in higher dimensional systems and in the presence of multiplicative noise , a non - linear magnetic system with a strongly periodic current can show behavior similar to that of sr but only for frequencies below the ferromagnetic resonance ( fmr ) frequency of the system .
such a phenomena can provide an effective way to suppress low frequency noise in spin valve magnetic sensors .
the application of stochastic resonance @xcite to increase the signal to noise ratio ( snr ) is now well established in many physical systems @xcite and has also been shown to increase the stability of an unstable system in some cases @xcite .
particle production in a quantum field can also be enhanced in the presence of noise @xcite .
the snr is enhanced by adding noise to the system as it is being driven by an external periodic force .
however until now only one - dimensional systems with additive noise and a simple bistable potential , e.g. , @xmath0 , or a potential with one metastable state , e.g. , @xmath1 , have been studied and shown to have this interesting property of increasing the snr with the addition of noise as it is being driven by a time - dependent force .
this counter intuitive response is inherently related to the non - linearities present in these systems .
one might therefore ask if a similar behavior can be found in more complicated systems , in particular magnetic systems . in this letter , we give one example of such systems ; a spin valve of two magnetic layers separated by a normal conductor .
the magnetization is inhomogeneous in the free layer and a current perpendicular to the plane ( cpp ) traverses the structure . in this case
a spin torque between the layers can exist and will play in the following an important role in the sr effect . until recently ,
the ideas of stochastic resonance have not been exploited in magnetic systems .
as far as i know , reference @xcite provides the only application we are aware of ; it applies sr to the measurement of hysteresis loops in magnetic systems described by preisach model . in the following , we study a magnetic system with multiplicative noise and show that a stochastic resonance type behavior exist in this system and can be taken advantage of to _ selectively _ suppress frequencies from the noise spectrum .
the system we treat is very realistic and hence more complex than previously treated systems @xcite .
the use of strong time - dependent currents implies that our system is in a non - equilibrium state and linear response methods are obsolete for our study @xcite .
in this letter we will attempt only a qualitative treatment of this subject aided by numerical integration of the non - linear landau - lifshitz - gilbert equation @xcite .
we study a current perpendicular to the plane ( cpp ) spin valve with two magnetic layers ( fig .
[ geometry ] ) , one with magnetization @xmath2 pinned along the x - axis ( reference layer rl ) while the other layer ( fl ) has a free magnetization @xmath3 .
figure [ zerocurrent ] shows the power spectral densities ( psd ) of the thre components of the magnetization in the absence of current .
the psd s have been normalized the same way in all the figures .
the noise in the z component is smallest due to the demagnetization field .
the current has an ac component in addition to the dc component .
the source of noise in the system is due to thermal fluctuations that activates switching between two states and hence it is intrinsic to the system . to get an energy surface with more than one local minimum
, we bias the spin valve with an external field that is close to being perpendicular to the fixed magnetization of the rl and take into account spin momentum transfer effects @xcite between the two layers of the spin valve .
the y - component of the external field is kept at @xmath4 while the one along @xmath5 is kept around @xmath6 .
the layers have dimensions @xmath7 with the easy axis along the x - axis and anisotropy @xmath8 .
the magnetization in the pinned layer is fixed by a large bias field . hence , in this study the pinned layer is studied micromagnetically on equal footing with the free layer .
the magnetization in this structure shows two configurations ( fig .
[ config ] ) which are non - homogeneous and are a result of the demagnetization field , the current field and any interaction between the layers @xcite .
in addition to stable and unstable states , the system has saddle points which are present due to the spin torque term .
spin valve structures are useful components of giant magneto - resistance ( gmr ) sensor devices and hence the manifestations of stochastic resonance in these systems may be of practical importance . to increase the sensitivity of a gmr sensor , the bias field on the free layer is almost perpendicular to the magnetization of the pinned layer . it is this kind of biasing that gives rise to the @xmath9-type noise studied here since it permits the system to hop between two - states
. the results reported here are valid even if the pinning of the bottom layer is not perfect .
in fact , large bias fields tend to distort the magnetization in the pinned layer but only slightly and will not contribute to the 1/f - type noise observed here .
to analyze this system , we use the landau - lifshitz - gilbert equation @xcite supplemented by the spin torque as first suggested by slonczweski @xcite , @xmath10 where @xmath11 is the gyromagnetic ratio and @xmath12 is the damping constant taken here to be equal to 0.01 .
the effective field @xmath13 includes exchange interactions , anisotropy , the demagnetization field and the field from the current .
the interlayer exchange is assumed small of the order of @xmath14 .
the value of @xmath12 chosen is larger than typical values of damping in the bulk of a permalloy . in a geometry such as ours ,
the damping is dominated by interfacial damping and 0.01 is a reasonable value for it @xcite .
the noise is assumed white gaussian , @xmath15 with @xmath16 and using the fluctuation - dissipation theorem ( fdt ) , @xmath17 is proportional to the temperature @xmath18 and inversely proportional to the volume @xmath19 , @xmath20 @xcite .
however , as we have shown in @xcite , in the presence of a bias voltage the fdt is strictly not applicable unless we restrict our region of interest to frequencies close to the fmr frequency of the system . in the following ,
we assume that fdt is still applicable .
@xmath21 is the direction of the magnetization in the rl .
the current @xmath22 has one static component and another one with frequency @xmath23 .
the form of the spin torque in the presence of ac currents still has the same form as in the slonczewski equation .
zhu et al .
@xcite calculated the ac spin torque in a magnetic tunnel junctions and found it to be very close in form to the static case . among other things
, they found that at low frequencies , the total spin torque can become less than the spin torque of the static component of the current .
hence an ac current can introduce a partial self - cancellation of the spin torque .
therefore we expect the ac spin torque to affect the noise spectrum of a magnetic system in a nontrivial way .
the noise or spectral density functions of the system are calculated numerically since for time dependent systems the fluctuation dissipation relations are invalid @xcite .
the current flows along the z - direction perpendicular to the xy plane of the magnetic layers .
the magnetization of the rl is fixed along the x - direction .
the fmr frequency for this system is around @xmath24 and @xmath25 .
the constant @xmath26 . in magnetic recording , e.g .
, such devices operate at frequencies below the fmr frequency and hence this system is considered noisy to be useful as a sensor . one possible way to address this problem is to seek a way to suppress the noise for frequencies less than @xmath27 and may be shift the noise to higher frequencies which is outside the operational range of the device .
in this sense we are selectively suppressing the noise below the fmr frequency only .
previous applications of stochastic resonance were interested in suppressing the noise around the frequency of the driving force which is usually the signal to be measured . from fig.[xac0 ] , we observe that it is the x - component of the magnetization of the fl that is the noisiest and hence this would interfere with the gmr signal .
the out - of plane z - component of the magnetization is very quiet due to the demagnetization field .
fig .
[ xtac0 ] shows that the source of the noise is from the switching of the magnetization between two configurations .
one with an average x - component of about @xmath28 and the other less stable one with an average x - component around @xmath29 . the spectral noise in the y and z components show more than the usual fmr peak since the system is not in equilibrium .
the higher order harmonics are due to non - homogeneity of the magnetization , i.e. , spin waves . in fig .
[ xt5nac0 ] , we change the sign of the current to show the effect of the spin torque on the magnetization and noise . in this case
, the magnetization spends almost equal time in both states .
this suggests the use of the spin torque itself as a regulator of the transition rate between two bistable states . in this letter
, we will not pursue this idea further and instead we want to explore the idea of adding a strongly periodic current in addition to the dc bias current in order to suppress the noise for frequencies below @xmath27 .
figure [ xac5 ] shows the result of applying an ac current with amplitude equal to the bias current and a frequency of about @xmath30 for the same system of fig . [ xac0 ] .
the noise spectrum for frequencies below @xmath31 is greatly suppressed by this additional current source .
however now the peak around the fmr frequency is much more pronounced and so does its width .
the second narrow peak is that of the ac current . a closer look at the real time behavior of the x - component of the magnetization , fig .
[ xac5 t ] , implies that the original energy barrier is no longer present .
the strong periodic current which in turn induces a periodic spin torque is driving the system and making it less sensitive to random thermal fluctuations which are the source of the low frequency . to be an effective suppressant , the frequency of the torque has to be outside the bandwidth of the sensor device .
finally , fig .
[ xac0ht1 ] shows that an increase in the frequency of the ac current degrades the effectiveness of the ac current component to suppress the low frequency noise .
we found for this example , that for current amplitudes at @xmath32 , the maximum frequency the current should have is about twice the fmr frequency of the system .
this latter criterion seems to depend strongly on the energy surface and hence there is no universal behavior as in the bistable one dimensional well where the frequency of the driving force is the inverse of twice the inter - well transition time . in summary
, we have shown that the ideas of sr are still attractive even for complex systems which can not be described by a one dimensional potential . in higher dimensions and in the presence of a spin torque ,
saddle point states become available to the system and may invalidate the ideas of stochastic resonance . in this letter
, we showed that a strongly periodic current can suppress thermally induced jumps between a stable and a metastable state and hence suppresses the low frequency noise and enhance the fmr peak in the spin valve .
_ this work is based on work done in the fall of 2004 at seagate research and will not be published elsewhere .
it is merely a compilation of numerical observations .
the author hopes to convince others that there is an interesting interplay of noise and time - dependent spin torques in cpp spin valves that is worth exploring in a more systematic way than attempted here . _ |
in last years , a lot of experimental and theoretical effort was made to study two - photon exchange ( tpe ) in the elastic electron - proton scattering .
this activity was motivated by the discovery of the problem in the proton form factor measurements : values of form factor ratio @xmath2 obtained by rosenbluth separation and polarization transfer methods were is strong disagreement .
it is now widely accepted that the discrepancy is caused by tpe , but still no direct experimental observations of tpe exist .
previously , several attempts were made to extract values of tpe amplitudes from available experimental data in more or less model - independent way @xcite . however , lack of precise data and/or theoretical understanding of tpe prevented from obtaining sufficiently accurate estimates for the amplitudes . in particular , it was difficult to determine the dependence of tpe amplitudes on the kinematical parameter @xmath3 , since the @xmath3 dependence of polarization observables was not known experimentally .
recently , a search for tpe effects in polarization observables was reported @xcite . in this experiment ,
ratio of transverse and longitudinal proton polarization components ( polarization ratio ) was measured with significantly improved precision for @xmath4 and wide range of the parameter @xmath3 . in the present paper we use latest experimental data to determine tpe amplitudes at @xmath4 following the ideas of ref .
@xcite with some improvements ( described below ) .
we will try to obtain as much information on tpe amplitudes as possible , while avoiding unnecessary assumptions . in our analysis
we only assume that tpe is the sole reason for the discrepancy between cross - section and polarization data , and rely on the following experimental and theoretical facts : 1 . the reduced cross - section exhibits no or small nonlinearity in @xmath3 ( @xcite , also verified in the present work ) .
2 . the polarization ratio does not vary significantly with @xmath3 @xcite .
3 . tpe amplitudes must vanish at @xmath5 ( because they can be represented by convergent dispersion integral ; @xmath5 implies @xmath6 , where @xmath7 is c.m .
energy squared ) .
the latter point was missing in ref .
there were no recent cross - section measurements in the @xmath8 region of our interest , so we have to use older data .
these data were analyzed before , but for our paper to be self - contained we repeat such an analysis here .
we have selected data in the range @xmath9 @xcite .
the corresponding reduced cross - sections were first multiplied by @xmath10 , to eliminate most of the @xmath8 dependence .
then , since in born approximation the reduced cross - section is _
r = g_m^2 + g_e^2 where @xmath11 and @xmath12 are electric and magnetic form factors , @xmath13 and @xmath14 is proton mass , the resulting values were fitted with the function a + b + c ( q^2 - 2.5 ^2 ) the last term takes into account @xmath8 dependence of @xmath15 term in @xmath16 .
we obtain rather acceptable fit with @xmath17 for 28 d.o.f . , indicating that the linearity of @xmath16 in @xmath3 is indeed supported by the data .
we will mainly need the quantity r_lt^2 = b / a which would be equal to @xmath18 at @xmath0 in born approximation .
we obtain @xmath19 , in agreement with the results ref . @xcite ( 0.1015 ) . in further calculations we use the first value .
we will use mostly the same notation as in ref .
we denote particle momenta according to e(k)+p(p)e(k)+p(p ) , and define @xmath20 , @xmath21 , @xmath22 , @xmath23 . in presence of tpe
, elastic electron - proton scattering amplitude has the form m = & |u_u |u ( f_1 ^- f_2 [ ^,^ ] + .
k_^ ) u where @xmath24 is fine structure constant , @xmath25 , @xmath26 ( @xmath27 , @xmath28 ) are initial and final electron ( proton ) spinors , and @xmath29 are scalar invariant amplitudes .
it is convenient to introduce linear combinations @xcite rclcl _ e & = & f_1 - f_2 + f_3/4m^2 & = & g_e + _ e + _ m & = & f_1 + f_2 + f_3/4m^2 & = & g_m + _ m + _ 3 & = & f_3/4m^2 & = & _ 3 where @xmath30 and prefix @xmath31 indicates tpe contribution .
the tpe amplitudes @xmath32 are complex , but only their real parts contribute to the observables discussed here . everywhere
below , speaking of the amplitudes , we will mean their real parts . neglecting terms of order @xmath33 , the reduced cross - section and polarization ratio
can be written as _
r & = & g_m^2 \ { + r_0 ^ 2 + 2 + 2r_0 ^ 2 } [ sigma_r ] + r & = & r_0 \ { 1 + - - } [ r ] where @xmath34 .
note that our definition of @xmath35 does not include a factor of @xmath36 , thus @xmath35 ( and @xmath37 ) is rather small quantity ( @xmath38 for @xmath39 ) .
utilizing this fact we will neglect last term in eq.([sigma_r ] ) ( the validity of this approximation will be checked afterwards ) . then , as it was argued in ref .
@xcite , the observed cross - section linearity in @xmath3 _ forces _ us to parameterize tpe amplitude @xmath1 as a linear function of @xmath3 . to vanish at @xmath5
, it must have the form [ cg_m ] _ m / g_m = a(1- ) then we have _
r = g_m^2 \ { + r_0 ^ 2 + 2 a ( 1 - ) } and the cross - section slope is r_lt^2 = from which we obtain [ a ] a = together with eq.([cg_m ] ) , this fully determines the amplitude @xmath1 . as a first approximation
, we replace @xmath40 by experimental value of polarization ratio , @xmath41 @xcite , and obtain numerically [ anum ] a = -0.0250 0.0035 thus extracted amplitude @xmath42 is shown in fig . [ fig1 ] with @xmath43 error band . the theoretical prediction @xcite is also shown and agrees rather well with our result .
( grey , @xmath43 band ) and its theoretical estimates , calculated with : elastic intermediate state only @xcite ( dashed ) , elastic + @xmath44 resonance @xcite ( solid).,scaledwidth=48.0% ] now we will take a closer look on the polarization ratio @xmath35 , which allows us to get some information about the amplitude @xmath45 .
first , we note that the last term in eq.([r ] ) should be very small , because the factor @xmath46 is not greater than 0.18 for @xmath47 .
thus we are left with [ r2 ] r = r_0 \ { 1 + - } since both @xmath1 and @xmath45 vanish at @xmath5 , we obviously have r_0 = .
r |_=1 , _ e - r_0 _ m = r - r_0 the experiment says that , at @xmath0 , there is no significant variation of @xmath35 with @xmath3 @xcite .
this implies [ cg_e ] r r_0 _ e r_0 _ m ( which also justifies using @xmath35 instead of @xmath37 in eq.([a ] ) ) .
now we can cross - check that the amplitude @xmath45 has small impact on the cross - section . using eqs.([cg_m],[a],[cg_e ] ) , we calculate the corresponding correction ( the last term of eq.([sigma_r ] ) ) and subtract it from the data
. then we repeat cross - section fitting and extraction of @xmath1 , as described above .
we obtain @xmath48 , i.e. practically no change with respect to ( [ anum ] ) .
eqs.([cg_e ] ) are , of course , a rough estimate . in particular , they do not allow to estimate the uncertainty of @xmath45 .
an accurate @xmath49 extraction with estimation of uncertainties requires determination of the small quantity @xmath50 , for which the present data hardly suffice .
nevertheless , note that theoretical calculations also show relative smallness of @xmath51 , which arises from significant cancellation between proton and @xmath44 resonance contributions ( fig .
[ fig2 ] ) .
, contribution of proton ( dashed ) , @xmath44 resonance ( dash - dotted ) , and total ( solid ) .
, scaledwidth=48.0% ] in the experiment @xcite , one more quantity was measured : longitudinal polarization of the final proton , @xmath52 . in principle , this data could help to determine the remaining amplitude @xmath53 .
the tpe correction to @xmath52 is given by p_l = - 2p_l \ { + } where @xmath54 is tpe correction to polarization ratio according to eq.([r2 ] ) .
as @xmath55 , the first term is negligible , and the deviation of @xmath52 from its born value is governed by the amplitude @xmath53 .
however , two available data points are too few to make any statements about @xmath3 dependence of @xmath53
. we can only compute @xmath53 at the @xmath3 values of experimental data .
the results are shown in fig .
[ fig3 ] . obviously , their precision is insufficient to make a meaningful comparison with theory .
the obtained values are compatible with zero , and do not contradict theoretical estimates as well . .
statistical and systematic errors are added in quadrature , inner bars pure statistical errors .
theoretical calculations with : elastic intermediate state only @xcite ( dashed ) , elastic + @xmath44 resonance @xcite ( solid).,scaledwidth=48.0% ]
in summary , we have tried to extract tpe amplitudes for the elastic electron - proton scattering at @xmath0 solely from the experimental data on cross - sections and polarization observables .
having defined three independent amplitudes @xmath1 , @xmath45 and @xmath53 , we found that the effect of these amplitudes on the observables is `` decoupled '' : the cross - section is mainly influenced by @xmath1 , the polarization ratio by @xmath45 and @xmath1 , longitudinal polarization component by @xmath53 .
the amplitude @xmath1 can be extracted with approximately 10% accuracy and is in good agreement with theoretical calculations .
the weakness of polarization ratio variation with @xmath3 implies approximate equality @xmath56 . as to the amplitude @xmath53 , present experimental data are consistent with @xmath57 . to allow for more accurate extraction of @xmath45 and @xmath53 , further polarization measurements at different @xmath3
are clearly needed .
it is also worth noting that momentum transfer value @xmath4 was not good choice for the experiment @xcite : just in this region tpe correction to polarization ratio is especially small because elastic and @xmath44 resonance contributions almost cancel each other .
recently , a preprint @xcite appeared , in which the same problem was considered .
however , authors have used certain parameterization of polarization component @xmath52 , which , to our opinion , is not well motivated by experimental data . their results strongly differ from ours as well as from theoretical calculations .
20 j. arrington , phys.rev .
c * 71 * , 015202 ( 2005 ) .
d. borisyuk , a. kobushkin , phys .
c * 76 * , 022201(r ) ( 2007 ) .
m. meziane _ et al .
_ , arxiv:1012.0339 [ nucl - ex ] .
v. tvaskis _ et al .
_ , phys .
c * 73 * , 025206 ( 2006 ) .
j. arrington , w. melnitchouk , j.a .
tjon , phys .
035205 ( 2007 ) . j. litt _
et al . _ ,
b * 31 * , 47 ( 1970 ) ; r.c .
walker _ et al .
_ , phys .
d * 49 * , 5671 ( 1994 ) ; l. andivahis _ et al .
_ , phys .
d * 50 * , 5491 ( 1994 ) ; m.e .
christy _ et al .
_ , phys .
c * 70 * , 015206 ( 2004 ) ; i.a .
qattan , arxiv : nucl - ex/0610006 .
d. borisyuk , a. kobushkin , phys .
c * 78 * , 025208 ( 2008 ) .
d. borisyuk , a. kobushkin , in preparation .
j. guttmann , n. kivel , m. meziane , m. vanderhaeghen , arxiv:1012.0564 [ hep - ph ] . | we extract two - photon exchange amplitudes for the elastic electron - proton scattering at @xmath0 from the unpolarized cross - section and recent polarization transfer measurements .
there are three independent amplitudes , but only one of them , @xmath1 , can be determined with a reasonable accuracy ( about 10% ) .
the result is in good agreement with theoretical predictions .
rough estimates for two other amplitudes are obtained . |
three characteristics are computed for each institution @xmath0 ( fig . [ fig2 ] ) : the institution size ( @xmath1 ) , representing the total number of distinct authors that published at least one paper at institution @xmath0 ; the number of papers ( @xmath2 ) published under affiliation @xmath0 ; the cumulative number of citations @xmath3 collected by all papers @xmath2 .
we find that @xmath4 follows a fat tailed distribution , indicating significant population heterogeneity among different institutions ( fig .
[ fig2]a ) . while most institutions are small , a few have a large number of scientists , often corresponding to large institutes or universities .
we observe similar disparity in @xmath5 ( fig .
[ fig2]b ) : few institutions acquire a large number of citations , while most research labs or universities receive few citations .
figures [ fig2]c - d show the correlation between the institution size @xmath6 and both the average publications impact @xmath7 and the average productivity @xmath8 of institutions .
the average productivity and impact of an institution are different but complementary measures of scientific performance .
we find the institution size has little influence on productivity ( @xmath9 ) ( fig .
[ fig2]d ) , yet it positively correlates with the impact of publications ( @xmath10 ) , indicating that large institutions offer a more innovative / higher impact environment than smaller ones as captured by citations per paper ( fig .
[ fig2]c ) .
also , as larger institutions have more internal collaborations , the number of co - authors in publications from large institutions might be larger and , as a consequence , attracts more citations @xcite .
many institutions are small with few citations , hence they account for very small portion of the data . for the rest of the paper , we will focus on the thousand most cited institutions , accounting for more than @xmath11 of papers .
they correspond to institutions with at least 698 citations within the aps data over the 120-year period ( shaded area in fig .
[ fig2 ] ) .
mobility is often important in furthering a professional career @xcite . in science ,
the best lab for the type of research you are doing is usually not where you are @xcite .
nowadays changing countries is a rite of passage for many young researchers who follow the resources and facilities @xcite .
as the patterns and characteristics of these migrations are blurry , we need to systematically study the mobility of scientists .
thanks to the large disambiguated data spanning the last 120 years that we have compiled , a systematic study of scientific mobility is now possible . the strong correlations between the three quantities ( @xmath12 ) indicate any of the three could characterise an institution , serving as a proxy of its ranking against others . here , we choose @xmath13 ( the total number of citations ) as our parameter to approximate the ranking by reputation .
other parameters such as the h - index of an institution or the number of papers @xmath14 could also be used @xcite .
but the results should be insensitive to this choice owing to good correlations between these quantities ( @xmath15 and @xmath16 respectively ) .
the top - ranked institutions all correspond to well - known universities or research labs with long tradition of excellence in physics ( fig .
[ fig3 ] ) , corroborating our hypothesis that @xmath13 is a reasonable proxy for ranking .
we can also observe the similarity and stability of other rankings when comparing with other metrics .
is a reasonable proxy for ranking , scaledwidth=45.0% ] we focus on authors with similar career longevity , restricting our corpus to those who began their career between 1950 and 1980 and published for at least 20 years without any interruption exceeding 5 years . following these criteria , we arrived at a subset of 2,725 scientists to study the mobility patterns and their impact on their careers .
a total of 5,915 career movements are recorded for this corpus . in figure
[ fig4]a we select three individuals as exemplary career histories .
each line represents one individual , with circles denoting his / her publications , allowing us to observe his / her location .
the size of the circle is proportional to citations the paper acquires in five years , approximating the impact of the work . by studying the whole corpus
, we compute @xmath17 , the probability for a scientist to have visited @xmath18 different institutions along his career ( fig .
[ fig4]c ) , finding that career movements are common but infrequent : only @xmath19 of them never moved at all ( @xmath20 ) .
for the ones that move , they mostly move once or twice , @xmath17 decaying quickly as @xmath18 increases .
we also compute @xmath21 , the probability to observe a movement at time @xmath22 , where @xmath23 corresponds to the date of the scientist s first publication .
we find that most movements occurred in the early stage of the career ( fig .
[ fig4]b ) , supporting the hypothesis that changing affiliations is a rite of passage for young researchers @xcite .
this likely corresponds to the postdoc period where graduates broaden their horizons through mobility .
this may also reflect the increasing cost of relocation and family constraints as family developed @xcite .
a third characteristic is the geographical distance of movements , @xmath24 .
existing literature hints for somewhat competing hypothesis in the role geography plays in career movements .
indeed , research on human mobility suggests that regular human movements mostly cover short distances with occasional longer trips , characterized by a power law distance distribution @xcite ; in contrast , country - level surveys find increasing cross - country movements mostly due to cultural exposure and life quality concerns , indicating potential dominance in long distance moves in career choices comparing with typical human travels @xcite .
we measure the distance distribution over all moves observed in our dataset , finding that our result is supported by a combination of both hypothesis .
we find the probability to move to further locations decays as a power law @xcite , whereas the null model predicts this probability to be flat ( fig .
[ fig4]d ) .
this observation is consistent with studies on human mobility , that short distance moves dominate career choices .
yet , when comparing the power law exponents , we find the exponent characterizing career moves ( @xmath25 ) is much smaller than those observed in human travel ( @xmath26 ) , corresponding to higher likelihood of observing long range movements .
this observation might be explained by the influence that scientific collaborations can have on career movements as similar low exponents are observed for collaboration network between cities @xcite . taken together
, the preceding results indicate that career moves mostly happen during the early stage of a career and are more likely to cover short distances .
the observed location in both time and space raises the question of how individual moves as a function of institutional rankings . to this end , denoting with @xmath27 the number of transitions from the institution of rank @xmath0 to the one of rank @xmath28 , we measure @xmath29 , the probability to have a transition from rank @xmath0 to rank @xmath28 as @xmath30 interestingly , we find that most movements involve elite institutions ( rank is small ) , and transitions between bottom institutions are rare ( fig .
[ fig5]a ) .
this is due to the fact that elite institutions are characterised by larger populations , hence translating into more events . to account for the population based heterogeneity , we compare the observed @xmath29 with the probability @xmath31 expected in a random model where we randomly shuffle the transitions from institution @xmath0 to @xmath28 while preserving the total number of transitions from and to each institution . formally , in this null model , we have @xmath32 and we compare @xmath29 with the null model by computing the matrix @xmath33 @xmath34 is the ratio between the probability @xmath29 to have a transition from rank @xmath0 to @xmath28 divided by the probability @xmath31 when the movements are shuffled , measuring the likelihood for a move to take place by accounting for the size of the institutions .
hence , @xmath35 indicates the amount of observed movements is about what one would expect if movements were random .
similarly , @xmath36 indicates that we observe more transitions from @xmath0 to @xmath28 than we expected , whereas @xmath37 corresponds to transitions that are underrepresented .
we find that career moves are characterized by a high degree of stratification in institutional rankings ( fig .
[ fig5]b ) .
indeed , we observe two distinct clubs ( red spots in fig . [ fig5]b ) , indicating that the overrepresented movements are the ones within elite institutions ( lower - left corner ) or within lower - rank institutions ( upper - right corner ) , and scientists belonging to one of the two groups tend to move to institutions within the same group . on the other hand , both upper - left and lower - right corners are colored blue , indicating cross group movements ( transitions from elite to lower - rank institutions and vice - versa ) are significantly underrepresented .
also , scientists from medium - ranked institutions move to the next institution with a probability that is indistinguishable from the random case . in other words ,
their movements indicate no bias towards middle , elite or lower - ranked institutions .
the high intensity of stratification in career movements raises an interesting question : how does individual performance in science relate to their moves across different institutional rankings ? to answer this question , we need to quantify the performance change for each individual before and after the move .
imagine that a scientist moves from @xmath0 to @xmath28 , and published @xmath38 papers at location @xmath0 and @xmath18 papers at @xmath28 .
the impact of a paper @xmath39 can be approximated by @xmath40 , the number of citations cumulated within 5 years after its publication @xcite .
let @xmath41 and @xmath42 be the lists of number of citations for papers published before ( @xmath43 ) and after ( @xmath44 ) the transition from @xmath0 to @xmath28 ( @xmath27 ) . to quantify the change in performance , we introduce @xmath45 where @xmath46 and @xmath47 are the average of @xmath44 and @xmath43 , respectively , and @xmath48 corresponds to the standard deviation of the concatenation of both @xmath44 and @xmath43 while preserving the moment when the movement took place ( see sm for more information about @xmath48 ) .
therefore , @xmath49 captures the statistical difference in the average citations between papers published before and after the movement normalized by the random expectation when the same author s publications were shuffled .
a positive @xmath49 indicates papers following the move on average result in higher citation impact , hence representing an improvement in scientific performance .
a negative value corresponds to a decline in performance . to quantify the influence of movements on individual performance , we divide all movements into two categories based on the performance change : movements associated with positive and negative @xmath49 , and measure @xmath50 and @xmath51 .
we find the observed stratification in career moves is robust against individual performance ( fig .
[ fig5]c - d ) .
that is , the two clubs emerge for both categories in a similar fashion as in figure [ fig5]b , indicating the pattern of moving within elite or lower - rank institutions is nearly universal for people whose performance is improved or decreased following the move .
comparing figure [ fig5]c and figure [ fig5]d , we find the red spot in lower - left corner is more concentrated in figure [ fig5]d than in figure [ fig5]c , hinting that being more mobile in the space of rankings may lead to variable performance . to test this hypothesis ,
for each transition @xmath27 we calculate the rank difference between the origin and destination ( @xmath52 ) .
a positive value of @xmath53 indicates @xmath54 , hence a movement to a lower - rank institution , whereas @xmath55 corresponds to transitions into institutions with a higher rank . in figure [ fig6 ]
we measure the relation between @xmath49 and @xmath56 .
when scientists move to institutions with a lower rank ( @xmath57 ) , we find that their average change in performance is negative , corresponding to a decline in the impact of their work .
yet , what is particularly interesting lies in the @xmath58 regime . indeed ,
when people move from lower rank location to elite institutions , we observe no performance change on average .
this is rather unexpected , as transitioning from lower - rank institutions to elite institutions is thought to provide better access to ideas and lab resources , which in turn should fuel scientific productivity .
a possible explanation may be that scientist who have the opportunity to make big jumps in the ranking space may have already had an excellent performance in their previous institutions . a move
therefore will not affect their impact . ) and the ranking difference ( @xmath56 ) associated to a transition shows that , when people move to institutions with a lower rank ( @xmath57 ) , their average change in performance is negative , corresponding to a decline in the impact of their work .
yet , what is particularly interesting lies in the @xmath58 regime .
indeed , when people move from lower rank location to elite institutions , we observe no performance change on average . ]
in summary , we extracted affiliation information from the publications of each scientist , allowing us to reconstruct their career moves between different institutions as well as the body of work published at each location .
we find career movements are common yet infrequent .
most people move only once or twice , and usually in the early stage of their career .
career movements are affected by geography . the distance covered by the move
can be approximated with a power law distribution , indicating that most movements are local and moving to faraway locations is less probable .
we also observe a high degree of stratification in career movements .
people from elite institutions are more likely to move to other elite institutions , whereas people from lower rank institutions are more likely to move to places with similar ranks .
we further confirm that the observed stratification is robust against the change in individual performance before and after the move .
when cross - group movement occurs , we find that while going from elite to lower - rank institutions on average results in a modest decrease in scientific impact , transitioning into elite institutions , does not result in gain in impact .
the nature of our dataset restricted our study on a sample of scientists . as a result of this selection process
, our results are biased towards physicists from 1960s to 1980s with high career longevity . yet
, these limitations also suggest new avenues for further investigations .
indeed , as datasets become more comprehensive and of higher resolution , newly available data sources like web of science or google scholar can provide new and deeper insights towards generalization of the results across different disciplines , temporal trends , and more .
further investigations regarding the influence of career longevity on scientific mobility should also be considered as it could reveal as well results of importance .
taken together our results offer the first systematic empirical evidence on how career moves affect scientific performance and impact .
* dataset . *
the data provided by the american physical society ( aps ) contains over @xmath59 publications , each identified with a unique number , corresponding to all papers published in 9 different journals , namely physical review a , b , c , d , e , i , l , st and review of modern physics , spanning a period of 117 years from 1893 to 2010 . for each paper
the dataset includes title , date of publication ( day , month , year ) , author names and affiliations of each of the authors .
a separate dataset also provides list of citations within the aps data only , using unique paper identifiers .
about 5% of publications with ambiguous author - affiliation links or massively authored were removed from this dataset ( see sm for more details ) . *
author name disambiguation .
* to derive individual information , one has to reconnect papers belonging to a single scientist . since no unique author identifier is present in the data , author names must be disambiguated .
the dataset contains about @xmath60 millions of author - paper pairs . to overcome the ambiguities present in the data , we design a procedure that uses information about the author but also metadata about the paper such as coauthors and citations . by computing similarities between authors ,
our procedure can successfully detect single authors as well as homonymies ( see sm for more details about the disambiguation method ) .
a total of @xmath61 distinct scientists are detected by our method .
* affiliation disambiguation . *
a major disadvantage when dealing with publication data is the inconsistencies and errors associated with affiliation names on papers .
a total of @xmath62 different affiliation names are identified in the dataset .
the disambiguation procedure for affiliations uses geocoded information as well as a similarity measure between affiliation names in order to disambiguate institutions .
the disambiguated set of authors also plays a crucial role in the procedure ( see sm for more details about the disambiguation method ) .
a total of @xmath63 distinct institutions are identified by our algorithm . * resolving individual career trajectory * based on the information present in the publications of a scientist , we can reconstruct his / her career trajectory . in order to detect career movements , i.e. changes in a scientist s institution
, one has to remove artificial movements induced by short - term stays and by errors and typos in the affiliation names on the papers .
to do so , only institutions reported in at least two consecutive papers are considered in a career trajectory .
* ranking the institutions * three variables are considered to rank an institution : ( i ) the total number of papers , @xmath2 , published with institution @xmath0 , ( ii ) the cumulated number of citations , @xmath3 , corresponding to institution @xmath0 , ( iii ) the h - index , @xmath64 , of institution @xmath0 . the variable @xmath3 is defined as @xmath65 where @xmath40 is the number of citations within the aps data of paper @xmath39 cumulated within @xmath66 years after its publications .
an institution has an h - index @xmath67 if @xmath67 of its @xmath14 papers have at least @xmath67 citations each , and the other @xmath68 papers have no more than @xmath67 citations each .
@xmath67 for papers indicates the cumulative number of citations obtained within @xmath66 years after the publication .
* binning the institutions . * about @xmath69 transitions between @xmath70 institutions are detected for our subset of scientists . in order to have a statistically significant number of transitions to derive the values of @xmath29 and @xmath34 ( fig .
[ fig5 ] ) , institutions are binned logarithmically according to their rank ( @xmath71 ) into five groups .
we thank nicolas boumal and colleagues from the center for complex network research ( ccnr ) for the valuable discussions and comments .
dw , cs , and alb are supported by lockheed martin corporation ( sra 11.18.11 ) , the network science collaborative technology alliance is sponsored by the u.s .
army research laboratory under agreement w911nf-09 - 2 - 0053 , defense advanced research projects agency under agreement 11645021 , and the future and emerging technologies project 317 532 `` multiplex '' financed by the european commission .
pd is supported by the national fund for scientific research ( fnrs ) and by the research department of the communaut franaise de belgique ( _ large graph _ concerted research action ) .
rs acknowledges support from the james s. mcdonnell foundation .
pd designed research , analysed the data and wrote the paper .
dw , rs , cs , vb and alb , designed research and wrote the paper .
* competing financial interests : * the authors declare no competing financial interests . | changing institutions is an integral part of an academic life . yet
little is known about the mobility patterns of scientists at an institutional level and how these career choices affect scientific outcomes . here , we examine over 420,000 papers , to track the affiliation information of individual scientists , allowing us to reconstruct their career trajectories over decades .
we find that career movements are not only temporally and spatially localized , but also characterized by a high degree of stratification in institutional ranking .
when cross - group movement occurs , we find that while going from elite to lower - rank institutions on average associates with modest decrease in scientific performance , transitioning into elite institutions does not result in subsequent performance gain .
these results offer empirical evidence on institutional level career choices and movements and have potential implications for science policy . despite their importance for education , scientific productivity , reward and hiring procedures , our quantitative understandings of how individuals make career moves and relocate to new institutions , and how such moves shape and affect performance , remains limited .
indeed , previous research on migration patterns of scientists @xcite tended to focus on large - scale surveys on country - level movements , revealing long - term cultural and economical priorities @xcite . at a much finer scale , research on human dynamics and mobility has emerged as an active line of enquiry @xcite , owing to new and increasingly available massive datasets providing time resolved individual trajectories @xcite . while these studies cover a much shorter time scale than a typical career , they uncover a set of regularities and reproducible patterns behind human movements @xcite .
less is known about patterns behind career moves at an institutional level and how these moves affect individual performance . here
we take advantage of the fact that scientists publish somewhat regularly along their career @xcite , and for each publication , the institution in which the work was performed is listed as an affiliation in the paper , documenting career trajectories at a fine scale and in great detail .
these digital traces , offering data on not only individual scientific output at each institution but also career moves from one institution to another , can provide insights for science policy , helping us understand how institutions shape knowledge , the typical moves of individual career development and help us evaluate scientific outcomes associated with professional mobility .
we use the physical review dataset to extract mobility information , publication record , and citations for individual scientists .
the data consists of 237,038 physicists and 425,369 scientific papers , out of which 4,052 different institutions are extracted after the disambiguation process for authors and affiliations ( see sm for disambiguation process ) . to reconstruct the career trajectory of a scientist , we use the affiliation given in each of his / her publications ( fig [ fig1 ] ) . for authors with multiple affiliations listed on a paper
we consider the first affiliation as primary institution .
we compute the impact of each paper by counting its cumulative citations collected 5 years after its publication @xcite . |
many of the physical challenges of solar and stellar physics arise from magnetic processes taking place in the outer regions of the atmosphere .
unfortunately , our present empirical knowledge on the magnetism of the chromosphere , transition region and corona is still very crude . in our opinion , there are two key avenues that should be pursued to improve the situation .
firstly , new instrumentation is needed , including a space - borne uv polarimeter and a large ground - based solar telescope optimized for ( spectro-)polarimetric observations ( e.g. , the advanced technology solar telescope ) . secondly , but equally importantly ,
we need appropriate diagnostic tools that will permit us to infer the magnetic field vector from the observed polarization in suitably chosen spectral ( chromospheric and coronal ) lines .
the present paper represents a contribution to this second line of research .
in particular , our aim here is to demonstrate that the determination of the magnetic field vector via the analysis of the observed polarization in the 10830 multiplet must be carried out considering the wavelength positions and the strengths of the zeeman components in the incomplete paschen - back effect regime .
previous research done with this multiplet has ignored this effect ( with the only exception of @xcite @xcite ) , which resulted in considerable confusion surrounding the formation physics of these lines .
the 10830 multiplet originates between a lower term ( @xmath1 ) and an upper term ( @xmath2 ) .
therefore , it comprises three spectral lines ( see , e.g. , @xcite @xcite ) : a ` blue ' component at 10829.09 with @xmath3 and @xmath4 ( hereafter transition 1 , or tr1 for abbreviation ) , and two ` red ' components at 10830.25 with @xmath5 ( hereafter , tr2 ) and at 10830.34 with @xmath6 ( hereafter , tr3 ) which appear blended at solar atmospheric temperatures . as shown in fig [ energy ] , the energy difference between the levels with @xmath6 and @xmath5 is such that their respective magnetic substates cross each other for magnetic strengths between 400 g and 1500 g , approximately
. therefore , _ at least _ in this range of strengths , the zeeman splitting of such @xmath7-levels is comparable to their energy separation . as a result , the perturbation theory of the familiar zeeman effect is no longer valid .
the energy positions of the magnetic substates as a function of the field strength have to be calculated in the incomplete paschen - back effect regime , which requires to diagonalize the _ total _ hamiltonian ( see , e.g. , @xcite @xcite ) .
interestingly , as we shall see below , the proper calculation of the zeeman components results in significant differences in the stokes profiles that emerge from a magnetized stellar atmosphere . furthermore
, such differences do not vanish in the limit of weak fields .
the outline of this paper is as follows .
section 2 presents a brief review of the paschen - back effect theory , illustrating the range of magnetic strengths for which the @xmath7-levels of the @xmath2 term cross .
section 3 shows a detailed comparison of the results that we have obtained assuming linear zeeman splitting ( lzs ) or incomplete paschen - back splittings ( ipbs ) , including an analytical demonstration that explains our numerical radiative transfer results in a milne - eddington model .
finally , section 4 gives our concluding remarks .
according to the theory of the zeeman effect ( see , e.g. , @xcite @xcite ) the corrections to the degenerate energy @xmath8 of any atomic level of total angular momentum quantum number @xmath7 are obtained by diagonalizing the matrix @xmath9 , where @xmath10 is the magnetic hamiltonian , given by : @xmath11 with @xmath12 erg g@xmath13 being the bohr magneton . by choosing the quantization z - axis of total angular momentum along the magnetic field vector , one finds : @xmath14 where @xmath15 is the land factor .
this equation shows that , to first order of perturbation theory , any atomic level of total angular momentum quantum number @xmath7 is split by the action of a magnetic field into @xmath16 _ equally spaced _
sublevels , the splitting being proportional to the land factor @xmath15 and to the magnetic field strength .
it is important to note that a state of the form @xmath17 is an eigenvector of the _ total _
hamiltonian @xmath18 , with eigenvalue @xmath19 , only when the z - axis of the reference system points in the direction of the magnetic field .
the previous well - known results of first - order perturbation theory are correct only if the splitting produced by the magnetic field on a @xmath20level is small compared to the energy separation between the different @xmath20levels of the @xmath21 term under consideration . in other words , the standard theory of the zeeman effect is valid only in the limit of `` weak '' magnetic fields . here ,
`` weak '' means that the coupling of either the spin or the orbital angular momentum to the magnetic field is _ weaker _ than the coupling between the spin and the orbital angular momentum ( the spin - orbit coupling ) .
this is the so - called _ zeeman effect regime_. in the opposite limit , the magnetic field is so `` strong '' that the spin - orbit interaction can be considered as a perturbation compared to the magnetic interaction . in this case
the magnetic field dissolves the fine structure coupling that is , @xmath22 and @xmath23 are practically decoupled and precess independently around @xmath24 .
therefore , the quantum number @xmath7 loses its meaning . in this so - called _ complete paschen - back effect regime _
the magnetic hamiltonian is diagonal on the basis @xmath25 , and the eigenvalues are given by : @xmath26 in this `` strong '' field regime , the term @xmath21 splits into a number of components , each of which corresponds to particular values of @xmath27 . in the presence of a magnetic field ,
the corrections to the energy of the state @xmath28 due to the spin - orbit interaction has the form @xmath29 , where @xmath30 is a constant with dimensions of energy .
it is of interest to note that , since the spin - orbit coupling increases rapidly with increasing nuclear charge , the conditions for a `` strong '' field are met at a much lower field with light atoms ( like helium ) than with heavy atoms . in fact , as shown in fig [ energy ] , the levels with @xmath31 and @xmath32 of the upper term @xmath33 of the 10830 multiplet cross for magnetic strengths between 400 g and 1500 g , approximately .
this level - crossing regime corresponds to the _ incomplete paschen - back effect regime _ ,
in which the energy eigenvectors are gradually evolving from the form @xmath34 to the form @xmath35 as the magnetic field increases .
this range between the limiting cases of `` weak '' fields ( zeeman effect regime ) and `` strong '' fields ( complete paschen - back regime ) is more difficult to analyze since it implies the evaluation of matrix elements of the form @xmath36 , which ( in general ) can only be done numerically . to this aim
, we have developed a numerical code capable of computing , for a magnetic field of arbitrary intensity , the strenghts and the splittings of the @xmath37 and @xmath38 components in a line multiplet arising from the transition between two @xmath21 terms .
this code is very similar to a previous one that solves the problem of the intermediate paschen - back effect in a line with hyperfine structure ( @xcite @xcite ) .
we have developed a numerical code for the synthesis and inversion of stokes profiles in a milne - eddington atmosphere , considering ipbs as described in [ ipb ] .
this code has been built with the aim of analyzing spectropolarimetric observations of solar active regions , but we shall use it here to explore the effects of ipbs on the 10830 polarization profiles .
we consider a reference model atmosphere and calculate the multiplet component positions and strengths for various magnetic fields ( ranging from 0 to 3 kg ) .
the zeeman components are calculated accounting for ipbs and then compared to the usual approximation of lzs
. the most relevant parameters of the chosen milne - eddington model are @xmath39 , @xmath40 m , @xmath41 , @xmath42 , @xmath43 ( see , e.g. , @xcite @xcite for definitions of the milne - eddington parameters ) .
sample profiles are depicted in fig [ fig : profiles ] for a 500 g field with an inclination of 50@xmath44 from the line of sight .
there are a few obvious differences between the ipbs and lzs stokes v profiles : a ) the blue component of the multiplet , tr1 , shows a net asymmetry with a global offset towards negative values ; components of this transition , with the red component being stronger . ]
b ) a small non - doppler blueshift of the same component ( which is actually related to the asymmetry ) ; c ) a reduced amplitude of the red and blue lobes of the red component of the multiplet ( tr2 and tr3 ) .
the linear polarization profiles exhibit also some small asymmetry .
these differences are due to ipbs effects on the zeeman components of the transitions , discussed in [ comp ] below .
cccccc 10829.09 & @xmath45s@xmath46 & @xmath45p@xmath47 & 2.00 & 0.111 + 10830.25 & @xmath45s@xmath46 & @xmath45p@xmath46 & 1.75 & 0.333 + 10830.34 & @xmath45s@xmath46 & @xmath45p@xmath48 & 1.25 & 0.556 + in order to compare the ipb results with the lzs ones , one needs the relative strengths and effective land factors of the three components .
these are listed in table [ tblatom ] .
note that some of the values do not coincide with those used by other authors in previous works .
for example , concerning the lzs calculations the effective land factor ( @xmath49 ) of tr3 is 1.25 , whereas @xcite ( @xcite ) and @xcite ( @xcite ) take it to be 0.875 .
the relative strengths , which is used here and in other previous works on this multiplet , refers to the relative strengths of the components .
these are not the `` oscillator strengths '' of the transitions , which are also commonly denoted by @xmath50 . ]
( @xmath50 ) of tr1 , tr2 and tr3 are here 0.111 , 0.333 and 0.556 , respectively .
our @xmath50 agree with those of @xcite ( @xcite ) , but not with @xcite ( @xcite ) ( these authors quote _ relative oscillator strengths _ , but the conversion is straightforward ) .
it is not completely clear to us what the source of the discrepancy is . for the @xmath49 we used its standard definition ( e.g. , @xcite @xcite ) applied to the quantum numbers shown in the table .
the @xmath50 are calculated according to the usual formulae ( e.g. , @xcite @xcite ) .
thus , both of these quantities are derived straightforwardly from well - known atomic parameters . as we discuss below
, it is not possible to reproduce the observations using the lzs approximation .
it is conceivable that other authors had to modify their atomic parameters in an attempt to compensate for ipbs effects .
this would allow them to fit observed profiles with those calculated using lzs .
for example , the ratio of stokes v amplitudes of the blue and red components calculated with lzs is significantly larger than it should be . artificially reducing the @xmath49 of tr3 reduces the amplitude of the red component without affecting the intensity profiles , thus bringing the profiles closer to the observations .
ccc @xmath51 & 0.524 & 0.106 + @xmath38 & 0.004 & 0.111 + @xmath52 & -0.516 & 0.116 + ccc @xmath51 & 0.542 & 0.136 + @xmath51 & 0.410 & 0.118 + @xmath38 & 0.149 & 0.118 + @xmath38 & 0.022 & 0.007 + @xmath38 & -0.111 & 0.215 + @xmath52 & -0.371 & 0.215 + @xmath52 & -0.499 & 0.190 + ccc @xmath51 & 0.495 & 0.091 + @xmath51 & 0.371 & 0.215 + @xmath51 & 0.260 & 0.333 + @xmath38 & 0.111 & 0.215 + @xmath38 & -0.025 & 0.215 + @xmath38 & -0.149 & 0.118 + @xmath52 & -0.260 & 0.333 + @xmath52 & -0.410 & 0.118 + @xmath52 & -0.546 & 0.027 + the positions ( with respect to the zero - field wavelength ) and strengths of the various @xmath38 and @xmath37 components of the multiplet are shown in fig [ components ] .
tables 2 to 4 list the actual ipbs values at 500 g , which are also useful to match the curves in the left and right panels .
note that the positions of the tr1 components are not significantly disturbed by the presence of the other levels ( @xmath7=1 and @xmath7=2 ) in the field range depicted in the plots .
however , the strength of the @xmath37 components is indeed perturbed , introducing an asymmetry between the left- and right - handed polarizations . for our discussions in [ weak ] below it is important to note that this asymmetry increases linearly with the magnetic field .
the positions of the tr2 and tr3 components feel a much stronger perturbation for fields stronger than @xmath0500 g. this is due to the close proximity in the energy diagram of the upper levels of these transitions ( the @xmath32 and @xmath31 levels ) . similarly to the case of tr1 , the component strengths of tr2 and tr3 become asymmetric .
the asymmetry is rather linear for weak fields . as the field increases above @xmath0500 g , however , the behavior of the component strengths becomes non - linear .
finally , it is interesting to note the presence of a @xmath38 component in tr2 that does not exist in the lzs approximation .
this new component appears for non - zero fields in ipbs and becomes more important as the field increases .
the asymmetries between left- and right - handed @xmath37 components in all three transitions , are important .
they ultimately manifest themselves as asymmetries in the profiles and lead to net circular polarization even in the absence of velocity or magnetic field gradients .
fig [ components ] shows that the zeeman components of the various transitions tend towards the lzs values as the magnetic strength goes to zero .
one may thus expect that the ipbs profiles ( plotted in the top row ) would become similar to the lzs profiles in the weak field limit . however , this is not the case .
the profiles obtained from our calculations maintain their shape unchanged down to the lower limit of our simulation ( 0.1 g ) .
this apparent paradox has a simple , albeit slightly counterintuitive , explanation .
the stokes v profiles can be described conceptually as the subtraction of two bell - type functions , similar to gaussians or lorentzians . in the weak field limit
the separation between these functions is much smaller than their width , and the resulting stokes v profile is essentially a small residual of the subtraction .
a slight asymmetry between the two functions will show more conspicuously the closer together they are .
therefore , reducing the field results in `` amplifying '' the red / blue asymmetries .
let us analyze this in some detail considering a very simple scenario .
suppose that we have a stokes v profile that is the result of subtracting two gaussians separated by a small distance @xmath53 : @xmath54 the zero - crossing wavelength ( @xmath55 ) of @xmath56 is given by the condition : @xmath57 which , taking logarithms on both sides , leads to : @xmath58 if the asymmetry is small ( @xmath59 ) , a taylor expansion of the logarithm yields : @xmath60 for weak fields , the asymmetry @xmath61 depends linearly on the field ( @xmath62 ) .
we then conclude that : @xmath63 i.e. , the zero - crossing wavelength is shifted and does _ not _ depend on the magnetic field . in the derivation above we have been concerned only with strength asymmetries , neglecting the effect of asymmetric component positions .
if we denote by @xmath64 and @xmath65 the ( different ) component positions , a similar straightforward reasoning can be conducted to conclude that : @xmath66 where @xmath67 .
it is evident from eq ( [ x02 ] ) that position asymmetries have only a second - order effect on @xmath55 , since @xmath68 is proportional to @xmath69 .
we verified this property with our simulations .
consider now the area asymmetry : @xmath70 in the weak - field limit , the denominator of eq ( [ area ] ) is directly proportional to the field .
thus : @xmath71 where @xmath72 is a constant . expanding the exponentials in eq ( [ prof ] ) in a taylor series and truncating to first order
, we obtain that for @xmath73 : @xmath74 thus : @xmath75 inserting this into eq ( [ area2 ] ) and taking into account the odd parity of the second term in eq ( [ vweak ] ) , it is evident that : @xmath76 therefore , the net area asymmetry @xmath77 remains constant even for fields approaching zero .
all of the asymptotic properties of stokes v derived in this section ( constancy of the zero - crossing wavelength , net asymmetry , relative importance of weight and position asymmetries ) have been confirmed by further modeling using our numerical code .
it is important to note that the results we present here do not violate the principle of spectroscopic stability , since the zeeman components ( and their center of gravity ) approach asymptotically the lzs values .
the asymptotic values that we obtained for tr1 are @xmath55=340 m s@xmath13 and @xmath78=0.11 ( with the red lobe having a larger area ) .
it must be noted that this value of @xmath77 is influenced by the blend with the blue wing of the red component .
we estimated the effect of this blend by calculating @xmath79 for the lzs profile ( which should be zero ) , obtaining @xmath80=0.04 . in this case
it is the blue lobe that has larger area .
therefore , the presence of the blend has the effect of reducing the @xmath78 of the ipbs profile .
the actual value of @xmath55 is significantly model - dependent .
in particular , it is very sensitive to the doppler width of the line . doubling this parameter in our model ,
resulted in a factor 4 increase of @xmath55 .
for the red component of the multiplet ( tr2 and tr3 ) , we obtained @xmath55=95 m s@xmath13 and @xmath78=0.05 . here , as opposed to tr1 , the zero - crossing is redshifted and the blue lobe has a larger area .
the dependence of @xmath55 and @xmath78 for both lines are plotted in figs [ zerocross ] and [ asym ] .
the calculations reported in this paper show that ipbs effects need to be taken into account for proper diagnostics based on the multiplet at 10830 .
this is a very interesting set of lines , with some unique properties that make it an ideal tool for the exploration of chromospheric and coronal magnetism .
care must be taken , however , to consider all the relevant physical processes that intervene in the formation of the lines .
the ( still few ) works existing in the literature regarding stokes observations of these lines do not seem to agree on such fundamental parameters as : the rest wavelength of the tr1 line ; the effective land factor of the tr3 line ; the relative strengths of the transitions ( compare the values given in @xcite @xcite ; @xcite @xcite ; this work ) .
these are basic parameters that should be well known , either from laboratory experiments or from very simple calculations .
we believe that , at least partly , the existing confusion might have its origin in neglecting ipbs effects .
we have shown that ipbs is responsible for a non - doppler blueshift of tr1 , systematic profile asymmetries and a reduced amplitude of stokes v in the tr2+tr3 blend . as an illustrative example , consider the simple line - ratio experiment represented in fig [ vratio ] . for a fixed model atmosphere ,
the ratio of stokes v amplitudes in the red ( tr2+tr3 ) and the blue ( tr1 ) components of the multiplet depends smoothly on the magnetic field .
it is clear from the figure that neglecting ipbs leads to an erroneous estimate of the field .
interestingly , the ipbs curve is steeper than the lzs one , indicating that it is more sensitive to the field .
some ipbs signatures described in this paper are probably present in published observations .
if one compares the tr1 stokes v profile depicted in our fig [ fig : profiles ] ( upper right panel ) with those of @xcite ( @xcite ) ( figs 4 and 5 ) , the resemblance is striking . more specifically , the ipbs blueshift and the net offset towards negative values both become immediately apparent .
profile asymmetries are also present in their observations .
however , it is not possible to assess whether these arise from ipbs or from velocity or magnetic field gradients . | we investigate the formation of polarization profiles induced by a magnetic field in the multiplet at 10830 .
our analysis considers the zeeman splitting in the incomplete paschen - back regime .
the effects turn out to be important and produce measurable signatures on the profiles , even for fields significantly weaker than the level - crossing field ( @xmath0400 g ) .
when compared to profiles calculated with the usual linear zeeman effect , the incomplete paschen - back profiles exhibit the following conspicuous differences : a ) a non - doppler blueshift of the stokes v zero - crossing wavelength of the blue component ; b ) area and peak asymmetries , even in the absence of velocity and magnetic gradients ; c ) a @xmath025% reduction in the amplitude of the red component .
these features do not vanish in the weak field limit .
the spectral signatures that we analyze in this paper may be found in previous observations published in the literature . |
with the information superhighway fast becoming a reality , the problem of designing networks capable of accommodating multimedia ( both audio and video ) traffic in a multicast ( simultaneous transmission of data to multiple destinations ) environment has come to assume paramount importance @xcite . as discussed in kompella ,
pasquale and polyzos @xcite , one of the popular solutions to multicast routing involves tree construction .
two optimization criteria ( 1 ) the minimum worst - case transmission delay and ( 2 ) the minimum total cost are typically sought to be minimized in the construction of these trees .
network design problems where even one cost measure must be minimized , are often @xmath0-hard .
( see section a2 on network design in @xcite . ) but , in real - life applications , it is often the case that the network to be built is required to minimize multiple cost measures simultaneously , with different cost functions for each measure .
for example , as pointed out in @xcite , in the problem of finding good multicast trees , each edge has associated with it two edge costs : the construction cost and the delay cost .
the construction cost is typically a measure of the amount of buffer space or channel bandwidth used and the delay cost is a combination of the propagation , transmission and queuing delays .
such multi - criteria network design problems , with separate cost functions for each optimization criterion , also occur naturally in information retrieval @xcite and vlsi designs ( see @xcite and the references therein ) . with the advent of deep micron vlsi designs , the feature size has shrunk to sizes of 0.5 microns and less . as a result , the interconnect resistance , being proportional to the square of the scaling factor , has increased significantly .
an increase in interconnect resistance has led to an increase in interconnect delays thus making them a dominant factor in the timing analysis of vlsi circuits .
therefore vlsi circuit designers aim at finding minimum cost ( spanning or steiner ) trees given delay bound constraints on source - sink connections .
the above applications set the stage for the formal definition of multicriteria network design problems .
we explain this concept by giving a formal definition of a bicriteria network design problem .
a generic bicriteria network design problem , ( @xmath1 , @xmath2 , @xmath3 ) , is defined by identifying two minimization objectives , - @xmath1 and @xmath2 , - from a set of possible objectives , and specifying a membership requirement in a class of subgraphs , - @xmath3 .
the problem specifies a budget value on the first objective , @xmath1 , under one cost function , and seeks to find a network having minimum possible value for the second objective , @xmath2 , under another cost function , such that this network is within the budget on the first objective .
the solution network must belong to the subgraph - class @xmath3 .
for example , the problem of finding low - cost and low - transmission - delay multimedia networks @xcite can be modeled as the ( diameter , total cost , spanning tree)-bicriteria problem : given an undirected graph @xmath4 with two weight functions @xmath5 and @xmath6 for each edge @xmath7 modeling construction and delay costs respectively , and a bound @xmath8 ( on the total delay ) , find a minimum @xmath9-cost spanning tree such that the diameter of the tree under the @xmath10-costs is at most @xmath8 .
it is easy to see that the notion of bicriteria optimization problems can be easily extended to the more general multicriteria optimization problems . in this paper , we will be mainly concerned with bicriteria network design problems . in the past
, the problem of minimizing two cost measures was often dealt with by attempting to minimize some combination of the two , thus converting it into a unicriterion problem .
this approach fails when the two criteria are very disparate .
we have chosen , instead , to model bicriteria problems as that of minimizing one criterion subject to a budget on the other .
we argue that this approach is both general as well as robust .
it is more general because it subsumes the case where one wishes to minimize some functional combination of the two criteria .
it is more robust because the quality of approximation is independent of which of the two criteria we impose the budget on .
we elaborate on this more in sections [ sec : equivalence ] and [ sec : general ] .
the organization of the rest of the paper is as follows : section [ sec : contributions ] summarizes the results obtained in this paper ; section [ sec : previous ] discusses related research work ; section [ sec : hardness ] contains the hardness results ; section [ sec : equivalence ] shows that the two alternative ways of formulating a given bicriteria problem are indeed equivalent ; section [ sec : general ] demonstrates the generality of the bicriteria approach ; section [ sec : parametric ] details the parametric search technique ; section [ sec : diameter ] presents the approximation algorithm for diameter constrained steiner trees ; section [ sec : treewidth ] contains the results on treewidth - bounded graphs ; section [ sec : concluding ] contains some concluding remarks and open problems .
the area of unicriterion optimization problems for network design is vast and well - explored ( see @xcite and the references therein . ) .
ravi et al .
@xcite studied the degree - bounded minimum cost spanning tree problem and provided an approximation algorithm with performance guarantee ( @xmath11 ) .
the ( degree , diameter , spanning tree ) problem was studied by ravi @xcite in the context of finding good broadcast networks . there
he provides an approximation algorithm for the ( degree , diameter , spanning tree ) problem with performance guarantee ( @xmath12 ) , on the degree he finds a tree whose total cost is at most @xmath13 times the optimal and whose degree is at most @xmath14 . ] .
the ( diameter , total cost , spanning tree ) entry in table 1 corresponds to the diameter - constrained minimum spanning tree problem introduced earlier .
it is known that this problem is @xmath0-hard even in the special case where the two cost functions are identical @xcite .
awerbuch , baratz and peleg @xcite gave an approximation algorithm with @xmath15 performance guarantee for this problem - i.e. the problem of finding a spanning tree that has simultaneously small diameter ( i.e. , shallow ) and small total cost ( i.e. , light ) , both under the same cost function .
khuller , raghavachari and young @xcite studied an extension called _ light , approximate shortest - path trees ( last ) _ and gave an approximation algorithm with @xmath15 performance guarantee .
kadaba and jaffe @xcite , kompella et al .
@xcite , and zhu et al .
@xcite considered the ( diameter , total cost , steiner tree ) problem with two edge costs and presented heuristics without any guarantees .
it is easy to construct examples to show that the solutions produced by these heuristics in @xcite , can be arbitrarily bad with respect to an optimal solution .
a closely related problem is that of finding a diameter - constrained shortest path between two pre - specified vertices @xmath16 and @xmath17 , or ( diameter , total cost , @xmath16-@xmath17 path ) .
this problem , termed the multi - objective shortest path problem ( mosp ) in the literature , is @xmath0-complete and warburton @xcite presented the first fully polynomial approximation scheme ( @xmath18 ) for it .
hassin @xcite provided a strongly polynomial @xmath18 for the problem which improved the running time of warburton @xcite .
this result was further improved by phillips @xcite .
the ( total cost , total cost , spanning tree)-bicriteria problem has been recently studied by ganley et al .
they consider a more general problem with more than two weight functions .
they also gave approximation algorithms for the restricted case when each weight function obeys triangle inequality .
however , their algorithm does not have a bounded performance guarantee with respect to each objective .
many @xmath0-hard problems have exact solutions when attention is restricted to the class of treewidth - bounded graphs and much work has been done in this area ( see @xcite and the references therein ) . independently , bern , lawler and wong @xcite introduced the notion of decomposable graphs .
later , it was shown @xcite that the class of decomposable graphs and the class of treewidth - bounded graphs are equivalent .
bicriteria network design problems restricted to treewidth - bounded graphs have been previously studied in @xcite .
in this paper , we study the complexity and approximability of a number of bicriteria network design problems .
the three objectives we consider are : ( i ) total cost , ( ii ) diameter and ( iii ) degree of the network .
these reflect the price of synthesizing the network , the maximum delay between two points in the network and the reliability of the network , respectively .
the _ total cost _ objective is the sum of the costs of all the edges in the subgraph .
the _ diameter _
objective is the maximum distance between any pair of nodes in the subgraph .
the _ degree _ objective denotes the maximum over all nodes in the subgraph , of the degree of the node .
the class of subgraphs we consider in this paper are mainly _ steiner trees _ ( and hence _ spanning trees _ as a special case ) ; although several of our results extend to more general connected subgraphs such as generalized steiner trees .
as mentioned in @xcite , most of the problems considered in this paper , are @xmath0-hard for arbitrary instances even when we wish to find optimum solutions with respect to a single criterion . given the hardness of finding optimal solutions , we concentrate on devising approximation algorithms with worst case performance guarantees . recall that an approximation algorithm for a minimization problem @xmath19 provides a * performance guarantee * of @xmath20 if for every instance @xmath21 of @xmath19 , the solution value returned by the approximation algorithm is within a factor @xmath20 of the optimal value for @xmath21 .
here , we extend this notion to apply to bicriteria optimization problems .
an @xmath22-approximation algorithm for an ( @xmath1 , @xmath2 , @xmath3)-bicriteria problem is defined as a polynomial - time algorithm that produces a solution in which the first objective ( @xmath1 ) value , is at most @xmath23 times the budget , and the second objective ( @xmath2 ) value , is at most @xmath24 times the minimum for any solution that is within the budget on @xmath1 .
the solution produced must belong to the subgraph - class @xmath3 .
analogous definitions can be given when @xmath1 and/or @xmath2 are maximization objectives .
table 1 contains the performance guarantees of our approximation algorithms for finding spanning trees , @xmath3 , under different pairs of minimization objectives , @xmath1 and @xmath2 . for each problem
cataloged in the table , two different costs are specified on the edges of the undirected graph : the first objective is computed using the first cost function and the second objective , using the second cost function .
the rows are indexed by the budgeted objective .
for example the entry in row @xmath1 , column @xmath2 , denotes the performance guarantee for the problem of minimizing objective @xmath2 with a budget on the objective @xmath1 .
all the results in table 1 extend to finding steiner trees with at most a constant factor worsening in the performance ratios .
for the diagonal entries in the table the extension to steiner trees follows from theorem [ better - scale - thm ] .
algorithm dcst of section [ sec : diameter ] in conjunction with algorithm bicriteria - equivalence of section [ sec : equivalence ] yields the ( diameter , total cost , steiner tree ) and ( total cost , diameter , steiner tree ) entries .
the other nondiagonal entries can also be extended to steiner trees and these extensions will appear in the journal versions of @xcite .
our results for arbitrary graphs can be divided into three general categories . * table 1 .
performance guarantees for finding spanning trees in an arbitrary graph on @xmath25 nodes .
asterisks indicate results obtained in this paper .
@xmath26 is a fixed accuracy parameter . *
+ first , as mentioned before , there are two natural alternative ways of formulating general bicriteria problems : ( i ) where we impose the budget on the first objective and seek to minimize the second and ( ii ) where we impose the budget on the second objective and seek to minimize the first .
we show that an @xmath22-approximation algorithm for one of these formulations naturally leads to a @xmath27-approximation algorithm for the other
. thus our definition of a bicriteria approximation is independent of the choice of the criterion that is budgeted in the formulation .
this makes it a robust definition and allows us to fill in the entries for the problems ( @xmath2 , @xmath1 , @xmath3 ) by transforming the results for the corresponding problems ( @xmath1 , @xmath2 , @xmath3 ) .
second , the diagonal entries in the table follow as a corollary of a general result ( theorem [ better - scale - thm ] ) which is proved using a parametric search algorithm .
the entry for ( degree , degree , spanning tree ) follows by combining theorem [ better - scale - thm ] with the @xmath13-approximation algorithm for the degree problem in @xcite . in @xcite
they actually provide an @xmath28-approximation algorithm for the weighted degree problem .
the weighted degree of a subgraph is defined as the maximum over all nodes of the sum of the costs of the edges incident on the node in the subgraph .
hence we actually get an @xmath29-approximation algorithm for the ( weighted degree , weighted degree , spanning tree)-bicriteria problem .
similarly , the entry for ( diameter , diameter , spanning tree ) follows by combining theorem [ better - scale - thm ] with the known exact algorithms for minimum diameter spanning trees @xcite ; while the result for ( total cost , total cost , spanning tree ) follows by combining theorem [ better - scale - thm ] with an exact algorithm to compute a minimum spanning tree @xcite .
finally , we present a cluster based approximation algorithm and a solution based decomposition technique for devising approximation algorithms for problems when the two objectives are different .
our techniques yield @xmath30-approximation algorithms for the ( diameter , total cost , steiner tree ) and the ( degree , total cost , steiner tree ) problems . *
* * consider , for example , the conjunction of this theorem with the results of goemans et al .
this leads to a host of bicriteria approximation results when two costs are specified on edges for finding minimum - cost generalized steiner trees , minimum @xmath31-edge connected subgraphs , or any other network design problems specified by weakly supermodular functions .
thus for example , we get @xmath15-approximation algorithms for the ( total cost , total cost , generalized steiner tree ) and ( total cost , total cost , @xmath31-edge connected subgraph)-bicriteria problems .
( see @xcite for the results on the corresponding unicriterion problems . ) similarly , given an undirected graph with two costs specified on each node we can get logarithmic approximations for the minimum node - cost steiner tree using the result of klein and ravi @xcite . as another example , with two edge - cost functions , and an input number @xmath31 , we can use the result of blum et al .
@xcite to obtain an @xmath15-approximation for the minimum cost ( under both functions ) tree spanning at least @xmath31 nodes . * * * * * * * * * * * we also study the bicriteria problems mentioned above for the class of treewidth - bounded graphs
. examples of treewidth - bounded graphs include trees , series - parallel graphs , @xmath31-outerplanar graphs , chordal graphs with cliques of size at most @xmath31 , bounded - bandwidth graphs etc .
we use a dynamic programming technique to show that for the class of treewidth - bounded graphs , there are either polynomial - time or pseudopolynomial - time algorithms ( when the problem is * np*-complete ) for several of the bicriteria network design problems studied here . a * polynomial time approximation scheme * ( ptas ) for problem
@xmath19 is a family of algorithms @xmath32 such that , given an instance @xmath21 of @xmath19 , for all @xmath33 , there is a polynomial time algorithm @xmath34 that returns a solution which is within a factor @xmath35 of the optimal value for @xmath21 . a polynomial time approximation scheme in which the running time grows as a polynomial function of @xmath36
is called a * fully polynomial time approximation scheme*. here we show how to convert these pseudopolynomial - time algorithms for problems restricted to treewidth - bounded graphs into fully polynomial - time approximation schemes using a general scaling technique
. stated in our notation , we obtain polynomial time approximation algorithms with performance of @xmath37 , for all @xmath33 .
the results for treewidth - bounded graphs are summarized in table 2 .
as before , the rows are indexed by the budgeted objective .
all algorithmic results in table 2 also extend to steiner trees in a straightforward way .
our results for treewidth - bounded graphs have an interesting application in the context of finding optimum broadcast schemes .
kortsarz and peleg @xcite gave @xmath28-approximation algorithms for the minimum broadcast time problem for series - parallel graphs . combining our results for the ( degree , diameter , spanning tree ) for treewidth - bounded graphs with the techniques in @xcite
, we obtain an @xmath38-approximation algorithm for the minimum broadcast time problem for treewidth - bounded graphs ( series - parallel graphs have a treewidth of @xmath39 ) , improving and generalizing the result in @xcite .
note that the best known result for this problem for general graphs is by ravi @xcite who provides an approximation algorithm performance guarantee ( @xmath40 ) . [
cols="^,^,^,^",options="header " , ] * table 2 .
bicriteria spanning tree results for treewidth - bounded graphs . *
the problem of finding a minimum degree spanning tree is strongly @xmath0-hard @xcite .
this implies that all spanning tree bicriteria problems , where one of the criteria is degree , are also strongly @xmath0-hard . in contrast , it is well known that the minimum diameter spanning tree problem and the minimum cost spanning tree problems have polynomial time algorithms ( see @xcite and the references therein ) .
the ( diameter , total cost , spanning tree)-bicriteria problem is strongly @xmath0-hard even in the case where both cost functions are identical @xcite .
here we give the details of the reduction to show that ( diameter , total cost , spanning tree ) is weakly @xmath0-hard even for series - parallel graphs ( i.e. graphs with treewidth at most @xmath39 ) .
similar reductions can be given to show that ( diameter , diameter , spanning tree ) and ( total cost , total cost , spanning tree ) are also weakly @xmath0-hard for series - parallel graphs .
we first recall the definition of the * partition * problem @xcite . as an instance of the * partition *
problem we are given a set @xmath41 of positive integers and the question is whether there exists a subset @xmath42 such that @xmath43 .
[ np - thm ] ( diameter , total cost , spanning tree ) is @xmath0-hard for series - parallel graphs .
reduction from the * partition * problem . given an instance @xmath44 of the * partition * problem
, we construct a series parallel graph @xmath45 with @xmath46 vertices , @xmath47 and @xmath48 edges .
we attach a pair of parallel edges , @xmath49 and @xmath50 , between @xmath51 and @xmath52 , @xmath53 .
we now specify the two cost functions @xmath54 and @xmath55 on the edges of this graph ; @xmath56 .
let @xmath57 .
now it is easy to show that @xmath45 has a spanning tree of @xmath10-diameter at most @xmath58 and total @xmath9-cost at most @xmath58 if and only if there is a solution to the original instance @xmath59 of the * partition * problem . * * * * * * * * * * we now show that the ( diameter , total - cost , steiner tree ) problem is hard to approximate within a logarithmic factor . an approximation algorithm provided in section [ sec : diameter ] .
there is however a gap between the results of theorems [ fix - dia ] and [ spanning - thm ] .
our non - approximability result is obtained by an approximation preserving reduction from the * min set cover*. an instance @xmath60 of the * min set cover * problem consists of a universe @xmath61 and a collection of subsets @xmath62 , each set @xmath63 having an associated cost @xmath64 .
the problem is to find a minimum cost collection of the subsets whose union is @xmath59 .
[ th : feige ] recently @xcite have independently shown the following non - approximability result : + it is @xmath65-hard to find an approximate solution to the * min set cover * problem , with a universe of size @xmath31 , with performance guarantee better than @xmath66 . * * * * recently feige @xcite has shown that , unless @xmath67 , the * min set cover * problem , with a universe of size @xmath31 , can not be approximated to better than a @xmath68 factor . * * * * * * * * * * * [ fix - dia ] there is an approximation preserving reduction from * min set cover * problem to the ( diameter , total cost , steiner tree ) problem . thus : unless @xmath69 , given an instance of the ( diameter , total cost , steiner tree ) problem with @xmath31 sites , there is no polynomial - time approximation algorithm that outputs a steiner tree of diameter at most the bound @xmath70 , and cost at most @xmath71 times that of the minimum cost diameter-@xmath70 steiner tree , for @xmath72 .
we give an approximation preserving reduction from the * min set cover * problem to the ( diameter , total cost , steiner tree ) problem .
given an instance @xmath60 of the * min set cover * problem where @xmath73 and @xmath74 , where the cost of the set @xmath63 is @xmath64 , we construct an instance @xmath45 of the ( diameter , total cost , steiner tree ) problem as follows . the graph @xmath45 has a node @xmath75 for each element @xmath75 of @xmath59 , a node @xmath76 for each set @xmath63 , and an extra `` enforcer - node '' @xmath25 . for each set @xmath63
, we attach an edge between nodes @xmath25 and @xmath76 of @xmath9-cost @xmath64 , and @xmath10-cost @xmath77 .
for each element @xmath75 and set @xmath78 such that @xmath79 we attach an edge @xmath80 of @xmath9-cost , 0 , and @xmath10-cost , @xmath77 .
in addition to these edges , we add a path @xmath81 made of two edges of @xmath9-cost , 0 , and @xmath10-cost , @xmath77 , to the enforcer node @xmath25 ( see figure [ fig : hardapprox ] ) .
the path @xmath81 is added to ensure that all the nodes @xmath75 are connected to @xmath25 using a path of @xmath10-cost at most 2 . all other edges in the graph are assigned infinite @xmath9 and @xmath10-costs .
the nodes @xmath75 along with @xmath25 and the two nodes of @xmath81 are specified to be the terminals for the steiner tree problem instance .
we claim that @xmath45 has a @xmath9-cost steiner tree of diameter at most @xmath82 and cost @xmath83 if and only if the original instance @xmath60 has a solution of cost @xmath83 .
note that any steiner tree of diameter at most @xmath82 must contain a path from @xmath75 to @xmath25 , for all @xmath84 , that uses an edge @xmath85 for some @xmath78 such that @xmath79 .
hence any steiner tree of diameter at most @xmath82 provides a feasible solution of equivalent @xmath9-cost to the original set cover instance .
the proof now follows from theorem [ th : feige ] .
in section [ sec : motivation ] , we claimed that our formulation for bicriteria problems is robust and general . in this section , we justify these claims . in this section , we show that our formulation for bicriteria problems is robust and general .
let @xmath45 be a graph with two ( integral ) cost functions , @xmath9 and @xmath10 ( typically edge costs or node costs ) .
let @xmath1 ( @xmath2 ) be a minimization objective computed using cost function @xmath9 ( @xmath10 ) .
let the budget bound on the @xmath9-cost '' or `` @xmath9-cost '' in this section to mean the value of the objective function computed using @xmath9 , and not to mean the total of all the @xmath9 costs in the network . ]
( @xmath10-cost ) of a solution subgraph be denoted by @xmath83 ( @xmath8 ) .
there are two natural ways to formulate a bicriteria problem : ( i ) ( @xmath1 , @xmath2 , @xmath3 ) - find a subgraph in @xmath3 whose @xmath1-objective value ( under the @xmath9-cost ) is at most @xmath83 and which has minimum @xmath2-objective value ( under the @xmath10-cost ) , ( ii ) ( @xmath2 , @xmath1 , @xmath3 ) - find a subgraph in @xmath3 whose @xmath2-objective value ( under the @xmath10-cost ) is at most @xmath8 and which has minimum @xmath1-objective value ( under the @xmath9-cost ) .
note that bicriteria problems are generally hard , when the two criteria are _ hostile _ with respect to each other - the minimization of one criterion conflicts with the minimization of the other .
a good example of hostile objectives are the degree and the total edge cost of a spanning tree in an unweighted graph @xcite .
two minimization criteria are formally defined to be hostile whenever the minimum value of one objective is monotonically nondecreasing as the budget ( bound ) on the value of the other objective is decreased .
let @xmath86@xmath87 be any @xmath22-approximation algorithm for ( @xmath1 , @xmath2 , @xmath3 ) on graph @xmath45 with budget @xmath83 under the @xmath9-cost .
we now show that there is a transformation which produces a @xmath27-approximation algorithm for ( @xmath2 , @xmath1 , @xmath3 ) .
the transformation uses binary search on the range of values of the @xmath9-cost with an application of the given approximation algorithm , @xmath86 , at each step of this search .
let the minimum @xmath9-cost of a @xmath88-bounded subgraph in @xmath3 be @xmath89 .
let @xmath90 be an upper bound on the @xmath9-cost of any @xmath8-bounded subgraph in @xmath3 .
note that @xmath90 is at most some polynomial in @xmath25 times the maximum @xmath9-cost ( of an edge or a node ) .
hence @xmath91 is at most a polynomial in terms of the input specification .
let @xmath92 ( @xmath93 ) denote the @xmath9-cost ( @xmath10-cost ) of the subgraph output by algorithm bicriteria - equivalence given below .
if @xmath45 contains a @xmath8-bounded subgraph in @xmath3 then algorithm bicriteria - equivalence outputs a subgraph from @xmath3 whose @xmath9-cost is at most @xmath23 times that of the minimum @xmath9-cost @xmath8-bounded subgraph and whose @xmath10-cost is at most @xmath94 . since @xmath1 and @xmath2 are hostile criteria it follows that the binary search in step [ binsearch - step ] is well defined .
assume that @xmath3 contains a @xmath8-bounded subgraph .
then , since @xmath86@xmath95 returns a subgraph with @xmath10-cost at most @xmath94 , it is clear that algorithm bicriteria - equivalence outputs a subgraph in this case . as a consequence of step [ step1 ] and the performance guarantee of the approximation algorithm @xmath86
, we get that @xmath96 . by step [ step2 ]
we have that @xmath97 and @xmath98 .
thus algorithm bicriteria - equivalence outputs a subgraph from @xmath3 whose @xmath9-cost is at most @xmath23 times that of the minimum @xmath9-cost @xmath8-bounded subgraph and whose @xmath10-cost is at most @xmath94 .
note however that in general the resulting @xmath27-approximation algorithm is , not _
strongly _ polynomial since it depends on the range of the @xmath9-costs .
but it is a _ polynomial - time _
algorithm since its running time is linearly dependent on @xmath99 the largest @xmath9-cost .
the above discussion leads to the following theorem .
[ equiv - thm ] any @xmath22-approximation algorithm for ( @xmath1 , @xmath2 , @xmath3 ) can be transformed in polynomial time into a @xmath27-approximation algorithm for ( @xmath2 , @xmath1 , @xmath3 ) .
our formulation is more general because it subsumes the case where one wishes to minimize some functional combination of the two criteria .
we briefly comment on this next . for the purposes of illustration
let @xmath1 and @xmath2 be two objective functions and let us say that we wish to minimize the sum of the two objectives @xmath1 and @xmath2 . call this an ( @xmath1 @xmath100 @xmath2 , @xmath3 ) problem .
let @xmath86@xmath87 be any @xmath22-approximation algorithm for ( @xmath1 , @xmath2 , @xmath3 ) on graph @xmath45 with budget @xmath83 under the @xmath9-cost .
we show that @xmath101 , there is a polynomial time @xmath102-approximation algorithm for the ( @xmath1 @xmath100 @xmath2 , @xmath3 ) problem .
the transformation uses simple linear search in steps of @xmath103 over the range of values of the @xmath9-cost with an application of the given approximation algorithm , @xmath86 , at each step of this search .
let the optimum value for the ( @xmath1 @xmath100 @xmath2 , @xmath3 ) problem on a graph @xmath45 be @xmath104 , where @xmath105 and @xmath106 denote respectively the contribution of the two costs @xmath9 and @xmath10 for @xmath1 and @xmath2 .
let @xmath107 ( @xmath108 ) denote the @xmath9-cost ( @xmath10-cost ) of the subgraph output by @xmath86@xmath87 .
finally , let @xmath109 denote the value computed by algorithm convert .
let @xmath86@xmath87 be any @xmath22-approximation algorithm for ( @xmath1 , @xmath2 , @xmath3 ) on graph @xmath45 with budget @xmath83 under the @xmath9-cost .
then , for all @xmath33 , there is a polynomial time @xmath110-approximation algorithm for the ( @xmath1 @xmath100 @xmath2 , @xmath3 ) problem .
* proof sketch : * consider the iteration of the binary search in which the bound on the @xmath9-cost is @xmath111 such that @xmath112 .
notice that such a bound is considered as a result of discretization of the interval @xmath113 $ ] .
then as a consequence of the performance guarantee of the approximation algorithm @xmath86 , we get that @xmath114 by step [ stepa ] , the performance guarantee of the algorithm @xmath86 , and the hostility of @xmath1 and @xmath2 , we have that @xmath115 .
thus @xmath116 .
since algorithm convert outputs a subgraph from @xmath3 the sum of whose @xmath9-cost and @xmath10-cost is minimized , we have that @xmath117 } \left ( heu_c({\cal c } ' ) + heu_d({\cal c } ' ) \right ) \leq ( 1 + \epsilon ) \max \ { \alpha,\beta \ } ( opt_{c + d}).\ ] ] a similar argument shows that an @xmath22-approximation algorithm @xmath86@xmath87 , for a ( @xmath1 , @xmath2 , @xmath3 ) problem can be used to find devise a polynomial time @xmath118 approximation algorithm for the ( @xmath1 @xmath119 @xmath2 , @xmath3 ) problem .
a similar argument can also be given for other basic functional combinations .
we make two additional remarks .
1 . algorithms for solving ( @xmath54(@xmath1 , @xmath2 ) , @xmath3 ) problems can not in general guarantee any bounded performance ratios for solving the ( @xmath1 , @xmath2 , @xmath3 ) problem .
for example , a solution for the ( total cost + total cost , spanning tree ) problem or the ( total cost / total cost , spanning tree ) problem can not be directly used to find a good @xmath22-approximation algorithm for the ( total cost , total cost , spanning tree)-bicriteria problem .
2 . the use of approximation algorithms for ( @xmath1 , @xmath2 , @xmath3)-bicriteria problems , to solve ( @xmath54(@xmath1 , @xmath2 ) , @xmath3 ) problems ( @xmath54 denotes a function combination of the objectives ) does not always yield the best possible solutions . for example problems such as ( total cost + total cost , spanning tree ) and ( total cost / total cost , spanning tree )
@xcite can be solved exactly in polynomial time by direct methods but can only be solved approximately using any algorithm for the ( total cost , total cost , spanning tree)-bicriteria problem . * * * * the above discussion points out that while a good solution to the ( @xmath1 , @xmath2 , @xmath3)-bicriteria problem yields a `` good '' solution to any unicriterion version , the converse is not true .
it is in this sense that we say our formulation of bicriteria network design problems is general and subsumes other functional combinations . *
* * * * * * * * *
in this section , we present approximation algorithms for a broad class of bicriteria problems where both the objectives in the problem are of the same type ( e.g. , both are total edge costs of some network computed using two different costs on edges , or both are diameters of some network calculated using two different costs etc . ) . as
before , let @xmath45 be a graph with two ( integral ) cost functions , @xmath9 and @xmath10 .
let @xmath83 denote the budget on criteria @xmath1 .
we assume that the @xmath9 and @xmath10 cost functions are of the same kind ; i.e. , they are both costs on edges or , costs on nodes .
let @xmath120@xmath121 be any @xmath20-approximation algorithm that on input @xmath45 produces a solution subgraph in @xmath3 minimizing criterion @xmath1 , under the single cost function @xmath54 . in a mild abuse of notation
, we also let @xmath120@xmath121 denote the ( @xmath54-)cost of the subgraph output by @xmath120@xmath121 when running on input @xmath45 under cost function @xmath54 .
we use the following additional notation in the description of the algorithm and the proof of its performance guarantee . given constants @xmath122 and @xmath123 and two cost functions @xmath54 and @xmath55 , defined on edges ( nodes ) of a graph
, @xmath124 denotes the composite function that assigns a cost @xmath125 to each edge ( node ) in the graph .
let @xmath126 denote the cost of the subgraph , returned by @xmath120@xmath127 ( under the @xmath128-cost function ) .
let the minimum @xmath10-cost of a @xmath83-bounded subgraph in @xmath3 be @xmath129 .
let @xmath92 ( @xmath93 ) denote the @xmath9-cost ( @xmath10-cost ) of the subgraph output by algorithm parametric - search given below .
let @xmath26 be a fixed accuracy parameter .
in what follows , we devise a @xmath130-approximation algorithm for ( @xmath1 , @xmath1 , @xmath3 ) , under the two cost functions @xmath9 and @xmath10 .
the algorithm consists of performing a binary search with an application of the given approximation algorithm , @xmath120 , at each step of this search .
[ binsearch - justify ] the binary search , in step [ psbinsearch - step ] of algorithm parametric - search is well - defined .
since @xmath131@xmath120@xmath132 is the same as @xmath120@xmath133 , we get that @xmath134 @xmath135 @xmath120@xmath136 @xmath120@xmath137 .
hence @xmath138 is a monotone nonincreasing function of @xmath139 .
thus the binary search in step [ psbinsearch - step ] of algorithm parametric - search is well - defined .
if @xmath45 contains a @xmath83-bounded subgraph in @xmath3 then algorithm parametric - search outputs a subgraph from @xmath3 whose @xmath10-cost is at most @xmath140 times that of the minimum @xmath10-cost @xmath83-bounded subgraph and whose @xmath9-cost is at most @xmath141 . by claim [ binsearch - justify ]
we have that the binary search in step [ psbinsearch - step ] of algorithm parametric - search is well - defined .
assume that @xmath3 contains a @xmath83-bounded subgraph .
then , since @xmath120@xmath142 returns a subgraph with cost at most @xmath143 , under the @xmath144-cost function , it is clear that algorithm parametric - search outputs a subgraph in this case . as a consequence of step [ psstep1 ] and the performance guarantee of the approximation algorithm @xmath120
, we get that @xmath145 by step [ psstep2 ] we have that the subgraph output by algorithm parametric - search has the following bounds on the @xmath9-costs and the @xmath10-costs
. @xmath146 @xmath147 thus algorithm parametric - search outputs a subgraph from @xmath3 whose @xmath10-cost is at most @xmath140 times that of the minimum @xmath10-cost @xmath83-bounded subgraph and whose @xmath9-cost is at most @xmath148 .
note however that the resulting @xmath149-approximation algorithm for ( @xmath1 , @xmath1 , @xmath3 ) may not be _ strongly _ polynomial since it depends on the range of the @xmath10-costs .
but it is a _ polynomial - time _
algorithm since its running time is linearly dependent on @xmath150 .
note that @xmath151 is at most some polynomial in @xmath25 times the maximum @xmath10-cost ( of an edge or a node ) .
hence @xmath152 is at most a polynomial in terms of the input specification .
the above discussion leads to the following theorem .
[ better - scale - thm ] any @xmath20-approximation algorithm that produces a solution subgraph in @xmath3 minimizing criterion @xmath1 can be transformed into a @xmath149-approximation algorithm for ( @xmath1,@xmath1,@xmath3 ) .
the above theorem can be generalized from the bicriteria case to the multicriteria case ( with appropriate worsening of the performance guarantees ) where all the objectives are of the same type but with different cost functions .
in this section , we describe algorithm dcst , our @xmath153-approximation algorithm for ( diameter , total cost , steiner tree ) or the diameter - bounded minimum steiner tree problem .
note that ( diameter , total cost , steiner tree ) includes ( diameter , total cost , spanning tree ) as a special case .
we first state the problem formally : given an undirected graph @xmath4 , with two cost functions @xmath9 and @xmath10 defined on the set of edges , diameter bound @xmath70 and terminal set @xmath154 , the ( diameter , total cost , steiner tree ) problem is to find a tree of minimum @xmath9-cost connecting the set of terminals in @xmath155 with diameter at most @xmath70 under the @xmath10-cost .
the technique underlying algorithm dcst is very general and has wide applicability .
hence , we first give a brief synopsis of it .
the basic algorithm works in @xmath156 phases ( iterations ) .
initially the solution consists of the empty set . during each phase of the algorithm
we execute a subroutine @xmath157 to choose a subgraph to add to the solution .
the subgraph chosen in each iteration is required to possess two desirable properties .
first , it must not increase the budget value of the solution by more than @xmath70 ; second , the solution cost with respect to @xmath2 must be no more than @xmath89 , where @xmath89 denotes the minimum @xmath9-cost of a @xmath8 bounded subgraph in * s*. since the number of iterations of the algorithm is @xmath13 we get a @xmath158-approximation algorithm .
the basic technique is fairly straightforward .
the non - trivial part is to devise the right subroutine @xmath157 to be executed in each phase .
@xmath157 must be chosen so as to be able to prove the required performance guarantee of the solution .
we use the solution based decomposition technique @xcite in the analysis of our algorithm .
the basic idea ( behind the solution based decomposition technique ) is to use the existence of an optimal solution to prove that the subroutine @xmath157 finds the desired subgraph in each phase .
we now present the specifics of algorithm dcst .
the algorithm maintains a set of connected subgraphs or _
clusters _ each with its own distinguished vertex or _ center_.
initially each terminal is in a cluster by itself . in each phase , clusters are merged in pairs by adding paths between their centers . since the number of clusters comes down by a factor of @xmath39 each phase , the algorithm terminates in @xmath159 phases with one cluster .
it outputs a spanning tree of the final cluster as the solution .
we make a few points about algorithm dcst : 1 .
the clusters formed in step [ dcst - match ] need not be disjoint .
2 . all steps , except step [ dcst - paths ] , in algorithm dcst can be easily seen to have running times independent of the weights . we employ hassin s strongly polynomial @xmath18 for step [ dcst - paths ] @xcite .
hassin s approximation algorithm for the @xmath70-bounded minimum @xmath9-cost path runs in time @xmath160 .
thus algorithm dcst is a strongly polynomial time algorithm .
3 . instead of finding an exact minimum cost matching in step [ dcst - minwtmatching ]
, we could find an approximate minimum cost matching @xcite .
this would reduce the running time of the algorithm at the cost of introducing a factor of @xmath39 to the performance guarantee .
we now state some observations that lead to a proof of the performance guarantee of algorithm dcst .
assume , in what follows , that @xmath45 contains a diameter @xmath70-bounded steiner tree .
we also refer to each iteration of step [ dcst - repeat ] as a phase .
[ log - lemma ] algorithm dcst terminates in @xmath159 phases
. let @xmath161 denote the number of clusters in phase @xmath84 .
note that @xmath162 since we pair up the clusters ( using a matching in step [ dcst - match ] ) .
hence we are left with one cluster after phase @xmath159 and algorithm dcst terminates .
the next claim points out as clusters get merged , the nodes within each cluster are not too far away ( with respect to @xmath10-distance ) from the center of the cluster .
this intuitively holds for the following important reasons .
first , during each phase , the graph @xmath163 has as its vertices , the centers of the clusters in that iteration . as a result
, we merge the clusters by joining their centers in step [ dcst - match ] .
second , in step [ dcst - match ] , for each pair of clusters @xmath164 and @xmath165 that are merged , we select one of their centers , @xmath166 or @xmath167 as the center @xmath168 for the merged cluster @xmath169 .
this allows us to inductively maintain two properties : ( i ) the required distance of the nodes in a cluster to their centers in an iteration @xmath84 is @xmath170 and ( ii ) the center of a cluster at any given iteration is a terminal node .
[ claim : radius ] let @xmath171 be any cluster in phase @xmath84 of algorithm dcst .
let @xmath172 be the center of @xmath173 .
then any node @xmath174 in @xmath173 is reachable from @xmath172 by a diameter-@xmath170 path in @xmath173 under the @xmath10-cost .
note that the existence of a diameter @xmath70-bounded steiner tree implies that all paths added in step [ dcst - match ] have diameter at most @xmath70 under @xmath10-cost .
the proof now follows in a straightforward fashion by induction on @xmath84 .
[ dcostlemma ] algorithm dcst outputs a steiner tree with diameter at most @xmath175 under the @xmath10-cost .
the proof follows from claims [ log - lemma ] and [ claim : radius ] .
this completes the proof of performance guarantee with respect to the @xmath10-cost .
we now proceed to prove the performance guarantee with respect to the @xmath9-costs .
we first recall the following pairing lemma .
@xcite [ pairing - claim ] let @xmath59 be an edge - weighted tree with an even number of marked nodes .
then there is a pairing @xmath176 , @xmath177 , @xmath178 of the marked nodes such that the @xmath179 paths in @xmath59 are edge - disjoint .
* * * * * a pairing of the marked nodes that minimizes the sum of the lengths of the tree - paths between the nodes paired up can be seen to obey the property in the claim above .
* * * * * * * * * * * * * [ cost - lemma ] let @xmath180 be any minimum @xmath9-cost diameter-@xmath70 bounded steiner tree and let @xmath89 denote its @xmath9-cost .
the weight of the largest cardinality minimum - weight matching found in step [ dcst - match ] in each phase @xmath84 of algorithm dcst is at most @xmath181 .
consider phase @xmath84 of algorithm dcst .
note that since the centers at stage @xmath84 are a subset of the nodes in the first iteration , the centers @xmath51 are terminal nodes .
thus they belong to @xmath180 .
mark those vertices in @xmath180 that correspond to the matched vertices , @xmath182 , of @xmath163 in step [ dcst - minwtmatching ] .
then by claim [ pairing - claim ] there exists a pairing of the marked vertices , say @xmath183 , and a set of edge - disjoint paths in opt between these pairs .
since these paths are edge - disjoint their total @xmath9-cost is at most @xmath89 .
further these paths have diameter at most @xmath70 under the @xmath10-cost .
hence the sum of the weights of the edges @xmath184 in @xmath163 , which forms a perfect matching on the set of matched vertices , is at most @xmath185 .
but in step [ dcst - minwtmatching ] of algorithm dcst , a minimum weight perfect matching in the graph @xmath163 was found .
hence the weight of the matching found in step [ dcst - match ] in phase @xmath84 of algorithm dcst is at most @xmath186 .
[ ccostlemma ] let @xmath180 be any minimum @xmath9-cost diameter-@xmath70 bounded steiner tree and let @xmath89 denote its @xmath9-cost .
algorithm dcst outputs a steiner tree with total @xmath9-cost at most @xmath187 .
from claim [ cost - lemma ] we have that the @xmath9-cost of the set of paths added in step [ dcst - match ] of any phase is at most @xmath188 . by claim [ log - lemma ] there are a total of @xmath159 phases and hence the steiner tree output by algorithm dcst has total @xmath9-cost at most @xmath189 . from lemmas [ dcostlemma ] and [ ccostlemma ] we have the following theorem . [ spanning - thm ] there is a strongly polynomial - time algorithm that , given an undirected graph @xmath4 , with two cost functions @xmath9 and @xmath10 defined on the set of edges , diameter bound @xmath70 , terminal set @xmath190 and a fixed @xmath33 , constructs a steiner tree of @xmath45 of diameter at most @xmath191 under the @xmath10-costs and of total @xmath9-cost at most @xmath192 times that of the minimum-@xmath9-cost of any steiner tree with diameter at most @xmath70 under @xmath10 .
in this section we consider the class of treewidth - bounded graphs and give algorithms with improved time bounds and performance guarantees for several bicriteria problems mentioned earlier .
we do this in two steps .
first we develop pseudopolynomial - time algorithms based on dynamic programming .
we then present a general method for deriving fully polynomial - time approximation schemes ( @xmath18 ) from the pseudopolynomial - time algorithms .
we also demonstrate an application of the above results to the minimum broadcast time problem . a class of treewidth - bounded graphs can be specified using a finite number of primitive graphs and a finite collection of binary composition rules .
we use this characterization for proving our results . a class of treewidth - bounded graphs @xmath193 is inductively defined as follows @xcite . 1
the number of primitive graphs in @xmath193 is finite .
each graph in @xmath193 has an ordered set of special nodes called * terminals*. the number of terminals in each graph is bounded by a constant , say @xmath31 .
3 . there is a finite collection of binary composition rules that operate only at terminals , either by identifying two terminals or adding an edge between terminals .
a composition rule also determines the terminals of the resulting graph , which must be a subset of the terminals of the two graphs being composed .
[ decomp - thm ] every problem in table 2 can be solved exactly in @xmath194-time for any class of treewidth bounded graphs with no more than @xmath31 terminals , for fixed @xmath31 and a budget @xmath83 on the first objective .
the above theorem states that there exist pseudopolynomial - time algorithms for all the bicriteria problems from table 2 when restricted to the class of treewidth - bounded graphs .
the basic idea is to employ a dynamic programming strategy .
in fact , this dynamic programming strategy ( in conjunction with theorem [ equiv - thm ] ) yields polynomial - time ( not just pseudopolynomial - time ) algorithms whenever one of the criteria is the degree .
we illustrate this strategy by presenting in some detail the algorithm for the diameter - bounded minimum cost spanning tree problem .
[ thm : dmsttwdth ] for any class of treewidth - bounded graphs with no more than @xmath31 terminals , there is an @xmath195-time algorithm for solving the diameter @xmath8-bounded minimum @xmath9-cost spanning tree problem .
let @xmath10 be the cost function on the edges for the first objective ( diameter ) and @xmath9 , the cost function for the second objective ( total cost ) .
let @xmath193 be any class of decomposable graphs .
let the maximum number of terminals associated with any graph @xmath45 in @xmath193 be @xmath31 .
following @xcite , it is assumed that a given graph @xmath45 is accompanied by a parse tree specifying how @xmath45 is constructed using the rules and that the size of the parse tree is linear in the number of nodes .
let @xmath196 be a partition of the terminals of @xmath45 .
for every terminal @xmath84 let @xmath197 be a number in @xmath198 . for every pair of terminals
@xmath84 and @xmath199 in the same block of the partition @xmath196 let @xmath200 be a number in @xmath201 .
corresponding to every partition @xmath196 , set @xmath202 and set @xmath203 we associate a cost for @xmath45 defined as follows : + @xmath204 = minimum total cost under the @xmath9 function of any forest containing + a tree for each block of @xmath196 , such that the terminal nodes + occurring in each tree are exactly the members of the corresponding + block of @xmath196 , no pair of trees is connected , every vertex in @xmath45 + appears in exactly one tree , @xmath197 is an upper bound on the maximum + distance ( under the @xmath10-function ) from @xmath84 to any vertex in the same + tree and @xmath200 is an upper bound the distance ( under the @xmath10-function ) + between terminals @xmath84 and @xmath199 in their tree . for the above defined cost
, if there is no forest satisfying the required conditions the value of @xmath205 is defined to be @xmath206 .
note that the number of cost values associated with any graph in @xmath193 is @xmath207 .
we now show how the cost values can be computed in a bottom - up manner given the parse tree for @xmath45 . to begin with ,
since @xmath193 is fixed , the number of primitive graphs is finite . for a primitive graph ,
each cost value can be computed in constant time , since the number of forests to be examined is fixed .
now consider computing the cost values for a graph @xmath45 constructed from subgraphs @xmath208 and @xmath209 , where the cost values for @xmath208 and @xmath209 have already been computed .
notice that any forest realizing a particular cost value for @xmath45 decomposes into two forests , one for @xmath208 and one for @xmath209 with some cost values .
since we have maintained the best cost values for all possibilities for @xmath208 and @xmath209 , we can reconstruct for each partition of the terminals of @xmath45 the forest that has minimum cost value among all the forests for this partition obeying the diameter constraints .
we can do this in time independent of the sizes of @xmath208 and @xmath209 because they interact only at the terminals to form @xmath45 , and we have maintained all relevant information . hence we can generate all possible cost values for @xmath45 by considering combinations of all relevant pairs of cost values for @xmath208 and @xmath209 .
this takes time @xmath210 per combination for a total time of @xmath211 . as in @xcite
, we assume that the size of the given parse tree for @xmath45 is @xmath212 .
thus the dynamic programming algorithm takes time @xmath213 .
this completes the proof .
the pseudopolynomial - time algorithms described in the previous section can be used to design fully polynomial - time approximation schemes ( @xmath18 ) for these same problems for the class of treewidth - bounded graphs .
we illustrate our ideas once again by devising an @xmath18 for the ( diameter , total cost , spanning tree)-bicriteria problem for the class of treewidth - bounded graphs .
the basic technique underlying our algorithm , algorithm fpas - dcst , is approximate binary search using rounding and scaling - a method similar to that used by hassin @xcite and warburton @xcite . as in the previous subsection ,
let @xmath45 be a treewidth - bounded graph with two ( integral ) edge - cost functions @xmath9 and @xmath10 .
let @xmath70 be a bound on the diameter under the @xmath10-cost .
let @xmath36 be an accuracy parameter .
without loss of generality we assume that @xmath214 is an integer .
we also assume that there exists a @xmath70-bounded spanning tree in @xmath45 .
let @xmath180 be any minimum @xmath9-cost diameter @xmath70-bounded spanning tree and let @xmath89 denote its @xmath9-cost .
let @xmath215@xmath216 be a pseudopolynomial time algorithm for the ( total cost , diameter , spanning tree ) problem on treewidth - bounded graphs ; i.e. , @xmath215 outputs a minimum diameter spanning tree of @xmath45 with total cost at most @xmath173 ( under the @xmath9-costs ) .
let the running time of @xmath215 be @xmath217 for some polynomial @xmath218 . for carrying out our approximate binary search we need a testing procedure procedure test(v ) which we detail below
: we now prove that procedure test(@xmath219 ) has the properties we need to do a binary search .
[ claim : testerhilo ] if @xmath220 then procedure test(@xmath219 ) outputs low . and , if @xmath221 then procedure test(@xmath219 ) outputs high .
if @xmath222 then since @xmath223 therefore procedure test(@xmath224 outputs low .
let @xmath225 be the @xmath9-cost of any diameter @xmath70 bounded spanning tree .
then we have @xmath226 . if @xmath221 then since @xmath227 therefore procedure test(@xmath219 ) outputs high .
[ claim : testerpoly ] the running time of procedure test(@xmath219 ) is @xmath228 .
procedure test(@xmath219 ) invokes @xmath215 only @xmath229 times . and
each time the budget @xmath173 is bounded by @xmath229 , hence the running time of procedure test(@xmath219 ) is @xmath228 .
we are ready to describe algorithm fpas - dcst - which uses procedure test(@xmath219 ) to do an approximate binary search .
[ correct ] if @xmath45 contains a @xmath70-bounded spanning tree then algorithm fpas - dcst outputs a spanning tree with diameter at most @xmath70 under the @xmath10-cost and with @xmath9-cost at most @xmath230 .
it follows easily from claim [ claim : testerhilo ] that the loop in step [ fpas - stepbound ] of algorithm fpas - dcst executes @xmath231 times before exiting with @xmath232 . since @xmath233
we get that step [ fpas - stepout ] of algorithm fpas - dcst definitely outputs a spanning tree . let @xmath234 be the tree output .
then we have that @xmath235 but since step [ fpas - stepout ] of algorithm fpas - dcst outputs the spanning tree with minimum @xmath9-cost we have that @xmath236 therefore @xmath237 this proves the claim .
[ runtime ] the running time of algorithm fpas - dcst is @xmath238 .
from claim [ claim : testerpoly ] we see that step [ fpas - stepbound ] of algorithm fpas - dcst takes time @xmath238 while step [ fpas - stepout ] takes time @xmath239 .
hence the running time of algorithm fpas - dcst is @xmath238 .
lemmas [ runtime ] and [ correct ] yield : for the class of treewidth - bounded graphs , there is an @xmath18 for the ( diameter , total cost , spanning tree)-bicriteria problem with performance guarantee @xmath240 .
as mentioned before , similar theorems hold for the other problems in table 2 and all these results extend directly to steiner trees . the polynomial - time algorithm for the ( degree , diameter , spanning tree)-bicriteria problem for treewidth - bounded graphs
can be used in conjunction with the ideas presented in @xcite to obtain near - optimal broadcast schemes for the class of treewidth - bounded graphs .
as mentioned earlier , these results generalize and improve the results of kortsarz and peleg @xcite . given an unweighted graph @xmath45 and a root @xmath241 , a _ broadcast scheme _ is a method for communicating a message from @xmath241 to all the nodes of @xmath45 .
we consider a telephone model in which the messages are transmitted synchronously and at each time step , any node can either transmit or receive a message from at most one of its neighbors .
minimum broadcast time problem _ is to compute a scheme that completes in the minimum number of time steps .
let @xmath242 denote the minimum broadcast time from root @xmath241 and let @xmath243 denote the minimum broadcast time for the graph from any root .
the problem of computing @xmath242 - the _ minimum rooted broadcast time problem _ - and that of computing @xmath244 - the _ minimum broadcast time problem _ are both @xmath0-complete for general graphs @xcite .
it is easy to see that any approximation algorithm for the minimum rooted broadcast time problem automatically yields an approximation algorithm for the minimum broadcast time problem with the same performance guarantee .
we refer the readers to @xcite for more details on this problem . combining our approximation algorithm for ( diameter ,
total cost , spanning tree)-bicriteria problem with performance guarantee @xmath240 for the class of treewidth bounded graphs with the observations in @xcite yields the following theorem .
* * * * * * * * * let @xmath245 denote the set of all spanning trees of graph @xmath45
. for spanning tree @xmath246 let @xmath247 denote the maximum , over all nodes , of the degree of the node in @xmath59 .
let the @xmath241-eccentricity of @xmath59 , @xmath248 , denote the maximum , over all nodes , of the distance of the node from @xmath241 in @xmath59 .
we generalize the definition of the _ poise _ of a graph from @xcite to that of the _ rooted poise _ of a graph : @xmath249 [ th : poise_log ] @xcite @xmath250 the proof of this theorem is constructive and implies that a good solution to the rooted poise of a graph can be used to construct a good solution for the minimum rooted broadcast time problem with an @xmath251-factor overhead . [
th : poise_approx ] given a treewidth bounded graph @xmath45 and root @xmath241 , in polynomial time a spanning tree @xmath59 can be found such that @xmath252 . since @xmath45 is an unweighted treewidth - bounded graph we can obtain a ( strongly ) polynomial - time algorithm to solve the ( degree , @xmath241-eccentricity , spanning tree)-bicriteria problem on @xmath45 , using the ideas outlined in subsection [ exact - algos ] .
run this algorithm with all possible degree bounds and choose the tree @xmath59 with the least value of @xmath253 . from theorem
[ th : poise_log ] and lemma [ th : poise_approx ] we get the following theorem * * * * * * * * * * * * * [ th : broadcast ] for any class of treewidth - bounded graphs there is a polynomial - time @xmath254-approximation algorithm for the minimum rooted broadcast time problem and the minimum broadcast time problem .
we have obtained the first polynomial - time approximation algorithms for a large class of bicriteria network design problems . the objective function we considered were ( i ) degree , ( ii ) diameter and ( iii ) total cost .
the connectivity requirements considered were spanning trees , steiner trees and ( in several cases ) generalized steiner trees .
our results were based on the following three ideas : 1 . a binary search method to convert an @xmath255-approximation algorithm for ( @xmath1 , @xmath2 , @xmath3)-bicriteria problems to a @xmath256-approximation algorithm for ( @xmath2 , @xmath1 , @xmath3)-bicriteria problems .
2 . a parametric search technique to devise approximation algorithms for ( @xmath1,@xmath1,@xmath3)-bicriteria problems .
we note that theorem [ better - scale - thm ] is very general .
given _ any _ @xmath20-approximation algorithm for minimizing the objective @xmath1 in the subgraph - class @xmath3 , theorem [ better - scale - thm ] allows us to produce a @xmath257-approximation algorithm for the ( @xmath1 , @xmath1 , @xmath3)-bicriteria problem .
3 . a cluster based approach for devising approximation algorithms for certain categories of ( @xmath1,@xmath2,@xmath3)-bicriteria problems . during the time when this paper was under review
, important progress has been made in improving some of the results in this paper .
recently , ravi and goemans @xcite have devised a @xmath37 approximation scheme for the ( total cost , total cost , spanning tree ) problem .
their approach does not seem to extend to devising approximation algorithms for more general subgraphs considered here . in @xcite , kortsarz and peleg consider the ( diameter , total cost , steiner tree ) problem .
they provide polynomial time approximation algorithms for this problem with performance guarantees @xmath258 for constant diameter bound @xmath70 and @xmath259 for any fixed @xmath260 for general diameter bounds . in @xcite , naor and schieber provide an elegant @xmath258 approximation algorithm for the ( diameter , total cost , spanning tree ) problem .
it is not clear how to extend their algorithms to the steiner tree case . improving the performance guarantees for one or more of the problems considered here
remains an interesting direction for future research . 1 .
devise algorithms that improve upon the performance guarantees presented in this paper . as a step in this direction ,
recently ravi and goemans @xcite have devised a @xmath37 approximation scheme for the ( total cost , total cost , spanning tree ) problem .
2 . devise approximation algorithms for the ( diameter , total cost , generalized steiner tree ) problem . for more information on the generalized steiner tree problem
look at @xcite and the references therein .
find good broadcast schemes which minimize the total cost of the spanning tree . in our language
this problem translates to finding an approximation algorithm for the ( degree , diameter , total cost , spanning tree ) problem .
* acknowledgements : * we would like to thank an anonymous referee for several useful comments and suggestions .
we thank sven krumke ( university of wrzberg ) for reading the paper carefully and providing several useful comments . in particular , both pointed an error in the original proof of theorem 5.3 .
we thank professors s. arnborg and h. l. bodlaender for pointing out to us the equivalence between treewidth bounded graphs and decomposable graphs .
we thank a. ramesh for bringing @xcite to our attention .
we also thank dr . v. kompella for making his other papers available to us .
finally , we thank the referees of _
icalp 95 _ for their constructive comments and suggestions . *
* * a. blum , r. ravi and s. vempala , `` a constant - factor approximation algorithm for the @xmath31-mst problem , '' to appear in the _ proceedings of the 28th annual acm symposium on the theory of computation _ ( 1996 ) . * * * * * * * * * * * * * * * * * m. frer and b. raghavachari , `` an @xmath262 approximation algorithm for the minimum degree spanning tree problem , '' _ proceedings of the 28th annual allerton conference on communication , control and computing _ , pp .
274 - 281 ( 1990 ) .
* * * * * * * * * * j. l. ganley , m. j. golin and j. s. salowe , `` the multi - weighted spanning tree problem , '' _ proceedings of the first conference on combinatorics and computing ( cocoon ) _ , springer verlag , lncs pp .
141 - 150 ( 1995 ) .
* * * m. x. goemans , a. v. goldberg , s. plotkin , d. b. shmoys , e. tardos and d. p. williamson , `` improved approximation algorithms for network design problems , '' _ proceedings of the fifth annual acm - siam symposium on discrete algorithms _
223 - 232 ( 1994 ) .
* * * * * * * * * * * * * * v.p .
kompella , j.c .
pasquale and g.c .
polyzos , `` two distributed algorithms for the constrained steiner tree problem , '' technical report cal-1005 - 92 , computer systems laboratory , university of california , san diego ( oct . 1992 ) . * * * * * * * r. ravi , m. v. marathe , s. s. ravi , d. j. rosenkrantz , and h.b . hunt iii , `` many birds with one stone : multi - objective approximation algorithms , '' _ proceedings of the 25th annual acm symposium on the theory of computing _ , pp .
438 - 447 ( 1993 ) .
( expanded version appears as brown university technical report tr - cs-92 - 58 . )
r. raz and s. safra , `` a sub - constant error - probability low - degree test , and a sub - constant error - probability pcp characterization of np , '' _ proc .
29th annual acm symposium on theory of computing _
, 475 - 484 ( 1997 ) . *
* * r. ravi , r. sundaram , m. v. marathe , d. j. rosenkrantz , and s. s. ravi , `` spanning trees short or small , '' in _ proceedings of the 5th annual acm - siam symposium on discrete algorithms _ ,
546 - 555 ( 1994 ) .
journal version to appear in _ siam journal on discrete mathematics_. | we study a general class of bicriteria network design problems .
a generic problem in this class is as follows : given an undirected graph and two minimization objectives ( under different cost functions ) , with a budget specified on the first , find a < subgraph from a given subgraph - class that minimizes the second objective subject to the budget on the first .
we consider three different criteria - the total edge cost , the diameter and the maximum degree of the network . here , we present the first polynomial - time approximation algorithms for a large class of bicriteria network design problems for the above mentioned criteria .
the following general types of results are presented .
first , we develop a framework for bicriteria problems and their approximations .
second , when the two criteria are the same we present a `` black box '' parametric search technique .
this black box takes in as input an ( approximation ) algorithm for the unicriterion situation and generates an approximation algorithm for the bicriteria case with only a constant factor loss in the performance guarantee .
third , when the two criteria are the diameter and the total edge costs we use a cluster - based approach to devise a approximation algorithms the solutions output violate both the criteria by a logarithmic factor .
finally , for the class of treewidth - bounded graphs , we provide pseudopolynomial - time algorithms for a number of bicriteria problems using dynamic programming .
we show how these pseudopolynomial - time algorithms can be converted to fully polynomial - time approximation schemes using a scaling technique . *
ams 1980 subject classification .
* 68r10 , 68q15 , 68q25 + * keywords . *
approximation algorithms , bicriteria problems , spanning trees , + network design , combinatorial algorithms .
+ = 1.2 |
in high - energy processes parton distribution functions ( pdf ) and parton fragmentation functions ( pff ) are used to describe the link between hadrons in initial and final states , respectively , as well as the quark and gluons that are used to describe the processes within the framework of quantum chromodynamics ( qcd ) .
the starting points are forward matrix elements of parton fields , such as the quark - quark correlator @xmath1 where a summation over color indices is understood , i.e. the operator structure @xmath2 is used .
this correlator replaces the fermionic correlator @xmath3 that one is used to for a single incoming fermion .
similarly one has the gluon correlator @xmath4 again with a color summation understood , i.e. @xmath5 is used as the operator structure if one uses for the gluon fields matrices in color triple representation .
besides a single quark , also a quark and a gluon from a particular hadron taking part in the hard process need to be accounted for .
this is included as a multi - parton correlator , such as the quark - quark - gluon correlator @xmath6 the basic idea is to factorize these hadronic correlators ( soft parts ) in a full diagrammatic approach and parametrize them in terms of pdfs .
this requires high energies in which case the momenta of different hadrons obey @xmath7 , where @xmath8 is the hard scale in the process .
the hadronic momenta can be treated as light - like vectors @xmath9 and the hard process brings in a conjugate light - like vector @xmath10 such that @xmath11 , for instance @xmath12 . with the light - like vectors one makes a sudakov expansion of the parton momenta , @xmath13 with @xmath14 . in any contraction with vectors outside the correlator
, the component @xmath15 contributes at order @xmath16 , the transverse component at order @xmath17 and the remaining component contributes at order @xmath18 .
this allows consecutive integration of the components to obtain from the fully unintegrated correlator in eq .
( [ e : phi - basic ] ) the tmd light - front ( lf ) correlator @xmath19 the collinear light - cone ( lc ) correlator @xmath20 or the local matrix element @xmath21 the importance of integrating at least the light - cone ( minus ) component @xmath22 is that the expression is at equal time , i.e. time - ordering is not relevant anymore for tmd or collinear pdfs.@xcite for local matrix elements one can calculate the anomalous dimensions , which show up as the mellin moments of the splitting functions that govern the scaling behavior of the collinear correlator @xmath23 .
we note that the collinear correlator is not simply an integrated tmd .
the dependence on an upper limit @xmath24 is found from the anomalous dimensions ( splitting functions ) .
one has an @xmath25 behavior of tmds that is calculable using collinear tmds and which matches to the intrinsic nonperturbative @xmath0-behavior.@xcite we note that in operator product expansion language , the collinear correlators involve operators of definite twist , while tmd correlators involve operators of various twist .
in order to determine the importance of a particular correlator in a hard process , one can do a dimensional analysis to find out when they contribute in an expansion in the inverse hard scale .
dominant are the ones with lowest canonical dimension obtained by maximizing contractions with @xmath10 , for instance for quark or gluon fields the minimal canonical dimensions dim[@xmath26 = dim[@xmath27
= 2 , while an example for a multi - parton combination gives dim[@xmath28 = 3 .
equivalently , one can maximize the number of @xmath9 s in the parametrization of @xmath29 .
of course one immediately sees that any number of collinear @xmath30 fields does nt matter .
furthermore one must take care of color gauge invariance , for instance when dealing with the gluon fields and one must include derivatives in color gauge - invariant combinations .
with dimension zero there is @xmath31 and with dimension one there is @xmath32 .
the color gauge - invariant expressions for quark and gluon distribution functions actually include gauge link operators , @xmath33 } = { \cal p}\exp\left(-i\int_0^{\xi } d\zeta_\mu a^\mu(\zeta)\right),\ ] ] connecting the nonlocal fields , @xmath34}(x , p_{{\scriptscriptstyle t}};n ) = \int \frac{d\,\xi{\cdot } p\,d^2\xi_{{\scriptscriptstyle t}}}{(2\pi)^3}\ e^{i\,p\cdot \xi } \
\langle p\vert \overline\psi_j(0)\,u_{[0,\xi]}\,\psi_i(\xi)\vert p\rangle \biggr|_{\textrm{\scriptsize{lf } } } , \label{e : qmatrix } \\ & & \gamma^{[u , u^\prime]\,\mu\nu}(x , p_{{\scriptscriptstyle t}};n ) = { \int}\frac{d\,\xi{\cdot}p\,d^2\xi_{{\scriptscriptstyle t}}}{(2\pi)^3}\ e^{ip\cdot\xi } \nonumber \\ & & \mbox{}\qquad\qquad\qquad \times \operatorname{tr}\,\langle p{,}s|\,f^{n\mu}(0)\ , u_{[0,\xi]}^{\phantom{\prime}}\ , f^{n\nu}(\xi)\,u_{[\xi,0]}^\prime\ , \label{e : matrix}\end{aligned}\ ] ] for transverse separations , the gauge links involve paths running along the minus direction to @xmath35 ( dimensionally preferred ) , which are closed with one or more transverse pieces at light - cone infinity .
the two simplest possibilities are @xmath36}^{[\pm]}$ ] = @xmath37}\,u^t_{[0_{{\scriptscriptstyle t}},\xi_{{\scriptscriptstyle t}}]}\,u^n_{[\pm\infty,\xi]}$ ] , leading to gauge link dependent quark tmds @xmath38}(x , p_{{\scriptscriptstyle t}})$].@xcite for gluons , the correlator involves color gauge - invariant traces of field - operators @xmath39 , requiring the inclusion of _ two _ gauge links @xmath36}$ ] and @xmath40}^\prime$ ] . again
the simplest possibilities are the past- and future - pointing gauge links @xmath41}$ ] , giving even in the simplest case four gluon tmds @xmath42}(x , p_{{\scriptscriptstyle t}})$ ] . in general , many gauge link structures are possible , connecting the positions @xmath43 and @xmath44 in different ways .
furthermore , there are also contributions containing wilson loops , given by @xmath45}=u_{[0,\xi]}^{[+]}u_{[\xi,0]}^{[-]}$ ] = @xmath36}^{[+]}u_{[0,\xi]}^{[-]\dagger}$ ] or @xmath45\dagger}$ ] = @xmath36}^{[-]}u_{[\xi,0]}^{[+]}$ ] = @xmath36}^{[-]}u_{[0,\xi]}^{[+]\dagger}$ ] .
a detailed list of useful gauge link structures can be found in ref .
. + ( a ) ( b ) + + ( c ) ( d ) + + ( e ) ( f ) for quark correlators there are two types of gauge link structures , @xmath46}\big\rgroup , \label{e : qtype1 } \\
\text{type 2:}&&\hspace{5mm}\textrm{tr}_c \big\lgroup \psi_i(\xi)\overline \psi_j(0 ) \,u_{[0,\xi]}\big\rgroup\,\frac{1}{n_c}\textrm{tr}_c \big\lgroup u^{[\text{loop}]}\big\rgroup .
\label{e : qtype2}\end{aligned}\ ] ] for gluon correlators there are three types of gauge link structures , namely @xmath47}^{\phantom{\prime}}\,f^{n\nu}(\xi)\,u_{[\xi,0]}^\prime\big\rgroup , \label{e : type1 } \\ \text{type 2:}&&\hspace{5mm}\textrm{tr}_c \big\lgroup f^{n\mu}(0)\,u_{[0,\xi]}^{\phantom{\prime}}\,f^{n\nu}(\xi)\,u_{[\xi,0]}^\prime\big\rgroup\,\frac{1}{n_c}\textrm{tr}_c \big\lgroup u^{[\text{loop}]}\big\rgroup , \label{e : type2 } \\ \text{type
3:}&&\hspace{5mm}\frac{1}{n_c}\textrm{tr}_c \big\lgroup f^{n\mu}(0)\,u^{[\text{loop}]}\big\rgroup\,\textrm{tr}_c \big\lgroup f^{n\nu}(\xi)\,u^{[\text{loop}^{\prime}]}\big\rgroup , \label{e : type3}\end{aligned}\ ] ] examples of which are given in fig .
[ f : gl ] .
which gauge link appears for a particular diagram has been calculated explicitly in ref . , but it is also closely related to the color flow in the process under consideration .
when looking at the hard scattering amplitudes , colored particles in the final state typically contribute to gauge links going through plus light - cone infinity , whereas colored particles in the initial state typically contribute to gauge links going through minus light - cone infinity .
an example of the latter is the @xmath48 production through gluon fusion , illustrated in fig .
[ f : fd]a , having a @xmath49 $ ] structure , since the only colored particles are in the initial state . for illustrative purposes , some color flow structures of diagrams in fig .
[ f : fd ] have been illustrated in fig .
[ f : colorflow ] .
we note that in the case of diffractive processes one has operator structures of yet another kind , @xmath50}\big\rgroup , \label{e : type0 } \\\ ] ] which could play a role in diffractive scattering.@xcite traced wilson loops typically appear whenever a hard scattering diagram contains contributions where it is possible to draw a closed color loop .
an example is fig .
[ f : fd]b , which receives a gauge link contribution with a traced wilson loop @xmath51 $ ] .
note that such diagrams typically contain multiple gauge link contributions , since multiple color flow structures are required in the full description . in fig .
[ f : fd]c the diagram even receives a contribution containing two separate color traces of the form @xmath52 $ ] .
the third type of gauge link structures appears whenever there exists a color flow structure which is singlet at the cut , which is e.g. one of the contributions for the diagram in fig .
[ f : fd]d .
the correlators in the eqs .
( [ e : qmatrix ] ) and ( [ e : matrix ] ) including a gauge link can be parametrized in terms of tmd pdfs@xcite depending on @xmath53 and @xmath54 , @xmath55}(x , p_{{{\scriptscriptstyle t}}};n ) = \bigg\ { f^{[u]}_{1}(x , p_{{\scriptscriptstyle t}}^2 ) -f_{1t}^{\perp[u]}(x , p_{{\scriptscriptstyle t}}^2)\ , \frac{\epsilon_{{{\scriptscriptstyle t}}}^{p_{{{\scriptscriptstyle t}}}s_{{{\scriptscriptstyle t}}}}}{m } + g^{[u]}_{1s}(x , p_{{\scriptscriptstyle t}})\gamma_{5 } \nonumber \\&&\mbox{}\qquad + h^{[u]}_{1t}(x , p_{{\scriptscriptstyle t}}^2)\,\gamma_5\,\slashed{s}_{{{\scriptscriptstyle t } } } + h_{1s}^{\perp [ u]}(x , p_{{\scriptscriptstyle t}})\,\frac{\gamma_5\,\slashed{p}_{{{\scriptscriptstyle t}}}}{m } + ih_{1}^{\perp [ u]}(x , p_{{\scriptscriptstyle t}}^2)\,\frac{\slashed{p}_{{{\scriptscriptstyle t}}}}{m } \bigg\}\frac{\slashed{p}}{2 } , \label{e : par}\end{aligned}\ ] ] with the spin vector parametrized as @xmath56 and shorthand notations for @xmath57}_{1s}$ ] and @xmath58}$ ] , @xmath59}_{1s}(x , p_{{\scriptscriptstyle t}})=s_{{{\scriptscriptstyle l } } } g^{[u]}_{1l}(x , p_{{{\scriptscriptstyle t}}}^2 ) -\frac{p_{{{\scriptscriptstyle t}}}\cdot s_{{{\scriptscriptstyle t}}}}{m}g^{[u]}_{1t}(x , p_{{{\scriptscriptstyle t}}}^2).\ ] ] for quarks , these include not only the functions that survive upon @xmath0-integration , @xmath60 , @xmath61 and @xmath62 , which are the well - known collinear spin - spin densities ( involving quark and nucleon spin ) but also momentum - spin densities such as the sivers function @xmath63 ( unpolarized quarks in a transversely polarized nucleon ) and spin - spin - momentum densities such as @xmath64 ( longitudinally polarized quarks in a transversely polarized nucleon ) .
for the correlator in eq .
( [ e : matrix ] ) the expansion in transverse momentum dependent distribution functions ( tmds ) at the level of leading twist is given by@xcite @xmath65}(x{,}p_{{\scriptscriptstyle t } } ) = -g_t^{\mu\nu}\,f_1^{g [ u]}(x{,}p_{{\scriptscriptstyle t}}^2 ) + g_t^{\mu\nu}\frac{\epsilon_t^{p_ts_t}}{m } \,f_{1t}^{\perp g[u]}(x{,}p_{{\scriptscriptstyle t}}^2 ) \nonumber\\&&\mbox{}\qquad + i\epsilon_t^{\mu\nu}\;g_{1s}^{g [ u]}(x{,}p_{{\scriptscriptstyle t } } ) + \bigg(\frac{p_t^\mu p_t^\nu}{m^2}\ , { -}\,g_t^{\mu\nu}\frac{p_{{\scriptscriptstyle t}}^2}{2m^2}\bigg)\;h_1^{\perp g [ u]}(x{,}p_{{\scriptscriptstyle t}}^2 ) \nonumber\\ & & \mbox{}\qquad -\frac{\epsilon_t^{p_t\{\mu}p_t^{\nu\}}}{2m^2}\ ; h_{1s}^{\perp g [ u]}(x{,}p_{{\scriptscriptstyle t } } ) -\frac{\epsilon_t^{p_t\{\mu}s_t^{\nu\ } } { + } \epsilon_t^{s_t\{\mu}p_t^{\nu\}}}{4m}\ ; h_{1t}^{g[u]}(x{,}p_{{\scriptscriptstyle t}}^2 ) .
\label{e : gluoncorr}\end{aligned}\ ] ] in these parametrizations , the process dependence due to the gauge links is absorbed in nonuniversal , gauge link dependent , tmds .
+ ( a ) ( b ) ( c ) ( d ) + ( a ) ( b )
for analyzing the process dependent behavior of the tmds , a formalism for probing these functions has to be used .
we will focus on gluons . for analyzing the structure of a single hadron correlator , we use weighting with transverse momenta , defined as @xmath66}(x)\equiv \int d^2p_{\scriptscriptstyle t}\ p_{\scriptscriptstyle t}^{\alpha_1}\ldots p_{\scriptscriptstyle t}^{\alpha_n}\,\gamma^{[u]}(x , p_{\scriptscriptstyle t}),\ ] ] where we weight the correlator with one or more factors of @xmath67 .
after a fourier transformation , these transverse momenta become derivatives in coordinate space , acting on all objects in the correlator depending on the coordinate @xmath68 , gauge links included.@xcite rather than giving the details for a transverse weighting for multiple factors of @xmath67 , we demonstrate the formalism using the single weighted case as an example , which has been described earlier in the refs . and , @xmath69}(x)\equiv\int d^2p_{\scriptscriptstyle t}\ p_{\scriptscriptstyle t}^{\alpha}\,\gamma^{[u]}(x , p_{\scriptscriptstyle t})=\widetilde\gamma_{\partial}^{\alpha}(x)+c_{g,1}^{[u]}\,\gamma_{g,1}^{\alpha}(x)+c_{g,2}^{[u]}\,\gamma_{g,2}^{\alpha}(x ) .
\label{e : moment1}\ ] ] we indicate the partonic operator structure in the correlators with the indices @xmath70 and @xmath71 , the latter of which stands for the gluonic pole contributions and are multiplied with gluonic pole coefficients .
these coefficients contain all the process dependence and have been tabulated , see e.g. the refs . and .
the basis equations on which the correlators in eq .
( [ e : moment1 ] ) are based are @xmath72}(x , x - x_1)=\int \frac{d\,\xi{\cdot}p}{2\pi}\frac{d\,\eta{\cdot}p}{2\pi}\ e^{ix_1 ( \eta\cdot p)}e^{i ( x - x_1)(\xi \cdot p ) } \nonumber \\ & & \hspace{23 mm } \times\textrm{tr}\langle p , s|f^{n\mu}(0)\big[u_{[0,\eta]}^{[n ] } id_{\scriptscriptstyle t}^{\alpha}(\eta)u_{[\eta,0]}^{[n]},u_{[0,\xi]}^{[n]}f^{n\nu}(\xi)u_{[\xi,0]}^{[n]}\big]|p , s\rangle\big|_{\textrm{\scriptsize{lc } } } , \nonumber \\ \label{e : gammad } \\ & & \gamma_{f,1}^{\mu\nu,\alpha[u]}(x , x - x_1)=\int \frac{d\,\xi{\cdot}p}{2\pi}\frac{d\,\eta{\cdot}p}{2\pi}\ e^{ix_1 ( \eta\cdot p)}e^{i ( x - x_1)(\xi \cdot p ) } \nonumber \\ & & \hspace{23 mm } \times\textrm{tr}\langle p , s|f^{n\mu}(0)\big[u_{[0,\eta]}^{[n ] } f^{n\alpha}(\eta)u_{[\eta,0]}^{[n]},u_{[0,\xi]}^{[n]}f^{n\nu}(\xi)u_{[\xi,0]}^{[n]}\big]|p , s\rangle\big|_{\textrm{\scriptsize{lc } } } , \nonumber \\ \label{e : gammaf1 } \end{aligned}\ ] ] where a third correlator , @xmath73}(x , x - x_1)$ ] , similar to @xmath74}(x , x - x_1)$ ] but with an anticommutator of gluonic fields instead rather than a commutator of them , has been omitted .
the correlators in eq .
( [ e : moment1 ] ) relate to these basis correlators through the relations @xmath75 the correlators with an index @xmath71 are the gluonic pole contributions , which correspond to multi - parton correlators with zero momentum gluons , see e.g. the refs .
, , , , and . in this , the index @xmath76 labels the different color configurations of the partonic operators .
the correlator with an index @xmath70 corresponds to a commutator of the partonic operator combination @xmath77 with the gluon field @xmath78 .
transverse weightings can not only be performed at the level of the matrix elements .
the expression involving the parametrization of tmds can be investigated in this way as well , using the definition @xmath79 in eq .
( [ e : tm ] ) , the label @xmath80 represents the number of transverse weightings .
see ref .
for the details of extending this mechanism to higher transverse weightings .
gluonic poles are t - odd , so correlators with an odd number of them are t - odd , while correlators containing an even number of gluonic poles or no gluonic poles at all are t - even .
by comparing the weighted expressions for operator matrix elements and those of the parametrization , one can identify transverse moments with specific operators .
this identification in the previous section can be used to find universal tmd functions . for this
we expand the process dependent gluon correlator in terms of definite rank functions depending on @xmath53 and @xmath54 , @xmath81}(x , p_{\scriptscriptstyle t } ) & \ = \ & \gamma(x , p_{\scriptscriptstyle t}^2 ) + \frac{p_{{\scriptscriptstyle t } i}}{m}\,\widetilde\gamma_\partial^{i}(x , p_{\scriptscriptstyle t}^2 ) + \frac{p_{{\scriptscriptstyle t } ij}}{m^2}\,\widetilde\gamma_{\partial\partial}^{ij}(x , p_{\scriptscriptstyle t}^2 ) + \frac{p_{{\scriptscriptstyle t } ijk}}{m^3}\,\widetilde\gamma_{\partial\partial\partial}^{\,ijk}(x , p_{\scriptscriptstyle t}^2 ) + \ldots \nonumber \\
& \quad + & \sum_c c_{g , c}^{[u]}\bigg(\frac{p_{{\scriptscriptstyle t } i}}{m}\,\gamma_{g , c}^{i}(x , p_{\scriptscriptstyle t}^2 ) + \frac{p_{{\scriptscriptstyle t } ij}}{m^2}\,\widetilde\gamma_{\{\partial g\},c}^{\,ij}(x , p_{\scriptscriptstyle t}^2 ) \nonumber \\ & & \hspace{20mm}+ \frac{p_{{\scriptscriptstyle t } ijk}}{m^3}\,\widetilde\gamma_{\{\partial\partial g\},c}^{\,ijk}(x , p_{\scriptscriptstyle t}^2 ) + \ldots\bigg ) \nonumber \\ & \quad + & \sum_c c_{gg , c}^{[u]}\left\lgroup\frac{p_{{\scriptscriptstyle t } ij}}{m^2}\,\gamma_{gg , c}^{ij}(x , p_{\scriptscriptstyle t}^2 ) + \frac{p_{{\scriptscriptstyle t } ijk}}{m^3}\,\widetilde\gamma_{\{\partial gg\},c}^{\,ijk}(x , p_{\scriptscriptstyle t}^2)+ \ldots\right\rgroup \nonumber \\ & \quad + & \sum_c c_{ggg , c}^{[u]}\left\lgroup\frac{p_{{\scriptscriptstyle t } ijk}}{m^3}\,\gamma_{ggg , c}^{ijk}(x , p_{\scriptscriptstyle t}^2 ) + \ldots \right\rgroup + \ldots \ , .
\label{e : tmdstructure}\end{aligned}\ ] ] in this expression the tmd correlators on the rhs are universal objects , multiplied with process dependent gluonic pole factors . just as for the moments , the functions with an odd number of gluonic poles are t - odd .
the rank of the correlators in eq .
( [ e : tmdstructure ] ) is equal to the rank of the momentum operator in front of it . for each correlator in eq .
( [ e : tmdstructure ] ) it is therefore clear what its rank and behavior under time - reversal symmetry is .
these two properties can be used to make a classification for the tmds in the eqs .
( [ e : par ] ) and ( [ e : gluoncorr ] ) as well , where quark tmds contain contributions up to rank 2 and gluons up to rank 3 . in table
[ t : spinhalfcol-1 ] all correlators are ordered according to their rank and the number of gluonic poles they contain , while in the tables [ t : quarkpdf-1/2 ] and [ t : gluonpdf-1 ] the same is done for the quark and gluon tmds respectively . by comparing these tables an identification can be made , identifying which tmd corresponds to which matrix element .
since it is known how many color structures exist for the matrix elements in table [ t : spinhalfcol-1 ] , it is then known how many different universal tmds exist .
an interesting observation is that not only t - odd functions have a process dependence , the t - even functions @xmath82 for quarks and @xmath83 for gluons do so as well .
one can go one step further , by using eq .
( [ e : tmdstructure ] ) and finding the gauge link dependent tmds in terms of the universal ones . this way , one finds for gluons @xmath84}(x , p_{\scriptscriptstyle t}^2)&=&\sum_{c=1}^2 c_{g , c}^{[u]}\,f_{1t}^{\perp g(ac)}(x , p_{\scriptscriptstyle t}^2 ) , \label{e : f1tperp } \\
h_{1t}^{g[u]}(x , p_{\scriptscriptstyle t}^2)&=&\sum_{c=1}^2 c_{g , c}^{[u]}\,h_{1t}^{g(ac)}(x , p_{\scriptscriptstyle t}^2 ) , \label{e : h1 t } \\
g[u]}(x , p_{\scriptscriptstyle t}^2)&=&\sum_{c=1}^2 c_{g , c}^{[u]}\,h_{1l}^{\perp g(ac)}(x , p_{\scriptscriptstyle t}^2 ) , \label{e : h1l } \\
h_1^{\perp g[u]}(x , p_{\scriptscriptstyle t}^2)&=&h_1^{\perp g ( a)}(x , p_{\scriptscriptstyle t}^2)+\sum_{c=1}^{4}c_{gg , c}^{[u]}\,h_1^{\perp g ( bc)}(x , p_{\scriptscriptstyle t}^2 ) , \label{e : h1perp } \\ h_{1t}^{\perp g[u]}(x , p_{\scriptscriptstyle t}^2)&=&\sum_{c=1}^2 c_{g , c}^{[u]}\,h_{1t}^{\perp g(ac)}(x , p_{\scriptscriptstyle t}^2)+\sum_{c=1}^{7}c_{ggg , c}^{[u]}\,h_{1t}^{\perp g(bc)}(x , p_{\scriptscriptstyle t}^2 ) , \label{e : h1tperp}\end{aligned}\ ] ] where the summations indicate the number of color structures that are possible for the different tmds . in the eqs .
( [ e : f1tperp])-([e : h1tperp ] ) , all functions on the rhs are universal and the gluonic pole factors appearing in these equations have been tabulated in ref . .
several of the quark and gluon tmds as they appear in the expansion of the correlators are process dependent , spoiling universality .
we have shown that they can be written as the sum of a finite number of universal tmds , where the gauge link dependence is isolated in multiplicative gluonic pole factors . for making the identification of which
tmds contain such a process dependence , an expansion of the gluon correlator in terms of transverse moments has been used . by using properties of the elements in this expansion , namely the rank and their behavior under time - reversal symmetry , an identification could be made .
furthermore , we find that also t - even functions can be process dependent . by studying the color flow of hard processes the appropriate gauge link(s ) for a given process can be found.@xcite
this contribution to the conference proceeding combines two separate talks at the qcd evolution workshop , namely `` tmds or definite rank for quarks and gluons '' by pjm and `` universality of gluon tmd distribution functions '' by mgab .
this research is part of the research program of the `` stichting voor fundamenteel onderzoek der materie ( fom ) '' , which is financially supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) '' .
am thanks the alexander von humboldt fellowship for experienced researchers , germany , for support
. also support of the fp7 eu - programmes hadronphysics3 ( contract no 283286 ) and qwork ( contract 320389 ) is acknowledged .
00 m. diehl and t. gousset , _ phys .
b _ * 428 * , 359 ( 1998 ) . j. c. collins and d. e. soper and g. f. sterman , _ nucl .
b _ * 250 * , 199 ( 1985 ) .
a. v. belitsky , x. ji and f. yuan , _ nucl .
b _ * 656 * , 165 ( 2003 ) . c. j. bomhof , p. j. mulders and f. pijlman , _ phys . lett .
b _ * 596 * , 277 ( 2004 )
. m. g. a. buffing , a. mukherjee and p. j. mulders , generalized universality of definite rank gluon transverse momentum dependent correlators , arxiv:1306.5897 [ hep - ph ] . c. j. bomhof , p. j. mulders and f. pijlman , _ eur .
j. c _ * 47 * , 147 ( 2006 ) .
f. dominguez , b - w .
xiao and f. yuan , _ phys .
lett . _ * 106 * , 022301 ( 2011 ) .
f. dominguez , c. marquet , b - w .
xiao and f. yuan , _ phys .
d _ * 83 * , 105005 ( 2011 ) .
a. bacchetta , m. diehl , k. goeke , a. metz , p. j. mulders and m. schlegel , _ jhep _ * 0702 * , 093 ( 2007 ) .
p. j. mulders and r. d. tangerman , _ nucl .
b _ * 461 * , 197 ( 1996 ) [ erratum - ibid .
_ b _ * 484 * , 538 ( 1997 ) ] .
p. j. mulders and j. rodrigues , _ phys .
d _ * 63 * , 094021 ( 2001 ) .
s. meissner , a. metz and k. goeke , _ phys .
d _ * 76 * , 034002 ( 2007 ) .
m. g. a. buffing and p. j. mulders , _ jhep _ * 1107 * , 065 ( 2011 ) .
m. g. a. buffing , a. mukherjee and p. j. mulders , _ phys .
d _ * 86 * , 074030 ( 2012 ) . c. j. bomhof and p. j. mulders , _ nucl .
b _ * 795 * , 409 ( 2008 ) . c. j. bomhof and p. j. mulders , _ jhep _ * 0702 * , 029 ( 2007 ) .
a. v. efremov and o. v. teryaev , _ sov .
phys . _ * 36 * , 140 ( 1982 ) .
a. v. efremov and o. v. teryaev , _ phys .
b _ * 150 * , 383 ( 1985 ) .
qiu and g. f. sterman , _ phys .
lett . _ * 67 * , 2264 ( 1991 ) .
qiu and g. f. sterman , _ nucl .
b _ * 378 * , 52 ( 1992 ) .
qiu and g. f. sterman , _ phys .
d _ * 59 * , 014004 ( 1998 ) .
y. kanazawa and y. koike , _ phys . lett .
b _ * 478 * , 121 ( 2000 ) . | transverse momentum dependent ( tmd ) parton distribution functions ( pdfs ) , in short referred to as tmds , also take into account the transverse momentum ( @xmath0 ) of the partons . just as the @xmath0-integrated analogues we want to link them to quark and gluon matrix elements using operator product expansion methods in qcd , involving operators of definite twist .
the tmds also involve operators of higher twist , which are not suppressed by powers of the hard scale , however . using the expression for tmds involving nonlocal matrix elements of quark and gluon fields
there is a gauge link dependence , which also introduces an inherent process dependence . using transverse moments , which are specific @xmath0-weightings
, we can establish the link with quark and gluon fields including the higher twist ones .
we introduce ( a finite number of ) universal tmds of definite rank and show how the process dependent tmds can be written as combinations of these universal functions . |
the interval of time @xmath0 over which a system moves by an amount of energy @xmath1 has a long history which starts with the time - energy uncertainty relation derived by heisenberg @xcite and was further generalized @xcite .
since then a considerable amount of work has developed from thereon , see @xcite .
it led to the derivation of different approximate estimations of the time related to the so - called quantum speed limit @xmath2 .
the first important contributions concerning the optimisation of the time duration of quantum jumps have been realized by mandelstam and tamm @xcite and followed by several other developments during the @xmath3 @xcite .
the results concerned essentially closed systems .
+ during the following decade up to present days the subject developed further and extended to the case of open systems , see in particular @xcite .
these developments were strongly motivated by the interest in the realization of fast electronic devices in which a given component interacts with its environment . +
a large series of very recent work was devoted to the study of different properties which govern the behaviour of @xmath2 in such systems such as the structure of the system space @xcite , the environment space @xcite , the importance of the interaction coupling with the environment and its optimization @xcite , the non - markovian memory effects @xcite , the initial state of the system @xcite , the role of a thermal environment @xcite , the influence of relativistic effects @xcite .
+ in the present work we aim to consider the behaviour of the time related to the lower bound of @xmath2 and related quantities .
we choose to perform a numerical analysis on a coherent open system described by a spin interacting with a bosonic environment .
we first recall the expression of the time inequality in section 2 and develop the model and the analytic expression of the inequality in the framework of the model in section 3 .
section 4 is devoted to the results . in section 5
we discuss and comment the results .
technical details are shown in the appendices .
we introduce here the generalized energy - time relation to open quantum systems following the developments of deffner and lutz @xcite .
denote by @xmath4 the density operator of an isolated system @xmath5 and by @xmath6 $ ] the density operator of an open quantum system obtained as the trace over the states of an environment @xmath7 coupled to @xmath5 , where @xmath8 is the density operator of the closed system @xmath9 . the system @xmath5 is characterized by its fidelity f(_s^(0)(0),_s(t))=tr_s[(_s^(0)(0))^1/2 _ s(t ) ( _ s^(0)(0))^1/2)^1/2]^2 [ eq1 ] which measures at time @xmath10 the deviation of the open system from its initial state at the initial time at which the system does not interact with an environment .
the deviation can be expressed in terms of an angle @xcite which reads b(_s^(0)(0),_s(t))= ( f(_s^(0)(0),_s(t ) ) [ eq2 ] which is the generalization in the case of mixed states of the angle characterizing the overlap of two states in hilbert space . in the general case the time evolution of the system @xmath5
is given by the master equation d _
s(t)/dt = d_t(_s(t ) ) [ eq3 ] where @xmath11 is a positive operator .
the average energy over a unit of time exchanged between @xmath5 and @xmath7 can be estimated by means of the quantity = _
0^t dt e(t ) [ eq4 ] where @xmath12 is taken as the smallest of the operator , trace or hilbert - schmidt norm of @xmath13 : the @xmath14 $ ] s are the time - dependent eigenvalues of @xmath13 which in practice is the time derivative of the known density operator @xmath15 , eq.(3 ) .
the minimum time corresponding to the quantum speed limit obeys the inequality given in ref .
@xcite : t t_lb(=t_qsl ) [ eq6 ] where the lower bound time is given by t_lb=|(b(_s^(0)(0),_s(t)))-1| [ eq7 ] ( @xmath16 in what follows ) .
+ this time corresponds to a lower limit of the evolution time of the system .
it is the smaller the larger the denominator corresponding to the energy exchange between @xmath5 and @xmath7 and the smaller the numerator , i.e. the larger the overlap beween the initial state and the state of the system at time @xmath10 .
the system @xmath5 is a spin which rotates around its @xmath17-axis and couples to a system of bosons .
the hamiltonian @xmath18 reads h= h_s+h_e+h_se [ eq8 ] with h_s = j_z + h_e = b^+b + h_se=(b^++b ) j^2 [ eq9 ] @xmath19 are boson operators , @xmath20 is the rotation frequency of the system , @xmath21 the quantum of energy of the bosonic oscillators and @xmath22 the strength parameter of the coupling interaction between @xmath5 and @xmath7 . in the present case
@xmath23 commutes with @xmath24 since @xmath25 and @xmath26 commute in the common basis of states @xmath27 $ ] which are the eigenstates of both operators .
it has been shown @xcite that this class of systems does not lead to decoherence . in the subspace of @xmath25
limited to a unique spin state in @xmath28 the projection of @xmath29 on @xmath5 space is diagonal in @xmath30 and reads j_1|h_se|j_2 = j_1(j_1 + 1)(b^++b)_j_1,j_2 [ eq10 ] as a consequence the density operator @xmath15 at time @xmath10 is obtained by taking the trace over the environment states of the total hamiltonian @xmath8 leading to _
s(t)=tr_e(t ) [ eq11 ] whose matrix elements read ^j_1 m_1 , j_2 m_2_s(t)=^j_1 m_1 , j_2 m_2_0(t)_e(j_1,j_2,t ) ( j_1,j_2 ) [ eq12 ] with ^(0 ) j_1 m_1 , j_2 m_2_0(t)= [ eq13 ] where @xmath31 , @xmath32 are the spin projections on the quantization axis @xmath17 . the bosonic environment contribution @xmath33 can be put in the following form _ e(j_1,j_2,;t)=1/n(t)_n=0^n_max_n,n^ " [ eq14 ] where @xmath34 is a normalization factor such that the trace of the density operator in the total space @xmath9 is constant in time , @xmath35 , @xmath36 are scalar quantities related to the coupling between @xmath5 and @xmath7 and @xmath37 is the upper limit of the bosonic quantas present in the environment . an exact analytical expression of these quantities can be obtained by using an infinite zassenhaus series @xcite whose development is given in appendix a. the explicit expressions of the polynomials @xmath38 are developed in appendix b. by simple inspection of these expressions it can be seen that the non - diagonal element of @xmath39 may cross zero with @xmath10 , oscillate and never reach and stay at zero whatever the length of the time interval which goes to infinity .
this is the signature of the fact that no decoherence of the system will be observable in this case . in the application of the model the spin of the system
is supposed to stay in a unique @xmath28 state , @xmath40 .
the energy - time expression leads to the lower bound t [ 15 ] which can be worked out analytically .
the index @xmath41 corresponds to different choices of the integrant @xmath42 which is chosen as the the smallest one given by eqs.(17 - 19 ) .
the fidelity takes the form f^j m_1,j m_2(_s^(0)(0),_s(t))= [ 16 ] which is a real quantity and consequently the expression of @xmath43 is obtained by means of eq.(2 ) . finally the energy denominator @xmath44 where @xmath45 $ ] is given by _
op(t)=1/t_0^t max[|_i(t)| ] dt [ 17 ] where @xmath46 are the eigenvalues of the operator @xmath47 and similarly _
tr(t)=1/t_0^t [ |_1(t)+|_2|(t)]dt [ 18 ] and _ hs(t)=1/t_0^t [ |_1(t)|^2+|_2(t)|^2]^1/2 dt [ 19 ] details concerning the derivation of the fidelity are given in appendix c.
we use the present model in order to analyze the fidelity and the behaviour of the quantum speed limit under different physical conditions .
we fix the energy parameters @xmath48 and @xmath49 and consider different strengths of the interaction @xmath50 ( weak coupling = w ) , @xmath51 ( intermediate coupling = i ) and @xmath52 ( strong coupling = s ) .
we choose different extensions of the bosonic environment , in practice @xmath53 .
the system space is restricted to @xmath54 . at time
@xmath55 the closed system is in a pure state , the state of lowest energy is |_s(0 ) = 1/2(|1/2 ; 1/2 + |1/2 ; -1/2 ) [ 20 ] the bosonic contribution at @xmath55 is fixed by @xmath56 .
+ we consider first the fidelity which is the essential ingredient in the r.h.s . of eq.(15 ) .
\a ) weak interaction : the fidelity shows an oscillating behaviour which indicates that the time dependent state vector @xmath57 rotates with time .
the evolution of @xmath58 for different times @xmath10 is the following : for @xmath59 , @xmath60 : @xmath61 , @xmath62 and @xmath63 for @xmath64 .
it oscillates for larger time intervals up to @xmath65 with sizable amplitudes comparable to those of small time intervals .
the behaviour is qualitatively similar for @xmath66 and @xmath67 .
one observes a saturation effect from @xmath66 to @xmath67 , the values of the fidelity are very similar to each other .
+ b ) intermediate interaction : the results are similar to those observed in the weak case .
the amplitude oscillations start with a similar decrease but diminish more quickly than in the former case .
for @xmath68 , @xmath60 : @xmath69 , @xmath70 and for @xmath71 @xmath72 , @xmath73 . for @xmath66
: @xmath61 , @xmath74 .
@xmath75 . the saturation effect from @xmath66 to @xmath67
is again seen here .
the pseudo - period of oscillation is however somewhat shorter than in the weak case for @xmath66 and @xmath67 .
oscillations persist over larger time intervals .
+ c ) strong interaction : the fidelity is large at the very beginning and gets quickly very small for early times .
the oscillation period is much shorter than in the former cases .
it goes on oscillating but with small amplitudes for larger time intervals .
for @xmath76 , @xmath60 : @xmath77 , @xmath78 , @xmath79 .
it is similar for @xmath66 : @xmath80 , @xmath81 , @xmath82 and for @xmath67 : @xmath80 , @xmath81 , @xmath83 .
the fidelity goes on oscillating over larger period of times .
the saturation effect is also present here .
+ in summary the fidelity is an oscillating function of the length of the time interval for any strength of the interaction .
it is large for small times with an overall decrease in amplitude with increasing time .
its pseudo - period decreases with increasing interaction strength between the system and its environment .
one observes a saturation effect with an increasing number of states in the environment .
the average energy over a unit time interval @xmath84 which enters the expression of the lower bound @xmath85 of @xmath2 is defined in eq.(4 ) .
the essential trend of its temporal evolution can be characterized as shown below .
+ a ) weak interaction : the average energy oscillates with a similar amplitude of the order of unity for small time intervals @xmath86 $ ] and @xmath53 . for large times
( @xmath87 $ ] ) @xmath84 gets smaller and the oscillation amplitudes get reduced .
+ b ) intermediate interaction and strong interaction : the behaviour is qualitatively the same as in the weak case with smaller amplitudes and reduced oscillations over the time interval @xmath88 $ ] . the lower bound @xmath85
defined in eq.(7 ) is the physical quantity of interest in the present study .
it depends on the behaviour of the fidelity ( numerator ) and the average energy per time unit present in the system ( denominator ) .
the analysis is done with the same set of parameters as those used for the fidelity .
+ a ) weak interaction : the lower bound @xmath85 is quite smaller than the evolution time interval , @xmath89 , whatever the extension of the bosonic space @xmath53 .
it is very similar in magnitude in all cases .
it is rather stationnary for @xmath90 and then increases slowly and quasi linearly , see fig.1 . for larger times ( @xmath91 $ ] ) @xmath85 remains of the same order of magnitude and oscillates .
+ \b ) intermediate interaction : for @xmath60 @xmath85 is quite smaller than the time interval @xmath10 except for small @xmath90 where it increases strongly with a steep slope , see fig.2 . + for @xmath92 one observes a saturation effect for which the values of @xmath85 come very close to each other .
otherwise the behaviour is the same as in the case @xmath60 , except for the fact that the magnitude of the oscillations are smaller , see fig.2 .
this remains so for larger time intervals .
the magnitudes of @xmath85 do not increase with time for @xmath93 as it is the case in the weak regime , they tend even to decrease compared to their behaviour in the former regime .
+ \c ) strong interaction : like in the intermediate strength case @xmath85 is somewhat larger for @xmath60 than for @xmath92 but the oscillations in time follow rather closely in time and are more or less in phase with each other .
the graph in fig.3 shows sizable oscillations for @xmath86 $ ] . for larger times ( @xmath94 $ ] ) oscillations of @xmath85
remain but they are smaller than those observed for small times . like in the case of the fidelity , the difference between the case @xmath66 and @xmath67 is negligible , the curves can not be distinguished from each other . + summing up : * the lower bound of the time @xmath85 is generally closer to the time interval @xmath10 for small time intervals than for long time intervals and shows oscillations which are essentially generated by the behaviour of the fidelity which appears in the numerator . * for all three interaction strengths there is no qualitative difference between the behaviour of @xmath85 in the case where @xmath60 and @xmath92 except over short time intervals . the fact that @xmath85 diminishes relative to increasing time intervals @xmath10 in the case of an intermediate and strong interaction is due to the fact that in both cases the fidelity @xmath58 decreases and the energy denominator @xmath84 varies slowly .
this is not the case when the interaction is weak . +
* the cases @xmath66 and @xmath67 are very similar , one observes a saturation effect with respect to an extension of the environment space in the size of @xmath85 .
we used the time - energy inequality of deffner and lutz @xcite in order to examine the behaviour of the speed limit in a non - decoherent open quantum system @xcite .
+ the time which charaterizes the speed with which the open system evolves in time shows a lower bound @xmath85 which is governed by the fidelity @xmath58 and an accumulated average energy per time unit @xmath84 .
both quantities oscillate in time .
this is due to the nature of the density matrix of the spin @xmath95 system which shows oscillatory matrix elements . + we considered an environment space which contains a set @xmath96 of non - interacting bosons with different numbers of excitation quantas and studied the evolution of the system for different values of the interaction strength between the system and the environment .
+ we first analyzed the behaviour of the fidelity @xmath58 which leads to the bures angle @xcite and showed how the state vector of the open system evolves in time .
it oscillates in the interval @xmath97 $ ] for all values of the interaction with different amplitudes and periods depending on the strength chosen to be weak , intermediate and strong .
the oscillatory character is induced by the structure of the density operator of the system .
the oscillating average energy per time unit @xmath84 varies due to the energy exchange between the system and its environment .
+ the physically observable of interest is the lower time limit @xmath85 .
different strengths lead to different lower time bounds @xmath85 . in the case of a weak amplitude @xmath85
is close to the time interval @xmath10 for small @xmath10 .
+ for an intermediate and strong strengths of the interaction @xmath85 comes closer to the time interval @xmath10 for small times .
it gets smaller if the bosonic space contains several excited states .
this is essentially due to the decrease of the numerator of the expression of @xmath85 governed by the fidelity , to some extent to the variation of the energy @xmath84 .
the size of @xmath85 is the smallest when the interaction is weak . +
as it can be seen on the figures , one observes a rather systematic `` saturation '' effect in the sense that @xmath85 starts to be insensitive to an enlargement of the environment space , i.e. it remains the same when one increases the number of excitation quanta from @xmath98 to @xmath99 excitations .
this is related to the fact that higher contributions to the quantity @xmath100 ( eq.14 ) which enters the expression of the fidelity ( eq.16 ) get small when the number of excitation quanta @xmath96 increases .
+ it has been shown in a preceding work that the evolution of an open quantum system obeys the divisibility property which characterises a markovian behaviour if the environment space reduces to a unique state @xcite .
it would be of interest to analyze the behaviour of @xmath85 in this specific case .
if @xmath101 and @xmath102 e^x+y = e^xe^ye^-c_2(x , y)/2!e^-c_3(x , y)/3!e^-c_4(x , y)/4 ! ... [ eq31 ] where @xmath103 $ ] + @xmath104,y]+[[x , y],x]$ ] + @xmath105,y],y]+[[[x , y],x],y]+[[x , y],[x , y]$ ] , etc .
+ the series has an infinite number of term which can be generated iteratively in a straightforward way @xcite . if @xmath106=0 $ ] the truncation at the third term leads to the factorisation of the @xmath107 and the @xmath108 contribution .
if @xmath106=c$ ] where @xmath109 is a c - number the expression corresponds to the well - known baker - campbell - hausdorff formula . here
the expressions of the evolution operator @xmath110 can be rigorously worked out analytically and the series can be formally summed up to infinity for any time by means of analytic continuation arguments .
the expressions of the bosonic contributions to the density matrix @xmath111 are given by e_n , n(j_1,t)=e^-it_nn_2,n_3n_2_n_3n_4,nn_4(-i)^n+n_3 ( -1)^n+n_2-n_4 + e^_1(t ) [ eq23 ] and e^*_n^",n(j_2,t)=e^it_n^"n_2,n_3n_2_n_3n_4,nn_4i^n^"+n_3 ( -1)^n+n_2-n_4 + e^_2(t ) [ eq24 ] the different quantities which enter @xmath112 are _ 1(t)= [ eq25 ] _ 1(t)= [ eq18 ] ( j_1)=j_1(j_1 + 1 ) [ eq26 ] _ 1(t)=-[+ ] [ eq27 ] and for @xmath113 : _ 2(t)= [ eq28 ] _ 2(t)= [ eq29 ] ( j_2)=j_2(j_2 + 1 ) [ eq30 ] _ 2(t)=-[+ ] [ eq31 ]
define the operator | we study the minimum time related to the quantum speed limit that characterizes the evolution of an open quantum system with the help of a simple model in the short and long time limits .
we compare in particular the situation corresponding to a unique state and several states in the environment space . for short time
intervals the results show a sensible difference in the behaviour of the system for different strengths of the interaction between the system and its environment .
keywords : open quantum systems , quantum speed limit , time scales + pacs numbers : 03.65.-w , 03.65.yz , 05.30.jp + |
the binary system / is formed by a young 48 ms radio pulsar in an eccentric orbit of 3.4 years around a massive main - sequence star @xcite .
the parameters of the system are shown in table [ table : system ] . the spectral type of the massive star , o9.5ve , and some of the binary parameters have been recently updated by @xcite , who obtained a distance to the system of @xmath3 kpc .
close to the periastron passage the system displays non - thermal unpulsed emission that has been detected in radio @xcite , x - rays @xcite , hard x - rays up to 200 kev @xcite , and very high energy ( vhe ; 0.1100 tev ) @xmath4-rays above @xmath5 gev @xcite . in the range @xmath00.1100
gev strict upper limits were obtained by egret close to the 1994 periastron passage @xcite , and the source has not yet been observed by _
agile _ and _ fermi _ close to periastron .
the vhe emission is variable on orbital timescales , and is interpreted as the result of inverse compton upscattering of stellar uv photons by relativistic electrons , which are accelerated in the shock between the relativistic wind of the young non - accreting pulsar and the wind of the stellar companion ( see @xcite , and references therein ) .
the high - mass binaries and have also been detected at vhe , and show a broadband spectral energy distribution ( sed ) similar to that of / @xcite .
in contrast to x - ray binaries , all three sources have the peak of the sed at mev - gev energies . for these reasons
, they can be considered gamma - ray binaries .
however , the nature of the compact objects in and is unknown because their masses are not well constrained by the system mass functions @xcite , and no pulsations have been found .
these systems have been extensively observed at vhe during several orbital cycles , while the observations of , the only system with a confirmed pulsar , are scarce due to the long orbital period .
any observational link between the three gamma - ray binaries would shed light in the understanding of this kind of systems .
l c c c pulsar period & @xmath6 & @xmath7 ms & 1 + period derivative & @xmath8 & @xmath9 & 1 + characteristic age & @xmath10 & @xmath11 yr & 2 + surface magnetic field & @xmath12 & 3.3@xmath13 g & 2 + spindown luminosity & @xmath14 & @xmath15 erg s@xmath16 & 3 + spectral type & & o9.5ve & 4 + effective temperature & @xmath17 & @xmath18@xmath19 k & 4 + surface gravity & @xmath20 & @xmath21@xmath22 & 4 + radius & @xmath23 & @xmath24@xmath25 & 4 + optical luminosity & @xmath26 & @xmath27 erg s@xmath16 & 4 + mass & @xmath28 & @xmath29 & 4 + distance & @xmath30 & @xmath3 kpc & 4 + mass function & @xmath31 & 1.53 @xmath32 & 5 + terminal wind velocity & @xmath33 & @xmath34 km s@xmath16 & 6 + orbital period & @xmath35 & 1236.72432(2 ) days & 1 + reference epoch & @xmath36 & mjd 48124.34911(9 ) & 1 + semimajor axis & @xmath37 & @xmath38 au & 4 + orbit inclination & @xmath39 & @xmath40 & 4 + eccentricity & @xmath41 & 0.8698872(9 ) & 1 + argument of periastron & @xmath42 & @xmath43 & 1 + longitude of ascending node & @xmath44 & @xmath45 & see the text + proper motion ( rigth ascension ) & @xmath46 & @xmath47 mas yr@xmath16 & 7 + proper motion ( declination ) & @xmath48 & @xmath49 mas yr@xmath16 & 7 + the radio emission from the system is described most recently in @xcite , which includes multiwavelength data from atca observations obtained during the periastron passages of 1994 , 1997 , 2000 , and 2004 ( hereafter we will refer to the epoch of each periastron passage as @xmath50 ) .
the emission has two non - thermal components .
the first one is pulsed emission with a flux density of @xmath025 mjy at 2.5 ghz and a nearly flat spectral index , which disappears approximately from 16 days prior to periastron ( @xmath5116 ) to 15 days after periastron ( @xmath5215 ) .
this has been interpreted as an eclipse of the pulsar when it crosses behind the equatorial circumstellar disk present around the massive star @xcite .
the second component is transient unpulsed synchrotron emission that appears at @xmath5120 days and shows two peaks centered around @xmath5110 and @xmath5220 days , with flux densities at 2.5 ghz up to @xmath01520 mjy and @xmath03050 mjy , respectively .
after the post - periastron peak , the flux density of the unpulsed emission decreases continuously , and it has been detected up to @xmath53100 days . this transient emission remains optically thin during the outbursts .
the broadband transient emission of / is produced around periastron , when strong interaction is produced between the stellar and pulsar winds . the shocked material is contained by the stellar wind behind the pulsar , producing a nebula extending away from the stellar companion . along this
adiabatically expanding flow , the accelerated particles produce synchrotron emission from radio to x - rays @xcite .
the expected morphology depends on the magnetization parameter of the pulsar wind , @xmath54 , defined as the upstream ratio of magnetic to kinetic energy .
these models predict that the radio emission extends up to several au ( corresponding to several milliarcseconds , mas , at 2.3 kpc ) , and that its structure should be variable on orbital timescales .
this radio behavior has not been tested in , which only emits over a period of a few months every 3.4 year . here
we present the first vlbi radio images of , obtained close to its 2007 periastron passage .
the high - resolution images at three different orbital phases provide a direct view of the small - scale morphology of the source , which is comparable to those previously observed in and .
c c c c c c c c a & 54309.0454309.46 & 4.65 & 5 & 1.1(2)1.5(2 ) & 0.00087(14)0.00120(14 ) & 9.1(1.4)12.5(1.4 ) & 191 + b & 54328.9754329.40 & 3.75 & 5 & 21.0(2)21.4(2 ) & 0.01698(14)0.01733(14 ) & 98.3(3)99.1(3 ) & 279 + c & 54623.2954623.67 & 4.20 & 4 & 315.3(2)315.7(2 ) & 0.25496(14)0.25527(14 ) & 165.97(1)165.99(1 ) & 346 + was observed with the australian long baseline array ( lba ) at 2.3 ghz ( 13 cm ) on three epochs : 2007 july 28 ( run a ) , 2007 august 17 ( run b ) , and 2008 june 6 ( run c ) .
the lba observations were performed with five antennas of the array : parkes , atca , mopra , hobart ( not present in run c ) and ceduna .
the observational parameters of each of the @xmath55 hr runs are shown in table [ table : observations ] .
the small number of antennas provides a rather poor @xmath56-coverage that makes the data calibration and imaging more difficult than for arrays with more elements , because we have less baseline redundancy .
the data were recorded at a bit rate of 512 mbps per telescope distributed in eight sub - bands ( four for each right- and left - handed polarization ) with a bandwidth of 16 mhz , each of them correlated using 64 frequency channels , two - bit sampling , and 2 s of integration time .
hobart and ceduna recorded at 256 mbps ( only two sub - bands for polarization ) .
the data were correlated at swinburne university using the difx software correlator @xcite without applying pulsar gating or binning .
the observations were performed using phase referencing on the calibrator j1337@xmath576509 ( b1334@xmath57649 ) , which has an angular separation of @xmath58 from and was correlated at @xmath59 and @xmath60 . this reference position from the global vlbi solution
2006d_astro has an uncertainty of 13 mas .
the cycle time was 6 minutes , spending half of the time on the phase calibrator and the target source alternatively .
the total flux density of the phase calibrator was @xmath61 , @xmath62 , and @xmath63 mjy for runs a , b , and c , respectively .
the source j0538@xmath574405 ( b0537@xmath57441 ) was used as a fringe finder for runs a and b , and j1337@xmath576509 ( b1349@xmath57439 ) was used for run c. no astrometric check source was observed during the runs . the data reduction was performed in aips .
total electron content models based on gps data obtained from the cddis data archive were used to correct phase variations due to the ionosphere .
standard instrumental corrections were applied ( parallactic angle , instrumental offsets , and slopes between and within bands and bandpasses ) .
fringe fitting on the phase calibrator was performed with the aips task fring , and the solutions were applied to the target source .
a self - calibrated model of the calibrator was used as an input model for calib , which was used to reduce the amplitude instabilities on timescales greater than 10 minutes .
the data were averaged in frequency and time , and clean images were produced . for run
b only , a single round of phase self - calibration was applied , to mitigate the residual atmospheric phase instabilities , which were more noticeable at this epoch .
the final images were produced using a natural weighting scheme ( robust 5 within the aips task imagr ) . for run b , robust 2 and tapering
were applied to avoid the presence of possible unreliable high - resolution features due to sidelobes of the synthesized beam .
self - calibration slightly affects measured properties such as extent and position angle ( p.a . ) , while it preserves the morphology .
the resulting vlbi images at 2.3 ghz are shown in figure [ fig : f1 ] .
extended emission is detected at distances up to 5055 mas ( @xmath64@xmath65 au at @xmath3 kpc ) during the two runs shortly after the periastron passage . the emission becomes gradually fainter from the peak toward the northwest , and no individual components have been found .
the p.a . of the extended emission with respect to the peak is @xmath66 for run a and @xmath67 for run b. the emission in run c , 315 days after the periastron passage , is dominated by a point - like source of a few mjy .
cccccccccccccc a & 19.9 @xmath68 1.4 & 10.4 @xmath68 0.2 & 28.9 @xmath69 26.1 & @xmath70 & 50 & 120 @xmath68 20 & @xmath5767 & 11.3 @xmath68 0.4 & 14.0 @xmath68 0.5 & ( 1020 ) @xmath68 3 & ( 2446 ) @xmath68 7 + b & 46.7 @xmath68 1.0 & 32.7 @xmath68 0.4 & 31.0 @xmath69 25.1 & @xmath71 & 55 & 132 @xmath68 22 & @xmath5750 & 4.2 @xmath68 0.1 & 11.3 @xmath68 0.1 & ( 514 ) @xmath68 3 & ( 1231 ) @xmath68 7 + c & 3.0 @xmath68 0.4 & 2.8 @xmath68 0.4 & 50.3 @xmath69 25.1 & @xmath72 & @xmath732.8 & @xmath736.7 @xmath68 1.1 & & 0.0 @xmath68 0.6 & 0.0 @xmath68 1.1 & & + like all vlbi arrays , the absolute flux calibration of the lba relies on noise calibration injection and so the absolute flux values reported in table [ table : results ] should be taken as uncertain at the @xmath74 level .
the flux densities of runs a and b are compatible with previous atca observations at the corresponding orbital phases ( the atca data from the current observations were not correlated as an independent array ) .
the flux density in run c is compatible with the flux density of the pulsar .
this is expected considering the lack of unpulsed emission at @xmath75 and @xmath76 in previous atca observations @xcite .
the phase - referenced observations allow us to obtain relative astrometry between runs .
since no astrometric check source was observed , we do not have real measurements of the astrometric uncertainties , and we will use the formal errors of a gaussian fit obtained with jmfit within aips . given the relatively large calibrator target separation of @xmath77 , we expect an additional systematic component to the position error due to the unmodeled ionosphere of 15 mas @xcite . as a reference position for the plots we use the peak position of run c , @xmath78 and @xmath79 , which we consider to represent the pulsar position at mjd 54623.48 .
assuming a mass of the neutron star of 1.4 @xmath32 and a stellar mass of 31@xmath80 @xmath32 ( see table [ table : system ] for the system parameters ) , the mass function provides @xmath81 and the semimajor axis of the pulsar orbit is @xmath82 au , or @xmath83 mas for a distance to the system of @xmath84 kpc .
the red crosses in figure [ fig : f1 ] mark the region where the pulsar should be located in each epoch .
their centers are placed at the position of run c corrected for proper motion at the corresponding epochs ( mjd 54309.25 for run a and mjd 54329.18 for run b ) . since we do not know the orientation of the orbit in the sky , we plot as error bars the projected orbital separation of the pulsar with respect to run c.
the error bars also include the 1@xmath54 uncertainties on the mass of the star ( and hence the distance uncertainty ) , on the astrometry of run c , and on the offset due to the proper motion .
finally , we have included the astrometry in runs a and b to compute the range of possible separations between the peak of the emission and the pulsar , as shown in table [ table : results ] .
we consider all possible values of the longitude of the ascending node ( @xmath44 ) , which determines the orientation of the orbit in the plane of the sky .
the radio morphology at a given epoch depends on the spatial distribution of synchrotron emitting particles and their emission processes .
given the limitations of our data ( only two images , and without accurate astrometry ) , we have used a simple kinematical model to check if it can trace the extended structures detected .
we have considered the shock between the relativistic pulsar wind and a spherical stellar wind .
the shock is produced at the standoff distance , the region where the pulsar and stellar wind pressures balance , as described in @xcite .
the evolution of the nebular flow after the shock is described in @xcite .
this should be considered as a first approximation to the much more complex hydrodynamic behavior of a shocked flow in a binary system , as shown in @xcite . in the kennel & coroniti approximation of a non - turbulent adiabatically expanding flow
, the flow speed depends only on the magnetization parameter @xmath54 when assuming @xmath85 .
this allows us to compute the past trajectory of the flow produced behind the standoff distance , which depends on the components separation along the orbit , the mass loss rate of the star ( @xmath86 yr@xmath16 for a typical o9 star ; @xcite ) , the terminal wind velocity , @xmath33 , and the spin - down luminosity of the pulsar , @xmath14 . with these restrictions ,
the only free parameters are the longitude of the ascending node , @xmath44 , which describes the orientation of the orbit , and the magnetization parameter , @xmath54 . projecting the past trajectory of the flow on our vlbi images of runs a and
b , we found that the best match with the detected morphologies is obtained for @xmath87 and a magnetization parameter of @xmath88 , assuming an orbital inclination of @xmath81 .
the obtained trajectories are shown in figure [ fig : f2 ] for three different values of @xmath54 .
we note that the uncertainty in the distance to the system scales the size of the contours , but not the orbit and the trajectories , which are computed in au .
the range of magnetizations in the plots has been chosen to approximately show the effect of keeping @xmath89 and changing either the distance from 1.9 to 2.7 kpc , either @xmath44 from @xmath5735 to @xmath90 , or a variation of the product @xmath91 of two orders of magnitude .
the parameters of the circumstellar equatorial disk of the star are uncertain , but considering @xmath92 yr@xmath16 , @xcite , and a wind velocity of @xmath010 km s@xmath16 , the product @xmath93 is @xmath0100 times smaller than the spherical wind contribution .
if this equatorial component dominates during certain orbital phases , the flow trajectory would be closer to the orbit than the dashed models in figure [ fig : f2 ] for some specific regions of the more recent part of the trajectory .
however , we do not attempt to accurately model this circumstellar disk , as its density , velocity , and crossing time are unknown .
we note that the astrometric errors can be of the order of @xmath010 au ( see above ) .
we also emphasize that this simple model can not account for the complex magnetohydrodynamical turbulences of the real flow and should be considered as a first approximation to constrain @xmath54 .
previous studies have assumed a magnetization parameter @xmath54 of around 0.010.02 for this pulsar ( see @xcite ; @xcite ) , considerably higher than our best - fit value , although comparable lower values for @xmath54 have been suggested for other systems ( e.g. , the crab pulsar ; * ? ? ?
the results presented in this letter show that the particle accelerator within / can produce a flow of particles emitting synchrotron radiation that can travel several au .
the total projected extent of the nebula is @xmath94 mas , or @xmath95 au , and the peak of the emission is clearly displaced from the binary system orbit ( see figure [ fig : f1 ] and table [ table : results ] ) .
similar morphologies and displacements have been found in the other two known gamma - ray binaries , and , although for smaller sizes and on shorter timescales @xcite .
there is an ongoing debate on the nature of the compact object and particle acceleration mechanisms in and .
although initially suggested to be accreting / ejecting microquasar systems @xcite , they are now thought to contain young non - accreting pulsars @xcite , which can explain their multiwavelength emission @xcite . however , there are three basic issues that are not well understood for and , and comparisons with / can help to clarify the situation .
first , the putative pulsar properties of these two sources are unknown , as no pulsations have been detected for these systems .
the lack of pulsations can be explained by the intense stellar wind that produces an extremely high absorption for these small orbits , a 3.9 day orbit with separations between 0.1 and 0.2 au for , and 26.5 days with separations of 0.10.7 au for @xcite . as a reference ,
the separation for is in the range 0.913.4 au , and the pulsations disappear for distances below @xmath96 au .
second , it is not clear if the massive stellar wind can confine the pulsar wind @xcite .
vlbi images , like the ones presented here , can shed light on the shock conditions and geometry .
third , the observed sed and variability at gev energies is not well understood @xcite .
_ fermi _ and _ agile _ are observing for the first time during the 2010 periastron passage , and are providing gev data that can be compared with those already obtained for and . in this context , the high - resolution vlbi radio observations presented here establish a common link to test the similarities between the three systems . in conclusion ,
our results provide the first observational evidence that non - accreting pulsars orbiting massive stars can produce variable extended radio emission at au scales .
similar structures are also seen in and , in which the nature of the compact object is unknown because the detection of pulsations is challenging .
the discovery presented here for the young non - accreting pulsar reinforces the link with these two sources and supports the presence of pulsars in these systems as well . planned lba observations of during the 2010 periastron passage
will allow us to provide a full comparison with the behavior observed in these sources , which have been extensively monitored during several orbital cycles .
we have also shown that the orientation of the orbit and the magnetization of the pulsar can be inferred from vlbi observations of the source .
several images at different orbital phases covering a wider range of true anomalies will allow for a complete modeling of the orbital changes of the extended emission .
finally , accurate vlbi observations of the pulsed emission during several orbits can provide the pulsar trajectory , from which we can directly obtain the proper motion of the binary system , the inclination and @xmath44 of the orbit , and the distance to the system .
j.m . , m.r . , and j.m.p .
acknowledge support by dgi of the spanish ministerio de ciencia e innovacin ( micinn ) under grants aya2010 - 21782-c03 - 01 and fpa2010 - 22056-c06 - 02 .
this work has been supported by the consejera de innovacin , ciencia y empresa of junta de andaluca as research group fqm-322 , and excellence grant fqm-5418 . j.m .
acknowledges support by micinn under grant bes-2008 - 004564 .
m.r . acknowledges financial support from micinn and european social funds through a _
ramn y cajal _ fellowship .
the australian long baseline array is part of the australia telescope which is funded by the commonwealth of australia for operation as a national facility by csiro .
is a jansky fellow of the national radio astronomy observatory ( nrao ) .
the nrao is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
casares , j. , ribas , i. , paredes , j. m. , mart , j. , & allende prieto , c.2005a , mnras , 360 , 1105 casares , j. , rib , m. , ribas , i. , paredes , j. m. , mart , j. , & herrero , a. 2005b , mnras , 364 , 899 cerutti , b. , dubus , g. , & henri , g. 2008 , , 488 , 37 moldn , j. , rib , m. , & paredes , j. m. , high - energy emission from pulsars and their systems : proceedings of the 1st session of the sant cugat forum of astrophysics ( april 12 - 16 , 2010 ) .
nanda rea & diego f. torres ( editors ) , 2011 , springer .
( arxiv:1101.4153 ) torres , d. f. , high - energy emission from pulsars and their systems : proceedings of the 1st session of the sant cugat forum of astrophysics ( april 12 - 16 , 2010 ) .
nanda rea & diego f. torres ( editors ) , 2011 , springer .
( arxiv:1008.0483 ) | is a 48 ms pulsar in a highly eccentric 3.4 year orbit around the young massive star . during the periastron passage the system displays transient non - thermal unpulsed emission from radio to very high energy gamma rays .
it is one of the three galactic binary systems clearly detected at tev energies , together with and .
we observed after the 2007 periastron passage with the australian long baseline array at 2.3 ghz to trace the milliarcsecond ( mas ) structure of the source at three different epochs . we have discovered extended and variable radio structure .
the peak of the radio emission is detected outside the binary system near periastron , at projected distances of 1020 mas ( 2545 au assuming a distance of 2.3 kpc ) .
the total extent of the emission is @xmath050 mas ( @xmath0120 au ) .
this is the first observational evidence that non - accreting pulsars orbiting massive stars can produce variable extended radio emission at au scales .
similar structures are also seen in and , in which the nature of the compact object is unknown .
the discovery presented here for the young non - accreting pulsar reinforces the link with these two sources and supports the presence of pulsars in these systems as well .
a simple kinematical model considering only a spherical stellar wind can approximately trace the extended structures if the binary system orbit has a longitude of the ascending node of @xmath1 and a magnetization parameter of @xmath2 . |
our knowledge of the physical conditions during the very early phases of high - mass star formation ( @xmath4 ) has been relatively limited in the past ( e.g. , * ? ? ?
* ) , due to significant observational challenges .
consequently , the qualitative and quantitative differences in star - formation processes over different mass regimes are still strongly debated ( cf . * ? ? ?
* ) . in the endeavor
to enhance our understanding of star formation beyond the well - studied low mass stellar nurseries , infrared dark clouds ( irdcs ) have become more and more important targets .
initially irdcs were discovered by the infrared space observatory ( iso , * ? ? ?
* ) and the midcourse space experiment ( msx , * ? ? ?
they appear as filamentary opaque structures against the diffuse galactic mid - infrared background .
as irdcs are known to be cold ( @xmath5k ) , dense , and exhibiting high column densities @xmath6@xmath7 , they are suitable environments to form young stellar objects that can evolve into stars of intermediate to high mass ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
irdcs were widely discussed in the context of massive star formation , but a certain fraction of irdcs only host sub - solar to intermediate - mass star formation @xcite .
still , @xcite argue that most stars in our galaxy could form in irdc - like molecular structures .
most of the current theories about the formation of irdcs treat the collapsing gaseous medium as isothermal . with the _ herschel _ observatory
, we have the spatial resolving power , spectral coverage , sensitivity and dynamic range to study the temperature and column density structure of irdcs in more detail .
already one of the early _ herschel _ studies using spatially resolved sed fitting toward an irdc indicated a non - isothermal temperature structure @xcite .
@xcite compared _ herschel _ observations of irdcs without embedded ir sources to radiative transfer models of strictly externally heated structures .
they find resulting temperature profiles that go down to 11.5k ( sample average ) in the center .
two further studies analyzed the temperature structure using sed - fitting and a modified blackbody model .
@xcite present large scale temperature maps of irdcs . @xcite studied the irdc18454 near the w43 mini - starburst and present dust temperature maps .
both studies found temperatures higher by more than 5k compared to the irdcs from @xcite .
further investigations have to clarify if the majority of irdcs show similar minimum temperatures or if clumps with higher central temperatures exist .
this classification is essential for the understanding of the formation process of massive stars .
the comparison between dust and gas temperatures can shed light on the efficiency of the gas - dust coupling at densities typical for irdcs .
furthermore , a well constrained dust temperature structure can provide critical input for chemical models of irdcs , and in particular for the dust grain surface reactions .
the goal of this paper is to accurately characterize the physical conditions of ( high - mass ) star formation in a relatively isolated irdc , [email protected] ( henceforth called `` 48 '' ) .
this case study is based on observations taken as part of the earliest phases of star formation ( epos ) herschel guaranteed time key program @xcite .
we present sub - parsec scale dust temperature and column density maps and place these properties in context with the kinematic environment , probed by various molecular gas tracers ( , , , , ) .
herschel also enables us to characterize the embedded protostellar core population and , through the use of evolutionary models , try to understand the likely outcome of star formation in this cloud .
the kinematic distance to 48 was determined to be @xmath9kpc @xcite 2.5kpc @xcite or 2.6kpc ( this work ) .
48 appears to be projected near the gmc w51 ( figure [ fig : rgb_overview ] , left ) , which is in fact twice as distant as 48 ( * ? ? ?
* see also section [ sec : dist ] ) .
48 appears isolated ; and therefore we can assume it is not strongly influenced by local heating sources .
furthermore 48 has a quite simple monofilamentary structure . by comparing the observational results from this isolated irdc to more complex irdcs ( c.f .
* ) , we can constrain the dominant effects at work for star formation in irdcs . @xcite
estimated from 3co observations an average density for h@xmath10 associated with 48 of @xmath11 , a peak column density of @xmath12 and a lte mass of @xmath13 .
@xcite identified three different clumps with masses from 91 to @xmath14 using scuba and msx observations . by modeling the internal velocity structure of these clumps using @xmath15 and @xmath16 single - dish data they found that the line - widths decrease outside the cloud , indicating a decrease in turbulent motions outwards of the clumps .
this suggests that the star formation activity inside these clumps down to the core scale is driven by turbulence .
@xcite identified 13 ysos in the vicinity of 48 , based on _ spitzer _ mid - infrared maps .
the majority of these objects belong to `` stage i '' , referring to the classification scheme of @xcite , and have low to intermediate masses . from a comparison of [ ci ] and data , @xcite inferred the presence of a young photodissociation region ( pdr ) ionized by the ysos inside of 48 . throughout the literature
the limits of sizes and masses for clumps and cores vary . in this work the spatial definition from @xcite is applied : cores range from @xmath17pc , while clumps extend over @xmath18pc .
given the isolated nature and simple structure of 48 , we use this source to study the collapse and fragmentation of a filamentary irdc . through the characterization of the point sources the evolutionary stage
can be assessed a final stellar mass can be estimated from different models .
furthermore , the _ herschel _ maps are utilized to study the temperature and column density structure of 48 .
molecular line observations provide information about the morphology of the clumps and clues to the pressure support therein .
we assess the level of freeze - out by studying the co - depletion in the central part of the irdc .
the next section describes the observations , the data reduction and some basic data processing steps for _ herschel _ , _ spitzer _ , sub - millimeter continuum and molecular line observations .
section [ sec : results ] details the scientific analysis and the direct results . in section
[ sec : morpho ] the structure and environment of 48 is described .
the measured distances for 48 and other sources in the near projected distance are discussed in section [ sec : morpho ] .
the temperature mapping , utilizing _
herschel _ , scuba and atlasgal data , is delineated in section [ sec : tmap ] .
the subsequent section [ sec : colmap ] compares the column densities with other observations and estimates mass for the whole irdc .
section [ sec : pointsrc ] addresses the fitting and classification of point sources from the mips 24@xmath19 m and pacs bands .
section [ sec : molecline ] contains the analysis of the molecular line data .
the concluding section [ sec : conclusion ] discusses the results in respect to previous studies of irdc and theoretical considerations .
the last section [ sec : summary ] summarizes the main results .
in this paper we present results from the _ herschel _ gto key program - `` early phases of star formation '' ( epos ) .
the high - mass portion of epos is compiled from different surveys .
the sample includes a sub - sample of `` classical '' irdcs , a selection of irdcs in the near vicinity of high - mass proto - stellar objects ( hmpos ) and cold , massive , compact sources detected by the iso serendipity survey ( isoss ) . for a detailed discussion of this sample and the underlining selection strategy we refer the reader to the overview paper by @xcite .
the epos observation were supplement with data from the the `` herschel infrared galactic plane survey '' ( hi - gal ) project , done with the photometer array camera and spectrometer ( pacs , * ? ? ? * ) and the spectral and photometric imaging receiver ( spire , * ? ? ? * ) on board of the _ herschel space observatory _ @xcite .
the epos pacs scan maps at 70@xmath19 m , 100@xmath19 m and 160@xmath19 m were obtained with the medium scan speed of @xmath20/s and the nominal scan leg length of 7@xmath21 .
the data were processed to level-1 , i.e. calibrated and converted to physical units , using hipe @xcite .
we applied a second level deglitching to flag bad data values by @xmath22-clipping flux values mapped on to each pixel .
the ` time ordered ' option was used , and a 25@xmath22-threshold was applied .
the level-2 products , i.e. fully reduced fits maps , were obtained using scanamorphos @xcite version 5.0 instead of applying the nominal hipe highpass filtering and photproject reduction step , due to the improved performance of scanamorphos ( see * ? ? ? * ) . the maps in each band
were reduced with a pixel - size of @xmath23/px .
the epos spire scan maps at 250@xmath19 m , 350@xmath19 m and 500@xmath19 m were obtained with a scan speed of 30/s . they were reduced up to level-1 using hipe version 5.0with the calibration pool 5.1 .
the final level-2 reduction was done with scanamorphos version 9.0 using the ` galactic ' option .
the pixel sizes are 6.0/px at 250@xmath19 m , 10.0/px at 350@xmath19 m and 14.0/px at 500@xmath19 m .
to include a larger coverage we obtained the raw data products from the hi - gal open time key project .
these observations were carried out in the pacs / spire parallel mode with a scan speed of 60@xmath24/s , resulting in large scale maps .
the 70@xmath19 m and 160@xmath19 m pacs maps have been reduced using hipe 8.0 , applying the same options as for the epos pacs data .
level-2 processing has been performed with scanamorphos version 15 .
the pixel size is 32/px for both maps .
the 250@xmath19 m , 350@xmath19 m and 500@xmath19 m spire maps were reduced up to level-1 using hipe version 8.0 with the calibration tree 8.1 .
scanamorphos version 15 was used to produce the level-2 maps .
the pixel size is the same as for the epos spire maps .
the final maps are presented in the left panel of figure [ fig : rgb_overview ] .
the hi - gal maps are used for the temperature mapping in section [ sec : tmap ] . due to the diffuse and heterogeneous background emission varying on a wide range of spatial scales , it is more robust to perform psf photometry than aperture photometry .
therefore the ` starfinder ` psf photometry software @xcite was used for the point source detection and flux extraction for the pacs bands .
the latest empirical psf of the vesta asteroid was used ( od 160 ) , obtained from the herschel science archive ( hsa ) , since an empirical psf derived from sources in 48 is noise dominated and therefore less acceptable . for consistency
the psf maps were reprocessed with scanmorphos and rotated to the orientation of the maps .
the noise in the pacs maps is not white noise , since it suffers from pixel - to - pixel correlations induced , e.g. , by non - thermal low - frequency noise . to calculate the average noise in the map
a gaussian distributed noise was assumed , but corrected by a factor for the pixel - to - pixel correlation .
the major source of confusion in the source detection is the complicated background .
in particular the bright background towards the gmc returns numerous false detections . we only include regions of the pacs maps that correspond to regions in the atlasgal map ( see section [ sec : submm ] ) with fluxes above a @xmath25 level . to further improve the detection efficiency an unsharp - mask - filter was applied in each pacs - band by subtracting
a median smoothed map from the respective science - frame . with this procedure
we avoid false detections , and miss three obvious sources for the automated detection run .
these sources were then included in the catalog identified by eye as clear detections . with the compiled fixed source positions the final photometry was performed on the non - filtered science frames .
to be considered a detection a source must meet the following conditions : it must have a source flux above the @xmath26 noise level and be detected in all pacs bands within the fwhm of the 160@xmath19 m pacs psf ( @xmath27 ) .
the results of the point source extraction are reported in section [ sec : pointsrc ] .
the given fluxes are already color - corrected for the source temperatures .
further details of the source extraction are given in the epos high - mass overview paper @xcite .
we obtained the _
irac and mips maps from the galactic legacy infrared mid - plane survey extraordinaire ( glimpse , * ? ? ?
* ) and the mips inner galactic plane survey ( mipsgal , * ? ? ?
the 24@xmath19 m mipsgal maps were processed with the mips data analysis tool starting from the ` raw ` data products .
we used the 8@xmath19 m irac map to create an extinction map .
psf photometry was performed on this map , using the ` starfinder ` package .
an averaged , empirical psf was used to remove all the point sources in this field .
the residual map was used to compute the optical depths .
one of the key problem in calculating an optical depth in ir data , is to determine the foreground and background intensities .
several different methods are described in the literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , to estimate the optical depth in one single cloud we used the method described in @xcite .
@xmath28 where @xmath29 is the flux derived from the irdc directly , @xmath30 is measured of `` off - cloud '' in a nearby region meant to represent the characteristic unattenuated mir emission .
@xmath30 ( and @xmath29 ) contains contributions from both the foreground and background ( i.e. @xmath31 ) , and only @xmath32 is needed to computed the optical depth . for @xmath33 , we assume the zodiacal light emission provided in the header of the irac data product .
this is a lower limit to the true foreground .
nevertheless , the uncertainties in the column densities are dominated by what is assumed about the extinction law ( factor 2 to 3 ) , whereas the @xmath33 contributes about an error less than 50% .
the point - source fluxes from mips 24@xmath19 m used in section [ sec : pointsrc ] were also obtained with ` starfinder ` . for mips
psf photometry we utilized a model psf computed by the ` stinytim ` program , based on ` tinytim ` for the hst @xcite .
the spectrum , shown in figure [ fig : spec ] , was obtained from the infrared spectrograph ( irs ) on - board the _ spitzer space telescope_. the observation ( aot : 12093696 ) was taken from the spitzer heritage archive ( sha ) as part of the `` infrared dark clouds in the inner galaxy '' program .
we used a pipeline partly based on the smart - package and the feps spectral extraction tools @xcite .
the background subtracted on - source spectrum and the background spectra from the off - source nod - positions is shown in figure [ fig : spec ] . only the long wavelength - order ( ll )
is reduced , since the slit for the short wavelength did not cover any point source in the irdc .
the ll slit did only cover about 60% of the smart - source psf , which leads to a flux leakage in the continuum flux .
but line shapes and line fluxes should be unchanged for the source spectrum , except for a lower s / n - ratio .
a strong [ siii ] emission appears over the whole irs - slit , leading to an artifact in the source spectrum due to an over - subtraction of the background .
the source spectrum was corrected for this artifact using a polynomial background fit(see section 3.2.3 in * ? ? ?
we utilized sub - millimeter data from the scuba legacy catalogue @xcite .
the 450 and 850@xmath19 m observations were part of the fundamental catalog with a effective beamsize of @xmath34 and @xmath35 , respectively . given the better sensitivity of the survey , the 870@xmath19 m observations from the `` apex telescope large area survey of the galaxy '' ( atlasgal , * ? ? ?
* ) were used in the temperature fitting ( section [ sec : tmap ] ) and to determine column densities ( section [ sec : colmap ] ) .
the corresponding beam size for the atlasgal data is around @xmath36 .
the rms noise level for atlasgal is @xmath37 .
we obtained molecular line data using the 22 meter mopra radio telescope and the mopra spectrometer ( mops ) , operated by the australian national facility ( atnf ) .
the observations were performed with the 3 mm band receiver ( 90ghz ) in the `` on - the - fly mapping '' mode ( otf ) with a beam size of @xmath38 at 86ghz and @xmath39 at 115ghz . using mops in the
`` zoom '' setup allowed us to observe 13 lines simultaneously ( @xmath40cs , c@xmath10h , ch@xmath41cch , ch@xmath41cn , h@xmath40cn , hco@xmath42 , h@xmath40co@xmath42 , hcccn , hcn , hnc , hnco , n@xmath10h@xmath42 , sio ) .
most of these lines are good tracers for dense gas ( see * ? ? ?
* for a detailed description of each molecular species ) .
however , the rms noise for all mopra maps is only slightly below 0.1k , that allows us to detect line emission above 3@xmath22 only in the hcn , hnc , hco@xmath42 and n@xmath10h@xmath42 maps . the spectrometer configuration results in a velocity resolution of @xmath43kms@xmath44 .
the main beam efficiency varies between @xmath45 at 86ghz and @xmath46 at 100ghz @xcite .
the typical system temperatures were between 210 and 290k .
the @xmath40co ( 1 - 0 ) data products were obtained from the galactic ring survey ( grs , * ? ? ?
* ) performed by the five college radio astronomy observatory ( fcrao ) . with a angular resolution of @xmath47
the beam of fcrao is somewhat larger than that of mopra .
the rms noise level in the grs map is below 0.9k .
we obtained ( 2 - 1 ) and ( 2 - 1 ) maps using the 230ghz heterodyne receiver array ( hera , * ? ? ?
* ) at the iram 30 m telescope .
the average fwhm of the hera beam is 113 , which is much better compared to mopra or fcrao .
the rms noise level is below 0.5 and 0.3k in the and hera maps , respectively . to acquire methylacetylene ( ) data observations
were performed with the iram 30 m telescope in position switching mode . for these observations
the emir receiver with fast fourier transform spectrometer ( fts ) as a backend was used .
total time on source and off source was 33min .
the off source position was chosen 600@xmath24 away from the on source position .
the on source position was pointed at @xmath48 @xmath49 ( fk5 , j2000 ) .
the pointing was checked every hour , giving a pointing accuracy better than 3@xmath24 .
the typical system temperature during observations was about 200k . at 153ghz
the telescope beam size is 16@xmath24 and the beam efficiency is 0.73 .
the data were analyzed using gildas / class software .
a list of physical parameters for each observed line is given in table [ tab : line_prop ] .
lcccccclll hco@xmath42 & @xmath50 & 89.188523 & 0.166 & 3.0 & 4.28 & @xmath51 & @xmath52 & mopra & 2010 jul 7 - 9 + hcn & @xmath50 & 88.631602 & 0.172 & 2.4 & 4.25 & @xmath53 & @xmath52 & mopra & 2010 jul 7 - 9 + hnc & @xmath50 & 90.663568 & 0.172 & 2.690 & 4.35 & @xmath54 & @xmath52 & mopra & 2010 jul 7 - 9 + n@xmath10h@xmath42 & @xmath50 & 93.173700 & 0.166 & 3.6 & 4.47 & @xmath55 & @xmath52 & mopra & 2010 jul 7 - 9 + @xmath40co & @xmath50 & 110.201354 & 0.166 & 0.0063 & 5.29 & @xmath56 & @xmath57 & fcrao & archive data@xmath58 + @xmath40co & @xmath59 & 220.398684 & 0.166 & 0.0604 & 15.87 & @xmath60 & @xmath61 & iram / hera & 2012 jan 23 + c@xmath62o & @xmath59 & 219.560319 & 0.163 & 0.0601 & 15.81 & @xmath63 & @xmath61 & iram / hera & 2012 jan 23 + & 9@xmath64 & 153.8172 & 0.142 & 8.75 & & & @xmath65 & iram / emir - fts & 2011 jun 10/12 + & 9@xmath66 & 153.8142 & 0.142 & 8.44 & & & @xmath65 & iram / emir - fts & 2011 jun 10/12 + & 9@xmath67 & 153.8054 & 0.142 & 7.55 & & & @xmath65 & iram / emir - fts & 2011 jun 10/12 + & 9@xmath68 & 153.7907 & 0.142 & 12.52 & & & @xmath65 & iram / emir - fts & 2011 jun 10/12 + ch@xmath41c@xmath10h & @xmath69 & 85.455665 & 0.142 & 0.1948 & 19.53 & & @xmath52 & mopra & 2010 jul 7 - 9 + h@xmath40cn & @xmath50 & 86.339860 & 0.169 & 2.2256 & 4.14 & @xmath70 & @xmath52 & mopra & 2010 jul 7 - 9 + h@xmath40co@xmath42 & @xmath50 & 86.754288 & 0.164 & 3.8534 & 4.16 & @xmath71 & @xmath52 & mopra & 2010 jul 7 - 9 + sio & @xmath59 & 86.846960 & 0.135 & 2.927 & 6.25 & @xmath72 & @xmath52 & mopra & 2010 jul 7 - 9 + c@xmath10h & @xmath59 & 87.316925 & 0.179 & 0.1528 & 4.19 & @xmath73 & @xmath52 & mopra & 2010 jul 7 - 9 + hnco & @xmath74 & 87.925238 & 0.137 & 0.9024 & 10.55 & @xmath75 & @xmath52 & mopra & 2010 jul 7 - 9 + hc@xmath41n & @xmath76 & 90.978865 & 0.125 & 5.812 & 24.01 & @xmath77 & @xmath52 & mopra & 2010 jul 7 - 9 + ch@xmath41cn & @xmath69 & 91.985316 & 0.140 & 6.0800 & 20.39 & @xmath78 & @xmath52 & mopra & 2010 jul 7 - 9 + @xmath79cs & @xmath59 & 92.494308 & 0.134 & 1.4124 & 6.66 & @xmath80 & @xmath52 & mopra & 2010 jul 7 - 9 [ tab : line_prop ]
the left panel of figure [ fig : rgb_overview ] shows the galactic neighborhood of 48 .
the image is composed of the 70 , 160 and 250@xmath19 m hi - gal maps in blue , green and red , respectively .
a global brightness gradient towards the outer parts of the galactic plane is present over the whole image . in the vicinity of 48
the gmc w51 is observed .
the cloud complex w51a and w51b have an angular distance of @xmath81 and @xmath82 to 48 , respectively .
the epos maps ( figure [ fig : rgb_overview ] , top right panel ) , in particular the pacs observations , show a brightness gradient towards w51 and the infrared bubble n100 @xcite .
w51 and 48 are not physically connected since recent observations place w51 twice as far away as 48 ( see next section ) in the _ spitzer _ irac 8@xmath19 m and the mips 24@xmath19 m channel the irdc appears as the typical dark filamentary absorption feature against the diffuse galactic emission background ( see figure [ fig : stamps ] ) .
48 is extended over @xmath83 in the north - south direction .
the dark cloud can be roughly divided into three parts .
the central region of the irdc appears as a elongated structure of @xmath84 in east - western orientation .
this region has the brightest sub - millimeter emission .
the western part of the central region shows the highest extinction in the _ spitzer _ maps .
only this part of the irdc appears in clear absorption in all three pacs channels .
furthermore this part is clearly associated with the most massive sub - millimeter clump , called `` extinction peak '' ( ep , with a mass of @xmath85m@xmath86 from scuba observation ) in @xcite .
the eastern part of the central irdc only shows clear extinction in the _ spitzer _ bands . in the _ herschel _
pacs maps this region is dominated by compact background emission from embedded point sources , in particular source 1 ( see section [ sec : pointsrc ] ) .
the central region , as the rest of 48 , appears as a diffuse emission structure in the spire maps , but two compact sources can be identified in the eastern central part .
the peak positions of these emission features detected at 250@xmath19 m and 350@xmath19 m are in good agreement with the peak positions of two sub - millimeter clumps ( p1 with 91 m@xmath86 and p2 with 100 m@xmath86 ) identified by @xcite .
due to the lower resolution of the 500@xmath19 m spire channel the two positions are not resolved .
the structures found south- and northwards of the central part of the irdc appear more diffuse in the _ spitzer _ bands and are barely recognizable in the pacs channels .
in contrast , in the spire bands this diffuse absorption features appear as an elongated emission structure . while this emission is diffuse in the northern part , compact emission is detected in good agreement with the scuba and atlasgal data in the southern part of the irdc .
the associated sub - millimeter peak was called `` p3 '' in @xcite ( see figure [ fig : stamps ] ) .
the lsr velocities found for the different clumps and molecular species do not differ by more than @xmath87 .
this results in a weighted mean velocity of @xmath88kms@xmath44 ( see section [ sec : molecline ] ) .
the resulting kinematic distance for 48 is @xmath89kpc .
this value is used for all calculations in this paper .
we derive kinematic distances using the @xcite algorithm , which takes into account trigonometric parallax measurements of masers in high - mass star forming regions in the galaxy , applied to a disk galaxy model .
we use their standard derived rotational parameters of @xmath90 = 254 km s@xmath44 and @xmath91 = 8.4kpc .
the hi - gal three color image ( figure [ fig : rgb_overview ] ) shows 48 at a close projected distance to the gmc w51 and several objects , as e.g. , hii regions , infrared bubble and other bright point - sources .
water maser observation estimates a distance of @xmath925.4kpc for w51 @xcite .
known distances of infrared bubbles ( sarah kendrew 2012 , private communication ) and sources from the rms survey catalog ( e.g. , massive yso , hii regions * ? ? ?
* ) in the vicinity of 48 have a distance of more than @xmath93kpc .
moreover , no co is detected in the global grs maps outside of 48 at its typical lsr velocities .
we therefore conclude that 48 is an isolated region making it an unique irdc .
to create temperature and column density maps of 48 we used _
herschel _ maps and ground based sub - millimeter data . in brief , the basic steps we have taken to fit the temperature and column density maps are as follows : 1 .
the global background flux offsets are estimated from the large scale hi - gal maps and subtracted from the maps .
2 . all data ( pacs at 160@xmath19 m , spire at 250 and 350@xmath19 m , scuba at 450@xmath19 m , and atlasgal 870@xmath19 m )
were convolved to a common limiting beam size ( spire 350@xmath19 m ) and converted to uniform units ( jy / arcsec@xmath94 ) .
the inclusion of the also available spire 500@xmath19 m data would force us to commonly convolve all other maps to it coarse resolution of around @xmath38 .
we therefore neglect these 500@xmath19 m data for the temperature mapping .
3 . in the final fitting process
an sed was fitted to each pixel using a modified black body function , requiring at minimum three data points for the sed . for the final pixel - to - pixel sed fitting a careful preparation of the initial data
is needed .
a complete description of the data treatment and fitting process is given in @xcite .
the observed globules in @xcite are more isolated and suitable background flux offsets can be calculated within the epos maps .
unfortunately , the larger extent of the irdc , the imprint of the adjacent gmc and infra - red bubble ( see section [ sec : morpho ] ) , and the limited map sizes prevent a reasonable background determination . to overcome these limitations we utilized _ herschel _ maps from the hi - gal program which cover a much larger area .
we extracted the background fluxes from a low - emission region off the galactic plane located at @xmath95 @xmath96 ( j2000 ) with a box size of 500@xmath97 . the actual method to extract the background fluxes in this field with relatively low spatial variations
is adopted from @xcite .
it is an iterative gaussian fitting algorithm applied to the pixel flux distribution in the selected region .
the best fit to the background levels is found by omitting the non - gaussian components .
finally background levels of 0.040jy / pixel and 1.434 , 1.191jy / beam were computed for 160 , 250 and 350@xmath19 m , respectively .
@xcite used a large scale gaussian profile to remove the galactic background from their observed hi - gal maps toward two different fields before the sed fitting . _ herschel _ temperature maps of a sub - region close to the star - forming complex w43 were created by @xcite without any global background subtraction . for lower temperatures ( @xmath98k )
both methods just differ by @xmath99k .
our maps appear to be more sensitive to the background determinationk inside the irdc ) .
therefore , we decided to estimate the background offsets further outside of the galactic plane .
] , however changing our estimated background level by @xmath100 does not have a significant impact on the derived temperatures . for the sed fitting all _ herschel _ maps ( pacs 160@xmath19 m , spire 250 and 350@xmath19 m ) were convolved to the spire 350@xmath19 m beam , using the kernels from @xcite .
for the sub - millimeter maps ( scuba 450@xmath19 m and apex 870@xmath19 m ) a gaussian kernel was used . for this step
all maps were calibrated in units of jy / arcsec@xmath94 . in the resulting maps the remaining emission per pixel is assumed to be well represented by modified blackbody emission : @xmath101 with @xmath102 where @xmath103 is the observed flux density for a given frequency @xmath104 and solid angle @xmath105 ( all fluxes are normalized to @xmath106 ) , @xmath107 is the optical depth , @xmath108 the planck function , @xmath109 the dust temperature , @xmath110 the proton mass , @xmath111 the dust - to - gas mass ratio , and @xmath112 the total hydrogen column density . for the column density map ( figure [ fig : nh2_tmap ]
, bottom left ) we assume that all hydrogen is present as molecular gas in 48 . for the dust opacity @xmath113
we adopt the model from @xcite for thin ice mantels and a gas density of @xmath114@xmath115 . for the hydrogen - to - dust mass ratio we adopted a value of @xmath116 ( e.g. ,
* ) . to account for helium and heavy elements one would choose the total gas - to - dust mass ratio 1.36 times higher
a modified black body function , as in equation [ eq : bbtmap ] was fitted individually for each pixel using @xmath117 and @xmath109 as free parameters with a @xmath118-minimization .
the hi - gal data includes only the 70 and 160@xmath19 m maps for the pacs bands .
to avoid contributions from stochastically heated small grains or optical thick emission , we exclude the 70@xmath19 m band ( e.g. , * ? ? ?
* ; * ? ? ?
usually the fluxes in the three remaining _ herschel _ bands ( 160 , 250 , and 350@xmath19 m ) are well constrained and provide a good fit to the modified black body function .
therefore , every shown pixel in the temperature map ( figure [ fig : nh2_tmap ] ) results from an sed fit of at least these three herschel bands .
if one of the submillimeter ( 450 or @xmath119 m ) fluxes deviate by more than @xmath120 from the sed - fluxes constrained in the other bands it is not used in the temperature and column density calculation .
the temperature map ( figure [ fig : nh2_tmap ] , bottom right ) shows a steep drop at the boundaries of 48 within the resolution limits set by the size of the convolution kernels ( a 19k contour is overplotted on the irac map at 8@xmath19 m and pacs at 100@xmath19 m in figure [ fig : stamps ] ) . in the central region of 48 , with the strongest sub - millimeter emission , the temperatures drop below 18k ( white contour in figure [ fig : nh2_tmap ] , bottom right ) .
the local minima agree with the position of the sub - millimeter peaks p1 ( 17.1k ) , p2 ( 16.8k ) , and ep ( 17.1k ) . in general our observed line - of - sight temperatures are higher than the ones ( 10 - 15k ) found by @xcite for p1 , p2 , and ep .
however , the later temperature profiles from @xcite are obtained from fitting observational results to core models ( similar to * ? ? ?
* see section [ sec : temp_disc ] ) . to assess the uncertainties of our line - of - sight - averaged temperatures
, we investigated how the change of different parameters influence the resulting temperatures and column densities .
when altering the subtracted global background level , that is subtracted from the herschel maps , by @xmath100 before the fitting process and using different beam sizes for the convolution , the resulting temperatures are found to differ by no more than 0.3k inside the irdc , where dust emission is detected in the atlasgal maps . in the temperature fitting routine , we assume mean absolute background flux levels that were derived for regions well outside the galactic plane ( see * ? ? ? * ) and might not be appropriate for this irdc . for lack of better estimates of the local absolute background levels
, we have tested the effect of simply amplifying those background levels by a factor of 100 . the resulting temperatures and column densities in the irdc
were found to change by no more than 0.1k and 2.5% , respectively .
the effect would be stronger for lower column densities , which we do not consider here .
hence , we can robustly conclude that the effect of the uncertain knowledge in the absolute local background flux levels on the derived temperatures and column densities in the irdc is smaller than the uncertainties introduced by the calibration , data reduction , and dust opacity models , and is thus negligible .
the total relative uncertainties introduced by calibration and data reduction are found to be 2.5% for the temperatures and
10% for the column densities .
the largest uncertainties are introduced by the uncertain knowledge of the underlying dust opacity model .
therefore , we assume an overall temperature uncertainty of @xmath121k . a more detailed discussion of the different contributions to the uncertainties can be found in @xcite .
it should be mentioned that the temperatures outside 48 are not very reliable because of the brightness gradient introduced by w51 and n100 , as mentioned earlier . in the column density and temperature maps ( figure [ fig : nh2_tmap ] )
the point source positions and temperatures are indicated .
the temperatures derived at the locations of the point sources in the large scale temperature map are discrepant from the temperatures derived by fitting the individual seds of said point - sources , the later of which were extracted at native resolution .
this discrepancy is not surprising as it is caused by beam convolution in the temperature map and the inclusion of longer wavelengths . to estimate the column densities for 48 we utilized three different methods ; mid - infrared extinction map from irac at 8@xmath19 m , sub - millimeter dust emission and modified blackbody fitting including _ herschel _ and sub - millimeter data ( see section [ sec : tmap ] ) .
the irac 8@xmath19 m extinction maps provides us with an estimate of the optical depth per pixel ( see section [ sec : obs_spitzer ] ) .
these optical depths were converted into column densities ( figure [ fig : nh2_tmap ] , bottom left ) using @xcite : @xmath122 where @xmath123 is the optical depth per pixel and @xmath124 is the extinction cross - section .
the nature of the mid - infrared extinction law below 10@xmath19 m in irdcs and massive star - forming regions is still a matter of debate ( e.g. , * ? ? ?
* ) . in agreement with findings of @xcite we used an extinction factor of @xmath125 to parametrize the extinction cross - section @xmath124 .
we choose @xmath126 from ( * ? ? ?
* ( case b ) ) .
the derived column densities also depend on the assumed @xmath127 value and , therefore , the used extinction cross - section .
the extinction factor @xmath128 is more representative of the diffuse ism and would result in significantly higher column densities .
another effect for the 8@xmath19 m extinction is the saturation of the column density for high values , therefore the masses based on irac 8@xmath19 m discussed further below can only be lower limits ( for a detailed discussion , see * ? ? ?
the second method to determine the hydrogen column densities utilizes the dust emission detected in the 870@xmath19 m map from atlasgal : @xmath129 where @xmath130 is the sub - millimeter flux , @xmath116 the gas - to - dust ratio , @xmath131 the planck function for a given wavelength ( 870@xmath19 m ) and dust temperature ( @xmath132k or 19.2k , see below ) , @xmath133 the mass per hydrogen molecule , and @xmath105 the beam size .
we adopt a mass absorption coefficient of @xmath134cm@xmath94g@xmath44 from ( * ? ? ?
* thin ice mantel , @xmath136@xmath115 , as in section [ sec : tmap ] ) .
we have three independently derived column density maps : irac 8@xmath19 m , atlasgal 870@xmath19 m , and the sed fitted map combining _ herschel _ and atlasgal .
each one requires us to make different and independent assumptions in order to derive column densities ; furthermore , they also vary widely in angular resolution .
for example , the 8@xmath19 m extinction map , which has a resolution of @xmath137 , requires an assumption of a certain extinction factor for the dust extinction cross - section , optically thin emission and the local background contribution ( @xmath138 in equation [ eq : irac_tau ] ) , but is temperature independent .
furthermore , the residuals of the psf removal for bright irac point - sources with compact background emission can result in an underestimate of local column density ( see e.g. , source 1 in figure [ fig : nh2_tmap ] , top left ) . for the atlasgal 870@xmath19 m map , with a resolution of @xmath139 , always a fixed temperature is assumed , it requires optically thin emission and a dust opacity choice at the wavelength of interest .
while the sed - fitted map also requires a dust model assumption , it contains information from multiple wavelengths and simultaneously derives @xmath140 and temperature for each pixel , under the assumption that the mass is well - represented by a modified black body and assuming a fixed spectral index @xmath141 .
therefore the derived column densities can be considered more robust compared to the two previous methods .
however , the sed - fitted map also has the lowest resolution of the three methods considered here , with final convolved beam size of 25@xmath24 , equivalent to @xmath142pc at the distance of 48 .
here we discuss the integrated masses for these three independent methods .
the total mass of 48 can be calculated within a contour of interest using the column densities mask ( shown in figure [ fig : nh2_tmap ] ) : @xmath143 where @xmath144 is the pixel area , @xmath145 and @xmath146 the column density per pixel .
the masking contours for the irac @xmath147 m extinction map were chosen by visual inspection .
this contour results in a threshold of @xmath148 , which correspond to a column density of @xmath149 . a total mass of @xmath150m@xmath86 is enclosed in these contours ( see figure [ fig : nh2_tmap ] , top left panel ) .
we assume a temperature of 17.3k ( the median point source temperature , see section [ sec : pointsrc ] ) for the atlasgal derived column density map ( figure [ fig : nh2_tmap ] ; top right panel ) .
the column density threshold is adopted to be at 3@xmath151the rms noise level ( @xmath152mjy / beam ) , corresponding to @xmath153 .
the resulting total mass inside this column density contour is @xmath154 . for comparison , the median _ herschel _ derived temperature ( figure [ fig : nh2_tmap];bottom right panel ) in this same region is 19.2k .
if we adopt this temperature instead of 17.3k , as assumed above , the column density threshold becomes @xmath155 and the total enclosed atlasgal mass is @xmath156 . for reference ,
applying a column density threshold of @xmath157 ( derived from the from irac 8@xmath19 m column density map ) is equivalent to applying a 18k isotherm to the _ herschel _ derived column density map , and results in thresholds that encompass approximately the same spatial extent of the three scuba peaks from @xcite .
the mass enclosed within these thresholds differ by only @xmath158 compared to the total mass ( 320 m ) of these three peaks .
the total mass for 48 derived from the 8@xmath19 m extinction ( @xmath159 ) and the 870@xmath19 m ( @xmath160 ) emission are remarkably similar .
furthermore , the average column density threshold for both maps corresponds to a _
herschel _ derived total mass of @xmath161 , also in excellent agreement with the other estimates . considering the uncertainties in the various assumptions , such as the dust properties , uncertainties in background flux levels , and variations in beam sizes we find that our mass agreement for 48 is strikingly good across different methods and compared to previous work .
in general the column density for the diffuse emissions in 48 are comparable for all three methods . for ep , p1 , p2 , p3 and p4 the peak column densities agree within 18% for the atlasgal and the _ herschel _ maps .
however , the peak column densities derived from the 8@xmath19 m extinction are significantly higher then the fitted _ herschel _ maps .
the peak columns can be higher up to @xmath162 .
this might be an effect introduced by the higher spatial resolution of irac .
ccc|rrrr|rrrr|r[th ] 1 & 19:21:49.65 & + 13:49:31.26 & 102.7 & 1764 & 4086 & 6931 & 18.3 & 48.2 & 14.3 & 36.1 & s5 , p1 + 2 & 19:21:50.24 & + 13:49:39.28 & 241.1 & 3702 & 6931 & 10712 & 17.9 & 35.5 & 11.9 & 38.0 & s6 , p1 + 3 & 19:21:50.26 & + 13:48:18.96 & 119.5 & 1730 & 3659 & 4745 & 20.9 & 17.7 & 2.4 & 14.6 & s2 , p3 + 4 & 19:21:47.95 & + 13:49:20.37 & 2.8 & 460 & 1807 & 4717 & 17.2 & 15.1 & 6.3 & 2.8 & s15 , p2 + 5 & 19:21:50.13 & + 13:47:57.00 & 27.7 & 474 & 902 & 1474 & 19.8 & 5.5 & 1.0 & 4.5 & s1 + 6 & 19:21:40.76 & + 13:51:17.42 & 102.7 & 365 & 494 & 693 & 16.8 & 2.4 & 1.2 & 6.5 & + 7 & 19:21:45.58 & + 13:49:20.51 & 8.7 & 299 & 842 & 1370 & 19.5 & 4.8 & 1.0 & 1.7 & s12 , ep + 8 & 19:21:48.48 & + 13:49:25.26 & - & 210 & 865 & 4668 & 17.4 & 13.3 & 5.3 & - & p2 + 9 & 19:21:53.01 & + 13:48:19.15 & 34.7 & 139 & 174 & 357 & 16.5 & 1.1 & 0.6 & 2.8 & s3 + 10 & 19:21:36.30 & + 13:52:20.23 & 31.8 & 132 & 780 & 2209 & 16.8 & 7.9 & 3.8 & 1.6 & s11 + 11 & 19:21:35.95 & + 13:51:59.97 & 115.5 & 121 & 177 & 521 & 15.6 & 1.7 & 1.3 & 4.4 & s9 + 12 & 19:21:40.73 & + 13:50:54.91 & 20.2 & 115 & 303 & 894 & 15.6 & 3.2 & 2.4 & 1.5 & s13 + 13 & 19:21:31.93 & + 13:52:51.82 & 4.6 & 61 & 131 & 516 & 14.5 & 1.7 & 1.9 & 1.0 & [ tab : pnt_src ] in table [ tab : pnt_src ] we present the fluxes of the 13 point sources which were detected in all three pacs bands . except for source 8 ,
all pacs detections have counterparts in mips 24@xmath19 m .
source 13 was not included in the @xcite sample , but is detected as a point source at 24@xmath19 m ( e2 in figure [ fig : stamps ] ) .
if one allowed point source detection in only two pacs bands this would add only one more obvious point source .
this source is also detected at 24 micron ( e1 ) but appears as a diffuse blob of emission at 100 micron , not well - matched to the characteristic psf we expect .
however , the projected distance to source 13 ( e2 ) is @xmath163pc which is close to the median separation discussed in section [ sec : conclusion ] , and therefore would not change the outcome of this discussion .
all detected sources are aligned along the mid - infrared filaments of the irdc .
even though the emission gradient towards the gmc ( on the eastern edge of our maps ) is brighter and contains some compact emission features , no pacs point sources are detected outside the irdc filaments .
pc ) , are plotted against bolometric luminosities .
bolometric luminosities are derived from irac , mips and pacs bands .
the numbers identify the point sources characterized in section [ sec : pointsrc ] .
the light - to - mass ratio derived from the sub - milliliter observation for different peaks ( p1,p2 and ep ) is plotted as well . in this
observations the area of the traced mass reservoir is larger .
the dashed lines represent the evolutionary tracks for yso from @xcite .
the finale stellar masses ( zams ) of those tracks are given above the lower axis . ] as in section [ sec : tmap ] , we used a modified blackbody function to constrain the parameters of these point sources ( figure [ fig : blackbodies ] ) , utilizing the dust model by ( * ? ? ? * thin ice mantles , @xmath114@xmath115 ) .
the sources detected at 24@xmath19 m can not be fitted by a single component fit , since the warmer part of the envelope in the direct vicinity of the evolving protostar is contributing significantly to the 24@xmath19 m fluxes . due to the rapidly increasing optical depth towards the innermost regions around the point - sources we only observe surface contributions of surrounding layers of warm dust .
the calculations of mass and temperature assume optically thin material .
the 24@xmath19 m data are highly sensitive to geometric effects in protostellar cores ( see * ? ? ?
* ; * ? ? ?
a proto - star in which outflows have carved a cavity in its natal core will be detectable at 24@xmath19 m if our line of sight is aligned with the cavity .
in particular the shocked molecular hydrogen , visible as green structures in figure [ fig : rgb_overview ] ( right , bottom panel ) , traces high velocity outflows .
they originate from the rotational s9-s11 transition of excited in the irac 4.5@xmath19 m band .
they are generally referred to as either `` green and fuzzy features '' @xcite or `` extended green objects '' ( egos , * ? ? ?
* ) . because the 24@xmath19 m flux arises primarily from the warm material immediately surrounding the proto - star ( and not the cold outer region of the core )
, we do not include the 24@xmath19 m flux in our sed fit to the pacs data but instead fit a second modified blackbody component ( see figure [ fig : blackbodies ] ) .
the best fit to the sed results from the sum to these two components .
for source 8 , we used one modified blackbody function .
the fit parameters are presented in table [ tab : pnt_src ] .
the four most massive sources in our sample are located in the central part of the irdc in the direct vicinity of the sub - millimeter peaks p1 and p2 .
they cover a mass range from 5.4 to 14.3m@xmath86 . from these sources ,
source 8 , is the only pacs source without a mips 24@xmath19 m counterpart .
source 1 and 2 are the brightest and most massive sources in the sample .
they are associated with the scuba peak p1 and have a close separation of @xmath164 .
source 1 appears on top of a compact and diffuse background emission complex .
only a faint source with 0.9m@xmath86 is associated with the ep - scuba - peak of @xcite . the point sources associated to p3 in the southern part of 48 have only @xmath165m@xmath86 . in figure
[ fig : lbol_vs_menv ] we present the bolometric luminosities ( irac to pacs ) , the envelope masses ( @xmath166pc ) , and the temperatures derived from cold components of the modified blackbody fits . in the luminosity - mass plot
the dotted line follows the relation @xmath167 and corresponds approximately to a threshold , where 50% of the initial core mass being converted into stellar mass @xmath168 ( c.f .
* ; * ? ? ?
sources with @xmath169 and a high sub - millimeter to bolometric luminosities ratio @xmath170 are considered to be class 0 @xcite .
source 1 and 2 as well as source 4 and 8 are detected as single spire sources .
but the fitted blackbodies indicate that the sub - millimeter fluxes should be equal for both sources in each case . for each source
we calculate the sub - millimeter luminosity @xmath171 with 50% of the corresponding spire fluxes .
all four sources mentioned before are good class 0 candidates since @xmath172 and @xmath169 .
source 1 and 2 have the highest envelope masses , and , as indicated by evolutionary tracks ( adopted from * ? ? ?
* ) , both could evolve into high - mass zero age main sequence stars ( zams ) with @xmath173 .
these evolutionary tracks assume exponentially declining accretion rates .
a recent study by @xcite , comparing observations to various accretion models , showed that models with increasing accretion are in best agreement with the observational results .
@xcite used a model with increasing accretion rates to create similar evolutionary tracks . according to these tracks source 1 and 2
would only evolve into intermediate - mass stars ( @xmath174 ) .
however , it is still unclear on which spatial range a proto - star is accreting material .
source 1 and 2 are associated to the sub - millimeter peak p1 . due to the resolution of scuba compared to pacs a larger area
is traced .
given the light - to - mass ratio for p1 derived by @xcite , still two massive stars ( @xmath173 ) could form in the model from @xcite . assuming these sources are in a very young evolutionary stage , they are completely dominated by the accretion luminosity @xcite . to determine the parameters from the mapping line data , we used the gildas / class package .
as mentioned before we have line detections ( cl@xmath175 ) associated to the irdc in the hco@xmath42 , , hcn , hnc , @xmath40co and c@xmath62o maps . while the line fitting for hco@xmath42 , hnc and co
is done straightforward with a single gaussian profile , n@xmath10h@xmath42 and hcn have a resolved hyperfine structure with multiple components .
for n@xmath10h@xmath42 a model profile containing seven hyperfine - structure components was fitted utilizing the gildas / hsf routine .
the presented n@xmath10h@xmath42 maps containing the integrated line flux , line width and velocity correspond to the line center of the strongest component ( 93.17378ghz , @xmath176 , @xmath177 , @xmath178 ) .
the hyperfine - structure of hcn shows a triplet of components , and in contrast to n@xmath10h@xmath42 , the relative line strength of each component varies for different irdcs @xcite .
non - lte effects especially in the 10 lines seem common ( see * ? ? ?
given the noise level of the hcn map we could not retrieve an averaged profile .
the fitting of the physical parameters is done on the strongest component .
rcc[th ] n@xmath10h@xmath42 & 5.0 & 5.0 + hco@xmath42 & 5.8 & 5.8 + hcn & 8.5 & 3.4 + hnc & 18.6 & 9.3 + @xmath40co ( 1 - 0 ) & 35.2 & 2.3 + @xmath40co ( 2 - 1 ) & 31.4 & 6.3 + c@xmath62o ( 2 - 1 ) & 8.8 & 5.5 [ tab : clumpfinder ] the derived integrated line fluxes , velocities and line widths are compiled in table [ tab : radio_lines ] .
this table gives also the effective radius ( see below ) , the total column density , the abundance and the traced mass . and are good tracers for dense gas because of their high critical densities .
has a lower critical density ( see table [ tab : line_prop ] ) , but is not strongly effected by depletion from freeze - outs on grain surfaces . for co
the critical density is , depending on the transition , one or two orders of magnitudes lower than for .
the total column density for a certain molecular species can be calculated with @xmath179 @xmath180 where @xmath181 is the einstein coefficient , @xmath104 the rest frequency of the transition , @xmath182 and @xmath183 the upper state column density and energy , @xmath184 the rotational partition function and @xmath185 the statistical weight of the upper level .
the general physical parameters ] for the observed lines are given in table [ tab : line_prop ] .
the abundance for a molecule is given in relation to the h@xmath10 column density derived from the sed - fitted map from section [ sec : tmap ] .
the temperature estimates needed to calculate the different molecular abundances were obtained from the same sed - fitting .
the line emissions from , , and can be optical thick , therefore the column densities and abundances provide only lower limits . for the molecular line maps without any @xmath186 detection the upper limits for line intensity and column density , based on the noise estimates ,
are given in table [ tab : radio_lines_upperlimits ] .
rcc[th ] ch@xmath41c@xmath10h & 0.50 & + h@xmath40cn & 0.53 & 3.8 + h@xmath40co@xmath42 & 0.52 & 0.7 + sio & 0.51 & 1.2 + c@xmath10h & 0.52 & + hnco & 0.48 & 6.4 + hcccn & 0.48 & 2.0 + ch@xmath41cn & 0.50 & 2.9 + @xmath40cs & 0.53 & 5.9 [ tab : radio_lines_upperlimits ] following @xcite the mass of each clump observed in the line maps is estimated using an assumed distance of 2.6kpc , an approximate size based on the extent of molecular emission , the molecular column densities , and approximate abundance calculated at the peak of absorption . in figure
[ fig : mopra ] the spire 250@xmath19 m images are shown and overplotted by the different line intensity maps ( , , , and ) .
we used the ` clumpfind ` @xcite package to identify the different clumps in the molecular tracers .
four different clumps were found by clumpfinder , hereafter named c1-c4 .
the detection parameters for ` clumpfinder ` are given in table [ tab : clumpfinder ] .
c1 , c2 and c3 are detected in hco@xmath42 , hcn and .
an additional clump c4 is detected in hcn . in the @xmath40co ( 1 - 0 ) grs map
only c1 is detected . while c1 and c2 are detected in the hera co ( 2 - 1 ) maps and in hnc . because ` clumpfind ` does not make any assumption about the clump morphology an effective circular radius @xmath187{a/\pi } \label{eq : reff}\ ] ]
is given in table [ tab : radio_lines ] , where @xmath188 is the area fitted by ` clumpfind ` . because the resolution of the hera observations is three times better than mopra , the emission is less diluted and we therefore derive smaller @xmath189 .
all line maps show the strongest emission in the central part of the irdc .
the maps hco@xmath42 , , hnc , and @xmath40co show an elongated structure in east - western orientation .
hcn shows two distinct peaks ( c1 , c4 ) ; these peaks are shifted by @xmath190 south - west of the compact spire emission .
all other mopra molecular tracers peak at c1 , and while the maps appear shifted from the compact spire emission , it is within the pointing error of mopra .
the peak positions for the hera co maps are much better aligned with the spire peaks .
in contrast to our observed hco@xmath42 ( 1 - 0 ) emission , the hco@xmath42 ( 3 - 2 ) observation from @xcite show multiple emission peaks .
their eastern emission peak is centered on the scuba clump , but the western emission is more diffuse . with the given spatial resolution of mopra ( @xmath52 )
it is likely that we only observe a single integrated emission peak for hco@xmath42 from these multiple unresolved emitting sources . about @xmath191 southwards of the central region another fainter emission peak ( c2 )
can be found for hco@xmath42 , hcn , hnc , , @xmath40co(2 - 1 ) and c@xmath62o .
it is not coincident with the compact spire and sub - millimeter emission in this region . in the integrated line flux map for @xmath40co(1 - 0 ) , no compact emission clump was found by ` clumpfind ` .
the emission appears just as a diffuse extended lobe from the central peak . in the north - western part of the irdc , where only diffuse emission appears in the spire and sub - millimeter regime , the line data are much more noisy . for hco@xmath42 , hcn and a compact emission peak ( c3 )
can be identified , while @xmath40co shows a elongated diffuse emission lobe .
the three scuba clumps modeled by @xcite incorporate a mass reservoir of @xmath192 .
the masses derived for this region from the hco@xmath42 , hcn , hnc , n@xmath10h@xmath42 and @xmath40co observations differ by up to 20% .
given the uncertainties for the mass estimates , this is a good agreement .
is a good temperature probe in dense molecular clouds ( e.g. , * ? ? ?
we detect four transitions : @xmath193 , @xmath194 with iram single pointing observations positioned on the sub - millimeter peak p1 ( as indicated in the bottom left panel of figure [ fig : nh2_tmap ] ) . from the measured integral line intensities we derived the column densities for every particular transition using formulae 1 - 2 from @xcite .
finally we computed the excitation diagram and estimated the excitation temperature and column density in the region .
the resulting temperature is @xmath195 and the column density @xmath196 .
all necessary values , like @xmath183 - the upper state energy , @xmath185 - degeneracy and others , were taken from the cdms data base @xcite co ( left ) and c@xmath62o ( right ) is plotted over the inversed pacs 70@xmath19 m image . for @xmath40co ( hera )
the blue contour is integrated between @xmath197 and the red contour between @xmath198 .
the integration in the c@xmath62o hera map is done over a more narrow range , due to higher noise level , with @xmath199 ( blue ) and @xmath200 ( red ) .
the middle plot shows the linewidth for the hera co data .
the overplotted contours are the isotherms of 17.5 , 18.5 , 19.0 , and 19.5k ( see figure [ fig : nh2_tmap ] ) .
the bottom plot shows the position - velocity diagram for @xmath40co and c@xmath62o .
the spatial orientation of the p - v - plot is indicated in the top panels by a black vertical line . ] in figure [ fig : hera_co ] ( bottom panel ) the position - velocity ( p - v ) plots are shown for @xmath40co and c@xmath62o ( 2 - 1 ) . in the p - v plot for c@xmath62o only a single velocity component is present . for @xmath40co two components
are present at a 50@xmath24 offset , which refers to the spatial position of the source 1 in the pacs points - source catalog and the sub - millimeter peak p1 . due to the small spread of just @xmath201
it is questionable that this structure originates from an outflow cone .
other possible explanations could be a large rotational envelope or just two core - like dense co structures along the line - of - sight . with the lack of additional observations with higher sensitivity and spatial resolution , no robust conclusion can be drawn .
the linewidth maps in figure [ fig : hera_co ] are fitted with a single gaussian .
a single gaussian profile is also fitted there two components are found in the map ( figure [ fig : hera_co ] , middle plot ) .
the resulting does not differ more then @xmath202 from a fit only on the strongest component . with typical isotopic abundance ratios of @xmath203/[\ce{^{13}c}]=60}$ ] and @xmath204/[\ce{^{18}o}]=500}$ ]
@xcite the hera and abundances correspond to abundances of @xmath205 .
the expected co abundance at a given galactocentric distance ( @xmath206 ) can be computed from the following relationship ( * ? ? ? * and ref . within ) :
@xmath207 for 48 with @xmath208kpc this results in an expected co abundance of @xmath209 thus consequently @xmath210 and @xmath211 . comparing this expected values with the observed co abundances gives the depletion factor : @xmath212 using the preceding isotopic abundance ratios results in average depletion factors of @xmath213 and @xmath214 for 48 . we detected ice in the _ spitzer _ spectra . in figure
[ fig : spec ] the _ spitzer_/irs spectrum for source 2 is shown .
the observed shape of the ice absorption feature at 15.2@xmath19 m is different for the source spectrum compared to the background spectra ; the on - sources profile is significantly broader and different in shape compared to the nodded background profiles .
this implies different ice temperatures and mixing ratios ( with e.g. , and , see * ? ? ?
the low spectral resolution of the irs / ll2 spectral order does not allow a proper fitting of temperature and mixing ratio to the laboratory data .
most likely the source profile originates from ice coated grains in the dust envelope of the embedded proto - star .
the presence of -ice does not provide direct evidence for freeze - out , since the conditions under which the surface conversion @xmath215 co2 + h}}$ ] takes place are still under debate ( e.g. , * ? ? ?
* ) . in the gas - phase
is predominantly produced by the following reaction @xmath216 to the last two reactions an activation energy barrier of 80k or higher is assigned @xcite .
therefore , gaseous creation becomes effective in cores which already have an internal heating source .
the ice can not be easily hydrogenated on dust grain surfaces though , so it will remain a terminal species of the co evolution in lukewarm regions ( dmitry semenov 2012 , private communication ) .
48 is a quite isolated irdc at a distance of 2.6kpc .
consequently , 48 provides nonpareil observations of the collapse and fragmentation of a irdc without the effects of external feedback , except for the diffuse isrf .
we have derived line of sight averaged _ herschel _ column density and temperature maps for this source , and find minimum temperature of 16.8k , and outer temperatures of @xmath217k .
the corresponding column densities range from @xmath218 at in central part of 48 to @xmath219 at the outer edges where our sensitivity drop sharply . in this section
we will discuss these results in combination with a wealth of other observations with the goal placing this unique source in an evolutionary context and determining its physical properties .
the comparison of the line data for 48.66 with other irdcs show results for which are similar to other irdcs , while the results are different .
the average line width of is 2.5kms@xmath44 which is lower compared to other irdcs , with typical values of @xmath220 @xcite .
for we observed an average line width of 1.8kms@xmath44 .
similar mean line widths were observed for other irdcs with 1.8kms@xmath44 by @xcite and 2.0kms@xmath44 by @xcite .
the mean column density of @xmath221 for in our sample is similar to the column density found for other irdcs by @xcite . comparing the amount of freeze out
one finds that the depletion factor is strongly dependent on the chemical age of the irdc , hence the depletion varies drastically for different irdcs . @xcite and @xcite reports a depletion factor of 5 for g035.39.-0033 which is a young irdc of @xmath222yr similar to 48 .
in contrast @xcite found no depletion for their sample , and argued that this it is connected to a more evolved stage of their irdcs where the increased star formation activity leads to sublimation of from the grain surfaces .
@xcite reports higher values of @xmath223 for several southern irdcs , the reason for this higher depletion factors is not completely clear , but might be explained by tracing denser regions of the cores by utilizing higher 18o ( 3 - 2 ) transitions , compared to our 18o ( 2 - 1 ) observations . in the gaseous inter clump medium is destructed in the following reaction : .
the presence of is consistent with the low abundance induced by the freeze - out on the dust - grain surfaces . by comparing their observations to clump - models @xcite found an increase of turbulent motions towards the inner part of the clump .
their observations have higher resolution by a factor of two .
therefore , our observations are not good enough to trace down turbulence in the same way , but the increasing mopra linewidths inwards of the clumps are in agreement with the findings of @xcite .
our hera data have a higher resolution than their observations , but due to the low critical density of we trace a larger column that might be dominated by envelope material also along the line of sight .
therefore , does not qualify as a good turbulence tracer in dense environments as in our observations . to assess the dynamical state of the molecular clumps we calculate the virial mass , using the line width @xmath224 : @xmath225 whereas @xmath226 for a @xmath227 profile @xcite , @xmath111 is the effective radius of the molecular clumps , and @xmath224 is the fwhm of the respective line in @xmath228
. we also calculated the virial parameter @xmath229 which is given by the ratio between the kinetically energy @xmath230 and the potential energy @xmath231 : @xmath232 a value of @xmath233 represents a virial equilibrium , while @xmath234 is present for pressure confined clumps @xcite , @xmath235 indicates a gravitationally unbound clump , and for @xmath236 the clump is collapsing .
the virial mass and the virial parameter are presented in table [ tab : radio_lines ] .
we see a large range , from less than 1 to over 7 , in virial parameters depending on the species , making it difficult to assess the stability of each clump with certainty .
however , while c1 seems to be off equilibrium and just confined by the surface pressure @xmath237 , c2 and c3 appear to be gravitationally unbound ( @xmath238 and @xmath239 , respectively ) .
another crucial parameter is the virial line mass .
an isothermal , nonmagnetic , self - gravitating filament in the absence of external pressure is in hydrodynamical equilibrium as long as the mass per unit length does not exceed @xmath240 , where @xmath22 is the velocity dispersion in the gas .
if the filament is only thermally supported the velocity dispersion is given by the thermal sound speed @xmath241 @xcite , while in the case of additional turbulence support the velocity dispersion is given by @xmath242 @xcite , where @xmath224 is the velocity dispersion .
the critical line mass for a thermally supercritical filament in 48 is @xmath243m@xmath244 , whereas the critical line mass calculated by the mean linewidth of 48 , to account also for turbulent motions , is @xmath245m@xmath244 . to highlight the derived linemass , we straightened the filament of 48 utilizing a set of non - uniform cubic splines @xcite and calculated the mass in each pixel . binning the pixels results in an average line mass of 174m@xmath86/pc
( see figure [ fig : linemass ] ) .
the observed line masses are @xmath246 larger than the critical line mass , indicating that the 48 filament does not contain enough internal support to counter the collapse while the contributions from magnetic fields are expected to be negligible ( see below ) .
line masses observed for other irdc seem to exceed @xmath247 .
however , for strongly fragmented cores with significantly lower mass in taurus line masses similar to @xmath247 , hence predominately thermal support , were observed @xcite .
in contrast to 48 several irdcs show line masses below or similar to @xmath248 .
this can not be explained by the evolutionary stage of the irdcs alone .
the young and starless coalsack cloud and the adjacent g304.74 + 01.32 show line masses of less or equal to @xmath248 ( * ? ? ?
* respectively ) . also @xcite found the young h filament of g035.390.33 to be virial undercritical , although these results rely several uncertain morphological assumptions about the irdc .
48 and the nessie irdc are more evolved , showing signs of more active star formation such as `` green and fuzzy '' features , but the latter region still shows line widths a factor of 5 below the critical value for turbulence support @xcite
. furthermore the presence of magnetic fields does not change the critical line mass significantly @xcite . however
, the higher line mass of 385m@xmath86 for the more evolved orion a cloud @xcite supports the view of increasing line masses along the evolutionary sequence . in the central part of the irdc a concentration of point sources
the projected median separation of pacs - point - sources is @xmath249pc , which is approximately the average separation in the central part of 48 .
this separation is about three times smaller than in other irdcs .
the median separation in the r1 fragment of the coalsack dark cloud is even an order of magnitude higher compared to 48 @xcite , although here this study uses gas / dust peak separations instead of point source separation . the key question is how to explain these different values .
one could assume that we just suffer from a projection effect in 48 .
considering we see an elongated filament and given the observed morphology it is a feasible assumption that 48 is aligned approximately face - on towards the line - of - sight . even if one would assume that 48 is aligned off by @xmath250 this would just account for a factor of @xmath251 .
in contrast to 48 the mass reservoir in the irdcs mentioned above is significantly higher ; these other dark clouds do not share the isolated nature of 48 . while external feedback and the overall evolutionary stage of the parent cloud might have an important impact on the fragmentation , evaluation of the entire high - mass part of the epos sample in @xcite
show that the average point source separation is not correlated with the cloud mass ( pitann et .
al , in prep . ) .
recent molecular line observation of g28.34 + 0.06 found a separation of just 0.16pc @xcite to 0.19pc @xcite .
the interferometric data analyzed in @xcite show substructures on a @xmath252pc scale inside the cores .
given the wide spread of source separation it should be discussed whether 48 exhibits the same mode of collapse as the irdcs with larger separation or if some substantially different physical processes can account for the smaller separation . to address this question
we compare the observational results to analytical considerations .
a first simple analysis is to compare the source separation to the jeans length .
the separation in 48 is still several times larger than the thermal jeans length ( @xmath253 at @xmath254 , 19k ) .
therefore , a thermal jeans instability is not the dominant fragmentation mode .
fragmentation below this spatial scale has been observed for several massive cores ( e.g. , * ? ? ?
* ) . a way to assess the spatial separation scale is to consider the collapse of a cylinder ( with a radius @xmath255 ) of an incompressible fluid .
it is predicted that several cores form , periodically placed at a characteristic length scale @xmath256 @xcite . for an infinite cylinder of isothermal self - gravitating gas
the characteristic fragmentation scale is @xmath257 , where @xmath258 is the isothermal scale height @xmath259 and @xmath241 the thermal sound speed , @xmath260 the gravitational constant and @xmath261 the central density in the cylinder @xcite .
when a isothermal cylinder with a given radius is embedded into a uniform medium the separation is @xmath262 for @xmath263 , but for the opposite case @xmath264 the length scale is @xmath265 @xcite . given the measured temperatures the typical thermal sound speed is @xmath266kms@xmath44 in 48 .
with @xmath267pc the width of the filament is much larger than the scale height .
the calculated length scale @xmath268pc ( with @xmath269@xmath115 , @xmath266kms@xmath44 ) is the same as the measured separation in 48 .
however , @xmath270 is affected by uncertainties , since the real value of the central density is unknown .
moreover , the most regular positioning of sources occur in the central part of the irdc along a clearly changing temperature profile which is a distinct derivation from the isothermal assumption .
the typical filament width seen in figure [ fig : linemass ] is @xmath267pc .
@xcite and @xcite find typical filament widths of @xmath271pc in two different low - mass star - forming regions in the gould - belt _ herschel _ gto project .
these widths are a factor @xmath272 smaller compared to the spire 350@xmath19 m .
irac provides a much better resolution , but using the irac column density contours shown in figure [ fig : nh2_tmap ] , results in an estimated filament width of @xmath273pc . because this widths a based on the column density threshold , they might be larger than the corresponding fwhm of a fitted gaussian profile .
@xcite calculated the line mass for their filaments .
all filaments have line mass below the average line mass for 48 and most of the filaments are in contrast to 48 below the critical thermal line mass .
also most of the filaments in the pipe nebula are thermally subcritical @xcite .
these results might indicate a trend of larger filament widths towards supercritcal filaments ( with respect to turbulence ) .
recent studies have shown that the temperature structure of star - forming regions is non - isothermal .
dust temperature maps for low - mass star - forming bok - globules are shown in , e.g. , ( * ? ? ?
* cb244 ) .
preliminary results for irdcs were shown by @xcite for g030.49 + 0.96 using sed - fitting .
the minima in their temperature maps are aligned with the column density peaks as in 48 and the subsequent studies .
the minimum temperatures for g030.49 + 0.96 are lower compared to 48 and the two following studies .
the work of @xcite and @xcite show similar temperature ranges as for 48 .
while @xcite provided a large scale temperature and column density map , @xcite present a more detailed study of irdc 18454 adjacent to the mini - starburst w43 .
in contrast to irdc 18454 , 48 provides a temperature map of an externally undisturbed filament with the better resolution of the convolution to the spire 350@xmath19 m beam ( as for g030.49 + 0.96 , irdc 18454 data were convolved to the spire 500@xmath19 m beam ) .
the low temperatures in the cloud centers indicate that they are very young structures .
@xcite used a different approach .
they compared radiative transfer models of starless purely externally heated cores with the observed embedded cores in irdcs in the _ spitzer _ and _ herschel _ bands .
they found core temperature profiles with minimum temperatures of @xmath274k which increase by @xmath275k towards the edge of the core .
while our temperature mapping only observe an average line - of - sight temperature of the optically thin material , the rt model can retrieve temperature profiles along the line of sight . since these profiles probing also the optically thick regime , the minimal temperatures are obviously lower . on the other hand ,
the rt models require an assumption about the geometry of the cores and can not reassemble the diffuse temperature structure of the whole irdc . the temperature profile for 48 shows a drop towards the inner part of the detected sub - millimeter peaks .
theoretical considerations show that gas - phase coolants , as c@xmath42 and co , are efficient for low densities , while the energy transfer between gas and dust is the dominant cooling process for higher densities ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . with the presence of irac and mips 24@xmath19
m point sources one would expect to see the imprint of this embedded heating sources in the temperature maps . the spire 350@xmath19 m beam size limits the resolution of our temperature maps to @xmath276 so that the embedded heating sources remain unresolved .
the dust temperature mapping as presented in this work can provide a versatile tool to test the predicted coupling between gas and dust .
methylacetylene ( ) can be used to determine gas temperatures in massive star - forming regions ( cf .
iram pointed observations of towards the peak position of the sub - millimeter peak p1 results in a gas temperature of @xmath277k .
although this gas temperature is 7k higher than the dust temperature the difference is not significant due to the high uncertainty in the gas temperature estimate .
however the trend to higher gas temperatures is a good motivation for further observations of or .
especially gas temperature mapping can provide useful information on the gas - dust coupling at higher densities in molecular clumps .
future work should study the relation between dust temperature structures and gas temperature maps as provided by ammonia ( as in * ? ? ?
* ; * ? ? ?
* ) or methylacetylene line observations .
such data should provide further observational evidence for the predicted strong thermal coupling between gas and dust at higher densities .
the northern and southern extent of the irdc ( hosting c2 and c3 ) show a much flatter temperature profile with higher temperatures compared to the central part of the filament .
furthermore , the column densities are lower towards c2 and c3 .
this leads to the conclusion that the gas and dust densities are lower in these parts of the 48 .
pure isothermal cylindrical collapse can hardly explain this density gradient .
@xcite showed by modeling the density structures for the orion a cloud that , even in the isothermal case , a simple initial arrangement of a slightly rotating gas sheet can result in filamentary structures with complex density structures .
as in the studies mentioned above , it is a clear finding of this work that the dust temperatures are non - isothermally distributed in an irdc .
what impact could the modeling of thermal profiles have ?
thermal line broadening is significantly smaller compared to the observed linewidths .
therefore , a change of local temperatures should not have a notable influence on the turbulence structure inside the core . on the other hand the thermal profile towards the core can play an important role in the support against fragmentation , since the jeans mass is strongly dependent on temperature as @xmath278 .
theoretical studies of clustered star formation utilizing polytropic equations of states @xmath279 and temperature dependent density profiles @xmath280 indicate that the number of fragments in a cloud is strongly dependent on the polytropic index @xmath281 @xcite .
for the non - isothermal case ( @xmath282 ) the polytropic index reflects the balance between cooling and heating .
deriving dust temperatures maps as in this work is the first step towards a better understanding of spatial distributed cooling and heating inside the core .
a more detailed discussion of thermal properties can be found in e.g. , @xcite and @xcite .
the following gives a brief summary of the results presented in this paper : * 48 is a quite isolated , mono - filamentary irdc .
the gmc w51 observed in the same field is twice as distant as 48 .
* the dark filament observed by _
spitzer _ does not appear as a large coherent extinction silhouette in the pacs bands .
only the region of highest extinction in the central part is seen in extinction .
the filament is clearly seen in emission in the spire bands beyond 250@xmath19 m .
this is in good agreement with the sub - millimeter emission maps from scuba and atlasgal .
* we present reliable temperature and column density maps based on pixel - by - pixel sed fitting of _ herschel _ , atlasgal and scuba legacy maps .
regions of low emissions outside of the galactic plane were utilized to estimate the global background level for the _ herschel _ maps .
the location of the temperature minima are well aligned with the sub - millimeter and column density peaks .
the temperatures drop to 17.5k in the central part of 48 .
the heating by embedded mips 24@xmath19 m point sources is not detected , due to beam dilution .
* we derived a total mass of 390mfor 48 by applying a threshold of @xmath283 to the sed fitted column density maps .
* `` green and fuzzy '' features reveal the central parts of the irdc as the sites of most active ongoing star formation .
* we detected 13 point sources in all three pacs bands .
only one source lacks a mips 24@xmath19 m counterpart .
* sed fitting to these sources result in core masses of @xmath284 m , while the bolometric luminosity ranges between 15 and 630l .
the two most massive cores ( source 1 and 2 ) have mass of 14.3 and 11.9m@xmath86 and a bolometric luminosity of 630 and 475l , respectively . comparing their light - to - mass ratio to evolutionary models , both sources are likely to evolve into high - mass stars with @xmath285 . applying the definition from low - mass star formation , four cores are good candidates for `` class 0 '' objects . *
the average line mass of the 48 filament is 174mpc@xmath44 . compared to theoretical predictions
the whole filament is collapsing , even under the assumption of internal turbulent support .
* the point source separation in 48 is significantly smaller than in other irdcs , but in good agreement with the predictions by the isothermal cylindrical collapse model . *
according to our virial analysis the molecular clumps found in 48 are gravitationally unbound or just pressure confined .
* our observations are consistent with previous observations by @xcite , which deduced increasing turbulent motion towards the inner parts of the clumps .
* we detected several molecular species , such as , , , , , ( and ) .
the mapping observation showed several molecular clumps associated to the sub - millimeter peaks .
* we found co depletion towards the molecular clumps c1 + c2 with factor of @xmath286 . *
the low abundance of co towards the sub - millimeter peaks p1 and p2 permit relative high abundances of .
we thank zoltan balog and marc - andre bessel for their help with the optimal reduction of the irac / glimpse and mipsgal maps .
we appreciate the help of bernhard sturm with the polynomial background fitting of the _ spitzer_/irs spectra .
we are grateful to dmitry semenov for the enlightening discussions regarding astro - chemical problems .
+ this work is based in part on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology , under a contract with nasa .
this research made use of tiny tim / spitzer , developed by john krist for the spitzer science center .
the center is managed by the california institute of technology under a contract with nasa .
this paper is using data from the mopra radio telescope .
the mopra radio telescope is part of the australia telescope national facility which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
this publication makes use of molecular line data from the boston university - fcrao galactic ring survey ( grs ) .
the grs is a joint project of boston university and five college radio astronomy observatory , funded by the national science foundation under grants ast-9800334 , ast-0098562 , & ast-0100793 .
this research has made use of nasa s astrophysics data system .
this research has made use of the simbad database , operated at cds , strasbourg , france .
this research made use of aplpy , an open - source plotting package for python hosted at http://aplpy.github.com the work of s.r . and
a.m.s . was supported by the deutsche forschungsgemeinschaft priority program 1573 ( `` physics of the interstellar medium '' ) .
t.v . wishes to thank the national science foundation ( us ) for its funding of the astrochemistry program at the university of virginia .
cllcc ccccc cc & & & & & & & & & + c1 & 19:21:48.958 & @xmath28713:49:17.04 & 60 & 33.90 ( 0.16 ) & 2.50 ( 0.11 ) & 3.84 & @xmath288 & @xmath289 & 310 & 605 & 1.95 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 50 & 33.92 ( 0.12 ) & 2.06 ( 0.15 ) & 1.64 & @xmath290 & @xmath291 & 100 & 343 & 3.43 + c3 & 19:21:35.569 & @xmath28713:52:02.04 & 45 & 34.20 ( 0.16 ) & 2.99 ( 0.26 ) & 1.40 & @xmath291 & @xmath292 & 150 & 649 & 4.33 + & & & & & & & & & + c1 & 19:21:48.958 & @xmath28713:49:17.04 & 40 & 34.0 ( 0.4 ) & 3.6 ( 0.7 ) & 1.29 & @xmath293 & @xmath294 & 230 & 837 & 3.64 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 42 & 33.8 ( 0.3 ) & 2.3 ( 0.5 ) & 0.84 & @xmath295 & @xmath296 & 50 & 359 & 7.18 + c3 & 19:21:35.569 & @xmath28713:52:02.04 & 40 & 33.7 ( 0.4 ) & 2.5 ( 0.5 ) & 1.04 & @xmath297 & @xmath298 & 115 & 403 & 3.50 + c4 & 19:21:46.898 & @xmath28713:49:17.04 & 41 & 34.4 ( 0.5 ) & 2.6 ( 0.4 ) & 1.04 & @xmath299 & @xmath300 & 165 & 448 & 2.72 + & & & & & & & & & + c1 & 19:21:48.958 & @xmath28713:49:17.04 & 63 & 33.81 ( 0.11 ) & 2.50 ( 0.21 ) & 1.27 & @xmath301 & @xmath302 & 450 & 636 & 1.41 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 47 & 33.77 ( 0.12 ) & 1.66 ( 0.18 ) & 2.01 & @xmath303 & @xmath304 & 95 & 209 & 2.20 + c3 & 19:21:35.569 & @xmath28713:52:02.04 & 39 & 33.91 ( 0.28 ) & 2.4 ( 0.4 ) & 0.83 & @xmath305 & @xmath292 & 140 & 363 & 2.59 + & & & & & & & & & + c1 & 19:21:48.958 & @xmath28713:49:17.04 & 72 & 33.60 ( 0.17 ) & 1.47 ( 0.25 ) & 4.69 & 68.7 & 27.0 & 370 & 251 & 0.68 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 71 & 33.59 ( 0.11 ) & 1.10 ( 0.22 ) & 2.92 & 45.0 & 31.5 & 85 & 139 & 1.64 + c3 & 19:21:35.569 & @xmath28713:52:02.04 & 50 & 33.8 ( 0.5 ) & 2.9 ( 0.7 ) & 2.06 & 31.4 & 20.5 & 165 & 679 & 4.15 + & & & & & & & & & + c1 & 19:21:48.449 & @xmath28713:49:17.90 & 76 & 33.72 ( 0.11 ) & 2.11 ( 0.03 ) & 8.53 & @xmath306 & @xmath307 & 380 & 546 & 1.44 + & & & & & & & & & + c1 & 19:21:48.449 & @xmath28713:49:17.90 & 19 & 35.1 ( 0.22 ) & 2.53 ( 0.06 ) & 14.9 & @xmath308 & @xmath309 & 60 & 196 & 3.27 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 16 & 34.9 ( 0.3 ) & 2.49 ( 0.06 ) & 5.9 & @xmath310 & @xmath311 & 40 & 160 & 4.00 + & & & & & & & & & & + & & & & & & & & & + c1 & 19:21:48.449 & @xmath28713:49:17.90 & 18 & 34.2 ( 0.3 ) & 1.54 ( 0.11 ) & 2.3 & @xmath312 & @xmath313 & 55 & 69 & 1.25 + c2 & 19:21:51.016 & @xmath28713:47:47.03 & 15 & 34.02 ( 0.20 ) & 1.88 ( 0.15 ) & 1.6 & @xmath314 & @xmath315 & 30 & 86 & 2.87 [ tab : radio_lines ] | we present continuum observations of the infrared dark cloud ( irdc ) g48.660.22 ( 48 ) obtained with _
herschel _ , _ spitzer _ , and apex , in addition to several molecular line observations .
the _ herschel _
maps are used to derive temperature and column density maps of 48 using a model based on a modified blackbody .
we find that 48 has a relatively simple structure and is relatively isolated ; thus this irdc provides an excellent target to study the collapse and fragmentation of a filamentary structure in the absence of complicating factors such as strong external feedback .
the derived temperature structure of 48 is clearly non - isothermal from cloud to core scale .
the column density peaks are spatially coincident with the lowest temperatures ( @xmath0k ) in 48 .
a total cloud mass of @xmath1 is derived from the column density maps . by comparing the luminosity - to - mass ratio of 13 point sources detected in the _
herschel_/pacs bands to evolutionary models , we find that two cores are likely to evolve into high - mass stars ( @xmath2 ) .
the derived mean projected separation of point sources is smaller than in other irdcs but in good agreement with theoretical predications for cylindrical collapse .
we detect several molecular species such as , , , and .
co is depleted by a factor of @xmath3 compared to the expected interstellar abundance , from which we conclude that freezes out in the central region .
furthermore , the molecular clumps , associated with the sub - millimeter peaks in 48 , appear to be gravitationally unbound or just pressure confined .
the analysis of critical line masses in 48 show that the entire filament is collapsing , overcoming any internal support . |
while we are used to think of a diffusion process in terms of the msd ( [ msd ] ) calculated as the spatial average of @xmath19 over the probability density function @xmath20 , single particle tracking experiments typically provide few but long individual time series @xmath21 .
these are evaluated in terms of the time averaged msd @xcite @xmath22 where @xmath23 is the length of the time series ( measurement time ) and @xmath24 the lag time . in an ergodic system for sufficiently long @xmath23 ensemble and time averages
provide identical information , formally , @xmath25 , and @xmath26 for different trajectories are identical . for anomalous diffusion the behaviour of the msd ( [ msd ] ) and the time averaged msd ( [ tamsd ] ) may be fundamentally different .
simultaneously @xmath27 for different trajectories may become intrinsically irreproducible .
this phenomenon of irreproducibility and disparity @xmath28 is usually called weak ergodicity breaking @xcite and was discussed in detail for the subdiffusive ctrw @xcite . in the following , for simplicity
we neglect the explicit dependence of @xmath26 on the measurement time @xmath23 . for sbm
the mean of the time averaged msd over multiple trajectories @xmath29 , @xmath30 can be derived from the langevin equation ( [ langevin ] ) , yielding @xcite @xmath31}{t-\delta}.\ ] ] in the limit @xmath32 , we recover the behaviour @xcite @xmath33 and when the lag time @xmath24 approaches the measurement time @xmath23 the limiting form @xmath34 describes a cusp around @xmath35 . result ( [ tamsd_sbm_full ] ) is important to deduce the full behaviour of @xmath36 when @xmath24 approaches @xmath23 , as shown in fig .
[ fig_msd_cusp ] in the appendix .
( top ) and @xmath37 ( bottom ) .
the simulations of the sbm - langevin equation ( [ langevin ] ) show excellent agreement with the analytical results , eqs .
( [ msd ] ) and ( [ tamsd_sbm_full ] ) .
we also show results for the time averaged msd for 20 individual trajectories . apart from the region where the lag time @xmath24 approaches the length @xmath23 of the time series and statistics worsen
, there is hardly any amplitude scatter between individual @xmath38.,title="fig:",width=302 ] + ( top ) and @xmath37 ( bottom ) .
the simulations of the sbm - langevin equation ( [ langevin ] ) show excellent agreement with the analytical results , eqs .
( [ msd ] ) and ( [ tamsd_sbm_full ] ) .
we also show results for the time averaged msd for 20 individual trajectories . apart from the region where the lag time @xmath24 approaches the length @xmath23 of the time series and statistics worsen
, there is hardly any amplitude scatter between individual @xmath38.,title="fig:",width=302 ] in fig .
[ fig_tamsd ] we show results from simulations of the sbm process for both sub- and superdiffusion , observing excellent agreement with the analytical findings .
we also see that the scatter between individual trajectories is very small .
such scatter is a characteristic for anomalous diffusion models and can be used , e.g. , to reliably distinguish fbm from subdiffusive ctrw processes @xcite . for ctrw subdiffusion
we find pronounced scatter between @xmath36 from individual trajectories even in the limit of extremely long trajectories @xmath39 @xcite , while for fbm the scatter vanishes for longer @xmath23 , a characteristic of the ergodic nature of fbm @xcite .
the amplitude scatter of individual @xmath26 can be characterised in terms of the dimensionless variable @xmath40 .
if @xmath41 has a narrow distribution around @xmath42 and the width decreases with increasing @xmath23 , the process is usually considered ergodic .
this width is characterised in terms of the ergodicity breaking parameter @xmath43 . for sbm
it was found @xcite @xmath44 \frac{1}{3}(\delta / t)\ln(t/\delta),&\alpha=1/2\\[0.2 cm ] \frac{4\alpha^2}{3(2\alpha-1)}\left(\delta / t\right),&\alpha>1/2 \end{array}\right.\ ] ] with the integral @xmath45 ^ 2 $ ] @xcite .
the @xmath46 parameter for sbm thus clearly decays to zero for increasing @xmath23 .
for @xmath47 we obtain the known form @xmath48 for brownian motion .
the @xmath49 scaling also characterises fbm for @xmath50 @xcite . our analysis for sbm shows that the scatter distribution @xmath51 is indeed narrow and of gaussian shape , albeit it is broader than the gaussian form predicted for fbm in ref .
@xcite ( not shown ) .
it decreases with the ratio @xmath49 and thus indicates a reproducible behaviour between individual trajectories .
the coexistence of the disparity @xmath52 and asymptotically vanishing ergodicity breaking parameter , @xmath53 is a new class of non - ergodic processes the more detailed mathematical nature of which remains to be examined . from the analysis
so far for free motion we can see that the sbm process has a truly split personality . thus its pdf is identical to that for free fbm .
in contrast to the ergodic behaviour @xmath54 of fbm for sufficiently long @xmath23 @xcite , however , sbm exhibits weak ergodicity breaking , as demonstrated in eq .
( [ tamsd_sbm ] ) .
this scaling form @xmath55 exactly matches the result for ctrw subdiffusion @xcite or diffusion processes with space - dependent diffusivity @xcite . unlike the weakly non - ergodic dynamics of the latter two , for sbm the fluctuations around the mean @xmath56 measured by the distribution @xmath51 are narrow and decrease with longer @xmath23 .
we now show that the behaviour of confined sbm is also unconventional .
an important physical property of a stochastic process is its response to external forces or spatial confinement . from an experimental point of view , this is of relevance to tracer particles moving in the confines of cellular compartments or when the particle is traced by the help of optical tweezers , which exert an hookean restoring force on the particle @xcite .
we study the paradigmatic case of an harmonic potential @xmath57 , for which the motion is governed by the fokker - planck equation @xmath58 which follows directly from the langevin equation ( [ langevin ] ) with the additional hookean force term @xmath59 .
note that in absence of the external force ( @xmath60 ) this equation is that of batchelor @xcite .
for the msd we find an exact expression in terms of the kummer function @xmath61 @xcite , @xmath62 from which we obtain the initial free subdiffusion ( [ msd ] ) for @xmath63 and the scaling form @xmath64 in the long time limit @xmath65 . for subdiffusion ( @xmath6 )
the msd thus has a power - law decay to zero , while for superdiffusion ( @xmath16 ) it grows infinitely .
this counterintuitive behaviour of sbm is due to the fact that the time dependence of the diffusivity @xmath66 corresponds to a time dependent temperature @xcite or a time dependent viscosity .
thus sbm is a most non - stationary process that never reaches stationarity .
[ harmonic ] corroborates this analytical result with simulations based directly on the langevin equation ( [ langevin ] ) : after the free anomalous diffusion behaviour of the msd for a particle starting at the vertex of the potential , we observe a turnover to a power - law behaviour with negative or positive scaling exponent .
and time averaged msd @xmath67 for @xmath68 ( top ) and @xmath69 ( bottom ) in an harmonic potential . in each case
we consider the force constants @xmath70 and @xmath71 .
convergence of the corresponding ensemble and time averages at @xmath72 can be shown numerically .
the shown analytical curve is based on the full solution for @xmath67 provided in the appendix.,title="fig:",width=302 ] + and time averaged msd @xmath67 for @xmath68 ( top ) and @xmath69 ( bottom ) in an harmonic potential . in each case
we consider the force constants @xmath70 and @xmath71 .
convergence of the corresponding ensemble and time averages at @xmath72 can be shown numerically .
the shown analytical curve is based on the full solution for @xmath67 provided in the appendix.,title="fig:",width=302 ] the result for the time averaged msd is similarly remarkable . as shown in fig .
[ harmonic ] for sbm simulations based on the langevin equation ( [ langevin ] ) with the hookean forcing , it exhibits a pronounced apparent plateau for lag times @xmath73 for both sub- and superdiffusive sbm .
this behaviour is in excellent agreement with the full analytical solution ( [ tamsd_harm_full ] ) provided in the appendix in terms of kummer functions . taking the limit @xmath32 we obtain @xmath74,\ ] ] which indeed features the extended plateau and provides a good approximation for @xmath73 .
note that when @xmath24 approaches the measurement time @xmath23 the time averaged msd @xmath75 converges to the value of the msd @xmath76 due to the pole in expression ( [ tamsd ] ) .
this behaviour is analysed in more detail in fig .
[ fig_msd_cusp ] in the appendix .
( left ) , @xmath47 ( centre ) , and @xmath69 ( right ) for three different confinement strengths @xmath77 .
the fluctuations are stationary only in the brownian case @xmath47.,width=370 ] fig .
[ harmtraj ] analyses this behaviour in the harmonic potential further by showing the time series @xmath21 for subdiffusive sbm , brownian motion , and superdiffusive sbm .
indeed we see that for subdiffusive and superdiffusive confined sbm the fluctuations continue to decrease and increase with time , while those in the brownian limit become stationary .
this is the direct effect of the time dependent temperature or viscosity encoded in the sbm diffusivity @xmath66 .
if the system is stationary or thermalised this behaviour clearly underlines the unsuitability of sbm for the description of anomalous diffusion .
the behaviour of sbm dynamics under confinement is the central result of our study .
the continued temporal decay or increase of the msd that we obtained for sbm is in stark contrast to the behaviour of confined ctrw subdiffusion , for which the msd @xmath78 saturates to the thermal plateau , while the time averaged msd continues to grow in the power - law form @xmath79 up until the lag time @xmath24 approaches @xmath23 @xcite .
it strongly differs from fbm , which relaxes to a plateau for both the msd and the time averaged msd , and for which only a transient disparity between @xmath80 and @xmath26 exists @xcite .
finally , sbm is also at variance with heterogeneous diffusion processes with a space dependent diffusion coefficient relax to a plateau for both @xmath78 and @xmath26 @xcite .
as we showed sbm is a truly paradoxical stochastic process . somewhat similar to a chameleon , each time we compare sbm with other established anomalous diffusion processes we find certain similarities .
looking at the sum of its features , however , sbm is a truly independent process with a range of remarkable properties . for free sbm
the probability density function @xmath20 equals that of fbm , despite the fact that both processes are governed by different stochastic ( langevin ) equations .
sbm s time averaged msd scales equally to those of the weakly non - ergodic ctrw subdiffusion and diffusion processes with space - dependent diffusivity . despite this weakly non - ergodic character of the mean time averaged msd @xmath81 , the amplitude scatter between the time averaged msd @xmath26 of individual realisations is small and
the distribution has a gaussian shape , as otherwise observed for the ergodic fbm or for brownian motion of finite time @xmath23 . in that sense sbm
represents a new class of non - ergodic processes .
the most striking behaviour of sbm is , however , its strongly non - stationary behaviour under confinement . instead of relaxing to a plateau
the msd acquires a power - law decay or growth mirroring a continuously decreasing or increasing temperature encoded in sbm s time dependent diffusion coefficient @xmath66 .
sbm is thus at variance with the currently available experimental observations in complex liquids using single particle tracking by video microscopy or by optical tweezers tracking of single submicron particles .
thus the free anomalous diffusion data garnered so far was classified into fbm - like and ctrw - like behaviour , or combinations thereof @xcite .
note that also for fluorescence correlation spectroscopy experiments recent analysis tools corroborated an fbm nature of the data @xcite . for optical tweezers tracking of lipid granules in different complex liquids
the time averaged msd either continues to grow under confinement , reflecting the non - ergodic features of ctrws @xcite , or it relaxes to a plateau value mirroring an ergodic dynamics @xcite .
how can fbm and sbm have the same distribution ( [ prop ] ) in free space ?
simply put for fbm , the viscoelastic properties of the environment effecting the long - range correlations of fbm lead to a frequency dependent response of the environment to a disturbance , while the materials properties remain unchanged in time . for free fbm
this gives rise to the subdiffusive msd ( [ msd ] ) , in which the time dependent diffusivity effectively encodes the frequency dependent response of the viscoelastic environment .
the distribution ( [ prop ] ) for fbm and its description in terms of a fokker - planck equation of the form ( [ fokker ] ) is treacherous , however .
this can already be seen when we use the pdf ( [ prop ] ) or the dynamics equation ( [ fokker ] ) to calculate the first passage behaviour .
this procedure leads to the wrong result @xcite , and the full analytical description of fbm in the presence of boundaries remains elusive , a difficulty imposed by the highly correlated fractional gaussian noise driving its langevin equation .
in contrast , sbm , according to its langevin description ( [ langevin ] ) , is driven by uncorrelated noise but the environment itself is changing as function of time , effecting an extremely non - stationary process .
the equivalence of the pdf ( [ prop ] ) of both processes is thus simply due to the fact that a gaussian pdf is completely defined by its second moment , the msd ( [ msd ] ) . with its interesting behaviour that is so different from the other conventional anomalous diffusion models ,
sbm may indeed have relevant applications to weakly coupled or fully adiabatic systems , as well as for active systems , in which the existence of a temperature is not meaningful .
in particular , in the superdiffusive case sbm or analogous dynamics with other increasing effective diffusivities may represent an alternative approach to active brownian motion @xcite . the difference of sbm to other processes
can also be seen in fig .
[ trajs ] . for the subdiffusive case a sample trajectory of sbm
is compared to that of fbm and the noisy ctrw @xcite , in which the pure subdiffusive ctrw is superimposed with ornstein - uhlenbeck noise to accommodate the thermal noise of the environment observed in experiments @xcite .
we see that sbm with its gaussian probability density function and uncorrelated driving noise appears more similar to the noisy ctrw motion , albeit it has a more pronounced tendency to reach larger amplitudes than the ctrw with its waiting time periods .
the fluctuations of sbm are dramatically less than those of the anticorrelated fbm , which also shows the largest amplitudes in @xmath21 . for superdiffusion , we compare sbm with a noisy version of the lvy walk process @xcite and again with fbm .
note the vastly different size of the window shown on the ordinate .
this time , the persistent fbm shows pronouncedly larger excursions and a distinct reduction of the noise compared to sbm .
the shape of @xmath21 of the latter does not appear to be qualitatively different from the subdiffusive case .
the noisy ctrw is fundamentally different from both sbm and fbm . despite their disparate physical nature and their dissimilarity when setting one against the other in a direct comparison , we note that generally it is difficult to identify a stochastic process solely from the appearance of the recorded trajectory .
( top ) and superdiffusion with @xmath69 ( right).,title="fig:",width=302 ] + ( top ) and superdiffusion with @xmath69 ( right).,title="fig:",width=302 ] from this discussion it is obvious that one thing is to have at our disposal a handy and easy - to - use description for anomalous diffusion processes .
sbm with its gaussian and uncorrelated nature appears deceivingly simple and is therefore easy to implement in numerical analyses and descriptions such as diffusion limited reactions .
however , the other face of the medal is about the physical relevance of a model process .
using sbm with its time - dependent diffusion coefficient violates the physical setting in typical experiments , in which the system is held at approximately constant temperature and its predictions are at odds with actual observations .
the unphysical nature for the kind of processes we have in mind is most obvious for confined motion .
in contrast , fbm and fractional langevin equation motion are ergodic processes with the physical background of an effective particle motion in a viscoelastic multi - body environment .
fbm and fractional langevin equation motion exhibit a transient disparity between the msd and the time averaged msd under confinement @xcite .
weakly non - ergodic ctrw processes emerge due to immobilisation periods imprinted on the dynamics by the structure of the environment of binding events to the environment .
finally , diffusion processes with space dependent diffusivities arise naturally in non - homogeneous systems such as biological cells or subsurface aquifers . to extract physically meaningful information from anomalous diffusion data one needs to have some physical insight into the observed process and
properly analyse the data using complementary tools @xcite before settling for the appropriate physical model .
we finally note that it will be interesting to compare the predictions of the sbm model with that of active processes in viscoelastic environments @xcite .
we acknowledge funding from the academy of finland within the fidipro scheme .
the covariance of the position for sbm in an harmonic potential follows from the langevin equation ( [ langevin ] ) .
our result is @xmath82 for @xmath83 . in absence of the confinement ( @xmath60 ) the covariance reduces to the msd ( [ msd ] ) and in the brownian limit
@xmath47 we recover the familiar covariance @xmath84 from eq .
( [ cova ] ) we derive the exact result for the time averaged msd ( [ tamsd ] ) , @xmath85)\\ \nonumber & & \hspace*{-1.2 cm } + \frac{2k_{\alpha}}{(1+\alpha)(t-\delta)}\big[t^{1+\alpha}e^{-2kt}m(1+\alpha , 2+\alpha,2kt)\\ \nonumber & & \hspace*{0.6cm}-\delta^{1+\alpha}e^{-2k\delta}m(1+\alpha,2+\alpha,2k\delta)\big]\\ \nonumber & & \hspace*{-1.2 cm } -\frac{4k_{\alpha}}{1+\alpha}(t-\delta)^{\alpha}e^{-2kt+k\delta}\\ & & \times m(1+\alpha,2+\alpha,2k[t-\delta ] ) . \label{tamsd_harm_full}\end{aligned}\ ] ] to derive the limit ( [ tamsd_harm ] ) of the time averaged msd for confined sbm in the long time limit @xmath39 we use the property @xmath86 of the kummer function @xcite . to the msd @xmath87 at @xmath88 for @xmath68 and @xmath69 .
the measurement time @xmath89 .
top : free sbm . for the superdiffusive case the dashed black line of unity slope aids in demonstrating the non - linear behaviour of @xmath75 .
bottom : sbm in an harmonic potential with @xmath71 .
note the different ranges of the abscissa in the two cases .
for the extremely small window for the subdiffusive case needed to illustrate the cusp at @xmath88 and the relatively large variation of @xmath75 the msd @xmath87 appears almost constant.,title="fig:",width=302 ] + to the msd @xmath87 at @xmath88 for @xmath68 and @xmath69 .
the measurement time @xmath89 .
top : free sbm . for the superdiffusive case the dashed black line of unity slope aids in demonstrating the non - linear behaviour of @xmath75 .
bottom : sbm in an harmonic potential with @xmath71 .
note the different ranges of the abscissa in the two cases .
for the extremely small window for the subdiffusive case needed to illustrate the cusp at @xmath88 and the relatively large variation of @xmath75 the msd @xmath87 appears almost constant.,title="fig:",width=302 ] | anomalous diffusion is frequently described by scaled brownian motion ( sbm ) , a gaussian process with a power - law time dependent diffusion coefficient .
its mean squared displacement is @xmath0 with @xmath1 for @xmath2 .
sbm may provide a seemingly adequate description in the case of unbounded diffusion , for which its probability density function coincides with that of fractional brownian motion . here
we show that free sbm is weakly non - ergodic but does not exhibit a significant amplitude scatter of the time averaged mean squared displacement .
more severely , we demonstrate that under confinement , the dynamics encoded by sbm is fundamentally different from both fractional brownian motion and continuous time random walks .
sbm is highly non - stationary and can not provide a physical description for particles in a thermalised stationary system .
our findings have direct impact on the modelling of single particle tracking experiments , in particular , under confinement inside cellular compartments or when optical tweezers tracking methods are used . the passive , thermally driven diffusion of fluorescently labelled molecules or optically visible submicron particles is routinely measured inside living cells and complex liquids by methods such as single particle tracking @xcite and fluorescence correlation spectroscopy @xcite .
often the observed diffusion patterns deviate from the laws of brownian motion and display anomalous diffusion characterised by the mean squared displacement ( msd ) @xmath3 with the anomalous diffusion coefficient @xmath4 of dimension @xmath5 .
depending on the magnitude of the anomalous diffusion exponent one distinguishes subdiffusion ( @xmath6 ) and superdiffusion ( @xmath7 ) @xcite . in biological contexts
one often invokes the concept of apparent subdiffusion , which is in fact a transient crossover between free normal diffusion with @xmath8 and the thermal plateau of the msd in confinement @xcite .
however , there is clear experimental evidence of long - ranged subdiffusion ( [ msd ] ) in living biological cells @xcite and in artificially crowded or structured liquids @xcite .
the physical mechanisms for such anomalous diffusion are very varied , even considering the most common approaches : we mention the continuous time random walk approach @xcite , in which subdiffusion is effected by multiple trapping or sticking events with an emerging power - law distribution of waiting times , as seen in the data of refs .
@xcite .
alternatively , subdiffusion arises in fractal geometries with their dead ends and bottlenecks across spatial scales @xcite , an additional mechanism pinpointed in ref .
@xcite .
we also mention diffusion processes in which the anomaly arises from the spatial dependence of the diffusion coefficients @xcite , or from interaction of the tracer particle with a structured environment @xcite . a special role play the gaussian models for anomalous diffusion which belong to two families , that can best be distinguished on the langevin equation level . the mandelbrot - van ness fractional brownian motion ( fbm )
follows the stochastic equation @xmath9 , where @xmath10 represents fractional gaussian noise characterised by the covariance @xmath11 which has long - ranged negative or positive correlations , respectively , for subdiffusion ( @xmath6 ) or superdiffusion ( @xmath7 ) @xcite .
fbm and the closely related fractional langevin equation motion are physically related to the motion in a viscoelastic environment @xcite . in various experiments
this type of motion was identified as single or partial component @xcite .
here we focus on the second family of gaussian anomalous diffusion models , namely , scaled brownian motion ( sbm ) @xcite .
sbm is governed by the langevin equation @xmath12 where @xmath13 is white gaussian noise with normalised covariance @xmath14 .
the prefactor in eq .
( [ langevin ] ) is the power - law time dependent diffusion coefficient @xmath15 which decreases or increases in time for @xmath6 and @xmath16 , respectively .
sbm is used to model anomalous diffusion in a wide range of systems @xcite , in particular , in frap experiments @xcite . in terms of the time dependent diffusivity ,
the msd ( [ msd ] ) can be rewritten as @xmath17 .
originally , the diffusion process with a time dependent diffusivity was introduced by batchelor in the description of relative diffusion in turbulence @xcite . for unconfined sbm
the probability density function is the gaussian @xcite @xmath18 it has exactly the same form as the free pdf of fbm @xcite , if only both processes are started at the origin . despite this deceiving similarity
, we show that sbm is fundamentally different from fbm and all other anomalous diffusion models mentioned above .
most importantly , we show that it can not represent a physical model for anomalous diffusion in stationary or thermalised systems . concurrently
, the remarkable properties of sbm revealed here may have other relevant applications , as briefly discussed below . |
a major challenge of contemporary biology is to ascribe interpretation to high - throughput experimental or computational results , where each considered entity ( gene or protein ) is assigned a value .
biological information is often summarized through controlled vocabularies such as gene ontology ( go ) @xcite , where each annotated term includes a list of entities .
let @xmath0 denote a collection of values , each associated with an entity .
given @xmath0 and a controlled vocabulary , enrichment analysis aims to retrieve the terms that by statistical inference best describe @xmath0 , that is , the terms associated with entities with atypical values .
many enrichment analysis tools have been developed primarily to process microarray data @xcite . in terms of biological relevance ,
the performance assessment of those tools is generally difficult .
it requires a large , comprehensive ` gold standard ' vocabulary together with a collection of @xmath0 s processed from experimental data , and with true / false positive terms corresponding to each @xmath0 correctly specified .
this invariably introduces some degree of circularity because the terms often come from curating experimental results . before declaring efficacy in biological information retrieval that is nontrivial to assess , an enrichment method should pass at least the statistical accuracy and internal consistency test . in their recent survey , @xcite list 68 distinct bioinformatic enrichment tools introduced between 2002 and 2008 .
most tools share a similar workflow : given @xmath0 obtained by suitably processing experimental data , they sequentially test each vocabulary term for enrichment to obtain its p - value ( the likelihood of a false positive given the null hypothesis ) . since many terms are tested , a multiple hypothesis correction , such as bonferroni @xcite or false discovery rate ( fdr ) @xcite , is applied to p - value of each to obtain the final statistical significance .
the results are displayed for the user in a suitable form outlining the significant terms and possibly relations between them .
note that the latter steps are largely independent from the first . to avoid confounding factors
, we will focus exclusively on the original enrichment p - values .
based on the statistical methods employed , the existing enrichment tools can generally be divided into two main classes . the singular enrichment analysis ( sea ) class contains numerous tools that form the majority of published ones @xcite . by ordering values in @xmath0 ,
these tools require users to select a number of top - ranking entities as input and mostly use hypergeometric distribution ( or equivalently fisher s exact test ) to obtain the term p - values .
after the selection is made , sea treats all entities equally , ignoring their value differences .
the gene set analysis ( gsa ) class was pioneered by the gsea tool @xcite .
tools from this class use all values ( entire @xmath0 ) to calculate p - values and do not require pre - selection of entities .
some approaches @xcite in this group apply hypergeometric tests to all possible selections of top - ranking entities .
the final p - value is computed by combining ( in a tool - specific manner ) the p - values from the individual tests .
other approaches use non - parametric approaches : rank - based statistics such as wilcoxon rank - sum @xcite or kolmogorov - smirnov - like @xcite . when weights are taken into account , such as in gsea @xcite , statistical significance
must be determined from a sampled ( shuffled ) distribution .
unfortunately , limited by the number of shuffles that can be performed , the smallest obtainable p - value is bounded away from 0 . the final group of gsa methods computes a score for each vocabulary term as a sum of the values ( henceforth used interchangeably with weights ) of the @xmath1 entities it annotates . in general , the score distribution @xmath2 for the experimental data is unknown . by central limit theorem ,
when @xmath1 is large , gaussian @xcite or student s t - distribution @xcite can be used to approximate @xmath2 .
unfortunately , when the weight distributions are skewed , the required @xmath1 may be too large for practical use .
evidently , this undermines the p - value accuracy of small terms ( meaning terms with few entities ) , which are biologically most specific .
it is generally found that , given the same vocabulary and @xmath0 , different enrichment analysis tools report diverse results .
we believe this may be attributed to disagreement in p - values reported as well as that different methods have different degree of robustness ( internal consistency ) . instead of providing a coherent biological understanding , the array of diverse results questions the confidence of information found . furthermore , other than microarray datasets , there exist experimental or computational results such as those from chip - chip @xcite , deep sequencing @xcite , quantitative proteomics @xcite and _ in silico _ network simulations @xcite , that may benefit from enrichment analysis .
it is thus imperative to have an enrichment method that report accurate p - values , preserves internal consistency , and allows investigations of a broader range of datasets . to achieve these goals , we have developed a novel enrichment tool , called saddlesum , that founds on the well - known lugananni - rice formula @xcite and derives its statistics from approximating asymptotically the distribution function of the scores used in the parametric gsa class .
this allows us to obtain accurate statistics even in the cases where the distribution function generating @xmath0 is very skewed and for terms containing few entities .
the latter aspect is particularly important for obtaining biologically specific information .
we distinguish two sets : the set of entities @xmath3 of size @xmath4 and the controlled vocabulary @xmath5 . each term from @xmath5 maps to a set @xmath6 of size @xmath7 . from experimental results ,
we obtain a set @xmath8 and ask how likely it is to randomly pick @xmath1 entities whose sum of weights exceeds the sum @xmath9 .
assume that the weights in @xmath0 come independently from a continuous probability space @xmath10 with the density function @xmath11 such that the moment generating function @xmath12 exists for @xmath13 in a neighborhood of 0 .
the density of @xmath14 , sum of @xmath1 weights arbitrarily sampled from @xmath0 , can be expressed by the fourier inversion formula @xmath15 where @xmath16 denotes the cumulant generating function of @xmath11 .
the tail probability or p - value for a score @xmath17 is given by @xmath18 we propose to use an asymptotic approximation to ( [ eqt : pval1 ] ) , which improves with increasing @xmath1 and @xmath17 .
@xcite derived an asymptotic approximation for the density @xmath19 through saddlepoint expansion of the integral ( [ eqt : int3 ] ) while the corresponding approximation to the tail probability was obtained by @xcite .
let @xmath20 and @xmath21 denote respectively the density and the tail probability of gaussian distribution .
let @xmath22 be a solution of the equation @xmath23 then , the leading term of the lugananni - rice approximation to the tail probability takes the form @xmath24 where @xmath25 and @xmath26 .
appropriate summary of derivation of ( [ eqt : tail ] ) is provided in supplementary materials .
@xcite has shown that eq.([eqt : saddle1 ] ) has a unique simple root under most conditions and that @xmath27 increases with @xmath17 , with @xmath28 for @xmath29 where @xmath30 is the mean of @xmath10 .
while the approximation ( [ eqt : tail ] ) is uniformly valid over the whole domain of @xmath11 , its components need to be rearranged for numerical computation near the mean . when @xmath31 , @xmath32 dominates and the overall error is @xmath33 @xcite .
saddlesum , our implementation of lugananni - rice approximation for computing enrichment p - values , first solves eq.([eqt : saddle1 ] ) for @xmath22 using newton s method and then returns the p - value using ( [ eqt : tail ] ) .
the derivatives of the cumulant generating function are estimated from @xmath0 : we approximate the moment generating function by @xmath34 , and then @xmath35 and @xmath36 . since the same @xmath0 is used to sequentially evaluate p - values of all terms in @xmath5 , we retain previously computed @xmath22 values in a sorted array .
this allows us , using binary search , to reject many terms with p - values greater than a given threshold without running newton s method or to bracket the root of ( [ eqt : saddle1 ] ) for faster convergence .
more details on the saddlesum implementation and evaluations of its accuracy against some well - characterized distributions are in section 2 of supplementary materials . when run as a term enrichment tool , saddlesum reports e - value for each significant term by applying bonferroni correction to the term s p - value .
the assignment of human genes to go terms was taken from the ncbi gene2go file ( ftp://ftp.ncbi.nih.gov/gene/data/gene2go.gz ) downloaded on 07 - 02 - 2009 .
after assigning all genes to terms , we removed all redundant terms if several terms mapped to the same set of genes , we kept only one such term . for our statistical experiments we kept only the terms with no less than five mapped genes within the set of weights considered and hence the number of processed terms varied for each realization of sampling ( see below ) .
@xcite is an implementation of the framework for exploring information flow in interaction networks @xcite .
information flow is modeled through discrete time random walks with damping at each step the walker has a certain probability of leaving the network .
although offers three modes : emitting , absorbing and channel , we only used the simplest , emitting mode , to provide examples illustrating issues of significance assignment .
the emitting mode takes as input one or more network proteins , called sources , and a damping factor @xmath37 . for each protein node in the network
, the model outputs the expected number of visits to that node by random walks originating from the sources , thus highlighting the network neighborhoods of the sources .
the damping factor determines the average number of steps taken by a random walk before termination : @xmath38 corresponds to no termination while @xmath39 leads to no visits apart from the originating node . for our protein - protein interaction network examples
, we used the set of all human physical interactions from the biogrid @xcite , version 2.0.54 ( july 2009 ) .
the network consists of 7702 proteins and 56400 unique interactions .
each interaction was represented by an undirected link .
a link carries weight 2 if its two ends connect to the same protein and 1 otherwise . from the ncbi gene expression omnibus ( geo )
@xcite , we retrieved human microarray datasets with expression @xmath40 ratios ( weights ) provided , resulting in 34 datasets and 136 samples in total . for each sample , when multiple weights for the same gene were present , we took their mean instead .
this resulted in a @xmath0 where each gene is assigned a unique raw weight . for evaluations
, we also used another version of @xmath0 where negative weights were set to zero .
this version facilitated investigation of up - regulation while keeping the down - regulated genes as part of statistical background . by definition ,
a p - value associated with a score is the probability of that score or better arising purely by chance .
we tested the accuracy of reported p - values reported by enrichment methods via simulations on ` decoy ' databases , which contained only terms with random gene assignments . for each term from the decoy dataset and each set of weights based on network or microarray data , we recorded the reported p - value and thus built an empirical distribution of p - values . if a method reports accurate p - values , the proportion of runs , which we term empirical p - value , reporting p - values smaller than or equal to a p - value cutoff , should be very close to that cutoff .
we show the results graphically by plotting on the log - log scale the empirical p - value as a function of the cutoff . for each given list of entities @xmath3 , be it from the target gene set of a microarray dataset or the set of participating human proteins in the interaction network
, we produced two types of decoy databases .
the first type was based on go .
we shuffled gene labels 1000 times .
for each shuffle , we associated all terms from go with the shuffled labels to retain the term dependency .
this resulted in a database with approximately @xmath41 terms ( 1000 shuffles times about 5000 go terms ) . in the second type , each term , having the same size @xmath1 , was obtained by sampling without replacement @xmath1 genes from @xmath3 .
the databases from this type ( one for each term size considered ) contained exactly @xmath42 terms .
the evaluation query set of 100 @xmath0 s from interaction networks was obtained by randomly sampling 100 proteins out of 7702 and running with each protein as a single source .
the weights for source proteins were not considered since they were prescribed , not resulting from simulation .
each run used @xmath43 , without excluding any nodes from the network . for microarrays , the set of 136 samples was used . since both query sets are of size @xmath44 , the total number of @xmath0term matches was @xmath45 .
similar to saddlesum , t - test approaches are based on sum - of - weights score but use the student s t - distribution to infer p - values .
as before , let @xmath46 denote the weight associated with entity @xmath47 , let @xmath48 denote the set of @xmath1 entities associated with a term from vocabulary and let @xmath49 .
for any set @xmath50 of size @xmath51 , let @xmath52 denote the mean weight of entities in @xmath53 and let @xmath54 be their sample variance .
gage @xcite enrichment tool uses two sample t - test assuming unequal variances and equal sample sizes to compare the means over @xmath3 and @xmath48 .
the test statistic is @xmath55 and the p - value is obtained from the upper tail of the student s t - distribution with degrees of freedom @xmath56 t - profiler @xcite compares the means over @xmath48 and @xmath57 using two sample t - test assuming equal variances but unequal sample sizes .
the pooled variance estimate is given by @xmath58 and the test statistic is @xmath59 the t - profiler p - value is then obtained from the tail of the student s t - distribution with @xmath60 degrees of freedom .
methods based on hypergeometric distribution or equivalently , fisher s exact test , use only rankings of weights and require selection of ` significant ' entities prior to calculation of p - value .
we first rank all entities according to their weights and consider the set @xmath61 of @xmath62 entities with largest weights .
the number @xmath62 can be fixed ( say 50 ) , correspond to a fixed percentage of the total number of weights , depend on the values of weights , or be calculated by other means .
the score @xmath17 for the term @xmath48 is given by the size of the intersection , @xmath63 , between @xmath61 and @xmath48 .
this is equivalent to setting @xmath9 with @xmath64 for @xmath65 and 0 otherwise .
the p - value for score @xmath17 is @xmath66 hence , the p - value measures the likelihood of score @xmath17 or better over all possible ways of selecting @xmath62 entities out of @xmath3 , with @xmath1 entities associated with the term investigated . in each of our p - value accuracy experiments we used two variants of the hypergeometric method , one taking a fixed percentage of nodes and the other taking into account the values of weights . for microarray datasets , the fist variant took 1% of available genes ( hgem - pn1 ) while the second select genes with four fold change or more ( hgem - f2 ) . in experiments based on protein networks
, we took 3% of available proteins ( 231 entities ) for the first variant ( hgem - pn3 ) and used the participation ratio formula to determine @xmath62 in the second ( hgem - pr ) .
participation ratio @xcite is given by the formula @xmath67 we chose a smaller percentage of weights for microarray - based data ( 1% vs 3% for data derived for networks ) because the microarray datasets generally contained measurement for more genes than the number of proteins in the network .
instead of making a single , arbitrary choice of @xmath62 and applying hypergeometric score , mhg method implemented in the gorilla package @xcite considers all possible @xmath62 s .
the mhg score is defined as @xmath68 where @xmath69 is the number of entities annotated by the term @xmath48 among the @xmath62 top - ranked entities . the exact p - value for mhg score is then calculated by using a dynamic programming algorithm developed by eden _
_ @xcite . for our experiments we used an implementation in c programming language that was derived from the java implementation used by gorilla .
the implementation uses a truncated algorithm that gives an approximate p - value with improved running speed . to evaluate consistency of investigated methods
, we compared the sets of significant terms retrieved from go using different numbers of nonzero weights as input . for each @xmath0 , we sort in descending order the weights associated with entities .
with each @xmath62 selected , we kept @xmath62 largest weights unchanged and set the remaining to 0 to arrive at a modified set of weights @xmath71 .
we did not totally exclude the lower weights but kept them under consideration to provide statistical background .
we submitted @xmath71 for analysis and obtained from each statistical method a set of enriched terms ordered by their p - value . in fig.[fig : mainexamples1]a and supplementary fig.s3 , we displayed the actual five most significant terms retrieved with their p - values for selected examples of weight sets . to investigate on a larger scale the retrieval stability to @xmath62 changes
, we computed for each method the overlap between sets of top ten terms from two different @xmath62 s for the @xmath0 sets mentioned in ` evaluating accuracy of p - values ' and then took the average ( fig.[fig : mainexamples1]b ) .
we compared our saddlesum approach against the following existing methods : fisher s exact test ( hgem @xcite ) , two sample student s t - test with equal ( t - profiler @xcite ) and unequal ( gage @xcite ) variances , and mhg score @xcite . based on data from both microarrays and simulations of information flow in protein networks ,
the comparison shown here encompassed ( in order of importance ) evaluation of p - value accuracy , ranking stability and running time .
accurate p - value reflects the likelihood of a false identification and thus allows for comparison between terms retrieved even across experiments .
incorrect p - values therefore render ranking stability and algorithmic speed pointless .
accurate p - values without ranking stability question the robustness of biological interpretation . for pragmatic use of an enrichment method , even with accurate statistics and stability , it is still important to have reasonable speed
. the term p - value reported by an enrichment analysis method provides the likelihood for that term to be enriched within @xmath0 . to infer biological significance using statistical analysis ,
it is essential to have accurate p - values .
we analyzed the accuracy of p - values reported by the investigated approaches through simulating @xmath45 queries and comparing their reported and empirical p - values .
results based on querying databases with fixed term sizes are shown in supplementary figs.s1 and s2 . shown in fig.[fig : stats1 ] are the results for querying go - based gene - shuffled term databases , which retain the structure of the original go as a mixture of terms of different sizes organized as a directed acyclic graph where small terms are included in larger ones .
the curves for all methods in fig.[fig : stats1 ] therefore resemble a mixture of curves from supplementary figs.s1 and s2 albeit weighted towards smaller - sized terms .
for weights from both network simulations and microarrays , saddlesum as well as the methods based on fisher s exact test ( hgem and mhg ) report p - values that are acceptable ( within one order of magnitude from the theoretical values ) . for hgem and mhg , this is not surprising because our experiments involved shuffling entity labels and hence followed the null model of the hypergeometric distribution . on the other hand , the null model of saddlesum and
the t - test methods assumes weights drawn independently from some distribution ( sampling with replacement ) . for terms with few entities ( @xmath72 ) ,
the difference between the two null models is minimal and the p - value accuracy assessment curves for saddlesum run as close to the theoretical line as those for hgem methods . for @xmath73
, saddlesum gives more conservative p - values for terms with large sums of weights ( supplementary figs.s1 and s2 ) . in practice
, this has no significant effect to biological inference .
large terms would be still selected as significant given a reasonable p - value cutoff and accurate p - values are assigned to small terms that are biologically specific .
two - sample t - test with unequal variances as used by gage package reports p - values so conservative that they are often larger than 0.01 and hence not always visible in our accuracy plots .
this effect persists even for @xmath1 as large as 500
. this might be because the number of degrees of freedom used is considerably small .
in addition , its test statistic ( eq.([eq : ttst2 ] ) ) emphasizes the estimated within - term variance @xmath74 that is typically larger than the overall variance @xmath75 . on the other hand ,
t - profiler generally exaggerates @xcite significance because it uses the t - distribution with a large number of degrees of freedom ( @xmath76 ) .
although some small terms may appear biologically relevant ( as in fig.[fig : mainexamples1 ] ) , one should not equate these exaggerated p - values with sensitivity . for microarray data , the @xmath40 ratios are almost symmetrically distributed about 0 ( supplementary fig.s4 ) .
the distribution of their sum is close to gaussian .
however , t - profiler still significantly exaggerates p - values for terms whose @xmath77 ( supplementary fig.s2 ) .
the statistical accuracy of t - profiler worsens when negative @xmath40 ratios are set to 0 .
the reason for doing so is that allowing weights within each term to cancel each other may not be biologically appropriate .
go terms may cover a very general category where annotations may not always be available for more specific subterms .
subsequently , terms may get refined and new terms may emerge .
in such situation , it is desirable to discover terms that have genes that are significantly up - regulated even if many genes from the same term are down - regulated .
p - value accuracy , although the most important criterion , measures only performance with respect to non - significant hits , that is , the likelihood of a false positive .
it is also necessary to consider the quality of enrichment results in terms of the underlying biology .
testing the quality directly , as described in the introduction , is not yet feasible .
instead we evaluated internal consistency of each method with respect to the number of top - ranked entities used for analysis .
fig.[fig : mainexamples1]a shows the change of p - values reported for the top five go terms with respect to the number of selected entities using two examples with weights respectively from network flow simulation and microarray .
additional examples are shown in supplementary fig.s3 .
results from evaluating the overall consistency of the best ten terms retrieved are shown in fig.[fig : mainexamples1]b . both hgem and mhg methods are highly sensitive to the choice of @xmath62 , the number of entities deemed significant . with a small @xmath62 ,
their sets of significant terms resemble the top terms obtained by saddlesum , while large values of @xmath62 render very small p - values for large - sized terms ( often biologically non - specific ) .
this is mainly because hgem and mhg treat all selected significant entities as equally important without weighting down less significant entities , the collection of which may out vote the most significant ones .
hence , although mhg considers all possible @xmath62 values , to obtain biologically specific interpretation , it might be necessary to either remove very large terms from the vocabulary or to impose an upper bound on @xmath62 . in that respect
, mhg is very similar to the original gsea method @xcite , which also ignored weights .
the authors of gsea noted that the genes ranked in middle of the list had disproportionate effect to their results and produced an improved version of gsea @xcite with weights considered .
gage does not show strong consistency because many p - values it reports are too conservative and fall above the 0.01 threshold we used .
consequently , the best overlap between various cutoffs is about 5 ( out of 10 ) for network flow examples and 4 for microarray examples
( fig.[fig : mainexamples1]b ) .
t - profiler shows great internal consistency .
unfortunately , as shown in fig.[fig : stats1 ] , supplementary figs.s1 and s2 , it reports inaccurate p - values , especially for small terms .
this is illustrated in the top panel of fig.[fig : mainexamples1]a , where t - profiler selects as highly significant the small terms ( with 5,6 and 9 entities ) , which are deemed insignificant by all other methods .
the same pattern can be observed in supplementary fig.s3 , although the severity is tamed for microarrays . using weights for scoring terms , saddlesum is also stable with respect to the choice of @xmath62 but with accurate statistics . in terms of algorithmic running time ( table [ tbl : speed ] ) , parametric methods relying on normal or student s
t - distribution require few computations .
methods based on hypergeometric distribution , if properly implemented , are also fast . on the other hand ,
non - parametric methods can take significant time if many shufflings are performed . based on dynamic programming
, mhg method can also take excessive time for large terms .
saddlesum has running time that is only slightly longer than that of parametric methods . .running times of evaluated enrichment statistics algorithms ( in seconds ) .
we queried go ten times with each of the five examined enrichment methods using weights from 100 network simulation results and 136 microarrays ( same datasets used for p - value accuracy experiments ) . running times for p - value calculations on dual - core 2.8 ghz amd opteron 254 processors ( using a single core for each run ) aggregated over all samples
are shown on the left , while average times per query are shown on the right . the hgem method used 100-object cutoff .
[ cols= " < , > , > , > , > " , ]
approximating the distribution of sum of weights by saddlepoint method , our saddlesum is able to adapt itself equally well to distributions with widely different properties .
the reported p - values have accuracy comparable to that of the methods based on the hypergeometric distribution while requiring no prior selection of the number of significant entities .
while our results show that gage method suffers from reduced sensitivity , it should be noted that it forms only a part of gage algorithm .
gage was designed to compare two groups of microarrays ( for example disease and control ) by obtaining an overall p - value . in that scheme ,
the p - values we evaluated are used only for one - on - one comparisons between members of two groups . by combining one - on - one p - values ( which are assumed independent )
, the overall p - value obtained by gage can become quite small .
the assumed null distribution by t - profiler @xcite is close to gaussian .
it has been commented @xcite that its statistics are similar to that of page @xcite , which uses z - test .
naturally , the smallest , and likely exaggerated , p - values occur when evaluating small terms .
for that reason , page does not consider terms with less than 10 entities , which we included in our evaluation solely for the purpose of comparison .
our network simulation experiments produce very different weight profiles ( supplementary fig.s4 ) than that of microarrays .
these weights are always positive and skewedly distributed . even after summing many such weights ,
the distribution of the sum is still far from gaussian in the tail . therefore , t - profiler and gage are unable to give accurate statistics .
overall , our evaluations clearly illustrate the inadequacy , even for large terms , of assuming nearly gaussian null distribution when the data is skewed .
while central limit theorem does guarantee convergence to gaussian for large @xmath1 , the convergence may not be sufficiently fast in the tail regions , which influence the statistical accuracy the most . as presented here
, saddlesum uses given @xmath0 both for estimating the @xmath1-dependent score distribution and for scoring each term .
if a certain distribution of weights are prescribed , it is possible to adapt our algorithm to take a histogram for that distribution as input and use experimentally obtained weights for scoring only .
a possible way to improve biological relevance in retrieval is to allow for term - specific weight assignment .
for example , a gene associated with a go term can be assigned a ` not ' qualifier to indicate explicitly that this gene product is not associated with the term considered .
a way to use this information would be to change the sign of the weight for such a gene ( from positive to negative or vice versa ) , but only when scoring the terms where the qualifier applies .
hence , potentially every term could be associated with a specific weight distribution .
while all methods using weights can implement this scheme , saddlesum is particularly suitable for it because it handles well the small terms and skewed distributions , where changing the sign for a single weight can have a considerable effect .
this procedure can be generalized so that each gene in a term carries a different weight .
several authors @xcite have raised the issue of correlation between weights of entities : generally the weights of biologically related genes or proteins change together and therefore a null model assuming independence between weights may result in exaggerated p - values . in principle , a good null model is one that can bring out the difference between signal and noise .
to what level of sophistication a null model should be usually is a trade - off between statistical accuracy and retrieval sensitivity . using protein sequence comparison for example ,
ungapped alignment enjoys a theoretically characterizable statistics @xcite but is not as sensitive as the gapped alignment @xcite , where the score statistics is known only empirically because the null model allows for insertions and deletions of amino acids . incorporating insertion and deletion into the null model made all the difference in retrieval sensitivity .
this is probably because insertions / deletions do occur abundantly in natural evolution of protein sequences .
the ignorance of protein sequence correlations , assumed by both ungapped and gapped alignments , does not seem to cause much harm in retrieval efficacy .
although saddlesum assumes weight independence and thus bears the possibility of exaggerating statistical significance of an identified term , it mitigates this issue by incorporating the entire @xmath0 in the null distribution .
it includes the entities with extreme weights that clearly represent ` signal ' and not ` noise ' , bringing higher the tail of the score distribution and thus larger p - values .
indeed , as shown by examples in fig.[fig : mainexamples1]a and supplementary fig.s3 , saddlesum does not show unreasonably small p - values .
it should also be noted that saddlesum is designed for the simple case where a summary value is available for each entity considered its use for analyzing complex microarray experiments with many subjects divided into several groups is beyond the scope of this paper and care must be exercised when using it in this context .
saddlesum is a versatile enrichment analysis method .
researchers are free to process appropriately their experimental data , produce a suitable @xmath0 as input , and receive accurate term statistics from saddlesum .
since it does not make many assumptions about the distribution of data , we foresee a number of additional applications not limited to genomics or proteomics , for example to literature searches .
this work was supported by the intramural research program of the national library of medicine at national institutes of health .
this study utilized the high - performance computational capabilities of the biowulf linux cluster at the national institutes of health , bethesda , md .
( http://biowulf.nih.gov ) .
we thank john wootton and david landsman for useful comments , roy navon for providing us with the java source code for the statistical algorithms of gorilla , weijun luo for his help with using gage package and the anonymous referees for comments that helped improve the first version of this paper .
al - shahrour , f. _ et al . _
( 2007 ) . from genes to functional classes in the study of biological systems . , * 8 * , 114 .
altschul , s. f. _ et al . _
gapped blast and psi - blast : a new generation of protein database search programs .
, * 25 * , 33893402 . ashburner , m. _ et al . _ ( 2000 ) . gene ontology : tool for the unification of biology . the gene ontology consortium . , * 25 * , 2529 .
backes , c. _ et al . _
( 2007 ) . advanced gene set enrichment analysis .
, * 35*(web server issue ) , w186192 .
barrett , t. _ et al . _
( 2009 ) . .
, * 37*(database issue ) , d885890 .
ben - shaul , y. _ et al . _
( 2005 ) . identifying subtle interrelated changes in functional gene categories using continuous measures of gene expression . , *
21*(7 ) , 11291137 .
benjamini , y. and hochberg , y. ( 1995 ) . controlling the false discovery rate : a practical and powerful approach to multiple testing .
, * 57 * , 289300 .
bleistein , n. ( 1966 ) .
uniform asymptotic expansions of integrals with stationary points and algebraic singularity . , * 19 * , 353370 .
blom , e .-
_ et al . _
( 2007 ) . : functional information viewer and analyzer extracting biological knowledge from transcriptome data of prokaryotes . , * 23*(9 ) , 11611163 .
boorsma , a. _ et al . _
: scoring the activity of predefined groups of genes using gene expression data .
, * 33*(web server issue ) , w592595 .
boyle , e. i. _ et al . _
open source software for accessing gene ontology information and finding significantly enriched gene ontology terms associated with a list of genes . ,
* 20 * , 37103715 .
breitkreutz , b. _ et al . _
( 2008 ) . .
, * 36*(database issue ) , d637640 .
breitling , r. _ et al . _
iterative group analysis ( iga ) : a simple tool to enhance sensitivity and facilitate interpretation of microarray experiments .
, * 5 * , 34 .
breslin , t. _ et al . _
( 2004 ) . comparing functional annotation analyses with catmap . , * 5 * , 193 .
daniels , h. e. ( 1954 ) .
saddlepoint approximations in statistics . , * 25 * , 631650 .
daniels , h. e. ( 1987 ) . tail probability approximations . , * 55*(1 ) , 3748 .
eden , e. _ et al . _
discovering motifs in ranked lists of dna sequences .
, * 3*(3 ) , e39 .
eden , e. _ et al . _
( 2009 ) . : a tool for discovery and visualization of enriched go terms in ranked gene lists . ,
* 10 * , 48 .
goeman , j. and bhlmann , p. ( 2007 ) . . ,
* 23*(8 ) , 980987 . gold , d. _ et al . _
( 2007 ) . .
, * 8*(2 ) , 7177 .
hochberg , y. and tamhane , a. c. ( 1987 ) . .
huang , d. w. _ et al . _
bioinformatics enrichment tools : paths toward the comprehensive functional analysis of large gene lists . , * 37*(1 ) , 113 .
jensen , j. l. ( 1995 ) . .
clarendon press , oxford .
karlin , s. and altschul , s. f. ( 1990 ) .
methods for assessing the statistical significance of molecular sequence features by using general scoring schemes . , * 87 * , 22642268 .
kim , s .- y . and volsky , d. j. ( 2005 ) .
: parametric analysis of gene set enrichment . , * 6 * , 144 .
lugannani , r. and rice , s. ( 1980 ) .
saddle point approximation for the distribution of the sum of independent random variables .
, * 12*(2 ) , 475490 .
luo , w. _ et al . _
: generally applicable gene set enrichment for pathway analysis . , * 10 * , 161 .
mootha , v. k. _ et al . _
genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes . , *
34*(3 ) , 267273 . press , w. h. _ et al . _
( 2007 ) . .
cambridge university press , 3 edition .
sharma , k. _ et al . _
proteomics strategy for quantitative protein interaction profiling in cell extracts .
* 6*(10 ) , 741744 . smid , m. and dorssers , l. c. j. ( 2004 ) .
: functional analysis of gene expression data using the expression level as a score to evaluate gene ontology terms .
, * 20*(16 ) , 26182625 .
stojmirovi , a. and yu , y .- k .
information flow in interaction networks .
, * 14*(8 ) , 11151143 .
stojmirovi , a. and yu , y .- k .
: analyzing information flow in protein networks .
, * 25*(18 ) , 24472449 .
subramanian , a. _ et al . _
gene set enrichment analysis : a knowledge - based approach for interpreting genome - wide expression profiles . , *
102*(43 ) , 1554515550 .
sultan , m. _ et al . _
( 2008 ) . a global view of gene activity and alternative splicing by deep sequencing of the human transcriptome . ,
* 321*(5891 ) , 956960 .
wood , a. t. a. _ et al . _
saddlepoint approximations to the cdf of some statistics with nonnormal limit distributions . , *
88*(422 ) , 680686 .
references about saddlepoint approximations of the tail probabilities of random variables are abundant @xcite . for completeness of our exposition
we here present the derivation of the lugannani - rice formula @xcite , relying extensively on expositions by daniels @xcite and woods , booth and butler @xcite .
let @xmath78 be a continuous random variable supported on a subset of @xmath79 .
we will assume that its probability density function ( pdf ) , denoted by @xmath80 exists and that its moment generating function ( mgf ) , defined by @xmath81 converges for real @xmath82 $ ] where @xmath83 .
recall that @xmath84 gives the characteristic function of @xmath78 , that is , the fourier transform of @xmath80 and that @xmath80 can hence be recovered by the fourier inversion formula : @xmath85 where @xmath86 denotes the cumulant generating function ( cgf ) of @xmath78 .
the tail probability or p - value for a value @xmath87 ( with respect to @xmath78 ) , which we will denote by @xmath88 can be expressed as @xmath89 where @xmath90 is a constant introduced to avoid the pole at @xmath91 .
let @xmath14 denote the sum of @xmath1 independent , identically distributed random variables .
we write @xmath92 , where @xmath93 for all @xmath94 .
our goal is to derive an asymptotic approximation for the tail probability @xmath95 .
it can be easily shown that @xmath96 and hence by ( [ eq : inv_formula2 ] ) @xmath97 to produce our approximation we note that the main contributions to the integral ( [ eq : inv_formula2 ] ) occur in the neighborhood of the pole at @xmath91 and in the neighborhood of the saddle point @xmath98 where the exponent @xmath99 has a maximum , that is , where @xmath100 . the saddlepoint condition is thus @xmath101 or alternatively @xmath102 let @xmath103 denote the expectation of @xmath78 .
daniels @xcite has shown that eq.([eq : saddle1 ] ) has a unique simple root under most conditions . the value of @xmath27 increases with @xmath104 , with @xmath105 . when @xmath106 , the contribution of the pole at @xmath91 to ( [ eq : inv_formula2 ] ) is very small and to obtain an asymptotic approximation to @xmath95 one can proceed by expanding @xmath107 as a taylor s series about @xmath108 and integrating the resulting integral term - by - term @xcite . however , as @xmath104 gets closer to the mean @xmath109 , such approximation performs poorly and in fact is unbounded at the mean .
the essence of the method of @xcite as applied to @xmath95 by lugannani and rice @xcite is to produce a transformation of the integral ( [ eq : inv_formula2 ] ) that would take into account the pole and hence to produce an approximation uniformly valid over the whole range of @xmath14 .
make a transformation from @xmath13 to a new variable @xmath110 by @xmath111 where @xmath112 denotes the gaussian random variable with pdf @xmath113 and @xmath114 and @xmath104 satisfies ( [ eq : saddle1 ] ) . the value @xmath115 is chosen so that the minimum of the left side is equal to the minimum of the right side , which occurs when @xmath98 . since @xmath116 , eq .
( [ eq : transform1 ] ) becomes @xmath117 to find @xmath115 , we set @xmath118 and @xmath108 in ( [ eq : transform2 ] ) to get @xmath119 or , taking the sign for @xmath115 to be equal to the sign of @xmath27 , @xmath120 the transformation ( [ eq : transform2 ] ) maps the region @xmath121 $ ] in @xmath13-space into the region @xmath122 $ ] in @xmath110-space .
the local behavior of @xmath123 , which vanishes at @xmath91 and has zero derivative at @xmath108 is reproduced by @xmath124 with similar behavior at @xmath125 and @xmath118 .
let @xmath126 .
then , @xmath127 expanding @xmath128 about @xmath108 we have @xmath129 where @xmath130 and @xmath131 .
hence , @xmath132 it follows that @xmath133 and @xmath134 are nonzero for all @xmath135 $ ] and @xmath136 $ ] , respectively .
since @xmath137 is analytic in the region of interest , @xmath138 and @xmath139 are analytic over the same intervals .
obviously , the same conclusion follows for @xmath110 as a function of @xmath13 and @xmath13 as a function of @xmath110 . by the inverse function theorem ,
the transformation @xmath140 can be extended to a bijection between complex neighborhoods of @xmath121 $ ] and @xmath122 $ ] .
the integral ( [ eq : inv_formula2 ] ) now transforms ( using cauchy s theorem ) into @xmath141 where @xmath142 . for small @xmath13 , we can write @xmath143 when @xmath144 and hence @xmath145 , differentiating ( [ eq : transform2 ] ) we obtain @xmath146 while when @xmath147 , ( [ eq : expansion1 ] ) implies @xmath148 when @xmath13 is small . thus , @xmath149 and therefore , for small @xmath13 , @xmath150 where @xmath151 is a constant .
let @xmath152 by expanding @xmath128 about @xmath91 , it can be shown that , @xmath153 and , since @xmath154 is analytic , @xmath155 is analytic in the neighborhood of @xmath125 that includes @xmath115 . therefore , we can rewrite the integral ( [ eq : inv_formula3 ] ) as @xmath156 the singularity has now been isolated into ( [ eq : inv_formula4 ] ) , which , by comparing with ( [ eq : inv_formula2 ] ) , we recognize to equal @xmath157 .
on the other hand , @xmath155 can be expanded as a taylor s series around the saddlepoint @xmath118 and integrated to obtain an asymptotic series for ( [ eq : inv_formula5 ] ) .
for the first - order approximation , that is , the leading behavior , we only take the constant term at @xmath115 .
let @xmath158 then , @xmath159 and the integral ( [ eq : inv_formula5 ] ) becomes @xmath160 thus , we have obtained the lugananni - rice formula : @xmath161 with @xmath162 given by ( [ eq : saddle1 ] ) and ( [ eq : hatz2 ] ) .
as mentioned in the main text , our saddlesum algorithm approximates term p - values by first solving eq.([eq : saddle1 ] ) for @xmath22 using newton s method and then using the lugananni - rice formula ( [ eq : tail ] ) .
the key step is estimation of @xmath27 .
since the moment - generating function @xmath163 of the underlying space @xmath10 is not known , we estimate it ( and its derivatives ) using @xmath0 . given sufficiently many weights ( @xmath164 )
, the results can be quite accurate ( see below ) .
one limitation of this approach is that our approximation can only accept scores not greater than @xmath1 times maximal weight ( @xmath27 becomes infinite at this bound ) .
thus , the approximation can be inaccurate for very large scores , causing a larger than usual relative error in p - values ( fig .
however , occurence of such extreme scores is rarely seen in practice .
theoretically , lugananni - rice formula is valid over the whole range of the distribution , for small and large scores and both near the mean and in the tails @xcite .
however , the form ( [ eq : tail ] ) becomes numerically unstable close to the mean of the distribution ( i.e. when @xmath27 is close to 0 ) .
alternative asymptotic approximations exist that are numerically stable near the mean @xcite . for saddlesum , we were mainly interested in the tail probabilities and we therefore decided not to attempt to approximate the p - values of the scores smaller than one standard deviation from the mean ( saddlesum returns p - value of 1 for all such scores ) .
terms with such scores are never significant in the context of enrichment analysis .
when processing a terms database , we retain previously computed values of @xmath27 with associated scores and parameters for lugananni - rice formula in a sorted array . since @xmath27 and the p - value are monotonic with respect to the score , using binary search we can certify for many terms that their p - value is larger than a given cutoff and hence eliminate them without running newton s method .
furthermore , binary search provides a bracket for @xmath27 and hence newton s method usually converges in very few iteration .
we use the bracketed version of newton s method recommended in the numerical recipes book @xcite ( section 9.4 ) .
this combines the classical newton s method with bisection and has guaranteed global convergence .
we show evaluations of saddlesum performance against some theoretically well - characterized distributions in fig .
s5 and s6 .
it can be seen that the relative error between the saddlesum approximation and the theoretical p - value is generally very small except for extremely large scores , when p - values are very small . in the context of the enrichment analysis ,
this discrepancy is not important because such terms will be evaluated as highly significant even if the p - value is off by few orders of magnitude . to further illustrate the quality of our approximation ,
we have computed the kullback - leibler ( kl ) divergences ( relative entropies ) between the tail distribution implied by saddlesum and the theoretical distribution .
prior to computation of kl divergence , both distributions were normalized over the region where saddlesum is valid ( i.e. the tail with scores larger than one standard deviation over the mean ) .
all kl divergence values are extremely small and are comparable between distributions .
s7 shows relative errors of saddlesum compared to the empirical distributions using the same weights and term sizes as for fig .
s1 and s2 . in this case
however , in agreement with the null model of saddlesum , we sampled weights with replacement .
our results indicate that , except for small @xmath1 with weights coming from network flow simulations , the relative error of the saddlesum is similar to that obtained in comparison with well - characterized distributions . | term enrichment analysis facilitates biological interpretation by assigning to experimentally / computationally obtained data annotation associated with terms from controlled vocabularies .
this process usually involves obtaining statistical significance for each vocabulary term and using the most significant terms to describe a given set of biological entities , often associated with weights .
many existing enrichment methods require selections of ( arbitrary number of ) the most significant entities and/or do not account for weights of entities .
others either mandate extensive simulations to obtain statistics or assume normal weight distribution .
in addition , most methods have difficulty assigning correct statistical significance to terms with few entities . implementing the well - known lugananni - rice formula ,
we have developed a novel approach , called saddlesum , that is free from all the aforementioned constraints and evaluated it against several existing methods . with entity weights properly taken into account ,
saddlesum is internally consistent and stable with respect to the choice of number of most significant entities selected . making few assumptions on the input data , the proposed method is universal and can thus be applied to areas beyond analysis of microarrays . employing asymptotic approximation , saddlesum provides a term - size dependent score distribution function that gives rise to accurate statistical significance even for terms with few entities . as a consequence
, saddlesum enables researchers to place confidence in its significance assignments to small terms that are often biologically most specific .
our implementation , which uses bonferroni correction to account for multiple hypotheses testing , is available at http://www.ncbi.nlm.nih.gov / cbbresearch / qmbp / mn / enrich/. source code for the standalone version can be downloaded from ftp://ftp.ncbi.nlm.nih.gov / pub / qmbpmn / saddlesum/. [email protected] * robust and accurate data enrichment statistics via distribution function of sum of weights * aleksandar stojmirovi and yi - kuo yu .2 in national center for biotechnology information + national library of medicine + national institutes of health + bethesda , md 20894 + united states |
let @xmath12 be a connected graph with @xmath13 vertices @xmath14 and having edge set @xmath15 .
a _ configuration _ @xmath16 of @xmath17 pebbles on @xmath0 is an assignment of @xmath17 indistinguishable pebbles to the vertices of @xmath0 .
the notation @xmath18 means that @xmath19 pebbles have been assigned to the vertex @xmath20 .
( notation such as @xmath21 means that @xmath19 pebbles have been assigned to each of the vertices @xmath22 . )
the parameter @xmath17 is known as the _ size _ of @xmath16 , which is also denoted by @xmath23 .
a _ pebbling step _ from vertex @xmath24 to vertex @xmath20 involves the removal of two pebbles from @xmath24 and the placement of one of them onto @xmath20 .
a configuration is @xmath25-_solvable _ if it is possible to place a pebble on the _ root _
vertex @xmath25 via a ( possibly empty ) sequence of pebbling steps .
a configuration is _ solvable _ if it is @xmath25-solvable for all choices of a root @xmath25 .
the _ pebbling number _ of a graph @xmath0 , denoted @xmath1 , is the least number @xmath17 for which every configuration of size @xmath17 is solvable .
one should read @xcite for the history and main results of the theory of pebbling in graphs .
clearly @xmath26 for every @xmath0 , for the configuration which places no pebble on @xmath25 and one pebble on all other vertices is not @xmath25-solvable .
the authors of @xcite defined @xmath0 to be _ class 0 _ when @xmath27 . examples of class 0 graphs include cliques ( via the pigeonhole principle ) and cubes ( see @xcite ) , as well as the 5-cycle , the petersen graph , and many others .
they proved that all 3-connected , diameter 2 graphs are class 0 , and conjectured that fixed diameter graphs with large enough connectivity are also class 0 .
this conjecture was proved in @xcite , where the result was used to prove that the random graph in which each edge appears independently with probability @xmath28 for some @xmath29 is almost surely class 0 .
let @xmath30 be the set of all connected bipartite graphs with @xmath31 vertices in each part .
it was proven in @xcite that every regular graph in @xmath30 having degree at least @xmath32 is class 0 . in this paper
we derive a more general and stronger result for large @xmath31 .
let @xmath33 be the minimum number so that every @xmath34 having minimum degree at least @xmath35 is class 0 .
[ bt ] for all @xmath4 , @xmath36 .
we prove this in section [ bp ] .
we also derive an analogous result for connected graphs .
let @xmath37 be the set of all connected graphs on @xmath9 vertices , and let @xmath38 be the minimum number @xmath39 so that every @xmath40 having minimum degree at least @xmath39 is class 0 .
we prove the following in section [ gp ] .
[ gt ] for all @xmath2 , @xmath41 .
we next consider a randomized version of pebbling , introduced in @xcite , in which we consider the probability space of all configurations of @xmath17 pebbles , each equally likely .
the pebbling number is the minimum @xmath17 for which the probability that a configuration is solvable equals 1 .
now we wish to find @xmath17 so that this probability is nearly 1 . to be more precise ,
let us introduce some notation . for two functions @xmath42 and
@xmath43 we say that @xmath44 ( @xmath45 ) if @xmath46 as @xmath47 .
we set @xmath48 and @xmath49 .
we also write @xmath50 whenever @xmath51 as @xmath47 .
further , we set @xmath52 , and similarly @xmath53 . finally we define @xmath54 .
we consider sequences @xmath55 of graphs for which the number of vertices increases with @xmath9 ( e.g. @xmath56 has @xmath9 vertices ) . for a function @xmath57
we denote by @xmath58 the probability that a randomly chosen configuration of @xmath17 pebbles on @xmath56 is solvable .
a function @xmath59 is a _ pebbling threshold _ for @xmath60 if @xmath61 for all @xmath62 and @xmath63 for all @xmath64 .
we denote by @xmath65 the set of all pebbling thresholds for @xmath60 .
it is not immediately evident that every graph sequence has such a threshold , but it is proven so in @xcite .
the first threshold result , from @xcite , established that @xmath66 , where @xmath67 is the sequence of complete graphs .
more results on the pebbling thresholds of paths , cubes , and other sequences appear in @xcite .
for instance , it is known that if @xmath68 for some graph sequence @xmath60 , then @xmath69 . for our purposes
let us define @xmath70 to be the set of all connected graphs on @xmath9 vertices having minimum degree at least @xmath71 .
let @xmath72 denote any sequence of graphs with each @xmath73 . in section [ dp2 ]
we prove the following theorem .
[ dt2 ] for every function @xmath74 , @xmath75 . in particular ,
if in addition @xmath76 then @xmath77 .
we define , for each @xmath31 , the bipartite graph @xmath80 as follows .
let @xmath81 with @xmath82 and @xmath83 so that @xmath84 and @xmath85 . let the induced subgraphs on @xmath86 and on @xmath87 each be complete bipartite with one missing edge .
let the two missing edges be @xmath88 , with @xmath89 and @xmath90 , and @xmath91 , with @xmath92 and @xmath93
. finally include the two edges @xmath94 and @xmath95 .
note that for @xmath96 the graph @xmath80 has minimum degree @xmath6 , and is @xmath6-regular when @xmath31 is even .
now we define a configuration @xmath16 of size @xmath97 , and show that it is unsolvable when @xmath79 .
we choose the root @xmath98 and define @xmath99 .
we find @xmath100 and define @xmath101 and @xmath102 .
finally we define @xmath103 for all other vertices @xmath20 . clearly , @xmath104 . in order that @xmath16 is @xmath25-solvable one must be able to move 2 pebbles onto either @xmath105 or @xmath106 , and consequently 4 pebbles onto either @xmath19 or @xmath107 .
it is not difficult to see that both cases are impossible , since at most three pebbles can be put in motion via pebbling steps from @xmath108 , and @xmath109 .
hence , for all @xmath79 , @xmath80 is not class 0 and so @xmath78 .
let @xmath111 have bipartition @xmath112 and minimum degree at least @xmath5 , where @xmath81 .
choose any configuration @xmath16 of size @xmath113 and let @xmath25 be any chosen root ( which we may assume lies in @xmath114 ) .
we assume that @xmath16 is @xmath25-unsolvable and derive a contradiction .
we will make use of the following two observations about @xmath115 .
first , every pair of vertices in the same part has a common neighbor .
second , from this it is clear that the diameter of @xmath115 is at most 3 .
we will derive a contradiction by accumulating 8 pebbles on some vertex , from which we can obviously pebble to @xmath25 .
denote the neighborhood of a vertex @xmath20 by @xmath116 , and the union of neighborhoods of a set @xmath117 of vertices by @xmath118 .
then we must have @xmath119 and @xmath120 for all @xmath121 .
we also know that @xmath122 for all @xmath123 .
let @xmath124 , @xmath125 , and @xmath126 .
we let @xmath127 , with @xmath128 , @xmath129 , @xmath130 , @xmath131 , and @xmath132 defined analogously . _
_ the claim is trivial if @xmath134 so we assume otherwise and pick some @xmath135 .
now we know that we can not move another pebble to @xmath136 .
we note that @xmath137 is nonempty because @xmath138 is nonempty .
also we note that if there is some vertex @xmath20 with @xmath139 , then we can move a pebble to either @xmath25 or @xmath136 .
hence we assume @xmath122 for all @xmath20 .
moreover , we note that it must be impossible to ever put 4 pebbles on any vertex .
suppose that there is a vertex @xmath17 having @xmath140 ; without loss of generality we assume that @xmath141 ( if @xmath142 , we think of @xmath136 as our new root and argue similarly ) . since @xmath143 , we know that @xmath144
. moreover , @xmath145 implies @xmath133 , so we assume otherwise and pick @xmath146 .
then @xmath147 , and so @xmath148 , which implies that @xmath149 .
if there is a vertex @xmath150 then we can argue as follows .
let @xmath151 , @xmath152 , and @xmath153 .
of course , @xmath154 ( @xmath155 since otherwise we could place 4 pebbles on @xmath17 ) , and @xmath156 .
therefore we have that which implies that @xmath133 .
if instead there is a vertex @xmath157 then we know that , either @xmath133 because @xmath158 , or there is some @xmath159 . in the latter case we move a pebble from @xmath160 to @xmath161 and argue as above .
henceforth we may assume that @xmath162 for all @xmath20 .
consequently the equality @xmath163 tells us that @xmath164 , and so @xmath165 .
therefore , again without loss of generality , @xmath166 , say @xmath167 .
if @xmath155 then we may copy the above argument that @xmath168 implies @xmath133 .
otherwise we may move a pebble from @xmath161 through @xmath169 to @xmath17 , find @xmath170 and use the original argument for the case that @xmath140 .
this completes the proof of the claim .
now we can use the relations @xmath171 to see that @xmath172 so that @xmath173 . from this
we can assume , without loss of generality , that @xmath174 , so that the number of edges with one end in @xmath131 is more than @xmath175 .
since @xmath4 there must be some @xmath176 having at least 8 neighbors in @xmath131 , so that we can put 8 pebbles on @xmath19 , a contradiction .
we define , for each @xmath9 , the graph @xmath56 as follows .
let the vertex set @xmath178 , with @xmath179 and @xmath180 .
let the induced subgraphs on @xmath114 and on @xmath181 each be complete with one missing edge .
suppose the edge @xmath88 is missing from the subgraph on @xmath114 , and the edge @xmath91 is missing from the subgraph on @xmath181 .
finally include the two edges @xmath94 and @xmath95 .
note that for @xmath182 the graph @xmath56 has minimum degree @xmath6 , and is @xmath6-regular when @xmath9 is even .
next we define a configuration @xmath16 of size @xmath183 , and show that it is unsolvable when @xmath2 .
we choose the root @xmath184 and define @xmath99 .
we find @xmath185 and define @xmath101 and @xmath102 .
finally we define @xmath103 for all other vertices @xmath20 . clearly , @xmath104 . in order that @xmath16 is @xmath25-solvable one must be able to move 2 pebbles onto either @xmath105 or @xmath106 , and consequently 4 pebbles onto either @xmath19 or @xmath107 .
it is not difficult to see that both cases are impossible .
hence for all @xmath2 , @xmath56 is not class 0 , and so @xmath186 .
let @xmath0 be graph with minimum degree @xmath3 .
we suppose that @xmath0 is not class 0 and derive a contradiction . because complete graphs are class 0
, @xmath0 has diameter at least 2 , and because every pair of vertices of @xmath0 has a common neighbor , the diameter of @xmath0 is exactly 2 .
it is proven in @xcite that every graph @xmath0 of diameter two has pebbling number @xmath189 or @xmath190 ( _ class 1 _ ) . in @xcite
we find the following characterization of class 1 graphs of diameter two ( see figure [ class1 ] ) . in the figure
, @xmath191 is any ( possibly empty ) graph , @xmath192 is any nonempty graph , and @xmath193 and @xmath194 are any nonempty connected graphs .
the solid lines indicate an edge from the given vertex to every vertex in the corresponding set .
at least two of the three dashed lines must be present , and the arrows indicate that every vertex in @xmath191 must have at least one edge to @xmath195 .
suppose that @xmath0 is labeled as in figure [ class1 ] .
choose any vertices @xmath196 , and recall that @xmath197 denotes the neighborhood of a vertex @xmath19 .
since @xmath198 is not an edge , @xmath199 and @xmath200 , we have @xmath201 , and so at least one of @xmath202 or @xmath203 has degree smaller than @xmath3 , a contradiction .
hence @xmath0 is class 0 . in this section
we prove theorem [ dt2 ] .
the proof is divided into two steps .
first , we will show that it is possible to partition the vertices of @xmath205 into @xmath206 subgraphs of diameter at most two .
second , we prove that if a distribution has enough pebbles then there will be in every subgraph `` many '' vertices with two pebbles each .
the following procedure constructs a @xmath212-star partition @xmath213 of @xmath208 with @xmath214 .
select @xmath215 arbitrarily and let @xmath216 . for a general step ,
suppose @xmath217 have been selected and let @xmath218 .
either every vertex from @xmath219 has a neighbor in @xmath210 for some @xmath220 , in which case we stop the process with @xmath221 and @xmath222 , or there is a vertex @xmath223 such that @xmath224 . in the latter case
we set @xmath225 and continue the process .
assume that @xmath226 with @xmath227 and let @xmath16 be a pebbling distribution with @xmath228 pebbles where @xmath229 is such that @xmath230 . consider the random variable @xmath231 , which is equal to one if @xmath232 and zero otherwise , and let @xmath233 .
then @xmath234 = \frac{{t+n-4 \choose t-2}}{{t+n-1 \choose t}}\ ; \ ] ] using the assumptions about @xmath17 and @xmath9 it is easy to check that @xmath235 \leq \e[x_{ij}]\e[x_{ik}]$ ] for @xmath236 . consequently @xmath237
= \e[x_i^2 ] - \e[x_i]^2 \leq \e[x_i]\ , \ ] ] and by chebyshev s inequality @xmath238}{2}\ ] \leq \pr\[|x_i - \e[x_i]| \geq \frac{\e[x_i]}{2}\ ] \leq \frac{4}{\e[x_i]}\ .\ ] ] thus the probability that there exists an @xmath239 such that @xmath240/2 $ ] is @xmath241)$ ] .
but , with our choice of @xmath17 and @xmath9 , @xmath242 = \w\(\frac{\d t^2}{n^2}\ ) = \w\(\frac{\w^2 n}{\d}\)\ .\ ] ] thus , with probability tending to one , for all @xmath243 , we have @xmath244/2 $ ] , which is at least 8 for large enough @xmath9 .
therefore , with probability tending to one , 4 pebbles can be accumulated on the center of every star , and since every vertex is within distance two of some center , it is possible to move a pebble to any given root vertex .
99 a. bekmetjev , g. brightwell , a. czygrinow and g. hurlbert , _
thresholds for families of multisets , with an application to graph pebbling _ , discrete math .
, to appear .
chung , _ pebbling in hypercubes _ , siam j. disc .
* 2 * ( 1989 ) , 467472 .
t. clarke , _ pebbling on graphs _ , master s thesis , arizona st .
t. clarke , r. hochberg , and g. hurlbert , _ pebbling in diameter two graphs and products of paths _ , j. graph th .
* 25 * ( 1997 ) , 119128 .
a. czygrinow , n. eaton , g. hurlbert and p.m. kayll , _ on pebbling threshold functions for graph sequences _ , discrete math .
, * 247/1 - 3 * ( 2002 ) , 93105 .
a. czygrinow , g. hurlbert , h. kierstead and w.t .
trotter , _ a note on graph pebbling _ , graphs and combin . , to appear .
a. czygrinow and m. wagner , _ on a conjecture about pebbling thresholds _ , preprint ( 2002 ) .
a. godbole , m. jablonski , j. salzman and a. wierman , _ an improved upper bound for the pebbling threshold of the @xmath9-path _ , preprint ( 2002 ) .
g. hurlbert , _ two pebbling theorems _ , congress
* 135 * ( 1998 ) , 5563 .
g. hurlbert , _ a survey of graph pebbling _ , congress . numer .
* 139 * ( 1999 ) , 4164 .
l. pachter , h.s .
snevily , and b. voxman , _ on pebbling graphs _ , congress .
numer . * 107 * ( 1995 ) , 6580 . | a configuration of pebbles on the vertices of a graph is solvable if one can place a pebble on any given root vertex via a sequence of pebbling steps .
the pebbling number of a graph @xmath0 is the minimum number @xmath1 so that every configuration of @xmath1 pebbles is solvable .
a graph is class 0 if its pebbling number equals its number of vertices .
a function is a pebbling threshold for a sequence of graphs if a randomly chosen configuration of asymptotically more pebbles is almost surely solvable , while one of asymptotically fewer pebbles is almost surely not .
here we prove that graphs on @xmath2 vertices having minimum degree at least @xmath3 are class 0 , as are bipartite graphs with @xmath4 vertices in each part having minimum degree at least @xmath5 .
both bounds are best possible .
in addition , we prove that the pebbling threshold of graphs with minimum degree @xmath6 , with @xmath7 , is @xmath8 , which is tight when @xmath6 is proportional to @xmath9 .
* 2000 ams subject classification : * 05d05 , 05c35 , 05a20 p s ( ( ) ) ^ @xmath10 ) @xmath11 ] |
* dual time scales in simulated annealing * we here provide technical details and some results complementing those in the main paper .
details of the simulated annealing and parallel tempering procedures are given in sec .
a and b , respectively . the scaling formalism underlying the extraction of the dynamic exponent is discussed in sec . c. illustrations for large systems in the power - law velocity scaling regime are presented in c-1 , and in c-2 data for small systems approaching equilibrium are presented , to show the ability of the scaling form to account for both velocity regimes . in sec .
d we presented data - collapse plots for @xmath63 and @xmath116 , complementing the @xmath60 graphs shown in the main paper .
the scaling of the minimum energy reached in the sa runs is discussed in sec .
e. we code the spins @xmath117 of the model ( [ ham ] ) as bits of long ( 64-bit ) integers , thus using @xmath56 integers @xmath118 for a system of @xmath56 spins and propagating 64 replicas of the same system ( with the same random couplings ) .
the bimodal couplings @xmath6 are also encoded as bits @xmath119 , and most of the operations involved in computing energy differences for the metropolis acceptance probabilities for single - spin flips ( with the same spin considered in all replicas ) can then be carried out simultaneously on all 64 replicas by using standard bit - vise logical operations on the stored integers . in the beginning of each repetition of the sa ,
we generate new random couplings and initialize the spins to a random configuration .
we then carry out 10 mc sweeps at the initial temperature @xmath57 .
we found that this small number of initial steps is sufficient for reaching very close to thermal equilibrium at this high temperature ( and note that any deviation from equilibrium at this stage can be regarded as just a different initial state and will not affect the scaling when @xmath7 at low velocities ) .
in the subsequent sa run we carry out @xmath58 mc sweeps and lower the temperature after each sweep according to the protocol in eq .
( [ tprotocol ] ) , carrying out measurements of the ea order parameter and the energy after the final @xmath0 sweep . to compute the ea order parameter ( [ ea ] ) we form 32 pairs out of the 64 replicas and again use bit operations , thus obtaining 32 contributions from each measurement .
the safest way to ensure independent propagation of the replicas is to generate different random numbers for the final accept / reject step for each replica , in which case the generation of the random numbers consumes a large fraction of the computation time . strictly speaking ,
uncorrelated replicas are required only when computing the ea order parameter ; any other correlations are unbiased ( provided that the random number generator is not flawed ) , thus not causing systematical error in the final results but potentially reducing the effective amount of statistical data .
for example , if the same random number is used for each replica , if ever two replicas go into the same state they will stay in the same state for the remainder of the simulation , thus reducing the number of independent replicas .
no statistical bias is introduced in computed mean values , however .
once the system size is reasonably large , it is very unlikely for replicas to lock to each other in this way , and one can sefely use the same random numbers within the two groups of 32 replicas between which the ea order is computed . in our pt simulations @xcite
, we again use the bit representation but now all the bits @xmath120 correspond to different temperatures on a uniform grid , @xmath121 . attempts to swap spin configurations of runs at adjacent temperatures
@xmath122 are carried out after each mc sweep over the spins .
the goal of the pt simulations is to obtain @xmath7 equilibrium results for the ea order parameter and the ground state energy ( results shown in fig .
[ fig1 ] ) . for the latter , we do not use the thermal energy average but simply keep track of the lowest energy reached in each run and average it over the coupling samples .
we here present results showing proper convergence as the number of sweeps is increased .
we choose the lowest temperature @xmath123 such that the @xmath2 dependence of the ea order parameter is insignificant in the neighborhood of this temperature for the system sizes studied , i.e. , @xmath2 is below the size - dependent entropic cross - over temperature @xmath30 @xcite discussed in the main paper .
the highest temperature should be high enough for the thermal fluctuations to migrate efficiently from high to low temperatures , thereby enhancing the ergodicity of the pt simulations relative to independent fixed-@xmath2 runs .
significant migration of the fluctuations also necessitates a sufficiently small @xmath124 , and in principle optimal simulations would also have @xmath124 decreasing with @xmath49 and the number of temperatures increasing .
here we use @xmath125 replicas and the spacing is @xmath126 or @xmath127 .
figure [ ptconv ] shows examples of the convergence of the ea order parameter and the lowest energy as functions of the number of sweeps in the pt simulations .
we use the same number of sweeps for equilibration and for collecting data .
the @xmath88-axis of fig .
[ ptconv ] corresponds to the sweeps for data collection only ( i.e. , the total number of sweeps is twice this number ) and each successive point corresponds to doubling the number of sweeps . for these system sizes , @xmath128 and @xmath129
, the ground state energy converges faster than the order parameter , but this trend is clearer for @xmath128 than @xmath17 .
it is likely that the energy will converge _ slower _ than the energy for large sizes , as in the sa simulations ( see also sec .
e below ) .
we have not studied the scaling properties of the pt scheme .
for acceptable convergence , we require statistically indistingusiable results from at least the last two runs in a series of runs such as those in fig .
[ ptconv ] . based on this criterion
we have converged results for @xmath44 up to @xmath17 and for @xmath110 up to @xmath16 . to ensure that we obtain @xmath115 results , it is also important to check the temperature dependence of the results .
[ m24 ] shows results for @xmath17 , from pt runs with the largest number of sweeps [ corresponding to the last point in fig .
[ ptconv](b ) ] . here
we can see that there is very little @xmath2-dependence below @xmath130 .
we estimate that the very small remaining finite - temperature effect at @xmath131 is much smaller than the statistical error .
according to the general non - equilibrium finite - size scaling form in @xcite , adapted to the present case where @xmath132 is replaced by @xmath98 , the squared order parameter @xmath133 can be written in the following way in three different regimes : @xmath134 where @xmath135 is the generalized kz velocity separating the near - equilibrium and power - law scaling behaviors .
recall that @xmath133 is the rescaled quantity obtained by dividing @xmath44 by the multiplicative finite - size correction @xmath136 , with @xmath137 and @xmath138 corresponding to the fitted constants in fig .
[ fig1 ] of the main paper , and this quantity approaches @xmath139 for large system sizes ( provided that the form of the correction is correct ) in equilibrium ( @xmath140 ) .
the factors @xmath141 in eq .
( [ eq : q_scaling ] ) represent the overall size dependence in the limit where @xmath142 and the exponent @xmath88 has to be given by @xmath143 in order for the two expressions on the middle line to be equal .
the taylor - expandable near - equilibrium behavior on the first line of eq . ( [ eq : q_scaling ] ) should smoothly connect to the first power - law form on the second line , through a cross - over region in the scaling function @xmath144 [ eq .
( 5 ) in the main paper ] . in the high - velocity limit , the third case above , the behavior can be expressed as a series in @xmath145 , and this series has to be smoothly connected to the form @xmath146 on the second line . forms analogous to eq .
( [ eq : q_scaling ] ) hold for other singular quantities , where the rescaling on the left - hand side of eq .
( [ eq : q_scaling ] ) always corresponds to dividing out the critical size - dependennce in equilibrium , i.e. , the factor @xmath147 in eq .
( 2 ) ( and in principle finite - size corrections can also be divided out , as we do here ) and @xmath38 on the second and third lines should be replaced by @xmath148 ( or , alternatively , @xmath149 should be left out on both sides ) .
we here focus on the ea order parameter but the scaling procedures are completely analogous for other quantities , e.g. , the excess energy for which we present some additional results in sec . c-2 . for convenience
we denote the often occurring generalized kz exponent by @xmath150 , @xmath151 one can use eq .
( [ eq : q_scaling ] ) for given @xmath72 to extract this exponent either from the high - velocity side , by fitting a straight line to @xmath152 versus @xmath153 , the slope of this line being the exponent @xmath154 in eq .
( [ x_exponent ] ) , or by adjusting @xmath150 so that @xmath133 versus @xmath155 in the power - law and equilibrium regimes collapse onto a common scaling function for different @xmath49 .
an analysis from the high - velocity side can include data only in the power - law regime , unless high - velocity corrections are taken into account . the data - collapse method potentially can lead to better statistical precision if a substantial amount of data are available in the low - velocity cross - over and equilibrium regimes , where the power - law scaling no longer holds . to disentangle eq .
( [ a ] ) and obtain the exponents @xmath85 and @xmath98 ( if its value is not known or one would like to test consistency with a known value , as we do here ) , it is in principle sufficient to work with two different values of @xmath72 in the annealing protocol , eq.([tprotocol ] ) , and extract @xmath156 and @xmath157 . here , as a further consistency check we use three different values , @xmath158 , and fit the resulting @xmath150 to the expected form ( [ a ] ) with @xmath85 and @xmath98 optimized for the best fit .
the procedure was illustrated in fig .
[ fig2](c ) in the main paper , and the error bars on the final results for the exponents were estimated by repeating the fitting procedure several hundred times with gaussian noise added to the original @xmath150 values . here
we give further details on the data - collapse procedures .
below , in sec .
c-1 we first consider @xmath133 for large systems in the power - law regime , and in c-2 we discuss both the ea order parameter and the excess energy for smaller systems in the regime where equilibrium is approached .
we here take @xmath60 as an example and in the following simply use @xmath159 for @xmath160 . in fig .
[ fig2](a , b ) we already illustrated how data are collapsed by optimizing @xmath159 . to characterize the goodness of the data collapse
, we fit a high - order polynomial to a set of data points @xmath161 for different @xmath62 and @xmath49 , sweeping over @xmath159 on a dense grid and locating the optimal value ( minimum @xmath162 for the fit ) .
if a satisfactory collapse , @xmath163 , can not be achieved we systematically eliminate small system sizes and/or high - velocity points until a statistically good fit is obtained .
typically tens of data points are left in the good fit . to estimate error bars , we perform bootstrapping , repeating the fitting procedure with many bootstrap samples and computing the standard deviation of the optimal @xmath159 . of the goodness of the linear fit of @xmath152 versus @xmath164
when the slope of the line fixed at @xmath165 .
the data used here is shown as the middle set ( triangles ) in fig .
[ fig : linear ] .
a scan is performed as a function of the exponent @xmath166 and the minimum of @xmath162 is identified for the optimum @xmath159.,width=283 ] here , for illustration puroses we discuss a slightly simpler method for analyzing only the power - law regime and including only the three or four lowest available velocities for three system sizes ; @xmath167 . for these sizes , even at the lowest velocity that we have studied , @xmath168 , the systems are far from equilibrium but , as we will show , they fall within the power - law scaling regime described by the middle line in eq .
( [ eq : q_scaling ] ) .
graphing on a log - log scale , we then expect all points to fall on a common line with slope @xmath88 given by eq .
( [ x_exponent ] ) if the horizontal axis is appropriately rescaled as @xmath169 .
we use the required @xmath60 line slope @xmath165 to constrain the fit to the form @xmath170 , i.e. , for given @xmath159 in the scaling procedure @xmath138 is the only adjustable parameter .
we scan over a dense grid of @xmath159 values , perform the constrained line fit for each case , and keep track of @xmath162 to locate the minimum value ; see fig .
[ fig : x2sigma ] for an illustration .
the optimum @xmath159 value is the result .
alternatively , according to the second form of the middle line in eq .
( [ eq : q_scaling ] ) , we could also just consider @xmath171 versus @xmath153 and extract the slope @xmath172 ( and again a good @xmath162 value would be an indication of being within the power - law scaling regime ) . the approach discussed here can , however , also be generalized to include low - velocity data , where the power - law scaling no longer holds but the behavior is described by the scaling function @xmath173 , of which the power - law constitutes the limiting form for large @xmath169 .
this latter part can be fitted to a line , and points deviating from it for smaller @xmath169 can be simultaneously fitted to a polynomial @xcite .
here we just consider the linear part , while in fig .
[ fig2 ] we also included lower-@xmath62 data but did not constrain the collapse by the line slope ( instead obtaining the slope as a consistency check ) .
( black ) , @xmath174 ( red ) , and @xmath175 ( blue ) .
the three lines are the best fits to the form @xmath176 , and the values of @xmath177 shown for each case corresponds to a @xmath162-minimum such as the one in fig .
[ fig : x2sigma].,width=302 ] figure [ fig : linear ] shows three different data sets along with the corresponding slope - constrained line fits .
the middle set of points is the original data set , while the left and right sets correspond to the extreme cases out of @xmath178 bootstrap samples .
the standard deviation of @xmath159 computed from the bootstrap samples directly gives the error bar ; in this case @xmath179 .
this value is completely consistent with the value in the caption of fig .
[ fig2 ] , but the error bar is larger because only the linear regime was used and the number of data points is smaller . note that the same coupling realizations are used in sa runs with all velocities ( where @xmath62 is if the form @xmath180 for positive integers @xmath181 ) , and the data points for the same system size but different @xmath62 are therefore strongly correlated ( the sample - to - sample fluctuations being much larger than the mc sampling noise ) .
the covariance predominantly corresponds to common up or down fluctuations of the value of the order parameter , and therefore the optimum line slope , as extracted above , is not significantly affected , and it is not necessary to use the full covariance matrix in the fitting procedure .
the bootstrapping procedure properly account for the covariance since the same reandom bins are randomly chosen for all velocities for a given @xmath49 . using the same system sizes and velocities and repeating the same procedures for @xmath63 and @xmath64 , we obtain @xmath182 and @xmath183 . combining these results and performing a fit to the expected @xmath72 dependence of @xmath150 , eq .
( [ a ] ) , we obtain @xmath184 , and @xmath185 , as shown in fig .
[ fig : a(r ) ] .
these values are consistent with those presented in the main paper , but again the error bars are larger due to the smaller amount of data used
. we can then conclude that the inclusion of also smaller sizes and lower velocities ( including some data away from the power - law regime ) in fig .
[ fig1 ] did not change the exponents to a noticable degree relative to the case here , where only large system sizes far from equilibrium were used . in the next section
we go even deeper into the quasi - static regime . for @xmath158 along with a fit to the form ( [ a ] ) .
the individual exponents ( fitting parameters ) are @xmath184 , and @xmath185 , where the error bars were computed by repeated fits with gaussian noise added [ with standard deviaton equal to the error bars on @xmath150].,width=283 ] in the main paper and the previous section we discussed velocity scaling for systems sufficiently large for no significant subleading finite - size scaling corrections to remain ( to within the statistical precision of the data ) . for large system sizes we can reach well into the power - law scaling regime ( the linear part of the scaling function graphed on a log - log scale ) , but not very far into the cross - over into the regime where the systems approach and reach equilibrium , i.e. , corresponding to the first line in eq .
( [ eq : q_scaling ] ) .
it is important to test the scaling behavior also here , to make sure that the final relaxation stage is governed by the same dynamic exponent as the power - law regime .
figure [ qesmall ] shows @xmath60 results for lattice sizes in the range @xmath187 to @xmath188 , with the exponent @xmath159 adjusted for best overall data collapse . here
the data - collapse procedure included all the system sizes shown , again excluding high velocities where no data collapse can be expected .
we obtain the exponents @xmath189 and @xmath190 for the ea order parameter and the energy , respectively .
these values are very close to those obtained in figs .
[ fig1 ] and [ fig2 ] in the main paper , @xmath191 and @xmath192 , respectively , demonstrating the stability of the results .
we did not include the smallest systems in the analysis in the main paper because , although the exponent values do not differ much , we can not obtain a statistically satisfactory value of the goodness of the fit ( @xmath162 per degree of freedom close to @xmath139 ) when all data are included in a common fit , given the small error bars of the sa data and small but statistically significant effects of neglected finite - size scaling corrections for the smaller systems .
given the good agreement we have demonstrated here and in the main paper between different system sizes and velocity regimes , we judge that the significant difference between the dynamic exponent for the excess energy and the ea order parameter , @xmath193 , can not be explained by neglected scaling corrections .
the dual time scales are therefore a real aspect of the relaxation of the 2d @xmath194 spin glass , and , as explained around fig .
[ fig4 ] in the main paper , this unusual physics also fits naturally within the droplet theory of the 2d spin glass . for completeness we here present the data for @xmath63 and @xmath116 , analyzed in the same way as the @xmath60 data in figs .
[ fig2](b ) and [ fig3 ] .
data - collapse plots for the ea order parameter and the excess energy are presented in figs .
[ r2r4q ] and [ r2r4e ] , respectively .
the exponent values are given in the figure captions .
when optimizing with sa , it is better to keep track of the minimum energy @xmath108 reached during the entire run for each disorder sample ( as we also do in the pt simulations discussed above in sec .
a ) . even in very slow annealings
this energy is occasionally lower than the energy after the final mc step at @xmath0 .
therefore , the disorder - averaged @xmath109 should be lower than @xmath110 .
an important question then is whether the scaling of the two quantities is the same or not .
figure [ emin ] shows results for @xmath60 , scaled using the same exponents as in fig . [ fig3 ] in the main paper .
the scaling collapse is still very good , and the optimized scaling exponent for this case is also statistically equal to the one obtained before .
overall the minimum energy values are , as expected , below those for the mean energy . with the range of system sizes used here we can see the full equilibrium behavior ( the flat portion , where the value corresponds to the prefactor of the @xmath38 correction in fig . [ fig1 ] of the main paper ) as well as the cross - over into the power - law scaling regime . in the graph
we also draw a straight line with exactly the same parameters as the line drawn through the power - law scaling portion of the collapsed mean energy data in fig .
[ fig3 ] . in @xmath195
we observe that larger systems are needed to observe the same slope
we see that the scaling function ( onto which the data collapse ) exhibits some curvature .
nevertheless , with increasing size the functional form appears to approach a line with the same slope as before .
it is possible that it actually approaches exactly the same line ( not just the same slope but also the same constant ) as the one for @xmath196 in fig .
if so the asymptotic power - law scaling of the two quantities would be exactly the same , and the advantages ( in optimization applications ) of monitoring @xmath108 instead of @xmath197 would only appear as the behavior crosses over toward the equilibrium behavior .
we conclude that the minimum energy collected during sa runs converges to the ground state energy on the same time scale @xmath198 as the convergence of the mean energy .
the scaling functions are different , reflecting an overall lower value of the minimum energy than the mean energy for given scaled velocity @xmath199 . in fig .
[ function ] we show the two @xmath60 scaling functions explicitly , by combining data from other figures but only including the points that fall sufficiently close to a common scaling function . here one can read off that @xmath108 ultimately converges ( the curve flattens out to a constant ) about @xmath200 times faster than @xmath197 .
this factor depends on the details of how @xmath108 is computed in the simulations . in our case
, we carried out 64 simulations in parallel for each coupling sample and monitored the lowest energy in any of these simulations .
clearly , upon increasing the number of parallel runs @xmath108 will converge faster , thus pushing the scaling function further to the right . | we apply a generalized kibble - zurek out - of - equilibrium scaling ansatz to simulated annealing when approaching the spin - glass transition at temperature @xmath0 of the two - dimensional ising model with random @xmath1 couplings . analyzing the spin - glass order parameter and the excess energy as functions of the system size and the annealing velocity , we find scaling where the energy relaxes slower than the order parameter , i.e. , there are two different dynamic exponents .
we argue that such behavior arises as a consequence of the entropy - driven ordering mechanism within droplet theory . in the context of optimization
, our study provides an example of robust dense - region solutions for which the excess energy ( the conventional cost function ) may not be the best measure of success .
a simulated annealing ( sa ) process @xcite carried out on a system with a continuous phase transition exhibits scaling with the system size and the annealing velocity ( the rate of change of the temperature @xmath2 versus time ) . following the seminal analysis by kibble @xcite and zurek @xcite ( kz ) of the `` freezing '' of defects close to a critical point
, a compelling picture has emerged of the combined effects of finite size and velocity on physical observables in sa @xcite .
the generalization of kz scaling to quantum systems ( where a system parameter is changed as a function of time at low @xmath2 ) @xcite has found applications in studies of cold atom systems @xcite , and should be of relevance also in the quantum annealing ( qa ) @xcite ( or quantum - adiabatic @xcite ) approach to solving hard optimization problems by adiabatically evolving a programmable qubit system from a trivial to a complex ground state @xcite .
an untested application of classical kz scaling is to systems with critical temperature @xmath3 .
a prominent example of such a case is the two - dimensional ( 2d ) ising spin glass , which is interesting not only in its own right but also in the context of quantum annealing , e.g. , the devices produced by d - wave systems @xcite are laid out with a particular 2d connectivity .
numerous studies of 2d ising spin glasses have been carried out recently in order to compare sa and qa , to gain insights into the nature of the quantum and thermal fluctuations in qa devices , and to develop methods for analyzing the efficiency of annealing protocols @xcite .
the kz scaling formalism has not been applied , however .
we here present such a study of the 2d ising spin glass with bimodal couplings and find an unusual behavior where two different dynamic exponents govern the equilibration of the spin - glass ( edwards - anderson , ea ) order parameter and the excess energy , with the energy relaxing slower than the ea order parameter .
we argue that this unusual behavior is a consequence of the entropy - driven ea ordering process within droplet theory @xcite ) .
the 2d ising spin glass is defined by the hamiltonian @xmath4 with random nearest - neighbor couplings @xmath5 drawn from , e.g. , a bimodal or normal distribution .
the equilibrium properties of the bimodal @xmath6 model , which we focus on here , have been controversial .
a long - standing issue has been to distinguish between exponential @xcite and power - law @xcite scaling as @xmath7 .
the nature of the state at @xmath0 has also been difficult to ascertain .
until recently it was widely believed that the @xmath6 system does not harbor spin - glass order ( unlike the model with normal - distributed couplings ) , only power - law decaying critical ea spin - spin correlations @xcite .
more recent studies @xcite point to significant long - range order .
in particular , thomas _ et al . _
@xcite evaluated the pfaffian form of the partition function on larger lattices and lower temperatures than in previous monte carlo ( mc ) studies .
a quantitative picture was presented for finite - size corrections to the long - range order at @xmath0 , power - law scaling at @xmath8 , and a size - dependent cross - over temperature @xmath9 , with @xmath10 when @xmath11 , below which the discreteness of the couplings is important .
we here use out - of - equilibrium ( sa ) mc simulations to study the model with @xmath12 spins on periodic square lattices .
we generate the @xmath13 couplings independently with probability @xmath14 and use bit representations for both the spins and the couplings ( as discussed in detail in supplemental material , sm @xcite ) , runing 64 independent parallel simulations for each realization ( sample ) of the couplings and repeat for many samples .
in addition to sa , where we go to system sizes up to @xmath15 , we have also used parallel tempering ( pt ) @xcite to obtain equilibrium @xmath7 results for smaller systems ( up to @xmath16 for the energy and @xmath17 for the ea order parameter ) ; see sm @xcite for convergence tests . although larger systems ( with open boundaries ) can be studied with ground - state methods @xcite , proper thermodynamic averages of the ea order parameter , with equal weighting of degenerate states , are difficult to obtain @xcite .
we discuss the equilibrium pt results first because they will be important in the kz scaling analysis .
averages were computed over millions of coupling samples for the smaller sizes and about @xmath18 samples for the largest systems .
the standard finite - size scaling ansatz @xcite for a quantity @xmath19 scaling as @xmath20 in the thermodynamic limit is @xmath21 $ ] , where the exponent @xmath22 governs the correlation length , @xmath23 , and the scaling function @xmath24 must have the form @xmath25 for @xmath26 to ensure the correct thermodynamic limit . with this form ,
the singular behavior in a system of finite length is cut off at @xmath27 , i.e. , at @xmath28 . in the 2d @xmath29 spin glass @xmath3 and a low size - dependent energy scale
was identified , @xmath30 , where @xmath31 is an exponent quantifying the entropy due to zero - energy clusters ; flipping a cluster of linear size @xmath32 changes the entropy by @xmath33 @xcite .
the finite - size scaling relation then changes to @xmath34 thomas _ et al .
_ showed that the specific heat exponent is @xmath35 @xcite .
then , at @xmath0 , eq . ( [ alt ] ) with @xmath36 predicts that the finite - size energy correction ( per spin ) should be @xmath37 .
this form was obtained based on a different scenario in @xcite and was consistent with data for periodic systems . in our pt simulations we generated a much larger number of samples for all systems up to @xmath16 , to obtain a more reliable estimate of the @xmath38 term . as shown in fig .
[ fig1](a ) , the agreement with the prediction is excellent .
the extrapolated infinite - size energy , @xmath39 , is in good agreement with the best previous result , @xmath40 , from open - boundary systems @xcite ( see also @xcite ) . using the more precise value to constrain the fit we obtan @xmath41 with @xmath42 .
-2 mm -2 mm the ea order parameter is defined with two replicas ( independent simulations with the same couplings ) as @xmath43 we evaluate @xmath44 and extrapolate it to infinite size , as shown in fig .
[ fig1](b ) . since long - range order
is expected at @xmath0 , the exponent @xmath45 in eq .
( [ alt ] ) and the size dependence reflects a correction of the form @xcite @xmath46 .
with data for @xmath47 our independent estimate of the exponent is @xmath48 for even @xmath49 and @xmath50 for odd @xmath49 . fixing @xmath51 , as was also done in ref .
@xcite , fits for both even and odd sizes are good and mutually consistent for @xmath52 .
the extrapolated order parameter is then @xmath53 , which is roughly consistent with the previous estimate , @xmath54 , from large systems at low but non - zero temperature @xcite .
turning to the sa simulations , we measure the time @xmath55 in the standard way in units of mc sweeps , where each sweep consists of @xmath56 attempted metropolis flips of randomly selected spins . starting in equilibrium at @xmath57 , we anneal to @xmath0 in @xmath58 mc sweeps according to the following generic power - law protocol : @xmath59 we define the velocity for @xmath60 as @xmath61 and use this definition of @xmath62 as the inverse of the total annealing time also for @xmath63 and @xmath64 .
we collect expectation values at the final temperature @xmath0 , using millions of samples for smaller sizes and several thousand for the largest systems .
there is significant self - averaging and the error bars are small even for the largest sizes ( much smaller than the plot symbols in the graphs below ) . in standard kz scaling , for a process stopping at the critical point ,
a singular quantity @xmath19 depends on the velocity and the system size as @xmath65 , where @xmath66 is the velocity separating fast and slow processes , @xmath67 is the dynamic exponent , and @xmath68 when @xmath69 .
variants of this ansatz have been confirmed in uniform systems @xcite as well as in the 3d ising spin glass ( where @xmath70 ) @xcite .
it has proved to be a reliable way to extract the dynamic exponent , especially if @xmath71 and @xmath22 are known .
if @xmath22 is not known , it can be obtained along with @xmath67 by combining results for different @xmath72 values in the protocol @xcite , and in principle @xmath71 can be determined by using an extended scaling ansatz @xcite .
the dynamic exponent relates the relaxation time scale @xmath73 to the correlation length ; @xmath74 . for given velocity , in the thermodynamic limit the correlation length saturates at @xmath75 and @xmath76 , i.e. , @xmath77 demarks the `` freezing '' of the system .
however , in the @xmath29 model the entropy scale @xmath78 present in the equilibrium should translate to velocity scaling as well . expressed as a length scale , the entropic scale is , @xmath79 , where presumably @xmath80 @xcite as in the model with normal - distributed couplings ( though this has been difficult to confirm @xcite ) . in any case ,
@xmath81 and @xmath82 . in analogy with the equilibrium scaling @xcite , saturation in sa
should then require @xmath83 and we arrive at a modified kz scaling form : @xmath84 here the renormalized dynamic exponent @xmath85 directly gives the critical velocity @xmath86 , which is the original kz ansatz with @xmath22 replaced by @xmath87 .
results for the ea order parameter in linear sa runs are shown in fig .
[ fig2 ] . for fixed velocity ,
the squared order parameter @xmath44 drops rapidly with increasing system size . in eq .
( [ avt ] ) we have @xmath45 as in the equilibrium , but since the correction to the asymptotic value of @xmath44 is large we first divide it out based on the fitted form in fig . [ fig1](b ) . as shown in fig .
[ fig2](b ) , we then rescale the @xmath88-axis by the size - dependent kz velocity , optimizing the value of the exponent @xmath89 for the best data collapse for large systems and low velocities ( see sm @xcite for technical details and alternative scaling procedures ) .
the linear behavior on the log - log plot corresponds to velocities such that @xmath90 . in this regime
@xmath91 , where @xmath92 is the dimensionality .
consistency with the @xmath49 dependence in eq .
( [ avt ] ) then requires that @xmath93 . a line with the corresponding slope @xmath94 ( here @xmath60 ) in fig .
[ fig2](b ) shows perfect consistency with the data .
note that the data collapse procedure does not only involve this linear regime but also lower velocities where the behavior is turning toward a constant as equilibrium is approached .
we do not go very far into this regime because of the exceedingly long running times for large systems . for smaller @xmath49
we reach close to equilibrium and in sm @xcite we show that the scaling there delivers a compatible exponent @xmath95 , with small size corrections . for @xmath60 ,
the exponent settles to @xmath96 as small systems are gradually excluded . at high velocity , the results in fig .
[ fig2](b ) deviate from the power law due to a cross - over @xcite , at an essentially size - independent velocity , reflecting the equilibrium behavior at the initial temperature .
-1 mm -1 mm repeating these procedures for @xmath97 , we can extract the exponents @xmath85 and @xmath98 . as shown in fig .
[ fig2](c ) , the @xmath72 dependence agrees with the form of the kz exponent , and we obtain @xmath99 and @xmath100 . fixing @xmath51 does not significantly alter the value of @xmath85 . -2 mm -2 mm to study the singular part of the energy , we subtract the infinite - size value @xmath101 from the velocity dependent energy @xmath102 and use @xmath103 in eq .
( [ avt ] ) .
we again optimize the data collapse with small systems and high velocities excluded .
fig .
[ fig3 ] shows @xmath60 results and similar @xmath97 plots are presented in sm @xcite .
combining the results for @xmath104 for the different @xmath72 values , we can again , as in fig .
[ fig2](c ) , disentangle the exponents .
interestingly , here we obtain a clearly different dynamic exponent , @xmath105 , than @xmath99 governing the ea order parameter , while @xmath106 is consistent with the previous value .
fixing @xmath51 we can reduce the error bar on the dynamic exponent ; @xmath107 .
using small systems close to equilibrium gives a similar value @xcite .
we also studied the lowest energy @xmath108 reached in each sa run .
naturally , @xmath109 is somewhat lower than @xmath110 for each @xmath111 , but the scaling exponent @xmath104 is still the same .
the scaling function @xmath112 is different for the two quantities @xcite .
the existence of two different dynamic exponents contradicts the standard picture of relaxation , where the slowest mode is associated with the fluctuation of the order parameter .
the coupling of the energy to the order parameter ( via defects ) normally implies that the asymptotic energy relaxation is also determined by @xmath85 .
thus , in the standard scenario , there is a single exponent governing the dynamic scaling of all quantities , except ones that are explicitly constructed to only sense _ faster _ modes .
the dual dynamic scales in the @xmath29 model should be related to the phenomenon of droplet entropy stabilizing the ea order parameter when @xmath7 in equilibrium .
the backbone of the spin - glass cluster has a fractal dimension @xmath113 @xcite and , thus , does not have long - range order on its own @xcite .
the ground states are strongly clustered within a small region ( and its spin - reflected counterpart ) , which implies that these states are related to each other by flipping small ( compared to the system size ) droplets ; flipping large droplet throws the system into atypical regions that are statistically inisgnificant in the thermodynamic limit .
although the absence of order at @xmath8 implies that low - energy excitations must be spread out over a large region of the configuration space , the ground state region should also have a much higher density of low - energy states than other regions ( since these states can be obtained from the ground states by flipping small clusters ) .
thus , we have a region of typical ground states and low - energy states , illustrated in fig . [ fig4 ] , and the large entropy drives the system toward this region under annealing . in the typical region , the ea order parameter , eq .
( [ ea ] ) , has essentially the same distribution for replicas in low - energy states as for those strictly in ground states , and , therefore , the order parameter can converge even when a significant fraction of the replicas remain in excited states .
thus , our scaling results show that the final relaxation of the system involves transitions between excited states into ground states located in the same high - density region , and that the time scale for this is significantly longer ( approximately by a factor @xmath114 ) than that for reaching the high - density region .
-2 mm -2 mm dynamic scaling is also interesting in the context of optimization .
it has recently been argued that the best measure of optimization is not necessarily just the energy ( the standard cost function ) , but the stability of the solution is also important and should be enhanced if the solution belongs to a dense region of similar soloutions @xcite .
a method was presented to enhance the ability to reach such regions , by using coupled replicas of the system .
the 2d @xmath29 ising spin glass is an extreme case of a system harboring a dense region of low - energy states , and we have shown here that sa finds this region efficiently even without artificial replicating , as evidenced by the entropy - driven order parameter converging faster than the energy . in optimization ,
one may be willing to accept a slightly sub - optimal solution , as measured by the energy , for a solution in a dense region that can be found on a much shorter time scale .
clustering of solutions is also important when discussing the efficiency of qa protocols , where the measure of success is also ambiguous and solution stability may be a desirable feature .
qa of systems with discrete coupling distributions may also be affected by dual time scales , due to mechanisms similar to those discussed here .
it would clearly be interesting to also study the kz dynamics of the model with normal - distributed couplings , which has a unique ground state and likely different dynamic scaling .
kz scaling of @xmath115 sa simulations can also be used in other systems that do not order at @xmath8 .
stimulated by the present work , the procedures were already applied to a planar vertex model encoding a class of reversible computing problems @xcite . _
acknowledgments_.we thank anatoli polkovnikov and claudio chamon for discussions .
the research was supported by the nsf under grant no .
dmr-1410126 .
computations were carried out on boston university s shared computing cluster .
99 s. kirkpatrick , c. d. gelatt , jr . , and m. p. vecchi , science * 220 * , 671 ( 1983 ) .
t. w. b. kibble , j. phys .
a * 9 * , 1387 ( 1976 ) .
w. h. zurek , nature ( london ) * 317 * , 505 ( 1985 ) .
f. zhong and z. xu , phys .
rev .
b * 71 * , 132402 ( 2005 ) .
a. chandran , a. erez , s. s. gubser , and s. l. sondhi , phys .
rev .
b * 86 * , 064304 ( 2012 ) . c .- w .
liu , a. polkovnikov , and a. w. sandvik , phys .
rev .
b * 89 * , 054307 ( 2014 ) . c .- w .
liu , a. polkovnikov , a. w. sandvik , and a. p. young , phys . rev .
b * 92 * , 022128 ( 2015 ) .
a. polkovnikov , phys .
rev .
b * 72 * , 161201(r ) ( 2005 ) .
w. h. zurek , u. dorner , and p. zoller , phys .
rev .
lett .
* 95 * , 105701 ( 2005 ) .
j. dziarmaga , phys .
rev .
lett . * 95 * , 245701 ( 2005 ) .
g. lamporesi , s. donadello , s. serafini , f. dalfovo , and g. ferrari , nature phys .
* 9 * , 656 ( 2013 ) .
l. w. clark , l. feng , c. chin , arxiv:1605:01023 .
t. kadowaki and h. nishimori , phys .
rev .
e * 58 * , 5355 ( 1998 ) .
e. farhi , j. goldstone , s. gutmann , j. lapan , a. ludgren , and d. preda , science * 292 * , 472 ( 2001 ) .
c .- w .
liu , a. polkovnikov , and a. w. sandvik , phys .
rev .
lett . * 114 * , 147203 ( 2015 ) .
m. w. johnson _
et al . _ , nature ( london ) * 473 * , 194 ( 2011 ) .
s. boixo , t. f. rnnow , s. v. isakov , z. wang , d. wecker , d. a. lidar , j. m. martinis , and m. troyer , nature . phys . * 10 * , 218 ( 2014 ) .
h. g. katzgraber , f. hamze , and r. s. andrist , phys .
rev .
x * 4 * , 021008 ( 2014 ) .
b. heim , t. f. rnnow , s. v. isakov , and m. troyer , science * 348 * , 215 ( 2015 ) .
h. g. katzgraber , f. hamze , z. zhu , a. j. ochoa , and h. munoz - bauza , phys .
rev .
x * 5 * , 031026 ( 2015 ) .
d. s. fisher and d. a. huse , phys .
rev .
b 38 , 386 ( 1988 ) . c. k. thomas , d. a. huse , and a. a. middleton , phys .
rev .
lett . * 107 * , 047203 ( 2011 ) .
j. wang and r. h. swendsen , phys .
rev .
b * 38 * , 4840 ( 1988 ) . i. a. campbell , a. k. hartmann , and h. g. katzgraber , phys . rev .
b * 70 * , 054429 ( 2004 ) .
n. kawashima , n. hatano , and m. suzuki , j. phys . a : math
. gen . * 25 * 4985 ( 1992 ) .
h. rieger , l. santen , u. blasum , m. diehl , m. jnger , and g. rinaldi , j. phys .
a : math .
gen .
* 29 * , 3939 ( 1996 ) .
h. g. katzgraber , l. w. lee , and i. a. campbell , phys .
rev .
b * 75 * , 014412 ( 2007 ) . i. morgenstern and k. binder , phys . rev .
b * 22 * , 288 ( 1980 ) .
r. n. bhatt and a. p.
young , phys . rev .
b 37 , 5606 ( 1988 ) .
j. poulter and j. a. blackman , phys .
rev .
b * 72 * , 104422 ( 2005 ) .
t. jrg , j. lukic , e. marinari , and o. c. martin , phys .
rev .
lett . *
96 * , 237205 ( 2006 ) .
f. roma , s. risau - gusman , a. j. ramirez - pastor , f. nieto , and e. e. vogel , phys .
rev .
b * 82 * , 214401 ( 2010 ) .
see supplemental material following the main paper .
k. hukushima , k. nemoto , j. phys .
. jpn . * 65 * , 1604 ( 1996 ) .
r. g. palmer and j. adler , int .
j. mod .
phys .
c * 10 * , 667 ( 1999 ) .
a. w. sandvik , europhys . lett .
* 45 * , 745 ( 1999 ) .
m. n. barber , in _ phase transitions and critical phenomena _ , edited by c. domb and j. lebowitz , vol .
8 ( academic , london , 1983 ) .
l. saul and m. kardar , phys .
rev .
e * 48 * , r3221 ( 1993 ) .
h. g. katzgraber , l. w. lee , and a. p.
young , phys .
rev .
b * 70 * , 014417 ( 2004 )
. a. k. hartmann , phys .
rev .
b * 77 * , 144418 ( 2008 ) . c. baldassi , a. ingrosso , c. lucibello , l. saglietti , and r. zecchina , phys
rev .
lett . * 115 * , 128101 ( 2015 ) . c. baldassi , c. borgs , j. chayes , a. ingrosso , c. lucibello , l. saglietti , and r. zecchina , arxiv:1605:06444 . c. chamon , e. r. mucciolo , a. e. ruckenstein , z .- c .
yang , arxiv:1604.05354 . |
galaxies reside in three different environments : clusters , groups and voids . the majority ,
more than 50 per cent , of galaxies are found in groups and clusters @xcite .
this is particularly true for elliptical galaxies , as the morphology - density relation has shown @xcite .
the merger hypothesis @xcite suggests that the product of the merger of two spiral galaxies will be an elliptical galaxy .
if this holds , then the probability to find an elliptical galaxy grows in environments with high densities and low velocity dispersions , e.g. groups of galaxies .
therefore , local environment is thought to play a crucial role in galaxy formation and evolution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in observations
it is relatively easy to study cluster and group galaxies . because of this and the potential formation mechanism of elliptical galaxies , it is not surprising that most studied ellipticals are found in clusters and groups .
therefore , the detailed properties of isolated field elliptical ( ife ) galaxies have not been extensively studied or very well understood ; there are only a few observational studies and the surveys are small .
moreover , the formation mechanisms and evolutionary paths of these `` lonely '' elliptical galaxies are not yet well understood . in the past two decades several observational projects
have identified and studied the properties of isolated field elliptical galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
in many studies , based on observational data , different possible formation scenarios have been proposed ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , ranging from clumpy collapse at an early epoch to multiple merging events .
also equal - mass mergers of two massive galaxies or collapsed groups have been suggested .
theoretical studies predict that isolated ellipticals are formed in relatively recent mergers of spiral galaxy pairs , while large isolated ellipticals may be the result of merging of a small group of galaxies ( e.g. * ? ? ?
* ; * ? ? ?
observational studies have shown that several ifes reveal a number of features such as tidal tails , dust , shells , discy and boxy isophotes and rapidly rotating discs ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) indicating recent merger and/or accretion events .
@xcite concluded that at least @xmath13 per cent of their isolated elliptical galaxies show some kind of mophological distortion , and suggested that these galaxies suffered late dry mergers .
@xcite compiled a sample of @xmath14 candidates of isolated early - type galaxies and studied their properties , and concluded that a major merger of two massive galaxies could explain most observed features .
they also concluded that a collapsed poor group of a few galaxies is a possible formation scenario .
however , @xcite studied a similar sample of isolated early - type galaxies , and concluded that isolated systems are underluminous by at least a magnitude compared with objects identified as merged group remnants .
@xcite concluded that mergers at different redshifts of progenitors of different mass ratios and gas fractions are needed to reproduce the observed properties of ifes .
however , some ifes have not shown any evidence of recent merging activity ( e.g. * ? ? ?
* ; * ? ? ?
@xcite , who studied a sample of nine isolated elliptical galaxies , identified five galaxies that were located in environments similar to those of loose groups , while the environments of the remaining four galaxies were confirmed to be on low density .
all galaxies showed smooth , azimuthally symmetric profiles , with no obvious indications of dust lanes , nascent spiral structure or star - forming regions .
it is possible that the merging events have happened in distant past , and all signs of merging events have been wiped out .
@xcite found out , by using a combination of numerical simulation and synthesized hubble space telescope ( hst ) wide - field planetary camera 2 ( wfpc2 ) images , that merger remnants appear morphologically indistinguishable from a `` typical '' elliptical @xmath15 gyrs after the galaxies merged , while @xcite estimated from numerical simulations that the time might be even less than @xmath16 gyrs .
indeed , these times are very short in time scales of galaxy evolution , making it difficult to find observational evidence to back up formation via merging events . on the other hand ,
it is equally possible that some or even all of these galaxies have initially formed in underdense regions and developed quietly without any major mergers .
observational studies of isolated field elliptical galaxies use similar isolation criteria as optical studies of fossil groups .
both classes show a large magnitude gap between the first- and the second - ranked galaxy
. however , there is no criterion for ifes that would require a presence of extended x - ray emission , as in case of fossil groups .
additionally , fossil groups are not necessarily found in low - density environments @xcite like ifes . despite the differences , it is possible that the formation mechanisms of the two systems are similar or related to each other .
therefore , it is interesting to compare isolated field elliptical galaxies and fossil groups , and see if these systems share a common origin . in this paper
, we use the millennium simulation @xcite together with a semi - analytical model @xcite of galaxy formation within dark matter haloes to identify isolated field elliptical galaxies , to study their properties and formation history , and to compare them with observational data .
this paper is organised as follows . in section [ sample ] , we discuss the millennium simulation , the semi - analytical galaxy catalogue used , and sample selection . in section [ results ] , we compare the properties of the millennium ifes with observations , while in section [ resultsa ] we show that the ifes form a population that is different from the regular ellipticals . in section [ formation ]
we concentrate on evolution of ifes and possible formation mechanisms , while in section [ s : discussion ] we discuss our results , the formation of field elliptical galaxies and their possible connection to fossil groups .
finally , in section [ summary ] we summarise our results and draw the conclusions . throughout this paper
, we adopt a parametrized hubble constant : @xmath17 km s@xmath4 mpc@xmath4 .
we use a simulation that covers a big enough spatial volume and has a sufficient mass resolution the dark matter
only millennium simulation ( ms ; * ? ? ?
* ) , a @xmath18-particle model of a co - moving cube of size @xmath19 mpc , on top of which a publicly available semi - analytical galaxy model @xcite has been constructed .
the cosmological parameters used in the ms were : @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 where the hubble constant is characterized as @xmath26 km s@xmath4 mpc@xmath4 ( for detailed description of the ms , see * ? ? ? * ) .
these values were inferred from the first - year wmap ( wilkinson microwave anisotropy probe ) observations @xcite .
the galaxy and dark matter halo formation modeling of the ms data is based on merger trees , built from @xmath27 individual snapshots . from these time - slices ,
merger trees are built by combining the information of all dark matter haloes found at any given output time .
this enables us to trace the formation history and growth of haloes and subhaloes through time .
properties of galaxies in the ms data are obtained by using semi - analytic galaxy formation models ( sams ) , where star formation and its regulation by feedback processes is parametrized in terms of analytical physical models .
the semi - analytical galaxy catalogue we use contains about nine million galaxies at @xmath28 down to a limiting absolute magnitude of @xmath29 .
a detailed description of the creation of the ms galaxy catalogue , used in this study , can be found in @xcite .
semi - analytical galaxy formation models are known to be less than perfect .
there are several free parameters in each sam that all have a different impact on the properties of galaxies .
the free parameters usually control the feedback effects and cooling of hot gas .
the sam also controls how gas is stripped from a galaxy when it approaches another galaxy , and possible starburst events .
the physics of these processes described is not well known .
however , in general , sams can reproduce the observed galaxy luminosity function and other statistics well . as our study is statistical by nature
, it is likely that the sam used for the millennium simulation galaxy catalogue is accurate enough to predict the properties of ifes and their evolution and formation history .
the ms galaxy catalogue does not predict galaxy morphologies . to assign a morphology for every galaxy , we use their bulge - to - disk ratios .
@xcite found a correlation between the @xmath1band bulge - to - disc ratio and the hubble type @xmath30 of galaxies from observations .
the mean relation may be written : @xmath31 where @xmath32 is the difference between the bulge magnitude and the total magnitude and @xmath33 .
we classify galaxies with @xmath34 as ellipticals , those with @xmath35 as s0s , and those with @xmath36 as spirals and irregulars .
galaxies without any bulge are classified as type @xmath37 .
there are a few different selection criteria that are being used to define isolated field elliptical galaxies in observations ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
despite the differences in selection criteria , all studies identifying ifes share a common ideology . in general , ifes are defined as elliptical galaxies that do not have nearby optically bright companion galaxies , usually in the @xmath1band .
this is tested by using an isolation sphere or a cone and choosing a minimum magnitude difference between the brightest and the second brightest galaxy @xmath38 inside the sphere or the cone . in observational studies , the isolation cone
is often taken as a circle in the sky that expands in redshift space . due to peculiar velocities ,
accurate line - of - sight distances of galaxies are unknown complicating the identification .
thus , observational studies differ in the numerical values of the isolation criteria . despite these differences , discussed next
, we use all possible available observational data for comparison as the number of observed targets is small . for completeness , we show in section [ s : abundance ] how the number of ifes depends on the selection criteria .
@xcite adopted rather strict rules for isolation , as they required that ifes can not have catalogued galaxies with known redshifts within a projected radius of @xmath39mpc and a velocity of @xmath40 km s@xmath4 , while for the morphology of the ifes they required hubble types @xmath41 . however , as their source of galaxies was the third reference catalogue of bright galaxies ( rc3 , * ? ? ?
* ) , their sample ifes might have faint companion galaxies due to the incompleteness of the source catalogue .
missing companions do not affect the study of bright isolated galaxies per se , but their environment is affected by the incompleteness and missing companions .
@xcite based their initial sample of ifes on the same catalogue .
they adopted an even stricter criterion in projected radius , a minimum projected physical distance of @xmath42 mpc to any nearest neighbour brighter than @xmath43 .
@xcite used less strict criteria in their study and required only that an ife candidate has no neighbours with brightness difference @xmath44 mag , in the @xmath1band , within @xmath45 mpc in the plane of sky and @xmath46 km s@xmath4 in recession velocity .
further , their elliptical galaxies are also of hubble type @xmath41 . for the selection of ifes from the millennium simulation
we adopt the criteria used by @xcite , similar to those adopted by @xcite.we concentrate on galaxies that are incontrovertibly ellipticals and thus adopt a strict morphology limit @xmath47 .
we further limit the brightness of the ife candidates with the magnitude cut - off of @xmath48 . for the isolation criterion
we adopt two isolation spheres with the radii of @xmath49 and @xmath39mpc . within these isolation spheres , we require that the @xmath1band magnitude difference of the first- and second - ranked galaxies @xmath38 must be @xmath50 and @xmath51 mag , respectively . the application of the criteria is illustrated in ( * ? ? ?
* fig.@xmath52 ) .
our adopted values of magnitude differences correspond to factors of about eight and two in luminosity for the small and large sphere , respectively .
this choice ensures that possible companions are small and light enough not to produce any major perturbations in the gravitational potential of the system .
note , that in case of simulations our isolation criteria operate in real space rather than in redshift space .
therefore , we can use isolation spheres rather than cones and we are not plagued by interlopers due to inaccuracies in distance measurements . to identify isolated field elliptical candidates in the ms
, we chose five independent cubic volumes inside the simulation box .
initially , we chose each volume to have a side length of @xmath53 mpc , while none of the cubes overlap .
however , we later limited the volume from where the possible ife candidates can be identified to have a side length of @xmath54 mpc .
this was done to enable the study of the surroundings of each ife candidate in the same fashion . moreover , without the limitation , we could have accidentally identified a field elliptical candidate @xmath55mpc from an edge of the box , leading to a possible false identification .
the five volumes were chosen to overcome computational issues and for easier study of ife environment .
with five independent volumes it is also possible to make comparisons between the ifes of each volume . at first all galaxies inside each volume
are treated as possible ife candidates . after identifying all candidates with @xmath47 and @xmath56
we apply the isolation criteria discussed in the previous section .
this produces an initial list of @xmath57 galaxies that fulfill the criteria for isolated field elliptical galaxies . after compiling the initial list of ifes , we inspected every galaxy individually , and found that nine candidates were not the main galaxy of the dark matter halo they reside in . therefore , all galaxies that fulfill the observationally motivated ife criteria are not the central galaxies of their friends - of - friends dark matter groups . in a strict sense
these are not isolated galaxies , as they belong to a dark matter halo containing galaxies more massive and more luminous than the ife candidate .
therefore , we omit them from the final list of ifes , which contains @xmath58 galaxies .
we also compile two control samples of elliptical galaxies for comparisons .
the first control sample contains in total @xmath59 elliptical ( @xmath47 ) galaxies brighter than @xmath0 in the @xmath1band at the redshift @xmath28 and it is named as ellipticals ( abridged as es ) . for the second control sample ,
called main ellipticals ( abridged as mes ) , we only select galaxies of the first control sample that are the main galaxies of their dark matter haloes .
this requirement further limits the ellipticals sample , and we are left with @xmath60 galaxies in the main ellipticals sample .
the removed galaxies are interesting from another point of view than regular ifes , as these nine non - standard ifes are subhalo galaxies ( i.e. satellites of a larger galaxy ) and have multiple nearby companions , from @xmath61 to @xmath62 inside the large isolation sphere .
thus , these galaxies are not isolated in galaxy number density , but reside in a cluster rather than in the field .
physically it is obvious that a satellite galaxy is not the dominant galaxy in its dark matter halo . since we find these non - standard ifes , our observationally motivated isolation criteria are not strict enough and the radii of the isolation spheres should be slightly larger to avoid any false identifications . but
as the number of non - standard ifes is small , we can keep the criteria and remove the non - standard galaxies from our final sample , as described above .
the satellite non - standard `` isolated '' elliptical galaxies can shed some light on the observational result @xcite where an ife galaxy ( ngc 1600 ) was also found to be located off from the dynamical centre of the system .
@xcite found that the x - ray emission is centered slightly to the northeast of ngc 1600 , suggesting that the galaxy is not at the centre of the gravitational potential .
the dynamical study and the x - ray observations together suggest that ngc 1600 is surrounded by a massive halo extending out to several hundred kiloparsecs .
a subhalo galaxy , fulfilling the optical ife criteria , could explain the observations of ngc 1600 .
all nine galaxies in our simulations reside inside a large dark matter halo , and in observations would be found to be off from the centre of the potential well , as in the case of ngc 1600 .
however , a massive and more luminous galaxy than ngc 1600 has not been detected at the centre of the potential well in observations , while this is often the case in simulations .
thus , a more straightforward explanation , where the ife orbits around the centre of the potential well and therefore seems to be shifted from the centre , seen in x - ray observations , is more plausible .
this is in agreement with an x - ray study of another isolated elliptical ngc 4555 , which is assumed not to lie in the centre of a massive group - scale dark matter halo @xcite .
below we compare the properties of simulated ifes with those of the observed ifes .
if these are close enough , it will justify the theoretical study of the formation , evolution and merger histories of ifes , using the simulated ife population . due to the large number of simulated ifes we can actually make theoretical predictions for properties , formation mechanisms and evolution of ifes that can be tested against observational data when the sample of observations is large enough . in following sections
we use the kolmogorov@xmath63smirnov ( ks ) two - sample test to study the maximum deviation between the cumulative distributions of two samples .
the null hypothesis of the ks test is that the two samples are from the same population .
we give our results as probabilities ( p - values ) that the difference between the two samples could have arisen by change if they are drawn from the same parent population .
the fraction of isolated field elliptical ( @xmath47 and @xmath56 mag ) galaxies among all galaxies of any brightness , morphology or dark matter halo status in the ms is merely @xmath64 per cent , corresponding to a number density of @xmath65h@xmath66 mpc@xmath3 .
this is a very low number density , however , such a straightforward calculation is not totally justified .
more meaningful number to conside is the fraction of ifes among elliptical galaxies .
moreover , our morphology limit ( @xmath47 ) for a simulated ife is very strict , biasing the fraction of elliptical galaxies to significantly lower fractions than observed in the real universe .
thus , we also identified isolated field elliptical galaxies with relaxed morphology limit @xmath34 and quote the fractions below .
about @xmath67 per cent of all simulated elliptical ( now @xmath34 ) galaxies of any brightness , can be identified as ifes , when the criteria of section [ selection ] are adopted for ifes .
if we relax the strict morphology limit of the ifes and require that ifes also have @xmath34 the fraction rises to @xmath68 per cent as about @xmath42 times more ifes can be identified . relaxing the morphology limit will therefore change the number density of ifes to @xmath69h@xmath66 mpc@xmath3 . when the brightness of the simulated elliptical galaxies and their morphologies are limited to @xmath56 at @xmath70 and @xmath47 , respectively , about @xmath71 per cent of these galaxies are identified as ifes at redshift zero . if we further limit our simulated elliptical galaxies to galaxies that are the main galaxies of their dark matter haloes ( as in case of ifes ) , @xmath72 per cent of ellipticals are now identified as ifes .
thus , ifes are very rare objects among all galaxies , but at the same time , over @xmath61 per cent of main elliptical galaxies brighter than @xmath0 mag in @xmath1band can be classified as ifes .
@xcite found from magnitude limited ( @xmath73 ) observations that less than three per cent of all galaxies can be classified as isolated .
this value should be interpreted as an upper limit for the number of ifes , as it was based on all types of galaxies .
@xcite further found that early - type galaxies outnumber s@xmath74 galaxies @xmath75 in very isolated areas .
however , as spiral galaxies are the most dominant galaxy type in low densities ( e.g. * ? ? ?
* ) , the observable estimate of the number of ifes is significantly lower than one per cent .
@xcite concluded that early - type galaxies amount to only @xmath76 per cent among all isolated galaxies , lowering the observational estimate of ifes to @xmath77 per cent .
this fraction is in good agreement with our findings as about @xmath78 per cent of all simulated galaxies can be identified as ifes when @xmath34 morphology limit has been adopted .
as observational studies adopt different criteria for isolation , we illustrate their influence on the number of ifes , with a stricter set of parameter values . for simplicity , we adopt only one isolation sphere with a radius of @xmath79 mpc , and require that the magnitude difference between the brightest and the second brightest galaxy @xmath38 is @xmath80 mag inside the isolation sphere . for the hubble type , we adopt the same requirement as previously for our ifes : @xmath47 .
such values were used , e.g. , by @xcite . with these values
we find that only @xmath81 per cent of all ms galaxies can be classified as ifes , corresponding to an extremely low number density of @xmath82h@xmath66 mpc@xmath3 .
this is approximately ten times less than before .
it is obvious that changing the isolation criteria has a major impact on the number of ifes and therefore their statistical properties , as stricter parameter values require a galaxy that is located in a true void .
this result also shows that galaxies with comparable luminosities tend to reside in relatively close proximity rather than being truly isolated from any other reasonable sized companion . the results of this section show that the fraction of isolated field elliptical galaxies depends highly on the isolation criteria adopted , but also on the morphological type that is required .
as the morphological type of simulated galaxies is less than precise our number density for ifes should not be taken as a strict limit , but as a guideline when observations are being planned .
thus , higher fractions of ifes can be expected if isolation criteria or morphology limits are being relaxed .
the easiest property of ifes to observe must be the luminosity , and thus the colour - magnitude relation .
this simple yet powerful relation is studied in figs .
[ colourmag ] and [ colourmag2 ] , which show the colour - magnitude diagrams ( cmds ) of simulated and observed ifes .
[ colourmag ] shows a comparison between the simulated ifes and observed field elliptical galaxies presented in @xcite , while fig .
[ colourmag2 ] shows a comparison to the sample of ifes presented in @xcite .
we also plot the ellipticals ( our control sample ) for completeness .
note , however , that in this section we limit the discussion of the colour - magnitude diagrams to the comparison of simulated and observed ifes , while the differences between simulated ifes and es are discussed in section [ comparisona ] .
colours and associated errors and standard deviations of observed and simulated ifes and simulated elliptical galaxies ( es ) are listed in table [ tb : colours ] . fig .
[ colourmag ] shows that the observed ifes lie in slightly bluer region in the cmd than simulated ifes of the same brightness and that they seem to follow a tighter correlation than simulated ifes .
the @xmath83 colour scatter of simulated ifes is slightly larger compared to observed ifes : all simulated ifes show a @xmath84 deviation of @xmath85 , however , as the observations of @xcite are limited to relatively bright ( @xmath86 ) galaxies , we calculate the colour scatter of bright simulated ifes using magnitude cut - off of @xmath86 and find it to be @xmath87 .
the colour scatter of the bright simulated ifes is close to the observed @xmath83 colour scatter ( @xmath88 ) , yet slightly higher .
however , one should bear in mind that the number of observed ifes is still small
. .colours of simulated and observed ifes and simulated non - isolated elliptical galaxies ( es ) . [ cols="<,^,^,<,^",options="header " , ] + fig .
[ f : formationetctimes ] ( top panel ) shows that isolated field elliptical galaxies have assembled at lower redshifts than galaxies of mes .
thus , in general , stars of ifes form later than stars of mes and thus ifes are younger .
this result is natural for hierarchical cold dark matter models . if the conventional theory , that higher density areas collapse earlier , holds , this implies that ifes have formed originally in less dense regions than mes .
[ f : formationetctimes ] ( second panel ) also shows that ifes undergo their major merging events at significantly lower redshifts than the ellipticals of mes .
the difference in redshift is significant ( see table [ tb : times ] ) and implies that a different formation mechanism is behind the evolution of ifes and mes .
[ f : formationetctimes ] ( third panel ) shows that stars that will eventually form an ife galaxy are present already at higher redshifts than for mes , in agreement with observational findings ( e.g. * ? ? ?
these results indicate that ifes can form stars more efficiently than mes ( see also fig .
[ f : massassembly ] ) .
note however , that we find the formation time ( @xmath89 ) to be highly dependent on the mass of the dark matter halo , as noted in @xcite .
if we limit mes to galaxies with dark matter halo mass greater than @xmath90m@xmath91 we find the median formation time to be at very high redshift ( @xmath92 ) .
this shows that galaxies in massive dark matter haloes form the bulk of their stars already at very early cosmic epochs , in agreement with the general hierarchical halo mass growth scenario . ) , identity ( @xmath93 ) , formation ( @xmath89 ) and last merging ( @xmath94 ) times measured in redshifts .
the blue solid lines corresponds to ifes , while the red dashed lines mark the mes.,width=317 ] we define a major merger as an event where the two progenitors both contain at least @xmath95 per cent of the stellar mass of the descendant galaxy .
almost half ( @xmath96 per cent ) of isolated field elliptical galaxies have experienced a major merger at some point of their formation history .
the percentage of major merging events for ifes is higher than for mes , where only one third ( @xmath97 per cent ) of galaxies have experienced a major merging event .
if all non - isolated e - type galaxies , independent of their luminosity or status in their dark matter haloes , are considered , only about four per cent experience a major merger .
the difference is significant and shows that it is possible to form elliptical galaxies , isolated or not , without a major merger via disc instabilities .
let us study the redshifts of the last merging event @xmath94 ( see fig .
[ f : formationetctimes ] bottom panel and tables [ tb : times ] and [ tb : timeksresults ] ) and the total number of merging events .
ifes have their last merging event at lower redshifts than mes .
the median redshift of the last merging event is @xmath98 and @xmath99 for ifes and mes , respectively .
this shows that ifes have merging activity also at very late stages of their evolution in agreement with @xcite .
@xcite quoted an age estimate of @xmath100 gyr for isolated elliptical galaxies since the last merger .
this age estimate from observational data seems to disagree with our median time of the last merging , which is more than twice as high : @xmath101 gyr .
however , we do find ifes that have experienced their last merging event only @xmath102 gyr ago .
thus , the discrepancy can arise from the small number of observed ifes in @xcite .
it is possible that evolution of ifes is suppressed by the low density environment they reside in , thus the major merging events happen later stage in their evolution if at all .
the larger fraction of ifes having major mergers than mes can also be due to the requirement of the magnitude gap not only the region they formed in .
ifes of denser areas with comparable sized galaxies must clean their environment effectively , leading to a larger fraction of major merging events . while studying the redshifts of the last merging events we noticed that @xmath9 ifes did not experience a single merging event during their evolution .
these galaxies have developed in truly isolated areas ; although , maybe , a greater mass resolution in simulations would reveal one or more minor merging events . despite the resolution limitations ,
the result is in agreement with observations where some ifes do not show any signs of merging activity .
typically , simulated ifes experience one to @xmath103 merging events ( above the mass limit of @xmath104m@xmath91 ) during their lifetime , the average being @xmath105 and the median six mergers .
these numbers are significantly higher than for mes , for which we find the average and median of @xmath106 and one merging event , respectively .
this suggests that ifes accrete multiple smaller haloes ( and galaxies ) during their formation , cleaning their neighbourhood from dwarf galaxies effectively .
this leads to a question how effectively ifes accrete mass as a function of time compared to mes .
[ f : massassembly ] shows the dark and stellar matter mass assembly as a function of redshift for both ifes and mes .
the figure shows that ifes start to form around the same epoch as mes , however , ifes accumulate stellar matter much faster . according to our findings
ifes can form stars more efficiently than mes , while mes seem to accrete dark matter slightly faster than ifes .
we confirm that ifes form the bulk of their stars at @xmath107 , as suggested in @xcite .
we also note that at redshift one ifes have formed over half of their stars ( stellar mass ) and have gathered as much as @xmath62 per cent of their final dark matter .
the galaxies of mes have accreted roughly the same fraction of dark matter as ifes at @xmath108 , however , they have formed as little as @xmath61 per cent of their stars compared to the final stellar matter at @xmath70 .
this difference is significant and shows that ifes form their stars quickly and are rather dark - matter poor compared to other elliptical galaxies ( see also fig .
[ f : masscomparison ] for stellar vs. dark matter ) .
it is also noteworthy that ifes continue to accrete dark matter till @xmath28 while mes have gathered @xmath109 per cent of their final dark matter already at @xmath110 .
all this points towards a different formation mechanism for these two galaxy classes and suggests that late merging events can be a significant part of the evolution of an ife . .
lines show the median value at given redshift.,width=317 ] to further illustrate the mass assembly of ifes , we plot the sample of blue , light and faint ifes separately in fig . [
f : binnedmassassembly ] . we have divided the ifes sample into two : light ( @xmath111m@xmath91 ) and heavy ( @xmath112m@xmath91 ) ifes .
fig [ f : binnedmassassembly ] shows that there is a significant difference in mass accretion of light and heavy ifes .
the stellar mass of heavy ifes follows closely the dark matter mass accretion , while light ifes evolve differently .
the heavy ifes have created half of their stellar matter already at @xmath113 , while the light ifes have created barely @xmath103 per cent of their stellar matter , in agreement with the findings of @xcite .
we note that the light ifes create half of their stellar matter by @xmath114 .
the heavy ifes also form stars extremely efficiently compared to the light ifes , as the stellar mass follows dark matter accretion closely .
[ f : binnedmassassembly ] also shows that heavy ifes host older stellar populations than light ifes , as @xmath109 per cent of their stellar mass has been accumulated already by @xmath115 .
thus , more massive ( and luminous ) ifes have old stellar populations , while the lighter ones have formed a signification fraction of their stellar mass relatively recently . when the sample of ifes has been divided by dark matter halo mass .
lines show the median value at a given redshift.,width=317 ] so far we have shown that minor and major merging events take place during the formation of an ife .
thus , in the following subsections , we try to identify general formation mechanisms based on different types of merger trees and merging events .
we study merger trees of all ifes individually and identify three typical formation scenarios .
we illustrate merger histories in merger plots .
the isolated field elliptical itself lies at the bottom of the plot at @xmath28 , and all its progenitors ( more massive than @xmath104m@xmath91 ) are plotted upwards going back in time .
galaxies with stellar mass larger than @xmath116m@xmath91 are shown as symbols , and are colour - coded as a function of their rest - frame @xmath83 colour .
[ f : alone ] shows an example of an isolated field elliptical galaxy that has developed undisturbed , alone , and has not undergone even a single merging event .
we group ifes that form in this fashion to a single group that we call solitude formation class .
even though the fig .
[ f : alone ] does not show a single merging event during the formation history , this may not be the whole story ; the plot does not show mergers smaller than the resolution limit ( @xmath117m@xmath91 ) of the millennium simulation .
therefore , it is possible or even likely that ifes belonging to this class have had one or even several minor merging events during their evolution .
even so , the merging events would have involved very light dark matter clumps , and it is likely that no significant observational evidence would exist . in this context
small merging events can be interpreted as accretion of dark matter making solitude a proper formation mechanism .
we identify six ifes , corresponding to mere two per cent of all ifes , that belong to the solitude class .
this implies that either ifes that develop truly alone in underdense regions are extremely rare or that some of the ifes have been misclassified .
it is possible that some ifes with one or more small mergers could belong to this formation class , especially if the motivation of classification is based on observability of each class .
it is not obvious how massive a merging event is required to find observational evidence of a merger complicating the classification .
the six ifes identified all show unsteady evolution in their colour as a function of redshift .
this is somewhat surprising as one would assume that a passively evolving galaxy should show a steady colour evolution from blue to red while the stellar population ages .
the bulge formation of solitude class is assumed to happen via disc instabilities .
solitude ifes reside inside lighter dark matter haloes than ifes of other class , with typical dark matter halo mass of @xmath118m@xmath91 or even less .
this makes ifes that belong to the solitude class the lightest ifes in the ms .
the formation , when the dark matter mass grows larger than @xmath104m@xmath91 , epoch of solitary ifes is in the redshift range @xmath119 .
however , we find one solitude ife that formed as late as @xmath120 , making it a very late bloomer , showing that , ifes can have formed very recently .
solitude ifes are in agreement with observations of e.g. @xcite and @xcite , who did not find any evidence of merging activity while studying their samples of ifes .
colour and their area scales with the stellar mass .
only progenitors more massive than @xmath104m@xmath91 are shown.,width=317 ] fig .
[ f : equal ] shows an example of an isolated field elliptical that has undergone at least one equal " size merger during its evolution .
ifes that have experienced at least one equal sized merger comprise a formation class named coupling .
our definition for an equal size merging is the following : the lighter of the merged galaxies had at least @xmath121 per cent of the stellar mass of the more massive one .
this is far larger than in our definition of a major merger ( @xmath95 per cent ) that was applied when calculating the identity time @xmath93 .
thus , an equal sized merger guarantees that the merging event has had a great impact not only on the morphology , but also on the evolution and other properties of the ife .
the motivation for this formation class resides in observations ; a major merging event should be visible in observational data .
thus , an equal sized merging should leave distinct marks to the descendant galaxy , which should be observable for at least few gyrs if not longer .
it has been suggested that isolated field ellipticals have formed in relatively recent mergers of spiral galaxy pairs , i.e. in merging of two comparable sized galaxies .
@xcite , @xcite and @xcite have found several isolated galaxies that show signs of recent morphological disturbances , while @xcite ( see also * ? ? ?
* ) have found that @xmath122 per cent of isolated galaxies show kinematically distinct cores ( kdc ) .
kdcs are generally believed to be the result of a major or an equal sized merger @xcite .
thus , it is possible that these observations have already identified galaxies that belong to this formation class .
however , it is also possible that kdcs can form in an early collapse without subsequent mergers @xcite , complicating the identification of the formation mechanism .
we identify in total @xmath123 ifes , corresponding to @xmath72 per cent of all ifes , that belong to the coupling class of formation scenarios .
ifes that belong to the coupling formation class show colour evolution that is in agreement with conventional theory .
the time from the last equal sized merging correlates well with the colour of the galaxy at @xmath28 ; galaxies with late merging are bluer than galaxies that experienced their equal sized merging a long time ago .
the mean redshift of the last equal sized merger is @xmath124 while the median is @xmath125 .
ifes that belong to this formation class start to form usually in the redshift range @xmath119 .
we also identify a few ifes that have started to form as early as @xmath126 , making ifes that belong to the coupling class older , in a statistical sense , than ifes that belong to the solitude formation class .
the coupling formation mechanism can explain several observed ifes .
the colour - magnitude diagrams of @xcite show a slope and scatter that is in agreement with equal - mass mergers .
they find @xmath127 per cent of their ifes to show boxy isophotes that can form when equal sized galaxies merge .
@xcite argue that all except one of their ife have luminosities that would , at most , be consistent with a single equal - mass merger event . note that all these observations are best explained by the ifes of the coupling class .
colour and their area scales with the stellar mass .
only progenitors more massive than @xmath104m@xmath91 are shown.,width=317 ] fig .
[ f : cannibal ] shows a typical isolated field elliptical galaxy that has developed and accreted dark matter and stellar mass through multiple , small and larger merging events , but has not experienced any equal sized mergers .
all ifes that form via multiple mergers form a formation class called cannibalism .
the merger trees of cannibal ifes show a large number of mergers , with also a relatively large ones .
it is obvious from the merger trees that the colour of the ife does not change due to a minor merger .
this is due to the fact that many small merging events are likely to be dry , and therefore , do not induce significant star formation that would make the global colour of the ife bluer .
however , the largest merging events can have a significant impact on the ife s morphology and colour ( see fig .
[ f : cannibal ] and the merging event around 0.8 gyr ago ) .
some merging events , visible in the merger trees of cannibal ifes , can be classified as major mergers , and these could be visible in observational data .
most ( @xmath128 ) of the ifes in the ms belong to the cannibal class , corresponding to @xmath129 per cent of all ifes .
the large number of ifes belonging to this class shows that merging events are important for the formation and evolution of isolated field elliptical galaxies .
cannibals often show a large number of merging events , more than @xmath95 , while the number of major merging events ranges from one to three . the possibility of major mergers can complicate the identification of ifes of this class , especially as we can not identify any preferred time for the last major merger . in general , ifes of the cannibal class form early , in the redshift range @xmath130 .
however , we identify few cannibal ifes that form as late as @xmath131 .
the cannibal ifes that form late could also be classified as solitudes , especially as these ifes do not , in general , show a single major merging event only few minor ones .
this formation class can explain a few observed ifes .
@xcite found low - luminosity dwarfs close to an isolated galaxy that avoid accretion .
three of their ifes show high central space density , but they also found quite strong morphological disturbances requiring an accretion of a fairly large galaxy .
this is in agreement with cannibal ifes that have had a major merging event that would explain the morphological disturbances , while a large number of smaller satellite dwarfs could have survived due to the large dark matter halo and long dynamical friction times .
colour and their area scales with the stellar mass .
only progenitors more massive than @xmath104m@xmath91 are shown.,width=317 ]
figs . [ colourmag ] and [ colourmag2 ]
show a separate population of blue and faint ifes .
this population was also confirmed to comprise isolated field elliptical galaxies that reside in light , @xmath132 , dark matter haloes .
moreover , the ife samples of @xcite and @xcite do not reveal a single observed ife that would belong to this predicited population of ifes .
@xcite studied isolated galaxies and used modern observations of sloan digital sky survey ( sdss ) data release ( dr ) 6 .
the base sample of their galaxies were the catalogue of isolated galaxies ( cig ) in the northern hemisphere combiled by @xcite . although the isolation criteria in the study of @xcite and in our study are not the same , it is still interesting to compare if any of the elliptical galaxies could belong to the population of faint and blue ifes that our results predict .
@xcite noted that four early - type galaxies of the cig sample showed blue ( @xmath133 ) colours .
three of their blue galaxies were found to be disky while the other one was found to be boxy . from the morphological inspections they concluded that three of the blue galaxies were ellipticals while one was s@xmath74-type .
thus , below we will inspect if any of the three blue elliptical galaxies noted in @xcite , could belong to the faint and blue population of ifes that simulations predict .
as the magnitudes of @xcite are in the sdss system , we need a method to transform their magnitudes for a comparison . for a crude comparison we can use galaxy colours of @xcite and derive a transformations : @xmath134 for @xmath135band magnitudes for a typical elliptical .
because of our ifes show bluer colours than typical ellipticals , this transformation is not very accurate , however , for our purposes it should be adequate . to transform the sdss colours
, we use the following equation : @xmath136 which has been derived from the works of @xcite and @xcite .
again , we note that this may not provide exact colour transformation , but should provide satisfactory conversion for our crude comparisons .
the three isolated ellipticals of @xcite show @xmath137 colours of @xmath138 , @xmath139 and @xmath140 . with the equation [ eq : colour ] we can approximate that these colours correspond to @xmath83 colours of @xmath141 , @xmath142 , and @xmath143 .
clearly at least one of their galaxy is blue enough ( @xmath144 ) to belong to the separate population of faint and blue ifes . given the inaccuracy of equation [ eq : colour ] it is possible that all three of their galaxies would belong to the blue population of ifes .
the blue galaxies of @xcite have absolute @xmath145band magnitudes @xmath146 , @xmath147 , and @xmath148 , respectively . with the help of equation [ eq : r - mag ]
we can approximate their @xmath135band magnitudes to be @xmath149 , @xmath150 , and @xmath151 , respectively .
thus , at least two of their galaxies are faint enough ( @xmath152 ) to be part of the population . above magnitudes and colours
show that possibly at least one and up to three isolated field elliptical galaxies that belong to the predicited population of faint and blue ifes has already been observed .
one of the isolated ellipticals of @xcite is significantly fainter than any of our ifes , and would not qualify as an ife in our study .
moreover , as the isolation criteria of cig galaxies are different that ours , we can not conclude without a doubt that any of the blue and faint population ifes have been observed .
thus , more observations of blue and faint isolated elliptical galaxies are required to confirm our prediction .
section [ s : mergerhistories ] introduced three typical formation classes of isolated field elliptical galaxies ; solitude , coupling , and cannibalism .
our results show that the majority of ifes experience numerous merging events during their evolution and belong to the cannibalism class .
a smaller , yet a significant fraction of ifes belong to the coupling class , whose members have experienced an equal sized merging , while only a small fraction of ifes show insignificant merging activity and belong to the solitude class .
if merging events are so important for the evolution of ifes , an obvious question remains : is it possible to identify the formation scenario based on observational data ?
therefore , in this section we briefly discuss the properties of ifes that can be observed and that at the same time would readily indicate the formation mechanism of the ife .
ifes that belong to the coupling class form later than cannibal ifes and are found to reside in lighter dark matter haloes than cannibal ifes : the median masses are @xmath153 and @xmath154m@xmath91 , respectively .
the coupling class ifes can therefore be merger remnants of nearby galaxy pairs having a modest sized dark matter halo at redshift zero .
unfortunately , dark matter haloes can not be observed directly .
ifes of equal sized mergings can show a weak x - ray emission ; however , it is not clear how well this could be detected if at all .
moreover , differentiating coupled ifes from cannibals might be difficult from x - ray data only .
the mean number of companion galaxies of ifes that have formed through coupling is lower than for the cannibal ifes , but larger than for the solitary ifes .
unfortunately , the distribution of stellar mass , @xmath1band magnitude and colour of ifes of the coupling class is indistinguishable from the cannibal ifes .
the redshifts of equal sized merging events suggest that to be able to identify galaxies belonging to the coupling class , observations should concentrate on intermediate redshifts in the range @xmath155 .
however , to find evidence of equal sized merging may not be simple ; it is not clear how different are the traces an equal size merger leaves , compared to a major merger .
study of stellar populations could provide a way to identify ifes that have experienced an equal sized merging , but this may not be applicable for high redshifts . a cannibal ife can be a merger remnant of a small or compact group .
we find cannibal ifes to reside in more massive and larger dark matter haloes than other types of ifes .
however , their dark matter haloes are less massive than haloes of typical groups , leaving only small or compact groups to consider .
multiple merging events of cannibal ifes could also be visible in their stellar populations , especially if merging events were gas rich , so - called wet mergers .
observational evidence shows that it is possible to detect ifes in x - ray observations ( e.g. * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
due to their more massive dark matter haloes , x - ray bright ifes are the best candidates for the cannibal formation class .
have fossil groups formed in a similar way as isolated field elliptical galaxies ? are ifes an intermediate product while they develop to become fossil groups , or vice versa ?
these questions are justified as the selection criteria of ifes are very similar to those used for the identification of fossil groups , especially if only optical data are available .
objects of both classes are required to show a large magnitude gap between the brightest and the second brightest galaxy in optical . in general , fossil groups
are also required to show extended x - ray emission while it is not demanded for ifes . however ,
several ifes have shown some amount of extended x - ray emission even it is not required . due to similar selection criteria objects of both classes might have similar evolution and assembly histories and a similar formation mechanism , being actually the same class seen only at different phases of evolution
. the majority of fossil groups seem to have experienced the last major merging event longer than @xmath156 gyr ago and they have assembled half of their final mass by @xmath157 @xcite .
moreover , @xcite found from simulations that only @xmath158 per cent of fossil groups experience the last major merger less than @xmath159 gyr ago , and at least @xmath121 per cent had the last major merger longer than @xmath9 gyr ago .
these timescales are in modest agreement with our findings .
our results show that ifes have experienced their last major merging , on average , @xmath160 gyr ago , while ifes have half of their final mass assembled by @xmath108 . for fossil groups
this number is @xmath161 @xcite .
@xcite study the evolution of the brightest galaxies of fossil groups .
they identified fossil groups from the millennium simulation and adopted the same galaxy catalogue @xcite as in this study , making it interesting to compare their findings to ours . when comparing the medians of assembly , formation and identity times we find that isolated field elliptical galaxies have assembled their stars earlier than the brightest galaxies of fossil groups
; @xmath162 and @xmath161 , respectively . however , fossil groups form significantly earlier than ifes ; @xmath163 and @xmath164 , respectively .
the median time of the last major merging of the ifes is almost twice ( @xmath165 ) the identity time of the brightest galaxies of fossil groups ( @xmath166 ) .
these discrepancies in formation and evolutionary times cast a serious doubt over the idea of similar formation mechanisms .
@xcite argue that the primary driver for the large magnitude gap is the early infall of massive satellites that is related to the early formation time of fossil groups .
we do not find infall of massive satellites at early time in our sample of simulated isolated field elliptical galaxies .
the early evolution of an ife includes relatively minor mergers , except for some ifes that experience an equal sized merging .
however , the time of the equal sized mergers is usually in a later stage of the development of the ife , not early as argued for fossil groups .
our study shows that the primary driver for the large magnitude gap of ifes is either a merging of a comparable sized galaxy pair or effective mass accretion and sweeping up of surrounding galaxies ( coupling and cannibalism , respectively ) .
@xcite argue that fossil groups can be identified from dark matter only simulations if one selects dark matter haloes more massive than @xmath167m@xmath91 .
they argue that above that limit , all optical fossil groups in the millennium simulation have enough hot gas to show x - ray emission , therefore qualifying as x - ray fossils .
our table [ tb : properties ] shows that the median mass of the dark matter haloes of ifes is only @xmath168m@xmath91 , while even the @xmath156rd quartile is just @xmath169m@xmath91 .
this comparison shows that ifes reside in significantly lighter dark matter haloes than the brightest galaxies of fossil groups .
@xcite argued that fossil groups have formed early and that more than @xmath170 per cent of their mass accumulated as early as @xmath171 gyr ago .
moreover , they suggest that x - ray fossil groups are not a distinct class of objects but rather that they are extreme examples of groups which collapse early and experience little recent growth .
this formation scenario does not apply for ifes , as cannibal ifes can have merging activity till the redshift @xmath172 , and ifes start to form later than fossil groups .
although a large magnitude gap seen in both fossils and ifes could imply a similar formation mechanism , the above comparison does not support this idea . considering the assembly and formation times
it seems unlikely that fossil groups , and especially their brightest galaxies , could share the same formation mechanism as isolated field elliptical galaxies . the large differences in dark matter halo masses of fossil groups and the most massive ifes makes it improbable that fossil groups and ifes are the same class of objects seen at different phases of their evolution .
we therefore conclude that fossil groups and ifes form two distinct classes .
the aim of this paper is twofold : 1 ) to compare simulated field elliptical galaxies with observed ones and to make predictions on their properties , and 2 ) to define the formation mechanism and history of isolated field elliptical galaxies .
we also discuss if isolated field elliptical galaxies are related to fossil groups and if they can share a common formation mechanism .
our results show that it is possible to identify isolated field elliptical galaxies ( ifes ) from cosmological @xmath173body simulations with semi - analytical models of galaxy formation , that have similar properties to observed ifes .
we show that simulated ifes are in good agreement with observations when similar identification criteria are adopted .
the colour - magnitude diagram of simulated ifes agree with observations , and the average colour of simulated and observed ifes are within their error limits when all simulated ifes are considered .
unfortunately , observational data sets are still small complicating more detailed comparison .
an agreement in age and mass estimates of ifes between observations and simulations is found . however , the age comparisons are less robust due to different age definitions .
our results show that isolated field ellipticals are very rare objects ; we find their total number density to be as low as @xmath65h@xmath66 mpc@xmath3 .
our result agrees with observational estimates of number of ifes , which however , are inaccurate at best .
our ifes have a small number of companion galaxies , ranging from only a few dwarf companions to as much as about @xmath95 .
thus , ifes are not completely isolated , although they are found to be located in underdense regions .
our results show that ifes reside in relatively light dark matter haloes . however , at the same time , the baryonic to dark matter ratio is higher in ifes than in es generally .
the stellar mass of ifes grows almost linearly with the dark matter mass with a mass ratio ( dark / stellar ) of @xmath174 , whereas our comparison ellipticals have a lower stellar to dark matter ratio .
therefore , ifes are good candidates for galaxies with low dark matter mass to stellar light ratio .
we also find a flat @xmath1band luminosity function for our simulated ifes . when studying the basic properties of ifes we find that ifes populate different regions in colour - magnitude diagrams than regular elliptical galaxies , which are found mostly in the red sequence .
this is due to the bimodality of the distribution of colours , as ifes populate not only the red sequence , but also the blue cloud . from simulation data we find a separate population of blue and faint ifes . on average ifes
are found to be bluer than our control sample ellipticals , howevever , this result is biased because of the separate population of ifes .
the bluest ifes are found to reside in light dark matter haloes , while red ifes are usually found inside more massive ( @xmath175 ) dark matter haloes .
we note that simulations predict a previously unobserved population of blue , dim and light galaxies that fulfill observational criteria to be classified as isolated field elliptical galaxies .
these galaxies have only a few companions , which are usually located many times further away than the virial radius of their dark matter halo .
these blue , dim and light ifes have formed their stars only lately and have continued to accrete dark matter mass till redshift zero .
this population of ifes is very interesting as it has not been detected yet in observations .
our results also show that ifes start to form around the same epoch as the galaxies of the control sample ( mes ) ; however , ifes seem to accumulate stellar mass much faster .
ifes form stars more efficiently than mes , while mes accrete dark matter slightly faster than ifes .
we confirm that ifes form the bulk of their stars at @xmath107 , as suggested in @xcite . by the redshift one ifes
have formed over half of their stars ( stellar mass ) and have gathered as much as @xmath62 per cent of their final dark matter .
the galaxies of mes sample have accreted roughly the same fraction of dark matter as ifes by @xmath108 .
however , they have formed as little as @xmath61 per cent of their stars compared to the final stellar matter at @xmath28 .
this difference shows that ifes form their stars quickly and are rather dark - matter poor compared to other elliptical galaxies .
moreover , ifes continue to accrete dark matter till @xmath28 while mes have gathered @xmath109 per cent of their final dark matter already by @xmath110 .
we note that more massive ( and luminous ) ifes have older stellar populations , while the lighter ones have formed a signification fraction of their stellar mass relatively recently . while studying the evolution of ifes we note that almost half ( @xmath176 per cent ) of ifes have experienced at least one major merger during their formation history , while only about four per cent ( es ) up to one third ( mes ) of control sample ellipticals experience a major merger .
major merging events happen later in ifes evolution than for control sample galaxies , the average of the latest major merging being @xmath8 for ifes while it is @xmath10 for mes .
we also find ifes that have not experienced a single merger event above the mass resolution limit of the ms during their evolution .
therefore , it is possible to form elliptical galaxies without major mergers . when inspecting the merger trees of simulated ifes , we identify three typical formation scenarios : solitude , coupling , and cannibalism , which can all lead to a formation of an ife .
the scenarios range from a solitary growth ( solitude class ) with quiet mass accretion and star formation to more violent evolution with multiple mergers ( cannibalism class ) .
we also identify a formation scenario where two comparable sized galaxies merge to form an ife ( coupling class ) .
our merger trees show that merging events are important for ifes that form through cannibalism or coupling .
all three formation classes are in agreement with observational findings .
comparison between isolated field elliptical galaxies and fossil groups show that these two classes are distinct .
galaxies of both classes show some similarities , but many properties and evolution times are significantly different .
ifes reside in significantly lighter dark matter haloes and we do not find an infall of massive galaxies at early times for ifes as argued for fossil groups @xcite .
however , we can not exclude that some fossil groups could not share the formation mechanism of the most massive ifes , namely cannibalism .
smn acknowledges the funding by the finnish academy of science and letters and the nordic optical telescope scientific association ( notsa ) .
smn would like to thank drs .
elena donghia and henry ferguson for enlightening and inspiring conversations , dr .
gerard lemson for invaluable help with the millennium simulation database , and ms carolin villforth for multiple inspiring conversations .
es thanks the university of valencia ( vicerrectorado de investigacin ) for a visiting professorship , the support by the estonian science foundation grant no .
8005 and by the estonian ministry for education and science , grant sf0060067s08 .
we thank the anonymous referee for detailed reading of the manuscript and comments that helped us to improve the original manuscript .
the millennium simulation databases used in this paper and the web application providing online access were constructed as part of the activities of the german astrophysical virtual observatory .
g. , de vaucouleurs a. , corwin jr . h. g. , buta r. j. , paturel g. , fouque p. , 1991
, third reference catalogue of bright galaxies .
volume 1 - 3 , xii , 2069 pp .
7 figs .. springer - verlag berlin heidelberg new york jester s. , schneider d. p. , richards g. t. , green r. f. , schmidt m. , hall p. b. , strauss m. a. , berk d. e. v. , stoughton c. , gunn j. e. , brinkmann j. , kent s. m. , smith j. a. , tucker d. l. , yanny b. , 2005 , aj , 130 , 873 d. n. , verde l. , peiris h. v. , komatsu e. , nolta m. r. , bennett c. l. , halpern m. , hinshaw g. , jarosik n. , kogut a. , limon m. , meyer s. s. , page l. , tucker g. s. , weiland j. l. , wollack e. , wright e. l. , 2003 , apjs , 148 , 175 springel v. , white s. d. m. , jenkins a. , frenk c. s. , yoshida n. , gao l. , navarro j. f. , thacker r. , croton d. j. , helly j. , peacock j. a. , cole s. , thomas p. , couchman h. , evrard a. , colberg j. m. , pearce f. , 2005 , nature , 435 , 629 | we study the properties , evolution and formation mechanisms of isolated field elliptical galaxies . we create a mock catalogue of isolated field elliptical galaxies from the millennium simulation galaxy catalogue , and trace their merging histories . the formation , identity and assembly redshifts of simulated isolated and non - isolated elliptical galaxies are studied and compared .
observational and numerical data are used to compare age , mass , and the colour - magnitude relation .
our results , based on simulation data , show that almost seven per cent of all elliptical galaxies brighter than @xmath0 mag in @xmath1band can be classified as isolated field elliptical galaxies .
results also show that isolated elliptical galaxies have a rather flat luminosity function ; a number density of @xmath2 mpc@xmath3 mag@xmath4 , throughout their @xmath1band magnitudes .
isolated field elliptical galaxies show bluer colours than non - isolated elliptical galaxies and they appear younger , in a statistical sense , according to their mass weighted age .
isolated field elliptical galaxies also form and assemble at lower redshifts compared to non - isolated elliptical galaxies . about @xmath5 per cent of isolated field
elliptical galaxies have undergone at least one major merging event in their formation history , while the same fraction is only @xmath6 per cent for non - isolated ellipticals .
almost all ( @xmath7 per cent ) isolated elliptical galaxies show merging activity during their evolution , pointing towards the importance of mergers in the formation of isolated field elliptical galaxies .
the mean time of the last major merging is at @xmath8 or @xmath9 gyrs ago for isolated ellipticals , while non - isolated ellipticals experience their last major merging significantly earlier at @xmath10 or @xmath11 gyrs ago . after inspecting merger trees of simulated isolated field elliptical galaxies ,
we conclude that three different , yet typical formation mechanisms can be identified : solitude , coupling and cannibalism .
our results also predict a previously unobserved population of blue , dim and light galaxies that fulfill observational criteria to be classified as isolated field elliptical galaxies .
this separate population comprises @xmath12 per cent of all ifes .
[ firstpage ] methods : numerical - galaxies : formation - galaxies : evolution - galaxies : ellipticals - galaxies : isolated - large - scale structure of universe . |
the ( 2 + 1)-dimensional physics system presents many interesting surprises , both experimentally and theoretically .
for example , in last three decades , it has been realized that while in three and higher space dimensions the particles must be either bosons or fermions , in two space dimensions the particles can have any fractional spin and can satisfy any fractional statistics which interpolates between the familiar two.@xcite the particles obeying such statistics are called as anyons , whose existence in real physics system have been proved by the fractional quantum hall effect@xcite where anyons act as fractionally charged excitations of the imcompressible quantum fluid . in ( 2 + 1)-dimensional spacetime ,
a new type of gauge theory named chern - simons theory which is different from maxwell theory plays a very significant role in many aspects.@xcite one of the most remarkable properties of the chern - simons action is that it does not depends on the metric tensor , and always has the same form no matter the spacetime is flat or not .
hence , the chern - simons theory is an important example of the topological field theory.@xcite there are both topological and nontopological self - dual vortex solutions in the chern - simons model.@xcite and topology is indispensable even in the situation where only the nontopological vortex solutions are present . on the other hand , for over four decades
there has been a wide interest in the condensed - matter systems with several coexisting bose condensates.@xcite recent research@xcite has revealed that a charged two - condensate ginzburg - landau model can be mapped onto a version of the nonlinear @xmath1 @xmath2 model .
this leads to many efforts to study the topological property behind the two - condensate system.@xcite in light of @xmath0-mapping topological current method,@xcite the present paper investigates the self - dual vortices in the abelian chern - simons model with two complex scalar fields , which is a kind of ( 2 + 1)-dimensional field theory generalization of the two - condensate system .
to be self - contained basically , in sec .
[ secmap ] , we briefly review the @xmath0-mapping topological current method . in sec .
[ secmodel ] , we introduce the model to be considered , and derive a set of self - dual equations . in sec .
[ secvortex ] , we find a nontrivial equation with a topological term for each scalar field , which is the development of our previous work,@xcite and also obtain an equation which relates the two scalar fields by their topological terms . besides , the angular momentum and the magnetic flux of the system are calculated . in sec .
[ secbranch ] , we briefly consider the evolution processes of the vortices , and find that because of the present of the vortex molecule , the detailed evolution processes of the vortices in the double scalar field model is more complicated than the vortex evolution processes in the corresponding single scalar field model , which have been investigated in ref . in detail .
because the @xmath0-mapping topological current method is unfamiliar to most of the readers , we will briefly review it in the following . consider a d - dimensional smooth manifold with metric tensor @xmath3 and local coordinates @xmath4 .
define a map @xmath5 : @xmath6 and introduce the direction unit field of @xmath7 : @xmath8 using @xmath9 , we can construct a topological tensor current : @xmath10 it is not difficult to find that @xmath11 is completely antisymmetric and identically conserved
. defined the jacobian tensor @xmath12 , and the @xmath0-space green function @xmath13 then using @xmath14 and the green function relation @xmath15 we can find @xmath16 the @xmath17-function in eq .
( [ jmn01 ] ) implies that @xmath18 only at the points where @xmath19 .
generally , these points would correspond to the submanifolds where the topological defects are located . denoting the @xmath20-th above - mentioned submanifold by @xmath21 , and defining a corresponding normal submanifold @xmath22 which is spanned by the parameter @xmath23 @xmath24 with the metric tensor @xmath25
, we can get the only intersection point of @xmath21 and @xmath22 denoted by @xmath26 . according to the implicit function theorem,@xcite at the regular points of @xmath7
, there exists the only solution to the equations @xmath19 , and eq .
( [ jmn01 ] ) can be expanded as @xmath27 where the jacobian @xmath28 , @xmath29 is the @xmath17-function on the submanifold @xmath21 , and @xmath30 denotes the winding number of the @xmath20-th topological defect .
some important physical quantities concerning the topology , say magnetic flux , can be calculated from eq .
( [ jmn02 ] ) . by virtue of the implicit function
theorem,@xcite the irregular points of @xmath7 correspond to the branch points of the topological current @xmath11 , and the branch processes of @xmath11 occur at these very points .
these branch processes can describe various evolutions of the topological defects , and the total topological charge of the system will keep unchange during these evolutions since @xmath11 is a conserved current .
we consider the ( 2 + 1)-dimensional chern - simons system described by the lagrangian @xmath31 where @xmath32 @xmath33 , and @xmath34 here , coupling @xmath35 , @xmath36 and @xmath0 are complex scalar fields , @xmath37 denotes the chern - simons gauge field .
the specific form of the potential ( [ u ] ) gives rise to a self - dual system .
the variation of the action with respect to @xmath38 leads to a gauss law : @xmath39 where @xmath40 , @xmath41 $ ] and @xmath42 $ ] . for finite - energy configurations , the magnetic flux @xmath43 and charges @xmath44 and @xmath45 are related by eq .
( [ gau1 ] ) , namely @xmath46 set @xmath47 , the energy function is @xmath48.\ ] ] employing the gauss laws and integrating by parts , one obtains the bound @xmath49 this bound is saturated by the fields satisfying the following self - dual equations : @xmath50
from the potential ( [ u ] ) , it is obvious that there is a symmetric phase as well as an asymmetric or broken phase . in the asymmetric phase , static finite energy solutions require both scalar fields @xmath36 and @xmath0 to be at most a pure phase on the circle at infinity , i.e. the vacuum manifold for both fields @xmath36 and @xmath0 is @xmath51 .
so the first homotopy group @xmath52 provides the topological guarantee for the existence of the vortex solutions . in the symmetric phase , nontopological static soliton solutions are possible , which make the situation more complicated than the self - dual maxwell system . for static case , from eq .
( [ gau1 ] ) we obtain @xmath53 from eq .
( [ sd ] ) , we have @xmath54 and furthermore @xmath55 at this point , expressing @xmath36 as @xmath56 , we can write the second term on the right - hand side of eq .
( [ f120 ] ) as @xmath57 this expression has an exact form of topological current ( [ jmn0 ] ) .
so , using the @xmath0-mapping method , we can rewrite eq .
( [ f120 ] ) as @xmath58 where the jacobian @xmath59 and @xmath60 is the @xmath17-function in @xmath36 space .
the presence of @xmath60 in eq .
( [ f121 ] ) identifies the existence of the vortex solutions in model ( [ l ] ) , and implies that the cores of the vortices locate at the zeros of @xmath36 . to be specific
, we denote the isolated zeros of @xmath36 by @xmath61 and the isolated zeros of @xmath0 by @xmath62 . according to the implicit function theorem,@xcite
when the zeros @xmath63 are the regular points of @xmath36 , i.e. when @xmath64 there exists one and only one continuous vortex solution to the equations @xmath65 .
furthermore , under the condition ( [ d1 ] ) , we can express @xmath60 as@xcite @xmath66 where @xmath67 is the winding number of the @xmath68-th isolated zero of @xmath36 .
if we consider the case with @xmath36 vanishing at infinity in the following way @xmath69 where @xmath70 . then @xmath71 therefore eq . ( [ f121 ] ) can generally be rewritten as @xmath72 even though @xmath36 vanishes at infinity in some kind of specific way .
using eqs .
( [ sd ] ) , ( [ a0 ] ) and ( [ f122 ] ) , we find a nontrivial equation to be satisfied by @xmath36 : @xmath73 the above nontrivial equation includes a topological term on its right - hand side , which is missing in many references . under @xmath74 , @xmath75 , @xmath76 and @xmath77
( the winding number of the @xmath78-th isolated zero of @xmath0 ) , all conclusions and expressions we obtain above still hold for @xmath0 . especially , corresponding to eq .
( [ sde1 ] ) , we have @xmath79 from eqs .
( [ sde1 ] ) and ( [ sde2 ] ) , we find the following equation for @xmath36 and @xmath0 : @xmath80.\ ] ] obviously , the above equation relates @xmath36 and @xmath0 by their topological terms , and displays the topological relation between them .
a characteristic feature of chern - simons vortices is that they carry nonzero angular momentum which is given by @xmath81.\ ] ] furthermore , from eq .
( [ ai ] ) and its @xmath0 counterpart as well as eq .
( [ sd ] ) , eq .
( [ j1 ] ) becomes @xmath82\nonumber\\ % = \mp2e\!\!\!\int\ !
d^2x\epsilon^{ij}x^ia_0 % ( |\psi|^2\epsilon^{jk}\partial_{k}\ln|\psi|+ % |\phi|^2\epsilon^{jk}\partial_{k}\ln|\phi|)\\ % = \pm e\!\!\!\int\ ! d^2xx^ia_0\partial_{i}\eta\nonumber\\ & & = \frac{1}{\kappa}~\!\!\!\int\ !
d^2xx^i(e_1|\psi|^2+e_2|\phi|^2-v^2 ) \partial_{i}(e_1|\psi|^2+e_2|\phi|^2 ) \nonumber\\ & & = \left\{% \begin{array}{ll } -\frac{1}{\kappa}~\!\!\!\int\ !
d^2x(e_1|\psi|^2+e_2|\phi|^2-v^2)^{2 } , & \hbox{asymmetric phase ; } \\
-\frac{1}{\kappa}~\!\!\!\int\ ! d^2x(e_1|\psi|^2+e_2|\phi|^2)(e_1|\psi|^2+e_2|\phi|^2 - 2v^2 ) , & \hbox{symmetric phase .
} \\ \end{array}% \right .
% -\frac{1}{\kappa}\!\!\!\int\ !
% d^2x(e_1|\psi|^2+e_2|\phi|^2-v^2)^{2 } , \hspace{7 mm } asymmetric phase,\nonumber\\ % & & = -\frac{1}{\kappa}\!\!\!\int\ !
% d^2x(e_1|\psi|^2+e_2|\phi|^2)(e_1|\psi|^2+e_2|\phi|^2 - 2v^2 ) , % \hspace{7 mm } symmetric phase.\nonumber\\\end{aligned}\ ] ] this conclusion is exactly the generalization of the angular momentum of single complex scalar chern - simons vortices.@xcite in the following , we concentrate on asymmetric phase and assume @xmath83 where @xmath84 and @xmath85 satisfy @xmath86 . because different choices for the vacuum values are inequivalent , we need consider the case with @xmath87 and the case with one of the vacuum values is zero respectively . in the case with @xmath87 , from eq .
( [ f122 ] ) and its @xmath0 counterpart , the flux is given by @xmath88 which implies that the regular vortex solutions may exist only if @xmath89 is rational and equals to the ratio of the total winding number of zeros of @xmath36 and @xmath0 .
on the other hand , in the case with one of the scalar fields , say @xmath36 , vanishes at infinity in the way of eq .
( [ bc1 ] ) , from eq .
( [ f122 ] ) and its @xmath0 counterpart the flux will be @xmath90 where @xmath91 is for @xmath92 for the nontopological soliton solutions allowed in symmetric phase , the situation is different from the topological soliton solutions in some aspects .
for example , suppose that for @xmath70 and @xmath93 @xmath94 then , as before , we can find the flux is given by @xmath95 where @xmath91 is for @xmath96 and @xmath97 is for @xmath98 .
in ref . , the evolution of the chern - simons vortices in the model with single complex scalar field was discussed in detail . as for the present system
, we can directly generalize the results in ref . to the vortex evolution processes concerning the individual scalar field ,
say @xmath36 .
the key issue here is the evolution processes concerning both @xmath36 and @xmath0 . for complex scalar field @xmath36 ,
define topological current @xmath99 it is easy to see that @xmath100 is identically conserved , i.e. @xmath101 . similar to the case in sec .
[ secvortex ] , we can obtain @xmath102 . according to the implicit function theorem as before ,
when @xmath103 there exists one and only one continuous vortex solution to the equations : @xmath104 which can be expressed as @xmath105 and @xmath60 can be expressed as @xmath106 using eqs .
( [ j ] ) and ( [ del2 ] ) , we can find the topological charge corresponding to the current @xmath100 is the winding number @xmath67 .
since the topological current @xmath100 is identically conserved , the topological charge @xmath67 will remain unchanged in various evolution processes .
obviously , the above discussion also apply to @xmath0 .
therefore , at this level , the topological evolution of the vortices for @xmath36 and @xmath0 is relatively independent . at this point ,
a question rises : is there any other nontrivial topological configuration formed in the system due to the interaction between the fields @xmath36 and @xmath0 ? to answer this question , let us pay attention to coherently coupled two - component bose - einstein condensates studied in ref . .
such a system is not described by a gauge theory , but it has the same complex scalar field structure with the present system .
so we can use the method presented in ref . to get an insight into the topological configuration formed by the internal coherent coupling between @xmath36 and @xmath0 . to be specific , writing @xmath107 and @xmath108
, we can introduce an internal space vector @xmath109 where @xmath110 .
field @xmath36 and @xmath0 tend to their vacuum values at the space infinity .
this boundary condition compactifies space @xmath111 into @xmath112 .
so vector @xmath113 defines a mapping : @xmath114 . since @xmath115
, there exists another kind of topological soliton named skyrmions . in the general nonaxisymmetric case , such a skyrmion will split into a pair of vortex - antivortex , or a vortex molecule , which is formed by a vortex in @xmath36 ( @xmath0 ) and an antivortex in @xmath0 ( @xmath36 ) .
the concrete amounts of the bound vortex molecules and unbound vortices in the system depend on the winding numbers of @xmath36 and @xmath0 as well as the skyrmion numbers of the system . because of the existence of the vortex molecule , the detailed evolution processes of the vortices in model ( [ l ] ) is more complicated than the vortex evolution processes in the corresponding single scalar field model .
for example , the processes in which one vortex in a vortex molecule is replaced by another vortex is absent in the single scalar field case .
we will leave this subject to the future studies .
making use of @xmath0-mapping topological current method , we have discussed the self - dual vortices in the abelian chern - simons model with the chern - simons gauge field coupling to two complex scalar fields .
this model is a kind of ( 2 + 1)-dimensional field theory generalization of the two - condensate system , which receives wide attention in condensed - matter physics.@xcite for each scalar field , we analytically derived an exact nontrivial equation with a topological term which is missing in many references .
each of these equations holds for a very wide range of the boundary conditions . besides
, the equation relating the two scalar fields by their topological terms was also obtained .
we calculated the angular momentum of the chern - simons vortices , and the result is exactly the generalization of the angular momentum of single scalar field chern - simons vortices .
we also calculated the magnetic flux of the system for different boundary conditions , and found this magnetic flux relates the total winding numbers for the two scalar fields in a specific way which depends on the given boundary condition .
furthermore , we briefly discussed the evolution processes of the vortices , and found that because of the present of the vortex molecule , the detailed evolution processes of the vortices in the present model is more complicated than the vortex evolution processes in the corresponding single scalar field model .
it is a pleasure to thank dr .
yong - qiang wang for discussions .
this work was supported by the national natural science foundation of the people s republic of china ( no . 502 - 041016 and no .
10705013 ) and the fundamental research fund for physics and mathematics of lanzhou university ( no .
lzu07002 ) . | making use of @xmath0-mapping topological current method , we discuss the self - dual vortices in the abelian chern - simons model with two complex scalar fields . for each scalar field ,
an exact nontrivial equation with a topological term which is missing in many references is derived analytically .
the general angular momentum is obtained . the magnetic flux which relates the two scalar fields
is calculated .
furthermore , we investigate the vortex evolution processes , and find that because of the present of the vortex molecule , these evolution processes is more complicated than the vortex evolution processes in the corresponding single scalar field model . |
many galaxies , over the whole hubble sequence , show nucleated central regions often referred to as `` nuclear star clusters '' ( nscs ) .
these systems are among the densest star clusters observed , with effective radii of a few parsecs and central luminosities up to @xmath4 @xcite .
the total frequency of nucleation is as large as @xmath5 for galaxies fainter than @xmath6 , while nscs tend to disappear in galaxies brighter than this magnitude .
there is no consensus on how nscs form .
a _ dissipative _ origin bases on the hypothesis of radial gas inflow into the galactic center and requires efficient dissipation mechanisms to work ( e.g. , * ? ? ?
* ) . in this model ,
a nsc consists mostly of stars that formed locally @xcite .
alternatively , nscs could have a _
origin in which massive stellar ( globular - like ) clusters migrate to the center due to dynamical friction and merge to form a dense nucleus growing up to the size of observed nscs @xcite .
observations of nscs in de galaxies suggest that the majority of these systems could be the result of accumulating mass in the form of orbitally decayed globular clusters .
numerical studies have also shown that such a formation model is consistent with the measured sizes and luminosities of nuclei @xcite .
in addition , stellar population synthesis studies of nsc spectra suggest that most of the mass usually resides in stars as old as typical globular clusters ( @xmath7gyr ; * ? ? ? * ; * ? ? ?
* ) . on the other hand
, the observed correlation between colors and luminosities together with the complex star formation histories that often characterize the central region of galaxies , may be difficult to explain on the basis of a dissipationless origin , unless there is some contribution from continuous or recurrent star formation in addition to the ancient globular cluster stars @xcite .
this suggests that dissipation and dissipationless processes are not exclusive , and nscs might indeed originate from a combination of the two processes .
due to their small sizes and their crowded stellar fields , galactic nuclei beyond the local group are typically unresolved , and the only quantities that can be determined are integrated properties such as half - mass radius and total luminosity . a radial density profile and velocity structure can be reliably determined only for the milky way ( mw ) nsc @xcite , which , due to its proximity ( @xmath8kpc ) , can be resolved into individual stars @xcite .
in addition to the mw , ngc 205 also has a spatially - resolved nsc ( figure 2 of * ? ? ? * ) .
the mw nsc has an estimated mass of @xmath9 @xcite and hosts a massive black hole ( mbh ) whose mass , @xmath10 , is uniquely well determined @xcite .
a number of other galaxies also contain both a nsc and a mbh @xcite , that have comparable masses . in models of nscs , the dynamical influence of a mbh
should therefore be considered , at least in bulges brighter than about @xmath11 which are believed to always contain a mbh @xcite .
more recently , we have presented large - scale @xmath12-body simulations of the inspiral and merger of massive clusters in the inner regions of the galaxy @xcite .
we showed that current observational constraints are consistent with the hypothesis that a large fraction of the mw nsc mass is in old stars brought in by infalling globular clusters . in this paper , we expand on this previous work and use simple analytical considerations to investigate the possibility that globular clusters can migrate through dynamical friction in the center of galaxies and form compact stellar nuclei . in particular , we focus on how the presence of mbhs at the center of galaxies can impact the merger hypothesis for the formation of their nsc . in order to highlight the role of mbhs in nsc formation
, we will systematically present our results for both cases of galaxy with and without central mbh .
this paper is organized as follows .
we start in 2 by comparing some of the observed scaling relations for nscs with the same relations predicted in both the merger and the gas model . in 3 and
4 we develop a simple analytical model for studying the orbital decay of globular clusters in galaxies like the mw , and explore the effect that a central mbh in the galaxy has on the properties of the resulting nsc . in 5 we use an approach similar to that of 3 to follow the orbital evolution of globular clusters near massive mbhs in the central regions of bright elliptical galaxies .
in 6 we discuss a variety of astrophysical implications of a dissipationless origin of nscs and conclude in 7 .
the study of nscs has revealed a number of correlations between their masses and several global properties of their host galaxies , such as velocity dispersion and bulge mass .
the existence of such correlations might indicate a direct link among large galactic spacial scales and the much smaller scale of the nuclear environment .
while it was for long believed that nscs and mbhs followed similar scaling relations with their host galaxies @xcite , it is now well established that ncs do not follow any of the scaling relations defined by mbhs @xcite
. observations also reveal that , unlike globular clusters , nsc half - light radii are luminosity dependent , increasing with increasing total mass @xcite .
this relation might contain important information on the processes that shaped the central regions of galaxies and their nscs .
we start here by comparing such observed correlations with predictions from both the dissipative and dissipationless formation models . of the global - to - nucleus relations ,
the most frequently referred to is the tight correlation between nsc mass , @xmath14 , and the host galaxy s velocity dispersion , @xmath15 .
@xcite argued that such a correlation has a slope which is consistent with that of the well - known @xmath16 relation obeyed by the mbh mass , @xmath17 .
more recently , a series of papers reached a different conclusion , suggesting that the nsc scaling relations are instead substantially shallower than the corresponding mbh scaling relations @xcite .
the version of the @xmath18 relation given in @xcite is @xmath19 we mention here that the @xmath18 relation might be a non - primary correlation , instead resulting from a projection of the fundamental plane , given the observed correlation between @xmath14 and the total host galaxy mass ( @xmath20 ) : @xmath21 ( e.g. , see * ? ? ?
a predicted @xmath18 relation can be easily derived in the globular cluster merger model if we assume that the globular cluster distribution initially follows the stellar light and by using an isothermal sphere density model : @xmath22 , where @xmath15 is the 1d velocity dispersion and @xmath23 the gravitational constant .
the cumulative mass within @xmath24 is @xmath25 , with @xmath26 the velocity of a circular orbit .
the dynamical friction coefficient for a stellar cluster of mass @xmath27 moving in an isothermal sphere model is @xcite : @xmath28~,\ ] ] with @xmath29 the coulomb logarithm and @xmath30 . noting that @xmath31 where @xmath32 is the orbital angular momentum , for a cluster moving on a circular orbit we obtain @xmath33 } /\sqrt{2}r\sigma~.\ ] ] integrating equation ( [ eq3 ] ) yields @xmath34 t/\sigma \right)^{1/2}~.\ ] ] clusters initially within @xmath35 reach the center at a time @xmath36 . assuming that the total mass accumulated in the center is equal to the total mass in globular clusters that are initially within @xmath35
, then we obtain ( e.g. , * ? ? ?
* ) @xmath37 \langle m_{\rm cl
t / g \right)^{1/2 } \nonumber\\ & & \times f \frac { \langle m_{\rm cl } \rangle}{m_\star } \sigma^{3/2}~,\end{aligned}\ ] ] where @xmath38 is the mass of the field stars , @xmath39 is the average globular cluster mass , and @xmath40 is the initial number fraction of globular clusters ( after galaxy formation ) to the total number of stars in the galaxy .
since the term in square brackets in equation ( [ eq5 ] ) is a constant , the globular cluster merger model predicts @xmath41 . in the mw , the initial total number of globular clusters , @xmath42 ,
can be recovered based on their observed luminosity function and semi - analytical modeling of mass - loss due to stellar evolution and due to tidal interaction with the galactic environment .
@xcite derived a survival fraction of @xmath43 for a minimum initial cluster mass @xmath44 . assuming 100 present - day globular clusters we have that the total initial number of clusters in the galaxy is @xmath45 . considering only those gcs that are associated with the bulge on a hubble time , this gives @xmath46 , which compared to the total stellar mass of the bulge , @xmath47 , yields @xmath48 .
we can rewrite equation ( [ eq5 ] ) as @xmath49 despite its simplicity , our model reproduces the observed @xmath13 relation both in slope and normalization .
@xcite proposed a nsc in - situ star formation model regulated by momentum feedback .
this model is an extension of the argument proposed by @xcite to explain the @xmath16 relation , and invokes the formation of a massive nuclear cluster due to gas inflow and accumulation at the center of the galaxy during the early phases of galaxy evolution .
stellar winds and supernovae from a young nuclear cluster with a standard imf produce an outflow with momentum flux given by @xmath50 where @xmath51 is the eddington luminosity calculated by the stellar mass , @xmath52 is the speed of light , @xmath53 is related to the mass fraction of massive stars , and @xmath54 is the electron scattering opacity .
the outflow is initially momentum conserving and produces an outward force on the gas in the bulge , whose weight is @xmath55 , with @xmath56 the enclosed gas mass and @xmath57 the total enclosed mass of the galaxy . for an isothermal potential
one finds @xmath58 where @xmath59 is the baryonic mass fraction @xcite . requiring that the momentum output produced by the nuclear cluster balances the weight of the gas leads to the relation @xmath60 a similar scaling relation can be derived in protogalaxies with non - isothermal dark matter halos @xcite .
this model is very attractive because it contains no free parameters .
in addition , equation ( [ eq7 ] ) is in good agreement with the @xmath13 correlation reported by @xcite : @xmath61 but , as previously mentioned , it is at odds with @xcite who finds nsc scaling relations considerably shallower than the corresponding mbh scaling relations .
the difference between these two studies was due to the proper exclusion of nuclear disks in the sample of @xcite and the larger sample of nscs used in this latter work .
for these reasons we consider the results of graham more robust and conclude that the mclaughlin et al .
model provides a poor description of the observed @xmath13 correlation .
this suggests that momentum feedback may be not relevant , which would be expected if the nscs originated elsewhere and were subsequently deposited into their host galaxy centers .
the size of galactic nuclei clearly correlates with their luminosity , in the sense that brighter nscs have larger effective radii .
the relation is approximately @xcite @xmath63 where @xmath64 is the nsc effective radius ( or half - mass radius ) and @xmath65 its total luminosity .
this relation is consistent with the extrapolation of the @xmath66 relation for elliptical galaxies . in the merger model the radius of the nucleus increases with increasing total mass as globular clusters merge .
the brighter , larger mass nuclei are therefore predicted to be spatially very extended . @xcite used simple energy arguments to derive the size - mass relation for nscs in galaxies with no central mbh . for the sake of completeness we repeat this simple calculation in what follows .
after a merger the nsc energy , @xmath67 , equals the energy of the nucleus before the merger , @xmath68 , plus the energy brought in by the globular cluster : @xmath69 with @xmath70 the internal binding energy of the cluster , and @xmath71 its orbital energy before the merger .
the equations above permit expressing the mass , energy , and radius of the nucleus recursively as @xmath72 where the subscript 1 denotes the initial nucleus , and , by assumption , @xmath73 .
equations ( [ rec1]-[rec3 ] ) imply @xmath74 at the time the mass of the nucleus is still comparable to the mass of the infalling clusters .
after many mergers the nucleus is much more massive than one globular cluster and the relation steepens to @xmath75 .
after 25 mergers , equations ( [ rec1]-[rec3 ] ) imply @xmath76 .
the half - mass radii of globular clusters are @xmath77pc @xcite , irrespective of their luminosity , so for a nucleus assembled from 25 mergers , @xmath78 pc .
such a value is in reasonable agreement with the measured half - mass radii for the brightest nuclei .
we conclude that a model that attributes the origin of nscs to the mergers of globular clusters at the centers of galaxies is consistent with the sizes and luminosities of the nuclei . in a number of galaxies nscs
are observed to coexist with mbhs .
an example of such systems is the mw for which the mass of the nsc , @xmath79 @xcite , is somewhat comparable to the mass of the central black hole , @xmath80 @xcite .
a mbh will disrupt clusters that pass within the radius @xmath81 where @xmath82 is the one - dimensional velocity dispersion of a globular cluster , @xmath83 its core radius , @xmath84 is the velocity dispersion in the nsc and @xmath85 is the influence radius of the mbh . in the presence of a mbh ,
the dependence of the half - mass radius @xmath64 on @xmath14 is substantially weaker than the observed correlation , @xmath86 @xcite , due to the fact that the size of the nsc is determined by the fixed tidal field from the mbh .
when a stellar cluster is disrupted , stars that were initially within the cluster core will redistribute locally at a distance @xmath87 from the mbh .
the density profile of the nsc will therefore have a core of characteristic size @xmath88 .
although the gas model remains somewhat more qualitative there are some indications that the observed @xmath62 relation might be difficult to reconcile with a purely dissipative scenario .
@xcite performed fully self - consistent chemodynamical simulations to investigate how nscs can form through dissipative gas dynamics .
he found that compact nuclei can be formed via dissipative , repeated merger of gaseous ( or stellar ) clumps that develop from nuclear gaseous spiral arms due to local gravitational instability .
these computations showed that fainter nscs are likely to have a more diffuse configuration due to more negative feedback from sne ii which in turn can prevent nscs from forming in faint galaxies .
the fact that nsc formation is more - strongly suppressed by stronger feedback effects in less - luminous galaxies would explain why brighter dwarf galaxies ( de ) are more likely to contain nscs @xcite .
however , simulations also show that more massive nscs are less extended than their lower mass counterparts , which is the opposite of the observed trend ( see figure 15 of * ? ? ?
further investigation of the above inconsistency is needed in order to determine how the results depend on the resolution limit of the simulations and on the adopted initial conditions .
the above discussion gives some level of reliability to the hypothesis of a dissipationless , merging formation for nscs .
we now wish to test whether the merger of globular clusters can result in a nsc similar to that of the mw . as a first step one can derive the globular cluster parameters required to give a peak density , @xmath89 , equal to the observed density of the nsc in the mw , and see whether such parameters are reasonable when compared to typical globular cluster properties .
the stellar cluster central phase space density is given by @xmath90 where @xmath91 is the cluster core density .
after inspiral , we require @xmath92 , or @xmath93 so that @xmath94 the observed nsc density within @xmath95pc at the galactic center ( gc ) is @xmath96 @xcite . comparing this with equation ( [ e1 ] ) ,
we obtain @xmath97 this crude calculation shows that for a mw - like galaxy , the globular cluster parameters required to give the observed peak density of the nsc are quite reasonable .
a simple analytical model is developed in what follows to evolve a population of stellar clusters subject to migration and dissolution in a galactic bulge and to calculate the influx of migrating clusters into the center of the galaxy . in this way we can address the possibility that a substantial fraction of the nsc mass in galaxies like the mw could have been assembled through cluster migration and mergers .
we assume that the central properties of a stellar cluster ( i.e. @xmath82 and @xmath83 ) remain unchanged during inspiral and that @xmath98 , where @xmath99 is the cluster tidal ( limiting ) radius given by @xcite @xmath100 with @xmath101 the galactic potential , and @xmath102 a `` form factor '' that depends on the density distribution within the cluster .
for a king model with a large central concentration , @xmath103 .
equation ( [ rtrking ] ) includes the tidal force due to the radial gradient in the galaxy potential , the centrifugal force from the cluster s orbit and assumes that the density of the cluster goes to zero roughly at the radius where the force acting to remove a star is balanced by the attracting force from the rest of the cluster .
the tidal radius of a king model orbiting in a galaxy with a power - law density profile , @xmath104 , and containing a mbh at its center , is @xmath105^{-1/2}~.\ ] ] the radius at which the tidal force from the mbh starts to dominate the tidal force from the stellar cusp ( in galaxies containing both ) is @xmath106 a king model also satisfies the relation : @xmath107 where @xmath108 is the truncated mass of the globular cluster whose radius is limited by the external tidal field . given the cluster central velocity dispersion , equations ( [ rtr2 ] ) and ( [ mt ] ) can be combined to evaluate , at any radius , the cluster mass permitted by the galaxy tidal field and the mass dispersed along the orbit .
this corresponds to a density enhancement with respect to the galactic background of @xmath109 the resulting radial dependence of the density profile of stars in the growing nsc is easily found to be @xmath110 steeper density cusps in the distribution of background stars give lower values of the density slope for the resulting nsc .
larger values of @xmath17 also give smaller densities near the center since the cluster models start being truncated at larger radii and their mass is dispersed over larger spatial scales .
these basic formalism assumes that mass loss from the clusters is entirely due to their interaction with the external tidal field and it omits mass loss ( evaporation ) due to internal dynamics . since the only clusters that can contribute to the growth of a nsc are
very massive systems with relaxation times longer than their dynamical friction timescale this simplification is quite reasonable and does not alter our results in any important way .
however , dynamical evaporation due to internal dynamics is also accounted for , in the sense that we do not follow the evolution of clusters which are disrupted ( evaporated ) on timescales shorter than the relevant dynamical friction timescales .
luminosity profiles of galaxies are well approximated by power laws with @xmath130 at radii smaller than the effective radius of the stellar spheroid ( @xmath131 ; * ? ? ?
* ) . since for mw - like galaxies the only clusters that can reach the gc in one hubble time are those initially at @xmath132 @xcite , and that within these radii the baryonic matter dominates the galactic potential , we represent the galaxy bulge by using a simple power - law density model : @xmath133 , with @xmath134 , for @xmath135 . ]
@xmath136 , @xmath137pc and @xmath138 ( e.g. , * ? ? ?
the scale length @xmath139 , is related to the bulge effective radius via @xmath140 for @xmath141 .
we assume that the clusters have initially the same distribution of stars in the galaxy ( e.g. , * ? ? ?
* ) and we assign their masses using the cluster initial mass function ( cimf ) , @xmath142 @xcite , and limiting mass values of @xmath143 ; @xmath144 . the fraction of all star formation that occurs in gravitationally bound stellar clusters is usually referred to as `` cluster formation efficiency '' .
the cluster formation efficiency in the mw is @xmath145 @xcite .
accordingly , in our model we assume that the total mass in clusters is initially @xmath146 .
we start by identifying a dissolution time " for globular clusters due to dynamical evaporation as a function of their initial mass @xcite : @xmath147 with @xmath148 , @xmath125 if the cluster is truncated , @xmath149 and @xmath150 the angular velocity at the galactocentric radius @xmath24 .
we then remove clusters with total lifetime , @xmath151 , shorter than the dynamical friction time - scale , @xmath152 .
this procedure reduces the initial total mass in globular clusters from @xmath153 to @xmath154 for @xmath155 , and to @xmath156 for @xmath157 . by comparing equation ( [ tdf - sfc2 ] ) with equation ( [ tdiss ] )
one can easily show that if @xmath158 and @xmath159 initially , this latter condition remains satisfied during inspiral .
we set the core radius of the globular clusters to be @xmath160pc , roughly equal to the median value of the core radii listed in the harris s compilation @xcite of galactic globular clusters .
if a cluster reached its tidal disruption radius its core mass is redistributed uniformly at the radius of disruption over a region of extent @xmath161pc .
but we note that our results do not strongly depend on the particular value for this parameter , which only affects the nsc density profile at very small radii ( @xmath162pc ) . within these central regions
we expect that other dynamical processes ( e.g. , two - body relaxation , mass segregation ) will modify the nsc density profile with respect to the simple predictions of our monte - carlo experiments ( we discuss this point in more detail below ) .
when a cluster enters the inner @xmath163pc , its mass was redistributed using a plummer sphere model of total mass @xmath164 and core radius @xmath83 .
in addition , we make the reasonable assumption that the distribution of field stars is constant when computing the relevant dynamical friction time - scales .
given that the total mass in clusters is only @xmath165 of the bulge mass initially , the density profile of the galaxy at intermediate and large radii ( @xmath166pc ; e.g. , see figure 4 of * ? ? ?
* ) and consequently the dynamical friction timescales are not significantly affected by such simplification . however , to approximately account for the mass that is progressively accumulated into the nucleus , the orbital decay of clusters was computed in order of increasing migration time ; then , for any inspiral , we added the quantity @xmath167 to the terms in square brackets in equation ( [ rtr2 ] ) , with @xmath168 the accumulated mass within @xmath24 due to clusters with shorter orbital decay times . finally , we computed the nsc density at any radius by summing up the contribution , @xmath169 , of all clusters that within @xmath170yr reached that radius .
after this time the total mass left in stellar clusters is @xmath171 , almost independent on the initial value of @xmath172 .
the initial mass of the globular cluster system is therefore reduced from @xmath173 to @xmath174 of the total bulge mass after one hubble time due to dynamical dissolution of the less massive systems , and also due to massive clusters that inspiral into center of the galaxy and dissolve locally to form the stellar nucleus .
figure [ fig1 ] gives the results of such calculations . in agreement with @xcite
we found that the mass of the forming nsc is mostly ( @xmath175 ) composed of clusters with initial masses @xmath176 , suggesting that the low mass clusters scarcely contribute to nsc formation in galaxies .
instead , our results very much depend on the value of @xmath172 , since only very massive clusters can arrive in the central regions of the galaxy in a reasonable time without being destroyed by the galactic tidal forces in the process . in galaxy models without a mbh ,
nsc formation is more efficient , and even for @xmath155 a nsc forms in the central few parsecs of the galaxy .
if a mbh is present at the center of the galaxy , star clusters that come closer than @xmath177 from the center are disrupted and , as a result , the nsc density is limited inside this radius ( left panels in the figure [ fig1 ] ) . for a mbh mass of @xmath178 and @xmath155 ,
no clusters penetrate radii smaller than @xmath179pc . from the lower - left panel in figure [ fig1 ]
we can see that , in this case , the presence of a mbh will strongly inhibit the formation of a nsc . taking @xmath157 ( upper - left panel )
results in a smaller radius of the nsc core and higher central densities , since the most massive clusters can reach smaller galactocentric radii ( @xmath180pc ) . a model that starts with a high mass truncation of @xmath157 reproduces the observed mass density profile of stars in the galactic nsc ( red - dotted lines in figure [ fig1 ] ) outside @xmath181pc .
we conclude that a scenario in which a large fraction of the mass of the milky way nsc is due to infalling globular clusters is in good agreement with current observational constraints .
the results presented here should be interpreted with caution when comparing the detail of the nsc density profile to observations .
our modeling includes a series of simplifications that can have some impact on the inner structure of the resulting nsc .
for example , consideration of the internal dynamics and mass spectrum in the globular clusters could enable mass segregation which can increase @xmath82 and then allow the cluster stars to reach much smaller radii .
a realistic treatment of these effects will require careful @xmath12-body simulations and is beyond the scope of this paper .
nevertheless , we note that the results of our computations that include a central mbh are very similar to those obtained in @xcite via @xmath12-body simulations . in both cases , the density profile that results after the final inspiral event
is characterized by a core of roughly the tidal disruption radius of the most massive clusters , @xmath182pc , and an envelope with density that falls off as @xmath183 .
these properties are similar to those of the mw nsc , with the exception of the core size , which in the mw is somewhat smaller ( @xmath184pc ) .
@xcite showed that such a core will shrink substantially via gravitational encounters in a time of 10 gyr as the stellar distribution evolves toward a bahcall - wolf cusp @xcite . in our computations
cluster inspiral occurs more or less continuously over the lifetime of the galaxy .
the core resulting from the combined effects of cluster inspiral and relaxation would therefore be somewhat smaller than that we found above , and closer to the observed core size ( we discuss this point in more details below in [ mbh ] ) .
yr in galaxy models with inner density profile slope @xmath185 . in these computations we set @xmath136 , @xmath186 , @xmath187 , and @xmath146 .
the high mass truncation of the cimf is @xmath157 .
the largest nsc central densities are found in models with small values of @xmath113 . in the middle and lower panels the properties of the nsc models are compared to those of real nscs in early type galaxies ( dots are data from * ? ? ?
* ; * ? ? ?
the middle panel plots the nuclei in the size ratio versus mass ratio plane .
open circles correspond to the nsc models of the top panel with @xmath188 , where the relaxation time and effective radius of the nsc decrease with increasing @xmath113 .
the blue - filled squares represent the model in the upper right panel of figure [ fig1 ] .
lines indicate the critical value of @xmath189 over which nuclei expand for galaxy models described by einasto indices @xmath190 ( continue line ) , @xmath191 ( dashed line ) and @xmath192 ( dot - dashed line ) , ( for comparison see figure 1c in * ? ? ?
the lower panel shows the dependence of nsc half - mass relaxation time on its total mass .
, title="fig : " ] + figure [ fig2 ] shows the results of additional computations that explore the effect that varying @xmath17 has on the structure of the nsc . in these plots we set @xmath157 and we relate the mass of the galaxy to @xmath17 through the relation @xmath193 \approx 8.4 + 1.9 \log \left [ m_{\rm sph}/7\times 10^{10}m_{\odot}\right]$ ] @xcite .
the galaxy velocity dispersion , @xmath15 , is derived from the @xmath16 relation @xcite , while the model scale length , @xmath139 , is obtained from the effective radius : @xmath194 , where here the normalization reproduces approximately the observed effective radii of elliptical galaxies @xcite . in the left panel of figure [ fig2 ] we show the increasing dominance of the central mbh over the nsc stars for more massive galaxies .
our results agree reasonably well with studies of galaxies at high mass end with @xmath195 , and which exclude the presence of stellar nuclei in such systems ( e.g. , * ? ? ?
the middle panel of figure [ fig2 ] displays how the total central mass in the nuclear component ( i.e. , mbh + nsc ) divided by the stellar mass of the host spheroid varies with this latter quantity .
also in this case our results agree reasonably well with the observed correlation ( dashed line in the plot ) . in order to reproduce the nsc masses obtained from observations ,
our model would require a high mass truncation of the cimf of @xmath157 and that @xmath196 of stars in galaxies originate in stellar clusters .
such a result is in agreement with the conclusion of @xcite that the globular cluster infall model strongly under - predicts the observed nsc masses when assuming the present day number of globular clusters in galaxies .
this assumption might be incorrect however , given that the number of globular clusters in a galaxy drops rapidly with time due to dynamical disruptions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition , the cluster formation efficiency and the cluster disruption rate also vary substantially with cosmic time , peaking at early epochs , in gas - rich disk galaxies @xcite .
in fact , up to 30 per cent of all stars in the universe could have been formed in bound stellar clusters @xcite .
the right panel of figure [ fig2 ] gives the density profile of a nsc that forms around a mbh of mass @xmath197 assuming an initial mass in clusters @xmath146 .
smaller black hole masses correspond to smaller tidal disruption radii for the infalling globular clusters . as a consequence ,
the nsc peak density becomes progressively larger and the nsc core radius smaller as @xmath17 decreases .
the mass delivered in the inner region of the galaxy is also a strong function of @xmath17 . for @xmath198 , very little mass ( @xmath199 )
can be deposited in the inner @xmath200pc of the galaxy . in this case
the nsc density profile remains below the stellar density of the background galaxy . to summarize the results presented in this section ,
if we postulate that nscs grow slowly through globular cluster migration and merging , then in order to form a nsc similar to that of the mw , our model will require @xmath201 and that @xmath202 per cent of the bulge mass originated in stellar clusters .
we showed that this fraction reduces to roughly @xmath174 of the total bulge mass after @xmath170yr due to dynamical evaporation of the less massive clusters , and also because more massive clusters inspiral toward the galactic center and dissolve due to their interaction with the galactic tidal field .
the presence of a pre - existing mbh at the center of the galaxy can have a strong impact on the inner structure of the forming nucleus .
if a mbh of mass @xmath203 sits at the center of the galaxy , massive globular clusters are disrupted at radii of order @xmath77pc and the nsc density profile will have a core of roughly this radius .
mbhs less massive than about @xmath204 do not influence very much the structure of the stellar nucleus during its formation since the tidal disruption radius of massive gcs is in this case of the order ( or less than ) the core radius of the most massive clusters , and in either models , with or without mbh , the nsc density profile will have have a central core of typical size @xmath205 . in low mass spheroids after a few gyr such a core will most likely relax to a bahcall - wolf cusp or , in the absence of mbh , undergo core - collapse .
the central peak density of the resulting nsc progressively declines as @xmath17 increases .
nscs forming around mbhs with @xmath206 have such low central densities that they would be more difficult to observe as distinct galactic components ; this appears to be in agreement with observations which reveal a lack of nscs in stellar spheroids brighter than about @xmath207 .
in the above discussion we have considered dehnen s models with inner density profile slope @xmath208 . in figure [ fig3 ]
we relax this assumption and explore the formation of nscs in galaxy models with different values of @xmath113 . in these computations
we adopt @xmath157 and @xmath146 .
the structural parameters defining the galaxy model are @xmath136 , @xmath186 , and @xmath187 . on the basis of equation ( [ deltarho ] )
we would expect the final ( after @xmath170yr ) densities in the inner @xmath209pc of the forming nsc to be lower in galaxy models with steeper density profiles , provided that the dynamical friction timescale of massive stellar clusters remains shorter than the hubble time .
the results illustrated in the upper panel of figure [ fig3 ] agree with this basic prediction , suggesting that , independently on @xmath113 , the sinking timescale of massive clusters in these models is short enough that they can always reach the center of the galaxy and form a compact nucleus .
the middle and bottom panels in figure [ fig3 ] compare the structural properties of our model nscs to the properties of nscs in early - type galaxies that were found to be nucleated in @xcite .
for each of the nsc models an approximation of its effective radius is obtained as the effective radius of the best fitting srsic model to the nsc projected density profile within a galactocentric radius of @xmath210pc .
given @xmath64 , the total nsc mass , @xmath14 , is then obtained as the total mass within a radius twice the effective radius .
a useful reference time is the relaxation time computed at @xmath64 .
setting @xmath211 , and ignoring the possible presence of a mbh , the half - mass relaxation time is @xcite : @xmath212^{3/2 } n^{1/2}}{(m / m_{\odot})^{1/2}}{\rm yr}~,\ ] ] where @xmath12 is the total number of stars and @xmath213 is the mass of a single star ( @xmath214 in figure [ fig3 ] ) . from figure [ fig3 ]
it is evident that our nsc models have structural properties that are similar to those of real nuclei . in the absence of a central mbh
the dynamical evolution of a nuclear cluster is a competition between core collapse , which causes densities to increase , and heat input from the surrounding galaxy , which causes the cluster to expand and densities to decrease @xcite . in a double - plummer - law galaxy model , the maximum size of a nsc of mass @xmath215 in order to resist expansion
is @xmath216 @xcite .
this simple condition , when compared to the middle panel of figure [ fig3 ] , implies that the core - collapse time of our models are always shorter that the time for the nucleus to absorb energy from the rest of the galaxy .
however , when using more realistic einasto profiles to describe the surrounding galaxy our nsc models appear to be close to the transition region between systems that are in the prompt core collapse phase and systems for which the evolution is driven by heating from the galaxy . our models will therefore either expand or contract depending on the `` compactness '' of the surrounding galaxy .
stellar spheroids more luminous than @xmath217 often host high mass black holes ( @xmath218 ) at their center , while they show no evidence for nucleation .
rather , the density profile of such `` giant '' elliptical galaxies is observed to be flat inside the influence radius of the mbh @xcite .
two factors have been invoked to explain the observed lack of nscs in such systems : ( i ) nscs and mbhs grow in competition from the same gas reservoir @xcite .
if a mbh forms first this can prevent , through its feedback , a nsc from growing if the gas accretion rate is smaller than the eddington rate @xcite .
( ii ) massive black hole binaries forming during the last galaxy - galaxy major merger destroyed their host nscs by ejecting stars from the inner galactic regions @xcite .
observations seem to point against both scenarios . picture ( i )
seems to break down for nsc - dominated galaxies , given the extremely low accretion rates observed in bulgeless galaxies and that at least few of them also contain central mbhs .
the model of @xcite is disfavored due to the fact that what sets nsc disappearance does not seem to be galaxy morphology .
rather , there is a limit to the mbh / nsc mass ratio that fixes a sharp transition from galaxies with nsc and mbh - dominated galaxies @xcite . .
even after @xmath170yr the cluster orbital radii have hardly changed from their initial values .
the lower panel gives the density profile of the galaxy , assuming that the clusters did have enough time to migrate to the center : dotted , dashed and dot - dashed curves show the results of the accumulation of @xmath219 clusters with @xmath220 and @xmath221 respectively .
red lines correspond to computations without a mbh at the center of the galaxy .
the galaxy background model is shown as a solid black line .
vertical marks correspond to the orbital radius of the cluster in the model with an mbh after @xmath170yr and starting from @xmath222pc .
arrows give the tidal disruption radii of the clusters due to the mbh . at radii smaller than these our computations including a mbh are no longer valid . ] of course , it is possible that nuclei are absent in bright galaxies because some mechanism prevents them from forming in the first place , or because they did not have time to reform after they were destroyed by the scouring effect of binary black holes . in
what follows , we show that nscs might in fact be difficult to form in such galaxies , due to long dynamical friction time - scales of star clusters and also the little mass that the clusters can deliver to the center due to the strong mbh tidal field . in order to do so
, we repeat a similar analysis to that presented in [ mw ] but for a galaxy like m87 , a giant elliptical with no evidence for nsc .
we stress that the problems discussed here and in [ mw ] are substantially different .
the influence radius of the mbh in a galaxy like m87 extends much further out ( a few hundreds of parsecs from the center ) than the tidal disruption radius of a massive globular cluster ( @xmath7pc ) , while for the mw we found @xmath223 . in the former case
, the clusters will therefore move in a stellar background whose velocity distribution is directly influenced by the presence of a mbh , which will strongly affect the timescale for cluster inspiral as we now show .
the distribution of field - star velocities has the following form near a mbh : @xmath224 where the normalizing constant corresponds to unit total number .
this expression gives the local distribution of velocities at a radius where the circular velocity is @xmath225 , assuming that the density of field stars follows @xmath226 .
the phase space density is zero for @xmath227 .
the fraction of stars that move slower than the local circular velocity can be computed as @xmath235 this function varies smoothly from @xmath236 when @xmath237 to @xmath238 when @xmath239 .
for the fast moving stars an ( ad - hoc ) approximation is @xmath240 \nonumber \\ & \simeq&0.1721 + 0.5280\gamma-0.1812\gamma^2 -529.2\exp(-14.46\gamma)\label{fms}~. \nonumber\end{aligned}\ ] ] for @xmath241 , @xmath242 , and when @xmath243 the contribution from the slow stars is of order @xmath244 larger than the frictional force due to the fast moving stars . for a test particle of mass @xmath27 , integrating equation ( [ am ] ) yields @xmath254 m_{\rm cl}\times t\big]^{\frac{1}{\gamma-3/2}}~,\nonumber\end{aligned}\ ] ] for @xmath255 , and @xmath256 m_{\rm cl}\times t \right)~\ ] ] for @xmath257 .
equation ( [ rt1 ] ) corresponds to a characteristic decay time for angular momentum loss ( for @xmath258 ) : @xmath259^{-1}}{\gamma-3/2}\\ & & \times m_{\bullet,9.5}^{3/2}\tilde{\rho}_{50}^{-1 } \tilde{r } _ { 300}^{-3/2}m_{{\rm cl},6}^{-1}\left(\frac{r_{in } } { \tilde{r}}\right)^{\gamma-3/2 } \nonumber\end{aligned}\ ] ] where @xmath260 , @xmath261 , and @xmath262 .
we note that , for @xmath263 , @xmath264 becomes negative and , formally , equation ( [ am ] ) gives an infinite decay time to the center . in this latter case
the dynamical friction timescale can be re - defined as the time required to reach a radius which is some fraction , @xmath265 , of its initial value : @xmath266 , with @xmath264 from equation ( [ taubh ] ) . in the case of extended objects , we assume that the central properties of the globular cluster ( i.e. @xmath82 and @xmath83 ) remain unchanged during inspiral .
the cluster s limiting radius is given by equation ( [ rtr2 ] ) , which permits expressing the satellite mass as a function of radius .
setting @xmath125 in equation ( [ am ] ) and assuming that the galactic potential is dominated by the mbh , we obtain @xmath267\sigma_k^3\times t\big]^{\frac{1}{\gamma-3}}~,\nonumber\end{aligned}\ ] ] and @xmath268^{-1}}{3-\gamma } \left(\zeta^{\gamma-3}-1\right ) \nonumber\\ & & \times m_{\bullet,9.5}^{2 } \tilde{\rho}_{50}^{-1}\tilde{r}_{300}^{-3 } \sigma_{k,10}^{-3 } \left(\frac{r_{in } } { \tilde{r}}\right)^{\gamma-3}~.\end{aligned}\ ] ] we used equations ( [ rt1 ] ) and ( [ rvst ] ) to trace the orbital evolution of massive stellar clusters in the core of our galaxy model .
we take a limiting value for the mass of @xmath269 , corresponding to the non - truncated model , when the cluster is far from the center .
the simulated orbits ( upper panel of figure [ fig4 ] ) demonstrate that the dynamical friction time - scale in the core of bright elliptical galaxies is extremely long .
also very massive clusters ( @xmath270 ) starting well within the galaxy core ( @xmath271pc ) , do not reach the center after one hubble time . in the lower panel of figure [ fig4 ] ,
we use equations ( [ rtr2 ] ) and ( [ deltarho ] ) to compute the density profile of a nsc resulting from the accumulation of @xmath219 equally massive globular clusters in the center of our m87 galaxy model . at radii smaller than @xmath177 ( marked by black arrows in the plot ) the clusters are fully disrupted by the interaction with the mbh and our integrations are no longer valid .
@xmath12-body simulations show that the density profile of stars will be very flat or even declining toward the mbh inside @xmath177 , as the stars accumulate near the radius of disruption @xcite . in order to hightligh
the role of the central mbh in determining the structure of the growing nucleus figure ( [ fig4 ] ) also shows the density profile of the nucleus when the mass of the mbh is set to zero . in this case , most of the cluster mass is delivered in the inner @xmath209pc and consequently the nsc is much more centrally concentrated .
after @xmath219 inspiral events , the nsc appears to be distinct from the galaxy background density profile .
apparently , nscs assembled around mbhs have a spatially more diffuse configuration and lower densities than those forming in galaxies with no mbh .
we conclude that the formation of nscs in giant ellipticals might be inhibited by the presence of their central mbh for two basic reasons : ( i ) the long dynamical friction timescale of massive objects in the galaxy core ; ( ii ) clusters are disrupted by the strong tidal field of the mbh producing a merger remnant with density profile that rises only modestly above that of the background galaxy .
we note in passing that the latter argument may also apply to the case in which a giant elliptical galaxy accretes a smaller galaxy containing a dense central nucleus .
the large cores observed in the central light profile of bright ellipticals are usually interpreted in terms of the scouring effect of mbh binaries forming during major merger events @xcite .
it is less clear however how such cores can be preserved up to the present epoch despite the large number of minor mergers that are predicted to occur by standard cosmological models .
for instance , an m87 like galaxy is expected to have accreted few galaxies of the size of the mw in the last @xmath272gyr @xcite .
a plausible explanation for the lack of central dense stellar concentrations in the brightest galaxies was provided by @xcite .
these authors performed high resolution @xmath12-body simulations of the accretion of high - density dwarf galaxies by low - density giant galaxies .
they found that the cusp of the secondary galaxy is disrupted during the merger by the giant galaxy mbh tidal field , producing a remnant with a central density that is only slightly higher than that of the giant galaxy initially ; removing the black hole from the giant galaxy allowed the smaller galaxy to remain essentially intact and led to the formation of a high central density cusp , contrary to what is observed .
in the previous sections we showed that nscs resulting from the accumulation of globular clusters around mbhs have lower central densities and larger cores than those forming in galaxies without mbhs . such a result is consistent with the results obtained using more sophisticated @xmath234body simulations @xcite .
our analysis however did not account for the effects of internal dynamics occurring in the nsc during and after its formation . in a preexisting nsc ,
the presence of a mbh would inhibit core collapse , causing instead the formation of a @xcite cusp on the two - body relaxation timescale , followed by a slow expansion as stars are tidally disrupted @xcite .
whether or not a bahcall - wolf cusp will form depends on the initial core size of the nucleus and on its relaxation time . for nuclei similar to the mw nsc ,
relaxation times are too long to assume that they have reached such a collisionally relaxed state , but they are still sufficiently short that gravitational encounters would substantially affect their structure over the age of the galaxy .
two body gravitational interactions will cause the central density of the nucleus to increase and the core to shrink as the stellar distribution evolves toward the bahcall - wolf form ( e.g. , * ? ? ?
the time evolution of the core radius can be approximately described by the expression @xcite @xmath273 ~,\ ] ] where @xmath274 is the relaxation time computed at the radius of influence of the mbh . for the mw , @xmath275gyr and @xmath276 .
a black hole of this mass corresponds to a core of @xmath182pc ( figure 1 ) . from equation ( [ evrel ] )
, we see that gravitational encounters occurring during and after the formation of the nsc will reduce the core size to @xmath277pc after @xmath170yr . such a value is more consistent with the size of the core observed in the distribution of late - type ( old ) stars in the inner parsec of the mw @xcite .
whether nscs in other `` power - law '' galaxies will turn out to have stellar cores similar to that of the mw remains to be seen .
nscs containing mbhs are expected to be the main birthplace of gravitational wave ( gw ) sources for space - based interferometers @xcite .
these include the capture of stellar - mass black holes ( bhs ) by mbhs , also called `` extreme mass - ratio inspirals '' ( emris ; * ? ? ?
* ; * ? ? ?
understanding systems like the mw nsc and their origin is therefore crucial for making predictions about the event rates for low frequency gravitational wave detectors . most of the emri rate calculations reported in the literature are derived under the assumption that the galactic nucleus had enough time to reach a state of mass segregation , which implies a high density of bhs near the center ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this appears to be in conflict with observations which suggest an unrelaxed state for the distribution of stars at the gc . @xcite and @xcite demonstrated that in the absence of an initial cusp in the stars , even after @xmath278 gyr the density of bhs could remain substantially below the densities inferred from steady - state models .
all these previous studies , however , assumed that the stellar bhs had the same phase - space distribution initially as the stars .
this is most likely to be a poor assumption if a substantial fraction of the nsc mass comes from orbitally decayed globular clusters . in the merger model
the resulting distribution of stellar remnants will reflect their distribution in their parent clusters just before they reach the center of the galaxy . in a dense stellar cluster , bhs formed by the supernova explosions of the most massive stars tend to segregate into the cluster core and form a sub - cluster of bhs which dynamically decouples from the rest of the cluster @xcite . when the central density of bhs becomes large enough
, bh - bh binary formation becomes efficient .
subsequent dynamical interactions involving binary bhs and higher multiplicity systems will tend to eject the bhs from the cluster until only a few of them are left ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
we may consider two possibilities : ( i ) the dynamical friction timescale of star clusters is short compared to the encounter driven evaporation timescale of their bh sub - cluster . in this case , the mass - segregated bh sub - clusters , due to their high central densities , can reach galactocentric radii as small as @xmath279pc @xcite . the preferential removal of stars from the outer parts of the clusters by the strong galactic tidal field might lead to the formation of a nsc with a central over - abundance of bhs when compared to predictions from standard mass functions @xcite .
( ii ) the orbital decay timescale is shorter than the evaporation timescale of the cluster bh sub - system .
most of the bhs will be dynamically ejected in this case before the cluster reaches the center of the galaxy . as a consequence of this ,
very few bhs are delivered to the vicinity of the mbh .
all of the above mentioned topics require a dedicated study which we defer to a future paper .
our work shows that a purely dissipationless merger scenario can explain , without obvious difficulties , the basic physical properties of nscs .
it is important to note that we adopted a rather idealized model of an isolated galactic spheroid ( with / without mbh ) .
this idealized model enabled us to neglect some gradients of the role of galaxy merging in nsc formation which can only be addressed by means of comprehensive cosmological hydrodynamical simulations .
simulations demonstrate that tidal torques in major mergers of gas - rich galaxies can induce rapid inflow of gas into the center of a galaxy followed by intense nuclear starbursts which results in a central light `` excess '' in the surface brightness profile of the galactic spheroid ( e.g. , * ? ? ?
* ; * ? ? ?
unfortunately , the spatial resolution ( @xmath280pc ) of current cosmological simulations is not high enough to address the role of galaxy mergers in the context of nscs , at least for low and low - intermediate mass galaxies .
in addition , it is not clear whether this picture would apply to elliptical , early - type galaxies which lack the large gas reservoirs of spirals , and thus should not experience frequent central starbursts .
on the other hand , the fact that the nuclei formation histories were in part governed by local and dissipative factors is supported by a wide range of observational phenomena . as an example , the gas distribution and the kinematics of the gas in the nearby spiral ngc6946 suggest the presence of a central , small - scale , s - shaped stellar bar which appears to funnel gas towards the galaxy nucleus ( within the inner @xmath7pc ) where about @xmath281 of molecular gas have been accumulated .
star forming events triggered by the rapid inflow of gas in the center of the galaxy may then contribute to the growth of its nsc @xcite .
support to a dissipative origin is added by the fact that nscs in general tend to have a wide range of stellar ages , including young stellar populations @xcite .
our gc contains for instance a large population of young massive stars that most likely formed locally following the infall and fragmentation of a dense gaseous clump in the vicinity of sgr a * @xcite .
such bursts may occur continuously over the age of the galaxy and thus produce a significant fraction of its nsc mass .
stellar population synthesis studies also show that the gc appears to have undergone continuous and recurrent star formation over the last @xmath282gyr , but it is not possible to fit the observations with ancient burst models , such as would be appropriate for an old population of stars that originated in globular clusters @xcite .
previous work incorrectly used the latter arguments to argue against a dissipationless origin for nscs ( e.g. , * ? ? ?
* ; * ? ? ?
it is important to stress that although observations probe recent and episodic star formation they do not exclude that the bulk of the gc stellar population is in old stars .
in fact , the gc luminosity function appears to be consistent with a star formation history in which a large fraction ( about 1/2 ) of the mass consists of old ( @xmath202 gyr ) stars and the remainder is due to continuous star formation @xcite .
accordingly , @xcite found that about 80 per cent of the stellar mass in the inner parsec of the galaxy is in old stars that formed more than @xmath283gyr ago .
we add that , although globular clusters in our galaxy are exclusively old stellar systems ( e.g. , * ? ? ?
* ) , this might not be always the case in other galaxies .
it is therefore not clear what fraction of the mass in nscs is contributed by local gaseous fragmentation .
more investigation needs to be done in order to explore the implications of in - situ star formation as the origin of galactic nuclei .
the morphology and kinematics of nscs are of great importance for understanding their origin ( e.g. , * ? ? ?
aspherical nscs are commonly observed in external galaxies .
@xcite found that the three edge - on late - type galaxies ic 5052 , ngc 4206 and ngc 4244 have nuclei that are strongly elongated along the plane of their host galaxies disks .
such clusters show evidence for multiple morphological components , with a disk - like young stellar population superimposed on an older more spherical component .
the radial velocity map of the nucleus in ngc 4244 , the nearest of these three galaxies ( @xmath284mpc ) , shows evidence for strong rotation , @xmath221 at @xmath282pc from the center , compared to a central velocity dispersion of @xmath285 @xcite .
there is also evidence for flattening and rotation in the m33 nucleus @xcite .
the m33 nsc is elongated along the major - axis of the galaxy and rotates at @xmath286 , while the central velocity dispersion is @xmath287 @xcite .
the mw nsc also appears to be rotating parallel to the overall galactic rotation @xcite .
unfortunately , as a consequence of the strong interstellar extinction along the gc line of sight , our knowledge of the galactic nsc morphology and size remains very limited .
is the observed rotation of nscs consistent with the predictions of a dissipationless model for nsc formation ?
of course if a large fraction of the nsc mass was formed via accretion of clusters isotropically distributed throughout the galaxy , no net rotation would be expected , since there will not be any preferred direction for inspiral .
if the primary formation process is instead gas accretion from the galactic disk , this would naturally explain both the flattening and the fact that nscs rotate parallel to their host galaxy rotation . although young and rotating components of nscs might be difficult to reconcile with a globular cluster origin ( due to the long time scale for inspiral compared to the ages of the stars ) , we believe that the evidence for nsc rotation alone , at least in late - type galaxies , is not a strong argument against a dissipationless origin . for instance , clusters falling into the gc could have originated in the inner part of the galactic disk and they will therefore share in its rotation @xcite .
another possibility is that globular clusters crossing the galactic disk experience a greater frictional force from the increased local stellar / gas density @xcite and hence they can be dragged down into the disk plane and transported into the central region of the galaxy where they then accumulate to form a dense nucleus . in either case
, the forming nsc will appear to rotate in the same sense of the galaxy .
these hypotheses lead to basic predictions that might be testable with future observations ; specifically : ( i ) nscs in early - type galaxies , which do not have extended massive disk - like structures , might have slower rotation with respect to nscs of spiral galaxies .
( ii ) there should be some mild depletion of stellar clusters in the galactic disk as compared to off the disk plane .
in conclusion , comparing kinematic data with simulations for distinguishing between gas and cluster accretion may be difficult , since the detailed structure of simulated nscs varies in an important way with the orbital initial configuration of the infalling clusters .
future observational work should be able to provide more reliable models for the spatial distribution of globular clusters near the center of galaxies , where timescales for infall are reasonably short .
galaxies fainter than @xmath288 do not contain prominent stellar nuclei @xcite .
the lack of nscs in dwarf spheroidals was explained in the context of the merger formation scenario in @xcite .
these authors used high resolution @xmath12-body simulations to study the orbital decay of globular clusters in dark matter halos with a core - density structure in the central regions , similar to what inferred by observations @xcite . in these models the standard chandrasekhar s dynamical friction formula breaks down @xcite and one finds that a massive object stalls at roughly the dark matter core radius ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; the number of globular clusters predicted to inspiral would be then @xmath289 .
this suggests that the lack of nscs in the fainter spheroidals could be simply a consequence of the arbitrarily long sinking timescales of stellar clusters in these systems .
alternatively , the lack of nuclei in faint galaxies could be related to the small total number , or even absence , of globular clusters in these galaxies . in the local group ,
globular clusters seem in fact to disappear in stellar spheroids fainter than @xmath290 ( e.g. , * ? ? ?
* ) , which would make the formation of nscs through globular cluster merging impossible in these systems ( see also * ? ? ?
in this paper we considered a dissipationless formation model for nscs where globular clusters orbital decay and merge at the center of a galaxy to form a compact nucleus .
our main results are summarized below . *
the observed scaling relation between nsc masses and the velocity dispersion of their host spheroids , the @xmath18 relation , is difficult to reconcile with a purely dissipative formation model for the nuclei .
these models predict that @xmath14 is a steeply rising function of @xmath15 .
the observed @xmath18 relation is instead in agreement with the predictions of a dissipationless formation model .
dissipationless formation modes produce relations that are substantially shallower than the corresponding mbh scaling relations and are therefore more consistent with observations . *
the globular cluster merger model , in the absence of a central mbh , naturally reproduces the observed relation between the size of galactic nuclei and their total luminosity , @xmath1 . when a mbh is present , the dependence of the nsc radius on its mass is substantially weaker than the observed relation because the size of the nsc is mainly determined by the fixed tidal field of the mbh .
* we derived explicit expressions for the orbits of globular clusters owing to dynamical friction and subject to mass - loss due to their tidal interaction with the galaxy ( equation [ sfc2 ] ) or with a central mbh ( equation [ rvst ] ) .
these expressions were used to ( i ) address the possibility that nscs in galaxies similar to the mw could have been assembled via cluster migration and mergers ; and ( ii ) to follow the orbital evolution of globular clusters in the core of bright elliptical galaxies .
* nscs that form through the mergers of globular clusters have a density profile characterized by a parsec - scale core and an envelope that falls off as @xmath183 .
these properties are similar to those of the mw nsc .
a nsc with mass comparable to that of the mw nsc is obtained by assuming an initial total mass in stellar clusters which is consistent with the cluster formation efficiency inferred from observational studies of embedded clusters in galactic molecular clouds ( figure [ fig1 ] ) . *
a pre - existing mbh at the center of the stellar spheroid has a strong impact on the structure of the growing stellar nucleus ( figure [ fig2 ] ) .
tidal stresses from a mbh disrupt the clusters when they pass within their tidal disruption radius , @xmath177 , limiting the density within that radius .
hence , the density profile of the resulting nsc will also have core size of @xmath88 . removing the mbh from the galaxy
allows the stellar clusters to keep their inner structure almost unchanged leading to the formation of a nsc with higher peak densities .
we find that separating the contribution of the nucleus from that of the galaxy could more difficult in stellar spheroids with more massive mbhs .
* for globular clusters orbiting in the core of a massive elliptical galaxy like m87 , the timescale to reach the center is much longer than one hubble time ( figure [ fig4 ] ) .
the essential reason for this is that the phase - space density of a shallow density cusp of stars around a mbh falls to zero at low energies : inside the galaxy core there are no stars at any radius that move slower than the local circular velocity .
the standard chandrasekhar s formula predicts little or even no frictional force in this case .
in addition , when a mbh is included , the tidal field near @xmath291 becomes much stronger , resulting in faster mass loss and fewer stars deposited to the center .
based on these facts , we conclude the presence of central mbhs might inhibit the formation of compact nuclei in the brightest galaxies , in agreement with what is observed .
i am grateful to d. merritt for numerous discussions about the ideas presented in this paper and for his detailed comments to an earlier version of this manuscript .
i thank r. capuzzo - dolcetta for encouraging me to study this problem , d. kruijssen and a. madigan for useful discussions and m. alvarez and a. graham for their comments to an earlier version of this paper .
this material is based upon work supported in part by the national science foundation grant no .
1066293 and the hospitality of the aspen center for physics .
agarwal , m. & milosavljevi 2011 , apj , 729 , 35 alexander , t. , & hopman , c. 2009 , apj , 697 , 1861 amaro - seoane , p. et al .
2012 , cqgra , 2914016 antonini , f. & merritt , d. 2012 , apj , 745 , 83 antonini , f. , capuzzo - dolcetta , r. , mastrobuono - battisti , a. & merritt , d. 2012 , apj , 750 , 111 antonini , f. 2012 , in preparation balcells , m. , graham , a. w. , & peletier , r. f. 2007 , apj , 665 , 1084 baumgardt , h. , & makino , j. 2003 , mnras , 340 , 227 bahcall , j. n. , & wolf , r. a. 1976 , , 209 , 214 bartko et al .
2010 , apj , 708 , 834 bekki , k. & graham , a. 2010 , apj , 714 , l313 bekki , k. 2010 , mnras , 401 , 2753 banerjee , s. , & kroupa , p. 2011 ,
apj , 741 , l12 banerjee , s.,baumgardt , h. , kroupa , p. 2010 , mnras , 402 , 371 bker , t. , laine , s. , van der marel , r. p. , sarzi , m. , rix , h. , ho , l. c. , & shields , j. c. 2002 , aj , 123 , 1389 bker , t. , sarzi , m. , mclaughlin , d. e. , van der marel , r. p. , rix , h. , ho , l. c. , & shields , j. c. 2004 , aj , 127 , 105 bker , t. 2010 , ihea.book , 99 bekki , k. , couch , w. j. , drinkwater , m. j. , & shioya , y. 2004 apj , 610 , l13 bekki , k. 2007 , pasa , 24 , 77b bik , a. , lamers , h. j. g. l. m. , bastian , n. , panagia , n. , & romaniello , m. 2003 , a&a , 397 , 473 buchholz , r. m. , schdel , r. , & eckart , a. 2009 , a&a , 499 , 483 capuzzo - dolcetta , r. , 1993 , apj , 415 , 616 capuzzo - dolcetta , r. , miocchi , p. , 2008 ,
mnras , 388 , l69 casertano , s. , phinney , e. s. , & villumsen , j. v. , 1987 , iau symp .
127 , structure and dynamics of elliptical galaxies , ed .
t. de zeeuw ( dordrecht : reidel ) , 475 chandrasekhar , s. 1943 , apj , 97 , 255 ct et al .
2006 apjs , 165 , 57 de grijs , r. , anders , p. , bastian , n. , lynds , r. , lamers , h. j. g. l. m. , & oneil , e. j. 2003 , mnras , 343 , 1285 dehnen , w. 1993 , mnras , 265 , 250 de lorenzi , f. , hartmann , m. , debattista , v. p. , seth , a. c. , & gerhard , o. 2012 , arxiv:1208.2161 do , t. , ghez , a. m. , morris , m. r. , lu , j. r. , matthews , k. , yelda , s. , & larkin , j. 2009 , apj , 703 , 1323 downing , j. m. b. , benacquista , m. j. , giersz , m. , & spurzem , r. 2010 , mnras , 407 , 1946 downing , j. m. b. , benacquista , m. j. , giersz , m. , & spurzem , r. 2011 , mnras , 416 , 133 fakhouri , o. , ma , c. , & boylan - kolchin , m. 2010 , mnras , 406 , 2267 ferrarese , l. et al . 2006 , apj , 644 , l21 ferrarese , l. , & ford , h. 2005 , ssrv , 116 , 523 ferrarese , l. , & merritt , d. 2000 , apj , 539 , l9 figer , d. f. , rich , r. m. , kim , s. s. , morris , m. , & serabyn , e. 2004 , apj , 601 , 319 freitag m. , grkan m. a. , rasio f. a. 2006 , mnras , 368 , 141 fall , s. m. , & zhang , q. 2001 , apj , 561 , 751 forbes , d. a. , lasky , p. , graham , a. w. , & spitler , l. 2008 , mnras , 389 , 1924 fujii , m. , iwasawa , m. , funato , y. , & makino , j. 2009 , apj , 695 , 1421 fujii , m. , iwasawa , m. , funato , y. , & makino , j. 2010 , apj , 716 , l80 gebhardt , k. , et al .
2001 , apj , 539 , l13 gebhardt , k. , et al .
2001 , aj , 122 , 2469 gehbardt , k. , et al .
2011 , apj , 2 , 119 genzel , r. et al . 2003 , apj , 594 , 812 goerdt , t. , moore , b. , read , j. i. , stadel , j. , & zemp , m. 2006 , mnras , 368 , 1073 ghez , a. m. et al .
, 2008 , apj , 689 , 1044 gieles , m. 2009 , mnras , 394 , 2113 gieles , m. , & baumgardt , h. 2008 , mnras , 389 , l28 gillessen , s. , eisenhauer , f. , trippe , s. , alexander , t. , genzel , r. , martins , f. , ott , t. 2009 , apj , 692 , 1075 graham , a. w. , & guzmn , r. 2003 , aj , 125 , 2936 graham , a. , & worley , c. 2008 , mnras , 388 , 1708 graham , a. w. & spitler , l. r. 2009 , mnras , 397 , 2148 graham , a. w. , driver , s. p. 2007
, apj , 655 , 77 graham a. w. 2012 , mnras , 422 , 1586 graham , a. w. 2012b , apj , 746 , 113 harris , w. 1996 , aj , 112 , 1487 hashimoto , y. , funato , yoko , & makino , j. 2003 , apj , 582 , 196 hring , n. , & rix , h. 2004 , apj , 604 , l89 hartmann , h. , debattista , v. p. , seth , a. , cappellari , m. , & quinn , t. r. , 2011 , mnras , 418 , 2697 hopman , c. , & alexander , t. , 2006 , apj , l645 hopkins , p. f .. ; hernquist , l. , cox , t. j. ; dutta , s. n. & rothberg , b. 2008 , apjs , 181 , 486 hopkins , p. f. , cox , t. j. , dutta , s. n. , hernquist , l. , kormendy , j. , & lauer , t. r. 2009 , apjs , 181 , 135 hughes , s. a. 2003 , annals of physics , 303 , 142 kruijssen , j. m. d. , & portegies zwart , s. f. 2009 , apj , 698 , l158 kruijssen , j. m. d. , pelupessy , f. i. ; lamers , h. j. g. l. m. , portegies zwart , s. .f . ; bastian , n , & icke , v. 2012 , mnras , 421 , 1927 kruijssen , 2012 , ( in preparation ) lada c. j. , lada e. a. , 2003 , ara&a , 41 , 57 lamers , h. j. g. l. m. , baumgardt , h. , & gieles , m. 2010 , mnras , 409 , 305 lauer , t. r. , et al .
1992 , aj , 110 , 2622 lauer , t. r. , faber , s. m. , ajhar , e. a. , grillmair , c. j. , & scowen , p. a. 1998 , aj , 116 , 2263 lauer , t. r. et al .
2002 , aj , 124 , 197 launhardt , r. , zylka , r. , & mezger , p. g. 2002 , a&a , 384 , 112 leigh , n. , bker , t. , knigge , c. arxiv macchetto , f. , marconi , a. , axon , d. , capetti , a. , sparks , w. , & crane , p. 1997 , apj , 489 , 579 matthews , l. d. , & gallagher , j. s. iii 1997 , aj , 114 , 1899 matthews , l. d. , et al .
1999 , aj , 118 , 208 mclaughlin , d. e. , king , a. r. , & nayakshin , s. 2006 , , 650 , l37 mcmillan , p. j. , & dehnen , w. 2007 , mnras , 378 , 541 mcquillin , r. & mclaughlin , d. 2012 , mnras , 423 , 2162 merritt , d. , & cruz , f. 2001 , apj , 551 , l41 merritt , d. , piatek , s. , portegies zwart , s. , hemsendorf , m. 2004 , apj , 608 , l25 merritt , d. 2009 , apj , 694 , 959 merritt , d. , 2010 , apj , 718 , 739 merritt , d. , alexander , t. , mikkola , s. , & will , c. 2011,phrvd , 84 , 4024 milosavljevi , m. 2004 , apj , 605 , l13 milosavljevi , m. & merritt , d. 2001 , apj , 563 , 34 nayakshin , s. , wilkinson , m. i. , king , a. 2009 , mnras , 398l , 54 nadeen , s. et al . 2012 , arxiv1203.2625 neumayer , n. , & walcher , c. j. 2012 , adv .
astron . , 2012 , 709038 paumard , t. 2006 , apj , 643 , 1011 peng , e. w. , 2008 , apj , 681 , 197 preto , m. , merritt , d. , spurzem , r. 2004 , apj , 613 , l109 pryor c. , kormendy j. , 1990 , aj , 100 , 127 pfuhl , o. et al . 2011 ,
apj , 741 , 108 read , j. i. ; , goerdt , t. , moore , b. , pontzen , a. p. ; stadel , j. , lake , g. 2006 , mnras , 373,1451 rosenberg , a. , saviane , i. , piotto , g. , & aparicio , a. 1999 aj , 118 , 2306 rossa , j. et al .
2006 , aj , 132 , 1074 seth a. c. , dalcanton j. j. , hodge p. w. , debattista v. p. , 2006
, aj , 132 , 2539 seth , a. et al . 2008 ,
apj , 678 , 116 schinnerer , e. et al .
2006 , apj , 649 , 181 schinnerer e. , bker t. , meier d. s. , & calzetti d. 2008 , apjl , 684 , l21 scott , n. , & graham , a. w. 2012 , arxiv1205.5338 shlosman i. , & begelman m. c. 1989 , apj , 341 , 685 schdel , r. , eckart , a. , alexander , t. , et al .
2007 , a&a , 469 , 125 schdel , r. , merritt , d. , & eckart , a. 2008 , journal of physics conference series , 131 , 012044 schdel , r. , merritt , d. , & eckart 2009 a&a , 502 , 91 spitzer , l. 1987 , dynamical evolution of globular clusters ( princeton : princeton univ . press ) spergel , d. n. , et al .
2003 , apjs , 148 , 175 terzi , b. , & graham , a. w. 2005 , mnras , 362 , 197 tremaine , s. d. , ostriker , j. p. , & spitzer , l. , jr .
1975 , apj , 196 , 407 tremaine , s. , & weinberg , m. d. 1984 , mnras , 209 , 729 trippe et al .
2008 , a&a , 492 , 419 turner et al . , 2012 ,
apjs , 203 , 5 van den bergh , s. , aj , 1986 , 91 , 271 vesperini , e. , 1997 , mnras , 287 , 915 vesperini , e. , 1998 , mnras , 229 , 1019 | the centers of stellar spheroids less luminous than @xmath0 are often marked by the presence of nucleated central regions , called `` nuclear star clusters '' ( nscs ) .
the origin of nscs is still unclear .
here we investigate the possibility that nscs originate from the migration and merger of stellar clusters at the center of galaxies where a massive black hole ( mbh ) may sit .
we show that the observed scaling relation between nsc masses and the velocity dispersion of their host spheroids can not be reconciled with a purely `` in - situ '' dissipative formation scenario . on the other hand
, the observed relation appears to be in agreement with the predictions of the cluster merger model .
a dissipationless formation model also reproduces the observed relation between the size of nscs and their total luminosity , @xmath1 .
when a mbh is included at the center of the galaxy , such dependence becomes substantially weaker than the observed correlation , since the size of the nsc is mainly determined by the fixed tidal field of the mbh .
we evolve through dynamical friction a population of stellar clusters in a model of a galactic bulge taking into account dynamical dissolution due to two body relaxation , starting from a power - law cluster initial mass function ( cimf ) and adopting an initial total mass in stellar clusters consistent with the present - day cluster formation efficiency of the milky way ( mw ) .
the most massive clusters reach the center of the galaxy and merge to form a compact nucleus ; after @xmath2 years , the resulting nsc has properties that are consistent with the observed distribution of stars in the mw nsc .
when a mbh is included at the center of a galaxy , globular clusters are tidally disrupted during inspiral , resulting in nscs with lower densities than those of nscs forming in galaxies with no mbhs .
we suggest this as a possible explanation for the lack of nscs in galaxies containing mbhs more massive than @xmath3 .
finally , we investigate the orbital evolution of globular clusters in giant elliptical galaxies which are believed to always host a mbh at their center rather than a nsc . in these systems
an additional mechanism can prevent a nsc from forming : the time for globular clusters to reach the center of the galaxy is much longer than the hubble time . |
in einstein - maxwell ( em ) theory in @xmath1 dimensions the unique family of stationary asymptotically flat black holes with non - degenerate horizon comprises the rotating kerr - newman ( kn ) and kerr black holes and the static reissner - nordstrm ( rn ) and schwarzschild black holes .
em black holes are uniquely determined by their mass , their electric and magnetic charge , and their angular momentum ( see @xcite and references therein ) .
the generalization of black hole solutions to higher dimensions was pioneered by tangherlini @xcite for static vacuum black holes , and by myers and perry ( mp ) @xcite for stationary vacuum black holes . whereas myers and perry @xcite also obtained charged static black holes in higher dimensional em theory , higher dimensional charged rotating black holes
have not yet been obtained in closed form in pure em theory , although such black holes are known for some low energy effective actions related to string theory ( including additional fields , though ) @xcite .
higher dimensional black holes received much interest in recent years also with the advent of brane - world theories , raising the possibility of direct observation in future high energy colliders @xcite , which makes the theoretical study of these higher dimensional objects particularly appealing .
however , finding black holes solutions in higher dimensions is a difficult task due to the size and complexity of the equations .
moreover , contrary to what happens in @xmath0 the topology of the horizon is not unique in higher dimensions ( for instance , ringlike configurations are allowed @xcite ) , which makes the problem even harder . for these reasons
the number of solutions known analytically in closed form is very limited .
this forces us to use alternative techniques to study these black holes . here
we will briefly analyze such black holes by two methods : the numerical method and the perturbative method .
we will concentrate on einstein - maxwell - dilaton ( emd ) theory in @xmath2 dimensions , as an example how these methods may be applied .
the classical emd action reads @xmath3 where @xmath4 is the scalar curvature , @xmath5 is the dilaton field , @xmath6 is the electromagnetic field tensor , and @xmath7 is the electromagnetic potential .
@xmath8 is the dilaton coupling constant .
the equations of motion can be obtained by varying the action with respect to the metric @xmath9 , the dilaton field @xmath10 and the gauge potential @xmath7 , yielding @xmath11 \ , \label{fe1}\ ] ] @xmath12 for @xmath13 the theory reduces to pure em theory .
these equations admit an analytical solution when the dilaton coupling constant @xmath8 takes the kaluza - klein ( kk ) value @xmath14 @xcite @xmath15 in order to generate charged rotating solutions for generic values of @xmath8 we will assume that the topology of the horizon is spherical .
we will further assume that the black holes are stationary and axisymmetric .
stationary axisymmetric black holes in @xmath2 dimensions possess @xmath16 $ ] independent angular momenta @xmath17 associated with @xmath18 orthogonal planes of rotation @xcite .
( @xmath19 $ ] denotes the integer part of @xmath20 , corresponding to the rank of the rotation group @xmath21 . ) these black hole solutions then fall into two classes : even-@xmath2 and odd-@xmath2-solutions . when all @xmath18 angular momenta have equal - magnitude , the symmetry of these black holes is strongly enhanced .
in fact , in odd dimensions , the symmetry then increases from @xmath22 to @xmath23 , thus changing the problem from cohomogeneity-@xmath18 to cohomogeneity-1 . as a consequence
the original system of partial differential equations in @xmath18 variables then reduces to a system of ordinary differential equations ( ode s ) .
in what follows we will focus on the case where @xmath2 is odd , all angular momenta have the same magnitude , and the topology of the horizon is that of a sphere . that case is tractable numerically and perturbatively . under those assumptions
the metric , the gauge potential and the dilaton field can be parametrized with their angular dependence explicitly known @xmath24 + \frac{n}{f } r^2 \sum_{k=1}^n \left ( \prod_{l=0}^{k-1 } \cos^2 \theta_l \right ) \sin^2\theta_k \left({\varepsilon}_k d{\varphi}_k - \frac{\omega}{r } dt\right)^2 \nonumber \\ & & + \frac{m - n}{f } r^2 \left\ { \sum_{k=1}^n \left ( \prod_{l=0}^{k-1 } \cos^2 \theta_l \right ) \sin^2\theta_k d{\varphi}_k^2 \right . -\left .
\left[\sum_{k=1}^n \left ( \prod_{l=0}^{k-1 } \cos^2 \theta_l \right ) \sin^2\theta_k { \varepsilon}_k d{\varphi}_k\right]^2 \right\ } -f dt^2 \ , \label{metric}\end{aligned}\ ] ] @xmath25 where @xmath26 , @xmath27 $ ] for @xmath28 , @xmath29 , @xmath30 $ ] for @xmath31 , and @xmath32 denotes the sense of rotation in the @xmath33-th orthogonal plane of rotation .
the spacetime possesses @xmath34 commuting killing vectors , @xmath35 , and @xmath36 , for @xmath37 .
this parametrization employs isotropic coordinates and is appropriate for numerical purposes @xcite .
however , for perturbative purposes boyer - lindquist coordinates simplify the algebra @xcite . in order that the solutions correspond to black holes
they must possess an event horizon located at @xmath38 , characterized by the condition @xmath39 @xcite .
such an event horizon is in fact a killing horizon since the killing vector @xmath40 is null at and orthorgonal to the horizon , @xmath41 representing the constant horizon angular velocity .
the system of ode s resulting when eqs .
( [ metric]-[gauge_phi ] ) are substituted in eqs .
( [ fe1]-[fe3 ] ) must be supplemented with the appropriate set of boundary conditions .
this is obtained by imposing that the horizon is regular and the spacetime is asymptotically flat @xcite .
then the system of ode s is solved by employing a numerical solver ( e.g. , colsys ) , providing accurate results .
the numerical parameters are @xmath42 , with @xmath43 being the electric charge .
varying these parameters we obtain families of solutions , with their corresponding values of the physical properties like the mass @xmath44 , the angular momentum @xmath45 , etc .
an important information the numerical method provides is the domain of existence of black holes in higher dimensions .
that domain consists of the region in the parameter space bounded by the extremal solutions . in figure 1 ( left )
we exhibit the domain of existence of emd black holes in terms of the scaled electric charge @xmath46 and angular momentum @xmath47 .
clearly the domain depends both on the dilaton coupling constant @xmath8 and the dimension @xmath2 .
the accuracy of the numerical method may be tested by means of the analytical kk solutions , as well as by exact relations ( like the smarr mass formula ) these black holes satisfy @xcite . as a complementary method one may employ perturbations to generate black hole solutions in higher dimensions @xcite .
the advantage of this method is that it produces analytical formulae , which the numerical method does not .
however , the range of validity of these formulae is limited by the accuracy of the order of the perturbations .
a good perturbative parameter to produce charged rotating black hole solutions in higher dimensions is the electric charge .
one starts from the mp solution @xcite and expands the metric , the gauge potential and the dilaton in series expansions in the electric charge around the mp solution ( see @xcite for details ) .
those series are substituted in eqs .
( [ fe1]-[fe3 ] ) and solved order by order in the electric charge . for lower orders the equations are easy to solve but
as the order of the perturbations increases the equations become more and more difficult to solve .
an important point when applying the perturbative method is the way the integration constants are fixed .
one has to impose regularity of the solutions at the horizon together with the condition of asymptotic flatness .
the smarr mass formula is a useful means to check the accuracy of the solutions .
when the perturbative method is applied to extremal solutions the formulae simplify very much .
for instance , one obtains the following formula of the gyromagnetic factor @xmath48 to second order @xmath49 \left(\frac{q}{m}\right ) ^2 + o\left(\left(\frac{q}{m}\right)^2\right ) \ , \label{gfactor}\ ] ] showing that the gyromagnetic factor is not constant for pure em solutions ( @xmath13 ) .
although both methods have a different range of applicability there is a region of the parameter space where both are valid , namely , when the electric charge is small . in that region
one can check the consistency of the methods and their level of agreement @xcite .
this is presented in figure 1 ( right ) where the perturbative formula eq .
( [ gfactor ] ) for extremal solutions is compared to the corresponding non - perturbative numerical results for @xmath50 .
we observe that for @xmath51 both approches are in good agreement .
in fact , the perturbative method may be modified to analyze also slowly rotating solutions , using as the unperturbated initial solution the tangherlini solution and using the angular momentum as the perturbative parameter . to summarize , these two methods are complementary and provide us with a powerful tool to analyze black holes in higher dimensions when analytical solutions are not available . | in recent years higher - dimensional black holes have attracted much interest because of various developments in gravity and high energy physics .
but whereas higher - dimensional charged static ( tangherlini ) and uncharged rotating ( myers - perry ) black holes were found long ago , black hole solutions of einstein - maxwell theory , are not yet known in closed form in more than 4 dimensions , when both electric charge and rotation are present . here
we therefore study these solutions and those of einstein - maxwell - dilaton theory , by using numerical and perturbative methods , and by exploiting the existence of spacetime symmetries .
the properties of these black holes reveal new interesting features , not seen in @xmath0 . for instance , unlike the @xmath0 kerr - newman solution , they possess a non - constant gyromagnetic factor . |
authors thank professors d.v .
shirkov , a.e .
dorokhov , i.l .
solovtsov , and n. stefanis for the stimulating discussions .
thanks the organizers of qcd 04 for their hospitality .
the work has been supported by grants rfbr ( 02 - 01 - 00601 , 04 - 02 - 81025 ) , ns-2339.2003.2 , and cicyt fpa20002 - 00612 .
nesterenko and j. papavassiliou , in preparation .
cornwall , phys .
d * 26 * , 1453 ( 1982 ) .
cornwall and j. papavassiliou , phys .
d * 40 * , 3474 ( 1989 ) .
j. papavassiliou and j.m .
cornwall , phys .
d * 44 * , 1285 ( 1991 ) .
aguilar , a. mihara and a.a .
natale , int .
j. mod .
a * 19 * , 249 ( 2004 ) .
b. haeri and m.b .
haeri , phys . rev .
d * 43 * , 3732 ( 1991 ) . | we study the dependence on the pion mass of the qcd effective charge by employing the dispersion relations for the adler @xmath0 function .
this new massive analytic running coupling is compared to the effective coupling saturated by the dynamically generated gluon mass .
a qualitative picture of the possible impact of the former coupling on the chiral symmetry breaking is presented .
-2.5 mm the basic idea behind the analytic approach to quantum field theory is to supplement the perturbative treatment of the renormalization group ( rg ) formalism with the nonperturbative information encoded in the corresponding dispersion relations @xcite .
the latter , being based on the `` first principles '' of the theory , provide one with the definite analytic properties in the kinematic variable of a physical quantity at hand @xcite . in practice the analytization procedure @xcite amounts to the restoration of the correct analytic properties for a given quantity @xmath1 by imposing the representation @xmath2 here the spectral function @xmath3 can be defined by the initial ( perturbative ) expression for @xmath1 : @xmath4.\ ] ] a distinctive feature of the model for the qcd analytic invariant charge @xmath5
@xcite is the application of the analytization procedure ( [ defan ] ) to the perturbative expansion of the rg @xmath6 function : @xmath7^{j+1}\!\right\}_{\!\!\mbox{\scriptsize an}}\!\!.\ ] ] here @xmath8 denotes the @xmath9-loop perturbative running coupling , @xmath10 , and @xmath11 is the @xmath6 function expansion coefficient . at the one - loop level
the renormalization group equation ( [ anrgeq ] ) can be solved explicitly : @xmath12 where @xmath13 stands for the spacelike momentum .
the solution to eq .
( [ anrgeq ] ) can also be represented in the form of the integral @xmath14 where the one - loop spectral density is @xmath15 and the explicit expression for the @xmath9-loop @xmath16 can be found in ref .
@xcite .
the model ( [ aic1l ] ) shares all the advantages of the analytic approach : it contains no unphysical singularities at @xmath13 and possesses good higher loop and scheme stability .
besides , it has proved to be successful in description of hadron dynamics of the both perturbative and intrinsically nonperturbative nature .
it is worth noting that the massless analytic effective charge ( [ aichlkl ] ) incorporates the ultraviolet asymptotic freedom with the infrared enhancement ( i.e. , the singular behavior at @xmath17 ) in a single expression , see figure [ fig:1 ] . ) ) ( solid curve ) , @xmath18.,title="fig:",width=226 ] -7 mm -5 mm in this talk we will outline how the behavior of the running coupling ( [ aichlkl ] ) is affected by the pion mass entering the adler @xmath0 function , thus giving rise to an infrared finite value for this coupling .
we will also argue that this analytic effective charge may be relevant to the study of chiral symmetry breaking ( csb ) through the schwinger - dyson equations .
a certain insight into the nonperturbative aspects of the strong interaction can be provided by the adler @xmath0 function @xcite @xmath19 with @xmath20 being hadronic vacuum polarization function . in particular , eq .
( [ adlerdef ] ) is related to the measurable ratio @xmath21 of the @xmath22 annihilation into hadrons through the dispersion relation @xcite @xmath23 in turn , this equation implies the definite analytic properties in the @xmath24 variable for @xmath25 : it is an analytic function in the complex @xmath24-plane with the only cut beginning at the two pion threshold @xmath26 along the negative semiaxis of real @xmath24 . in the ultraviolet domain
the adler @xmath0 function is usually computed in the framework of perturbation theory : @xmath27,\ ] ] where @xmath28 is the charge of the @xmath29-th quark , @xmath30 + d_2\left[\frac{{\alpha{^{}_{\mbox{\scriptsize s}}}}(q^2)}{\pi}\right]^2 + \ldots,\end{aligned}\ ] ] @xmath31 , @xmath32 , and @xmath33 is the number of active quarks
. however , such approximation of eq .
( [ adlerpert ] ) violates the analyticity condition that @xmath25 must satisfy , due to the spurious singularities of the perturbative running coupling @xmath34 .
nevertheless , this difficulty , which is an artifact of the perturbative treatment , can be eliminated by imposing the analyticity requirement of the form @xmath35 on the right hand - side of eq .
( [ adlerpert ] ) .
therefore , the qcd effective charge itself has to satisfy the integral representation @xmath36 where @xmath16 is the @xmath9-loop spectral density @xcite .
the behavior of the one - loop massive analytic charge ( [ aicm ] ) in the spacelike and timelike infrared domains is shown in figure [ fig:2 ] .
the value of @xmath37mev is found by making use of the experimental data on the inclusive @xmath38 lepton decay @xcite .
it is worth noting that the nonvanishing pion mass drastically affects the infrared behavior of the analytic coupling in hand : instead of the enhancement in the massless case ( [ aichlkl ] ) one has here a finite infrared limiting value of the effective charge ( [ aicm ] ) ( see also refs .
@xcite for the details ) . ) in the spacelike and timelike domains .
the values of parameters are : @xmath39 active quarks , @xmath40mev.,title="fig:",width=226 ] -7 mm -5 mm based on the study of the _ gauge invariant _ schwinger - dyson equations , cornwall proposed a long time ago that the self - interactions of gluons give rise to a dynamical gluon mass , while preserving at the same time the local gauge symmetry of the theory @xcite .
this gluon `` mass '' is not a directly measurable quantity , but must be related to other physical parameters , such as the glueball spectrum , the energy needed to pop two gluons out of the vacuum , the qcd string tension , or the qcd vacuum energy .
one of the main phenomenological implications of this analysis is that the presence of the gluon mass @xmath41 saturates the running of the strong coupling , forcing it to `` freeze '' in the infrared domain .
in particular , the nonperturbative effective coupling obtained in ref .
@xcite is given by @xmath42},\ ] ] where @xmath43 is the dynamical gluon mass @xmath44^{-12/11}.\ ] ] the coupling ( [ jmc ] ) has the infrared finite limiting value @xmath45^{-1}$ ] . for a typical values of @xmath46mev and @xmath47mev ,
one obtains for the case of pure gluodynamics ( @xmath48 ) an estimation @xmath49 .
an independent analysis @xcite yields a maximum allowed value for @xmath50 of about 0.6 .
the incorporation of fermions into the effective charge @xcite @xmath51 does not change the picture qualitatively ( at least for quark masses of the order of @xmath52 ) . in equation ( [ jprc ] ) @xmath53 , @xmath54 , @xmath55mev stands for the gluon mass and @xmath56mev is a light quark constituent mass . the effective coupling of eq .
( [ jmc ] ) was the focal point of extensive scrutiny , and has been demonstrated to furnish a unified description of a wide variety of the low energy qcd data @xcite .
however , an important unresolved question in this context is the incorporation of the qcd effective charge into the standard schwinger - dyson equation governing the dynamics of the quark propagator @xmath57 @xmath58 specifically , since qcd is not a fixed point theory , the usual qed - inspired gap equation must be modified , in order to incorporate the running charge and asymptotic freedom .
the usual way of accomplishing this eventually boils down to the replacement @xmath59 in the corresponding kernel of the gap equation , where @xmath60 is the qcd running coupling .
the inclusion of @xmath60 is essential for arriving at an integral equation for @xmath57 which is well - behaved in the ultraviolet .
however , since the perturbative form of @xmath60 diverges at low energies as @xmath61 when @xmath62 , some form of the infrared regularization for @xmath60 is needed , whose details depend on the specific assumptions one is making regarding the nonperturbative hadron dynamics . at this point
the issue of the critical coupling makes its appearance . specifically , as is well - known
, there is a critical infrared limiting value of the running coupling , to be denoted by @xmath63 , below which there are no nontrivial solutions to the resulting gap equation , i.e. , there is no csb , see fig .
[ fig:3 ] . on the limiting value @xmath64 , @xmath65.,title="fig:",width=226 ] -7 mm -5 mm the incorporation of the effective charge of eq .
( [ jmc ] ) into a gap equation has been studied for the first time in ref .
@xcite .
there it was concluded that csb solutions for @xmath66 could be obtained only for unnaturally small values of the gluon mass , namely @xmath67 .
this is so because the typical value of @xmath63 found in the standard treatment of the gap equation is @xmath68 , which is what the expression for @xmath50 yields for the above value of @xmath69 .
this issue was further investigated in ref .
@xcite , where a system of coupled gap and vertex equations was considered .
the upshot of this study was that no consistent solutions to the system of integral equations could be found , due to the fact that the allowed values for @xmath64 , dictated by the vertex equation , were significantly lower than @xmath50 , i.e. , not large enough to trigger chiral symmetry breaking . ) , solid curves ) with the effective charge ( [ jprc ] ) ( dashed curves ) .
the values of parameters are : @xmath39 active quarks , @xmath70mev ( a ) , @xmath71mev ( b ) , gluon mass @xmath72mev and @xmath73mev ( c ) , gluon mass @xmath74mev and @xmath75mev ( d).,title="fig:",width=226 ] -7 mm -7 mm in what follows we will suggest a possible resolution of this problem .
the basic observation is captured in figure [ fig:4 ] : the effective charge with a gluon mass ( dashed curves ) and the analytic charge ( [ aicm ] ) ( solid curves ) coincide for a large range of momenta , and they only begin to differ appreciably in the deep infrared domain ( @xmath76 ) . in this region the analytic charge ( [ aicm ] ) rises abruptly , almost doubling its size between @xmath77 and @xmath78 , whereas the running coupling ( [ jprc ] ) in the same momentum interval remains essentially fixed to a value of about 0.6 .
a possible picture that seems to emerge from this observation is the following .
it may be that the concept of the dynamically generated gluon mass fails to capture all the relevant dynamics in the very deep infrared , where confinement or other nonperturbative effects make their appearance . at that point
it may be preferable to switch to a description in terms of the analytic charge ( [ aicm ] ) , which ( i ) coincides with that of cornwall in the region where the latter furnishes a successful description of data , and ( ii ) in addition , because it overcomes the critical value @xmath63 , offers the possibility of accounting for csb at the level of gap equations .
it would be interesting to carry out a detailed study of the gap equation , with the analytic charged plugged into , in order to verify if indeed one encounters nontrivial solutions , whose size is phenomenologically relevant , and if a reasonable value of the pion - decay constant @xmath79 may be obtained . |
low - redshift quasars are often surrounded by massive ionized nebulae showing filamentary structures on scales of a few tens of kpcs . more specifically , about 40% of the quasars at @xmath6 that are also steep - spectrum ( i.e. , usually frii type ) radio sources show such extended emission - line regions ( eelrs ) with an [ oiii]@xmath7 luminosity greater than @xmath8 erg s@xmath9 . here
we will focus our discussion on these steep - radio - spectrum quasars , because ( 1 ) they are possibly the counterparts of frii radio galaxies like 3c79 , and ( 2 ) the presence of a powerful radio jet seems to be a necessary ( though not sufficient ) condition for producing a luminous eelr , as implied by the correlation between radio spectral index and extended optical emission @xcite .
an eelr typically has a mass of @xmath10 @xmath4 , displays globally disordered kinematics , and shows a complex morphology that bears no obvious relationships ether with the host galaxy or with the extended radio structure ( @xcite ; see @xcite for a review ) . in terms of quasar luminosity , host galaxy luminosity , and black hole mass ,
the quasars that show eelrs ( hereafter the eelr quasars " ) do not differ significantly from non - eelr quasars .
but the former are all low - metallicity quasars .
the gas metallicity in their broad - line regions ( blrs ) is significantly lower ( @xmath11 0.6 @xmath2 ) than that of non - eelr quasars ( @xmath12 @xmath2 ) @xcite .
although this gas - phase metallicity is unexpectedly low for galaxies with the masses found for these quasar hosts , it is consistent with that in their eelrs . combining all of the pieces together ,
the most likely scenario for the origin of an eelr is that it comprises gas that is native to a gas - rich galaxy that merged with the quasar host and triggered the quasar activity , after which a large fraction of the gas was impulsively swept out by a large - solid - angle blast wave ( i.e. , a superwind ) , in a manner similar to that envisioned for quasar - mode feedback in the early universe ( e.g.,@xcite ) .
3c79 ( @xmath13 , 1 = 4 kpc ) km s@xmath9 mpc@xmath9 , @xmath14 , and @xmath15 is a narrow - line radio galaxy having frii radio jets with a central component @xcite . _
hubble space telescope _
( _ hst _ ) wfpc2 broad - band images show a complex optical morphology a bright elliptical galaxy with an effective radius of 7.5 kpc , accompanied by multiple tidal arms " and two distinct cores " other than the nucleus @xcite .
as we will show in [ sec : morph ] , only the galaxy and the core " southwest of the nucleus are true continuum sources ; the rest are all line - emitting regions .
the nuclear spectra of 3c79 show stellar absorption lines and a red continuum , indicating an old stellar population @xcite .
a line - emitting region in a curving filament extending 12 to the northwest of the galaxy was seen in an h@xmath16 image and was subsequently confirmed by a slit spectrum covering the [ oiii]@xmath17 region @xcite .
the spectrum also shows emission - line clouds between the filament and the nucleus , as well as clouds extending 5 to the southeast . an [ oiii]@xmath7 image taken in much better seeing @xcite shows a very rich morphology and an emission - line luminosity comparable to that of the eelrs around steep - spectrum radio - loud quasars , establishing that 3c79 is associated with an eelr . in line with the orientation - based unification schemes of frii radio galaxies and radio - loud quasars @xcite , we believe that 3c79 belongs to the same class of object as the eelr quasars , such as 3c249.1 and 4c37.43 , but it is viewed at a different angle with respect to the molecular torus surrounding the central engine .
if this is true , then the unique geometry of radio galaxies like 3c79 could afford us a natural coronograph blocking the blinding glare from the central engine .
such objects can be especially useful for studies of their host galaxies , which potentially preserve a large amount of information regarding the forces that have sculpted these spectacular emission - line regions . in this paper
, we study the host galaxy and the eelr of 3c79 with extensive ground - based spectroscopy in combination with a re - analysis of archival _ hst _
wfpc2 multi - band images .
first , we briefly describe our observations and data reduction procedures in [ sec : obs ] . in [ sec : host ] we study the morphology , stellar kinematics and stellar populations of the host galaxy .
we then demonstrate the similarities between the 3c79 eelr and the ones around quasars , in terms of gas kinematics , pressure , ionization mechanism , and metallicity in [ sec : eelr ] .
finally , we discuss our main results in
[ sec : dis ] and close with a summary in
[ sec : sum ] . 0.6 in 0.1 in
we observed 3c79 with the integral field unit ( ifu ; @xcite ) of the gemini multiobject spectrograph ( gmos ; @xcite ) on the gemini north telescope .
the observations were performed on the night of 2006 december 21 ( ut ) .
three exposures of 2880 s were obtained using the half - field mode with the b600/g5303 grating at a central wavelength of 6242 . with this setting
, we had a field - of - view ( fov ) of @xmath18 , a wavelength range of 4100 to 6900 , a dispersion of 0.46 pixel@xmath9 and an instrumental full width at half - maximum ( fwhm ) of 4.5 pixels ( 2.1 ) .
the host galaxy was placed at the lower - left corner of the ifu field so that a large fraction of the western part of the eelr could be covered ( figure [ fig : obs ] ) .
feige34 was observed for flux calibration .
the data were reduced using the gemini iraf package ( version 1.8 ) .
the data reduction pipeline ( gfreduce ) consists of the following standard steps : bias subtraction , spectral extraction , flat - fielding , wavelength calibration , sky subtraction , and flux calibration .
cosmic ray rejection was performed by l.a.cosmic @xcite before running the data through the reduction pipeline , with careful adjustment of the parameters to avoid misidentification of real data .
spectra from different exposures were assembled and resampled to construct individual data cubes ( x , y,@xmath19 ) with a pixel size of 005 ( gfcube ) . to correct for the differential atmosphere refraction ( dar ) ,
we first binned the data cubes along the wavelength direction to increase the s / n ratio of the host galaxy ; then the centroid of the galaxy was measured for each bin , and a polynomial was fit to the ( x , y ) coordinates , restricted by the fact that dar only causes the centroids to vary along a straight line in ( x , y ) plane ( at the parallactic angle ) ; finally we shifted the image slices at each wavelength using the solution from the polynomial fit .
the three data cubes were then combined and binned to 02 pixels , the original spatial sampling of the ifu fiber - lenslet system , to form the final data cube .
we obtained optical long - slit spectroscopy of 3c79 on the night of 2007 october 17 ( ut ) with the low resolution imaging spectrograph @xcite on the keck i telescope .
the cassegrain atmospheric dispersion compensator ( cass adc ) was used during the observation .
we took two 1200 s exposures through a 1 slit centered on the nucleus at a position angle of @xmath20 ( fig .
[ fig : obs ] ) . on the blue arm we used the 600 groove
mm@xmath9 grism blazed at 4000 , whilst on the red arm we used the 400 groove mm@xmath9 grating blazed at 8500 , offering , respectively , wavelength ranges of 3100@xmath215600 and 5100@xmath218900 and spectral resolutions of 4 and 7 ( fwhm ) .
the spectra were taken at low airmass ( @xmath221.1 ) , and the seeing was about 07 throughout the night .
the spectra were reduced in the standard fashion using iraf tasks .
the blue - arm spectra were wavelength - calibrated using observations of arc lamps , whilst for the red - arm spectra we used night - sky lines .
background sky was removed by fitting low - order cubic splines to the regions on each side of the object along the slit direction using background .
one - dimensional spectra of 3c79 were first extracted with a wide ( 11 ) aperture , and they were flux - calibrated and corrected for atmospheric absorption using the spectrophotometric standard wolf1346 ( taken 5 min after the nautical twilight earlier in the night ) .
the calibrated spectra from the two arms agree perfectly in their overlapping region .
then , one - dimensional spectra of high s / n ratio were obtained by using a 2 extraction aperture .
these spectra were multiplied by appropriate smooth curves so that their continua match those of the wide - aperture spectra . finally , the blue and the red sides , after flux calibration and atmospheric absorption correction , were joined and binned to a common linear dispersion ( 1.86 per pixel ) using scombine .
the line - of - sight galactic reddening ( @xmath23 , @xcite ) was corrected with a standard reddening curve @xcite .
we obtained an absolute flux calibration by scaling the spectrum to the _ hst _ photometry result from [ sec : morph ] ( @xmath24 erg @xmath25 s@xmath9 @xmath9 between 6300 and 7300 ) . to study the host galaxy morphology , we obtained wfpc2/wfc f675w ( 4@xmath26565 s ) and fr680n ( 2@xmath261300 s ) images , and a wfpc2/pc f702w ( 2@xmath26140 s ) image of 3c79 , available from the archive of the _ hst_. the two broad - band images have been previously presented by @xcite and @xcite . the linear ramp filter ( lrf ; fr680n ) image has a central wavelength of 6284 and a bandpass fwhm of 82 at the position of 3c79 , i.e. , the filter is centered on the redshifted [ oiii]@xmath195007 emission line .
images observed with lrfs are not flat - field calibrated in the calibration pipeline , so we flattened the lrf image with an f631n flat field reference image .
additional cosmic rays after the standard pipeline reduction were identified and replaced by l.a.cosmic @xcite , again with careful adjustment of the parameters to avoid misidentification of real data .
though the wfc2 f675w image is dominated by continuum radiation , h@xmath27 and [ oiii ] @xmath284959,5007 lines fall on the shortward wing of the filter profile , where the transmissions are about 42 , 76 , and 85% , respectively .
to study the morphology of the host galaxy , one has to remove the line - emitting regions from the image .
we thus rotated the wfc3 lrf [ oiii ] image to the orientation of the f675w image , shifted and scaled the image to match the f675w image , and finally subtracted it off the wide - band image to obtain an emission - line free image ( hereafter the `` f675w@xmath29 '' image ; fig .
[ fig : subdzl ] ) . a close
companion 08 southwest of the major host galaxy is seen in the f675w@xmath29 image , which we designate as 3c79a .
the host galaxy itself is simply called 3c79 in this section .
3c79a is better sampled in the pc1 f702w image , so we used this image to study its morphology .
we used c. y. peng s galfit software @xcite to fit the two - dimensional galaxy profile , and we used an oversampled tiny tim psf . the result is shown in fig . [ fig:3c79a ] .
as indicated by the best fit srsic index of @xmath30 , the profile is close to an exponential .
as shown by the residuals , the exponential model clearly provides a better fit than the de vaucouleurs model .
we chose to use the f675w@xmath29 image to study the morphology of 3c79 .
we converted the geometric parameters from the best fit exponential model of 3c79a ( effective radius , axis ratio , and position angle [ pa ] ) to the wfc image , and froze them ; however , we allowed the position and magnitude of 3c79a vary in the modeling of 3c79 .
figure [ fig:3c79 ] shows that 3c79 can be described by a de vaucouleurs model , as indicated by the high srsic index .
table [ tab : morph ] summarizes the morphology results .
the pa of the minor axis of 3c79 is aligned within 5@xmath31 to the direction of the radio jets ( as defined by the hot spots ; see fig .
[ fig : obs ] ) . the companion galaxy 3c79a
is located 084 ( 3.3 kpc ) southwest of 3c79 , and its pa is aligned within 7@xmath31 to the direction of the host galaxy .
@xcite modeled the 3c79 host galaxy with a de vaucouleurs model ( plus a concentric psf to account for the narrow - line region [ nlr ] ) . for comparison ,
their results are also shown in table [ tab : morph ] .
@xcite used the original f675w image , i.e. , the emisson - line regions were not removed ; thus they overestimated the brightness of the galaxy and had a poor estimate for the axis ratio and pa .
we were able to reproduce their results using galfit when modeling the original f675w image .
we extracted the host spectrum from the final combined gmos data cube using a @xmath32 ( 2 kpc ) radius aperture .
although the gmos spectrum shows a poorer s / n ratio in comparison with the lris spectrum , it has a better spectral resolution ( 2.1 fwhm as opposed of 7 ) , making it best - suited for measuring stellar kinematics .
we chose to use the stellar synthesis models of @xcite to fit the data .
although these models have very limited spectral coverage compared with those of @xcite , they have a better spectral resolution ( @xmath22 1.8 ) , which is similar to that of the gmos spectrum , and they do cover the spectral region where the spectrum show the highest s / n ratio ( s / n @xmath22 33 at rest - frame wavelengths of 4800 @xmath33 5500 ) . for the modeling we used m. cappellari s idl programmcappell / idl/ ] implementing the pixel - fitting method of @xcite .
the program finds the best fit to the data by convolving the model templates with a line - of - sight velocity distribution ( losvd ) . since we used a gaussian function to parameterize the losvd , the program reports the mean stellar velocity ( @xmath34 ) and velocity dispersion ( @xmath35 ) . in figure [ fig : skin ] we plot our best fit model against the data .
following @xcite , we included a multiplicative legendre polynomial of degree 6 in the fit to correct the model continuum shape .
the kinematics results are insensitive to the assumed model age , although @xmath36 increases rapidly for models younger than 3 gyr and the oldest population yielded the minimum @xmath36 . using a monte - carlo approach , we found that the 1 @xmath37 uncertainties of @xmath34 and @xmath35 are both about 11 km s@xmath9 .
these errors should be regarded as lower limits since they do not account for the effect of the template and continuum mismatch . assuming a negligible amount of systematic rotation , the measured velocity dispersion of @xmath22236 km s@xmath9implies a virial mass of @xmath38 @xmath4within 2 kpc from the galaxy center .
we estimated the luminosity - weighted velocity dispersion within @xmath39 ( 7.2 kpc ) to be @xmath40 km s@xmath9 , with the aperture correction function for sauron elliptical galaxies , @xmath41 @xcite .
the virial mass inside @xmath39 is then @xmath42 @xmath4 . to study the stellar population of the host galaxy
, we have performed detailed continuum modeling to the 3c79 nuclear spectrum obtained from lris .
the host galaxy shows a strong uv excess ( fig .
[ fig : spop ] ) and a broad ( fwhm @xmath22 6500 km s@xmath9 ) mgii @xmath192798 line ( fig .
[ fig : mg2 ] ) , but no broad h@xmath16 or h@xmath27 lines are seen .
these characteristics seem to resemble closely the radio galaxy cygnus a @xcite , implying a significant scattered quasar component .
in addition , stellar absorption features are evident ( caii k @xmath193933 , the g band @xmath194300 , and the mgi _ b _ @xmath195175 , etc . ) , implying a dominating old stellar population ( osp ) . as a first guess , we tried to model the continuum with three components : ( 1 ) a nebular continuum ( at @xmath43 = 10,000 , 15,000 , and 20,000 k ) , ( 2 ) a power law ( @xmath44 in accordance with the composite uv / optical quasar spectrum ; @xcite ) , and ( 3 ) an osp ( 7 gyr @xmath45 age @xmath45 10 gyr = 0.256 the age of the universe is 10.4 gyr . ] , and @xmath46 = 0.2 , 0.4 , 1.0 , and 2.5 @xmath2 ; @xmath2 = 0.02 )
. the same idl program was used in the modeling as in the previous section , but this time we did _ not _ include polynomials to adjust the model continuum shape . since the nebular continuum scales with the balmer lines , we required that the h@xmath47 flux from the models to be within 5% of the measured flux .
figure [ fig : spop]_a _ shows the best fit three - component model ( i.e. , the model that produces the least @xmath36 ) , which is a combination of a 9-gyr - old 2.5 @xmath2 population , a nebular continuum at @xmath43 = 15,000 k , and a @xmath48 power law with @xmath49 .
this model fits both the continuum shape and most of the absorption features , however , it underpredicts the caii k line and the g band .
this inadequacy can be removed once we introduce a fourth component an intermediate - age stellar population ( 1 gyr @xmath45 age @xmath45 3 gyr , and @xmath46 = 0.2 , 0.4 , 1.0 , and 2.5 @xmath2 ) .
figure [ fig : spop]_b _ displays the best fit four - component model , which successfully reproduces the caii k line and the g band absorption .
the global best fit to the data is given by stellar population models with a metallicity of 2.5 @xmath2 . in order to gauge
how robust this metallicity result is , we compared the best fit four - component models of different metallicities .
the best fit 1.0 @xmath2 model gives a generally good fit to both the absorption features and the overall continuum shape , except that it can not reproduce the high s / n continuum shape around the mgi _ b _ feature between 5000 and 5500 (rest - frame ) , resulting in a 16% increase in the @xmath36 value relative to the 2.5 @xmath2 model .
models with sub - solar metallicities all fail to give an adequate fit , especially in the continuum below 4300 ( rest - frame ) .
we thus conclude that the stellar population in the host galaxy has a super - solar metallicity . a conservative range of metallcities that can give an adequate fit to the lris spectrum is between 1 and 2.5 @xmath2 .
we have also attempted to use a young stellar population ( @xmath50 0.1 gyr ) to replace the power law component , but the best fit model underpredicts the higher ( @xmath51 h7 ) balmer lines if it correctly predicts the h@xmath47 flux ( due to the balmer absorption lines from the young population ) , and it gives a much poorer fit to the rest - frame uv continuum ( 2900 @xmath52 3800 ) , in particular , it overpredicts the balmer limit break at 3650 . if we adopt the average total stellar mass from the two best fit models shown in fig .
[ fig : spop ] , then the stellar mass inside @xmath53 is @xmath54 @xmath4 .
the @xmath55 ratio implies a dark matter fraction of @xmath56% within one effective radius , in agreement with the dark matter fraction of nearby early - type galaxies @xcite .
we caution that there are significant uncertainties in both the stellar mass derived from single - burst stellar synthesis models and the dynamical mass using the virial theorem .
hence , this straightforward calculation of dark matter content is subject to an error of at least a factor of two .
we note that the absolute @xmath0-band magnitude - band vega system ( @xmath210.7 mag ) , after applying @xmath57-correction ( @xmath210.27 ) and passive evolution correction ( + 0.23 ) , following @xcite .
] of the galaxy is about @xmath58 , which is the same as that of the host galaxies of eelr quasars @xcite .
the compact companion galaxy 3c79a is detected in the gmos data cube .
the comparison of the spectrum extracted from 3c79a and a background spectrum from a symmetric location with respect to the minor axis of the host galaxy shows a red continuum , indicating an old stellar population for 3c79a .
this galaxy is reminiscent of the close companion galaxy 1 away from the eelr quasar 3c48 @xcite , which is also elongated towards the agn nucleus .
the s / n ratio of the spectrum does not allow a detailed modeling .
the _ hst _ lrf [ oiii ] image ( fig .
[ fig : obs ] ) shows not only the filament seen in the h@xmath16 image of @xcite , but also luminous emission - line `` arms '' at distances around 25 from the nucleus , and an `` inner arc '' 1 to the west . to compare the eelr of 3c79 with those around quasars
, we determined the total luminosity in the [ oiii]@xmath7 line within an annulus of inner radius 11.2 kpc and outer radius 44.9 kpc km s@xmath9 mpc@xmath9 and @xmath59 . ]
( @xmath60}$ ] ) .
we found a luminosity of @xmath60}$ ] = @xmath61 erg s@xmath9 , close to that of the most luminous quasar eelr at @xmath6 , i.e.,that of 4c37.43 , @xmath60}$ ] = @xmath62 erg s@xmath9 ( converted from @xcite ) .
as a reference , the least luminous detected eelr in the radio - loud quasar subsample of @xcite has a luminosity about 10 times lower ( pks2135@xmath21147 , @xmath60}$ ] = @xmath63 erg s@xmath9 ) .
kinematics of the ionized gas can be measured from strong emission lines . since a single 48 min gmos exposure is enough to acquire a good s / n ratio in the [ oiii]@xmath195007 line region , we derived the velocity fields from the single data cube that has the smallest airmass ( am = 1.172 ) and the best seeing . to study the velocity structure for the clouds in the inner arc that is only 1 from the nuclear nlr , we subtracted the light from the nlr and the bright nearby cloud 05 east to the nlr by simultaneously fitting two moffat psf profiles at each wavelength in the wavelength range where both clouds are apparent . ] .
after trials it was found that the best subtraction results came from using the best fit moffat profile from modeling the feige34 data cube over the same wavelength range , indicating both data cubes have the same resolution ( fwhm @xmath64 06 at the [ oiii ] region ) .
we then fixed the moffat parameters to those of the standard star and let the positions and magnitudes vary .
finally , we fixed the positions to the average values across the wavelength range from the previous fit and performed the fit with only the magnitudes as free parameters .
the resulting psf - subtracted data cube is presented as a set of velocity channel maps in fig .
[ fig : cmap ] .
the velocity field is shown in fig .
[ fig : vfield ] .
the velocity dispersion measurements have been corrected for the @xmath65 km s@xmath9 instrumental resolution . due to the imperfect psf subtraction process , no useful kinematics can be extracted from the psf - subtracted data cube in the nuclear region bordered by the white no - data pixels in fig .
[ fig : vfield ] .
hence , we measured the velocity field in this region from the original data cube and inserted the results into the final velocity field .
most of the extended emission comes from two main filaments the inner arc and the outer arm 25 from the nucleus .
both filaments show similar velocity gradients , with the northern part approaching towards us and the southern part receding from us at velocities of @xmath66 km s@xmath9 .
together , these two filaments are consistent with a common disk rotation with a flat rotation curve .
this sort of locally ordered kinematics of the brighter clouds is reminiscent of that of the three most luminous clouds southeast of 3c249.1 @xcite .
however , the much fainter clouds outside of the outer arc disrupt this simple picture by showing large approaching velocities .
the velocity dispersion over most of the region is around 50 to 120 km s@xmath9 .
the region showing large @xmath37 ( @xmath67 km s@xmath9 ) between the inner arc and the outer arm , although roughly coincident with the jet direction , in fact also has multiple velocity components that happen to contribute light to the same pixels .
overall , the velocity field of the 3c79 eelr is similar to those of the quasar eelrs , in the sense that all of them appear globally disordered but locally ordered , and they all show super - sonic velocity dispersions ( sound speed @xmath68 km s@xmath9 ) within the same range between 50 and 130 km s@xmath9 @xcite . to increase the s / n ratio , we have combined gmos spectra within various regions .
the spectra of the clouds in the inner arc ( @xmath69 ) are heavily contaminated by light from the nuclear nlr , making it difficult to measure their line intensities accurately .
we therefore concentrated on the eelr clouds in the outer arm and beyond . in fig .
[ fig : cmap ] we have shown the extraction apertures for the four eelr clouds .
we have also extracted a nlr spectrum using a circular aperture with a radius of 03 . in order to measure the key metallicity diagnostic the [ nii]@xmath70/h@xmath16 ratio
we extracted lris spectra from three regions along the slit : a 2-wide aperture centered on the nucleus was used for the nuclear nlr , and the apertures used for the _
n1 _ and _ s1 _ regions are shown in fig . [
fig : obs ] . unlike those for the eelr clouds , nlr spectra should be corrected for stellar absorption features .
we fitted the continuum using the same stellar synthesis model templates as in [ sec : spop ] and subtracted off the best fit model to obtain the clean emission - line spectrum of the nlr .
note that the line ratios for the nlr should be reliable even if the continuum subtraction were just barely correct , given the large equivalent widths of the emission lines that are affected by strong absorption features .
intrinsic reddening due to dust in front of _ or _ associated with the cloud was determined from the measured h@xmath47/h@xmath27 ratio for gmos spectra and from h@xmath16/h@xmath27 ratio for lris spectra , assuming theoretical ratios of 0.468 for h@xmath47/h@xmath27 ( the value for case b recombination at @xmath71 k and @xmath72 @xmath73 ; @xcite ) and 3.1 for h@xmath16/h@xmath27 . for clouds
@xmath74 and @xmath75 , where the s / n ratio of the h@xmath47 line is poor , we assumed zero intrinsic reddening . for clouds
@xmath76 and @xmath77 , where the measured h@xmath16/h@xmath27 is slightly lower than the theoretical value , we also assumed zero intrinsic reddening .
both the intrinsic reddening and galactic reddening were corrected using a standard galactic reddening law @xcite .
the reddening - free emission - line fluxes and 3 @xmath37 upper limits are tabulated in table [ tab : flux ] .
the quoted 1 @xmath37 errors were derived from the covariance matrices associated with the gaussian model fits , with the noise level estimated from line - free regions on either side of an emission line . there is enough overlapping area to show that the data from the two instruments agree fairly well : ( 1 ) the low intrinsic reddening for cloud @xmath78 is consistent with the measured h@xmath16/h@xmath27 ratio of 3.09 in region @xmath76 ; ( 2 ) the gmos nlr spectrum suggests an intrinsic reddening of @xmath79 = 0.28 , whilst the lris nlr spectrum shows @xmath79 = 0.36 ; ( 3 ) the individual line ratios relative to h@xmath27 of regions @xmath78 and nlr ( gmos ) are very much consistent with those measured from the lris @xmath76 and nlr spectra , respectively .
the [ oii ] luminosity - weighted average electron density ( @xmath80 ) and electron temperature ( @xmath81 ) can be measured from the intensity ratio of the [ oii]@xmath82 doublet and the [ oiii](@xmath83)/@xmath84 ratio , respectively @xcite .
our gmos data cube covers all of these five lines , allowing us to measure the density and temperature in individual emission - line clouds .
the [ sii]@xmath85 doublet is covered by the lris spectra , however , they fall into the region where strong oh night sky lines dominate and severe ccd fringing occurs , preventing reliable measurements .
we thus only discuss results from the [ oii ] doublet .
the [ oii ] doublet is barely resolved in our spectra .
however , for the nlr and the eelr clouds @xmath78 & @xmath86 , the high s / n ratio profile of the doublet allows a meaningful decomposition once we constrain their expected wavelengths to be at the same redshift as that of the nearby [ neiii]@xmath87 line .
uncertainties of the line fluxes from this fit tend to be low as the wavelengths are no longer free parameters . lifting this wavelength constraint
gives a more conservative error estimate .
here we have adopted the mean values of the error estimates from the two fits one with and one without wavelength constraints . the weak [
oiii]@xmath84 line is detected in all the clouds except @xmath75 .
we then derived @xmath81 and @xmath80 consistently using the iraf routine temden for the three regions where both [ oii ] and [ oiii ] ratios are available .
table [ tab : prop ] summarizes the results . for the clouds in the eelr , the temperatures are about 12,000 to 14,000 k , and the densities are about 100@xmath21200 @xmath73 , similar to those of quasar eelrs that were measured using the same technique @xcite
. the emission - line ratios of the 3c79 clouds are very similar to those of the quasar eelrs .
previously we have concluded that the spectra of quasar eelrs are most consistent with being photoionized by quasar continuum , and are inconsistent with shock or the self - ionizing shock + precursor " models @xcite .
now we re - emphasize this point by comparing strong line ratios of the eelrs of the radio galaxy 3c79 and the quasars 3c249.1 and 4c37.43 with emission - line galaxies at a similar redshift range .
the sloan digital sky survey data release 4 ( sdss dr4 ) provides high - quality optical spectra for a large number of galaxies .
the emission - line fluxes after correcting for stellar absorption and foreground extinction are publicly available .
we first created a subsample of sdss emission - line galaxies by including only those with s / n @xmath12 3 in the strong emission - lines [ oii]@xmath88 ( = @xmath89 ) , h@xmath27 , [ oiii]@xmath7 , h@xmath16 , [ nii]@xmath70 and [ sii]@xmath85 .
then we divided the sample into star - forming galaxies and agn using the empirical dividing line of @xcite , and corrected for the intrinsic reddening using the balmer decrement and the standard galactic reddening curve @xcite .
the theoretical values for the h@xmath16/h@xmath27 ratio are 2.85 for star - forming galaxies and 3.1 for agn . finally , we classified the galaxies into seyferts , liners , star - forming galaxies and agn / starforming composites , following the classification schemes of @xcite .
figure [ fig : class ] compares the locations of our eelrs and the distributions of the sdss galaxies in two line - ratio diagnostic diagrams . as can be seen , the eelrs share the same territory as the seyfert galaxies , but are clearly distinguishable from star - forming and composite galaxies , indicating that the eelrs are photoionized by an agn - type ionizing spectrum . the line - ratios from the 3c79 eelr match those from 3c249.1 and 4c37.43 very well , thus pure shock and shock + precursor " models can be ruled out , since it has been shown ( using the same line - ratios ) that these models do not fit these quasar eelrs ( see fig . 5 of @xcite for an example ) . with the knowledge that the eelr of 3c79 is photoionized by the hidden quasar
, we can now estimate the metallicity of the gas using photoionization models . in the following
we will compare our data with the dusty radiation - pressure dominated photoionization models of @xcite with a density of 1000 @xmath73 .
since the diagnostic line ratios we will use are almost entirely dependent on the abundances , models adopting a different density will not change our results . as the metallicity of a photoionized nebula increases , the temperature of the nebula decreases ( as a result of the increased metal cooling ) , and the relative strengths between hydrogen lines and lines from other heavier elements also decreases ( due to the increased abundance of the heavy elements relative to hydrogen ) .
the latter effect is especially strong for nitrogen lines since the element is dominated by secondary " production at @xmath90 0.2 @xmath2 .
hence , n abundance increases at a much faster rate with metallicity , i.e. , n / h @xmath91 . in figure [ fig : z]_a _ we use the temperature sensitive [ oiii]@xmath92 ratio to demonstrate the dependance of nebula temperature on metallicity .
the stratification among the grids of different metallicities is further improved by the fact that helium increases linearly with metallicity relative to hydrogen @xcite . in fig . [
fig : z]_b _ and fig .
[ fig : n2 ] , we show the strong dependance of [ ni]@xmath93/h@xmath27 and [ nii]@xmath70/h@xmath16 on metallicity . at each metallicity
, we show models spanning a range of ionization parameters ( @xmath94 log(@xmath95 ) @xmath45 0 ) and four power - law indices from @xmath96 to @xmath97 , representing the quasar ionizing continuum ( @xmath98 ) .
we have converted the modeled total metallicities to gas - phase metallicities using the solar abundances defined by @xcite ( 12+log(o / h)@xmath99 = 8.93 and @xmath2 = 0.02 ) , in order to be consistent with both previous studies on the metallicity of quasar blrs and metallicity results from stellar populations ( [ sec : spop ] ) .
one solar metallicity in @xcite corresponds to 1/3 @xmath2 in our figures , since approximately half of the metals are assumed to be depleted onto dust in the models , and their assumed solar metallicity is about 0.2 dex0.2 dex lower than those in @xcite .
hence , the mass fraction of metals ( i.e. @xmath46 ) is about 1.5 times lower @xmath2 = 0.013 than that of @xcite . ] lower than that of @xcite . as can be seen in figs .
[ fig : z ] & [ fig : n2 ] , the line - ratios of the eelr as well as the nlr of 3c79 , like those quasar eelrs and their blrs , are most consistent with a low - metallicity at about 1/3 to 2/3 solar .
[ fig : n2 ] illustrates the peculiarity of 3c79 more clearly by comparing it directly with sdss agn on the same line - ratio diagnostic .
the nuclear nlr along with the two other regions where we have [ nii ] measurements all lie on the lower metallicity side of the main seyfert branch , overlapping with quasar eelrs .
this result supports the orientation - based unification schemes of frii radio galaxies and radio - loud quasars @xcite . on the other hand
, 3c79 offers yet another example strengthening ( 1 ) the correlation between the presence of eelrs and the metallicity of the nuclear gas ( blrs and/or nlrs ) , and ( 2 ) the similar metallicity in eelrs and the gas in the nuclear region .
both correlations were initially discovered only among steep - spectrum radio - loud quasars @xcite .
the great majority of quasars show super - solar metallicities in their nuclear regions ( see * ? ? ? * for a recent review ) ; yet , as we have previously shown @xcite , quasars with luminous eelrs are drawn exclusively from the subset of steep - radio - spectrum quasars that have blrs with sub - solar metallicities .
the fact that the quasar host galaxies are almost always very massive and thus expected to have high metallicities suggests an external source for the low - metallicity gas .
furthermore , the apparent link between the metallicity of the gas in the blrs and the metallicity of the gas in the eelrs , which have minimum masses of @xmath100 @xmath4 , means that the external source of the gas must itself have been moderately massive . in 3c79
, we have an frii radio galaxy that shows the same pattern .
although we of course can not obtain a broad - line - region metallicity in this case , we have used the unresolved nuclear narrow - line region metallicity as a surrogate . aside from providing a consistency check on unified models for frii radio galaxies and quasars , the main question of interest is whether the clearer view of the inner region of the host galaxy gives us any insight into the origin of the low - metallicity gas and the mechanism that produced the eelr .
the host galaxy of 3c79 appears to be a fairly normal elliptical morphologically .
the most intriguing feature in the inner region is the extremely compact galaxy 08 from the center of the host galaxy .
the nearly exponential profile of this galaxy suggests that it could be a tidally stripped `` pseudobulge '' of a late - type galaxy ( e.g. , * ? ? ?
* ) ; the interstellar medium of such a galaxy would be quite plausible as a source for the low - metallicity gas in 3c79 .
we have suggested elsewhere @xcite that luminous eelrs may result from nearly spherical blast waves connected with the initiation of frii radio jets .
the connection with low - metallicity gas is less clear , although we can speculate that it may have something to do with the lower radiative coupling such gas would have to the quasar radiation field , allowing more efficient accretion .
it is interesting to note that related considerations have been invoked recently to explain the link between low metallicity and long - duration gamma - ray bursts @xcite on the scale of massive stars , although the physical mechanisms involved are certainly different .
armed with the latest photoionization models , @xcite identified @xmath2240 ( out of @xmath101 ) candidates of low - metallicity seyfert2s in sdss . a caveat to note
is that the authors restricted the sample to galaxies with stellar masses lower than @xmath102 @xmath4 while selecting their candidates , as guided by the mass - metallicity correlation ( e.g. * ? ? ?
although the eelr quasars are excluded from the sdss emission - line galaxy sample , 3c79 represents a population of low - metallicity agn that were missed by @xcite these agn are hosted by massive evolved galaxies with @xmath103 @xmath4 of stars and harbor @xmath104 @xmath4 black holes at their hearts , and they are likely to be more powerful than the seyfert2s . by cross - correlating the texas radio galaxies @xcite with sdss dr4 agn , we have identified a sample of low - metallicity radio galaxies . like that of 3c79 ,
the spectra of their host galaxies show red slopes and absorption features that are indicative of an old stellar population .
follow - up high - resolution imaging and spatially resolved spectroscopy of these galaxies and a comparable control sample are needed to finally nail down the source of the low - metallicity gas and the triggering mechanism for quasar superwinds .
in addition , detecting eelrs around these objects would provide a new test for the unification schemes .
based on extensive ground - based spectroscopy and archival _ hst _
wfpc2 images of the radio galaxy 3c79 , we have conducted a detailed analysis of its host galaxy and the eelr .
the host galaxy of 3c79 is a massive elliptical with @xmath105 .
the uv / optical spectral energy distribution of the host galaxy is best described by a combination of an intermediate - age stellar population ( 1.3 gyr ) , an old stellar population ( 10 gyr ) , a power law , and a nebular thermal continuum .
both stellar populations are metal rich ( 2.5 @xmath2 ) .
this best fit model indicates a total stellar mass within one effective radius of approximately @xmath106 @xmath4 , consistent with the virial mass derived from stellar kinematics ( @xmath107 @xmath4 ) .
the eelr of 3c79 is remarkably similar to the most luminous quasar eelrs . the velocity field , although only available in a small area , is locally ordered but globally disordered .
the eelr is almost certainly photoionized by the hidden quasar , and it shows densities of @xmath108 @xmath73 in the [ oii]emitting regions and temperatures around 13,000 k. most interestingly , the metallicity of the gas in both the eelr and the nlr is about 1/3@xmath212/3 solar , matching perfectly the metallicity in both the eelrs and the nuclear blrs of the eelr quasars .
there is a very compact close companion galaxy 3.3 kpc from and 4 magnitudes fainter than the host galaxy .
this companion galaxy shows an exponential profile , and is presumably a tidally stripped pseudobulge " of a late - type galaxy that might be the source for the low - metallicity gas as well as the starforming regions in 3c79 .
the exact correspondence between this eelr and the eelrs around quasars joins the already overwhelming evidence in support of the unification schemes for frii radio galaxies and radio - loud quasars .
the definitive trait of sub - solar metallicity in the nlr of 3c79 also provides an efficient tool for selecting frii radio galaxies that likely host luminous eelrs .
we thank the gemini staff for carrying out the gmos observations and nicola bennert for helping us obtain the lris spectra .
we also thank the anonymous referee for a careful reading of the manuscript and for cogent comments to help us clarify and improve the presentation .
this research has been partially supported by nsf grant ast03 - 07335 .
lccccccc 3c79a & srsic & & 22.19 & 0.22 & 1.23 & 0.49 & 18.2 + & exp & 22.24 & 22.23 & 0.22 & ( 1 ) & 0.48 & 18.0 + 3c79 & srsic & 18.06 & & 10.1 & 5.21 & 0.70 & 6.2
+ & de vauc & 18.24 & & 7.2 & ( 4 ) & 0.71 & 5.7 + dunlop@xmath109 & de vauc & 17.46 & & 7.5 & ( 4 ) & 1.0 & 13 lcccccccccc + @xmath74 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath117 & & + @xmath75 & @xmath118 & @xmath119 & @xmath120 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & & + @xmath78 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & & + @xmath86 & @xmath134 & @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath131 & @xmath139 & @xmath140 & & + @xmath141 & @xmath142 & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & & + + @xmath76 & @xmath150 & @xmath151 & @xmath152 & @xmath153 & @xmath154 & @xmath131 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + @xmath77 & @xmath159 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & @xmath168 + @xmath141 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & @xmath175 & @xmath176 & @xmath177 & @xmath178 lcccccccc @xmath74 & @xmath179 & 102 & ( 0.000 ) & @xmath180 & & @xmath181 & & + @xmath75 & @xmath182 & 91 & ( 0.000 ) & @xmath183 & & @xmath184 & & + @xmath78 & @xmath185 & 109 & 0.016 & @xmath186 & @xmath187 & @xmath188 & @xmath189 & @xmath190 + @xmath86 & @xmath191 & 132 & 0.392 & @xmath192 & @xmath193 & @xmath194 & @xmath195 & @xmath196 + @xmath141 & @xmath197 & 136 & 0.284 & @xmath198 & @xmath199 & @xmath200 & @xmath201 & @xmath202 | we present extensive ground - based spectroscopy and _ hst _ imaging of 3c79 , an frii radio galaxy associated with a luminous extended emission - line region ( eelr ) .
surface brightness modeling of an emission - line - free _ hst _
@xmath0-band image reveals that the host galaxy is a massive elliptical with a compact companion 08 away and 4 magnitudes fainter .
the host galaxy spectrum is best described by an intermediate - age ( 1.3 gyr ) stellar population ( 4% by mass ) , superimposed on a 10 gyr old population and a power law ( @xmath1 ) ; the stellar populations are consistent with super - solar metallicities , with the best fit given by the 2.5 @xmath2models .
we derive a dynamical mass of @xmath3 @xmath4 within the effective radius from the velocity dispersion .
the eelr spectra clearly indicate that the eelr is photoionized by the hidden central engine .
photoionization modeling shows evidence that the gas metallicity in both the eelr and the nuclear narrow - line region is mildly sub - solar ( @xmath5 @xmath2 ) significantly lower than the super - solar metallicities deduced from typical active galactic nuclei in the sloan digital sky survey .
the more luminous filaments in the eelr exhibit a velocity field consistent with a common disk rotation .
fainter clouds , however , show high approaching velocities that are uncoupled with this apparent disk rotation .
the striking similarities between this eelr and the eelrs around steep - spectrum radio - loud quasars provide further evidence for the orientation - dependent unification schemes .
the metal - poor gas is almost certainly not native to the massive host galaxy .
we suggest that the close companion galaxy could be the tidally stripped bulge of a late - type galaxy that is merging with the host galaxy .
the interstellar medium of such a galaxy is probably the source for the low - metallicity gas in 3c79 . |
we now look at whether there are other types of pulse sequences similar to those above which can achieve the same fidelity . the most general 3-pulse sequence is , @xmath65 as before we define axes @xmath66 and @xmath3 in the @xmath5-plane of the bloch sphere by the angles @xmath67 and @xmath8 respectively .
the first condition to satisfy is that the pulse sequence must be symmetric so that it can be reduced using the symmetric bch formula , thus keeping the fidelity for the total pulse sequence 6@xmath2 order in @xmath1 .
this means we need @xmath68 and @xmath69 . also to satisfy the condition that @xmath70 ,
we require , @xmath71 we are then left with sequences of the form , @xmath72 finally , we let the first derivative of the total pulse sequence with respect to @xmath1 equal zero at @xmath16 and obtain the constraint , @xmath73 giving @xmath74 with @xmath75 and @xmath76 the most general 3-pulse sequences are @xmath77 with @xmath67 determined from , @xmath78 for even @xmath79 , @xmath80 meaning that either @xmath81 or @xmath82 .
hence @xmath83 since @xmath84 .
similarily , @xmath85 , i.e. @xmath86 and @xmath87 .
the same is true for odd @xmath79 except @xmath88 .
we are left with the general 3-pulse sequences , @xmath89 the @xmath62 family is obtained from repeating the @xmath90 sequence @xmath58 times . when @xmath91 , the `` passband '' wimperis sequence @xcite is recovered .
moreover , these are the only 3-pulse sequences which achieve this 6@xmath2 order fidelity .
the first three of these composite sequences are plotted in fig .
in other 3-pulse sequences , the first order term in the fidelity always disappears due to the fact that by collapsing the entire pulse sequence using the bch formula to , @xmath92 the fidelity is @xmath93 and the leading term in @xmath94 is @xmath95 .
so there are no pulse sequences constructed from 3 pulses which achieve this 6@xmath2 order dependence for the fidelity for the resulting rotation other than the above family of pulse sequences .
there is , however , the option of creating a 5-pulse sequence by introducing a third axis .
the general form for such a sequence is , @xmath96 we define the axes @xmath97 and @xmath98 in the @xmath5-plane of the bloch sphere by the angles @xmath99 and @xmath8 respectively .
@xmath98 is now the axis around which we wish to rotate by @xmath7 .
again , we will require @xmath100,@xmath101 and @xmath102 in order to keep symmetry in the sequence and hence the 6@xmath2 order fidelity dependence . in order that the pulse sequence is the identity when @xmath16 , we need to satisfy , @xmath103 the first derivative of the total pulse sequence is zero at @xmath16 when @xmath104 , @xmath105 for positive integers @xmath79 and @xmath106 .
we arrive at general 5-pulse sequences of the form , @xmath107 where @xmath108 and @xmath51 , @xmath52 and @xmath109 are determined from , @xmath110 one solution to find 5-pulse sequences is at @xmath111 , analogously to the 3-pulse sequence case . in this case
@xmath112 and , as it must remain an integer , @xmath113 must be a multiple of 3 with now @xmath114 .
( 11 ) is satisfied for @xmath115 when @xmath116 , @xmath117 and @xmath118 for a rotation around the -@xmath10-axis .
however , while the fidelity displays a 6@xmath2 order dependence on the fractional error , the coefficient of the leading term , shown in table [ table1 ] , is so much larger that the sequence is only better than the error - prone pulse for small values of @xmath1 ( @xmath119 0.2 )
. the situation does not improve for higher values of @xmath79 and so these sequences are of no real practical use .
another 5-pulse sequence which achieves the same 6@xmath2 order dependence for the fidelity is found by setting @xmath120 , @xmath121 , @xmath122 .
( 11 ) is now satisfied when @xmath123 , @xmath124 , @xmath125 for a rotation around the -@xmath10-axis ( @xmath126 ) .
as before , other axes may be accounted for by phase - shifting each pulse by the appropriate angle .
the fidelity of this sequence is much better than the previous 5-pulse sequence and is quite close to that of the pb1 sequence as seen in fig .
other sequences can be constructed by varying @xmath79 , @xmath106 and @xmath127 .
we have presented an analysis of the composite pulse sequences presented by jones _
@xcite to combat systematic pulse - length errors in single - qubit rotations .
we have derived an explicit form for the fidelity and shown how it is possible to set up other 3-pulse sequences which achieve the same order error dependence for the fidelity .
we have shown that there are also 5-pulse sequences which do achieve the 6@xmath2 order dependency of the fidelity on the error . * acknowledgement * + d. mchugh kindly acknowledges support from enterprise - ireland basic research grant sc/1999/080 .
the work was also supported by the ec ist fet project qipddf - roses ist-2001 - 37150 .
we thank the referees for their comments .
99 d. deutsch , _ proc .
lond . a _ * 400 * , 97 ( 1985 ) . h. k. cummins , g. llewellyn , j. a. jones _ physical review a _ * 67 * , 042308 ( 2003 ) .
s. wimperis , _ j. magn .
a _ * 109 * , 221 - 231 ( 1994 ) .
a. iserles , _ j. comp
_ * 19 * , 15 - 26 ( 2001 ) .
j. a. jones , _ proc .
soc . a _ * 361 * , 1429 - 1440 ( 2003 ) .
j. a. jones , _ phys . lett . a _ * 316 * 24 - 28 ( 2003 )
. k. r. brown , a. w. harrow , i. l. chuang , quant - ph/0407022 ( 2004 ) .
f. schmidt - kaler et al .
_ nature _ * 422 * , 408 - 411 ( 2003 ) .
s. gulde et al .
_ nature _ * 421 * , 48 - 50 ( 2003 ) .
d. wineland et al .
lond . a _ * 361 * , 1349 - 1361 ( 2003 ) .
i. roos , k. molmer _ physical review a _ * 69 * , 022321 ( 2004 ) .
j. wesenberg , k. molmer _ physical review a _ * 68 * , 012320 ( 2003 ) . m. steffen , j. m. martinis , i. l. chuang _ physical review b _ * 68 * , 224518 ( 2003 ) .
p. bertet et al .
cond - mat/0405024 ( 2004 ) .
b. e. kane , _ fortschr .
phys . _ * 48 * 1023 ( 2000 ) .
a. j. fisher _ proc .
lond . a _ * 361 * , 1441 - 1450 ( 2003 ) .
coefficients @xmath3 in the fidelity expansion @xmath128 for six composite pulse sequences which compensate for an error - prone @xmath129-pulse around the -@xmath10-axis . [ cols="<,^,<,^",options="header " , ] | composite pulse sequences designed for nuclear magnetic resonance experiments are currently being applied in many quantum information processing technologies .
we present an analysis of a family of composite pulse sequences used to address systematic pulse - length errors in the execution of quantum gates .
it has been demonstrated by cummins _
et al .
_ in @xcite that for this family of composite pulse sequences , the fidelity of the resulting unitary operation compared with the ideal unitary operation is @xmath0 , where @xmath1 is the fractional error in the length of the pulse .
we derive an exact expression for the 6@xmath2 order coefficient , @xmath3 , and from this deduce conditions under which this 6@xmath2 order dependance is observed .
we also present new pulse sequences which achieve the same fidelity .
0.1 cm systematic errors are always present in any experimental set - up .
they are best dealt with by stripping down the experiment and finding them one by one . even after doing this
, there is still the possibility of not eliminating them all .
quantum computers operate universally with single - qubit rotations and a controlled not gate between two qubits @xcite . recently ,
cummins _
et al .
_ @xcite have addressed the issue of systematic errors in single - qubit gates due to pulse - length and off - resonance effects .
off - resonance errors result in a rotation around an axis tilted with respect to the desired rotation axis .
pulse - length errors result in a rotation through an angle which falls short of , or goes beyond , the desired angle of rotation , due to an error in the timing of the pulse .
composite pulse techniques have received much attention as a means to correct systematic errors @xcite , with @xcite a significant recent development .
first developed in the nmr setting , these techniques can be very useful in reducing the effects of systematic errors in a wide range of quantum computer proposals .
they have been identified as such in numerous quantum information processing technologies such as trapped - ion technologies @xcite , in rare - earth doped crystal technology @xcite , in superconducting technologies @xcite and in solid state quantum information processing technologies @xcite . the bb1 composite pulse sequence , first developed by wimperis @xcite , deals with pulse - length errors in a remarkably efficient manner .
it is shown in @xcite that the composite pulse sequence has a fidelity of @xmath0 when compared to the exact qubit rotation . in this article
we examine the origin of this @xmath4 order dependence on the error in the fidelity of the gate , and deduce some constraints on the possible composite pulse sequences one may use .
we also suggest some other pulse sequences whose fidelities display this 6@xmath2 order dependence . a general single - qubit rotation around an axis in the @xmath5-plane of the bloch sphere can be written as , @xmath6 where @xmath7 is the angle through which the qubit is rotated and @xmath8 is the angle of the axis in the @xmath5 plane of the bloch sphere , @xmath9 being the @xmath10 axis .
we now consider the effects of a systematic error , such as a pulse - length type error .
using a superscript to denote pulse sequences suffering this type of error , where all rotation angles are altered by the fractional error @xmath1 , the above one single - qubit rotation becomes @xmath11 where @xmath1 is the fractional error . in order to compare the error - prone unitary transformation ( @xmath12 ) with the error - free unitary transformation ( @xmath13 ) ,
the following fidelity definition is used , @xmath14 the composite pulse sequence as presented by wimperis takes the form , @xmath15 so that when @xmath16 , @xmath17 is simply the identity .
when this pulse sequence is carried out before or after the desired single - qubit rotation , the fidelity of the resulting composite pulse sequence is @xmath0 .
in fact , @xmath18 may be placed at any point during a rotation around a given axis , i.e. , @xmath19 with @xmath20 $ ] , and the same fidelity is observed . in order to show where this comes from , we need to examine the definition of the fidelity .
first we let , @xmath21 be three axes in the @xmath5-plane of the bloch sphere .
@xmath3 is the axis around which we wish to perform the rotation and @xmath22 and @xmath23 are the two axes in @xmath18 .
the fidelity is , @xmath24 the last step is due to the trace property of invariance under cyclic permutations .
@xmath18 can be simplified to , @xmath25 since @xmath26 .
@xmath27 is another axis in the @xmath5-plane of the bloch sphere .
the fidelity now takes on a more symmetric look when written as , @xmath28 this form for the fidelity turns out to be the reason the bb1 sequence performs so well . the symmetric baker - campbell - hausdorff formula is stated and derived in @xcite as , @xmath29 with , @xmath30+{\mathcal o}(t^5 ) % + t^5(-\frac{1}{1152}[r,[r,[r,[r , s]]]]\\&&-\frac{1}{1152}[s,[r,[r,[r , s]]]]+\frac{1}{1440}[r,[s,[r,[r , s]]]]\\&&+\frac{1}{360}[s,[s,[r,[r , s]]]]-\frac{1}{480}[s,[r,[s,[r , s]]]])+{\cal o}(t^7).\end{aligned}\ ] ] if we define @xmath31 , @xmath32 and @xmath33 to simplify the notation , we can reduce the expression in the trace above by twice applying the symmetric bch formula , yielding , @xmath34 letting @xmath35 , we find that , @xmath36\right.\\&&\left.-\frac{1}{192}[q,[q,2(r+s)]-\frac{1}{24}[r+s,[q , r+s]]\right)\\&&+{\cal o}(\epsilon^5).\end{aligned}\ ] ] given that all the operators in @xmath37 are proportional to the pauli operators , @xmath38 and therefore , @xmath39 for the next step we let the first derivative of the total pulse sequence with respect to the error , @xmath1 , equal the zero matrix at @xmath16 to find a relation between the rotation axis , @xmath3 , and the axes @xmath22 and @xmath23 , the total pulse sequence is given by , @xmath40 so that , @xmath41 @xmath42 this allows us to simplify the expression for @xmath37 , @xmath43 + \frac{1}{192}[q,[q , q]]\\&&-\frac{1}{96}[q,[q , q ] ] ) + { \cal o}(\epsilon^5)\\ & = & -\frac{1}{24}[r+2s,[r , s]]\epsilon^3 + { \cal o}(\epsilon^5).\end{aligned}\ ] ] next we expand the commutator @xmath44 $ ] to get , @xmath45)^2&=&-\pi^6([a+2aba,[a , aba]])^2\\ & = & -\pi^6(40{\mathbb{i}}+8(ab+ba)\\&&-20(abab+baba)\\&&-8(abab ab+bababa))\\ & = & -\pi^6[40 + 16\cos(\phi_2-\phi_1 ) \\&&\ \ \ \ \ \ \ \ \ \ -\ 40\cos 2(\phi_2-\phi_1 ) \\&&\ \ \ \ \ \ \ \ \ \ -\ 16\cos 3(\phi_2- \phi_1)]{\mathbb{i}}\end{aligned}\ ] ] since @xmath46 .
therefore , we can write the fidelity as , @xmath47)^2)+{\cal o}(\epsilon^8)\\ & = & 1-\frac{\epsilon^6\pi^6}{144}[5 + 2\cos(\phi_2-\phi_1)-5\cos 2(\phi_2-\phi_1 ) \\&&\ \ \ \ \ \ \ \ \ \ \ \ -2\cos 3(\phi_2-\phi_1)]+{\cal o}(\epsilon^8)\end{aligned}\ ] ] finally , by writing @xmath22 , @xmath23 and @xmath3 in terms of the pauli operators , @xmath10 and @xmath48 , the condition @xmath49 implies , @xmath50 and these two equations lead to the following rule for choosing @xmath51 and @xmath52 given @xmath7 and @xmath8 , @xmath53 to give a concrete example , we choose @xmath9 , @xmath54 corresponding to a @xmath55 rotation about the @xmath10 axis of the bloch sphere .
this gives us , @xmath56 and a fidelity , @xmath57 which agrees with the findings in @xcite . it can be readily seen that if @xmath58 copies of @xmath18 are carried out one after the other , then the condition that @xmath59 , @xmath60 and also @xmath61 .
this defines the general @xmath62 pulses introduced in @xcite as , @xmath63 and shows that they all perform with a fidelity @xmath64 .
we note that it is only necessary to show the sequence works for one angle , say @xmath9 , with other axes accounted for by phase - shifting each pulse by some angle @xmath8 . |
ultracold atomic gases provide an experimental playground for testing pairing phenomena , due to the ability to control the inter - atomic interactions via a magnetically - tuned feshbach resonance @xcite . over the past year
, tremendous progress has been made in realizing the crossover from the bose - einstein condensation ( bec ) of diatomic molecules to the bardeen - cooper - schrieffer ( bcs ) limit of weakly - bound cooper pairs in ultracold gases of fermionic atoms @xcite .
but there is much debate over where the crossover between the bcs and bec regimes occurs , following recent experimental studies on fermionic condensates @xcite .
therefore , it is timely to perform a thorough theoretical investigation of what the signature of the crossover is and how it might be observed experimentally , using a realistic potential with a finite range .
much of the theoretical work on systems of two types of fermions interacting via an adjustable , attractive potential have focussed on interactions that are governed by a single parameter , namely the s - wave scattering length @xmath0 @xcite .
such a description is valid provided we have @xmath1 and @xmath2 , where @xmath3 is the width of the potential and @xmath4 is the fermi momentum of the non - interacting system , so that the only independent dimensionless variable in the problem is @xmath5 .
thus , it is only applicable to dilute systems like the ultracold atomic gases , not the high density situation found in conventional superconductors @xcite .
the potential supports a 2-body bound state for @xmath6 , but this molecular state passes through zero energy and vanishes into the continuum at @xmath7 , the position of the feshbach resonance . as such
, the bcs and bec limits correspond to @xmath8 and @xmath9 , respectively .
the distinction between the bec and bcs regimes can be made sharper @xcite : the qualitative boundary between the bcs and bec ground states occurs when the chemical potential @xmath10 reaches @xmath11 , marking the disappearance of a fermi surface , and this coincides with the appearance of a bound state at @xmath7 only when the density is zero . at the point where
there is no longer a defined fermi surface , there are qualitative changes in the quasi - particle excitation spectrum : the momentum corresponding to the minimum in the excitation spectrum shifts from finite momentum in the bcs limit to zero momentum in the bec limit . at the same point , the value of the excitation gap goes from @xmath12 in the bcs limit to @xmath13 in the bec limit , where @xmath12 is the gap parameter .
we shall use these changes in the excitation spectrum as the definition " of the crossover point .
there have also been studies of the bcs - bec crossover using coulomb potentials in the context of excitons @xcite . here ,
the interaction is kept fixed and the density varied to achieve a smooth transition from the high density bcs limit to the dilute bec limit . note that it is not possible to obtain a density - driven crossover using an interaction that is only characterized by the scattering length , since this picture breaks down at high densities . in this paper , we shall study the bcs - bec crossover at zero temperature using a realistic gaussian potential and compare with results from single - parameter potentials .
we use a mean - field variational wave function to describe the crossover , which is expected to give an accurate description of the low temperature behavior , though not , of course , the finite temperature transition . the paper is organized as follows : we describe the model and variational scheme in sec .
[ formalism ] .
we discuss the characteristics of the wave functions and excitation spectra in sec .
[ spectra ] . in sec .
[ universal ] we compare our results with leggett s zero - range pseudo - potential result which is usually used to describe the mean - field limit for dilute fermi systems .
we conclude in sec .
[ conclusions ] .
we determine the ground - state wave function @xmath14 using conventional mean - field methods , briefly reviewed here .
we consider the hamiltonian @xmath15 where @xmath16 and @xmath17 denote momentum variables and spin states @xmath18 , respectively , and @xmath19 .
since experiments on atomic gases are carried out at low energies , we will restrict ourselves to the simple case of an _ s_-wave pairing interaction .
we calculate the ground state by minimizing the free energy @xmath20 where @xmath21 is the total number operator and @xmath10 is the chemical potential .
the standard bcs variational wave function is given by @xmath22 and it smoothly interpolates from the bcs to bec limits , giving an accurate description of the crossover . here ,
the bcs parameters @xmath23 , @xmath24 only depend on @xmath25 is the normalization constant , such that @xmath26
. with this ansatz , the free energy becomes @xmath27 where the _ normal _ and _ anomalous _ densities are @xmath28 and @xmath29 respectively , subject to the normalization condition , @xmath30 without loss of generality we can choose @xmath23 and @xmath24 to be real .
after minimization , the resulting equations we have to solve are @xmath31 with the added constraint that the total density @xmath32 of particles is constant @xmath33
( color online ) scattering length @xmath0 and effective range of the interaction @xmath34 as a function of the potential depth @xmath35 , where the energy scale @xmath36 .
, scaledwidth=48.0% ] ( color online ) momentum distributions @xmath37 ( top ) and the quasi - particle excitation spectra @xmath38 ( bottom ) as a function of @xmath39 , at fixed density ( @xmath4 ) and various potential scattering lengths @xmath0 , where @xmath40 is the fermi energy and @xmath41.,scaledwidth=48.0% ] numerical solutions are found for an attractive short - range interaction , described by a gaussian potential @xmath42 .
we explore the bcs - bec crossover by fixing the width of the potential @xmath43 and varying the potential depth @xmath44 , or implicitly the scattering length @xmath0 of the interaction .
[ fig1 ] depicts all the essential two - body physics of the problem . as the depth of the potential
is increased from zero , the scattering length grows , diverging to negative infinity at the first appearance of a two - body bound state . on the bound - state side of the feshbach resonance
the scattering length re - emerges from positive infinity .
note that this procedure does not keep the _ effective _ range @xmath34 of the potential fixed . in the low - energy limit ,
the effective range @xmath34 , like the scattering length @xmath0 , can be determined from the phase shift @xmath45 and is shown in fig . [ fig1 ] .
the variation of the effective range is small over the region of interest .
from now on , we shall focus on finite densities and the crossover to bcs that can occur in this regime . like previous theoretical studies @xcite
, we find that the momentum distribution @xmath37 at constant density smoothly evolves from the wave function of a molecule in momentum space at large positive @xmath46 to a fermi - like distribution at large negative @xmath46 , as depicted in fig .
[ fig2 ] ( top ) . during this transition , we see from fig .
[ fig2 ] ( bottom ) that the quasi - particle excitation spectrum @xmath47 develops a minimum at finite momentum , signifying the bcs - bec crossover point defined previously . because the density is non - zero , though small , the crossover point occurs for positive @xmath5 , where a two - body molecular bound state has already formed .
the appearance of the minimum in the quasi - particle excitation spectrum at non - zero momentum is clearly evident in the quasi - particle density of states . as a function of momentum ,
the density of states is defined as @xmath48 thus , @xmath49 is singular whenever there is a stationary point in the excitation spectrum ( @xmath50 ) , except when @xmath51 , where @xmath49 vanishes provided @xmath52 no faster than @xmath53 .
[ fig3 ] shows how @xmath49 varies with scattering length at constant particle density .
the energy minimum for @xmath54 shows up as a spike that migrates away from @xmath51 towards @xmath55 as @xmath56 , but the spike disappears when the energy minimum occurs at @xmath51 .
thus , the bcs - bec crossover is the point at which a spike first appears as we decrease @xmath57
. a more straightforwardly measurable quantity may be the density of states as a function of energy @xmath58 which is depicted in fig .
[ fig4 ] . approximating the gap parameter @xmath59 as a constant
, the bcs regime has a singularity at finite momentum ( @xmath60 ) of the form @xmath61 where the amplitude is @xmath62 we see that the size of the singularity will increase the further we enter the bcs phase . in the bec limit ,
the density of states is zero at the energy minimum occurring at @xmath51 , and deep within the bec phase it has the following form close to the minimum @xmath63 ( color online ) density of states @xmath49 ( as a function of momentum ) , at fixed @xmath4 and various potential scattering lengths @xmath0 , where @xmath41 . , scaledwidth=48.0% ] ( color online ) density of states @xmath64 ( as a function of energy ) , at fixed @xmath4 , where @xmath41 , and various potential scattering lengths @xmath0 . , scaledwidth=48.0% ] ( color online ) density of states @xmath49 ( as a function of momentum ) , at fixed potential scattering length ( @xmath65 ) , and various values of @xmath4 .
, scaledwidth=48.0% ] as well as considering the bcs - bec crossover at constant density arranged by tuning the interaction , we also examine the _ density - driven _ crossover at fixed interaction . fig .
[ fig5 ] illustrates the density of states as a function of momentum for various particle densities , where we have chosen @xmath0 to be positive so that a molecular bound state exists .
we observe the same disappearance of the spike in @xmath49 as @xmath66 but , unlike fig .
[ fig3 ] , the extreme bcs limit occurs when @xmath67 since this corresponds to infinite density .
note also that both fig .
[ fig3 ] and [ fig5 ] show @xmath68 const , as @xmath69 , so that the pair - breaking excitation spectrum has @xmath70 , where @xmath71 is the binding energy of the molecule .
this is exactly what we would expect for a non - interacting gas of bose molecules .
( color online ) momentum distribution of the condensate wave function , at fixed density ( @xmath4 ) and various potential scattering lengths @xmath0 , where @xmath41 .,scaledwidth=48.0% ] finally , fig .
[ fig6 ] shows the behavior of the condensate wave function throughout the crossover .
here we see the most pronounced features of the evolution from a molecular state to the bcs limit where pairing exists only on the fermi surface .
profiles of the chemical potential @xmath72 as a function of @xmath73 .
the points mark the bec - bcs crossover . , scaledwidth=48.0% ] a more complete understanding of the bcs - bec crossover can be gained from considering the phase diagram for our finite - range potential .
figure [ fig : fig7 ] shows the variation of chemical potential @xmath72 as a function of @xmath46 , for a set of different scattering lengths @xmath74 . on each curve
we mark the position of the bcs - bec crossover point , which we refer to as the `` critical '' pair ( @xmath75 , @xmath76 ) .
the crossover point always occurs for @xmath77 and there is no density - driven crossover when @xmath78 . since the size of @xmath0 dictates the size of the two - body molecular bound state while @xmath77 , the crossover point in fig .
[ fig : fig7 ] is driven to lower densities when @xmath0 is increased .
as the zero - density limit ( @xmath76=0 ) is approached , the critical chemical potential @xmath79 , which is equivalent to the bcs - bec crossover for a single - parameter potential initially discussed by leggett . however , as @xmath76 increases , the crossover point shifts to negative values of @xmath75 and larger @xmath46 .
it is useful to compare the results of our calculations with previous studies of the bcs - bec crossover that involve single - parameter potentials . in the dilute or low - energy limit , where @xmath80 , the effective interaction @xmath81 can be written as @xmath82 and it is independent of the details of the real interaction . substituting @xmath81 for the potential and eliminating high energies , eqs . and become , respectively @xmath83 \ = \ \pi \
> , \\
\int_0^\infty \ \mathrm{d } \varepsilon \ \sqrt \varepsilon \
\biggl [ 1 \ - \ \frac{\varepsilon - \tilde \mu}{\sqrt { ( \varepsilon - \tilde \mu)^2 + \tilde \delta } } \biggr ] \ = \
\frac{4}{3 } \>,\end{aligned}\ ] ] where we have introduced the notations @xmath84 , @xmath85 , and @xmath73 .
the above equations , first studied by leggett @xcite , are then solved for @xmath86 and @xmath87 . for obvious reasons
, we will refer to these functions as `` model independent '' or `` universal . ''
( color online ) comparison with leggett s `` universal '' curve : we keep @xmath4 constant , and vary the scattering length @xmath0 . , scaledwidth=48.0% ] we first compare our results for @xmath86 against leggett s predictions for the case where the scattering length @xmath0 is varied and the density is held fixed , as shown in fig [ fig8 ] . on the leggett curve ,
the bcs - bec crossover point ( @xmath88 ) occurs at @xmath89 , and our numerical @xmath86 converges to this universal curve as @xmath90 . since typical experimental parameters
correspond to @xmath91 , the ultracold atomic gases clearly lie on the leggett curve and are well described by a single - parameter potential . the value of @xmath72 for @xmath92 is usually referred to as the `` unitarity '' limit @xcite . in this limit one
expects that all sensitivity to the detailed features of the interaction is lost @xcite , and the system energy is determined entirely by the density . as a result
, this limit is particularly sensitive to many - body correlation effects , and the leggett curve predicts this value as being 0.59 ( see also @xcite ) .
recently , accurate calculations based on the green s function monte carlo ( gfmc ) method @xcite , have lowered the upper - bound on this result to ( @xmath93 ) , which shows that beyond mean - field effects account for at least a 25% improvement in the binding energy over the mean - field result .
( color online ) comparison with leggett s `` universal '' curve : the scattering length @xmath0 is held constant , the density ( @xmath4 ) is varied .
, scaledwidth=48.0% ] fig .
[ fig9 ] compares the numerical @xmath86 with the universal curve for the case where the density is varied and @xmath74 is fixed .
not surprisingly , the numerical @xmath86 approaches the universal curve as scattering length increases , and we require @xmath94 before @xmath86 touches the universal curve .
furthermore , we see that no matter how large we make @xmath0 , the universal curve breaks down in the region sufficiently close to @xmath95 .
this point corresponds to infinite density , the bcs limit in the density - driven crossover , so we must always have @xmath96 , which is consistent with our numerical results .
if we consider @xmath86 for which @xmath97 we observe that it collapses onto the universal curve when @xmath98 .
this translates into the condition @xmath99 for the universal equations to be valid , which is consistent with fig .
in this paper we have performed a study of the bcs - bec crossover , at zero temperature , in the single - channel model , using a realistic effective interaction . in the mean - field approximation
the crossover is described using a variational wave function , which interpolates smoothly between the bec and bcs limits .
we find that the crossover point always lies on the side of the feshbach resonance where a molecular bound state exists at zero density , and that it converges to that of a single - parameter potential in the low density limit .
moreover , our results reproduce the usual leggett picture in the limit of a dilute fermi gas , as expected . by using a potential of finite range , we have shown that the crossover for typical experimental parameters in ultracold atomic fermi gases can be described by a zero - range , single - parameter potential .
we emphasize that a clear signature of the crossover point appears in the density of states , which could be measured spectroscopically @xcite .
the mean - field description of the single - channel model presented here is not entirely satisfactory .
as described above , recent gfmc results provide an upper value for the energy density in the dilute limit ( @xmath92 ) , which represents a 25% improvement over the leggett limit .
therefore , additional work is required in order to include next - to - leading order effects in a quantitative way , and study the changes ( or lack thereof ) in the features of the bec - bcs crossover .
approximations schemes based on the two - particle irreducible effective action @xcite are readily available in quantum field theory where they have been recently employed to describe the dynamics of a system out of equilibrium , and the dynamics of phase transitions @xcite .
work is currently under way in order to apply such methods to the study of the bec - bcs crossover based on the hamiltonian .
the single - channel model is of course a simplification of detailed models for the feshbach resonance which include the resonance explicitly , either as a boson ( the fermi - bose model @xcite ) or more generally as three - level @xcite or four - level fermi systems . the fermi - bose model , as well as models with even numbers of fermions , all have large regimes of parameter space where the resonance is sufficiently detuned from the open channel that high energy degrees of freedom can be integrated out in favor of an effective interaction between two species , recovering the model discussed in this paper .
close enough to resonance the detailed fermionic structure of the resonance may become important .
models where the feshbach bound state shares a state with the open channel ( relevant to @xmath100 ) are different , and do not reduce to an effective single channel model in a simple way@xcite . in future work
, we will compare and contrast these different models @xcite using a generalization of the variational scheme given in this paper .
c. a. regal , c. ticknor , j. l. bohn , and d. s. jin , nature * 424 * , 47 ( 2003 ) ; s. jochim , m. bartenstein , a. altmeyer , g. hendl , c. chin , j. hecker denschlag and r. grimm , science * 302 * , 2101 ( 2003 ) ; m. greiner , c. a. regal , and d. s. jin , nature * 426 * , 537 ( 2003 ) ; m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , s. gupta , z. hadzibabic and w. ketterle , phys .
* 91 * , 250401 ( 2003 ) ; t. bourdel , l. khaykovich , j. cubizolles , j. zhang , f. chevy , m. teichmann , l. tarruell , s.j.j.m.f .
kokkelmans , and c. salomon , phys .
* 93 * , 050401 ( 2004 ) . c. a. regal , m. greiner , and d. s. jin , phys .
lett . * 92 * , 040403 ( 2004 ) ; m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
lett . * 92 * , 120403 ( 2004 ) .
m. randeria , in _ bose - einstein condensation _ , edited by a. griffin , d. w. snoke , and s. stringari ( cambridge university press , cambridge , 1995 ) .
j. r. schrieffer , _ theory of superconductivity _ , ( benjamin / cummings , reading , 1964 ) h. heiselberg , phys .
a * 63 * , 043606 ( 2001 ) ; r. combescot , phys .
* 91 * , 120401 ( 2003 ) .
g. a. baker , phys .
c 60 , 054311 ( 1999 ) ; t. papenbrock and g. f. bertsch , phys .
c * 59 * , 2052 ( 1999 ) .
j. stenger , s. inouye , a. p. chikkatur , d. m. stamper - kurn , d. e. pritchard , and w. ketterle , phys .
* 82 * , 4569 ( 1999 ) ; p. trm and p. zoller , phys .
lett . * 85 * , 487 ( 2000 ) ; t. bourdel , j. cubizolles , l. khaykovich , k. m. f. magalhaes , s. j. j. m. f. kokkelmans , g. v. shlyapnikov , and c. salomon , phys .
lett . * 91 * , 020402 ( 2003 ) ; l. viverit , s. giorgini , l. p. pitaevskii , and s. stringari , phys .
a * 69 * , 13607 ( 2004 ) .
j. m. luttinger and j. c. ward , phys .
rev . * 118 * , 1417 ( 1960 ) ; g. baym , phys . rev . * 127 * , 1391 ( 1962 ) ; h. d. dahmen and g. jona lasino , nuovo cimento , * 52a * , 807 ( 1962 ) ; j. m. cornwall , r. jackiw and e. tomboulis , phys . rev .
d * 10 * , 2428 ( 1974 ) .
k. b. blagoev , f. cooper , j. f. dawson and b. mihaila , phys.rev .
d. * 64 * , 125003 ( 2001 ) ; f. cooper , j. f. dawson and b. mihaila , phys .
d * 67 * , 051901(r ) ( 2003 ) ; f. cooper , j. f. dawson , and b. mihaila , phys .
d * 67 * , 056003 ( 2003 ) ; b. mihaila , phys . rev .
d * 68 * , 36002 ( 2003 ) [ hep - ph/0303157 ] .
m. holland , s. j. j. m. f. kokkelmans , m. l. chiofalo and r. walser , phys .
lett . * 87 * , 120406 ( 2001 ) ; e. timmermans , k. furuya , p. w. milonni and a. k. kerman , phys .
a * 285 * , 228 ( 2001 ) ; y. ohashi and a. griffin , phys .
rev . lett . * 89 * , 130402 ( 2002 ) . | we study the crossover from bec to bcs pairing for dilute systems but with a realistic finite - range interaction .
we exhibit the changes in the excitation spectrum that provide a clean qualitative distinction between the two limits .
we also study how the dilute system converges to the results from a zero - range pseudo - potential derived by leggett . |
quantum antiferromagnetism in lower dimensions is one of the most fascinating subjects in the condensed matter physics . in particular ,
the antiferromagnetic quantum heisenberg chains ( afqhc ) with small spins have attracted much attention on account of the rich quantum behaviour they display .
nevertheless , due to the strong quantum fluctuations the classical nel state is not more an eigenstate of the hamiltonian and thus , more interesting quantum phases should be expected to occur in the ground state . the nature of these phases , however , basically depends on the spin value of atoms .
in fact , as conjectured haldane @xcite in 1983 , the integer spin afqhc have a finite energy gap between the ground state and the first excited state , while the half - odd integer ones possess a gapless excitation spectrum .
another striking feature of the afqhc is the appearance of fractional magnetization plateaus in the magnetization process . extending the original lieb - schultz - mattis theorem @xcite oshikawa , yamanaka and affleck ( oya ) @xcite argued that the magnetization per site @xmath2 can be topologically quantized as : @xmath3 where @xmath4 is a period of ground state in the thermodynamic limit and @xmath5 denotes a total spin of an elementary unit .
however , this condition represents just the necessary condition for the plateau - state formation and does not directly prove its existence .
it is therefore of interest to investigate how the plateau state is related to the periodicity of specific model @xcite .
moreover , from the theoretical viewpoint the plateau state can also be regarded as a spin - gap state .
hence , the zero - temperature magnetization curves with plateaus bring an insight into the ground - state properties of the system , since the field - induced spin gaps reflect the gapped excitation spectrum . despite of an extensive theoretical effort focused on afqhc ,
there still exist only few exactly solvable models with pure heisenberg exchange interactions @xcite , especially for mixed - spin chains @xcite . on the other hand ,
an exact solution for the chains with alternating ising- and heisenberg - type exchange interactions can be attained in less sophisticated manner .
indeed , the exact solution for the ising - heisenberg bond alternating chain ( originally proposed and solved by lieb _ et al _ @xcite ) , has been recently successfully generalized to the case of anisotropic heisenberg interaction @xcite . in order to avoid mathematical complexities connected with the noncommutability of relevant spin operators
, we will introduce in this article another class of the ising - heisenberg chains ( with period @xmath6 ) , that can be treated exactly within the mapping transformation method .
however , the considered model naturally enables to analyse in detail the mutual competition between ising- and heisenberg - type interactions and moreover , it also proves to be very useful in view of the confirmation of multiplateau magnetization curves by an exact calculation .
this paper is organized as follows . in sec .
2 the detailed description of the model , as well as the fundamental aspects of transformation technique are presented . in sec .
3 we are concerned with the analysis of the most interesting numerical results for typical spin cases and finally , in sec .
4 some concluding remarks are also drawn .
in this article we will study a mixed spin-@xmath7 and spin-@xmath0 ( @xmath1 ) ising - heisenberg chain in a presence of external magnetic field .
the structure of the considered mixed - spin chain is depicted in fig.1 , where the black circles denote the spin-@xmath7 atoms and the grey ones represent the spin-@xmath0 atoms .
the total hamiltonian of the system is given by : @xmath8 + j_1 \sum_{k , l } \hat s_k^z \hat \mu_l^z - h_a \sum_l \hat \mu_l^z - h_b \sum_k \hat s_k^z , \label{r1}\end{aligned}\ ] ] where @xmath9 and @xmath10 ( @xmath11 ) denote the well - known components of standard spin-@xmath7 and spin-@xmath0 ( @xmath1 ) operators , respectively .
the parameter @xmath12 stands for the heisenberg interaction between nearest - neighbouring spin-@xmath0 atoms ( the grey atoms ) and @xmath13 is the anisotropy parameter that allows to control the anisotropic xxz interaction between an easy - axis regime ( @xmath14 ) and an easy - plane regime ( @xmath15 ) .
furthermore , the interaction parameter @xmath16 describes the ising - type exchange interaction between pairs of nearest - neighbouring spin-@xmath7 and spin-@xmath0 atoms and finally , the terms incorporating @xmath17 and @xmath18 respectively , describe the coupling of spin-@xmath7 atoms and spin-@xmath0 atoms to an external magnetic field . as we can see from fig.1 ,
all pairs of nearest - neighbouring heisenberg atoms are surrounded by ising - type atoms only , thus the model under investigation can also be viewed as an ising model the bonds of which are doubly decorated by the heisenberg atoms . in view of further manipulations , it is useful to rewrite the total hamiltonian @xmath19 as a sum of the bond hamiltonians , i.e. @xmath20 , where @xmath21 denotes the total number of ising - type atoms and the summation runs over all bonds of the original ( undecorated ) chain
. the bond hamiltonian @xmath22 contains all the interaction terms associated with the @xmath23th couple of heisenberg atoms ( see fig.1 ) , and it is given by @xmath24 + j_1 ( \hat s_{k1}^z \hat \mu_{k1}^z + \hat s_{k2}^z \hat \mu_{k2}^z ) \nonumber \\ & & - h_b ( \hat s_{k1}^z + \hat s_{k2}^z ) - h_a ( \hat \mu_{k1}^z + \hat \mu_{k2}^z)/2 . \label{r2}\end{aligned}\ ] ] now , exploiting the usual commutation relation for the bond hamiltonians ( i.e. @xmath25 = 0,$ ] for @xmath26 ) , the partition function of the system can be partially factorized , namely , @xmath27 in above , @xmath28 , @xmath29 being boltzmann constant and @xmath30 the absolute temperature .
@xmath31 means a trace over the degrees of freedom of the ising spins and @xmath32 denotes a trace over the @xmath23th couple of heisenberg spins . at this stage one
easily observes that the structure of relation ( [ r3 ] ) implies a possibility to introduce the decoration - iteration mapping transformation @xcite @xmath33 .
\label{r4}\end{aligned}\ ] ] as usual , the unknown transformation parameters @xmath34 , @xmath35 and @xmath36 can be attained by taking into account remaining degrees of freedom of both ising spins ( @xmath37 and @xmath38 ) . in this way one obtains for the transformation parameters
@xmath34 , @xmath35 and @xmath36 the following expressions @xmath39 where the functions @xmath40 , @xmath41 and @xmath42 depend on the spin value of heisenberg atoms , as well as on the parameters of hamiltonian ( [ r1 ] ) and they are summarized for both investigated spin cases in the appendix . here
, one should emphasize that the mapping relations ( [ r4])-([r5 ] ) enable to transform the ising - heisenberg mixed - spin chain onto a simple spin-1/2 ising chain with an effective exchange parameter @xmath35 , placed in an external magnetic field of the magnitude @xmath36 . indeed , substituting ( [ r4 ] ) into ( [ r3 ] ) gives the following equality @xmath43 which relates the partition function of ising - heisenberg chain @xmath44 and that one of the spin-@xmath7 ising chain @xmath45 .
since the explicit expression for @xmath45 is well known @xcite , we can then straightforwardly calculate all relevant thermodynamic quantities .
for example , the gibbs free energy @xmath46 of the mixed - spin ising - heisenberg chain is given by @xmath47 where @xmath48 denotes the gibbs free energy of the corresponding spin-@xmath7 ising chain .
next , by differentiating the gibbs free energy @xmath46 with respect to @xmath17 and @xmath18 respectively , one directly obtains solution for the total sublattice magnetization .
of course , other thermodynamic quantities can also be calculated on the basis of familiar thermodynamic relations , e. g. the entropy @xmath0 and the specific heat @xmath49 can be calculated from @xmath50 nevertheless , a similar thermodynamic approach can not be used for the calculation of other important quantities such as staggered magnetization , quadrupolar momentum or some correlations .
fortunately , eq . ( [ r6 ] ) in conjunction with the transformation formula ( [ r4 ] ) allows after an elementary algebra the derivation of following exact spin identities @xcite @xmath51 with arbitrary function @xmath52 depending exclusively on ising spin variables and the function @xmath53 depending on the spin variables from the @xmath23th bond only .
the superscript @xmath54 label the spatial components of spin operators and finally , the symbols @xmath55 and @xmath56 stand for the standard ensemble average in the ising - heisenberg and its equivalent simple ising model , respectively .
however , the above spin identities considerably simplify the calculation of a large number of quantities .
indeed , for the reduced sublattice magnetization ( @xmath57 , @xmath58 ) , the total single - site magnetization @xmath2 and the staggered sublattice magnetization ( @xmath59 , @xmath60 ) , one attains after straightforward algebra @xmath61 in above , @xmath62 , @xmath63 and @xmath64 represent the reduced magnetization , staggered magnetization and nearest - neighbour correlation of the corresponding undecorated ising chain and the coefficients @xmath40-@xmath65 are listed for both investigated spin cases in the appendix .
finally , let us define some pair correlation functions and the quadrupolar momentum , which are also very useful for understanding of magnetic properties of the system , namely , @xmath66 in these equations , the subscripts denote the type of atoms and superscripts the space direction .
one should also notice that the definition of the parameter @xmath67 is obviously meaningful for @xmath68 only .
although , the derivation of relevant equations for these quantities is straightforward , the calculation procedure by itself is rather lengthy and tedious , thus the details are not presented here .
before discussing the most interesting numerical results , it is worth mentioning that some preliminary results for the ferromagnetic version of the model ( @xmath69 , @xmath70 ) have already been published by the present authors elsewhere @xcite .
for this reason , in this article we will restrict our attention to the doubly decorated ising - heisenberg chain with antiferromagnetic interactions only ( i.e. @xmath71 ) .
the particular attention is focused on the ground - state analysis and the appearance of plateaus in the chains with different decorating spin @xmath0 . among other matters
, we will directly prove the existence of double - plateau magnetization curve in the spin @xmath72 chain , more precisely , the magnetization curve with plateaus at @xmath73 and @xmath74 . on the other hand , in the spin @xmath75 chain a greater diversity of magnetization process will be confirmed , in fact , we will prove the existence of double - plateau ( @xmath73 and @xmath7 ) , triple - plateau ( @xmath73 , @xmath74 and @xmath7 ) and quadruple - plateau ( @xmath73 , @xmath74 , @xmath76 and @xmath7 ) magnetization curves .
in addition , since each couple of the heisenberg atoms is surrounded by the ising atoms only , the relevant spin deviations can not propagate through the ising bonds and therefore , the quantum fluctuations are necessarily localized within the unit cell ( within the four - spin cluster consisting of the heisenberg spin pair and its nearest - neighbouring ising spins ) . owing to this fact
, the observed plateaus can not be considered as the oya plateaus , i. e. the plateaus with collective eigenstate extending over the whole chain .
nevertheless , it is interesting to note that all observed fractional magnetization satisfy the oya condition for @xmath77 ( the period of translational symmetry should be twice as large as the periodicity of hamiltonian as a consequence of an antiferromagnetic nature of ground state ) .
we begin our analysis by considering the effect of exchange anisotropy @xmath13 and uniform magnetic field ( i.e. @xmath78 ) on the ground - state phase boundaries of the spin @xmath79 chain .
for this purpose , we have displayed in fig.2 some typical ground - state phase diagrams in the @xmath80 plane for @xmath81 and @xmath82 .
as one can see from this figure , the relevant phase boundaries separate three or four distinct phases , namely , the antiferromagnetic ( af ) , ferrimagnetic i ( fi ) , ferrimagnetic ii ( fii ) and saturated paramagnetic ( sp ) phase .
one also observes that both ferrimagnetic phases ( fi and fii ) represent an intermediate phase between af and sp phase occurring due to the first - order metamagnetic transition .
as we have already mentioned before , different phases can be distinguished by analysing the magnetization and the correlation functions at @xmath83 . in this way
one finds the following ground - state results for particular phases : + antiferromagnetic phase ( af ) : @xmath84 where we have defined the function @xmath85 , in order to write the relevant expressions in more abbreviated and elegant form .
+ ferrimagnetic phase i ( fi ) : @xmath86 ferrimagnetic phase ii ( fii ) : @xmath87 saturated paramagnetic phase ( sp ) : @xmath88 moreover , it is also noteworthy that all ground - state phase boundaries can be expressed analytically as follows ( see fig.2 ) , a ) for @xmath89 @xmath90 b ) @xmath91 @xmath92 in order to demonstrate the diversity of the magnetization process , we have plotted in fig.3 the low - temperature magnetization curves for various exchange anisotropies @xmath13 .
the detailed examination of these dependencies reveals that the mechanism of the magnetization process depends basically on the ratio between ising and heisenberg interaction constants , more specifically , on whether @xmath89 or @xmath91 . in the former case
, the plateau state arises due to the alignment of the ising spins towards the external - field direction ( fi phase ) regardless of the anisotropy strength @xmath13 ( see fig.3a ) .
naturally , in this case there is no other possibility for the formation of intermediate plateau .
on the other hand , in the case of @xmath93 the metamagnetic transition to the fi phase can be observed for the stronger anisotropies @xmath13 only ( see fig.3d ) , while for the weaker anisotropies the fii phase becomes the stable one . in this phase
, the heisenberg ( ising ) spins align parallel ( antiparallel ) with respect to the external - field direction as it is apparent from fig.3b .
moreover , the situation for the most interesting point at which both intermediate phases ( fi and fii ) coexist , is depicted in fig.3c .
referring to this plot , the coexistence of both ferrimagnetic phases is also reflected in the mixed feature of the magnetization curve ( compare magnetization curve from fig.3c with those in figs.3b and 3d ) . in order to enable an independent check of the magnetization scenario ,
we have also studied the relevant low - temperature dependencies of the nearest - neighbour correlation functions introduced in eq .
( [ r11 ] ) . in fig.4a ,
the variations of the correlation functions with anisotropy @xmath13 are shown for the system without external magnetic field .
as one can see , the correlation function @xmath94 between heisenberg spins takes its saturation value irrespectively of @xmath13 , what means that all nearest - neighbouring heisenberg spin pairs align antiparallel with respect to each other .
moreover , the saturated value of correlation @xmath95 indicates a perfect antiferromagnetic alignment also in the ising sublattice ( between third nearest - neighbour ising spins ) .
contrary to this behaviour , the perfect antiparallel alignment between ising and heisenberg spins is destroyed as @xmath13 increases from zero ( see the correlation @xmath96 ) .
hence , the ising and heisenberg spins are oriented randomly with respect to each other , the degree of randomness being the greater , the stronger the exchange anisotropy @xmath13 .
in addition , it is clear that the anisotropy term @xmath13 is also responsible for the onset of an interesting short - range ordering ( nonzero @xmath97 ) in @xmath98 plane .
anyway , the value of correlation @xmath97 can be thought as a measure of the strength of local quantum fluctuations appearing in the spin system .
furthermore , in figs.4b-4d we have displayed the field dependencies of the correlation functions corresponding to the magnetization curves from figs.3b-3d .
the depicted behaviour for the correlation functions is in complete agreement with results for the magnetization curves and moreover , it enables better understanding of the magnetic ordering in relevant phases . as a typical example we can mention the ferrimagnetic phases .
although the total magnetization of both the fi and fii phases is the same , the results for other quantities indicate a fundamental difference between them .
in fact , in the fi phase the pairs of heisenberg spins create singlet dimers and thus , they do not contribute to the total magnetization which is nonzero due to the fully polarized ising spins only , as already stated before . this observation would suggest that the formation of fi plateau state should be based on the quantum mechanism with a significant influence of local quantum fluctuations .
in contrast to this , the fii phase represents the standard ferrimagnetically ordered phase usually observed in the pure ising systems .
indeed , the generation of the fii plateau state is nothing but the gapped excitation from the nel state , implying the magnetization process with a classical ising - like mechanism .
finally , one should notice that after exceeding the saturation field given by conditions ( [ 11 ] ) and ( [ 12 ] ) , the ground state becomes fully polarized ( sp phase ) and all spins are aligned in the external - field direction .
now , let us investigate the finite - temperature behaviour of the system .
firstly , we take a closer look at the thermal dependencies of total magnetization that are shown in fig.5 . as one can see
, the initial value of total magnetization takes one of three possible values @xmath73 , @xmath74 or @xmath7 for the af , fi ( fii ) or sp phase , respectively .
furthermore , there are two special cases which correspond to the coexistence of the relevant phases . in these cases
, we have obviously obtained @xmath99 or @xmath76 at @xmath83 .
it is also easy to observe here that the most interesting dependencies appear for external fields from the neighbourhood of phase boundaries .
hence , the observed rapid increase ( decrease ) of the magnetization can be attributed to the thermal excitations of huge number of spins which occur due to the competitive influence of both phases . for completeness
, we have also plotted the entropy ( fig.6a ) and the specific heat ( fig.6b-6d ) as a function of temperature .
as one can expect , the entropy of the system does not vanish for the boundary external - field values @xmath100 and @xmath101 , at which the relevant phases coexist in the ground state ( see fig.6a ) .
however , we should mention that for any other external fields the entropy vanishes as the temperature goes to zero .
another quantity which is interesting also from the experimental point of view represents the specific heat .
the thermal variations of this quantity are depicted in fig.6b for the same values of @xmath102 as for entropy in fig.6a .
the displayed behaviour indicates a round schottky - type maximum , whereas the stronger the external field , the flatter and the broader the maximum .
apart from this trivial finding , one can also observe the double - peak specific heat curves ( see figs.6c and 6d ) .
the first peak which occurs in the specific heat curve at lower temperature is evidently closely related to the rapid variation of the magnetization ( compare fig.6c - d with fig.5b ) .
moreover , it turns out that the maximum of this peak can be located approximately in the middle of ferrimagnetic region ( in our case around @xmath103 ) . on the other hand
, the second peak may be thought as a schottky - like peak resulting from the antiferromagnetic short - range order .
indeed , the relevant thermal dependencies for the nearest - neighbouring correlations strongly support this statement . in the following subsection
we examine the spin @xmath75 chain in order to clarify the influence of decorating spin on the magnetic properties of the system .
we start our discussion once again with the analysis of ground state . in order to establish correct ground - state phase boundaries all relevant quantities
have been examined in detail . from this analysis
one can conclude that depending on the ratio between @xmath16 , @xmath104 and @xmath105 , altogether six different phases can appear in the ground state ( see ground - state phase diagrams in figs.7a and 9a - b ) . in this part
, we will firstly describe details of the spin ordering emerging in the appropriate ground - state phases and then , we will show how the scenario of the magnetization process depends on the parameters of the model . in the antiferromagnetic phase ( af )
one finds a perfect antiferromagnetic alignment in the ising sublattice ( i.e. @xmath106 and @xmath107 ) regardless of the strength of the exchange anisotropy @xmath13 .
accordingly , the relevant spin order in the ising sublattice is completely identical to that one in the af phase of the spin @xmath72 chain .
nevertheless , in contrast to the spin @xmath72 case , the antiparallel alignment between nearest - neighbouring heisenberg spins is weakened along @xmath108-axis as @xmath13 increases from zero .
in fact , the correlation @xmath94 tends monotonically from its classical ising value @xmath109 at @xmath110 to @xmath111 for large @xmath13 ( see e. g. fig.8a ) .
in addition , one easily proves also the validity of relation @xmath112 in the whole af region .
this observation would suggest that @xmath13 supports the spin reorientation of heisenberg spin pairs , namely , from the antiparallel oriented heisenberg spin pair ( one spin in @xmath113 state , another one in spin @xmath114 state to be further denoted as the @xmath115 spin pair ) towards the spin pair with both heisenberg spins in the spin @xmath116 state ( the @xmath117 spin pair ) .
however , in the large @xmath13 limit both types of the heisenberg spin pairs ( @xmath115 as well @xmath117 ) are equally well populated and with a high probability also randomly distributed among ising spins .
this suggestion is strongly supported by results for the correlation @xmath96 and staggered magnetization @xmath60 which asymptotically tend to zero as @xmath13 @xmath118 @xmath119 ( fig.8a ) .
finally , one should also notice that all these effects could originate from the onset of the antiferromagnetic short - range - ordering in the @xmath98 plane ( nonzero @xmath97 ) that is the stronger ( up to the value @xmath120 ) , the greater the exchange anisotropy strength @xmath13 .
now , let us turn our attention to both the fi and fii ferrimagnetic phases .
as in the spin @xmath72 case , the nonzero total magnetization of the fi phase arises due to the fully polarized ising spins only ( @xmath121 and @xmath122 ) , while the heisenberg sublattice does not contribute to the total magnetization at all ( @xmath123 and @xmath124 ) .
since the fi phase can arise by increasing the external field from the af phase only ( see fig.7a , 9ab ) , it is of interest to compare the relevant spin ordering in both phases .
actually , one still finds @xmath112 to be valid , however , the nearest - neighbour correlation @xmath125 is weaker ( stronger ) in the fi phase with respect to that one in the af phase ( fig.8d ) .
these results are taken to mean that the number of nearest - neighbouring 00 spin pairs increases ( of course , only up to one half of the total number of pairs ) as one passes through the af - fi phase boundary .
moreover , the appearance of a massive short - range order in the @xmath98 plane strongly implies the relevance of local quantum fluctuations also in the fi phase . on the other hand , in the second ferrimagnetic phase fii
we have found the following results @xmath126 indicating the classical character of this phase that is usually observed also in the pure ising spin systems . as the relevant
spin - ordering is thoroughly analogous to that one in the fii phase of spin @xmath72 chain , for brevity we will not repeat its description here .
the most interesting spin order , however , can be found in the valence - bond phase ( vb ) and the intermediate phase ( ip ) .
really , for instance in the vb phase one finds @xmath127 as one can see , in contrast to the fully polarized ising sublattice ( @xmath128 and @xmath122 ) , each couple of the nearest - neighbouring heisenberg spins consists of one polarized spin ( @xmath114 ) and one spin in the @xmath116 state , i. e. the spin @xmath129 pair .
it is interesting to note that the symmetrization of both heisenberg spin states can be achieved using the valence - bond - solid ( vbs ) picture @xcite .
accordingly , each spin-@xmath130 atom splits into one polarized spin-@xmath7 variable with the fixed projection into the external field direction and one spin-@xmath7 variable with the unfixed projection creating a valence - bond .
thus , it is reasonable to assume that both heisenberg spins interchange their spin states ( tunneling between the @xmath116 and @xmath114 spin states ) and therefore , they effectively act on the surrounding ising spins as the spin-@xmath7 atoms ( see also the result for correlation @xmath96 ) .
however , the antiferromagnetic short - range order in the @xmath98 plane ( nonzero @xmath97 ) , as well as the fact that the vb phase can not be detected in the pure ising system ( i.e. for @xmath131 ) , imply again the obvious influence of the local quantum fluctuations . perhaps the most striking spin alignment has been discovered in the ip phase . among other matters ,
the perfect antiferromagnetic alignment between the ising spins ( @xmath106 and @xmath95 ) has been confirmed in ip phase , hence , the total magnetization of system ( @xmath132 ) is entirely determined by the contribution of heisenberg spins .
anyway , the results for the heisenberg sublattice magnetization @xmath133 , correlation function @xmath134 and quadrupolar momentum @xmath135 , clearly indicate that each couple of the nearest - neighbouring heisenberg atoms comprises of the spin @xmath129 pair . however , since both heisenberg spins are placed in the ip phase between two non - equivalent ising spins ( one up , other down ) , the spin state interchange between both heisenberg atoms would lead to @xmath124 what is in contradiction with our numerical result @xmath136 . on the other hand ,
if the heisenberg atom in spin @xmath114 state would be strictly antiferromagnetically coupled to its nearest - neighbouring ising spin ( spin down ) , then we would have @xmath137 .
nevertheless , our result for the correlation @xmath138 indicates only the partial antiferromagnetic order between ising and heisenberg spins .
this observation would suggest that the spin @xmath129 heisenberg pairs should be , to a certain degree , randomly distributed among ising spins .
finally , as one can expect , in the high field limit the system undergoes a phase transition towards the fully saturated paramagnetic phase ( sp ) . as before , in the sp phase all ising and heisenberg spins are completely aligned towards the external - field direction , thus , in the sp phase one attains @xmath139 now , let us proceed to examine the magnetization process of the system under investigation . for this purpose ,
we have shown in figs.7a and 9a - b the ground - state phase diagrams in the @xmath80 plane together with some typical examples of the magnetization curves ( figs.7b - d and 9c - d ) for three selected values of interaction parameters @xmath140 . in order to provide an independent check of the magnetization scenario from figs.7b - d , the corresponding field - dependencies of the correlation functions and quadrupolar momentum
are displayed in figs.8b - d .
obviously , all the above results are absolutely in accordance with the aforementioned ground - state spin ordering .
moreover , through the comparison of figs.7a and 9a - b one can realize that the nature of magnetization process depends basically on the ratio between the ising and heisenberg interaction parameters .
in fact , the stronger the ising interaction @xmath16 with respect to the heisenberg one @xmath104 , the broader the parameter region corresponding to the classical fii phase .
otherwise , the increasing influence of the heisenberg interaction @xmath104 causes the broadening of regions corresponding to the fi and vb phase , respectively , until the fii phase completely vanishes below @xmath141 , see e. g. fig.9a where the fii phase is already missing .
consequently , in the fi and vb phases one can expect that the effect of local quantum fluctuations plays an important role . although , the system exhibits a stepwise magnetization curve with an abrupt change of the magnetization in the whole range of parameters ( see figs.7b - d and 9c - d ) , there is a fundamental difference between the magnetization process in the classical ising - like regime and that one in the quantual regime . as a matter of fact
, in the former case one observes the double - plateau magnetization curve with the fii plateau state only , i. e. with the fii phase as an intermediate state between the af and sp phase ( see fig.7b ) .
contrary to this , in the latter case one can detect the double- ( fig.7c ) , triple- ( fig.7d ) , or quadruple - plateau ( fig.9c - d ) magnetization curves .
the double - plateau magnetization curves are rather rarely observable , since they arise due to the direct metamagnetic transition from the af phase to the vb phase in a relatively narrow region of @xmath13 only ( see fig.7a ) . however , when the mechanism of the magnetization process is driven by the heisenberg interaction , i. e. under the requirement of sufficiently small @xmath140 and sufficiently large @xmath13 , the triple - plateau magnetization curves with the transitions between af - fi - vb - sp phases are always preferred ( fig.7d ) .
finally , one should also remark that if the condition @xmath142 is satisfied , there also appear the extraordinary quadruple - plateau magnetization curves ( see fig.9b ) with the ip phase in a very narrow region of the external fields .
indeed , we found even two different possibilities for the quadruple - cascade transitions , namely , the af - fi - ip - fii - sp cascade transitions ( fig.9c ) as well as the af - fi - ip - vb - sp ones ( fig.9d ) . to conclude the analysis of the spin @xmath75 chain
, we will also briefly mention the finite - temperature behaviour of the system under consideration . for this purpose ,
the thermal variations of the total magnetization are plotted in fig.10 for @xmath143 and two selected values of the anisotropy @xmath13 . in agreement with the aforementioned arguments
, one observes here three and two field - induced transitions for the anisotropy strengths @xmath144 and @xmath145 , respectively . evidently ,
as the magnitude of external field varies , the various thermal dependencies result from the competition between the ising interaction , heisenberg interaction and magnetic field .
however , the most interesting dependencies arise again for the external fields from the vicinity of the phase boundaries . in such a case , the relevant thermal excitations result in a very rapid change of magnetization due to the competing influence of the phases separated by relevant transition line .
moreover , it turns out that the narrower interval of external fields corresponds to the relevant phase , the more robust change in the magnetization can be observed .
in the present article we have obtained the exact solution of the mixed spin-@xmath7 and spin-@xmath0 ( @xmath146 ) ising - heisenberg chain in the external magnetic field .
the most important result stemming from this study is the confirmation of multistep magnetization process by an exact calculation .
moreover , it has been proved that the character of magnetization process depends essentially on the ratio between the ising and heisenberg interactions , whereas the xxz anisotropy term @xmath13 also allows to control it in deciding manner .
since the presence of the non - diagonal interaction term @xmath147 is responsible for the onset of the local quantum fluctuations , as we have also shown , it basically modifies an otherwise trivial ising - like behaviour .
for example , in the case of the spin @xmath75 chain one finds instead of the double - plateau magnetization curve arising in the pure ising spin system ( @xmath110 ) , the double- , triple- , or even quadruple - plateau magnetization curves in the ising - heisenberg chain with @xmath148 .
altogether , the presented results indicate that the extraordinary rich ground - state phase diagrams result from the competition between the easy - axis interactions @xmath149 and the easy - plane interaction @xmath147
. one should also emphasize that our research on ising - heisenberg chains has been stimulated by the recent experimental works dealing with many quasi-1d mixed - spin chains @xcite .
although , we are not aware of any quasi-1d system in that two kinds of magnetic ions regularly alternate with @xmath6 periodic fashion ( aabaab ... ) , the recent progress in molecular engineering @xcite supports our hope that the synthesis of such a polymeric chain should be possible in the near future .
structural derivatives of a novel polymeric chain recently reported by mukherjeee _
et al _ @xcite , seem to be the most promising candidates from this point of view .
in fact , the crystal structure of the above mentioned polymeric chain consists of the spin-1/2 @xmath150 dimers linked through @xmath151 monomers .
unfortunately , as a consequence of the square - planar coordination of @xmath151 ions in @xmath152^{2-}$ ] bridging groups , the @xmath151 metal ions are diamagnetic and thus , they do not contribute to the magnetism .
finally , it should be stressed that the applied mapping transformation technique does not require any restriction to the dimensionality of the spin system and hence , it can be straightforwardly generalized to the mixed ising - heisenberg lattices in two- and three - dimensions , too @xcite .
the authors would like to thank referee for useful comments . +
this work was supported by the vega grant no .
1/9034/02 and apvt grant no .
20 - 009902 .
explicit expressions for the functions @xmath40-@xmath153 .
a ) spin-1/2 chain : @xmath154 b ) spin-1 chain : @xmath155 where the expressions @xmath156 and @xmath157 are given by : @xmath158 \bigr \ } , \nonumber \\ w_2 \ ! \ ! \ ! & = & \ ! \ ! \ ! \sum_{n=0}^2 \exp \bigl \{- 2 \beta p_2 \cos \bigl [ ( \phi_2 + 2 \pi n)/3 \bigr ] \bigr \ } , \nonumber \\ p_1 ^ 2 \ ! \ ! \ ! & = & \ ! \ ! \ ! ( j/3)^2 + 2 ( j \delta)^2/3 , \hspace{1.5 cm } p_2 ^ 2 = ( j/3)^2 + 2 ( j \delta)^2/3 + j_1 ^ 2/3 , \nonumber \\ q_1 \ !
\ ! & = & \ ! \ ! \ !
( j/3)^3 + j ( j \delta)^2/3 , \hspace{1.4 cm } q_2 = ( j/3)^3 + j ( j \delta)^2/3 - j_1 ^ 3 , \nonumber \\ \phi_1 \ ! \ ! \ ! & = & \ ! \ ! \ !
\arctan \big ( \sqrt{p_1 ^ 6 - q_1 ^ 2}/q_1 \bigr ) , \hspace{0.7 cm } \phi_2 = \arctan \bigl ( \sqrt{p_2 ^ 6 - q_2 ^ 2}/q_2 \bigr ) .
\nonumber \label{a3}\end{aligned}\ ] ] hida k 1994 _ j. phys .
_ * 63 * 2359 ; + okamoto k 1995 _ solid state commun . _
* 98 * 245 ; + tonegawa t , nakao t and kaburagi h 1996 _ j. phys .
jpn . _ * 65 * 3317 ; + pati s k , ramasecha s and sen d 1997 _ phys .
b _ * 55 * 8894 ; + sakai t and takahashi m 1998 _ phys . rev .
b _ * 57 * r3201 ; + totsuka k 1997 _ phys . lett .
a _ * 228 * 103 ; + totsuka k 1998 _ phys .
b _ * 57 * 3454 ; + sakai t and yamamoto s 1999 _ phys . rev . b _ * 60 * 4053 ; + yamamoto s and sakai t 2000 _ phys .
b _ * 62 * 3795 ; + okamoto k and kitazawa a 1999 _ j. phys . a : math .
gen . _ * 32 * 4601 ; + kitazawa a and okamoto k 1999 _ j. phys . : condens
. matter _ * 11 * 9765 ; + koga a , okunishi k and kawakami n 2000 _ phys .
b _ * 62 * 5558 ; + sakai t and okamoto k 2002 _ phys . rev .
b _ * 65 * 214403 ; + schulenburg j and richter j 2002 _ phys .
b _ * 65 * 054420 ; + chen x y , jiang q and wu y z 2002 _ solid state commun . _
* 121 * 641 bethe h a 1931 _ z. phys . _ * 71 * 205 ; + hulthn l 1938 _ ark . mat .
* 26a * 1 ; + orbach r 1958 _ phys .
_ * 112 * 309 ; + des cloizeaux j and gaudin m 1966 _ j. math .
* 7 * 1384 ; + yang c n and yang c p 1966 _ phys .
rev . _ * 150 * 321 ; + yang c n and yang c p 1966 _ phys .
rev . _ * 150 * 327 ; + yang c n and yang c p 1966 _ phys .
rev . _ * 151 * 258 ; + majumdar c k and ghosh d k 1969 _ j. math .
* 10 * 1388 ; + majumdar c k and ghosh d k 1969 _ j. math .
* 10 * 1399 ; + majumdar c k 1970 _ j. phys .
c : solid st .
phys . _ * 3 * 911 ; + kirillov a n and reshetikhin n yu 1987 _ j. phys . a : math
. gen . _ * 20 * 1565 ; + kirillov a n and reshetikhin n yu 1987 _ j. phys . a : math
. gen . _ * 20 * 1587 ; + parkinson j b 1989 _ j. phys . : condens
. matter _ * 1 * 6709 ; + long m w and fehrenbacher r 1990 _ j. phys . : condens
* 2 * 2787 ; + long m w and siak s 1990 _ j. phys . :
condens . matter _
* 2 * 10321 ; + de vega h j and woynarovich f 1990 _ j. phys . a : math .
gen . _ * 23 * 1613 ; + kubo
k 1993 _ phys .
b _ * 48 * 10552 de vega h j and woynarovich f 1992 _ j. phys . a : math .
gen . _ * 25 * 4499 ; + martins m j 1993 _ j. phys . a : math .
* 26 * 7301 ; + drfel b d and meissner st 1996 _ j. phys . a : math .
gen . _ * 29 * 1949 ; + drfel b d and meissner st 1996 _ j. phys . a : math .
* 29 * 6471 ; + drfel b d and meissner st 1997 _ j. phys . a : math .
* 30 * 1931 ; + drfel b d and meissner st 1998 _ j. phys .
a : math . gen . _ * 31 * 61 ; + drfel b d and meissner st 1999 _ j. phys . a : math .
* 32 * 41 gleizes a and verdauger m 1981 _ j. am .
* 103 * 7373 ; + verdauger m , julve m , michalowicz a and kahn o 1983 _ inorg .
_ * 22 * 2624 ; + drillon m , gianduzzo j c and gorges r 1983 _ phys .
a _ * 96 * 413 ; + gleizes a and verdauger m 1984 _ j. am .
soc . _ * 106 * 3727 ; + verdauger m , gleizes a , renard j p and seiden j 1984 _ phys .
b _ * 29 * 5144 ; + gorges r , drillon m and currely j 1985 _ j. appl .
phys . _ * 58 * 914 ; + kahn o 1987 _ struct .
bonding _ ( berlin ) * 68 * 91 ; + pei y , verdauger m , kahn o , sletten j and renard j p 1987 _ inorg .
* 26 * 138 ; + kahn o , pei y , verdauger m , renard j p and sletten j 1988 _ j. am .
soc . _ * 110 * 782 ; + pei y , kahn o , sletten j , renard j p , georges r , gianduzzo j c , curely j and xu q 1988 _ inorg .
_ * 27 * 47 ; + caneschi a , gatteschi d , renard j p , rey p and sessoli r 1989 _ inorg .
_ * 28 * 1976 ; + van koningsbruggen p j , kahn o , nakatani k , pei y , renard j p , drillon m and leggol p 1990 _ inorg . chem . _
* 29 * 3325 ; + kahn o , pei y and journaux y 1995 _ inorganic materials _ ed d w bruce and d ohare ( new york : wiley ) ; + hagiwara m , minami k , narumi y , tatani k and kindo k 1998 _ j. phys . soc .
jpn . _ * 67 * 2209 willet r d , wang z , molnar s , brewer k , landee c p , turnbull m m and zhang w 1993 _ mol . cryst .
* 233 * 277 ; + kahn o 1993 _ molecular magnetism _
( new york : verlag ) ; + verdauger m 2001 _ polyhedron _ * 20 * 1115 ; + ernk j , orend m , potok i , chomi j , orendov a , skorepa j and feher a 2002 _ coord .
rev . _ * 224 * 51 * part of the doubly decorated mixed - spin chain .
the black circles denote the spin-@xmath7 ising atoms of sublattice @xmath34 and the gray ones represent the decorating spin-@xmath0 heisenberg atoms of sublattice @xmath159 .
the ellipse demarcates a typical bond described by the hamiltonian @xmath22 introduced in eq .
( [ r2 ] ) . * ground - state phase boundaries in the @xmath13 - @xmath102 plane for @xmath81 and @xmath82 . * low - temperature ( @xmath160 ) magnetization curves for : a ) @xmath81 and @xmath161 ; b)-d ) @xmath162 and @xmath163 .
the solid and dotted ( dashed ) lines represent the total magnetization per one site and the single - site magnetization of the ising ( heisenberg ) sublattice , respectively . * low - temperature behaviour ( @xmath160 ) of several correlation functions : @xmath94 and @xmath164 ( dashed lines ) , @xmath97 ( dotted lines ) and @xmath96 ( solid lines ) .
+ a ) plot of the correlation functions against the anisotropy @xmath13 for the system without external field ; + b)-d ) field - dependence of the correlation functions depicted for @xmath162 and various anisotropies @xmath165 , @xmath166 and @xmath167 . * thermal variations of the total magnetization for some typical values of the external field , @xmath162 and two selected anisotropies @xmath165 and @xmath167 , respectively . * thermal behaviour of the system for @xmath162 , @xmath168 and several values of external field .
+ a ) entropy plot versus temperature ; + b)-d ) variations of the specific heat with temperature . *
ground - state phase diagram together with some typical examples of low - temperature magnetization curves when the ratio @xmath81 ; + a ) ground - state phase diagram in the @xmath80 plane ; + b)-d ) some typical examples of the low - temperature ( @xmath169 ) magnetization curves for various anisotropies @xmath13 .
* low - temperature ( @xmath169 ) behaviour of several correlation functions @xmath94 ( dashed ) , @xmath97 ( dotted ) , @xmath96 ( solid ) , @xmath164 ( dashed - dotted ) , quadrupolar momentum @xmath67 ( dashed - dotted - dotted ) and staggered sublattice magnetization @xmath60 ( solid line ) . + a ) zero - field variations of relevant quantities with anisotropy @xmath13 ; + b)-d ) field - dependencies of correlations for selected values of @xmath13 .
* ground - state phase diagrams together with some typical examples of low - temperature magnetization curves : + a)-b ) ground - state phase diagram in the @xmath80 plane for @xmath170 and @xmath82 ; + c)-d ) some typical examples of the low - temperature ( @xmath169 ) magnetization curves when @xmath162 and @xmath171 and @xmath172 , respectively .
* thermal variations of the total magnetization for @xmath81 and two selected values of @xmath13 . | the mapping transformation technique is applied to obtain exact results for the spin-1/2 and spin-@xmath0 ( @xmath1 ) ising - heisenberg antiferromagnetic chain in a presence of external magnetic field . within this scheme
, a field - induced first - order metamagnetic phase transition resulting in multiplateau magnetization curves , is investigated in detail .
it is found that the scenario of the plateau formation depends fundamentally on the ratio between ising and heisenberg interaction parameters , as well as on the xxz heisenberg exchange anisotropy strength .
pacs:05.50.+q , 75.10.jm , 75.10.hk ising - heisenberg model ; exact solution ; quantum antiferromagnetism |
measurement of gravitational waves from inspiraling compact binaries ( composed of neutron stars and/or black holes ) by forthcoming laser interferometric detectors ( such as ligo @xcite and virgo @xcite ) will rely heavily on theoretical templates obtained from approximations to the exact laws of general relativity @xcite .
one possible method to calculate these templates , which has been pushed to a high degree of accuracy @xcite , is the post - newtonian approximation , which is based upon the assumption that the orbital motion is slow .
another approach is black - hole perturbation theory @xcite , which is accurate only in the unrealistic situation of systems with small mass ratios .
currently , the post - newtonian approach appears to be the most promising for generating the templates .
unfortunately , however , there is considerable evidence that the post - newtonian approximation has poor convergence @xcite , if it converges at all .
it is important to understand the reasons for the poor convergence .
although this issue is still an open one , some preliminary steps were taken in this direction in a previous paper @xcite .
the amount of energy radiated per unit time ( denoted @xmath6 ) by a binary system in circular orbit can be expressed as an infinite sum over the multipole moments of the gravitational - wave field , which are related to the behaviour of the source . in a slow - motion approximation , each multipole moment is then expressed as a post - newtonian expansion in powers of @xmath0 , the ratio of the orbital velocity to the speed of light . since higher - order moments come with a higher power of the orbital velocity ,
the post - newtonian approximation to @xmath6 consists of truncating the multipole sum at a given multipole order , and in calculating the contributing moments to the appropriate post - newtonian order .
we will refer to this expansion of @xmath6 in powers of @xmath0 as the `` true post - newtonian series '' . at present , only a few terms of this post - newtonian series have been calculated @xcite , and the issue of its converge is not at all clear . however , since post - newtonian corrections can be computed for _ all _ the multipole moments , the simpler issue of the convergence of the _ multipole _ sum can be addressed . in reference
@xcite , the convergence of the multipole series was studied in the restricted context of a `` bare multipole truncation '' , in which the multipole moments were calculated only to leading - order in a post - newtonian expansion .
we refer to these bare moments as `` newtonian '' , and the bare multipole truncation therefore consists of discarding all post - newtonian corrections to each of the multipole moments . when an approximation to @xmath6 is constructed by summing over all these bare moments , it is found that the series converges for all orbital velocities less than a critical value .
this value depends on the mass ratio of the system ; in the case of very small mass ratios , to which we specialize in this paper , the condition for convergence is @xmath7 , where @xmath2 is the base of natural logarithms .
the convergent interval includes all physically relevant values for the orbital velocity , which must be smaller than its value at the innermost stable circular orbit , @xmath8 .
although the bare multipole truncation is an unphysical approximation which compares poorly with the numerically `` exact '' results @xcite , this analysis reveals that an eventual failure of the true post - newtonian series to converge can not originate from summing over the newtonian part of all the multipole moments . in this paper
, we extend the results of reference @xcite to a `` dressed multipole truncation '' , in which the leading - order moments are corrected with terms of relative order @xmath3 ( first post - newtonian , or 1pn , terms ) and @xmath4 ( 1.5pn terms ) .
this improved analysis relies on a previous paper of ours @xcite , in which the equations of black - hole perturbation theory are integrated for an arbitrary perturbing stress - energy tensor @xmath9 ( the unperturbed spacetime is assumed to be schwarzschild ) . applying the results of this paper to the particular case of a particle moving on a circular orbit returns the desired post - newtonian corrections to the bare multipole moments derived in reference @xcite .
the dressed multipole truncation gives three separate series for @xmath6 .
the first is the bare multipole series , in which we sum over the newtonian part of all the multipole moments .
the second and third series incorporate the 1pn and 1.5pn corrections to the bare moments , respectively .
we find that each of the series converges for @xmath1 , and we conclude that inclusion of these post - newtonian corrections does not affect the convergence of the multipole sum .
this shows that an eventual failure of the true post - newtonian series to converge can not originate from summing over the newtonian , 1pn , and 1.5pn part of all the multipole moments .
the following sections of the paper provide the details of the calculation .
the dressed multipole truncation is introduced in sec . [ 2 ] , and the convergence of the resulting series is established in sec .
[ 3 ] . finally , concluding remarks are provided in sec .
in this section , we extend the bare multipole calculation of reference @xcite by calculating the 1pn and 1.5pn corrections to the multipole moments of the gravitational - wave field .
such corrections have been calculated for general sources , first by blanchet @xcite using post - newtonian theory , and then by us @xcite using black - hole perturbation theory . to carry out this calculation
, we apply the results of reference @xcite ( hereafter referred to as lp ) to the specific case of a point particle moving on a circular orbit around a schwarzschild black hole .
we denote by @xmath10 the mass of the particle , and by @xmath11 the mass of the black hole .
it is assumed that @xmath12 .
we also denote the orbital radius by @xmath13 , and the angular velocity is @xmath14 .
finally , we define the orbital velocity by @xmath15 . throughout
we use units such that @xmath16 .
the gravitational - wave luminosity is given by @xcite @xmath17 , \label{power}\ ] ] where @xmath18 and @xmath19 are the mass and current multipole moments , respectively , while @xmath20 denotes retarded time .
( because the orbit is fixed , @xmath6 does not actually vary with @xmath20 . ) to evaluate @xmath6 , we shall use equations ( 5.37)(5.39 ) of lp , which give the fourier transform of the moments , @xmath21 and @xmath22 .
these equations incorporate post - newtonian corrections of up to order @xmath23 , relative to the leading - order , newtonian expressions . to evaluate the multipole moments
, we must first compute the source terms , which are constructed from the stress - energy tensor of the orbiting particle @xcite . using equations ( 5.24)(5.27 ) of lp
, we find that the only nonzero source terms are @xmath24 , @xmath25 , @xmath26 , and @xmath27 , where @xmath28 , \qquad \nonumber \\ & & \label{source } \\ _ 0 t & = & \frac{v^2}{2 } \left [ 1 + \frac { 3v^2}{2 } + o(v^4 ) \right ] , \qquad \delta = \frac { \mu}{r_0 ^ 2}\ , \delta ( r - r_0 ) \delta ( \cos \theta ) \delta ( \phi - \omega t ) , \nonumber \end{aligned}\ ] ] and @xmath29 .
the second step is to substitute these results into equations ( 5.37)(5.39 ) of lp .
after a few steps of algebra , the mass and current multipole moments become @xmath30^{1/2 } { \mathcal{t}}_l ( \omega)\ , ( - i \omega r_0)^l\ , a_l ( \omega ) \nonumber \\ & & \mbox { } \times ( r+ 2_0t)\ , y_{lm}({\textstyle \frac{\pi}{2}},0)\ , \delta(\omega - m \omega ) , \label{mass_1 } \\ \tilde{\cal s}_{lm } ( \omega ) & = & \frac { - 16 \pi i \mu}{(2l+1 ) ! ! } \left ( \frac{l+2}{l-1 } \right ) ^{1/2 } { \mathcal{t}}_l^{\ # } ( \omega)\ , ( - i \omega r_0)^l\ , b_l ( \omega ) \nonumber \\ & & \mbox { } \times \left [ { _ { -1}j } { _ { -1 } y_{lm}}({\textstyle \frac{\pi}{2}},0 ) + { _ 1j } { _ 1 y_{lm } } ( { \textstyle
\frac{\pi}{2}},0 ) \right ] \ , \delta(\omega - m\omega),\end{aligned}\ ] ] where @xmath31 while @xmath32 and @xmath33 are the spherical harmonics of spin weight 1 and @xmath34 , respectively @xcite .
the third step is to invert the fourier transform , which is trivial because of the delta function in the integrand . this yields @xmath35^{1/2 } { \mathcal{t}}_l ( m \omega)\ , ( - i m \omega r_0)^l\ , a_l ( m \omega ) \nonumber \\ & & \mbox { } \times ( r+ 2_0t)\ , y_{lm } ( { \textstyle \frac{\pi}{2}},0)\ , e^{- i m \omega u } , \label { ilm } \\ { \cal s}_{lm }
( u ) & = & \frac { - 16 \pi i \mu}{(2l+1 ) ! ! }
\left ( \frac { l+2}{l-1 } \right ) ^{1/2 } { \mathcal{t}}_l^{\ # } ( m \omega)\ , ( - i m \omega r_0)^l\ , b_l ( m \omega ) \nonumber \\ & & \mbox { } \times \left [ { _ { -1}j } { _ { -1 } y_{lm } } ( { \textstyle \frac{\pi}{2}},0 ) + { _ 1j } { _ 1
y_{lm}}({\textstyle \frac{\pi}{2}},0 ) \right ] \ , e^{- i m \omega u}. \label{slm}\end{aligned}\ ] ] it is now useful to write explicit expressions for the spin - weighted spherical harmonics @xcite .
first , if @xmath36 is even , then @xmath37^{1/2}}{(l - m)!!(l+m ) ! ! } , \label { ylm}\ ] ] while @xmath38 if @xmath36 is odd .
second , if @xmath36 is odd , then @xmath39^{1/2 } \frac { ( l+m)!!(l - m ) ! !
} { \left [ ( l+m)!(l - m ) !
\right]^{1/2 } } ; \label { ylm_-1}\ ] ] although @xmath40 for @xmath36 even , its expression will not be needed . finally , we note that @xmath41 .
looking at this last result and the fact that @xmath42 , we see that the quantity within the square brackets of equation ( [ slm ] ) may be simplified as follows : it vanishes identically for @xmath36 even , and it is equal to @xmath43 when @xmath36 is odd . using this while substituting equations ( [ ilm ] ) and ( [ slm ] ) into equation ( [ power ] ) , we obtain @xmath44 where @xmath45 is the quadrupole - formula result
. we also have @xmath46 where @xmath47 ^ 2 \label{zetaeven}\ ] ] if @xmath36 is even , while @xmath48 ^ 2 \label{zetaodd}\ ] ] if @xmath36 is odd .
the symmetry @xmath49 , used in equation ( [ power_1 ] ) , can easily be established from the previous expressions .
equations ( [ power_1])([zetaodd ] ) constitute the dressed multipole truncation of the gravitational - wave luminosity .
the right - hand side of equation ( [ power_1 ] ) is an infinite series in powers of @xmath50 , and our next task is to examine the convergence of this series .
the velocity dependence of all the terms in equation ( [ power_1 ] ) is made clear by substituting equations ( [ source ] ) and ( [ ab ] ) , which gives @xmath51 the various coefficients are @xmath52 ^ 2 , \label{alm } \\
b_{lm } & = & -\left [ \frac { ( l+9)m^2}{(l+1)(2l+3 ) } + 2l -1 \right]\ , a_{lm } , \label{blm } \\
c_{lm } & = & 2 \pi m\ , a_{lm } , \label{clm } \\
d_{lm } & = & 5 \pi\ , \frac { m^{2(l+1)}}{(2l+1)!!^2 } \frac { l+2}{l-1 } \left [ { _ { -1}y_{lm } } ( { \textstyle \frac{\pi}{2}},0 ) \right ] ^2 , \label{dlm } \\ e_{lm } & = & -\left [ \frac { ( l+4)m^2}{(l+2)(2l+3 ) } + \frac { 2(l-1)(l+2)}{l } - 3 \right]\ , d_{lm } , \label{elm } \\ f_{lm } & = & 2 \pi m\ , d_{lm}. \label{flm } \end{aligned}\ ] ] we now express equation ( [ power_2 ] ) in the following form , which displays a more transparent grouping of terms : @xmath53 \right \ } , \label{edot}\ ] ] where @xmath54 , \label{s_1 } \\
s_2(v ) & = & \sum_{l=3}^{\infty } v^{2(l-2 ) } b_{ll } \left [ 1 + \frac{1}{b_{ll } } \sum_{m=1}^{l-1 } \left ( b_{lm } + e_{l-1,m } \right ) \right ] , \label{s_2 } \\
s_3(v ) & = & \sum_{l=3}^{\infty } v^{2(l-2 ) } c_{ll } \left [ 1 + \frac{1}{c_{ll } } \sum_{m=1}^{l-1 } \left ( c_{lm } + f_{l-1,m } \right ) \right ] .
\label{s_3 } \end{aligned}\ ] ] the dressed multipole truncation therefore splits @xmath6 into three separate series : @xmath55 , the bare multipole series , @xmath56 , the 1pn correction , and @xmath57 , the 1.5pn correction .
it is important to point out that in equations ( [ s_1])([s_3 ] ) , the sums over @xmath58 are restricted to values such that @xmath36 is _
even_. the convergence of the dressed multipole series may be established by analyzing the convergence of each of the partial series .
we begin with @xmath55 , the bare multipole series of reference @xcite .
its convergence is established in the following manner .
first , it is shown ( numerically ) that the terms in the square brackets of equation ( [ s_1 ] ) approach the approximate value 1.01 as @xmath59 .
this implies that the behaviour of the series for large @xmath60 is dominated by the factor in front of the square brackets .
the series may therefore be approximated by @xmath61 , which is shown ( using the cauchy ratio test ) to converge for @xmath62 the convergence of @xmath63 is established in the same manner .
first , one shows numerically that the terms in the square brackets of equation ( [ s_2 ] ) also approach 1.01 as @xmath59 .
the series is then approximated by @xmath64 , and the convergence of this series is again analyzed via the cauchy ratio test . using equation ( [ blm ] ) , we find that @xmath65 is equal to @xmath66 multiplied by a ratio of terms quadratic in @xmath60 . in the limit @xmath59 , @xmath67 and we immediately conclude that the convergence of @xmath63 is also determined by equation ( [ v_range ] ) .
the same is true also for @xmath68 .
again , it can be shown that the terms in the square brackets of equation ( [ s_3 ] ) approach 1.01 as @xmath59 .
the series is then approximated by @xmath69 . from equation ( [ clm ] ) we see that in the limit @xmath59 , the ratio of successive coefficients is equal to @xmath66 .
this allows us to conclude that the convergence of @xmath68 is also determined by equation ( [ v_range ] ) . in summary ,
when 1pn and 1.5pn corrections are added to the bare multipole series , @xmath6 splits naturally into three separate series , which all converge for @xmath70 .
therefore , the dressed multipole truncation of @xmath6 must also converge in this interval .
we note that this convergence test ignores the fact that the combined series @xmath71 is alternating at large @xmath60 .
however , establishing the convergence of the alternating series using the leibnitz criterion does not result in an improvement on equation ( [ v_range ] ) .
furthermore , a numerical analysis of the convergence of @xmath72 confirms equation ( [ v_range ] ) .
we have shown that for circular binaries with small mass ratios , the dressed multipole truncation leads to a convergent series for the gravitational - wave luminosity , for all physically relevant values of the orbital velocity .
it appears likely that this conclusion will remain valid for binaries of comparable masses , but a separate calculation is required to confirm this .
our results imply that the poor convergence of the true post - newtonian series does not originate from summing over the newtonian , 1pn , and 1.5pn part of all the multipole moments .
we see in figure 1 that the dressed - multipole series leads to a significant improvement in accuracy , compared with the bare - multipole series . by including the negative 1pn corrections and the ( smaller ) positive 1.5pn corrections ,
our approximation to the luminosity coincides ( within 1 % ) with the numerically `` exact '' results @xcite over the interval @xmath73 .
one would therefore expect that the incorporation of further corrections would lead to a series that converges everywhere in the region of physical interest , and stays accurate over a larger interval .
meanwhile , the issue of the convergence of the true post - newtonian series remains an open one .
this research was supported by the natural sciences and engineering research council of canada . conversations with clifford
will were greatly appreciated .
abramovici a , althouse w e , drever r w p , grsel y , kawamura s , raab f j , shoemaker d , sievers l , spero r e , thorne k s , vogt r e , weiss r , whitcomb s e and zucker m e 1992 _ science _ * 256 * , 325 bradaschia c , del fabbro r , di virgilio a , giazotto a , kautzky h , montelatici v , passuello d , brillet a , cregut o , hello p , man c n , manh p t , marraud a , shoemaker d , linet j y , barone f , di fiore l , milano l , russo g , aguirregabiria j m , bel h , duruisseau j p , le denmat g , tourrenc ph , capozzi m , longo m , lops m , pinto i , rotoli g , damour t , bonazzola s , marck j a , gourghoulon y , holloway l e , fuligni f , iafolla v and natale g 1990 _ nucl
. instrum . & methods _ * a289 * , 518 cutler c and flanagan e e 1994 _ phys .
d _ * 49 * 2658 poisson e and will c m 1995 _ phys .
52 * 848 blanchet l , damour t , iyer b r , will c m and wiseman a g 1995 _ phys .
lett . _ * 74 * , 3515 will c m and wiseman a g 1996 _ phys .
d _ * 54 * , 4813 blanchet l 1996 _ phys .
d _ * 54 * 1417 tanaka t , tagoshi h and sasaki m 1996 _ prog .
_ 96 , 1087 leonard s w and poisson e 1997 _ phys .
, in press ( gr - qc/9705014 ) poisson e 1995 _ phys .
d _ * 52 * , 5719 ; erratum and addendum 1997 _ ibid .
_ * 55 * , 7980 simone l e , poisson e and will c m _ phys . rev
. d _ * 52 * , 4481 simone l e , leonard s w , poisson e and will c m 1997 _ class .
quantum grav . _
* 14 * 237 poisson e and sasaki m 1995 _ phys .
d _ * 51 * 5753 blanchet l 1995 _ phys .
d _ * 51 * 2559 thorne k s 1980 , _ rev .
phys . _ * 52 * 299 poisson e 1993 _ phys .
d _ * 47 * 1497 goldberg j n , macfarlane a j , newman e t , rohrlich f and sudarshan e c g 1967 _ j. math .
_ * 8 * 2155 | the gravitational radiation originating from a compact binary system in circular orbit is usually expressed as an infinite sum over radiative multipole moments . in a slow - motion approximation , each multipole moment is then expressed as a post - newtonian expansion in powers of @xmath0 , the ratio of the orbital velocity to the speed of light .
the `` bare multipole truncation '' of the radiation consists in keeping only the leading - order ( newtonian ) term in the post - newtonian expansion of each moment , but summing over _ all _ the multipole moments . in the case of binary systems with small mass ratios ,
the bare multipole series was shown in a previous paper [ simone _ et al . _
1997 , _ class .
quantum grav .
_ * 14 * , 237 ] to converge for all values @xmath1 , where @xmath2 is the base of natural logarithms .
( these include all physically relevant values for circular inspiral . ) in this paper , we extend the analysis to a `` dressed multipole truncation '' of the radiation , in which the leading - order moments are corrected with terms of relative order @xmath3 ( first post - newtonian , or 1pn , terms ) and @xmath4 ( 1.5pn terms ) .
we find that the dressed multipole series converges also for all values @xmath1 , and that it coincides ( within 1 % ) with the numerically `` exact '' results for @xmath5 .
although the dressed multipole series converges , it is an unphysical approximation , and the issue of the convergence of the true post - newtonian series remains uncertain .
however , our analysis shows that an eventual failure of the true post - newtonian series to converge can not originate from summing over the newtonian , 1pn , and 1.5pn part of all the multipole moments . |
let us suppose that we are dealing with a system with deterministic dynamics , on which external noise is acting .
that means that , in case we could remove the noise ( by switching it off if possible , by isolating the system , etc ) , its equations of motion would be purely deterministic : @xmath2 now , let us consider that gaussian white noise is acting on the system . to include its effects
, we add to ( [ eq : deterministic ] ) a term @xmath3 : @xmath4 where the functions of @xmath5 are determined by some physical considerations , and the components of @xmath6 are independent wiener processes .
a wiener process @xcite is a gaussian stochastic process , almost surely continuous , where non - overlapping increments are independent , and with @xmath7 equation ( [ eq : stochastic ] ) can be written as @xmath8 an issue arises when defining the integral over the wiener processes .
if we take a partition @xmath9 of the interval @xmath10 $ ] ( where @xmath11 ) , and being @xmath12 a value between @xmath13 and @xmath14 , @xmath15 , we would define the integral of any function @xmath16 as @xcite @xmath17,\ ] ] where @xmath18 .
the issue is that the limit above is different for different choices of @xmath19 .
the most common interpretations are the ito scheme , in which the function is evaluated at the left - hand endpoint of each sub - interval ( @xmath20 ) , and the stratonovich scheme , in which the function is evaluated at the midpoint of each sub - interval ( @xmath21 ) .
we usually write the stratonovich calculus with a circle before @xmath22 .
the solutions to the stochastic differential equations are different for both interpretations , given that the stochastic integrals have a different value for each one .
for example , the integral bellow in the ito calculus is @xcite , @xmath23 - \frac{1}{2}(t - t_0),\ ] ] while for stratonovich calculus , @xmath24.\ ] ] chosen one interpretation , one can always do the calculus in the other interpretation by modifying @xmath25 with the _ ito - stratonovich drift correction formula _ :
the two equations bellow @xmath26 have the same solution if their drifts fulfil the relationship @xmath27 when the equations ( [ eq : stochastic ] ) model a physical system , the white noise " @xmath22 is actually an idealization of noise with a small correlation time : the autocorrelation function of @xmath28 is not really a delta function , but a sharply peaked one ; i.e. , its correlation time is small , but positive .
therefore , the stochastic function @xmath28 is actually not singular , and the stochastic differential equations ( [ eq : stochastic ] ) have a well - defined solution . in the limit of correlation time of @xmath28
going to zero , such solution tends to the integration of the equations with the stratonovich interpretation ( see more details and a discussion in chapter ix.5 of @xcite , _ discussion of the ito - stratonovich dilemma _ ) .
that is why we will use the stratonovich interpretation .
we will consider three light - weight ( in terms of the complication of the formulae ) numerical integration methods leading to the stratonovich interpretation .
they all require using a sample of the discretized wiener process at each integration step .
let @xmath29 be the length of the time step used for the numerical integration .
therefore , according to the properties of the wiener process ( [ wiener_process ] ) , we need a sample of independent random variables @xmath30 at each of the values of the time @xmath31 in the discretization , distributed as @xmath32 , where @xmath33 stands for the normal distribution with mean @xmath34 and variance @xmath35 .
one of the simplest discretization schemes leading to the stratonovich interpretation is the heun method .
this is a predictor - corrector method : given the value of @xmath36 at a time @xmath37 of the discretization , we first obtain the predictors , or supporting values , with the euler integration scheme @xmath38 where @xmath39 .
then , we obtain @xmath40 as @xmath41\ , h\nonumber\\ & & \quad + \frac{1}{2}\left [ g_{ij } ( x(t_n ) , t_n)+ g_{ij } ( \bar{x } ( t_{n+1 } ) , t_{n+1})\right]\ , \xi_j(t_n).\end{aligned}\ ] ] note that we are using einstein s tensor convention of summing over repeated indices .
the milstein scheme requires the use of the derivatives of the diffusion coefficients . in the general case ,
it reads @xmath42 where all the terms at the right hand side of ( [ milstein_general ] ) are taken at time @xmath37 , and @xmath43 is a multiple stratonovich integral @xmath44 ( when using the ito interpretation , the corresponding multiple ito integral is defined in the same way as in ( [ multiple_strat_integral ] ) , but with the ito interpretation , and written as @xmath45 . ) these integrals can not be easily expressed in terms of the increments @xmath46 and @xmath47 of the components of the wiener process .
nevertheless , we can consider the particular case of diagonal noise , where each component of the system has its own , independent , noise , i.e. : there are as many independent wiener processes ( number of components of @xmath6 ) as the number of variables in the system ( the number of components of @xmath36 ) , each component @xmath48 of @xmath36 is affected only by the corresponding component @xmath49 of the wiener process , and the diagonal diffusion coefficient @xmath50 depends only on @xmath48 and maybe on time ( but not on the other components of @xmath36 ) .
in such a case , @xmath51 in ( [ milstein_general ] ) is not equal to zero only where @xmath52 , @xmath53 only where @xmath54 , and @xmath55 only where @xmath56 .
putting all things together , the last term in ( [ milstein_general ] ) is equal to @xmath57 ( it is clear that we are not summing over @xmath58 in the expression above , given that @xmath58 is an external index in ( [ milstein_general ] ) ; we will omit this remark wherever it would be easy to figure out over which repeated indices we are not summing up . ) regarding the last factor in the product above , ( [ multiple_strat_integral ] ) becomes for @xmath59 , @xmath60\circ \mathrm{d}w_i ( s_1)\nonumber\\ & = & \int_{t_n}^{t_{n+1 } } w_i ( s_1)\circ \mathrm{d}w_i ( s_1 ) - w_i(t_n)\int_{t_n}^{t_{n+1}}\circ\ , \mathrm{d}w_i ( s_1)\nonumber\\ & = & \frac{1}{2}[w_i(t_{n+1})^2 - w_i(t_n)^2 ] - w_i(t_n ) [ w_i(t_{n+1})-w_i(t_n ) ] \nonumber\\ & = & \frac{1}{2}[w_i(t_{n+1 } ) - w_i(t_n)]^2.\end{aligned}\ ] ] where ( [ integral_stratonovich ] ) has been used at the third equality .
therefore , for the numerical integration , we set @xcite @xmath61 as a result , for diagonal noise the milstein method ( [ milstein_general ] ) becomes , in the stratonovich interpretation , @xmath62 it is also possible to simplify the general expression ( [ milstein_general ] ) for the case of commutative noise @xcite , a case slightly more general than our case of diagonal noise .
the milstein method above requires the analytic specification of the first derivative of the diffusion term
. this can be sometimes inefficient , either because the analytic expression of the derivative is highly complex , or because we are trying many different functions for @xmath63 , etc .
in such cases , we can use a numerical approximation for the derivative of @xmath63 for use in the milstein scheme . here , we will only give the expression for diagonal noise .
the formulae for the ito interpretation can be seen on equations ( 1.3 ) and ( 1.4 ) of chapter 11 in @xcite .
first of all , the supporting values are obtained as @xmath64 then , starting from ( [ milstein_diagonal ] ) for the stratonovich interpretation , we get @xmath65\ , \xi_i^{\ 2}.\ ] ]
for each method , we are interested in checking its degree of strong convergence , or convergence over the path .
that means that , for any single realisation @xmath66 of the wiener process on a given time interval @xmath67 $ ] that must be chosen as the biggest time interval that we will use among all our simulations , and starting with given initial conditions @xmath68 for the system , the final point @xmath69 obtained with a good integration scheme , using a number of time steps big enough , must be close to the real solution of the differential equation , at time @xmath70 ( for that particular realisation of the wiener process @xmath66 and for this particular initial condition @xmath68 ) .
of course , when selecting any other realisation of the wiener process and any other initial condition , the final point at time @xmath70 obtained by the integration method must be close to the one of the real solution as well . of course
, we can not know the value of @xmath69 obtained with the real solution but , still , we can test whether the integration method gives something close enough to it , by studying the self - consistence .
if we have chosen a small number of integration steps ( a big value for the time step @xmath29 ) , the obtained @xmath69 will not be reliable and , when repeating the integration with more time steps ( but the same wiener process and the same initial conditions ) , we will obtain a new @xmath69 which will considerably differ from the previous one . on the other hand , we can be reasonably sure that we have already chosen a number of integration steps big enough , and therefore we have obtained a reliable value of @xmath69 if , after repeating the integration once more with a considerably bigger number of time steps ( e.g. , twice the former number of time steps , with same wiener process and same initial conditions ) , the new obtained @xmath69 stays close to the value previously obtained .
we are using here this method of self - consistency via the _ brownian tree _ : we will start with the maximum number of timesteps @xmath71 , for a chosen natural number @xmath72 , and we will then progressively reduce the number of integration steps by a half each time , i.e. , we will integrate with @xmath71 , @xmath73 , @xmath74 , etc , time steps , until the desired minimum power of 2 ( that can be , if desired , as little as @xmath75 , i.e. , a single time step ) . given the discretised wiener process @xmath76 for @xmath0 time steps , we obtain the discretisation of the same wiener process for half the number of time steps by summing up the two members of each couple , i.e. @xmath77 sometimes , one may first take a signal of gaussian white noise @xmath78 with mean zero and variance 1 ( i.e. , @xmath78 is distributed as @xmath79 ) , so the integration routine gets the increments @xmath80 of the wiener process from the 1correlated signal @xmath78 as @xmath81 , where @xmath29 is the time step that is used at that moment to integrate the differential equation ( see introduction to section [ sec : integration_methods ] ) .
then , when halving the total number of time steps ( and therefore doubling the time step to @xmath82 ) , the new signal @xmath78 still has to be 1correlated , so we have to make @xmath83,\quad j=0 , \dots , 2^{n-1}-1,\ ] ] so that @xmath84 is 1correlated as well as @xmath85 . also , that way , given that @xmath86 and @xmath87 , equation ( [ eq : brownian_correl_steps ] ) yields ( [ eq : brownian_wiener_steps ] ) . for a desired precision @xmath88 for the @xmath58-th component @xmath48 in the stochastic differential equation , we will consider that @xmath1 is a number of steps big enough if @xmath89 . in order to check that such accuracy holds for different realisations of the noise
, we will repeat the comparison of the @xmath0 and @xmath1integrations with different brownian paths @xmath90 : we call @xmath91 to the trajectory with the wiener process @xmath92 .
we will always use the same initial conditions @xmath93 . as a result
, we have for each component @xmath58 and each exponent @xmath94 a set of @xmath95 values consisting of the differences between the state at the final time @xmath70 with @xmath0 integration steps minus the state with @xmath1 integration steps : each value corresponds to each realisation of the wiener process . to be clear : we have for each component @xmath58 and each exponent @xmath96 a set @xmath97 , where @xmath98 . from each set @xmath99
, we can assess the reliability of the integration with @xmath100 time steps .
normally , if @xmath96 is not too small , the mean value of those differences @xmath101 will be around 0 , otherwise we can say that there is a preferred sign in the values of the differences @xmath102 and , therefore , a noticeable systematic error at the numerical integration
. then , the quality of the numerical integration is better the smaller the absolute values of the differences @xmath103 , i.e. , the narrower the distribution , and this being peaked around 0 . a way of measuring this is to obtain the mean and some central momenta , @xmath104 of the distribution .
given that , for a number of integration steps big enough , the absolute values of the differences @xmath105 should be as small as possible , we can say that the reliability is better the smaller ( in absolute value ) the mean and the central momenta . of course , a number of time steps that is acceptable for a given integration method ( heun , milstein , etc ) will be in general insufficient or , contrarily , higher than necessary , for another integration method .
we are using here a dynamical system as a model for paleoclimate .
our system consists of three variables : the volume of the ice @xmath106 , the @xmath107 concentration @xmath108 , and the variable @xmath109 that is related to the ocean s temperature and circulation .
the oscillatory astronomical forcing acting on the system can be approximated by two functions @xcite : @xmath110 carrying the effect of the precession of the earth s axis , and @xmath111 carrying the effect of the obliquity .
both functions are approximated by a sum of harmonic oscillations , @xmath112 , where the parameters @xmath113 , @xmath114 and @xmath115 are arranged as rows in two data files ( a file for precession and another one for obliquity ) , and we sum for each function the number of oscillations that we like according to the desired precision and computation time .
nine parameters appear in the equations that govern the deterministic system : an offset @xmath116 in the coupling of the variable @xmath109 to the ice volume , a coupling @xmath117 of the ice volume to the obliquity , a coupling @xmath118 of the ice volume to the precession , an offset @xmath119 in the coupling to the ice volume , a coupling @xmath120 of the @xmath107 to the precession , the relaxation times of each variable @xmath121 , @xmath122 , @xmath123 , and a parameter @xmath124
playing a role in the coupling to the variable @xmath106 . we will add to each equation a gaussian white noise variable : @xmath125 , @xmath126 , @xmath127 to the variables @xmath106 , @xmath108 , @xmath109 respectively .
the functions @xmath128 are independent , with zero mean and variance one ( the actual variances of the noises will be included in the functions that multiply the @xmath128 s ) : @xmath129 ( these functions @xmath128 correspond to the wiener processes @xmath22 in ( [ eq : stochastic ] ) . ) after inserting the noise , we will also have to take into account the variances of the noises , and maybe some more extra parameters in the functions that couple the variables @xmath106 , @xmath108 , @xmath109 to the noises ( i.e. , the functions that multiply the noises ) .
putting all things together , the equations of the system read : @xmath130 + g_v(v , t)\ , \eta_v\nonumber\\ \frac{\mathrm{d}c}{\mathrm{d}t } & = & - \frac{1}{\tau_c } ( c + v -\frac{1}{2}d - c_p\ , \pi)+ g_c(c , t)\ , \eta_c\nonumber\\ \frac{\mathrm{d}d}{\mathrm{d}t } & = & - \frac{1}{\tau_d } [ \varphi_3 ( 2d ) - ( v - v_t)]+ g_d(d , t)\ , \eta_d,\end{aligned}\ ] ] where the function @xmath131 is defined as : @xmath132 and the potentials are : @xmath133 as stated before , the variance of each noise is taken into the function that multiplies it . in the simplest case of additive noise ,
we just make @xmath134 , where @xmath135 is the variance of the noise acting on the component @xmath58 .
it is worth mentioning that we are performing two corrections by hand " at each integration step . on the first hand , in order to prevent @xmath136 , which would be devoid of physical meaning , we shall make @xmath137 by hand whenever we end up with @xmath138 after an integration step . on the other hand , in an attempt to reduce computing divergences , we will bound @xmath139 and make it actually be @xmath140 .
when coding , we have of course tried to avoid redundant calculations as much as possible .
this specially includes the figures that just depend on time ( and not on the state variables @xmath106 , @xmath108 , @xmath109 ) .
given that , in most cases , we will be propagating many particles , we will first obtain once and for all the values that depend only on time and are therefore the same for all the particles .
we start by generating the array @xmath141 of all the times at which the integration steps will take place , and then we calculate at all such times : the precession @xmath142 , the obliquity @xmath143 , and also the combination @xmath144 appearing in ( [ eq : function_r ] ) .
this will imply passing these three extra parameters to all the functions ( instead of passing just the parameter time ) and adds complexity to the code , but it saves invaluable time when running the programme , specially when the number of particles is high . for the tests that will follow in the next sections , we took 50 terms for the precession and 20 terms for the obliquity .
given that we are comparing three methods of numerical integration , we do not want to stay in the simple case of additive noise , but we want to test the different methods in the more general case of multiplicative noise .
we therefore use a simple kind of multiplicative noise : @xmath145 the values of the parameters used here for the comparison of the numerical integration schemes are the following ( one unit of time corresponding to 1000 years ) : @xmath146 the initial conditions used are @xmath147 , @xmath148 , @xmath149 .
the code was written in c and compiled with the intel compiler ( icc ) in order to take advantage of vectorisation .
the compilation commands were ` ice & & icc file.c -l / opt / gsl / lib -lgsl -lgslcblas ` ` -i / opt / gsl / include -o file ` ( we use the first command ` ice ` to switch to 64 bits . )
the cpu is ` intel(r ) xeon(r ) x5450 @ 3.00ghz ` running on ` suse linux 11.0 x86_64 ` .
trajectories for three particles were generated using the three integration schemes , with the same wiener process for the three schemes .
we integrated up to time equal to 2000 , using @xmath150 time steps per trajectory .
the time used to generate the data file for each integration scheme on the computing scenario described above was : * heun scheme : 0.335 seconds . * milstein scheme : 0.304 seconds . *
derivative - free milstein scheme : 0.357 seconds .
nevertheless , such small times are not reliable for comparing the different integration schemes in terms of time expenses , given that such execution times vary from one execution to another of the same executable file , depending on the other jobs in the whole computer .
we will compare the time expenses in section [ subsec : precision_brownian ] .
, scaledwidth=70.0% ] , scaledwidth=70.0% ] , scaledwidth=70.0% ] , scaledwidth=70.0% ] , scaledwidth=70.0% ] , scaledwidth=70.0% ] in order to check the agreement between the different integration schemes , we have plotted , in figures [ fig : trajectories - heun - milstein ] and [ fig : trajectories - heun - milstein - df ] , a two - by - two superposition of the same trajectories generated with the different integration schemes . for clarity
, we have only plotted the second half , from time 1000 to time 2000 .
the space between consecutive points has been intentionally made relatively large , in order to be able to see the overlapping in the same trajectory obtained from two different integration schemes .
we can see in figures [ fig : trajectories - heun - milstein ] and [ fig : trajectories - heun - milstein - df ] that all the trajectories overlap perfectly with the corresponding same trajectory obtained with the other integration scheme .
this says that the three integration methods are reliable for the number of integration steps stated above ( @xmath150 ) .
now , we will make in section [ subsec : precision_brownian ] a quantitative study of accuracy versus time expense of each integration scheme . to compare the rate of strong convergence of the three integration schemes
, we followed the method described in section [ sec : convergence_check ] .
we integrated @xmath151 particles from time 0 to time 400 with each integration scheme .
the maximum number of time steps was @xmath152 , and the minimum @xmath153 ( the biggest power of 2 , as we will see , that is small enough to cause divergences in the numerical integration for the three schemes ) . the same wiener processes were used for the three integration schemes . for the sample of differences , at the final time 400 , between the integrations with @xmath0 and @xmath1 time steps , we obtain the mean and central momenta ( [ central_momenta ] ) , for @xmath154 .
the time used to generate the data file for each integration scheme on the computing scenario described above was : * heun scheme : real 4m44.192s , user 4m29.721s , sys 0m4.396s . * milstein scheme : real 3m45.887s , user 3m41.430s , sys 0m4.008s . *
derivative - free milstein : real 4m21.324s , user 4m9.640s , sys 0m4.296s .
given that the most time consuming task of the programme is the numerical integrations , we can say from the figures above that , for the same number of time steps , the milstein scheme is the fastest method ; the next one would be the derivative - free milstein taking around 1.13 times longer than the former , and the slowest one is the heun scheme taking around 1.22 times longer than the milstein scheme ( beware , though , that these ratios may change if we use another computing environment ) . given that these ratios are close to 1 ( i.e. , the time expense of the integration schemes is roughly the same for the three ) , we shall choose the most accurate of our integration schemes . in order to compare the accuracies , we have rearranged the tables to display together the same relevant variables derived after integrating with the three different methods .
the means and central momenta explained above are displayed in tables [ table : mean][table : fourth - momenta ] .
the columns are named after the system s variable ( @xmath106 , @xmath108 or @xmath109 ) , and after the integration method : suffix `` -h '' for heun , `` -m '' for milstein , and `` -df '' for derivative - free milstein .
the names of the rows correspond to `` the small power of 2 '' , i.e. , @xmath96 means that we are considering the distribution of the differences , at the final time 400 , between the integrations with @xmath155 and @xmath100 time steps : small absolute values in the mean and central momenta indicate that the integration with @xmath100 time steps is in principle reliable . also , for the sake of visual clarity , all the means and central momenta have been multiplied by @xmath156 before displaying them in the tables .
.mean values times @xmath156 ; for the different variables , integration methods , and time steps . see text for an explanation on the notation . [
cols="^,>,>,>,>,>,>,>,>,>",options="header " , ] we can see that the numerical integration yields divergencies when done with a small number of time steps : in our case , they appear when we descend to @xmath157 integration steps for the two milstein methods , and when we descend to @xmath158 steps for the heun method . as we have stated in section [ subsec : manual - corrections ] , we have tried to remove the divergences coming from the equations , so we should think that the divergences found are due in principle only to the integration using too few , too big , time steps . if we check a given row ( corresponding a given number of integration steps ) in tables [ table : mean][table : fourth - momenta ] , we see that the mean and central momenta for the heun scheme are always smaller ( in absolute value ) compared to the figures for the other two schemes ( with a couple of exceptions in table [ table : third - momenta ] ) .
not only smaller , but the figures for the heun scheme are almost always ( except for some rows corresponding to a small number of time steps ) at least _ one order of magnitude smaller _ than the figures for the two milstein schemes .
if we were too strict , one might object that the heun scheme is ( just slightly ) more expensive in terms of computing time : however , in most cases , the figure for the heun scheme in a given row of the tables is smaller than the figure for the other schemes displayed one row above ( which corresponds to an integration with double number of time steps ) , so the slightly bigger time expense of running the heun scheme is largely made up for by the accuracy of the method . as a result
, we can state that the heun method is the best performing as we get much better accuracy , for a given computation time , compared to the milstein methods .
the better performance of the heun method was to some extend expectable , given that , in the case of additive noise , the milstein scheme reduces to the basic euler scheme ( as the derivative in ( [ milstein_diagonal ] ) vanishes ) , whereas heun s scheme is always ( also for additive noise , and even for no noise at all ) a second - order predictor - corrector method .
also , the idea of heun s method is easy to understand , and the method is easy to code ( in particular , it does not use any derivative ) .
all this makes the heun s scheme a very suitable and attractive method in our opinion .
the author would like to thank michel crucifix ( universit catholique de louvain ) and jonathan rougier ( university of bristol ) for useful advice .
the work was funded by the erc - starting grant integrated theory and observations of the pleistocene " . | three schemes , whose expressions are not too complex , are selected for the numerical integration of a system of stochastic differential equations in the stratonovich interpretation : the integration methods of heun , milstein , and derivative - free milstein .
the strong ( path - wise ) convergence is studied for each method by comparing the final points after integrating with @xmath0 and @xmath1 time steps .
we also compare the time that the computer takes to carry out the integration with each scheme . putting both things together ,
we conclude that , at least for our system , the heun method is by far the best performing one . |
we derive of the probability current given by eq . employing a generalized fox approximation @xcite to the _ coupled _ stochastic differential equations given by eq . .
the method detailed in appendix b of ref .
@xcite , which suggests the occurrence of a force derivative of the form @xmath114 , is missing two crucial points , which we will detail in the following .
firstly , in @xmath115 dimensions , eq . is vector valued : taking this into account properly , the force derivative becomes @xmath116 , where @xmath117 denotes a dyadic product of two vectors .
secondly , the forces @xmath118 are multivariate , resulting in the derivative term @xmath119 as entering in eq . .
introducing a component - wise notation ( compare , e.g. , ref .
@xcite ) for @xmath120-dimensional arrays @xmath121 we understand that the two points can be accounted for in the same way , as we rewrite in the form ( neglecting the brownian white noise @xmath122 for the moment ) @xmath123 with @xmath124 .
obviously , the correlator @xmath125 needs to be a @xmath126 tensor and the probability distribution functional @xcite @xmath127\!\propto\!\exp\!\!\left(\!-\frac{1}{2}\iint\!{\mathrm{d}}s\,{\mathrm{d}}s'\sum_{\alpha\beta } \chi_\alpha(s)k_{\alpha\beta}(s - s')\chi_\beta(s')\!\right ) \label{eq_foxapp2}\end{aligned}\ ] ] is equipped with a tensorial kernel @xmath128 , the inverse of @xmath129 .
the latter point is the basic content of the discussion in ref .
@xcite about how the one - dimensional single - argument case should be generalized . for our derivation , we calculate the formal solution @xmath130p_n[\{\chi_\alpha\}]\,\prod_{\alpha}\delta(y_\alpha - x_\alpha(t))\end{aligned}\ ] ] of eq . and its time derivative @xmath131p_n[\{\chi_\alpha\}]\left(-\sum_\beta\frac{\partial}{\partial y_\beta } \prod_{\alpha}\delta(y_\alpha - x_\alpha(t))\dot{x}_\beta(t)\right)\cr & = \sum_\beta-\frac{\partial}{\partial y_\beta } \bigg(d_\text{t}\beta f_{\beta}(\{y_\alpha\})p_n(\{y_\alpha\},t)\cr & \ \ \ \ \ \ + \int d[\{\chi_\alpha\}]p_n[\{\chi_\alpha\ } ] \bigg(\prod_{\alpha}\delta(y_\alpha - x_\alpha(t))\bigg)\chi_{\beta}(t)\bigg)\,,\cr \label{eq_foxappfp0}\end{aligned}\ ] ] using in the second step .
it is here necessary to account for each component @xmath132 and @xmath121 of the two vectors . then the exact starting point @xmath133p_n[\{\chi_\alpha\}]\left(\prod_{\alpha}\delta(y_\alpha - x_\alpha(t))\right)\chi_\beta(t)\cr & = -\sum_\gamma\int{\mathrm{d}}s'c_{\beta\gamma}(t - s')\int d[\{\chi_\alpha\}]\cr&\ \ \ \ \ \ \ \ \times\left(\frac{\partial}{\partial y_\beta}\prod_\alpha\delta(y_\alpha - x_\alpha(t))\right ) \frac{\delta
x_\beta(t)}{\delta\chi_\gamma(s')}p_n[\{\chi_\alpha\}]\cr \label{eq_foxapp3}\end{aligned}\ ] ] of the fox approximation follows from a partial integration @xcite . the most important difference to the ( approximate ) presentation in ref .
@xcite arises in how the variation in eq .
( b15 ) therein is determined . in the multivariate case
we find @xmath134 and obtain the tensorial solution @xmath135 which we approximated in the second step according to the fox scheme @xcite ( compare ( b20 ) therein ) while only taking into account the linear term in @xmath136 when expanding the exponent .
the matrix @xmath137 $ ] in the exponential has the components @xmath138 , where we recognize the desired generalization of the force derivative when switching back to the vectorial notation . using and the explicit correlator in yields at a sufficiently large time @xmath139 @xcite the representation @xmath140 of with @xmath141 given by eq .
we thus have established the full generalization of the fox approximation to three dimensions . after reintroducing into eq .
the contribution @xmath142 resulting from the brownian white noise , the probability current in the smoluchowski equation is given by eq . .
consider the following equation @xmath143 where @xmath144 is gaussian process with zero mean and the time autocorrelation @xmath145 and @xmath146 is colored noise with an autocorrelation that decays exponentially in time as @xmath147 . in numerical simulation of eq .
[ eqgen ] with @xmath148 as the integration time step , the time - updated @xmath53 can be written as @xmath149 where @xmath150 and @xmath151 are random processes defined as @xmath152 since @xmath144 is a gaussian process , @xmath150 is distributed according to a gaussian distribution with zero mean and variance @xmath153 , i.e. , @xmath154 .
the distribution corresponding to @xmath151 can be determined in the following way .
the process @xmath155 can be written in terms of a filtered white noise as @xcite @xmath156 where @xmath157 is gaussian noise with zero mean and the time correlation @xmath158 .
the formal solution to eq . [ filtered ] is given as @xmath159 following ref .
@xcite , we define @xmath160 and two gaussian processes @xmath161 with zero mean and correlations as @xmath162 with the mean and variance known the two gaussian processes can be expressed as @xmath163 we can now write the time - updated @xmath146 as @xmath164 using eq . in eq .
, we can write @xmath165 the trajectory of a particle governed by the stochastic equation of motion eq .
can be obtained by advancing time in small steps in eq . .
to calculate the rate of escape numerically , we place sinks at @xmath166 where @xmath167 is the location of the barrier of the bare potential .
a particle starting at @xmath168 at time @xmath169 is considered captured by the sink if @xmath170 .
once a particle is captured at the sink , it is reintroduced at the origin and the process is repeated at least 5000 times to obtain a reliable average of @xmath171 .
this average corresponds to the mean first passage time of the particle and the escape rate is obtained as simply the inverse of this quantity .
we note that the choice of the sink @xmath166 is arbitrary .
we have verified that our results are insensisitive to the choice of the location of sink by considering @xmath172 and @xmath173 . | the escape rate of a brownian particle over a potential barrier is accurately described by the kramers theory .
a quantitative theory explicitly taking the activity of brownian particles into account has been lacking due to the inherently out - of - equilibrium nature of these particles .
using an effective equilibrium approach [ farage _ et al .
_ , phys .
rev .
e 91 , 042310 ( 2015 ) ] we study the escape rate of active particles over a potential barrier and compare our analytical results with data from direct numerical simulation of the colored noise langevin equation .
the effective equilibrium approach generates an effective potential which , when used as input to kramers rate theory , provides results in excellent agreement with the simulation data .
the escape of a brownian particle over a potential barrier is a thermally activated process .
kramers theory accurately describes the the escape process by taking into account the force acting on a particle due to the confining potential and solvent induced brownian - motion .
kramers showed that in the limit of vanishing particle - flux across the barrier , the escape rate decreases exponentially with increasing barrier height @xcite .
in contrast to brownian particles , active particles undergo both brownian - motion and a self - propulsion which requires a continual consumption of energy from the local environment @xcite . due to self - propulsion , active particles
are expected to escape a potential barrier at a higher rate than their passive counterparts . however , a quantitative description of their escape rate , explicitly taking the activity into account has been lacking . the fact that active particles , in general , have coupled orientational and positional degrees of freedom @xcite
makes the theoretical treatment of escaping active particles over a potential barrier a difficult problem . in this paper
we show that a kramers - like rate expression can be obtained for a closely related model system of active particles in which the velocities are represented by a stochastic variable and the orientations are not considered explicitly . for small activity the steady - state properties obtained from this model exhibit intriguing similarities with an equilibrium system @xcite and several sedimentation and trapping problems are analytically tractable on the single - particle level @xcite . as a starting point for a theoretical treatment of the non - stationary case we will employ this model in the form of a a coarse - grained langevin equation for the particle position @xcite with activity of particles appearing as a colored - noise term .
such a langevin equation describes a non - markovian process and thus can not yield an exact fokker - planck equation .
the colored - noise langevin equation serves as the basis for effective equilibrium approaches which map an active system to a passive equilibrium system with modified interaction potential and an approximate fokker - planck equation @xcite .
an approximate modified potential is microscopically derived taking explicitly into account the activity on the two - particle level @xcite .
previously this approach has been applied to the structural properties of active brownian particles such as the pair correlation function and phase behavior @xcite . here
we show this approach can yield valuable insight into dynamical properties such as the rate of escape of active particles across a barrier .
the standard model system of active brownian particles in three spatial dimensions consists of spherical particles of diameter @xmath0 with coordinate @xmath1 and orientation specified by an embedded unit vector @xmath2 .
active motion of the particle is included by imposing a self - propulsion of speed @xmath3 in the direction of orientation .
the motion of the particle can be modeled by the equations @xmath4 where @xmath5 is the friction coefficient and @xmath6 the force on particle @xmath7 .
the stochastic vectors @xmath8 and @xmath9 are gaussian distributed with zero mean and have time correlations @xmath10 and @xmath11 , where @xmath12 and @xmath13 are the translational and rotational diffusion coefficients . disregarding the orientational degrees of freedoms , we will consider a theoretically tractable model of active particles evolving according to the langevin equations @xcite @xmath14 the ornstein - uhlenbeck process ( oup ) @xmath15 has the time correlation @xmath16 , where @xmath17 denotes an active diffusion coefficient and @xmath18 is the persistence time of active particle .
for a homogeneous system , the time correlation of the orientations @xmath19 of active particles evolving according to eq . can be conveniently mapped onto that of oups by choosing @xmath20 and @xmath21 .
this procedure @xcite may be viewed as a coarse - graining which effectively neglects the coupling of fluctuations in orientation and positional degrees of freedom .
escape of particles driven by colored noise in non - thermal systems has been studied in the past in different context @xcite . as we show below ,
our focus here is on an active system for which one can obtain an approximate fokker - planck equation and subsequently identify an effective interaction potential allowing us to explicitly use kramers approach to the active particles under consideration . due to the presence of colored noise in eq .
, an exact fokker - planck equation for the time evolution of probability density can not be obtained .
however , one can obtain an approximate fokker - planck equation following different schemes @xcite . using the method of fox @xcite , one obtains the approximate fokker - planck equation for eq .
( see appendix [ appendix : foxapproach ] ) with the one - body current given by @xmath22\boldsymbol{\psi}({\mathbf{r}},t)\ , , \label{jeff}\ ] ] where @xmath23 is the one - body configurational probability and @xmath24 is the effective force acting on particle with index @xmath25 with @xmath26 .
the activity enters in the description via @xmath27 , the components of an effective diffusion tensor @xmath28}$ ] which are given as @xmath29 where @xmath30 the effective force can be written as @xmath31 } ) , \label{feff}\ ] ] where the dimensionless diffusion tensor is defined as @xmath32}={{\mathbf{\stringd } } } _ { [ n]}/d_\text{t}$ ] . in this study
we restrict ourselves to study a one - dimensional problem of active particles with eq . as the equation of motion , allowing us to employ an effective potential without facing further caveats resulting from the general form of eq .
@xcite .
further considering only an external force on particle @xmath7 , generated from the one - body potential @xmath33 according to @xmath34 , the diffusion tensor @xmath28}$ ] becomes diagonal and the effective force in eq .
becomes the sum of single - particle forces .
introducing the dimensionless parameters @xmath35 and @xmath36 , we obtain from the single - particle limit of eq .
the effective external potential ( assuming that @xmath37 vanishes in the origin ) @xmath38 with the dimensionless effective diffusivity ( setting @xmath39 ) @xmath40 this result conforms with the approximations made in ref .
@xcite . following a different scheme ,
the unified colored noise approximation , the same effective potential can be obtained in the special case of non interacting active particles in a one - dimensional external potential @xcite .
we note that for interacting particles , where one is interested in obtaining an effective interaction potential , one must take into account the generalized form of the effective diffusivity in eq . which requires calculation of dyadic terms @xmath41 .
in ref . @xcite the dyadic product was approximated as a scalar product @xmath42 ( see appendix [ appendix : foxapproach ] ) which yields the same effective potential as obtained for the special case considered in this study .
however , in more than one dimension , the validity of this approximation is not obvious @xcite and a detailed discussion for interacting particles taking into account the full dyadic product will be presented elsewhere . , eq . , for different values of @xmath43 ( see legend ) . for the given parameters @xmath44 , @xmath45 , @xmath46 and @xmath47 , these results are indistinguishable from the numeric solution of eq .
we denote by @xmath48 the local minimum of the potential from where particles escape over the barrier at @xmath49 to the sink located at arbitrary @xmath50 . for numerical treatment @xmath50
will be obtained as the solution of @xmath51 .
the orange vertical arrow indicates the decreasing potential barrier with increasing activity .
the change in the curvature of the energy landscape is clearly evident .
we use the curvature at @xmath52 to approximate the curvature at the ( effective ) maximum which shifts slightly towards larger values of @xmath53 . ]
the main objective of this work is to apply the above effective interaction approach to an active particle trapped in the potential of the ( non - specific ) form @xmath54 where @xmath55 is the curvature of this bare potential at both its minimum @xmath48 and its maximum @xmath52 and the parameter @xmath56 can be used to tune the barrier height @xmath57 .
now we seek to obtain an expression for @xmath58 from eq . and employ kramers approach @xcite .
this will allow us to explicitly determine the rate of escape @xmath59 of an active particle over the given ( effective ) potential barrier in the limit of vanishing flux .
this requires determining the effective curvature @xmath60 at the minimum located at @xmath61 , as well as , the curvature @xmath62 and the maximal height @xmath63 of the effective potential at @xmath49 , as indicated in fig .
[ schematic ] .
in general , all these variables except @xmath64 are functions of @xmath55 , @xmath56 and the activity parameters @xmath65 and @xmath43 .
the most appealing aspect of the barrier - crossing problem considered is that the curvature @xmath66 of the chosen potential in the region of interest , @xmath67 , is bounded .
choosing the product @xmath68 of bare curvature and persistence time small enough , we can avoid some of the pitfalls of the effective - potential approach .
in contrast to most of the potentials describing realistic interactions between two particles @xcite , it is now justified to taylor expand the integrand in eq . in terms of @xmath65 , as the whole expression @xmath69 in the denominator of eq .
remains small within the potential well .
the unphysical divergence of @xmath70 resulting from the highly negative curvature @xmath71 at , say , @xmath50 is irrelevant for our calculations .
using eq . , the effective potential , up to linear order in @xmath65 , becomes @xmath72 where @xmath73 this analytical approximation reduces to the bare potential @xmath74 in the limit of @xmath75 .
it becomes apparent from fig .
[ schematic ] that introducing activity to the particles makes it easier for them to escape the effectively shrinking barrier .
to further quantify this observation , we assume @xmath76 independent of activity ( compare fig . [ schematic ] ) and obtain the simple expressions @xmath77 for the effective curvature and barrier height , respectively .
it can be easily seen that the effective potential barrier decreases with increasing @xmath43 . explicitly requiring @xmath78 we find the nearly trivial condition @xmath79 .
a more meaningful constraint @xmath80 for the maximal applicability of the effective potential approach to our problem in general , follows from demanding @xmath81 .
following kramers @xcite , we calculate the escape rate as @xmath82 where @xmath83 is the flux of an active particle across the potential barrier and @xmath84 is the probability of finding the particle in the potential well ( in the neighborhood of @xmath61 ) .
the one - dimensional probability distribution @xmath85 can be calculated exactly @xcite but we do not require its explicit form in the following . using @xmath85 one can calculate @xmath84 using the equilibrium approximation @xmath86 $ ] , which holds for vanishing flux across the potential barrier which is justified for sufficiently large potential barrier . under this assumption @xmath84
can be obtained as an integral @xmath87 over @xmath85 in a region around @xmath61 corresponding to the width @xmath88 of the ( effective ) potential well .
the integral expression on the second line is obtained by invoking a saddle point approximation for @xmath58 at @xmath61 and extending the integration domain to infinity .
the flux can be calculated from the one - dimensional version of eq . rewritten as @xmath89 assuming a constant flux of particles one can integrate eq .
from @xmath61 to @xmath50 to obtain an expression for @xmath83 as @xmath90dx}}{\mathlarger{\int_{x_\text{a}}^{x_c}\frac{e^{\beta v^{\rm eff}(x)}}{\mathcal{d}(x)}dx } } \\ & \approx \psi(x_\text{a})\frac{\mathlarger{d_\text{t}(1+\mathcal{d}_\text{a})e^{- \left(\beta e_b-\frac { \mathcal{d}_\text{a}\omega_\text{0}\tau}{1+\mathcal{d}_\text{a}}\right)}}}{\mathlarger{\sqrt{\frac{2\pi}{\beta |\omega_b| } } } } , \end{aligned } \label{jkramers}\ ] ] where the saddle point approximation ( at @xmath49 ) has been used to evaluate the integral in the denominator and the boundary condition at the sink is set to @xmath91 . as noted above , the effective potential can exhibit unphysical behavior .
however , the approximate result for @xmath83 remains reasonable as long as the condition @xmath92 holds . for @xmath92 , the unphysical behavior of the effective potential manifests itself for @xmath93 and thus does not obscure our calculations as @xmath94 and the location of @xmath50 do not explicitly enter in the second step in eq . .
when @xmath68 becomes too large such that @xmath92 is no longer valid , the saddle point approximation is unjustified and the above analytics do not hold .
[ cols= " > , > " , ] using eqs . and
, the rate of escape of active particles over a potential barrier can be written as @xmath95 this result is valid only for large barrier heights @xmath96 . employing our approximate effective potential by substituting eqs . , and into eq .
, we obtain the compact analytic representation @xmath97 where @xmath57 . in the second line
we have omitted the term in the square root . in eq . , we have identified the escape rate of passive particles as @xmath98 . in this form eq .
clearly demonstrates that activity significantly facilitates the escape of a particle .
equation is obtained based on the approximate form of the effective potential where the small @xmath65 approximation has been used . in principle , the effective potential can be obtained directly by numerical integration of eq . or in a lengthy analytical form generalizing eq .
by taking into account higher order terms in @xmath65 .
however , the escape rate obtained in this way does not differ significantly from the analytical approximation in eq . .
hence , eq . is expected to accurately capture the escape rate under the condition that the saddle - point approximations made in eqs . and
do not introduce any significant errors . the accuracy of eq .
can be assessed by comparing the analytical predictions with the numerical rate of escape obtained by solving the eq . with the boundary conditions @xmath99 and @xmath91 .
however , the comparison between eq . and the rate obtained from eq . serves only to benchmark the analytical ( kramers ) approximation against the numerical prediction from the effective equilibrium approach for low activity .
the most relevant test is to determine the rate directly from the governing langevin equations .
this is done by numerically solving the one - dimensional version of eq .
( appendix [ appendix : noisesim ] ) for a particle trapped in the potential given by eq . .
we calculate the mean - first - passage - time ( mfpt ) of a particle starting at @xmath100 escaping to a sink located sufficiently far from the potential barrier .
the equivalence of mfpt to kramers rate @xcite implies that kramers rate can be numerically obtained as the inverse of mfpt .
we first discuss the escape rate @xmath59 as a function of the reduced diffusion constant @xmath43 in fig .
[ fluxcomparison ] . as can be seen in fig .
[ fluxcomparison ] , the escape rate increases over several orders of magnitude with increasing @xmath43 .
in particular , the analytical prediction of eq . is in excellent agreement with the numerical prediction of eq . over the full range of @xmath43 considered in this study . in general , the escape rate obtained using the effective equilibrium approach agrees very well with the simulations based on the colored noise langevin equation , eq . .
it is apparent from fig .
[ fluxcomparison](c ) that the deviations resulting from the effective equilibrium mapping are only manifest at large @xmath43 . with increasing @xmath43 ,
the flux of particles across the potential barrier can become significantly large such that the equilibrium approximation to calculate @xmath84 in eq .
does not hold .
this is equivalent to stating that the barrier height must remain sufficiently large for the kramers approach to yield reliable estimate of escape rate across the barrier .
we also tested the analytical prediction for different values of @xmath65 as shown in fig .
[ fluxcomparison ] .
the excellent agreement suggests that in one dimension , the effective potential can yield accurate quantitative description of the escape process . to further assess the accuracy of our approach and
demonstrate its utility , we study the escape rate of active particles at constant activity but for some other combinations of parameters in the trapping potential , eq . .
the parameters must be chosen so as not to violate the constraints on @xmath55 and @xmath65 as determined from eqs . and .
the analytical prediction is expected to become inaccurate with increasing @xmath101 .
first , we fix the location @xmath102 of the potential barrier by setting @xmath103 . as a result
the height @xmath104 becomes a linearly increasing function of @xmath55 .
the evolution of the escape rate is primarily determined by the exponential function , the argument of which is linear in @xmath55 in both the active and the passive case .
therefore , the escape rate of active particles shown in fig . [ escaperae](a ) closely resembles that of passive ones and can thus be explained as if there was a higher effective temperature ( strictly speaking this analogy , which here would require @xmath105 up to a constant , only holds exactly for linear potentials @xcite ) .
we calculate the escape rate from eq . and plot it together with the numerically obtained rates in fig .
[ escaperae ] .
the numerically obtained escape rates are in agreement with each other over the entire range of @xmath55 considered as well as with the analytical prediction of eq . .
the high accuracy of the analytical approach in describing the escape rate suggests that the assumptions used in the kramers - like analytical approach in the derivation of eq . remain valid for a significant range of the barrier height .
the barrier height of the bare potential depends on @xmath55 and @xmath56 . by choosing @xmath106 ,
the barrier height @xmath107 remains independent of @xmath108 , but the width @xmath109 of the potential well decreases with increasing curvature .
another , more intuitive interpretation would be that of an increasing average slope @xmath110 of the potential barrier .
this particular choice of parameters allows us to discuss the rate of escape from a potential well of changing curvature but with a fixed barrier height .
it follows from eq .
that , with the barrier height of the bare potential fixed , the passive rate @xmath111 becomes a perfectly linear function of @xmath55 , whereas @xmath59 maintains its exponential form .
however , this deviation from linearity is almost negligible for the range of @xmath55 we are interested in . as shown in fig .
[ escaperae](b ) , the behavior of the escape rate is well represented by the linear function @xmath112 obtained from expanding eq . in terms of @xmath55 at constant @xmath113 . even for the small values of @xmath59 ,
the numerics and analytics are in good agreement .
a related barrier - crossing problem has recently been studied experimentally @xcite and theoretically @xcite .
the active particle moves in an energy landscape which is flat except having an asymmetric potential barrier .
the particle can cross the potential barrier from either side .
it was found that the transition rate was smaller for particle crossing the barrier from the side facing steeper slope of the barrier .
this can not be explained using the kramers approach in which , as shown above , the escape rate increases with increasing curvature for a fixed barrier height .
note that in the kramers approach one considers a potential well rather than a single barrier that does not surround the particle .
an ideal experimental study to test our theoretical results would correspond to studying the transition rate in a double - well potential of equal depth but different widths .
it will also be very interesting to extend the kramers approach to particles escaping over an asymmetric barrier . in conclusion , we derived an effective interaction potential for active particles in a one - dimensional potential well of finite depth . using this effective potential we calculated the escape rate of active particles over the potential barrier . for the problem considered ,
this approximate procedure turns out to be ( i ) well justified , as no tensorial effective quantities occur and no pairwise interaction forces are involved which both would require further approximations , ( ii ) highly accurate although the potential considered has a negative curvature , ( iii ) particularly simple , as all conditions are respected to justify expansion methods and ( iv ) the ideal link to kramers approach for passive particles . as our main result we obtained and discussed a closed analytic formula for the escape rate .
we find that upon increasing the activity or the curvature at the maximum of our model potential , the effective equilibrium approach only slightly overestimates the escape rate of active particles compared to computer simulations .
similar calculations can be made involving any other trapping potential .
it would be interesting to set up an experiment or adapt existing ones @xcite to test our theoretical predictions . |
galaxy properties , such as morphological appearance and star formation rate ( sfr ) , strongly depend on the physical conditions of the environment
. clusters of galaxies are dominated by red early - type galaxies with little star - forming activity , while the dominant population in the low - density field are blue late - type galaxies with significant star formation ( @xcite ; @xcite ; @xcite ) .
the trend that the fraction of early - type galaxies increases with the local density is found even in poor groups ( @xcite ; @xcite ; @xcite ) .
cosmological simulations predict that galaxies and groups are assembled into massive clusters moving along filamentary structures ( e.g. millennium simulation ; @xcite ) .
these surrounding regions are likely to play a critical role in the evolution of galaxies before becoming cluster galaxies .
it is reported that ultraviolet - optical colours change sharply from blue to red when going from the field to dense environments ( @xcite ; @xcite ) , implying that at least for some of the red cluster galaxies , star - forming activity was quenched through environmental effects before entering the cluster region .
moreover , galaxy groups , due to their low velocity dispersion , represent natural sites for the previous evolution of present - day cluster galaxies through tidal interactions @xcite , before the galaxies settled down in clusters .
it is thus likely that at least part of the morphological transformation of cluster galaxies took place in earlier epochs in different conditions than the ones observed in present day clusters @xcite .
this is called pre - processing its relevance has still been debated in recent simulation results ( @xcite ; @xcite ; @xcite ) . there is a range of known physical mechanisms that could be responsible for altering galaxy morphology and/or sfr in clusters .
ram pressure stripping @xcite , one of the most powerful effects , seems to have been observed directly in the virgo cluster @xcite , and simulations suggest that this process operates on short time - scales ( @xmath81 gyr , @xcite ) .
other processes include strangulation ( @xcite ; @xcite ) , which cuts off the cold gas supply for ongoing star formation in cluster galaxies .
galaxy mergers , tidal interactions , and harassment within the cluster potential @xcite are other effective processes in transforming observed properties of the galaxy populations . when investigating the effect of the environment on the evolution of galaxies , early - type galaxies and transitional dwarf galaxies in poor groups are the most attractive targets
. in general , dwarf galaxies with a shallow potential well are more sensitive to environmental effects than giant galaxies .
recent studies reveal blue - cored early - type dwarf galaxies in various environments : clusters ( @xcite ; @xcite , @xcite ) , groups ( @xcite ; @xcite ) , and the field @xcite .
theories recently evoked include that blue - cored early - type dwarf galaxies are the results of morphological evolution of late - type star forming galaxies by cluster environmental effects such as ram pressure stripping , harassment , tidal interaction or tidal stirring , and mergers ( @xcite ; @xcite ; @xcite ) .
it is also possible that blue - cored early - type dwarf galaxies in fields are formed via merging of gas rich small irregular galaxies @xcite . while all these hypotheses are still debatable , comparing the presence and properties of these transitional dwarf galaxies in different environments is thus a promising approach to shed light on the specific transformation mechanisms . in this context
, the ursa major cluster is an ideal target to investigate the evolution of dwarf galaxies .
the cluster is located at a distance of 17.4 mpc @xcite , slightly more distant than the virgo cluster ( 15.8 mpc , @xcite ; 15.9 mpc , @xcite ) but closer than the fornax cluster ( 20.3 mpc , @xcite ) .
@xcite mentioned that the ursa major cluster is situated in the junction of large - scale filamentary structures , which makes it intrinsically difficult to measure its dynamical properties and to define cluster membership for galaxies in this region . to overcome this problem
, @xcite defined the cluster members to be galaxies within a projected circle with [email protected] radius from the nominal cluster centre ( ra(b1950 ) = 11@xmath10 [email protected] and dec.(b1950 ) = @xmath1249@xmath9 22@xmath13 ) by means of a hierarchical grouping algorithm .
the ursa major cluster covers the velocity range between 700 and 1210 km s@xmath14 and has a small velocity dispersion of 148 km s@xmath14 @xcite .
@xcite identified 79 cluster members which consisted mainly of gas - rich late - types @xcite and two dozen dwarf galaxies @xcite .
morphological and photometric properties in the optical , near - infrared , and radio of individual galaxies are described in detail in @xcite and @xcite . in the following ,
we present a photometric study of the optical and ultraviolet ( uv ) properties of galaxies in the ursa major cluster .
we compile previously uncatalogued cluster galaxies using the spectroscopic data of the sloan digital sky survey ( sdss ) and the uv data of the galaxy evolution explorer ( @xmath15 ) . by comparing their properties to those in the more massive and more dense virgo cluster , we examine different environmental effects on galaxy properties .
this paper is arranged as follows : section 2 describes the criteria for the selection of the cluster galaxy samples and how the photometry is performed . in section 3 ,
the galaxy catalogue is listed and the properties of the ursa major cluster are presented in comparison to previous studies . in section 4 , we show our results on the galaxy properties and compare them to the virgo cluster . a discussion on the environmental dependence is given in section 5 , along with our conclusions .
.5 radius ( cf .
the hatched histogram denotes our 166 member galaxies .
the grey histogram is for the galaxies located in the same circle , but not selected as member galaxies .
the filled histogram is for the 79 member galaxies of @xcite , i.e. the 12 - 1 group from @xcite . ]
.5 radius centred at r.a.(b1950 ) = 11@xmath10 [email protected] and dec.(b1950 ) = @xmath12 49@xmath9 22@xmath13 ( sgl = [email protected] and sgb = [email protected] , red cross ) defines the region of the ursa major cluster @xcite .
all galaxies extracted from the sdss and ned with v@xmath16 @xmath1 2000 km s@xmath14 are presented by grey filled circles .
blue filled circles denote our 166 cluster member galaxies whereas black filled circles are the original 79 member galaxies from the @xcite catalogue .
galaxies in neighbouring groups defined by @xcite are also indicated : open squares for the 14 - 4 group , open squares with cross for the 12 - 2 group , and open circles with cross for the 12 - 3 group .
the galaxies 1136 + 46 and ugc 7129 from @xcite are excluded from our catalogue ; 1136 + 46 is a background galaxy with a large sdss redshift and ugc 7129 is just located outside of the [email protected] radius circle . ] our sample in the ursa major cluster is predominantly based on the sdss data release 7 ( dr7 ; @xcite ) .
the sdss imaging data provide reduced and calibrated ccd images taken in the @xmath17 , @xmath18 , @xmath19 , @xmath20 , and @xmath21 bands , which allow detailed morphological classification and accurate galaxy photometry .
we took advantage of the sdss spectroscopic data for determining cluster membership of galaxies .
following the definition of the region of the ursa major cluster by @xcite , we extract sdss spectroscopic data for all galaxies within a [email protected] distance from the cluster centre at ra ( j2000 ) = 11@xmath10 59@xmath11 [email protected] and dec.(j2000 ) = @xmath1249@xmath9 05@xmath13 [email protected] . as a first step , we selected 162 galaxies with radial velocities less than 2000 km s@xmath14 in this region . while the spectra of most sdss galaxies are obtained from the fibre aperture near the centre of galaxies , some large galaxies have multiple spectra observed at various fibre locations . in this case , we choose the radial velocity derived from the spectrum obtained at the nearest location to the galaxy centre .
the sdss spectroscopic observations cover galaxies as faint as r = 17.77 mag @xcite , but luminous galaxies were rejected in the sdss spectroscopic observations due to saturation problems ( blanton et al .
2005a , b ) . in order to fill in the incompleteness
, we added 69 galaxies from the nasa / ipac extragalactic database ( ned ) , which are located in the same region and have the same radial velocity range .
the additional galaxies from ned cover a wide range of magnitude ( 9.8 @xmath1 r @xmath1 17.7 ) , but are dominated by galaxies brighter than r = 14 mag . from the radial velocity distribution of the galaxies , we determined ursa major cluster membership using the biweighted scale estimator @xcite , which provides a statistically robust mean velocity and dispersion .
galaxies with radial velocities beyond 2@xmath24 from the mean value were rejected , and the process was repeated iteratively until convergence .
this identifies 166 member galaxies in the ursa major cluster .
their velocity distribution is given in fig .
[ f1 ] ( see hatched histogram ) .
following @xcite , we calculated local group velocity v@xmath16 = v@xmath25 + 300sin__l__cos__b _ _ , where _ l _ and _ b _ is galactic longitude and galactic latitude , respectively .
the resulting velocity range and velocity dispersion of our sample galaxies are 677 @xmath1 v@xmath16 @xmath1 1302 km s@xmath14 and @xmath26 = 158 km s@xmath14 .
these values are in good agreement with the results for the @xcite galaxies ( i.e. , 700 @xmath1 v@xmath16 @xmath1 1210 km s@xmath14 and @xmath26 = 148 km s@xmath14 , see filled histogram in fig .
[ f1 ] ) .
we calculated the number ratio of the member galaxies from sdss spectroscopic data to the total sample ( ned and sdss ) as a function of magnitude .
almost all faint galaxies are included in the sdss spectroscopic sample ( @xmath8 96 per cent at r @xmath27 14 mag ) , while only 63 per cent of galaxies brighter than r = 12 mag are included .
as we already mentioned , this is because bright galaxies with r @xmath1 14.5 were omitted in the sdss spectroscopic survey due to saturation problems ( blanton et al .
2005a , b ) .
additionally , fibre conflicts affect the completeness @xcite . in order to check for any potential members of the ursa major cluster that we might have missed
, we used the measurements from the sdss photometric pipeline to extract galaxies within the same magnitude , size , and colour range : @xmath4band petrosian magnitude between 9 and 18 , 3.4 @xmath1 petro r50 @xmath1 7 arcsec , and @xmath28 @xmath1 @xmath29 @xmath1 1.0 ( using petrosian @xmath3 and @xmath4band magnitudes ) , which corresponds to the member galaxies of the ursa major cluster .
this yields a sample of 270 objects classified as galaxies without any spectroscopic information by the sdss pipeline .
we visually inspected every object and excluded it if it was a star , a ` piece ' of a galaxy instead of a whole galaxy , or a small object that is obviously located at larger distance .
most of the 270 objects turned out to be stars or background galaxies .
we found 10 galaxies that bear some similarity to low surface brightness ( lsb ) members of the ursa major cluster and could not be excluded as cluster member candidates .
we thus conclude that our above sample of 166 member galaxies is complete to more than 94 per cent within 9 @xmath1 r @xmath1 18 magnitude .
the spatial distribution of galaxies selected as our final sample of the ursa major cluster is shown in fig .
[ f2 ] ( see blue filled circles ) . in comparison , the 79 members defined by tully et al .
( 1996 , i.e. , 12 - 1 group from @xcite ) are also overplotted ( black filled circles ) .
all but two galaxies ( 1136 + 46 and ugc 7129 ) of the @xcite sample are included in our catalogue .
the galaxy 1136 + 46 has a large redshift of z @xmath8 0.05 in the sdss spectroscopic measurement .
the galaxy ugc 7129 is located at [email protected] from the cluster centre , just outside of the [email protected] radius circle that defines our catalogue and is therefore not included .
our final catalogue thus remains at a total number of 166 galaxies .
we note that all galaxies in the foreground 14 - 4 group ( open squares ) in @xcite are excluded from our catalogue .
however , there is possibly a little contamination of background groups in the spatial and velocity distributions , as already mentioned by @xcite .
five galaxies in our sample overlap in velocity with those in background groups in @xcite ; two galaxies in the 12 - 2 group ( open squares with cross ) and three galaxies in the 12 - 3 group ( open circles with cross ) .
we note that none of our result depends on the inclusion of these galaxies .
the projected spatial distribution of the new ursa major cluster galaxies shows essentially a diffuse configuration and lack of concentration towards any core .
however , the new member galaxies in our catalogue seem to fill in the vacant regions between the @xcite cluster members .
consequently , the overall distribution of galaxies appears to be a more elongated , filamentary configuration in the direction of supergalactic longitude .
nuv ( top panel ) and fuv ( bottom panel ) detected galaxies in supergalactic coordinates .
crosses denote galaxies ( 18 and 25 in nuv and fuv , respectively ) which are not covered by @xmath15 observations .
the small open circle in the lower panel indicates u108 , which is observed but not detected in fuv .
open circles with different colours imply @xmath15 mosaic fields with different imaging modes . the solid large circle with [email protected] radius indicates the ursa major cluster region based on @xcite . ] in order to investigate the ultraviolet properties of our sample galaxies in the ursa major cluster , we used far - ultraviolet ( fuv ; 1350 @xmath30 1750 ) and near - ultraviolet ( nuv ; 1750 @xmath30 2750 ) images from the @xmath15 gr6 .
a @xmath15 pointing has a circular field of view of 1.@xmath925 diameter @xcite . in order to secure the best astrometric and photometric results we avoided the image distortions at the edge of the field and considered galaxies located only within the inner 0.55 degree radius field .
the @xmath15 plate scale is 1.5 arcsec pixel @xmath14 which adequately samples the 4.5 @xmath30 6 arcsec point spread function .
@xmath15 observed a total of 316 and 263 fields mostly overlapping in nuv and fuv , respectively , within a [email protected] radius circle from the centre of the ursa major cluster ( see fig .
[ f3 ] for mosaic fields for nuv and fuv ) .
unfortunately , due to some coverage gaps 18 and 25 galaxies were missed in nuv and fuv , respectively ( see crosses in fig .
the depth of each field also varies with the survey mode : deep imaging survey ( dis ) , medium imaging survey ( mis ) , nearby galaxy survey ( ngs ) , all - sky imaging survey ( ais ) , and guest investigator ( gi ) programme .
when several uv images are available for a galaxy , we used the deepest image for its uv source .
overall , 148 ( @xmath889 per cent ) and 140 ( @xmath884 per cent ) galaxies were detected in nuv and fuv , respectively .
this means that all observed galaxies were detected nuv sources and only a single one of those galaxies that were covered remained undetected in fuv ( u108 in fig .
all fuv detected galaxies are also detected in the nuv . in table 1
, we summarize the @xmath15 observations and the number of detected galaxies in different survey modes
. @lccc mode & observed field # & exposure time ( s ) & detected galaxy # + + dis & 7 & 8969 - 13756 & 12 + mis & 50 & 126 - 2982 & - + ngs & 6 & 109 - 1242 & 9 + ais & 216 & 71 - 856 & 102 + gi & 37 & 326 - 9482 & 25 + + dis & 7 & 8969 - 13756 & 12 + mis & 7 & 1115 - 2026 & - + ngs & 6 & 107 - 1242 & 9 + ais & 216 & 71 - 409 & 103 + gi & 27 & 1444 - 5891 & 16 + the sdss photometric pipeline is optimized for small , faint objects . galaxies with large angular extent , irregular morphology , and bright subclumps ( e.g. , hii regions and spiral arms ) are shredded into multiple , separate objects .
the standard sdss pipeline thus fails to derive a reliable total flux of these galaxies @xcite . in the sdss pipeline , a 256 @xmath31 256 pixel mask
was used for determining the sky background . however ,
large and/or bright galaxies in terms of apparent size and magnitude occupy a considerable fraction of the 256 @xmath31 256 pixel mask , resulting in an overestimation of sky value as high as 1 magnitude ( see @xcite ) . in order to avoid these two issues
, we performed our own sdss photometry for all selected galaxies in the ursa major cluster .
the photometry of member galaxies was performed using source extractor ( se ; @xcite ) . for each galaxy , we created sdss postage images in five passbands ( @xmath17 , @xmath18 , @xmath19 , @xmath20 , and @xmath21 ) .
source detection and deblending are carried out on the r - band image . the same source position and isophote of r - band image
are also used for subsequent photometry in other passbands .
we use the parameter back size = 256 pixels for most galaxies . in the case of large , extended galaxies , a larger mesh size ( i.e. , back size = 512 pixels )
is adopted to prevent an overestimation of the sky background .
for source detection , we usually chose parameters of detect threshold = 1 and detect minarea = 10 . in some cases of faint galaxies and galaxies affected by neighbouring bright sources , different parameters ( detect threshold = 0.5 @xmath8 1 , detect minarea = 5 @xmath8 50 ) were chosen preventing a large number of spurious detections .
the deblending parameters ( deblend mincont and deblend nthresh ) for each galaxy were selected iteratively by visual inspection of detected sources , enabling to separate neighbour sources which overlap with the target galaxy .
we adopted mag auto as the total magnitude of a galaxy , which provides the flux within an adjustable elliptical aperture that is scaled according to the isophotal light profile .
once a flux was calculated , we converted all fluxes to asinh magnitudes @xcite . as for the @xmath15 uv sources , we also performed photometry for all detected objects in a similar manner .
flux calibrations were applied to bring the final photometry to the ab magnitude system @xcite .
magnitude errors of sdss photometry were estimated using the equations given by @xcite in which several different sources of uncertainty are considered ( see sec .
4 of @xcite for details ) .
magnitude errors of @xmath15 photometry were estimated by adding the three different uncertainties in quadrature : photometric calibration uncertainty , signal - to - noise ( s / n ) uncertainty , and background uncertainty .
the calibration uncertainty is estimated to be 0.03 and 0.05 mag in the nuv and fuv , respectively @xcite . the s / n uncertainty is calculated using the equations given by the @xmath15 data guide on the mast homepage .
se assumes that the distribution of background counts is a gaussian distribution .
however , due to the low count rates in the uv images , a poisson distribution is a better model for the background count rate . to account for this effect , we used the equation provided by the @xmath15 data guide , @xmath32 where _ f _ is the flux of the source in counts s@xmath14 , _ s _ is the sky level in counts s@xmath14 pixel@xmath14 , _
@xmath33 _ is the area in which the flux is measured , and _
t _ is the effective exposure time in seconds .
the sky levels were measured from background images provided on the galex website .
the typical values of uncertainty are 0.03 and 0.07 mag in the nuv and fuv , respectively .
colour errors were obtained by adding the relative flux errors in quadrature in each band . the galactic extinction correction for each galaxy
is applied using the reddening maps by @xcite .
we used the reddening law of @xcite to derive the following : r@xmath34 = 8.16 , r@xmath35 = 8.90 , r@xmath36 = 5.05 , r@xmath37 = 3.89 , r@xmath0 = 2.82 , r@xmath38 = 2.18 , and r@xmath39 = 1.57 .
all galaxies of our catalogue were morphologically classified using the sdss images .
five members of our team ( mp , tl , sk , es , and hj ) independently carefully inspected both monochromatic @xmath3 , @xmath4 , @xmath5band images and @xmath3 , @xmath4 , @xmath5combined colour images .
the morphology of each galaxy was finalized if the classification of three or more classifiers agreed . if the classification for a galaxy differed for more than three classifiers , all classifiers inspected the galaxy again and agreed on a final classification .
galaxies were divided into four major classes : early - type galaxies [ elliptical ( e ) and lenticular ( s0 ) ] , early - type dwarf galaxies [ dwarf elliptical ( de ) and dwarf lenticular ( ds0 ) ] , spiral galaxies ( sa , sb , sc , and sd ) , and irregular galaxies .
our morphological classification scheme is basically in accordance with the hubble classification ( see @xcite for a recent reviews ) .
e galaxies are characterized by high surface brightness ( hsb ) spheroidal shape and a smooth featureless envelope with no sign of a disc .
s0 galaxies have a smooth bulge and a disc component with a distinct change in surface brightness at the transition radius , discriminating them from e galaxies . in usual cases ,
s0 galaxies also exhibit lens - like features .
the classification of spiral galaxies is based on the prominence of bulge and the appearance of the spiral arms such as degree of their tightness and resolution .
edge - on galaxies are spiral galaxies that are seen edge - on , preventing us from a more detailed morphological classification .
the ` s ' type includes spiral galaxies for which the spiral subclass could not be determined due to relatively small angular size and lack of unique features within the spiral galaxy classification scheme .
sm galaxies are defined as magellanic type spirals . as for the s0 and spiral galaxies , we also considered barred galaxies showing a linear bar structure in the centre .
in addition , we classify specific dwarf galaxies as described by @xcite .
de galaxies are lsb ones characterized by elliptical isophotes and smooth surface brightness profiles .
ds0 galaxies show an overall smooth structure but exhibit additional features deviated from the normal de galaxies ; a distinct bulge - disc transition , asymmetric features ( e.g. , bar , lense , and boxyness ) , or an irregularity in the central region ( see @xcite for details ) . in most cases of ds0 galaxies with central irregularity , their irregular structures turned out to be smaller than half - light radius returned from the optical photometry of galaxies . among early - type dwarf galaxies , we classify nucleated dwarf galaxies ( de , n and ds0 , n ) where an unresolved stellar nucleus is seen at the centre of the galaxy .
irregular galaxies are divided into hsb irregulars ( hsb irr ) and lsb irregulars ( lsb irr ) according to overall surface brightness of galaxies by visual inspection .
some irregular galaxies present elliptical shapes in outer part , but they show wide - spread irregular structures at their centres which are comparable to or larger than their half - light radius .
we conducted a thorough classification based on their morphology and also deliberately avoided intermediate classes like ` de / irr ' . in fig .
4 , we present sdss colour and @xmath4band images of galaxies from the different morphological classes .
the new catalogue of the ursa major cluster galaxies contains 166 objects .
table 2 presents the full catalogue including fundamental information such as membership , morphology , and optical / uv photometry .
the following is a brief description of each column of table 2 : column 1 : i d number .
i d is marked with the ra column 2 : ngc , ugc , or other name of galaxy from @xcite , @xcite , and ned columns 3 - 4 : right ascension ( j2000 ) and declination ( j2000 ) from se in degrees columns 5 - 6 : supergalactic longitude and latitude in degrees column 7 : heliocentric radial velocity ( v@xmath40 ) from the sdss in km s@xmath14 ( flag
n : adopted from ned ) column 8 : radial velocity that adjusted for the frame of the local group by adding 300sin__l__cos__b _ _ in km s@xmath14 ( v@xmath16 ) column 9 : membership of galaxy from different catalogue ( 1 = this study , 2 = @xcite , and 3 = @xcite ) column 10 : name of group defined by @xcite which the galaxy belongs to column 11 : morphology of galaxy classified from this study column 12 : rc3 t - type from @xcite column 13 : rc3 t - type from @xcite columns 14 @xmath30 18 : sdss @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 magnitudes and magnitude errors columns 19 @xmath30 20 : @xmath15 nuv and fuv magnitudes and magnitude errors column 21 : note for blue - cored early - type dwarf galaxies .the ursa major cluster catalogue [ cols="^,^,^,^,^,^,<,^,^,^,^,^,^,>,>,>,>,>,>,>,^ " , ] table 2 is presented in its entirety in the electronic edition of the monthly notices of the royal astronomical society .
a portion is shown here for guidance regarding its form and content . in fig .
[ f5 ] ( left - hand panel ) , we present the spatial distribution of the 166 cluster galaxies ( blue filled circles ) included in our catalogue , in comparison with 79 and 157 member galaxies from tully et al .
( 1996 , black dots ) and karachentsev et al .
( 2013 , open red circles ) , respectively .
in the case of the catalogue of @xcite , based on the galaxy group catalogue of @xcite , they define the ursa major cluster consisting of seven galaxy subgroups : ngc 3769 , ngc 3877 , ngc 3992 , ngc 4111 , ngc 4157 , ngc 4217 , and ngc 4346 groups . from our 166 galaxies , 141
are also included in the seven groups of @xcite . in the top right panel of fig .
[ f5 ] , we compare the velocity distribution of the galaxies between the three catalogues . at a glance ,
the strongly overlapping galaxy samples of all three catalogues cover a similar range in velocities .
we obtain a velocity range and dispersion for the @xcite sample as 653 @xmath1 v@xmath16 @xmath1 1361 km s@xmath14 and @xmath26 = 158 km s@xmath14 , which is in good agreement with those from ours and tully et al .
( 1996 , see also sec .
the bottom right panel of fig .
[ f5 ] shows a comparison of the r - band luminosity function among three catalogues .
we adopt a distance modulus of the ursa major cluster of 31.20 from @xcite .
the luminosity functions of our sample and @xcite are almost identical .
note that , since the galaxies selected from the sdss spectroscopic data span a wide range in luminosity ( @xmath41 @xmath1 m@xmath42 @xmath1 @xmath2 ) , the catalogues of ours and @xcite contain a large fraction of faint dwarf galaxies ( i.e. , @xmath7 @xmath1 m@xmath42 @xmath1 @xmath2 ) compared to the sample of @xcite .
these faint galaxies mostly consist of early - type dwarf galaxies ( de and ds0 ) and irr galaxies in our catalogue ( see also fig . [ f9 ] ) . in fig .
[ f6 ] , we present two - dimensional histograms comparing the morphologies of the galaxies in our catalogue with those of tully et al .
( 1996 , left - hand panel ) and karachentsev et al .
( 2013 , right - hand panel ) .
adopting table 1 in @xcite , we translate the rc3 t - type number @xcite employed in the catalogues of @xcite and @xcite to the hubble morphological classification scheme .
the overall trend in fig .
[ f6 ] is that , except for dwarf galaxies , our morphological classification is largely consistent with those of @xcite and @xcite .
a significant disagreement occurs in the classes of sm and irr galaxies for @xcite .
a large fraction of these galaxies are classified as early - type dwarf galaxies ( i.e. , de and ds0 ) in our catalogue ( see lower - right part of right - hand panel of fig . [ f6 ] ) .
we point out that , while some of these galaxies show somewhat irregular features at their centres , all early - type dwarf galaxies show clear elliptical shapes with overall smooth and regular appearances . we show in fig . [ f7 ] the images of all these dwarf galaxies we classified as early - type .
tully & trentham ( 2008 ) noted an unusual population of central starburst dwarfs in the luminosity range of @xmath6 @xmath1 m@xmath42 @xmath1 @xmath43 in the ngc5353/5354 group .
these authors concluded that this population would be classified as de based on their morphology .
these galaxies look like some of our early - type dwarf galaxies with central irregularities , but are partly even bluer and more irregular .
they may thus be considered transition - type dwarf galaxies . on the other hand
, the classification comparison also shows a small scatter at fixed morphological type of our catalogue .
a scatter is seen for edge - on galaxies in our catalogue , since we refrained from subdividing these further into subclasses . a difference with respect to @xcite and
@xcite is also seen for our irr galaxies in the sense that several of them have been classified as sm or sd in @xcite and @xcite as shown in fig .
the type - specific luminosity function is helpful for studying galaxy evolution .
detailed studies of the luminosity function for different morphological types were performed for virgo ( @xcite ; @xcite ) .
the dedicated study by @xcite found that the type - specific luminosity function of the centaurus and virgo cluster are very similar . in fig .
[ f9 ] , we show the @xmath4band type - specific luminosity function of the galaxies in the ursa major cluster . the ursa major cluster is dominated by late - type galaxies of spiral , edge - on , and irregular type ( about 80 per cent , i.e. , 133/166 ) , but comprise a considerable fraction of early - type galaxies ( about 20 per cent , i.e. , 33/166 ) .
it is shown that the luminosity function depends on the galaxy type in the sense that irr and late - type spiral galaxies are systematically fainter than s0 and early - type spiral galaxies ( see also sandage et al . 1985 ) . in the case of early - type dwarf galaxies ( de and ds0 ) , they are located in the faintest range of the distribution .
irregular galaxies show a bell - shaped distribution and are skewed towards fainter magnitude . however , due to small number statistics , it is difficult to draw conclusions on the precise shape of the type - specific luminosity function for all galaxy types in the ursa major cluster .
, @xmath4 , @xmath5combined images of irregular galaxies in our catalogue which are classified as different types in @xcite .
the i d of each galaxy is given in the upper left part of the image , along with our morphology .
the morphology from @xcite is also given in the lower right part of the image . ]
@xmath44 ) of the faintest galaxy in the catalogue of @xcite ] ( see the text ) .
different colours are for the seven subgroups adopted from @xcite .
the black circles represent galaxies which are not included in those seven subgroups .
we also mark the name and location ( cross ) of the brightest galaxy of each subgroup . ]
the fact that the overall spatial distribution of galaxies in the ursa major cluster is patchy ( see fig .
[ f5 ] ) is already a hint towards the existence of several subgroups .
based on the local supercluster galaxy group catalogue of @xcite , which was constructed by an algorithm of the modified friend - of - friend method , @xcite have suggested that the ursa major cluster includes seven subgroups . in order to confirm the existence of subgroups , we carried out the dressler@xmath30shectman test ( ds test ; @xcite ) which allows us to quantify the substructure in a galaxy cluster using the line - of - sight velocities and the projected distribution of galaxies .
the deviations with respect to the systemic velocity ( @xmath45 ) and the velocity dispersion ( @xmath46 ) of the cluster are calculated for each galaxy as @xmath47,\ ] ] where n is the number of the nearest galaxies , @xmath48 and @xmath49 are their mean velocity and velocity dispersion , respectively . here
we adopt n = @xmath50 from @xcite .
n@xmath51 is the total number of members in the cluster . in fig .
[ f10 ] , we present the ds test bubble plot for galaxies .
the circle size is proportional to @xmath52 .
this @xmath53 statistic is used as the significance of the substructure , measuring the local deviation of each galaxy from a smooth , virialized velocity and spatial distribution . for comparison
, we also overplot the seven subgroups of @xcite with different colours .
the location of the brightest main galaxy of each subgroup is also indicated with a cross . in the most prominent subgroup
ngc 3992 , the galaxies appear to be concentrated around the brightest galaxy of the subgroup .
the central region of the ursa major cluster consists of galaxies showing small @xmath53 values indicated by small circles and is dominated by galaxies associated with ngc 3877 as well as with the ngc 4346/ngc 4217 subgroups . in the northern part of the cluster , there is a distinct subgroup with large @xmath53 values ( red circles ) , which corresponds to the ngc 3992 subgroup .
in addition , one subgroup is found in the southern region , which might be associated with ngc 4111 ( yellow circles ) .
while the substructure of a cluster can be barely distinguished by the velocity distribution alone , by means of ds test we conclude that the ursa major cluster does consist of several subgroups , confirming the findings of @xcite . in fig .
[ f11 ] , we present the projected spatial distribution of early - type ( s0 , de , and ds0 ; upper panel ) and late - type ( spiral and irr ; bottom panel ) galaxies separately .
as @xcite noted , the galaxy distribution in the ursa major cluster is significantly elongated along the supergalactic equator ( see also fig .
[ f2 ] ) , suggesting an unrelaxed , dynamically young structure .
late - type galaxies in the ursa major cluster are distributed over the whole area of the cluster .
on the other hand , the early - type dwarf galaxies tend to be more clustered .
the largest number of early - type dwarf galaxies are found towards the ngc 3992 subgroup .
a two - dimensional kolmogorov - smirnov ( ks ) test on the spatial distributions between early - type galaxies and late - type galaxies yields a probability of only 2 per cent for the same parent distribution of the two subsamples . and nuv@xmath54 cmrs , the red thin solid lines represent the linear least - square fit to the early - type galaxies in the virgo cluster . in the nuv@xmath54 cmr ,
the red and green thick solid lines indicate the red sequence and blue cloud defined by @xcite .
the two vertical dashed lines indicate the magnitude range ( @xmath6 @xmath1 m@xmath42 @xmath1 @xmath7 ) of a possible gap between s0s and early - type dwarf galaxies in the ursa major cluster .
rms error bars for ursa major bright ( m@xmath42 @xmath1 @xmath6 ) and faint ( m@xmath42 @xmath27 @xmath6 ) galaxies are shown in the upper left and right corner of each cmr panel .
note that error bars in the uv cmrs are omitted because their sizes are much smaller than the symbol size .
even for faint red galaxies , colour errors rarely exceed 0.1 and 0.2 mag for nuv@xmath54 and fuv@xmath54 , respectively . ] in fig .
[ f12 ] , we present @xmath29 , nuv@xmath54 , and fuv@xmath54 colour - magnitude relations ( cmrs ) of galaxies in the ursa major cluster ( large coloured symbols ) . for comparison with galaxies in a different environment ,
we also overplot cmrs of the virgo cluster ( dots ) based on sdss optical and @xmath15 uv photometry of galaxies in the virgo cluster catalogue ( see @xcite ) .
we adopt a distance modulus of the virgo cluster of 31.01 from @xcite .
the cmrs of galaxies in the ursa major cluster follow the general trend in the sense that galaxies become progressively bluer with decreasing optical luminosity .
at all luminosities , the ursa major cluster sample is dominated by late - type galaxies in the blue cloud of the cmrs , in contrast to the virgo cluster . in the optical cmr , the early - type galaxies ( s0/de / ds0 ) of the ursa major cluster exhibit a well - defined sequence . while they do follow closely the red sequence of their virgo counterparts ( red thin solid line in fig .
[ f12 ] ) , the early - type dwarf galaxies with m@xmath42 @xmath27 @xmath7 mag show a small blueward offset : 57 per cent lie blueward of the virgo red sequence in g - r ( top panel ) .
this offset becomes more prominent when using uv - optical colours : 78 per cent of the ursa major early - type dwarf galaxies lie blueward of the virgo red sequence in nuv@xmath54 ( middle panel ) .
they occupy the region between red sequence and blue cloud as defined by wyder et al .
( 2007 , red and green thick solid lines ) , which was based on galaxies in the local universe with a redshift range of 0.01 @xmath1 z @xmath1 0.25 and @xmath4band magnitudes in the range of 14.5 @xmath1 r @xmath1 17.6 .
note that some fraction of the faint early - type dwarf galaxies in the virgo cluster also show blue nuv@xmath54 and fuv@xmath54 colours .
it has been suggested that most of these galaxies are in the outer region ( @xmath27 2@xmath9 from m87 ) of the virgo cluster and follow slightly steeper nuv@xmath54 and fuv@xmath54 cmrs than those of their counterparts in the inner region ( see @xcite for details ) . in the present case , we find that the faint early - type dwarf galaxies in the ursa major cluster have bluer uv@xmath54 colours than counterparts in the outskirts of the virgo cluster .
we calculate mean values of nuv@xmath54 colour of early - type dwarf galaxies in the ursa major and virgo outskirts are 3.7 and 4.2 with standard deviations of 0.81 and 0.63 mag , respectively , in the magnitude range of @xmath7 @xmath1 m@xmath0 @xmath1 @xmath55 .
the uv cmrs are particularly efficient tools to trace recent star formation , owing to their sensitivity to young ( @xmath1 1 gyr ) stellar populations ( e.g. , @xcite ; @xcite ) .
this indicates that most faint early - type dwarf galaxies in the ursa major cluster have undergone recent star formation ( @xcite ; @xcite ; @xcite ; @xcite , see also sec .
we present the @xmath29 cmr separately for the three most massive subgroups in the ursa major cluster ( ngc 3992 , ngc 4111 , and ngc 3877 ; see left - hand panels of fig .
[ f13 ] ) . due to small number statistics , the other four subgroups are not considered .
we adopt the velocity dispersions of the subgroups reported in @xcite .
we determine the mean fraction of early - type galaxies , f@xmath56 , for the three subgroups .
the ngc 3992 group , the most massive subgroup with @xmath57 = 122 km s@xmath14 , has a relatively well populated red sequence with f@xmath56 @xmath8 0.28 ( i.e. , 18/64 ) . about half ( 55 per cent , 18/33 ) of the early - type galaxies in the ursa major cluster seem to be associated with the ngc 3992 group .
the second massive subgroup , ngc 4111 ( @xmath57 = 106 km s@xmath14 ) , has f@xmath56 @xmath8 0.19 ( i.e. , 6/31 ) , whereas ngc 3877 ( @xmath57 = 65 km s@xmath14 ) contains only late - type galaxies .
this indicates a correlation between the fraction of early - type dwarf galaxies with the velocity dispersion of the subgroup ( see right - hand panel of fig .
[ f13 ] ) .
our result is in accordance with previous results based on the massive early - type sample in poor groups ( e.g. , @xcite ; @xcite ) .
magnitude relations of early - type dwarfs in the ursa major cluster .
the des and ds0s are represented by red filled circles and red crosses , respectively . the blue - cored early - type dwarf galaxies ( etdg(bc ) ) are additionally marked by blue triangles . for comparison
, the distributions of des and ds0s in the virgo cluster are overplotted by red and green contours , respectively .
the red dotted line and green dashed line represent linear least - squares fits to the des and ds0s in the virgo cluster , respectively ( kim et al .
the black open triangles indicate blue - cored early - type dwarfs in the virgo cluster defined by @xcite .
rms error bars for ursa major bright ( m@xmath42 @xmath1 @xmath58 ) and faint ( m@xmath42 @xmath27 @xmath58 ) galaxies are shown as error bars in the upper - left and upper - right corner of each cmr panel . ]
colour versus ew ( h@xmath59 ) for early - type dwarfs in the ursa major cluster , in comparison to virgo blue - cored early - type dwarfs ( black triangles ) .
the ew ( h@xmath59 ) of all galaxies are extracted from the sdss archive .
the distributions of nuv@xmath54 and ew ( h@xmath59 ) of galaxies are presented in the right and upper histograms , respectively . ] for early - type dwarf galaxies in the ursa major cluster .
the blue - cored early - type dwarfs and late - type galaxies in the ursa major cluster are denoted by blue triangles and grey filled boxes , respectively . for comparison
, we also plot blue - cored early - type dwarfs in the virgo cluster ( black triangles ) .
the adjacent panels show blue - cored early - type fraction of the ursa major cluster and virgo cluster as a function of density ( upper ) and m@xmath42 ( right ) , using poisson errors for the virgo cluster fractions and binomial errors for the ursa major cluster fractions . ] in order to confirm the presence of blue cores and positive colour gradients , we have examined the radial @xmath60 colour profiles of all ursa major early - type dwarf galaxies by performing surface photometry . for that purpose ,
background subtracted @xmath3 , @xmath4 , @xmath5band images were aligned and stacked to increase the s / n ratio .
elliptical annuli for each galaxy were determined using the ellipse fitting procedure of ellipse @xcite . in the ellipse fit
, we fixed the position angle , ellipticity , and centre of the isophotes using the values returned from se .
we then measured separately the @xmath18 and @xmath5band surface brightness profiles using the same elliptical annuli , so that the @xmath60 colour is derived from the same region . out of 23 early - type dwarf galaxies ,
16 galaxies show a blue cores based on their radial @xmath60 colour profiles ( see fig .
[ f14 ] ) .
while some of those 16 galaxies have relatively weak colour gradients ( e.g. , u31 and u159 ) , most galaxies with blue cores exhibit colour differences of more than 0.2 mag between the central and outer regions . in several galaxies , central irregularities
are also identified , which are presumably caused by gas and dust associated with ongoing or recent star formation activity ( e.g. , @xcite for virgo blue - cored early - type dwarfs ) . in fig .
[ f15 ] we also present the @xmath60 colour profiles and images of the seven early - type dwarf galaxies without blue cores . in fig .
[ f16 ] , we present @xmath29 , nuv@xmath54 , and fuv@xmath54 cmrs for the des ( red circles ) and ds0s ( red crosses ) in the ursa major cluster . for comparison , we overplot the distribution of the des ( red contours ) and ds0s ( green contours ) in the virgo cluster . the virgo des and ds0s show a similar distribution in the @xmath29 cmrs ( upper panel ) . however , the ds0s form a distinct sequence in the uv - optical cmrs ( middle and bottom panels ) which has a steeper slope compared to the des ( see also @xcite ) .
the ds0s in the ursa major cluster follow the ds0 sequence of the virgo cluster at the faint end in the uv - optical cmrs .
overall , ds0s in the ursa major cluster show bluer uv - optical colours than the des at a given magnitude , which is also found in the virgo cluster . on the other hand , a large fraction ( 83 per cent ) of ursa major des are bluer than the mean distribution of virgo des , especially in fuv@xmath54 cmr .
since the uv flux is particularly sensitive to the presence of young ( @xmath11 gyr ) stars , the blue uv - optical colours of blue - cored early - type dwarfs indicate that these galaxies have experienced recent or ongoing star formation activities in their central regions . in order to confirm this
, we examined integrated sdss spectra .
the sdss spectroscopic fibre diameter ( 3 arcsec ) corresponds to 253 pc at the distance of the ursa major cluster , providing spectral information for the very central region of the galaxy . in fig .
[ f17 ] , we show the distribution of equivalent width of the h@xmath59 emission line ( ew ( h@xmath59 ) ) for early - type dwarfs against the nuv@xmath54 colour .
h@xmath59 emission traces the most massive o- and b - type stars of ages less than a few million years @xcite .
as expected , nuv@xmath54 colour is correlated with ew ( h@xmath59 ) in the sense that bluer galaxies show higher ew ( h@xmath59 ) .
the majority of blue - cored early - type dwarfs ( triangles ) have a different distribution of ew ( h@xmath59 ) values than those of early - type dwarfs without blue core ( circles and crosses without triangles ) .
we note that the blue - cored early - type dwarfs tend to show more enhanced h@xmath59 emission line strengths ( see also upper histograms ) .
we find that 75 per cent of nuv - detected blue - cored early - type dwarfs exhibit residual star formation with ew ( h@xmath59 ) @xmath27 2 , which is a lower limit for star forming activity following @xcite .
furthermore , blue - cored early - type dwarfs in the ursa major cluster are distributed towards bluer nuv@xmath54 colour and higher ew ( h@xmath59 ) values than their virgo counterparts , indicating more active recent or ongoing star formation in the blue - cored early - type dwarfs in the ursa major cluster ( see right histograms in fig .
[ f17 ] ) . in fig .
[ f18 ] we present the projected local densities against the luminosities of the early - type dwarfs in the ursa major cluster comparing with blue - cored early - type dwarfs in the virgo cluster .
we also compare the densities of blue - cored early - type dwarfs with the distribution of spiral and irregular galaxies in the ursa major cluster .
following @xcite , we calculate the projected local density of each galaxy by defining a circular area around the galaxy that encloses the 10th nearest - neighbour galaxy .
the projected local density is then the number of galaxies per square mpc within this circle .
first of all , the density distribution of blue - cored early - type dwarfs in the ursa major cluster shows no significant difference with that of spiral and irregular galaxies .
a ks test gives the probability of 27 per cent that the densities of blue - cored early - type dwarfs and spiral / irregular galaxies are drawn from the same distribution .
this suggests that blue - cored early - type dwarfs share similar density environments of late - type galaxies .
@xcite reported that the blue - cored early - type dwarfs in the virgo cluster are found in moderate to low - density cluster regions with no centrally clustered projected spatial distribution , whereas normal early - type dwarfs dominate the high - density regions .
note that blue - cored early - type dwarfs in the ursa major cluster systematically reside in the lower density environments ( about a factor of 6 ) compared to those in the virgo cluster .
this indicates that these galaxies are also frequently found in more sparsely populated regions compared to the centre of the cluster . on the other hand ,
the blue - cored early - type dwarfs in the ursa major cluster are distributed towards lower luminosity compared to those in the virgo cluster .
note that virgo blue - cored early - type dwarfs are restricted to brighter magnitudes with m@xmath61 @xmath1 16 , and the decline of the number fraction of these galaxies as a function of magnitude is real and not caused by s / n effects ( see @xcite for details ) .
we present the results of optical and ultraviolet photometric properties of the ursa major cluster galaxies . also we performed a comparison study of the ursa major cluster and the virgo cluster to investigate the environmental dependence of galaxy properties .
a. we have constructed a new catalogue of the ursa major cluster galaxies with available spectroscopic data from sdss and ned .
membership was defined within a [email protected] radius circle centred at ra(b1950 ) = 11@xmath10 [email protected] and dec.(b1950 ) = @xmath1249@xmath9 22@xmath13 , following @xcite , and a velocity range of 677 @xmath1 v@xmath16 @xmath1 1302 km s@xmath14 . in total 166 galaxies brighter than m@xmath62 mag
were selected as members , more than twice as many as reported in @xcite . out of our 166 galaxies , 77 and 141 ones
are also included as members in @xcite and @xcite , respectively .
b. we derived sdss optical and @xmath15 uv photometric parameters of all cluster members .
morphological classification was carried out by examining sdss images . except for early - type dwarf galaxies ,
our morphological classification is largely consistent with those of @xcite and @xcite .
our new catalogue of the ursa major cluster includes fundamental information on galaxies such as morphology , radial velocity , and optical / uv photometry . c. we found evidence for the existence of several subgroups in the ursa major cluster based on the ds test , confirming previous findings of @xcite .
while the ursa major cluster is dominated by late - type galaxies , in the most massive subgroup ngc 3992 has a relatively large population of early - type galaxies ( s0/de / ds0 ) with early - type fraction of 28 per cent ( 18/64 ) .
about half of all early - type galaxies ( 55 per cent , 18/33 ) in the ursa major cluster are members of this subgroup .
the second massive subgroup around ngc 4111 has an early - type fraction of @xmath819 per cent ( 6/31 ) .
the ngc 3769 and ngc 3877 subgroups have only late - type galaxies .
d. we analysed the cmrs for galaxies in the ursa major cluster and compared them with the virgo cluster , using the @xmath29 , nuv@xmath54 , and fuv@xmath54 colours as a function of absolute @xmath4band magnitude . the ursa major cluster is significantly underpopulated in the red sequence . in particular , there are no luminous early - type dwarf galaxies ( brighter than m@xmath63 @xmath7 mag ) in the cluster . in the @xmath29 cmr ,
the slopes of the red sequences of the two cluster populations are very similar . the faint ( m@xmath0 @xmath27 @xmath7 ) early - type dwarf galaxies lie blueward of the virgo red sequence in uv@xmath54 cmrs . e. we discover blue - cored early - type dwarf galaxies , which are already known to exist in the outskirts of the fornax and virgo clusters ( @xcite ; @xcite ) as well as in the field @xcite
. out of 23 ursa major early - type dwarf galaxies , 16 galaxies ( @xmath870 per cent ) show evidence of central star formation .
the fraction of blue - cored early - type dwarfs in the ursa major cluster is much higher than that of the virgo cluster and extends to fainter magnitude .
their overall uv - optical colours are consistent with the ds0 sequence of the virgo cluster at the faint end .
the most striking result of our study is that about 70 per cent of the early - type dwarf galaxy population in the ursa major cluster have blue cores with evidence of ongoing or recent star formation .
in contrast , only about 5 per cent of such galaxies are found in the virgo cluster @xcite . while the frequency of the blue - cored early - type dwarfs in group environments seems to depend on the stage of dynamical evolution of the group , a substantial fraction ( more than 30 per cent ) of blue - cored early - type dwarfs is also found in nearby groups ( e.g. , ngc 5846 and ngc 5353/4 groups ; @xcite ; @xcite ) . especially in the ngc 5353/4 group , about two thirds of the early - type galaxies
are identified as candidates of blue - cored early - type dwarfs ( see @xcite ) .
this suggests that low - density environments are very favorable to the formation and/or survival of blue - cored early - type dwarfs .
based on the results of blue - cored early - type dwarfs in the virgo cluster , several possible scenarios for the formation of these galaxies in cluster environments have been suggested ( see @xcite ; @xcite for details ) .
galaxy harassment @xcite within the cluster appears to be a possible mechanism to transform the infalling late - type disc galaxies to galaxies with early - type morphology . in this process
, blue central region with enhanced density could be formed by funnelled gas into the centre @xcite and subsequent star formation . ram pressure stripping of dwarf irregular galaxies ( dirrs ) by the hot intracluster medium could be responsible for the shaping the blue - cored early - type dwarfs , in which interstellar gas in dirrs are removed except around the central region of the galaxy .
on the other hand , it is also plausible that blue - cored early - type dwarfs could be formed via star formation in central regions of dirrs triggered by tidal interaction with other galaxies . by several bursts of star formation ,
dirrs become galaxies classified as blue compact dwarf galaxies @xcite .
@xcite suggested that evolutionary connections between early - type dwarfs and blue compact dwarfs are possible based on the largely overlapping range of their structural parameters ( such as effective surface brightness and radius ) , for which only the mass - dominating old stellar population of the blue compact dwarfs was considered . in the ursa major cluster , more than half of irrs are hsbs ( 29/54 ) that could be potential candidates of blue compact dwarf galaxies .
as hypothesized by @xcite , the presence of dwarf galaxies exhibiting star formation at their centre may be less related to direct external influence on the galaxy , but instead more governed by how much of the original gas reservoir remains available when a galaxy enters the cluster potential @xcite . in this context
, it is noteworthy that @xcite recently found that the fornax cluster dwarf elliptical galaxy fcc046 , while having a centrally concentrated young stellar population , is surrounded by a significant gas reservoir ( @xmath810@xmath64 m@xmath65 of hi gas ) .
this galaxy is located in the outskirts of the fornax cluster , far from the x - ray halo .
this discovery strengthens the argument that accreting gas into the galaxy can be a fuel of central star formation .
it was found that the fraction of massive early - type galaxies ( e / s0 ) and dwarf elliptical galaxies is higher in a dynamically more evolved system with high velocity dispersion ( @xcite ; @xcite ; @xcite ) . especially in the case of the ngc 3992 subgroup , which is the most dynamically evolved ursa major cluster subgroup , considering its high frequency of early - type galaxies , morphological transformation to early - type galaxies by interactions and merging between galaxies may have occurred more frequently in the early stages of its dynamical evolution than in the other subgroups .
we find that bright early - type dwarf galaxies ( @xmath6 @xmath1 m@xmath42 @xmath1 @xmath7 in fig .
[ f12 ] ) are absent in the ursa major cluster at a significance level of 3@xmath24 .
this is based on bootstrap resampling @xcite using the luminosity function of galaxy types in the virgo cluster as reference @xmath1 m@xmath42 @xmath1 @xmath7 occurs in less than 0.2 per cent of cases , implying that this gap in the ursa major early types is significant at the 3@xmath24 level . ] .
the fact that environmental effects are only of weak to moderate strength is likely to be responsible for the absence of bright early - type dwarf galaxies : no late - type galaxies in this luminosity range were transformed into early types .
however , from the analysis of colour gradients of the early - type dwarf galaxies in the ursa major cluster ( figs .
[ f14 ] and [ f15 ] ) , the colour in their outer part is as red as those in the virgo cluster . the outer colour of most blue cored early - type dwarf galaxies in the ursa major cluster is within the @xmath662@xmath24 region around the red sequence of the virgo cluster early - type dwarfs .
in contrast to the virgo cluster environment , ram - pressure stripping as a mechanism regulating galaxy evolution is clearly less relevant in the ursa major cluster .
the spiral galaxies in the ursa major cluster do not show any prominent hint of hi deficiency or gas stripping ( see fig .
1 of @xcite ) , while many examples for ram pressure stripping are found in the virgo cluster ( e.g. , @xcite ) .
this is also supported by no detection of intracluster medium in the ursa major cluster @xcite .
galaxy harassment , which is a mechanism for transforming late - type galaxies into des through frequent high - speed encounters with massive galaxies , prevails in cluster environments @xcite .
however , this process would be insignificant in groups of relatively low density .
we adopt the concept of collision rate described by @xmath67 from @xcite to calculate how insignificant harassment is in the ursa major cluster as compared to the virgo cluster .
@xmath68 is the number density , @xmath24 is the velocity dispersion of cluster , and @xmath69 is the cross - section of the target galaxy .
the number density @xmath68 is calculated from the number of galaxies of m@xmath42 @xmath1 @xmath58 within 1.5 mpc of each cluster .
the result confirms that the galaxies in the ursa major cluster are barely affected by multiple tidal influence with low efficiencies @xmath70 for harassment . in group environments with relatively low velocity dispersion
, it has been known that interactions between member galaxies are more frequent ( @xcite ; @xcite ; @xcite ; @xcite ) .
this is supported by hi surveys of galaxies in the ursa major cluster in which a large fraction of galaxies show evidence of tidal interactions ( see @xcite ; @xcite ; @xcite ; @xcite ) . from their extensive hi observations of 43 spiral galaxies , @xcite found that 10 galaxies show clear signs of galaxy - galaxy interactions . additionally , about half of the sample galaxies also exhibit features of warps or asymmetries in their hi surface density profile of discs .
most recently , from a blind hi survey of the ursa major cluster , @xcite also found a fraction ( 33 per cent ) of candidates showing tidal interactions . furthermore , note that all three brightest s0 galaxies ( ngc 3998 , ngc 4026 , and ngc 4111 ) in the ursa major cluster show extended hi filaments , indicative of tidal gas stripping from the disc of nearby companions ( @xcite ; @xcite ; see also fig . [ f12 ] of @xcite for the extended hi envelope around ngc 4026 and ngc 4111 ) . all of these results provide direct evidence of considerable frequency of interactions between galaxies in the ursa major cluster and then support the view that tidal interactions may be the dominant mechanism that transforms late - type dwarf galaxies into early - types in group environments . in this context , tidally induced star formation of dirrs and their subsequent evolution should be the most plausible scenario related with the formation of the blue - cored early - type dwarfs in the ursa major cluster . in order to compare interaction probabilities
, we calculate the perturbation parameter @xmath71 for the galaxies in the ursa major cluster and the virgo cluster .
the @xmath71 value is defined by @xcite as @xmath72 where r is the size of the galaxy , d@xmath73 is the projected distance between the galaxy and the perturber , and m@xmath74 and m@xmath73 are the apparent magnitudes of the primary and perturber galaxies , respectively . the galaxy with the largest @xmath71 value is chosen to be the perturber for the primary galaxy .
this parameter means the ratio between the internal and tidal forces acting on the galaxy by a possible perturber .
the median @xmath71 values for virgo and ursa major early - type galaxies are @xmath75 and @xmath76 , with standard deviations of 1.21 and 1.47 , respectively .
their ratio is large ( 10@xmath77 @xmath8 0.008 ) .
however , if we assume that interactions between the primary and the perturber start in both clusters at the same time , the average interaction time - scale of galaxies in the ursa major cluster will be five times longer than virgo galaxies since velocity dispersion of virgo cluster ( 715 km s@xmath14 ) is larger than that of ursa major cluster ( 151 km s@xmath14 ) . while this longer interaction time could make both clusters more similar
, it is still not explained that the fraction of blue - cored early - type dwarfs in the ursa major cluster is larger than that of virgo cluster .
there is one possibility might solve this problem .
we hypothesize that the blue - cored early - type dwarf galaxies could be considered as possible transitional objects which are on the way to red early - type galaxies from the late - type progenitors ( @xcite ; @xcite ; @xcite ) .
although they are formed via tidal interactions in groups or clusters at the same rate , they have a different evolutionary path as a function of their mass or brightness and the environments where they were formed . in a severe environment like virgo
, less massive galaxies with a shallow potential well lose gas in their centres and will passively evolve to be seen as de / ds0 . in dense environments , only massive galaxies with a deep potential
well keep their gas in their centres , although losing it in their outer disk . on the other hand , in a less dense environment like ursa major , both massive and less massive galaxies
keep their gas in their centre .
consequently , ursa major possesses a much larger fraction of blue - cored early - type dwarfs than virgo .
the above hypothesis is supported by fig .
blue - cored early - type dwarfs in the virgo cluster are located on the brighter magnitude and higher density region than those in the ursa major cluster .
we conclude that blue - cored early - type dwarfs are found ubiquitously in all environments including the field ( e , g . , @xcite ) while their frequencies vary with environment and may probe the strength of environment . in a forthcoming paper ( kim et al . in preparation .
) , we will investigate blue - cored early - type dwarf candidates in the field environment in comparison with those in cluster ( e.g. , virgo ) and group ( e.g. , ursa major ) environments , which will allow to probe their formation and evolutionary path .
we are grateful to the anonymous referee for helpful comments and suggestions that improved the clarity and quality of this paper .
we thank marc a.w .
verheijen for useful discussions .
this research was supported by basic science research program through the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology ( nrf-2012r1a1b4003097 ) .
support for this work was also provided by the nrf of korea to the center for galaxy evolution research ( no .
2010 - 0027910 ) .
tl was supported within the framework of the excellence initiative by the german research foundation ( dfg ) through the heidelberg graduate school of fundamental physics ( grant number gsc 129/1 ) .
sk acknowledges support from the national junior research fellowship of nrf(no .
2011 - 0012618 ) .
abazajian k. n. et al . , 2009 , apjs , 182 , 543 barrow j. d. , bhavsar s. p. , sonoda d. h. , 1984 , mnras , 210 , 19p beers t. c. , flynn k. , gebhardt k. , 1990 , aj , 100 , 32 bekki k. , 2008 , mnras , 388 , l10 bertin e. , arnouts s. , 1996 , a&as , 117 , 393 binggeli b. , cameron l. m. , 1991 , a&a , 252 , 27 binggeli b. , sandage a. , tammann g. a. , 1988 , ara&a , 26 , 509 binggeli b. , tammann g. a. , sandage a. , 1987 , aj , 94 , 251 blanton m. r. et al . , 2005a , aj , 129 , 2562 blanton m. r. , lupton r. h. , schlegel d. j. , strauss m. a. , brinkmann j. , fukugita m. , loveday j. , 2005b , apj , 631 , 208 boselli a. , boissier s. , cortese l. , gavazzi g. , 2008 , apj , 674 , 742 boselli a. et al . , 2005 ,
apj , 629 , l29 buta r. j. , 2013 , in oswalt t. d. , keel w. c. , eds . , planets , stars and stellar systems , vol . 6 : galaxy morphology .
springer - verlag , berlin , p. 1
byrd g. , valtonen m. , 1990 , apj , 350 , 89 cardelli j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 cellone s. a. , buzzoni a. , 2005 , mnras , 356 , 41 chung a. , van gorkom j. h. , kenney j. d. p. , crowl h. , vollmer b. , 2009 , aj , 138 , 1741 chung a. , van gorkom j. h. , kenney j. d. p. , vollmer b. , 2007 , apj , 659 , l115 davies j. i. , phillipps s. , 1988 , mnras , 233 , 553 de looze i. et al . , 2010 ,
a&a , 518 , l54 de lucia g. , weinmann s. , poggianti b. m. , aragn salamanca a. , zaritsky d. , 2012 , mnras , 423 , 1277 de rijcke s. , buyle p. , koleva m. , 2013 , apj , 770 , l26 de rijcke s. , zeilinger w. w. , dejonghe h. , hau g. k. t. , 2003 , mnras , 339 , 225 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . , et al .
1991 , sky telesc .
, 82 , 621 dressler a. , 1980 , apj , 236 , 351 dressler a. et al . , 1997 , apj , 490 , 577 dressler a. , shectman s. a. , 1988 , aj , 95 , 985 ferguson h. c. , sandage a. , 1991 , aj , 101 , 765 gu q. , zhao y. , shi l. , peng z. , luo x. , 2006 , aj , 131 , 806 gunn j. e. , gott , iii j. r. , 1972 , apj , 176 , 1 hallenbeck g. et al . , 2012 , aj , 144 , 87 jedrzejewski r. i. , 1987 , mnras , 226 , 747 jerjen h. , 2003 , a&a , 398 , 63 jerjen h. , binggeli b. , barazza f. d. , 2004 , aj , 127 , 771 jerjen h. , tammann g. a. , 1997 , a&a , 321 , 713 karachentsev i. d. , nasonova o. g. , courtois h. m. , 2013 , mnras , 429 , 2264 kauffmann g. , white s. d. m. , guiderdoni b. , 1993 , mnras , 264 , 201 kaviraj s. , rey s .- c . , rich r. m. , yoon s .- j .
, yi s. k. , 2007 , mnras , 381 , l74 kennicutt , jr .
r. c. , 1998 , ara&a , 36 , 189 kim s. , rey s .- c .
, lisker t. , sohn s. t. , 2010 , apj , 721 , l72 kodama t. , smail i. , 2001 , mnras , 326 , 637 lisker t. , 2009 , astron .
, 330 , 1043 lisker t. , glatt k. , westera p. , grebel e. k. , 2006 , aj , 132 , 2432 lisker t. , grebel e. k. , binggeli b. , 2008 , aj , 135 , 380 lisker t. , grebel e. k. , binggeli b. , glatt k. , 2007 , apj , 660 , 1186 lisker t. , weinmann s. m. , janz j. , meyer h. t. , 2013 , mnras , 432 , 1162 lupton r. h. , gunn j. e. , szalay a. s. , 1999 , aj , 118 , 1406 mayer l. , governato f. , colpi m. , moore b. , quinn t. , wadsley j. , stadel j. , lake g. , 2001a , apj , 559 , 754 mayer l. , governato f. , colpi m. , moore b. , quinn t. , wadsley j. , stadel j. , lake g. , 2001b , apj , 547 , l123 mcgee s. l. , balogh m. l. , bower r. g. , font a. s. , mccarthy i. g. , 2009 , mnras , 400 , 937 mcintosh d. h. , guo y. , hertzberg j. , katz n. , mo h. j. , van den bosch f. c. , yang x. , 2008 , mnras , 388 , 1537 meyer h. t. , lisker t. , janz j. , papaderos p. , 2014 ,
a&a , 562 , a49 moore b. , katz n. , lake g. , dressler a. , oemler a. , 1996 , nature , 379 , 613 moore b. , lake g. , katz n. , 1998 , apj , 495 , 139 morrissey p. et al . , 2007 ,
apjs , 173 , 682 nair p. b. , abraham r. g. , 2010 , apjs , 186 , 427 oke j. b. , 1990 , aj , 99 , 1621 pinkney j. , roettiger k. , burns j. o. , bird c. m. , 1996 , apjs , 104 , 1 postman m. , geller m. j. , 1984 , apj , 281 , 95 roediger e. , 2009 , astron .
nachr . , 330 , 888 sandage a. , binggeli b. , 1984 , aj , 89 , 919 sandage a. , binggeli b. , tammann g. a. , 1985 , aj , 90 , 395 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 shaya e. j. , tully r. b. , 1984 , apj , 281 , 56 springel v. et al .
, 2005 , nature , 435 , 629 strauss m. a. et al .
, 2002 , aj , 124 , 1810 tanaka m. , kodama t. , arimoto n. , okamura s. , umetsu k. , shimasaku k. , tanaka i. , yamada t. , 2005 , mnras , 362 , 268 toomre a. , toomre j. , 1972 , apj , 178 , 623 tovmassian h. m. , plionis m. , 2009 , apj , 696 , 1441 tovmassian h. m. , plionis m. , andernach h. , 2004 , apj , 617 , l111 trentham n. , tully r. b. , 2002 , mnras , 335 , 712 trentham n. , tully r. b. , verheijen m. a. w. , 2001 , mnras , 325 , 385 tully r. b. , 1987 , apj , 321 , 280 tully r. b. , courtois h. m. , 2012 , apj , 749 , 78 tully r. b. , trentham n. , 2008 , aj , 135 , 1488 tully r. b. , verheijen m. a. w. , pierce m. j. , huang j .- s . , wainscoat r. j. , 1996 , aj , 112 , 2471 varela j. , moles m. , mrquez i. , galletta g. , masegosa j. , bettoni d. , 2004 , a&a , 420 , 873 west a. a. , garcia - appadoo d. a. , dalcanton j. j. , disney m. j. , rockosi c. m. , ivezi . , bentz m. c. , brinkmann j. , 2010 , aj , 139 , 315 wolfinger k. , kilborn v. a. , koribalski b. s. , minchin r. f. , boyce p. j. , disney m. j. , lang r. h. , jordan c. a. , 2013 , mnras , 428 , 1790 wyder t. k. et al . , 2007 , apjs , 173 , 293 yi s. k. et al .
, 2005 , apj , 619 , l111 zabludoff a. i. , mulchaey j. s. , 1998 , apj , 496 , 39 | using sdss - dr7 and nasa / ipac extragalactic database spectroscopic data , we identify 166 galaxies as members of the ursa major cluster with m@xmath0 @xmath1 @xmath2 mag .
we morphological classify all galaxies by means of carefully inspecting @xmath3 , @xmath4 , @xmath5band colour and monochromatic images .
we show that the ursa major cluster is dominated by late - type galaxies , but also contains a significant number of early - type galaxies , particularly in the dwarf regime .
we present further evidence for the existence of several subgroups in the cluster , consistent with previous findings .
the early - type fraction is found to correlate with the mass of the subgroup .
we also investigate environmental effects by comparing the properties of the ursa major early - type dwarf galaxies to those of the virgo cluster .
in contrast to the virgo , the red sequence of the ursa major cluster is only sparsely populated in the optical and ultraviolet colour - magnitude relations .
it also shows a statistically significant gap between @xmath6 @xmath1 m@xmath0 @xmath1 @xmath7 mag , i.e. the ursa major cluster lacks early - type dwarf galaxies at the bright end of their luminosity function .
we discover that the majority of early - type dwarf galaxies in the ursa major cluster have blue cores with hints of recent or ongoing star formation .
we suggest that gravitational tidal interactions can trigger central blue star forming regions in early - type dwarfs .
after that , star formation would only fade completely when the galaxies experience ram pressure stripping or harassment , both of which are nearly absent in the ursa major cluster .
[ firstpage ] galaxies : clusters : individual : ursa major - galaxies : dwarf - galaxies : evolution - galaxies : structure |
since 1979 it became clear that non - interacting electrons are always localized in a 2d disordered potential @xcite . at present
this result is firmly confirmed by different analytical and numerical studies ( see the reviews @xcite ) according to which all one - particle eigenstates are exponentially localized in the case of orthogonal or unitary symmetry .
delocalization is possible only for the symplectic symmetry which is however not realized in the absence of spin - orbit interaction @xcite .
therefore , the discovery of metallic behavior in 2d high mobility samples by kravchenko _ et .
@xcite attracted a great interest of the solid - state community and pushed forward the question about a role of electron - electron interactions in the localized phase .
indeed , in many experiments of different groups @xcite the ratio of the coulomb interaction energy @xmath4 to the fermi energy @xmath5 is characterized by a large dimensionless parameter @xmath6 , where @xmath7 is the electron density in 2d , and @xmath8 , @xmath9 , @xmath10 are the effective bohr radius , electron mass and dielectric constant , respectively . typically in the above experiments
the parameter @xmath0 is ranged in the interval 5 - 40 showing that these experiments are well outside of the perturbative regime corresponding to @xmath11 . generally , in these experiments
the insulating increase of resistivity at low temperature @xmath12 is observed at large @xmath0 values ( low charge density @xmath7 ) , while its metallic decrease at low @xmath12 is seen at lower @xmath0 ( higher @xmath7 ) .
in addition , experimentally it was found that a sufficiently strong magnetic field parallel to the 2d plane destroys the metallic behavior @xcite .
these results show an important role of the spin degrees of freedom .
however , we should note that the recent experiments @xcite with n - sige samples demonstrate that the metallic behavior persists even in very strong in plane magnetic field when all spins are polarized .
another important experimental indication was found for a localized phase with a variable range hopping ( vrh ) @xcite .
indeed , the experiments show that a prefactor in the exponential vrh resistivity dependence on temperature is phonon independent @xcite . for the theoretical explication of this result other physical mechanisms of vrh
should be found @xcite and it is possible that their origin is related to the electron - electron interaction as it had beed discussed long ago in @xcite . from the theoretical view point
the problem of interaction in the localized phase is rather nontrivial . indeed , due to localization of noninteracting states , the two interacting particles ( tip ) , with a short range repulsive / attractive interaction and located on a distance of one - particle localization length @xmath13 from each other , always return to their center of mass that enhances enormously their interaction . as a result the two particles can propagate coherently on a distance much larger @xmath13 .
this effect has been discussed recently by different groups for a short range interaction @xcite ( see also the early paper by dorokhov for a strongly attractive case @xcite ) . however , in the experiments discussed above the charge density is very low and the distance between electrons ( holes ) is much larger than @xmath14 ( @xmath15 ) . in this situation
the localized electrons interact via the long range coulomb interaction .
recently , it has been shown @xcite that in this situation two interacting electrons can be delocalized even when they are separated by a distance @xmath16 .
moreover , this delocalization goes in a way very similar to the anderson transition in 3d . in this paper
we discuss the physical origin of this delocalization ( section ii ) .
after that ( section iii ) we present the results of extensive numerical studies of up to @xmath2 electrons ( with polarized spins ) on 2d disordered lattice . in agreement with the first results presented in @xcite
we establish that the ground state remains localized ( nonergodic ) but at moderate values of parameter @xmath0 the electrons become ergodic at very low total energy where their spectral statistics demonstrates a transition from the poisson to the wigner - dyson distribution .
the conclusion is presented in section iv .
let us consider a system of electrons with polarized spins on a 2d disordered lattice ( anderson model ) .
such polarization can be reached by a strong in plane magnetic field but also one can speak about spinless fermions .
the system is described by the hamiltonian @xmath17 here @xmath18 is the fermion creation ( annihilation ) operator at site @xmath19 , the hopping between the nearest neighbors is @xmath20 , the diagonal energies @xmath21 are randomly distributed within the interval @xmath22 $ ] and @xmath23 is the strength of the coulomb interaction with @xmath24 being the distance between electrons at sites @xmath25 . in this notation @xmath26
is the occupation number at site @xmath19 .
the electrons ( particles ) are moving in a 2d cell of size @xmath27 with periodic boundary conditions .
the coulomb interaction is taken between electrons in one cell of size @xmath28 and with 8 charge images in nearby 8 cells as in @xcite . the number of particles @xmath3 and the cell size were varied within the intervals @xmath29 and @xmath30 . with the notations of eq .
( 1 ) , the parameter @xmath0 is given by @xmath31 , where @xmath32 is the filling factor and @xmath33 .
the majority of our data have been obtained for @xmath34 , @xmath35 that corresponds to @xmath36 .
when changing @xmath3 the filling factor was kept approximately constant ( nearest rational value ) by the appropriate choice of @xmath28 . for @xmath37 all states in the hamiltonian ( 1 ) are localized and the one - particle localization length varies exponentially with @xmath38 in the limit of weak disorder : @xmath39 .
the numerical diagonalization @xcite allows to determine the inverse participation ratio ( ipr ) for one particle @xmath40 , which approximately gives the number of lattice sites contributing in one eigenstate .
the values found at @xmath41 for different disorder strength at the ground state and the middle of the band are correspondingly : @xmath40 = 3.4 and 4.2 @xmath42 ; 5.2 and 11.2 @xmath43 ; 8.2 and 36.7 @xmath44 ; 13.5 and 84.2 @xmath45 .
this shows that the one - particle states at low energy and even at the band center ( except maybe @xmath46 ) are well localized @xmath47 . to understand how two electrons interact in the 2d localized phase we should estimate the interaction induced transition rate @xmath48 between unperturbed @xmath49 localized eigenstates .
following @xcite let us assume that the distance between electrons @xmath50 is much larger than @xmath13 ( @xmath16 ) and their energy is in the middle of the band .
in this case the two - body interaction has a dipole - dipole form since the lower order terms only slightly modify the one - particle effective potential .
the two - body matrix element between noninteracting eigenstates is then @xmath51 , where we used that the dipole is of the order of @xmath13 ( electron can not jump on a distance larger @xmath13 due to exponential localization ) .
the sum @xmath52 of the product of four random one - particle wave functions @xmath53 runs over @xmath54 sites for each electron so that @xmath55 and @xmath56 .
the density of directly coupled two electron states is @xmath57 and according to the fermi golden rule the hopping rate is @xmath58 .
the mixing of two - electron levels takes place when @xmath59 , where @xmath60 is the value of @xmath0 at the filling @xmath61 ( one electron in a box of size @xmath13 with @xmath62 ) ) . for @xmath63
this mixing takes place at @xmath64 . in this sense
the situation is qualitatively different from the case of short range interaction where the mixing is possible only for particles on a distance @xmath65 . for @xmath66
the pair of two electrons becomes delocalized in a way similar to the 3d anderson transition .
indeed , the center of mass can move in the 2d plane and in addition electrons can rotate around a ring of radius @xmath50 and width @xmath13 keeping their total energy @xmath67 constant . since @xmath68 the dynamics is going in an effective 3d space where the anderson transition to diffusion takes place at @xmath66 .
formally , the diffusion of two electrons will be finally localized since the finite length of the ring makes the situation analogous to quasi-2d anderson model ( finite number of planes ) .
however , the corresponding tip localization length @xmath69 jumps from @xmath70 at @xmath71 to exponentially large @xmath72 for @xmath66 . in a similar way for @xmath73
coupled 2d anderson models the one - particle localization length @xmath13 makes a jump near the 3d transition point ( e.g. @xmath74 ) from @xmath75 to @xmath76 where the conductance @xmath77 . in the delocalized phase
the diffusion rate or the conductivity in units @xmath78 can be estimated as @xmath79 .
the above estimates show that the border between the insulating ( localized at @xmath80 ) and metallic ( delocalized at @xmath81 ) phases is : @xmath82 the physical meaning of this result is very natural : larger @xmath0 corresponds to a lower density @xmath7 with larger distance between electrons and smaller two - body interaction between them .
the first numerical studies of two polarized electrons in 2d anderson model showed a transition in the level spacing statistics @xmath1 from the poisson distribution @xmath83 near the ground state to the wigner - dyson statistics @xmath84 at higher energies @xcite in a way similar to the results of shklovskii _ et al . _
@xcite for the 3d anderson model . in both cases
the intermediate statistics @xmath1 are very close at the critical transition point ( fig . 2 in @xcite ) .
the numerical studies were done in the range of parameters @xmath85 , @xmath86 , @xmath87 , @xmath88 ; more details can be found in @xcite .
independent recent studies @xcite of a similar model with coulomb interaction at the center of the energy band also show a transition to @xmath84 as a function of @xmath38 . since even analytical formulas for the coulomb matrix element between localized states are absent further investigations of this problem are still needed .
for numerical studies of the hamiltonian ( 1 ) we follow the approach developed in @xcite and rewrite the hamiltonian matrix in the basis of one particle eigenstates ( orbitals ) at @xmath37 using the computed two - body matrix elements between the orbitals .
only a finite number @xmath73 of low energy orbitals has been considered and the final many - body hamiltonian matrix was constructed on the basis of a pyramid rule for the one - particle orbital index @xmath89 : @xmath90 .
this procedure allows to make an efficient reduction of the resulting matrix size @xmath91 comparing to @xmath92 without any serious modification of the properties of low energy states . to the maximum we used @xmath93 corresponding to @xmath94 ( for @xmath2 we used up to @xmath95 ) .
we checked that our results at low energy are not sensitive to the variation of @xmath73 and @xmath91 ( see the illustrations below ) .
one of the main characteristics we extracted from the numerical studies is the level spacing statistics @xmath1 at a given total excitation energy @xmath96 measured from the ground state .
the disorder average has been performed over @xmath97 realization ( for low @xmath96 ) and @xmath98 ( for higher @xmath96 ) . in this way , the total statistics used for @xmath1 varied from @xmath99 at low @xmath96 to @xmath100 at high @xmath96 . to study the variation of @xmath1 with the interaction or disorder strength it is convenient to use the parameter @xmath101 which is defined as @xmath102 , where @xmath103 is the smaller intersection point of @xmath83 and @xmath104 . in this way
@xmath105 corresponds to @xmath83 and @xmath101=0 to @xmath104 .
we remind that for the one - particle 3d anderson model the localized phase is characterized by @xmath105 , since localized ( nonergodic ) eigenstates do not feel each other , while in the metallic phase the eigenstates are extended and @xmath106 @xcite .
the results presented in @xcite show that at @xmath107 the parameter @xmath101 evolves to 1 for the total energy @xmath108 and 0 for @xmath109 with the growth of the system size @xmath28 at fixed filling @xmath110 .
the critical energy @xmath111 and @xmath112 at which the transition takes place are @xmath113 , @xmath114 ( @xmath115 ) and @xmath116 , @xmath117 ( @xmath118 ) with @xmath119 . for @xmath120
no transition is found for @xmath121 where @xmath122 .
the new data for @xmath46 are presented in fig . 1 where the transition is seen at @xmath123 , @xmath124 .
at the critical point @xmath125 the spectral statistics @xmath1 is independent of the number of particles and the system size reached in our simulations ( see fig . 3 in @xcite ) .
in fig.2 we show an illustration that at low energy the parameter @xmath101 is not sensitive to the variation of the total matrix size @xmath91 by more than 3 times that confirms the validity of our numerical approach .
we note that not only the values of @xmath101 are close for different @xmath91 but also the whole distributions @xmath1 , as it is illustrated in fig.3 for @xmath126 . the results of fig .
1 are obtained at fixed parameter @xmath0 .
the variation of the spectral statistics @xmath1 at low energy for the change of @xmath0 from @xmath127 to @xmath128 is shown in fig .
3 . these data demonstrate a clear approach to the poisson distribution with growing @xmath0 .
the same tendency is seen in the @xmath101 dependence on energy shown at different @xmath0 in fig .
4 . these results are in agreement with the relation ( 2 ) and the argument given in the previous section according to which at large @xmath0 the distance between electrons grows and the two - body interaction between them drops .
since only the interaction between two particles is able to mix many - body levels , the electron dynamics becomes nonergodic at large @xmath0 that leads to the poisson distribution . at the same time even at the optimal values of @xmath129 the ergodicity and level mixing appear only at total energy @xmath109 while for @xmath108 including the ground state the ergodicity is absent in the range of parameters we studied .
formally , one can object against the statement that the value @xmath105 implies the localization of electrons in space giving as a counterexample the case of noninteracting delocalized electrons in the metallic phase where also @xmath105 for the multi - electron spectrum .
however , in the system we study all one - particle states are exponentially localized and it is very difficult to imagine that the interaction gives delocalization of charge , makes the system metallic and does not introduce complete ergodicity and level mixing , which should give @xmath106 instead of @xmath105 found for @xmath108 .
due to this reason we consider that the appearance of poisson statistics near the ground state is the direct evidence of localization at @xmath108 . at the same time
the emergence of random matrix statistics at @xmath109 implies ergodicity and delocalization for many electron states .
this delocalization apparently appears in a way similar to the two electron delocalization discussed in the previous section but in addition the interaction between larger number of particles decreases the delocalization border comparing to the tip case discussed in @xcite .
an interesting point is that the transition at @xmath130 takes place at zero temperature . indeed , according to the data of fig .
1 at the transition the energy per particle @xmath131 is much smaller than the fermi energy @xmath132 showing that the data are close to the thermodynamic limit with @xmath133 .
our results show that the coulomb interaction between two electrons in excited states leads to their delocalization for @xmath134 while for @xmath80 they remain localized .
the transition between these phases is similar to the anderson transition in an effective dimension @xmath135 .
the numerical studies of the spectral statistics for many polarized electrons in the 2d anderson model show that for @xmath136 and @xmath85 the ground state is nonergodic ( localized ) and is characterized by the poisson statistics for the total energy @xmath137
. however , the transition to quantum ergodicity and the wigner - dyson statistics takes place at a fixed total energy @xmath111 independent of system size ( for @xmath138 , @xmath139 and fixed filling @xmath110 ) .
this implies a delocalization at zero temperature @xmath12 . at the critical point
@xmath111 the parameter @xmath112 increases with the disorder strength @xmath38 : @xmath140 and the critical statistics approaches to the poisson limit . in a certain sence
the situation is similar to the anderson transition in high dimensions @xmath141 discussed in @xcite where also @xmath112 is growing with @xmath142 . in analogy with this result and the case of two electrons in 2d
we make a conjecture that in the hamiltonian ( 1 ) the transition at @xmath111 is similar to a transition in some effective dimension @xmath143 .
this @xmath144 is growing with the disorder strength @xmath38 .
the interaction induced ergodicity at @xmath133 is in favor of the metal - insulator transition observed experimentally , especially in the view of recently observed metallic behavior for polarized electrons @xcite
. however , our data are not sufficient to determine the behavior of resistivity on temperature in the ergodic phase at @xmath109 .
therefore , it is not excluded that at strong disorder this ergodic phase will show a resistivity growth at low @xmath12 . in this case
one can suppose that the ergodicity induced by the coulomb interaction is responsible for the phononless vrh conductivity as it was argued in @xcite and indicated by the experiments @xcite .
more detailed investigations are required to understand the properties of the coulomb ergodic phase at @xmath109 .
s. v. kravchenko , g. v. kravchenko , j. e. furneaux , v. m. pudalov , and m. diorio , phys .
b * 50 * ( 1994 ) 8039 ; s. v. kravchenko , w. e. mason , g. e. bowker , j. e. furneaux , v. m. pudalov , and m. diorio , phys . rev .
b * 51 * ( 1995 ) 7038 e. cuevas , phys . rev .
* 83 * ( 1999 ) 140 ; we note that for the short range interaction the majority of states is localized @xcite so that contrary to the claim of the paper @xmath145 for large @xmath28 at any @xmath146 . | we study the effect of electron - electron interaction on a two dimensional ( 2d ) disordered lattice . for the case of two electrons the analytical estimates
are presented showing a transition from localized to delocalized states in a way similar to the anderson transition in 3d .
the localized phase corresponds to large values of the parameter @xmath0 , which is determined by the ratio of the coulomb and fermi energies .
the numerical investigations of the spectral statistics @xmath1 in a system with up to @xmath2 spin polarized electrons show the transition from the poisson to the wigner - dyson distribution at a total electron energy independent of @xmath3 . the relation to experiments on the metal - insulator transition in 2d
is also discussed . |
codes are the first structured codes ( as opposed to random codes ) that provably achieve the symmetric capacity of binary - input memoryless channels ( bmcs ) @xcite .
this capacity - achieving code family is based on a technique called channel polarization . given a bmc @xmath0 , after performing the channel transform , which consists of the channel combining and splitting operations , over a set of independent copies of @xmath0 , a second set of synthesized channels is obtained . as the transformation size
, i.e. , the number of channel uses participated in the transform , goes to infinity , some of the resulting channels tend to be completely noised , and the others tend to be noise - free , where the fraction of the channels approaches the symmetric capacity of @xmath0 . by transmitting free bits over the noiseless channels and sending fixed bits over the others , polar coding with a very large code length
@xmath1 can achieve the symmetric capacity under a successive cancellation ( sc ) decoder with both encoding and decoding complexity @xmath2 . to construct a polar code , the capacities ( or equivalently , reliabilities ) of the polarized channels
can be estimated efficiently by calculating bhattacharyya parameters for binary - input erasure channels ( becs ) @xcite .
but for channels other than becs , computationally expensive solutions based on density evolution ( de ) @xcite and other modified methods are required to calculate the channel reliabilities @xcite , @xcite .
although polar codes are asymptotically capacity achieving , the performance under the sc decoding is unsatisfying in the practical cases with finite - length blocks .
several improved sc decoding schemes have been proposed to improve the finite - length performance of polar codes .
the successive cancellation list ( scl ) decoding @xcite , @xcite and successive cancellation stack ( scs ) @xcite decoding algorithms are introduced to approach the performance of ml decoding with acceptable complexity . by regarding the improved sc decoding algorithms as path search procedures on the code tree , the scl and scs decoding are the `` width - first '' and the `` best - first '' search , respectively . to provide a flexible configuration under the constraint of both the time and space complexities , an decoding algorithm called the successive cancellation hybrid ( sch )
is proposed by combining the principles of scl and scs @xcite . moreover , under these improved sc decoding algorithms , polar codes are found to be capable of achieving the same or even better performance than turbo codes or low - density parity - check codes with the help of cyclic redundancy check ( crc ) codes @xcite @xcite @xcite . therefore , polar codes are believed to be competitive candidates in future communication systems . shortly after the polar code was firstly put forward , the channel polarization phenomenon has been found to be universal in many other signal processing problems , such as multiple access communications @xcite @xcite , source coding @xcite @xcite , information secrecy @xcite @xcite and other settings . to improve the spectral efficiency , a @xmath3-ary polar coded modulation scheme is provided in @xcite . by regarding the dependencies between the bits which are mapped to a single modulation symbol as a kind of channel transformation , the polar coded modulation ( pcm ) scheme is derived under the framework of multilevel coding @xcite .
it is shown in @xcite that this polar coded modulation scheme can outperform the turbo coded modulation scheme in the 3gpp wcdma system @xcite by up to @xmath4db with @xmath5-ary quadrature amplitude modulation ( qam ) over additive white gaussian noise ( awgn ) channel . in this paper , the channel polarization technique is extended to the multiple - input multiple - output ( mimo ) transmission scenario .
similar to the polar coded modulation scheme @xcite , the transmission over the mimo channel is further combined into the channel transform . the mimo transmission , modulation and the conventional binary channel polarization form a three - stage channel polarization procedure . based on this generalized channel polarization , a jointly optimized space - time polar coded modulation ( stpcm ) scheme
is proposed .
the remainder of the paper is organized as follows .
introduces the system model concerned in this paper . provides a three - stage channel transform which can be seen as a joint processing of the conventional binary channel polarization , modulation and mimo transmission ; describes the construction , encoding and decoding of the proposed stpcm scheme . evaluates the performance of the proposed stpcm scheme under the rayleigh fading channel through simulations . finally , concludes the paper .
in this paper , the capital roman letters , e.g. @xmath6 , @xmath7 , are used to denote random variables . the lowercased letter
@xmath8 denotes a realization of @xmath6 . @xmath9 and @xmath10 are the real and image parts of a complex number @xmath8 , respectively .
the modulus of @xmath8 is written as @xmath11 .
the calligraphic characters , such as @xmath12 and @xmath13 , are used to denote sets , and we use @xmath14 to denote the number of elements in @xmath12 .
the cartesian product of @xmath12 and @xmath13 is written as @xmath15 , and @xmath16 stands for the @xmath17-th cartesian power of @xmath12 .
we use notation @xmath18 to denote an @xmath1-dimensional column vector @xmath19 and @xmath20 to denote a subvector @xmath21 of @xmath18 , @xmath22 .
when @xmath23 , @xmath20 is a vector without elements , and this empty vector is denoted by @xmath24 .
we write @xmath25 to denote the subvector of @xmath18 with odd indices ( @xmath26 ; @xmath27 is odd ) .
similarly , we write @xmath28 to denote the subvector of @xmath18 with even indices ( @xmath26 ; @xmath27 is even ) . for example , for @xmath29 , @xmath30 , @xmath31 , and @xmath32 .
further , given an index set @xmath33 , let @xmath34 denote the subvector of @xmath18 , which consists of @xmath35s with @xmath36 .
the matrices are denoted by bold letters , e.g. , @xmath37 .
the notations @xmath38 and @xmath39 stand for the transpose and conjugate transpose of @xmath37 , respectively .
the element in the @xmath40-th row and the @xmath41-th column of matrix @xmath37 is written as @xmath42 .
the @xmath41-th column of matrix @xmath37 is written as @xmath43 ; the @xmath40-th row of @xmath37 is written as @xmath44 , i.e. , the @xmath40-th column of @xmath38 .
furthermore , we write @xmath45 to denote the kronecker product of two matrices @xmath46 and @xmath47 , and @xmath48 to denote the @xmath17-th kronecker power of @xmath46 . throughout this paper
, @xmath49 means `` logarithm to base 2 '' , and @xmath50 stands for the natural logarithm . a block diagram of space - time coded modulation is depicted in .
the @xmath51 information bits are coded and modulated into a series of @xmath3-ary symbols , and then transmitted to the receiver through a mimo system with and antennas within @xmath1 time slots . at the transmitter , a sequence @xmath52 of @xmath51-length information bits , where @xmath53 , is fed into a binary channel encoder with code rate @xmath54 .
the encoded sequence @xmath55 is interleaved into an other binary sequence @xmath56 . after a modulation
, the @xmath57 bits are mapped into @xmath58 complex symbols @xmath59 .
these symbols are then partitioned into @xmath60 streams with @xmath1 symbols in each stream and respectively transmitted over @xmath60 antennas .
the transmitted symbols are represent by a @xmath61 matrix @xmath37 , where the rows and columns are corresponding to the transmitting antennas and time slots , respectively . in this paper , only qam is considered , and the average transmitting power of the transmitted symbols are normalized to one , i.e. , @xmath62=1 $ ] . at the receiver , @xmath63 antennas are configured .
thus , the mimo channel at the @xmath64-th time slot can be described by a @xmath65 matrix @xmath66 with @xmath67 .
the @xmath64-th column of @xmath68 , i.e. the received signals at the @xmath64-th time slot is @xmath69 where @xmath70 is an @xmath71 additive noise matrix , the elements of which are i.i.d .
complex circular gaussian random variables with mean zero and variance @xmath72 , i.e. @xmath73 . in this paper , the channels between all the transmit / receive antenna pairs are assumed to be independent memoryless discrete - time normalized rayleigh fast and uncorrelated fading channels , i.e. , for any time slot @xmath64 , the channel confident @xmath74 of @xmath66 satisfies @xmath75 .
we assume that an ideal channel estimation ( instant values of @xmath76 and @xmath77 ) is available at the receiver .
furthermore , due to the channel - aware property of polar coding , a precise knowledge of the noise variance @xmath77 is assumed to be available at the transmitter which can be usually obtained from a feedback link . after receiving @xmath68 , a series of signal processes , i.e. , mimo detection , demodulation , de - interleaving and channel decoding , are used to retrieve the information bits @xmath78 .
these generalized `` decoding '' operations can be done in either a separately concatenated manner or a jointly combined manner .
in this section , after a brief review of the existing works , the channel polarization is extended to the mimo transmission case . under the multilevel coding framework ,
a three - stage channel transform is derived , which is the basis of the proposed stpcm scheme . in the initial work of arikan @xcite
, a mapping @xmath79 is used to denote a bmc channel , where @xmath12 and @xmath13 are the input and output alphabets , respectively . since the channel input is binary , @xmath80 .
the channel transition probabilities are @xmath81 , @xmath82 and @xmath83 . after channel combining and splitting operations on @xmath1 independent uses of @xmath0
, we obtain @xmath1 successive uses of synthesized binary input channels @xmath84 , @xmath85 with transition probabilities @xmath86 where @xmath87 where @xmath88 the matrix @xmath89 , in which @xmath90 is the @xmath91 bit - reversal permutation matrix and @xmath92\ ] ] after this channel transform , part of the resulting channels @xmath93 with @xmath36 becomes better than the original channel @xmath0 , i.e. @xmath94 , where function @xmath95 is the symmetric capacity ( the maximum mutual information between the channel inputs and outputs under uniform input distribution ) ; while the others with @xmath96 become worse , where @xmath97 is the complementary set of @xmath33 .
arikan proved it in @xcite that when @xmath1 goes to infinity , @xmath98 with @xmath36 , @xmath99 with @xmath96 , and @xmath100 .
when dealing with the polar coded modulation problem @xcite , the channel becomes @xmath101 , where @xmath12 is the @xmath3-ary input alphabet , @xmath102 , and @xmath103 is the modulation order .
every @xmath104 bits @xmath105 are modulated into a single modulation symbol @xmath82 under a specific one - to - one mapping called constellation labeling @xmath106 thus , the channel can be equivalently written as with transition probabilities @xmath107 where @xmath108 is the inverse mapping of @xmath109 . by regarding the modulation process as a special kind of channel transform , @xmath104 synthesized bmcs @xmath110 can be obtained , where @xmath111 , with transition probabilities @xmath112 after that , a conventional binary - input channel polarization transform @xmath47 is performed on each of the resulting bmcs @xmath113 .
finally , a series of polarized bmcs @xmath114 is obtained , where @xmath85 and @xmath115 . in the @xmath116 mimo transmission scenario ,
the channel becomes @xmath117 , where @xmath12 and @xmath13 are the alphabet at each transmit or receive antenna , respectively . in order to simultaneously transmit @xmath60 streams
, we assume the number of the receive antennas is no less than that of the transmit antennas , i.e. @xmath118 , and the channel matrix @xmath119 is full rank . in each time slot , after transmitting the @xmath60 symbols of @xmath120 via the @xmath60 transmit antennas respectively , the received signal is @xmath121 where @xmath122 with @xmath123 .
suppose the data stream are detected in a successive cancellation manner , @xmath60 correlated channels @xmath124 are obtained , where @xmath125 and the transition function @xmath126 similar to the pcm in @xcite , a three - stage channel transform is depicted in fig.[fig_polarization ] . after performing the first stage channel transform which is induced by the mimo transmission in ( [ equ_mimo_polarization ] ) , each of the resulting @xmath3-ary input channels @xmath127 is transformed into a set of binary input channels @xmath128 with @xmath129 , @xmath130 then , by respectively performing binary channel polarization transform on @xmath1 uses of @xmath131 , totally @xmath57 bmcs @xmath132 are obtained , @xmath133 where , , , , and the set @xmath134 the one - to - one mapping from @xmath135 to @xmath37 is jointly determined by the binary channel polarization ( [ equ_binary_polarization ] ) , modulation process ( [ equ_constellation_labeling ] ) and the serial - to - parallel processing . after substituting ( [ equ_wkj ] ) into ( [ equ_wkji_tmp ] ) , a consistent representation of @xmath136 is obtained @xmath137 where @xmath138
when construct a practical stpcm scheme , similar to the conventional polar coding in @xcite and the pcm in @xcite , the most reliable @xmath51 channels of @xmath139 are selected for carrying the information bits @xmath140 . in the existing works where only awgn channels are considered , the channel reliabilities can be evaluated efficiently by using gaussian approximation ( ga ) of de @xcite .
however , the channel model considered in this paper is rayleigh fast fading channel , and no existing practical solution is available under this scenario .
therefore , we first propose a pcm scheme over the rayleigh fast fading channel , where the fading channel is approximated by an awgn channel with identical capacity .
after that , a stpcm scheme is derived based on the three - stage channel polarization discussed in the previous section . in this subsection ,
the channel transform of the mimo channel under a qr decomposition is proposed . since the channel transform in ( [ equ_mimo_polarization ] )
implicates a detecting order of the data streams , the transmitter and receiver should have an agreement on the specific mimo detection solution . in this paper , qr - decomposition
is applied to for each channel coefficient matrix @xmath119 at the receiver , @xmath141 where @xmath142 is an @xmath143 unitary matrix , and @xmath144 is an @xmath65 upper triangular matrix with @xmath145 for any @xmath146 and @xmath147 for any @xmath148 , where @xmath149 and @xmath150 .
the received signal in ( [ equ_channel ] ) after qr - decomposition detection is @xmath151 where the elements in @xmath152 is still i.i.d gaussian distributed , @xmath153 for any @xmath149 . after expanding the matrix operations in ( [ equ_ch_after_detec1 ] ) , we have @xmath154 the transmitted streams @xmath155 can be detected in a decreasing order in the antenna index @xmath27 , i.e.
, the stream from the @xmath60-th transmit antenna is first detected , then the @xmath156-th , @xmath157 , finally the @xmath158st .
under such a successive cancellation detection , when dealing with @xmath159 , the term @xmath160 in ( [ equ_ch_after_detec ] ) can be dropped .
therefore , the @xmath161 are equivalent to be transmitted over a fading channel with gains @xmath162 .
thus the resulting polarized channels \{@xmath127 } in ( [ equ_mimo_polarization ] ) ( redefined in a reversed order of @xmath27 ) are written as @xmath163 at the transmitter , the detecting order of the transmit streams and the noise variance @xmath77 are assumed to be notified through a feedback link .
however , the @xmath162 , or equivalently @xmath66 , is time - varying , and the instantaneous values of the channel coefficients are unavailable at the transmitter when the channel is assumed to be fast fading .
this is quite different from the conventional polar coding schemes where the precise knowledge of the channel state is known at both the transmitter and receiver .
the reliabilities of @xmath164 can not be precisely evaluated by the existing solutions , so the pcm scheme in @xcite also can not be applied directly on each @xmath127 . in the following part of this subsection
, we propose to construct pcm schemes by approximating the fading channels using a set of awgn channels which have the same capacities with the originals , and it will be used in the construction of the proposed stpcm scheme in section [ subsec_multilevel_stpcm ] . the @xmath3-ary qam with constellation @xmath12 is equivalent to two independent @xmath165-ary pulse amplitude modulations ( pam ) with constellations @xmath166 and @xmath167 , respectively . without loss of generality , we assume the real and image parts of the qam constellation are identical , i.e. , @xmath168 . the symmetric capacity of an awgn channel under @xmath3-ary qam with noise variance @xmath77 is @xmath169 where @xmath170 under the channel model described in , the channel coefficients @xmath171 are i.i.d normalized circular gaussian distributed .
according to ( * ? ? ?
* theorem 3.3 ) , twice the square of the elements @xmath172 in @xmath144 are scaled @xmath173-distributed with @xmath174 degrees of freedom , i.e. , @xmath175 for @xmath176 , where the probability density function ( pdf ) of @xmath177 for a given value @xmath178 and @xmath179 degrees of freedom is @xmath180 where @xmath181 is the gamma function @xmath182 thus , the ergodic capacity of a @xmath116 mimo channel @xmath0 under @xmath3-ary qam with noise variance @xmath77 can be calculated as @xmath183 where @xmath184 is the ergodic capacity of @xmath127 which is calculated as @xmath185 with @xmath186 . after approximating the fading channel @xmath127 using an awgn channel @xmath187 with @xmath188 , where @xmath189 the code construction and performance evaluation
is then performed over each of the equivalent awgn channels @xmath190 in the same way as that in the conventional awgn case @xcite . as that will be shown in
, the bounds obtained by ga under this equivalence well match the simulated block error rate ( bler ) curves . in this subsection ,
a stpcm scheme over qam modulated channels base on qr - decomposition is proposed . as for the scheme construction ,
@xmath1 independent uses of the mimo channel @xmath0 in @xmath1 time slots are transformed into a series of binary - input channels in a three - stage channel transform .
when the code length is finite , pcm under set - partitioning ( sp ) labeling is find to achieve the best performance over pam modulated channel @xcite . in this paper ,
an identical sp labeling rule is applied on the real and image parts of the qam constellation , respectively . moreover , to achieve the optimal utilization of the channel independencies , an additional transform is applied between the real - valued channel pairs corresponding to real and image parts of qam symbols . the detailed channel transform adopted by the proposed stpcm scheme is described below . 1 . [ s1 ] after a qr - decomposition of the mimo channel @xmath0 , a set of single - input multiple - output ( simo ) channels
@xmath164 are obtained , with @xmath125 .
2 . [ s2 ] under a @xmath3-ary qam with @xmath191 , the simo channel @xmath127 at the @xmath27-th transmit antenna is further transformed to @xmath104 binary - input channels @xmath128 . without loss of generally , the first half of the channels with @xmath192 correspond to the real parts of the modulation symbols , while the other half correspond to the image parts .
[ s3 ] for each @xmath193 and @xmath194 , @xmath1 uses of @xmath131 are further transformed by an @xmath1-scaled binary channel polarization .
since at the @xmath27-th transmit antenna , the channels corresponding to the real and image parts of the symbols , i.e. @xmath128 and @xmath195 with @xmath196 , are noised by independent real - valued awgns , and the labeling rules of the real and image parts of the qam constellation are set to be identical , one additional step of binary channel polarization transform can be performed between the channel pairs of @xmath128 and @xmath197 .
thus , for each @xmath198 and @xmath199 , the @xmath200 channel uses , which consist of @xmath1 uses of @xmath131 and @xmath201 respectively , are then transformed by an @xmath202 matrix @xmath47 into @xmath203 , where @xmath204 .
note that when the system is working over fading channels , the shared channel gain @xmath205 introduces correlationship between the channel pairs @xmath131 and @xmath201 for any specific time slot @xmath64 .
therefore , an additional interleaving process is required on the inputs of @xmath128 and @xmath197 to make the channel uses participated in the binary channel transform @xmath47 independent . following section [ sub_ch_trans_with_qr ] ,
when constructing the stpcm scheme , each of the channels @xmath206 obtained in s1 ) are approximated using an awgn channel , the reliabilities of the corresponding @xmath203 can then be evaluated by de or ga in the same way as that in the conventional pcm scheme . finally , the @xmath51 most reliable channels among @xmath203 are selected for carry the information bits @xmath140 , and the others are fixed to frozen bits @xmath207 , while the universal set @xmath208 , where , , } .
gives an illustration of the mapping from the information bits @xmath52 to the transmitted signal matrix @xmath37 . at the receiver , since both the mimo channel and the modulation procedures are combined into the channel polarization transform , the mimo detection , demodulation and binary polar decoding can be jointly processed . a successive cancellation ( sc )
algorithm can be used to decode this generalized polar code . given the received signal @xmath68 , the information bit @xmath209 is decoded with indices @xmath210 taking values from @xmath158 to @xmath57 under an sc manner @xmath211 where @xmath212 where the indices of the channel @xmath136 are calculated as , , @xmath213 , and @xmath214 is the floor function .
equation ( [ equ_sc ] ) is essentially the same with the conventional sc decoding rules in @xcite .
therefore , the improved sc decoding algorithms , scl @xcite@xcite and crc - aided scl ( cascl ) @xcite@xcite@xcite can also be used to decode the proposed stpcm scheme which can yield much better performance than sc .
the proposed stpcm scheme over @xmath116 mimo channel with @xmath3-ary qam is equivalent to a set of @xmath215 binary polar codes with code length @xmath200 . for binary polar coding ,
the encoding and decoding complexities are both @xmath216 .
compared to the component polar encodings , the complexities brought by the modulation and interleaving operations are negligible , so the encoding complexity of stpcm is @xmath217 . to decode the stpcm ,
a qr - decomposition is applied on each @xmath66 for @xmath67 .
since it is assumed that @xmath118 and the complexity of qr - decomposition operation is @xmath218 @xcite , the decoding complexity of stpcm is @xmath219 .
[ fig_sim_config ] in this section , the bler performance of stpcm scheme is analyzed via simulations .
the number of available mimo channel uses @xmath1 is in @xmath220 , and the code rate is in @xmath221 . for comparison ,
the the performance of a bit - interleaved turbo coded modulation ( bitcm ) scheme over channel is also provided . the turbo encoder and rate - matching algorithm used in 3gpp wcdma system @xcite
are adopted .
the punctured codeword is fed into a randomized interleaver , and then modulated and distributed to the transmit antennas .
the constellation rule of bitcm scheme is gray mapping . at the receiver , the mmse detection @xcite , demodulation , deinterleaving and log - map decoding ( with maximum @xmath222 iterations ) @xcite
are executed sequentially .
this transmission model is essentially applied in many practical wireless communication systems @xcite @xcite .
different from the separate signal processing in the above bitcm scheme , the propose stpcm can be regarded as a joint processing of the channel coding , modulation and the mimo transmission .
gives a block diagram of stpcm transmission .
the sc decoding algorithms in ( [ equ_sc ] ) is used to decode the stpcm . as stated in @xcite ,
when decoding the bitcm , @xmath223 metric updating operations in the trellis representations of the component convolutional codes is required : @xmath222 iterations over @xmath224 constituent @xmath222-state decoders with @xmath225 metric updates per trellis node , and the interleaver size @xmath226 .
when decoding the stpcm under sc , the number of required metric updates in trellis of the @xmath227 component polar codes is @xmath228 .
therefore , under the simulated configurations of @xmath229 and @xmath1 , the bitcm consumes about @xmath230 times of the computational complexity taken by stpcm under sc decoding . gives the bler performance of stpcm under sc decoding .
the simulated stpcm schemes are constructed after evaluating the reliabilities of the polarized channels using ga algorithm .
similar to the case of the conventional binary polar codes @xcite , the bler performances under the sc decoding of the proposed stpcm and the corresponding estimated values obtained by ga are well matched . to improve the performance of stpcm
, the cascl decoding is applied .
the searching width of the cascl decoder is set to @xmath231 , the complexity of which is upper bounded by @xmath231 times of sc .
taking the complexity - reducing implementation methods @xcite@xcite@xcite into consideration , the cascl decoding of stpcm under this configuration is with the comparable complexity with the log - map decoding of bitcm .
as shown in , the improvement in bler performance of stpcm under the cascl is about @xmath224db or more against that under the sc decoding .
furthermore , the stpcm scheme can even outperform the bitcm scheme by no less than @xmath232db .
the performance gain of stpcm scheme under cascl decoding against the bitcm scheme remains when higher modulation order and more antennas are assigned .
the performance over rayleigh fast and uncorrelated mimo channels of @xmath5qam with up to @xmath233 antennas are shown in .
as the figure shows , the performance gains are around @xmath232 to @xmath234db .
a comprehensive comparison of stpcm and bitcm schemes under different configurations is provided in .
the cascl decoding is used to decode the stpcm scheme . in the subfigures ,
the minimum required snrs to achieve bler @xmath235 are plotted , and the ergodic capacities ( [ equ_fadingcap ] ) of the corresponding transmission schemes are also provided . among all the simulated cases ,
the stpcm scheme can achieve a performance again of @xmath236 to @xmath234db against the bitcm scheme .
particularly , a significant @xmath237db gain is observed in the case of @xmath238 mimo with @xmath5qam ( ) , when the bitcm scheme suffers with a severe _ error floor _ effect around the bler @xmath239 .
a space - time coded modulation scheme based on polar codes is proposed following the multilevel principle , which can be seen as a joint optimization of the binary polar coding , modulation and multiple - input multiple - output ( mimo ) transmission .
similar to the multilevel approach of polar coded modulation , the mimo transmission process is combined into the channel transform . based on the generalized channel polarization , a space - time polar coded modulation ( stpcm ) scheme with qr - decomposition
is proposed for the @xmath3-ary modulated mimo channel .
in addition , a practical solution of polar code construction over the fading channels is also provided , where the fading channels are approximated by an awgn channel which shares the same capacity with the original .
the proposed stpcm scheme is simulated over uncorrelated mimo rayleigh fast fading channels . compared with the widely used bit - interleaved turbo coded modulation ( bitcm ) approach ,
the proposed stpcm scheme achieves a performance gain of @xmath236 to @xmath234db in all the simulated cases .
this work was supported in part by the national natural science foundation of china ( no . 61171099 ) , the national science and technology major project of china ( no . 2012zx03003 - 007 ) and qualcomm corporation .
e. arikan , `` channel polarization : a method for constructing capacity - achieving codes for symmetric binary - input memoryless channels , '' _ ieee trans .
inf . theory _ ,
55 , no . 7 , pp . 3051 - 3073 , jul . | the polar codes are proven to be capacity - achieving and are shown to have equivalent or even better finite - length performance than the turbo / ldpc codes under some improved decoding algorithms over the additive white gaussian noise ( awgn ) channels .
polar coding is based on the so - called channel polarization phenomenon induced by a transform over the underlying binary - input channel .
the channel polarization is found to be universal in many signal processing problems and has been applied to the coded modulation schemes . in this paper , the channel polarization
is further extended to the multiple antenna transmission following a multilevel coding principle .
the multiple - input multile - output ( mimo ) channel under quadrature amplitude modulation ( qam ) are transformed into a series of synthesized binary - input channels under a three - stage channel transform .
based on this generalized channel polarization , the proposed space - time polar coded modulation ( stpcm ) scheme allows a joint optimization of the binary polar coding , modulation and mimo transmission . in addition , a practical solution of polar code construction over the fading channels is also provided , where the fading channels are approximated by an awgn channel which shares the same capacity with the original .
the simulations over the mimo channel with uncorrelated rayleigh fast fading show that the proposed stpcm scheme can outperform the bit - interleaved turbo coded scheme in all the simulated cases , where the latter is adopted in many existing communication systems .
polar codes , space - time coding , coded modulation , multilevel coding , joint optimization . |
investigation of the decay properties of nuclear resonant states is an important part of the nuclear structure studies . usually in such investigations only two - body decay channels
are allowed for .
three - body decays are much more informative than the two - body ones , but the full description of a three - body decay is a very complicated problem .
thus it is very important to derive simple but accurate approximate approaches for the analysis of the experimental data and for making reliable theoretical predictions .
one of the approaches of such kind is the approach based on the democratic decay approximation .
a three - body continuum spectrum wave function in the asymptotic region is generally a superposition of the components describing two- and three - body decay channels @xmath9 .
the democratic decay approximation accounts for the three - body channel only ; two - body channels associated with the appearance of bound two - body subsystems and non - democratic subdivision of the system of the type ( 2 + 1 ) , are not allowed for in the approximation . therefore , this approximation is valid only for the study of a three - body system in the case when all two - body decay channels are closed and the only open channel is a three - body channel . generally speaking , if the three - body channel is the only open one , than the system can have two types of asymptotics @xmath9 .
one of the asymptotics corresponds to the situation when one of the particles is scattered by another one and the third particle is a spectator . from the general physical point of view , this asymptotics is supposed to be of little importance for a nuclear system excited in some nuclear reaction and decaying via a three - body channel . the alternative type of the asymptotics is a superposition of ingoing and outgoing six - dimensional spherical waves ( or , equivalently , six - dimensional plane wave and ingoing or outgoing six - dimensional spherical wave ) and corresponds to the situation when the decaying system emits ( or / and absorbs ) three particles from one and the same point in space .
the version of the scattering theory that allows only for three - body asymptotics of this type is called true three - body scattering theory , or , alternatively , @xmath10 scattering theory ; the approximation based on the allowance for @xmath11 scattering only in the examination of nuclear reactions is called democratic decay approximation . nuclear structure studies within the framework of democratic decay approximation have been started in 70th by prof .
r.i.jibuti with collaborators .
the reviews of the results obtained by the tbilisi group can be found in refs .
recently a considerable progress has been made in both theoretical and experimental studies of democratic decays .
below we shall briefly review some of the recent results .
in particular , we shall discuss the results of experimental studies of democratic decays of @xmath1 nuclei via the three - body channel @xmath13 , theoretical investigations of monopole excitations of @xmath3c nucleus in the cluster model @xmath14 and @xmath5li structure studies in the cluster model @xmath15li@xmath16 .
detailed experimental study of the three - body decay of @xmath17 , @xmath18 and @xmath19 states in @xmath20he , @xmath20li and @xmath20be nuclei has been performed by the group from kurchatov institute ( see @xmath21 and references therein ) .
the democratic decay concept has been for the first time experimentally tested in this investigation , and the democratic nature of the decay of these states via the only open channel @xmath22 has been unambiguously demonstrated .
the wave function in the democratic decay approximation has the following asymptotics : @xmath23\chi_{sm_s}\ , \label{1}\ ] ] where the first term is a six - dimensional plane wave while the second one is an ingoing six - dimensional spherical wave ; @xmath24 is the @xmath10 scattering amplitude ; @xmath25 and @xmath26 are jacobi coordinates , and @xmath27 and @xmath28 are associated canonically conjugated momenta ; hyper - radius @xmath29 , and @xmath30 ; @xmath31 is the energy measured from the three - body breakup threshold ; @xmath32 and @xmath33 are the total spin of the system and its projection while @xmath34 is the spin part of the wave function .
it is natural to expand the decay amplitude @xmath24 in hyperspherical harmonics @xmath35 and to couple the three - body orbital momentum @xmath36 and the spin @xmath32 into the total angular momentum @xmath2 . as a result , for the decay amplitude @xmath37 of a resonant state with the total spin @xmath2 and its projection @xmath38 we have the following expression @xmath39 : @xmath40_{jm}\ .
\label{2}\ ] ] in eq .
( [ 2 ] ) , @xmath41 is hypermomentum , the multi - index @xmath42 plays the role of a channel index for the @xmath43 scattering , and the multi - index @xmath44 stands for the set of all additional quantum numbers labelling different channels with the same quantum numbers @xmath45 .
it is convenient to use the yamanouchi symbol @xmath46 and the young scheme [ @xmath47 for @xmath44 if there are three identical particles in the final state ; otherwise orbital momenta @xmath48 and @xmath49 corresponding to jacobi coordinates @xmath25 and @xmath26 are usually used for @xmath44 .
( [ 2 ] ) is a generalization of the usual angular momentum decomposition of a two - body scattering amplitude and can be formally applied to any three - body decay .
note , that hypermomentum @xmath41 is not an integral of motion , and due to the two - body interactions components labelled by different values of @xmath41 mix together in the wave function , so , partial amplitudes @xmath50 with various values of @xmath41 are present in the expansion ( [ 2 ] ) . for a two - body decay channel
the expansion ( [ 2 ] ) is non - convergent ; in the case when two - body subsystems of the decaying three - body system have sharp resonances , the convergence can appear to be very slow .
if two - body subsystems of the three - body system have neither bound nor sharp resonant states , than the @xmath43 scattering amplitude @xmath37 converges rapidly and only few terms play an important role in eq .
( [ 2 ] ) .
thus only in the case of democratic decay of such system it is possible to extract partial amplitudes @xmath50 from the experimental data , i.e. , to solve the problem that is somewhat similar to the problem of the spin assignment to a resonant state by the analysis of two - body scattering data .
the hyperspherical harmonics @xmath51 are functions of five angles in the formal six - dimensional momentum space \{@xmath52}. one of the angles , @xmath53 , rules the energy distribution between two jacobi coordinates . therefore , in order to evaluate partial amplitudes @xmath54 , one should analyze energy spectra of emitted particles . the analysis of ref .
@xmath55 of the three - body decay @xmath20be@xmath56 will serve us as an example .
@xmath20be levels have been populated in the reaction @xmath20li(@xmath57he,@xmath57h)@xmath20be at @xmath58 mev .
@xmath57h has been detected in the coincidence with @xmath59 and/or @xmath60 from the decay @xmath20be @xmath61 .
the detection of @xmath57h of appropriate energy made it possible to select the levels of interest . taken from ref .
@xmath55 the spectrum of @xmath59-particles from the decay of the @xmath20be @xmath62 state in the beryllium rest frame is depicted on fig . 1
. the peculiar form of the spectrum can not be reproduced in the model of independent escape of the decay products ( phase volume calculations ) , nor can it be reproduced in the model of sequential decay via the resonant states of @xmath60-@xmath60 or @xmath59-@xmath60 subsystems . at the same time , as is seen from fig . 1 ,
the data are perfectly reproduced using expansion ( [ 2 ] ) and allowing only for the terms with @xmath63 and 2 .
nevertheless , the set of the partial amplitudes \{@xmath54 } can not be uniquely determined from the analysis of the @xmath59-spectrum because various sets of \{@xmath54 } give the fit to the data of the same quality .
one should involve in the analysis the energy spectrum of protons from the same reaction .
the predictions for the proton spectrum obtained using various sets of \{@xmath54 } fitted to the @xmath59-spectrum , differ significantly . from the fit to both @xmath59- and @xmath60-spectra , the relative weights of ( @xmath64)% ; ( @xmath65)% ; and ( @xmath66)% for the components @xmath63 ; @xmath67 ; and @xmath68 , respectively , have been obtained in ref .
@xmath55 .
the set of partial amplitudes \{@xmath54 } can be used for the predictions of the kinematically complete experiment based on the triple - coincidence @xmath57h @xmath69 measurements . as is seen from fig . 2
, the set of \{@xmath54 } fitted to @xmath59- and p - spectra , perfectly matches the triple - coincidence data .
note , that the kinematical conditions of the triple - coincidence experiment @xmath55 have been chosen in such a way , that the contribution of the @xmath70 decay has been strongly suppressed .
therefore , the fine agreement between the theoretical curve and the experimental data on fig .
2 confirms the good quality of the determination of the relative weights of @xmath63 and @xmath67 components only .
in the previous section we have discussed the application of the democratic decay approximation to the analysis of experimental data .
now we shall discuss the use of the true three - body scattering theory in calculations .
it is well known that the giant monopole resonance generates a well - pronounced peak of the monopole excitation strength in heavy nuclei exhausting an essential fraction of the energy - weighted sum rule ( ewsr ) . in light nuclei ,
the giant monopole resonance is very weak and exhausts , e.g. , in @xmath3c , only @xmath71% of ewsr @xmath72 .
however , the strong peak of @xmath73 strength exhausting @xmath74% of ewsr has been predicted in various microscopic calculations ( see , e.g. , ref .
@xmath75 ) .
it has been shown in ref .
@xmath15 that this discrepancy should be attributed to the continuum spectrum effects , i.e. , to the decay properties of the giant monopole resonance .
the 3@xmath59 cluster model has been used in ref .
the ground state wave function has been obtained by diagonalization of the three - body hamiltonian using the harmonic oscillator basis . for
@xmath62 continuum spectrum states only the decay channel @xmath3c @xmath76 has been allowed for .
this channel is the single open channel for low - lying @xmath62 states and is supposed to dominate for @xmath62 states with excitation energies of up to @xmath77 mev .
continuum spectrum wave functions have been calculated by means of the oscillator representation of scattering theory ( @xmath2-matrix method ) @xmath78 which had been generalized on the case of the true few - body scatting in ref .
@xmath5 . in the @xmath2-matrix approach , the continuum spectrum wave function @xmath79 for a given three - body decay channel @xmath80 , @xmath36 , and @xmath81 in the channel index @xmath82 and in the indexes of the functions used .
in the case of spinless particles ( e.g. , @xmath59-particles ) , @xmath83 , @xmath36 , and @xmath81 are integrals of motion , therefore , all functions are labelled with the same values of @xmath32 , @xmath36 , and @xmath81 and channels @xmath82 and @xmath84 with @xmath85 and/or @xmath86 are not coupled . ]
is expressed as an infinite series expansion in six - dimensional harmonic oscillator functions @xmath87 , @xmath88 the basic approximation of the @xmath2-matrix approach is in neglecting potential energy matrix elements @xmath89 with a large number of oscillator quanta @xmath90 and/or @xmath91 ; the infinite kinetic energy matrix @xmath92 is not truncated .
the exact wave functions of the hamiltonian with the infinite kinetic energy matrix and truncated potential energy matrix for any positive energy @xmath31 can be found by algebraic methods . to do it
, one should first diagonalize the truncated hamiltonian matrix , i.e. , calculate its eigenvalues @xmath93 and corresponding eigenvectors @xmath94 . next the matrix elements @xmath95 of the matrix @xmath96 , @xmath97 are calculated .
the @xmath43 scattering @xmath32-matrix can be easily computed by the expression @xmath98 where matrix elements of matrices @xmath99 and @xmath100 , @xmath101 @xmath102 and @xmath103 are the truncation boundaries of the hamiltonian matrix in the channels @xmath104 and @xmath105 , respectively ; the kinetic energy matrix elements @xmath106 ; the eigenvectors @xmath107 and @xmath108 of the infinite kinetic energy matrix t can be easily calculated by analytical expressions or by recurrent relations ( see ref .
@xmath5 for details ) . finally , for the wave function with outgoing six - dimensional spherical wave in the channel @xmath109 and ingoing six - dimensional spherical waves in all allowed channels @xmath110 , the expansion coefficients @xmath111 [ see eq .
( [ 3 ] ) ] in the asymptotic region @xmath112 are calculated by the expression @xmath113 \ , \label{7}\ ] ] while for calculation of @xmath114 in the interaction region @xmath115 the expression @xmath116 is used .
ali - bodmer @xmath117 potential @xmath3 has been used in ref .
@xmath15 ; coulomb interaction has been neglected .
the truncation boundary @xmath118 has been used in the calculations , i.e. , the truncated hamiltonian matrix accounted for all @xmath119 oscillator states up to @xmath120 .
the convergence for both the ground and continuum @xmath62 states is achieved at @xmath121 provided that @xmath122 mev . in the asymptotic region @xmath112 all channels @xmath110 with @xmath123 have been allowed for .
however , in the interaction region @xmath115 we have accounted for all oscillator states up to @xmath120 , thus the hyperspherical harmonics with @xmath124 have been allowed for in the interaction region .
the inclusion of components with @xmath125 in the channel space have not produced changes in the results .
the results of the calculations of ref .
@xmath15 are presented on fig .
the solid line in fig .
3 represents the @xmath31-dependence of the @xmath73 transition strength @xmath126 to the states above the @xmath3c @xmath127 threshold . here , @xmath128 is the matrix element of the @xmath73 transition from the ground state to the states with energy @xmath31 .
the dash - dotted curve displays the sum of the contributions to ewsr from all states with excitation energies less than @xmath31 , i.e. , the function @xmath129 where @xmath130 the monopole sum rule @xmath131 , and @xmath132 is the rms radius of the @xmath3c nucleus .
the dashed step - like line is the plot of the function @xmath133 calculated disregarding the effect of the @xmath119 continuum . in this case , the height of each step is equal to the contribution to the ewsr from the corresponding quasistationary state obtained by the diagonalization of the truncated hamiltonian matrix .
it is seen that the most essential contributions to ewsr are connected with the quasistationary levels with excitation energies 6.11 , 10.39 , 13.60 and 22.30 mev ( 31% , 29% , 13% and 13% , respectively ) .
all the remaining levels with excitation energies up to 35 mev exhaust about 7% of the ewsr . the @xmath3c binding energy in calculations of ref .
@xmath15 has been found to be @xmath134 mev ( the experimental value is 7.27 mev ) .
the first excited @xmath62 state with the excitation energy @xmath135 mev belongs to the discrete spectrum , so , the continuum does not affect its properties .
the experimental value for @xmath136 is 7.66 mev and this state belongs to the @xmath119 continuum .
however , its width is very small ( @xmath137 ev ) and its properties are rather close to those of the discrete spectrum states .
the allowance for the continuum slightly changes the properties of the 10.39 mev @xmath62 state .
namely , the @xmath138 maximum shifts down to @xmath139 mev , the state gets some width and its contribution to ewsr decreases to 26%
. however , in the 1325 mev excitation region the effect of the continuum appears to be essential .
the levels in this excitation energy region get considerable widths and merge each other , forming very broad ( of @xmath140 mev in width ) plateau of a peculiar shape , i.e. , the strength function @xmath138 is almost constant between 13 and 25 mev except for a bump of the width of 23 mev at an excitation energy of about 20 mev .
the bump should be related to the 20 mev monopole resonance observed in refs .
33% of the ewsr is exhausted in the 1226 mev excitation energy region , with 16% of the ewsr being exhausted in the 1822 mev interval .
the approach , as is seen from fig .
3 , reproduces quite correctly the positions of the strong @xmath73 absorption regions .
the broad and smooth pedestal of the 20 mev @xmath73 excitation strength maximum can be easily misinterpreted in experimental studies as a background .
the shape of the background that has been subtracted by eyrich et al @xmath72 in their study of the 20 mev resonance , is similar to the shape of the pedestal on fig .
if the experimental background is a real manifestation of the smooth pedestal of the @xmath73 strength , than it should be higher for small scattering angles . from fig . 3 of ref .
@xmath72 it is seen , that the background is really enhanced at small angles .
so , the experimental spectra of ref .
@xmath72 is in qualitative agreement with theoretical results of ref.@xmath15 .
the further investigation of the nature of the background is needed .
the results presented on fig . 3
give the explanation why the giant monopole resonance is nearly not seen in @xmath3c .
there are two physical reasons .
first one is the strong fragmentation of the @xmath73 strength in @xmath3c . as a result ,
more than 50% of the ewsr is exhausted by the 7.65 mev and 10 mev resonances , and not too much of the @xmath73 strength is left for the giant monopole resonance .
the second reason is the continuum spectrum effect . due to the democratic decay
, the giant monopole resonance gets a large width and merge with other @xmath62states . as a result
, the @xmath73 strength becomes very smooth , and the giant monopole resonance manifests itself only as a non - prominent bump on the smooth pedestal . about 15% of the ewsr is exhausted in the vicinity of 20 mev bump , but if the pedestal is treated as a background , than the contribution of the resonance to the ewsr is severely underestimated .
light neutron - rich nuclei like @xmath20he , @xmath142he , @xmath5li , @xmath143be , etc . , are extensively studied now ( see recent review papers @xmath144 and references therein ) .
these nuclei are often called exotic nuclei ; they are also often called barromean nuclei , because they have a well - pronounced cluster structure @xmath145 and after removing a neutron from any of these nuclei , the remaining system @xmath146 becomes unbound .
two excess loosely - bound neutrons have a wide spatial distribution with long tail reaching far out and form the so - called neutron halo .
the neutron halo manifests itself in a number of unusual properties of exotic nuclei , in particular , in a large value of the rms radius of the nucleus , in the existence of the soft dipole mode , etc .
it is natural to make use of the democratic decay approximation in the study of barromean nuclei . in ref .
@xmath147 , @xmath5li has been studied in the cluster model @xmath15li @xmath8 .
it has been shown in ref . @xmath147 that the democratic decay approximation within the framework of the @xmath2-matrix approach described in the previous section , is advantageous for the investigations not only of the continuum states , but of the ground state as well . in a usual variational calculation with the oscillator basis , it is impossible to describe the slowly - dying asymptotic tail of the halo neutron wave functions . in order to describe the neutron halo properties of the @xmath5li nucleus , one is pushed to account for the oscillator states with a very large number of oscillator quanta . in the @xmath2-matrix approach
, the infinite number of the oscillator states can be allowed for . for a bound state , the expansion coefficients @xmath148 [ see eq .
( [ 3 ] ) ] in the asymptotic region @xmath112 are of the form@xmath5 : @xmath149 the bound state asymptotic normalization constants , @xmath150 , are calculated by handling numerically the set of linear homogeneous equations@xmath5 @xmath151 eq .
( [ 12 ] ) can be solved if only @xmath152 using eq .
( [ 13 ] ) the energy of a bound state can be easily calculated .
this method of calculation of the bound state energies is equivalent to the location of the @xmath32-matrix pole@xmath5 corresponding to the bound state , as is easily seen from eq .
( [ 5 ] ) .
the expansion coefficients @xmath153 for the inner region @xmath154 are obtained using eq .
( [ 8 ] ) .
finally , the wave function should be normalized .
the calculations@xmath147 have been performed using the shallow gaussian @xmath155@xmath15li and @xmath155@xmath155 potentials of ref .
only the states with the spin of the halo - neutron pair @xmath83 have been considered .
the three - body angular momentum of the ground state of @xmath5li have been supposed to be equal to @xmath157 .
the parameter @xmath158 mev has been used in the calculations .
this value corresponds approximately to the minimum of the ground state energy @xmath159 .
the results for the @xmath5li ground state for various values of the truncation parameter @xmath160 are presented in the table 1 .
the variational ground state energies , @xmath161 , obtained by the pure diagonalization of the truncated hamiltonian matrix are listed in the second column , while the @xmath2-matrix results , @xmath159 , which are the solutions of the eq .
( [ 13 ] ) are listed in the third column .
.the effect of the lanczos smoothing on the ground state energy of @xmath5li ( see text for details ) . [ cols="^,^,^,^,^ " , ] + the values of the @xmath5li r.m.s .
radius , @xmath162 and @xmath163 , obtained by the pure diagonalization of the truncated hamiltonian matrix and with the allowance for the asymptotic tail of the wave function , respectively , are presented in the 4-th and the 5-th columns of the table .
it is seen , that by locating the @xmath32-matrix pole using eq .
( [ 13 ] ) that is equivalent to the allowance for the long asymptotic tail of the wave function , the convergence of the ground state is essentially improved .
the results presented in the table 1 have been obtained using lanczos smoothing of the three - body potential energy matrix@xmath164 .
the smoothing improves the convergency of calculations of the scattering characteristics .
nevertheless , the smoothing causes the underestimation of the binding energy of a three - body system .
the @xmath5li ground state energy obtained in the @xmath2-matrix calculations without the use of the smoothing , @xmath165 , shows a staggering as @xmath160 increases - matrix calculation is not a variational one , so , the ground state energy may behave non - monotonically as @xmath160 increases . ]
( see table 2 ) .
@xmath165 is seen to converge to the value of @xmath166 mev .
the smoothing results in a usual - type monotonic decrease of @xmath167 as @xmath160 increases , but @xmath168 for all values of @xmath160 . at the same time , the ground state wave functions obtained with the smoothing and without it , differ very slightly .
so , below we shall discuss only the results of calculations with the smoothed potential energy matrix .
the quality of the convergence of the wave function in the @xmath2-matrix approach can be demonstrated by the plots of momentum distributions .
figure 4 presents the transverse momentum distribution of the cluster @xmath7li in @xmath5li that is currently supposed ( the so - called serber reaction mechanism ) to be proportional to the @xmath7li transverse momentum distribution @xmath169 in the fragmentation of high - energy @xmath5li beams on target nuclei .
the experimental data for @xmath5li fragmentation@xmath170 are also presented on the figure .
the results of @xmath2-matrix calculations with @xmath171 and @xmath172 are so close to each other that their plots on fig.4 coalesce and can not be distinguished . it is seen that in calculation of the ground state , the allowance for the wave function asymptotics is very important for a weekly - bound system like @xmath5li .
the terms of expansion ( [ 3 ] ) with the number of oscillator quanta @xmath173 that can not be obtained in the usual oscillator - basis variational calculations , play an essential role in the formation of the transverse momentum distribution , r.m.s . radius , etc .
the convergence of @xmath174 , transverse momentum distribution and other properties of the wave function ( e.g. , of the weights of its components ) in the full @xmath2-matrix calculation is very good .
nevertheless , it is seen that the r.m.s .
radius converges to a value that is somewhat larger than the experimental one , and the calculated transverse momentum distribution is narrower than the experimental one .
these shortcomings can be overcome by the adjustment of @xmath155@xmath15li potential .
we have not aimed to fit the potential to the @xmath5li properties , we have just take its parameters from ref .
@xmath156 . the differences between the calculated and experimental values of r.m.s .
radius arise from the fact that in ref .
@xmath156 the potential has been fitted to the @xmath5li r.m.s . radius and the binding energy in the _ variational calculation _ that is unable to reproduce these quantities with high accuracy .
the reduced probability of the @xmath4 transition obtained in the cluster model , @xmath175 , is displayed on the figure 5 . in the @xmath2-matrix approach ,
zero - width pure variational peaks of @xmath176 gain finite widths and shift to lower energies .
it is seen that in the calculation of @xmath175 it is important to allow for the asymptotics not only for continuum states , but for the ground state as well . due to the wide spatial distribution of the halo neutrons density , large distances
contribute essentially to the @xmath4 strength in the low - energy region . to account for this contribution and to get a convergence , one should allow for the ground and continuum state wave function components with very large values of the total number of oscillator quanta @xmath177
is at a distance of @xmath178108 fm from the origin . ] in calculations of @xmath175 .
the large - distance @xmath4 transitions enhance the peak of the cluster @xmath175 and shift it to a lower energy .
obviously , this peak should be associated with the soft dipole mode .
thus , the neutron halo manifests itself in the appearance of the soft dipole mode that arises from the shift and enhancement of the @xmath175 peak .
this effect appears to be very important in the calculation of the @xmath5li electromagnetic dissociation cross section ( see below ) .
large distances contribute essentially to @xmath179 and @xmath73 transitions , too .
the shift and the enhancement of @xmath180 and of @xmath181 are even more pronounced .
nevertheless , @xmath179 and @xmath73 transitions do not play an important role in the electromagnetic dissociation , and we shall not discuss them here .
figure 6 shows the comparison of the results of our calculations@xmath147 of the cluster @xmath175 with the parametrization of experimental data of ref .
the agreement is reasonable .
the form of the @xmath175 peak is well reproduced , the discrepancy in the position of the @xmath175 maximum is supposed to be eliminated by the adjustment of the potentials .
the results of the @xmath175 calculations of refs .
@xmath183 are also depicted on fig .
the three - body cluster calculations with the allowance for democratic decay channels of ref .
@xmath184 performed using usual coordinate - space hyperspherical harmonics approach , nicely reproduce the energy of the soft dipole mode .
nevertheless , the form of the peak in our calculations is reproduced better .
note , that the authors of ref .
@xmath184 used another set of the potential parameters .
the calculations of ref .
@xmath185 with the allowance for two - body decay channels only , failed to reproduce the form of the @xmath175 peak and underestimate the energy of the soft dipole mode ..
the soft dipole mode exhausts about 90% of the cluster ewsr , @xmath186 , associated with the excitation of the cluster degrees of freedom only .
nevertheless , it is easy to show that @xmath187 where @xmath188 is the total ewsr accounting for excitations of all nucleons .
so , the contribution from the soft dipole mode to the total ewsr is relatively small . in the vicinity of the @xmath175 maximum at the excitation energy @xmath189 mev only @xmath1908% of the total ewsr is exhausted . nevertheless , the account for the soft dipole mode results in an essential increase of the electromagnetic dissociation cross section of 0.8 gev / nucleon @xmath5li beams on pb and cu targets .
the wide space distribution of the halo neutrons density is also well - manifested in the electromagnetic dissociation of @xmath5li beam .
the contribution of the large - distance @xmath191-transitions to the cross section is about 50% .
the calculations have been performed by the equivalent photon method .
the only parameter of the method is the minimal value of the impact parameter @xmath192 .
we use for @xmath192 the values of 9.0 fm for pb and 6.8 fm for cu target nuclei , respectively .
these quantities are the sums of the @xmath5li and target nucleus charge radii . with these values of @xmath192
we obtain for the electromagnetic dissociation cross sections the values of 0.966 barn for the pb target and 0.132 barn for the cu target ; the corresponding experimental values are @xmath193 barn and [email protected] barn , respectively@xmath195 . @xmath73 and @xmath179 transitions give only 1.2% contribution in the cross sections .
* acknowledgments * + i am thankful to yu.a.lurie , t.ya.mikhelashvili and yu.f.smirnov for a fruitful collaborate work some results of which have been discussed above .
valuable discussions with j.m.bang , b.v.danilin , r.i.jibuti , i.j.thompson and j.s.vaagen are acknowledged .
this paper has been presented as a talk at the conference perspectives of nuclear physics in the late nineties ( hanoi , march 1994 ) . the support of my participation the conference by the international science foundation is acknowledged .
* s.p.merkuriev and l.d.faddeev , _ quantum scattering theory for systems of few bodies _ ( nauka publishes , moscow , 1985 ) . *
r.i.jibuti , _ elem .
part . and
atom . nucl . _
* 14 * ( 1983 ) 741 . *
r.i.jibuti and n.b .
krupennikova , _ hyperspherical harmonics method in quantum mechanics of few bodies _
( metsniereba , tbilisi , 1984 ) . * o.v.bochkarev et al , _ nucl .
phys . _ * a505 * ( 1989 ) 215 . * o.v.bochkarev et al , _ yad .
j. nucl.phys.)_ * 55 * ( 1992 ) 1729 .
* b.v.danilin et al , _ yad .
j. nucl.phys.)_ * 46 * ( 1987 ) 427 . *
w.eyrich et al , _ phys .
_ * c36 * ( 1987 ) 416 . *
v.v.burov , v.n.dostovalov , m.kaschiev and k.v.shitikova , _ j. phys .
_ * g7 * ( 1981 ) 131 .
* t.ya.mikhelashvili , a.m.shirokov and yu.f.smirnov , _ j.phys .
_ * g16 * ( 1990 ) 1241 . *
h.a.yamani and l.fishman , _ j. math .
* 16 * ( 1975 ) 410 ; g.f.filippov , _ yad .
. phys . ) _ * 33*(1981 ) 928 ; yu.i.nechaev and yu.f.smirnov , _ yad .
( sov.j .
nucl . phys . )
_ * 35 * ( 1982 ) 1385 .
* yu.f.smirnov and a.m.shirokov , _ preprint _ * itp-88 - 47p * ( kiev , 1988 ) ; a.m.shirokov , yu.f.smirnov and s.a.zaytsev , _ to be published_. * s.ali and a.bodmer , _ nucl .
_ * 80 * ( 1966 ) 99 .
* d.lebrun et al , _ phys .
_ * 97b * ( 1980 ) 358 .
* c.a.bertulani , f.l.canto and m.h.hussein , _ phys . rep . _
* 226 * ( 1993 ) 281 .
* zhukov m.v .
et al , _ phys . rep . _
* 231 * ( 1993 ) 151 .
* yu.a.lurie , a.m.shirokov and yu.f.smirnov , _ nucl - th_/*9407011 * ; _ to be published . _ * l.johansen , a.s.jensen and p.g.hansen , _ phys . lett . _ * 244b * ( 1990 ) 357 . * j.rvai , m.sotona and j.zofka , _ j. phys . _ * g11 * ( 1985 ) 745 ; j.mares , _ czech .
j. phys .
_ * b37 * ( 1987 ) 665 . * t.kobayashi et al , _ phys .
* 60 * ( 1988 ) 2599 . *
d.sackett et al ,
_ preprint _ * msucl * ( michigan state university , 1993 ) .
* b.v.danilin , m.v.zhukov , j.s.vaagen , i.j.thompson and j.m.bang , _ preprint _ * nordita
93/34 n * ( copenhagen , 1994 ) ; _ private communication . _
* g.f.bertsch and h.esbensen , _ ann . of phys . _ * 209 * ( 1991 ) 327 ; h.esbensen and g.f.bertsch , _ nucl .
_ * a542 * ( 1992 ) 310 . * t.kobayashi et al , _ phys
_ * b232 * ( 1989 ) 51 . | results of some recent investigations of democratic three - body decays of nuclear systems are discussed . in particular ,
we consider experimental studies of three - body ( @xmath0)-decay of @xmath1 nuclei and @xmath2-matrix calculations of monopole excitations in @xmath3c and @xmath4 transitions in @xmath5li in the three - body - continuum cluster models @xmath6 and @xmath7li @xmath8 , respectively . 8.5 in |
the construction of a lorentzian wormhole @xcite or time machine @xcite requires violation of the null energy condition ( nec ) : @xmath0 where @xmath1 is the matter energy - momentum tensor and @xmath2 is any null - vector ( @xmath3 ) . for example , in a wormhole the throat geometry requires that converging null geodesics evolve into diverging null geodesics ( imagine the trajectories followed by light rays traversing the wormhole ) .
the raychaudhuri equation for the expansion @xmath4 of a hypersurface orthogonal null congruence is @xcite @xmath5 where @xmath6 is an affine parameter , @xmath7 the tangent vector field , and @xmath8 the shear .
the first two terms on the right - hand side of eq .
( [ theta ] ) are negative , and after using the einstein equation to relate @xmath9 and @xmath10 , we see that @xmath4 can increase only if the nec is violated .
in this paper we consider the exotic matter necessary to stabilize a traversable wormhole throat ( which may or may not be a time machine ) .
our analysis focuses on constraining its properties , using the fact that it violates the nec and some additional assumptions about the throat spacetime .
it is important to note that both the magnitude of @xmath11 and its violation of the nec must be large in order to construct a useful wormhole .
roughly , a wormhole with throat diameter of order @xmath12 requires energy - momentum densities of order @xmath13 , where @xmath14 is the planck length .
taking @xmath12 to be of order meters requires energy - momentum densities greater than @xmath15 @xcite .
similarly , it is easy to show using eq .
( [ theta ] ) that the time required to travel through the wormhole is determined by the degree of violation of the nec .
this travel time is @xmath16 , where @xmath17 . for travel time of order 1 year
, this requires @xmath18 .
thus , the exotic matter under consideration must exhibit large energy density and substantial violation of the nec .
we define two types of wormholes ( or time machines constructed using wormholes ) , henceforth referred to as `` devices '' : those with semi - classical spacetimes ( type a ) and those with strongly fluctuating spacetimes ( type b ) .
clearly , type a devices are preferable to type b as they can by definition be controlled more precisely , and perhaps pose less risk to users in their operation .
unfortunately , we will show that all type a devices are unstable to small perturbations , so time travel or travel via wormhole is possible only via an intrinsically quantum device with a `` fuzzy '' spacetime .
essentially , type a devices require exotic matter which behaves classically ( quantum effects are negligible in its dynamics ) , and classical matter which violates the nec is unstable @xcite . by a semi - classical state @xmath19 we mean one whose probability distribution is peaked about its central value in a particular basis ,
defined by some operator @xmath20 . in this case , @xmath21 up to small corrections ( assuming @xmath22 ) , and we say that @xmath19 is semi - classical with respect to the operator @xmath20 .
this condition of semi - classicality is weaker than the usual one , which requires that eq .
( [ z ] ) holds for _ all _ operators simultaneously .
a state might , according to our definition , be semi - classical with respect to @xmath20 but not with respect to some other operator @xmath23 ( i.e. , it might exhibit large fluctuations in measurements of @xmath23 ) .
for example , @xmath19 might be an eigenstate of @xmath20 , while @xmath24 \neq 0 $ ] ( e.g. , let @xmath20 be the momentum operator and @xmath23 the position operator ) . a central issue addressed in this paper is whether the time evolution of a device which gives rise to a semi - classical , non - zero @xmath11 is well - approximated by the classical equations of motion , or equivalently , whether it is simultaneously semi - classical with respect to both the field operators @xmath25 and their conjugate momenta @xmath26 .
if the time evolution of the exotic matter in the device is well - approximated by the classical equations of motion , we can apply a previous result @xcite , showing that in models built from gauge fields , scalars and fermions , any classical solution is unstable to small perturbations if it violates the nec . specifically , in ref .
@xcite it was shown that the kinetic term of the effective hamiltonian governing small fluctuations about a classical solution must exhibit a negative eigenvalue if the nec is violated .
therefore , there exist infinitesimal perturbations involving local spatial gradients which lower the energy of the solution .
in this section we discuss spacetime fluctuations @xcite , and argue that type a devices are necessarily semi - classical with respect to the matter energy - momentum tensor @xmath11 .
it is useful to distinguish between `` active '' and `` passive '' metric fluctuations .
active fluctuations are those involving quantum effects due to the graviton itself ; these are negligible as long as the spacetime curvature is small compared to the planck scale .
we focus on passive fluctuations which are induced on the metric through the einstein equations @xmath27 by fluctuations in @xmath11 . clearly , in order for the spacetime to be semi - classical
, the matter state must be semi - classical with respect to @xmath11 .
because of ultraviolet divergences , the property of semi - classicality is scale - dependent . on sufficiently small
length scales the passive fluctuations become arbitrarily large .
operator expectations as in eq .
( [ z ] ) require regularization ; for example , if @xmath28 , it involves products of fields at the same point in spacetime .
we assume a regulator such as point - splitting or smearing of the fields , controlled by a length scale @xmath29 .
short - distance contributions to operator averages are then of order @xmath30 to the appropriate power , leading , e.g. , to fluctuations @xmath31 ( no summation over indices ) . to investigate the effects of these fluctuations , we take @xmath29 to be some characteristic length scale associated with the spacetime or relevant experimental probes , such as test particles .
minkowski spacetime is type a on length scales @xmath32 ( where @xmath14 is the planck length ) , even though @xmath33 , so that @xmath34 for any @xmath29 .
this is because the effect of these fluctuations on the metric @xmath35 is suppressed by @xmath14 , as can be seen from the einstein equations : @xmath36 .
since @xmath37 is constructed of objects with two derivatives of @xmath35 , we see that deviations of @xmath35 from flat space on length scales @xmath29 are of order @xmath38 .
borgman and ford have studied the effect of vacuum energy fluctuations on null geodesics in minkowski space , and found them to be undetectable @xcite . now consider a device whose spacetime is nontrivial i.e. , not flat space .
if we take @xmath29 to be some characteristic length scale associated with the device spacetime , the regulated short - distance fluctuations are small compared to the central values of the energy momentum tensor : @xmath39 . in fact , from the einstein equations we expect @xmath40 . in other words , the energy and momentum densities required to warp spacetime on length scale @xmath29 are much larger than @xmath41 .
( as mentioned , in the wormholes studied by morris and thorne in ref .
@xcite , for a throat size of order meters the required energy - momentum densities are of order @xmath15 , provided perhaps by a cosmic string or similar object . )
then , the question of whether the spacetime is semi - classical depends not on the short - distance singularities in the operator products , but rather on whether the wavefunction of @xmath19 is highly peaked about its central value on length scales of @xmath29 or larger . _ the classicality of the spacetime is determined by the classicality of the energy - momentum tensor in the matter state @xmath19 . _ as an example , consider the casimir vacuum @xmath42 , which describes the interior of a cavity of size @xmath12 .
the nec is violated in the casimir state , which has negative ( renormalized ) energy density : @xmath43 . however , since the fluctuations @xmath44 are of order @xmath45 , they are larger than the central value even if we choose the largest possible length scale @xmath46 ; the casimir state never leads to a semi - classical energy - momentum tensor .
further , the casimir state is problematic for construction of an interesting device , since the curvatures induced are only of order @xmath47 .
spacetime within the cavity deviates from flat space only very slightly , and in an intrinsically fuzzy ( highly fluctuating ) manner .
in this section we show that semi - classicality with respect to @xmath48 implies semi - classicality with respect to the fields @xmath25 and their conjugate momenta @xmath26 .
hence the wavefunction is localized in phase space , and the quantum evolution is closely approximated by the classical evolution .
the proof proceeds in two steps .
first , we note that at least two components of @xmath11 do not commute with each other .
then , we derive a relation between the uncertainties in two such non - commuting operators and arbitrary functions of them . in the case of interest
, the original pair is two non - commuting components of @xmath11 and the functions map these operators into @xmath25 and @xmath26 .
the relation shows that small uncertainties in @xmath10 imply small uncertainties in the phase space variables .
first , using the translation operator , it is easy to see that @xmath49 $ ] is related to @xmath50 by a unitary transformation . therefore , if for some @xmath51 , @xmath52 for all @xmath53 , then @xmath54 identically .
however , if that were the case we would have , in particular , @xmath55=0 $ ] for all @xmath51 and @xmath53 . integrating over @xmath53 , we find that @xmath56 does not depend on time , which is possible only if @xmath57 .
thus , in any non - trivial model , for any @xmath51 there always exists an @xmath53 for which @xmath58 .
in fact , @xmath59 has support only for @xmath51 near @xmath53 , since it is proportional to the delta function @xmath60 or its spatial derivatives ; this follows from the canonical commutation relation @xmath61 = i \delta ( x - x')$ ] .
thus , there are at least two non - commuting components of the energy - momentum tensor . next ,
suppose we have a pair of non - commuting operators @xmath62 and @xmath63 , and consider a new operator which is an arbitrary function of them , @xmath64 .
the only limitation imposed on the function @xmath65 is that it can be expanded in a power series of its arguments .
( note that if @xmath62 and @xmath63 commute we can not necessarily express an arbitrary @xmath66 as a local function of them .
for example , let @xmath67 , @xmath68 and @xmath69 . ) it should be clear that if both @xmath62 and @xmath63 are semi - classical , than @xmath66 is semi - classical too .
indeed , let us expand the operator @xmath66 around its classical value @xmath70 : @xmath71 + { { \cal o}}(\delta^3),\end{aligned}\ ] ] where @xmath72 is of order @xmath73 or @xmath74 .
the fluctuation @xmath75 then is @xmath76 for semi - classical @xmath62 and @xmath63 , the third term on the right hand side of eq .
( [ delta.f ] ) is as small as the first two terms .
thus @xmath66 is semi - classical if @xmath62 and @xmath63 are .
if there is another operator @xmath77 , we also expand it around its classical value @xmath78 and find the commutator @xmath79=(\bar{{{\cal f}}}_{\bar{a}}\bar{{{\cal g}}}_{\bar{b } } -\bar{{{\cal f}}}_{\bar{b}}\bar{{{\cal g}}}_{\bar{a}})[a , b ] + { { \cal o}}(\delta^3).\label{jacobian}\end{aligned}\ ] ] alternatively , in the semi - classical limit the commutator @xmath80 $ ] turns into the poisson bracket @xmath81 .
since the brackets @xmath82 and @xmath83 are related via the jacobian @xmath84 , we arrive at eq .
( [ jacobian ] ) again .
fluctuations of two semi - classical operators @xmath62 and @xmath63 satisfy the relation @xmath85 \rangle$ ] . since @xmath66 and @xmath86 are semi - classical too , writing the similar relation for them and combining the result with eq .
( [ jacobian ] ) , we arrive at @xmath87 and @xmath88 spaces , up to possible degeneracy.,width=321 ] to complete our proof , we identify the operators @xmath89 with two non - commuting components of @xmath10 , and the operators @xmath90 with @xmath91 .
a state which has semi - classical energy - momentum tensor then must be localized in phase space , and therefore evolve according to the classical equations of motion . note that if either of the commutators in eq .
( [ jacobian ] ) vanish , the jacobian @xmath92 would be either zero or divergent , and localization in @xmath93 space would not imply localization in @xmath88 space , or vice versa .
it is possible that degeneracies ( either discrete or continuous ) lead to multiple small regions in the @xmath88 space which are mapped onto the same small region of the @xmath93 space ( see fig .
( [ figure ] ) ) .
in other words , there may be inequivalent configurations in phase space with the same energy - momentum tensor .
in such a case , the preceding equations should be modified by including the sums ( or integrals , in the continuous case ) over these regions .
for example , if two states @xmath94 and @xmath95 with distinct values of @xmath96 or @xmath97 lead to the same @xmath98 , we can obtain a semi - classical spacetime from a superposition of the two . nevertheless , _
each _ component of the superposition is localized ( behaves semi - classically ) and will separately exhibit an instability .
classical systems which violate the nec are unstable @xcite in a particularly violent way : they can lower their energy by increasing local spatial field gradients .
therefore , the exotic matter used to stabilize a wormhole throat must be quantum mechanical in nature ; in other words , quantum effects must play an important role in its dynamics and time evolution .
however , it is undesirable for a device to have a strongly fluctuating ( non - semi - classical ) spacetime .
such a device would presumably behave unpredictably and might transport its payload to an undesirable time or place .
we have shown that semi - classicality of the spacetime is a strong enough condition to imply phase space localization of the wavefunction of the stabilizing matter .
this means the time evolution of type a devices will be semi - classical and well - approximated by the classical equations of motion .
such devices are unstable in any region where the nec is violated .
wormholes can not be both predictable and stable .
m. s. morris and k. s. thorne , am .
j. phys .
* 56 * , 395 ( 1988 ) ; j. l. friedman , k. schleich and d. m. witt , phys .
71 * , 1486 ( 1993 ) [ erratum - ibid .
* 75 * , 1872 ( 1995 ) ] ; d. hochberg and m. visser , phys . rev .
d * 56 * , 4745 ( 1997 ) ; m. visser and d. hochberg , arxiv : gr - qc/9710001 .
for a review , see m. visser , _ lorentzian wormholes : from einstein to hawking _ , aip , woodbury usa , 1995 . c. i. kuo and l. h. ford , phys .
d * 47 * , 4510 ( 1993 ) ; n. g. phillips and b. l. hu , phys .
d * 55 * , 6123 ( 1997 ) ; b. l. hu and n. g. phillips , int .
j. theor .
phys . * 39 * , 1817 ( 2000 ) ; n. g. phillips and b. l. hu , phys .
d * 62 * , 084017 ( 2000 ) ; l. h. ford and c. h. wu , int . j. theor . phys .
* 42 * , 15 ( 2003 ) . | we show that lorentzian ( traversable ) wormholes with semi - classical spacetimes are unstable .
semi - classicality of the energy - momentum tensor of the exotic matter used to stabilize the wormhole implies localization of its wavefunction in phase space , leading to evolution according to the classical equations of motion .
previous results related to violation of the nec then require that the matter is unstable to small perturbations . |
the numerical solutions of the bk equation do not display exact geometric scaling , although they approach a solution showing such scaling at asymptotic @xmath36 . assuming the solutions to be of the form ( [ ngamma ] ) , where scaling violations are encoded in the `` anomalous dimension '' @xmath24
, therefore leads to the conclusion that @xmath1 is not a function of @xmath61 exclusively .
in particular , it is never simply a constant , not even at the saturation scale ( @xmath28 ) . at
asymptotically large rapidities , @xmath24 reaches a limiting function @xmath54 .
this function is universal for a large range of initial conditions @xcite . at the saturation scale
, @xmath57 equals approximately @xmath2 , which is considerably smaller than the corresponding values in the phenomenological models @xcite . for small values of @xmath70
the limiting function seems to reach @xmath29 @xcite , in accordance with the traveling wave results of refs .
@xcite . performing the replacement of @xmath71
does allow one to find a solution for which @xmath72 is kept fixed .
the behavior of @xmath40 is then for small rapidities qualitatively similar to the parameterization in @xcite .
however , the usually considered choice @xmath73 yields some unwanted features , i.e. @xmath74 being negative in a region above the saturation scale and the absence of solutions below the saturation scale , although the ansatz was not intended for that region . keeping @xmath72 fixed at a smaller value , e.g. at @xmath75 , seems more suitable @xcite , but it remains to be investigated whether such a choice allows for a good fit of all relevant dis , @xmath21-@xmath22 and @xmath76-@xmath76 data
. it would be interesting to consider modifications of phenomenological models for the dipole scattering amplitude that are compatible with both the bk equation and the data .
given the fact that the bk evolution does not respect geometric scaling around @xmath48 , phenomenological parameterizations that reflect this feature would seem a natural choice .
fortunately , the lhc and a possible future electron - ion collider will provide data over a larger range of momenta and rapidities , so that one can expect to test the evolution properties of the models more accurately .
99 slides : + ` http://indico.cern.ch/contributiondisplay.py?contribid=69&sessionid=15&confid=9499 ` a. h. mueller , nucl .
b * 335 * , 115 ( 1990 ) .
i. balitsky , nucl .
b * 463 * , 99 ( 1996 ) ; y. v. kovchegov , phys .
d * 60 * , 034008 ( 1999 ) .
k. golec - biernat and m. wsthoff , phys .
d * 59 * , 014017 ( 1999 ) .
a. m. stasto , k. golec - biernat and j. kwiecinski , phys .
lett . * 86 * , 596 ( 2001 ) .
a. h. mueller and d. n. triantafyllopoulos , nucl .
b * 640 * , 331 ( 2002 ) .
d. kharzeev , y.v .
kovchegov and k. tuchin , phys . lett .
b * 599 * , 23 ( 2004 ) .
a. dumitru , a. hayashigaki and j. jalilian - marian , nucl .
phys . a * 765 * , 464 ( 2006 ) .
a. dumitru , a. hayashigaki and j. jalilian - marian , nucl .
a * 770 * , 57 ( 2006 ) .
e. iancu , k. itakura and s. munier , phys .
b * 590 * , 199 ( 2004 ) .
s. munier and r. peschanski , phys .
lett . * 91 * , 232001 ( 2003 ) ; phys .
d * 69 * , 034008 ( 2004 ) .
r. enberg , `` bksolver : numerical solution of the balitsky - kovchegov nonlinear integro - differential equation '' , available at url : http://www.isv.uu.se/@xmath77enberg / bk/. | deep inelastic scattering at small @xmath0 can be described very effectively using saturation inspired dipole models .
we investigate whether such models are compatible with the numerical solutions of the balitsky - kovchegov ( bk ) equation which is expected to describe the nonlinear evolution in @xmath0 of the dipole cross section .
we find that the bk equation yields results that are qualitatively different from those of phenomenological studies .
geometric scaling is recovered only towards asymptotic rapidities . in this
limit the value of the anomalous dimension @xmath1 at the saturation scale approaches approximately @xmath2 , in contrast to the value @xmath3 commonly used in the models . at small @xmath0 , deep inelastic scattering ( dis ) can be described as the scattering of a color dipole , which the photon fluctuates into , off the proton @xcite .
the linear bfkl equation , which describes the dipole - proton interaction in terms of gluon ladders , predicts an exponential growth of the corresponding cross section as @xmath4 increases , potentially violating unitarity .
hence , interactions between bfkl gluon ladders may become important , which leads to a nonlinear evolution approximately described by the balitsky - kovchegov ( bk ) equation @xcite . as a consequence of the nonlinearity , the dipole cross section saturates with decreasing @xmath0 , thereby offering a resolution to the unitarity problem .
the inclusive hera data at low @xmath0 ( @xmath5 ) could be described well by a dipole cross section of the form @xmath6 , where the scattering amplitude @xmath7 is given by @xcite @xmath8 , \label{ngbw}\ ] ] @xmath9 denotes the transverse size of the dipole , @xmath10 and the @xmath0-dependence of the saturation scale is given by @xmath11 , where @xmath12 and @xmath13 .
the scattering amplitude depends on @xmath0 and @xmath14 through the combination @xmath15 only , which is known as geometric scaling and leads to the prediction that the structure function @xmath16 is a function of @xmath17 only .
this prediction was checked in a model independent way @xcite and holds widely even though the gbw model ( [ ngbw ] ) is not applicable at large @xmath18 .
it should be mentioned that the leading order bk equation leads to a faster evolution in @xmath0 @xcite ( @xmath19 where @xmath20 ) than the experimental data seem to favor ( @xmath13 ) .
this discrepancy can be reduced by introducing a running coupling constant .
hadron production in @xmath21-@xmath22 collisions can also be described by saturation inspired dipole models @xcite . however , these data seem to require geometric scaling violation . the dipole scattering amplitude modified in this respect
is given by @xcite @xmath23\ , .
\label{ngamma}\ ] ] the exponent @xmath24 is usually referred to as the `` anomalous dimension '' , although the connection of @xmath25 with the gluon distribution may not be clear for all cases considered below . following partly @xcite , in @xcite a few requirements were used to determine a parameterization of @xmath24 .
firstly , one assumes that @xmath1 approaches 1 in the limit @xmath26 .
therefore the `` dglap '' limit @xmath27 is recovered for all @xmath0 .
secondly , at the saturation scale , @xmath28 , @xmath24 should be constant to ensure geometric scaling in this region .
this constant @xmath29 is chosen to be @xmath30 .
the value of @xmath29 is motivated by a saddle point analysis of a solution of the bfkl equation with saturation boundary conditions @xcite and also shows up in the traveling wave approach @xcite .
thirdly , if one writes @xmath31 , then @xmath32 should decrease as @xmath33 for @xmath34 at fixed @xmath35 .
this ensures that geometric scaling is asymptotically recovered .
furthermore , the parameters were adjusted in such a way that geometric scaling holds approximately for finite @xmath36 in a growing region between @xmath37 and roughly @xmath38 .
note that the parameterization in @xcite is intended to describe @xmath39 in this so - called extended geometric scaling region only . to simplify the procedure of the required fourier transformation of @xmath25 ( [ ngamma ] )
, @xmath1 was replaced in @xcite by @xmath40 where @xmath41 is the transverse momentum of the scattered parton that will fragment into the final state hadron .
we want to check whether these requirements for @xmath1 are compatible with the nonlinear evolution of the dipole scattering amplitude @xmath25 .
the bk equation for @xmath25 reads @xcite @xmath42 & -n{\left}(|\vec{x}_t-\vec{y}_t|,x{\right})-n{\left}(|\vec{x}_t-\vec{z}_t|,x{\right } ) n{\left}(|\vec{z}_t-\vec{y}_t|,x{\right } ) \big ] .
\label{bk}\end{aligned}\ ] ] here @xmath43 .
we will not consider the impact parameter dependence of @xmath25 .
the bksolver program @xcite provides a numerical solution of the amplitude @xmath44 in momentum space . in order to use this solution of the bk equation ( [ bk ] ) to constrain @xmath1 ,
one first has to find @xmath39 by fourier transforming to coordinate space : @xmath45 using the ansatz ( [ ngamma ] ) one can extract @xmath1 from the resulting @xmath39 , @xmath46/ \log[r^2\,q_s^2(x)]\ , .
\label{gammar}\ ] ] this equation requires as a separate input the value of @xmath47 , which can be found by equating the right hand sides of eqs .
( [ ngamma ] ) and ( [ fourier ] ) for @xmath28 . combining the resulting values of @xmath48 with eq .
( [ gammar ] ) , we obtain a numerical result for @xmath1 , which is shown in fig .
[ gammar1_fig]a . a ) @xmath1 resulting from the relations ( [ fourier ] ) and ( [ gammar ] ) as a function of @xmath49 and @xmath50 .
b ) @xmath51 as a function of @xmath52 for various rapidities @xmath50.,title="fig:",width=260 ] a ) @xmath1 resulting from the relations ( [ fourier ] ) and ( [ gammar ] ) as a function of @xmath49 and @xmath50 .
b ) @xmath51 as a function of @xmath52 for various rapidities @xmath50.,title="fig:",width=260 ] the resulting @xmath1 has the following features : 1 . for @xmath53 , @xmath1 asymptotically approaches 1 . 2 . at the saturation scale , @xmath1 is not a constant .
3 . for decreasing @xmath0
, @xmath1 approaches a limiting curve , @xmath54 , indicated in fig .
[ gammar1_fig ] by @xmath55 .
hence , after a longer evolution one indeed recovers geometric scaling .
the fact that for small distances @xmath24 asymptotically approaches 1 is understandable from the bk equation , since in this limit it reduces to the bfkl equation . in the limit of small distances ,
the solution to the bfkl equation is dominated by either the saddle point or the initial condition , both leading to @xmath56 , since here we use the mv model as the initial condition , see @xcite for details .
it turns out that @xmath24 is clearly not constant , not even at @xmath28 , unlike in @xcite .
however , asymptotically geometric scaling is recovered as @xmath24 approaches @xmath57 .
writing @xmath58 it turns out that , similar to the parameterizations used in @xcite , @xmath59 decreases as @xmath33 for @xmath60 and fixed @xmath61 . at the saturation scale @xmath24
is given in the small-@xmath0 limit by @xmath62 which is significantly below @xmath63 .
this is not in disagreement with theoretical expectations @xcite .
rather it indicates that requiring @xmath24 in eq .
( [ ngamma ] ) to be constant at @xmath48 does not follow from the bk equation . in @xcite ,
@xmath39 ( [ ngamma ] ) was considered with @xmath1 replaced by @xmath40 .
this approximation scheme we will discuss next .
the procedure of extracting @xmath24 becomes quite different when @xmath24 depends on @xmath41 , since the dipole cross section @xmath25 then depends on both @xmath14 and @xmath41 , so that it is not related to @xmath64 by a straightforward inverse fourier transform ( [ fourier ] ) anymore : @xmath65{\right})\ , .
\label{nk2}\ ] ] instead of by using the inverse fourier transform , we will extract @xmath24 by numerically solving eq .
( [ nk2 ] ) , imposing the following condition .
in order to test the ansatz in @xcite , we will fix @xmath51 in such a way that it equals the constant @xmath66 at the saturation scale .
the @xmath0-dependence of @xmath48 is determined by explicitly solving eq .
( [ nk2 ] ) for @xmath67 and @xmath68 .
now we can extract @xmath24 from relation ( [ nk2 ] ) for any given value of @xmath0 and @xmath41 .
fig .
[ gammar1_fig ] shows the results for @xmath51 as a function of @xmath69 above @xmath48 , for a broad range of rapidities . for small rapidities
the resulting @xmath24 looks very similar to the one in @xcite ( cf .
fig . 4 of ref .
@xcite ) .
as one can see , for larger @xmath36 the resulting @xmath24 is not compatible with the parameterization in @xcite anymore ; it first decreases before it rises towards 1 asymptotically . for a discussion of additional important issues like the dependence on initial conditions , the @xmath0-dependence of @xmath48 in our approach and the running coupling case we refer to @xcite . |
the qcd description of exclusive processes has been successfully extended to meson pair production through two - photon collisions @xmath4 , in the kinematical region of large virtuality of one photon and of small center of mass energy @xcite .
the scattering amplitude of the process can be factorized into a perturbative calculable hard process @xmath5 or @xmath6 , and non - perturbative matrix elements describing the transition of the two partons into a hadron pair , which are called generalized distribution amplitudes ( gdas ) @xcite , to emphasize their close connection with the distribution amplitudes introduced many years ago in the description of exclusive hard process @xcite .
the gdas are related by crossing @xcite to the generalized parton distributions ( gpds ) @xcite , which enter the factorization of deeply virtual compton scattering and hard exclusive processes @xcite , and encode information of the quark orbital angular momentum contributing to the nucleon spin @xcite .
a systematic investigation of the pions gdas ( 2@xmath7das ) has been performed in refs .
@xcite , and the corresponding phenomenology at @xmath8 colliders has been developed @xcite in detail .
gdas can also be defined for more complicated systems .
the case of three pions has been studied in @xcite , two baryon production in @xcite ( although in a different kinematical region ) , and the case of two vector mesons ( @xmath9 mesons ) @xmath10 in @xcite , which has been measured first by the l3 collaboration @xcite at lep and the data has been analyzed in @xcite .
gdas appear in a variety of hard scattering processes other than the two - photon process .
studies have also been performed on exotic meson production @xcite , and exclusive heavy mesons decaying to multimesons @xcite , where a large scale is set by the heavy quark mass . in this paper
we consider another process that has not been studied so far , the pion pair production at the @xmath0-pole : @xmath11 . in this process
the mass of the @xmath0 boson provides a natural hard scale .
thus in the kinematical region where the invariant mass of pion pair @xmath12 is small , the process can be separated into a hard subprocess @xmath13 and a non - perturbative two - pion distribution amplitude , analogous to @xmath14 .
we give a leading order calculation for neutral and charged pion pair production at the @xmath0-pole .
using an available model for the 2@xmath7das we find that the cross - section for charged pion pair production is much larger than the case of neutral pion pair , which means that the process is dominated by the isovector channel , that is , the contribution from the @xmath2-odd part of the @xmath3das .
the same one also contributes to the hard exclusive electroproduction of pion pairs @xcite .
our study shows that combining the measurement of charged pair and neutral pair production at the @xmath0-pole can give information of the @xmath2-odd part of the @xmath3das .
the bremsstrahlung process @xmath15 is also considered in our analysis .
the underlying process we study is : @xmath17 the photon emission allows for a small invariance mass to be transferred to the pion pair .
denoting by @xmath18 and @xmath19 the momenta of the @xmath0 boson and final state photon , respectively , we use @xmath20 and @xmath21 to denote the momenta of the two - pion mesons , and define the sum of the two - pion momenta as @xmath22 . the lowest order diagrams of pion pair production at the @xmath0-pole are shown in fig .
[ zpole ] .
we choose the pion pair center of mass frame ( shown in fig .
[ kine]a ) as the reference frame , and define the @xmath0 boson momentum direction as the @xmath23 axis . in this frame we have @xmath24 respectively . here
@xmath25 and @xmath26 are two lightlike vectors which satisfy : @xmath27 , @xmath28 , @xmath12 is the invariant mass of the pion pair and @xmath29 .
the skewness parameter @xmath30 is the momentum fraction of plus momentum carried by @xmath31 with respect to the pion pair : @xmath32 the hadronic tensor of the process is @xmath33 which is expressed as a convolution of a hard coefficient function and a soft correlation function , @xmath34 is the momentum fraction of the quark with respect to the hadronic system . @xmath35 and @xmath36 are the neutral current and electromagnetic current respectively .
the lignt - cone coordinates used here are defined as @xmath37 , @xmath38 . in the above equation
the gauge - link ensuring the gauge invariance of the definition is implied .
the hard coefficient function can be calculated from the subprocess @xmath39 directly : @xmath40\nonumber\\ & = & \frac{ie_qg}{2\sqrt{2}\cos \theta_w}\left [ v^\mu\frac{(z-1)qv{\!\!\!/}-qv^\prime\!{\!\!\!/}}{(1-z)q^2}\gamma^\nu\right.\nonumber\\ & & \left.+\gamma^\nu\frac{zqv{\!\!\!/}+qv^\prime\!{\!\!\!/}}{zq^2}v^\mu\right ] , \end{aligned}\ ] ] where @xmath41 is the weak coupling constant , @xmath42 is the @xmath0-boson - quark vertex , with @xmath18 denoting the quark flavor . the vector and axial - vector coupling to the @xmath0 boson
are given by : @xmath43 here @xmath44 = @xmath45 for @xmath46 , and @xmath47 for @xmath48 and @xmath49 , @xmath50 is the electric charge of quark @xmath18 in units of the electron charge . expanding the above equation we arrive at : @xmath51v^\prime_\alpha\gamma_\beta + 2 i c^q_a\epsilon^{\mu\nu\alpha\beta}v_\alpha\gamma_\beta \nonumber\\&&- 2 i c^q_v \epsilon^{\mu\nu\alpha\beta}v_\alpha\gamma_\beta \gamma_5 + \left [ c^q_a s^{\mu \nu \alpha \beta } \frac{2z-1}{z(1-z)}\right . \nonumber\\ & & \left.\left.- c^q_v \epsilon^{\mu \nu \alpha \beta}\frac{i}{z(1-z)}\right ] v^\prime_\alpha\gamma_\beta \gamma_5\right \},\label{hardcoef}\end{aligned}\ ] ] where we have used the identities : @xmath52 convoluting @xmath53 with the soft matrix element , and keeping the leading twist contribution ( the high - twist contributions are irrelevant at the @xmath0-pole ) , we get the hadronic tensor : @xmath54 the function @xmath55 is the generalized distribution amplitude defined as @xcite : @xmath56 note that because the final hadrons are two pions ( pseudoscalar mesons ) , the third and fourth lines of ( [ hardcoef ] ) vanishes by parity invariance of the strong interactions . before further discussion
it is useful to recall some symmetry properties of @xmath3das .
charge conjugation invariance of the strong interactions implies @xcite : @xmath57 following the notation in @xcite the @xmath2-even and @xmath2-odd parts ( in @xcite they are also called isoscalar and isovector parts respectively ) of @xmath3 da can be defined as @xmath58,\label{cpm}\ ] ] where the superscript @xmath59 represents @xmath60 or @xmath61 .
then @xmath62 the properties of @xmath63 under the interchange @xmath64 and @xmath65 can be easily derived from ( [ cdecomp ] ) @xcite : @xmath66 the @xmath65 exchange means the interchange of the two pions , thus in the case in which the final two pions are @xmath60 we get @xmath67 therefore there is only a @xmath2-even part for @xmath68 .
this is true because the @xmath60 state is a @xmath2-even state .
now we turn back to eq .
( [ hdtensor2 ] ) .
the hard coefficient in the first term inside the brackets is antisymmetric under @xmath69 , while the second one is symmetric under @xmath69 . then by virtue of ( [ cdecomp],[ceven],[codd ] )
, we have @xmath70 the first term in ( [ hdtensor3 ] ) probes the @xmath2-even part of the @xmath3das , which is the same as that in the two - photon process except for the different coupling . here the main difference with the two - photon process , which can only project out the @xmath2-even part of the @xmath3das , is an additional term , the second term in ( [ hdtensor3 ] ) , which can probe the @xmath2-odd part of @xmath3das . the amplitude of @xmath71 can be calculated from @xmath72 where @xmath73 and @xmath74 are the polarization vectors of @xmath0 boson and real photon , respectively . in our frame
these vectors have the form : @xmath75 similar to the two - photon process , in leading twist only the transverse polarization contributes , and the polarizations of the @xmath0 boson and the photon must be the same .
this follows from angular momentum conservation in the collinear approximation .
the pion pair production at the @xmath0-pole can be realized in the @xmath8 annihilation process @xmath76 .
it is interesting to mention that except for the fact that the process occurs at the @xmath0-pole , this process is related by crossing to the virtual compton process @xmath77 , as well as the process @xmath78 . with the polarizations of the final photon summed the amplitude of the process is calculated from @xmath79 here @xmath80 is the @xmath0-boson - lepton vertex , and @xmath81 and @xmath82 have the same form as shown in ( [ cqv ] ) , and ( [ cqa ] ) where @xmath83 for the electron is @xmath47 .
the differential cross - section for the process @xmath84 is expressed as @xmath85 where @xmath86 is the center of mass energy squared of the lepton pair .
we then rearrange the kinematics of the phase space to @xcite @xmath87 in which ( @xmath88 ) is the momentum of pion 1 in the c.m .
frame of the pion pair , and @xmath89 is the angle of the photon in the rest frame of the @xmath0 boson ( that is , the c.m .
frame of @xmath8 ) . @xmath90 and @xmath91 are given by @xmath92 the symbols in the above equation are @xmath93 where @xmath94 . in the c.m .
frame of the lepton pair ( see fig . [
kine]b ) there is the relation @xmath95 , where @xmath96 is the angle of the photon with respect to the momentum of the incoming positron . thus @xmath97 is in the interval @xmath98 .
then we write the differential cross - section as @xmath99 the three terms in the above equation give the contribution from the @xmath2-even channel , the @xmath2-odd channel and the interference of the two channels , respectively . in the case of the charged pion pair production ,
however , there is also contribution from a bremsstrahlung type process that @xmath100 , where a time - like photon produced in a qed process decays into the pion pair . the differential cross - section of this process in the kinematics that we consider ( @xmath101 ) has the form @xmath102 to complete the analysis
, we give the the interference term between the production at the @xmath0-pole and the bremsstrahlung process : @xmath103.\label{intecs}\end{aligned}\ ] ] in eq .
( [ bremcs ] ) and ( [ intecs ] ) the @xmath104 suppressed terms are not kept , since they are irrelevant at the @xmath0-pole energy scale .
we now use a simple model for the 2@xmath7das to estimate the cross - section of pion pair production at the @xmath0-pole through eq .
( [ dcs ] ) . in the following we use @xmath63 to represent the @xmath2-even / odd part of @xmath105 .
isospin invariance implies @xcite @xmath106 therefore , only the first term in ( [ dcs ] ) contributes to the neutral pion pair production .
here we only consider the contributions from the @xmath107 and @xmath108 quark .
another consequence of isospin invariance is @xmath109 diehl et.al .
gave a simple model for @xmath110 based on the legendre polynomials expansion of the gdas , as follows @xcite @xmath111 , \label{phiplus}\end{aligned}\ ] ] in which @xmath112 and @xmath113 are the s wave and d wave phase shifts of elastic @xmath59 scattering , respectively .
the analysis for these phase shifts is available , for example , in ref .
this model is valid in the regime @xmath114 gev , and we restrict our calculation to this regime
. for @xmath115 we use the asymptotic form at large @xmath116 , corresponding to the p wave contribution @xcite : @xmath117 where @xmath118 is the time - like pion form factor , for which we use the parametrization @xmath119 given in ref . @xcite . in fig .
[ wdepend ] we show the @xmath12 dependence of the differential cross - section for the neutral pion and charged pion pair production using the 2@xmath7das given in ( [ phiplus ] ) and ( [ phiminus ] ) .
we also show the contribution from the interference term of the @xmath2-odd and @xmath2-even parts of the @xmath120 to the charged pion production at the @xmath0-pole .
the corresponding @xmath97 dependence and @xmath121 dependence of the differential cross - section are given in fig .
[ ydepend ] and fig .
[ thetadepend ] , respectively .
the figures show that the cross - section of the charged pion pair production is much larger than that of the neutral pion pair , nearly @xmath122 order of magnitude in the region @xmath123 gev @xmath124 gev .
the contribution of the interference term is significant compared with the cross - section of the neutral pion pair production , but still several times smaller than that of the charged pion production .
one can conclude that most of the contribution to the charged pion production comes from @xmath125 .
this is expected from the large magnitude of the time - like form factor of the pion and the different @xmath23 dependence of the @xmath126 given in ( [ phiplus ] ) and ( [ phiminus ] ) . in the @xmath12 range which fig .
[ wdepend ] covers , the contributions from the bremsstrahlung process and its interference with the production at the @xmath0-pole are not suppressed , due to the factor @xmath127 and @xmath12 in the denominator of eq .
( [ bremcs ] ) and eq .
( [ intecs ] ) .
the differential cross - sections contributed by these two terms at @xmath128 are depicted in fig .
[ bremin ] , where the interference contribution has been scaled by a factor 10 . in fig .
[ bremin ] we also give a comparison between bremsstrahlung with the production at the @xmath0-pole which is scaled by a factor @xmath3 .
the numerical results show that the contribution from the bremsstrahlung process is several times larger than that from the production at the @xmath0-pole ( note that in fig .
[ wdepend ] , [ ydepend ] and [ thetadepend ] the azimuthal angle @xmath129 has been integrated over ) .
the interference term given in ( [ intecs ] ) is about one order of magnitude less than the bremsstrahlung process .
the large contribution of the @xmath2-odd channel production at the @xmath0-pole compared to the production at the @xmath2-even channel provides an opportunity to obtain information on @xmath125 .
the same part also contributes in hard electroproduction @xcite and charm or @xmath130 meson decay @xcite .
however , in these cases the gdas convolute with non - perturbative gpds or meson distribution amplitudes . in the pion pair production case at the @xmath0-pole since there are no hadronic states in the initial state , gdas only convolute with the perturbative calculable hard coefficients , thus @xmath125 can be accessed more cleanly in this case .
we now discuss how to obtain the contribution coming from @xmath115 in @xmath8 annihilation process at the @xmath0-pole .
first one confront the large contribution from the bremsstrahlung process which need to be carefully treated .
because the modulus of @xmath131 has been well measured , the contribution from bremsstrahlung can be predicted quite well by eq .
( [ bremcs ] ) . the interference between the bremsstrahlung process and the production at the @xmath0-pole
is eliminated by integrating over azimuthal angle @xmath129 .
therefore the @xmath0-pole contribution can be separated out . again
notice that the interference term in ( [ dcs ] ) vanishes after integrating over @xmath97 ( or alternatively , integrating over @xmath121 ) , therefore the difference of the differential cross - sections of @xmath132 and @xmath133 production at the @xmath0-pole ( after integrating over @xmath97 ) is @xmath134 the above result depends only on the moment of @xmath115 . thus combining the measurement of charged pair production and neutral pair production at the @xmath0-pole , the moment of @xmath115 is obtained .
[ cs ] shows the contribution of @xmath115 to the charged pair production which is calculated from eq .
( [ coddm ] ) . due to the relative small cross - section ,
the extraction of @xmath115 is difficult through the pion pair production at the @xmath0-pole based on the lep - i data @xcite , while it is feasible at the possible giga-@xmath0 option @xcite of the planned international linear collider , at which the integrated luminosity can reach @xmath135 in about one yr of running .
we have shown that pion pairs with small invariance mass , produced at the @xmath0-pole through @xmath136 , can be factorized into a hard scattering coefficient convoluted with the non - perturbative two - pion distribution amplitudes .
we calculated the cross - sections for the production of a charged pion pair and of a neutral pion pair in leading order .
the comparison of these two cases show that in the regime 0.4 gev @xmath137 gev the cross - section of the charged pair production is much larger than that of neutral pair production , showing that the process is dominated by the contribution from the @xmath2-odd part of the @xmath3das . by virtue of isospin symmetry
, one can combine the measurements of the charged pair and neutral pair production at the @xmath0-pole to access the @xmath2-odd part of the @xmath3das cleanly .
the pion pair production at the @xmath0-pole at @xmath8 facilities can open opportunities to obtain valuable information on the 2@xmath7das , especially the @xmath2-odd part of 2@xmath7das , and can be viewed as the useful complement to the two - photon process .
j. h. khn and a. santamaria , z. phys .
c * 48 * , 445 ( 1990 ) .
the aleph , delphi l3 opal sld collaborations and the lep electroweak working group , the sld electroweak and heavy flavour groups , hep - ex/0509008 , to appear in physics report . | we investigate the exclusive two - pion production at the @xmath0-pole through the process @xmath1 . in the kinematical region where the invariant mass of the two pions is much smaller than the mass of the @xmath0 boson
, the process can be factorized into the convolution of a hard coefficient and a soft matrix element , the generalized distribution amplitude of two pions .
we calculate the differential cross - section of charged pion pair production and neutral pion pair production and find the former to be much larger than the later .
we show that combining the measurements of charged pion pair production and neutral pion pair production at the @xmath0-pole provides a convenient approach to access the @xmath2-odd part of the @xmath3 distribution amplitudes . |
the authors thank professors k. nakazawa , th .
a. rijken and b.f .
gibson for valuable discussions .
this work was supported by a grant - in - aid for scientific research from monbukagakusho of japan .
the numerical calculations were performed on the hitachi sr11000 at kek and fujitsu primequest580 at kyushu university .
k. nakazawa _
_ , nuclear physics ( the proceedings on the 10th international conference on on hypernuclear and strange particle physics ( hyp x ) , tokai , sept.14 - 18 , 2009 ) , in poress and to be submitted to phys . rev .
c : k. nakazawa , private communication ( 2010 ) . | energy levels of the double @xmath0 hypernucleus , @xmath1be are calculated within the framework of an @xmath2 five - body model .
interactions between constituent particles are determined so as to reproduce reasonably the observed low - energy properties of the @xmath3 , @xmath4 nuclei and the existing data for @xmath0-binding energies of the @xmath5 , @xmath6 , @xmath7 and @xmath8 systems .
an effective @xmath9 interaction is constructed so as to reproduce , within the @xmath10 three - body model , the @xmath11 of @xmath12he , which was extracted from the emulsion experiment , the nagara event . with no adjustable parameters for the @xmath2 system , @xmath11 of the ground and bound excited states of @xmath1be
are calculated with the gaussian expansion method .
the hida event , recently observed at kek - e373 experiment , is interpreted as an observation of the ground state of the @xmath1be .
15 pt in nuclear physics involving strangeness , the fundamental problem is to describe the different facets of the interactions among the baryon octet in a unified way .
our aim is to reveal various features of hyperon(@xmath13)-nucleon(@xmath14 ) and @xmath15 interactions through combined analyses for two- and many - body hyperonic systems .
for instance , @xmath16 interaction models have been constructed so far by utilizing various @xmath0 hypernuclear data to complement the limited @xmath16 scattering data .
the @xmath9 interaction is an important entry into the @xmath17 baryon - baryon interactions , decisive information about which is obtained from observations of double-@xmath0 hypernuclei and their separation energies for two @xmath0 s separated from a double-@xmath0 hypernucleus , denoted as @xmath11 . in the kek - e176/e373 hybrid emulsion experiments , there were observed several events corresponding to double-@xmath0 hypernuclei . among them
was the epoch - making observation of the nagara event , which was identified uniquely as @xmath18he in the ground state with a precise value of @xmath19 mev @xcite .
a second important observation was the demachi - yanagi event @xcite identified as @xmath20be with @xmath21 mev values of nagara and demachi - yamanagi events are given as @xmath22 mev and @xmath23 mev in @xcite and @xcite , respectively .
in ref.@xcite , however , the e176/373 emulsion events have been reanalyzed by using the new @xmath24 mass @xcite .
their newly - obtained values of @xmath11 for these events are shown in the text . ] though it was uncertain whether this event was interpreted to be the ground state or an excited state .
a newly observed double-@xmath0 event has been recently reported , called the hida event @xcite .
this event has two possible interpretations : one is @xmath25 with @xmath26 mev , and the other is @xmath27 with @xmath28 mev .
it is uncertain whether this is an observation of a ground state or an excited state . in the planned experiments at j - parc
, dozens of emulsion events for double-@xmath0 hypernuclei will be produced . in emulsion experiments , however , it is difficult to determine the spin - parity or even to know whether an observed event corresponds to a ground or excited state .
therefore , it is vitally important to compare the emulsion data with theoretical analyses to obtain a proper interpretation . in order to interpret the demachi - yanagi event , we studied in ref.@xcite the @xmath29be hypernucleus within the framework of the @xmath30 four - body cluster model , where the @xmath9 interaction was taken consistently with the nagara event .
our calculated value for the @xmath31 state was in good agreement with the observed data .
thus , the demachi - yanagi event is interpreted as an observation of the @xmath31 excited state of @xmath29be .
the aim of this paper is to interpret the new hida event on the basis of our theoretical study , adapting the method used for interpreting the demachi - yanagi event . at present
, the hida event has two possible interpretations : @xmath1be and @xmath32be . in this paper
, we assume this event is a @xmath1be hypernucleus .
it is reasonable to employ an @xmath2 five - body model for the study of @xmath1be , because , as mentioned above , the interpretation of the demachi - yanagi event for @xmath29be was possible on the basis of an @xmath30 four - body cluster model , and @xmath1be is composed of @xmath29be plus one additional neutron .
we further note that the core nucleus @xmath33be is well described by using an @xmath4 three - cluster model @xcite , and , therefore , it should be possible to model the structure change of @xmath33be due to the addition of the two @xmath0 particles as a five - body problem .
the @xmath2 five - body cluster model employed in this paper is quite challenging as a numerical computation , because of the following conditions : ( 1 ) there exist three species of particles ( @xmath34 , @xmath0 , and neutron ) , ( 2 ) five different kinds of interactions ( @xmath0-@xmath0 , @xmath0-neutron , @xmath0-@xmath34 , neutron-@xmath34 , and @xmath34-@xmath34 ) are involved , ( 3 ) one must take into account the pauli principle between the two @xmath34 particles and between the @xmath34 and neutron .
we have succeeded in developing our gaussian expansion method @xcite , used in the above mentioned study of @xmath29be , in order to perform this five - body cluster - model calculation .
it should be emphasized , before going to the five - body calculation , that all the interactions are determined so as to reproduce the observed binding energies of the two- and three - body subsystems ( @xmath3 , @xmath35 , @xmath4 , @xmath5 , @xmath6 , @xmath10 , and @xmath36 ) .
we then calculated the energies of the @xmath37 four - body subsystems , @xmath38be ( @xmath39 ) and @xmath29be ( @xmath40 ) , and found that they are simultaneously reproduced with no additional adjustable parameters needed .
we are then able to calculate both the ground and excited states of @xmath1be with no free parameter . on the basis of careful calculations
, we shall show that the recently reported hida event can be considered to be an observation of the ground state of @xmath1be . in order to take into account the full five - body degrees of freedom of the @xmath2 system and the full correlations among all the constituent five particles
, we describe the total wave function , @xmath41 , as a function of the _ entire _ 35 sets of jacobi coordinates @xmath42 in which the two @xmath0 s ( two @xmath34 s ) have been antisymmetrized ( symmetrized ) .
some of the important coordinate sets @xmath43 are shown in fig .
[ fig : be11-jacobi ] .
the total wave function is described as a sum of five - body basis functions @xmath44 , which is an extension of our previous work using four - body basis functions @xcite : @xmath45 where @xmath46 ( @xmath47 ) is the antisymmetrizer ( symmetrizer ) for two @xmath0 s ( two @xmath34 s ) , and @xmath48_i \ , \varphi^{(c)}_{n'l'}(\mbox{\boldmath $ \rho$}_c )
\big]_k \ , \phi^{(c)}_{n'l'}({\bf s}_c ) \ , \big]_{l } \nonumber \\ \,\ , & & \quad \times\ , \big[\ , [ \chi_{\frac{1}{2}}(\lambda_1 ) \chi_{\frac{1}{2}}(\lambda_2)]_\sigma \ , \chi_{\frac{1}{2}}(n ) \big]_s \,\ , \bigg]_{jm } \ , , \label{eq:5-bodybasis}\end{aligned}\ ] ] with @xmath49 denoting a set of the quantum numbers . in eq .
( [ eq:5-bodybasis ] ) , @xmath50 is the internal wave function of an @xmath34-cluster having @xmath51 configuration and is used in the folding procedures for the @xmath52 , and @xmath3 interactions .
the @xmath53 and @xmath54 are the spin functions of the @xmath0 and @xmath55 , respectively . following refs .
@xcite , the radial shapes of the basis function @xmath56 are taken to be gaussians with ranges postulated to lie in a geometric progression and similarly for @xmath57 and @xmath58 .
the expansion coefficients @xmath59 and the eigenenergy @xmath60 of the total wave function @xmath41 are determined by solving the five - body schrdinger equation using the rayleigh - ritz variational method . in the present @xmath2 five - body model for @xmath1be ,
it is absolutely necessary that any subcluster systems composed of the two , three , or four constituent particles are reasonably described by taking the interactions among these systems . in our previous work on double @xmath0 hypernuclei with @xmath61 within the framework of the @xmath62 four - body cluster model ( @xmath63 , and @xmath34 ) @xcite , the @xmath34-@xmath64-@xmath65-@xmath66-@xmath55 , and @xmath0-@xmath0 interactions
were determined so as to reproduce well the following observed quantities : ( i ) energies of the low - lying states and scattering phase shifts in the @xmath35 and @xmath3 systems , ( ii ) @xmath0-binding energies @xmath67 in @xmath68he ( @xmath69 ) , @xmath70he ( @xmath71 ) and @xmath72be ( @xmath73 ) , ( iii ) double-@xmath0 binding energies @xmath11 in @xmath12he @xmath74 ) , the nagara event .
we then predicted , with no more adjustable parameters , the energy level of @xmath29be ( @xmath75 ) , and found that , as mentioned before , the demachi - yanagi event was an observation of the @xmath31 excited state of @xmath29be . in the present five - body calculation
, we employ the interactions of ref .
@xcite so that those severe constraints are also successfully met in our two- , three- , and four - body subsystems .
but , the present core nucleus @xmath33be @xmath76 does not belong to the subsystems studied previously .
since use of the same @xmath3 and @xmath35 interactions does not precisely reproduce the energies of the low - lying states of @xmath33be measured from the @xmath4 threshold ( the same difficulty is seen in another microscopic @xmath4 cluster - model study @xcite ) , we introduce an additional phenomenological @xmath4 three - body force with a gaussian shape , @xmath77 , having @xmath78 fm , @xmath79 fm and @xmath80 for the negative - parity ( positive - parity ) state so as to fit the observed energies of the @xmath81 ground state and the @xmath82 and @xmath83 excited states in @xmath33be . for the latter three resonance states , the same bound - state approximation ( namely , diagonalization of the hamiltonian with the @xmath84-integrable basis functions ) was applied . simultaneously , we found that the calculated @xmath67 of the ground state @xmath85 of @xmath86 reproduced well the observed value . in order to reproduce the observed @xmath87 mev for @xmath88
, we tuned the @xmath9 interaction \{eq.(3.6 ) in ref .
@xcite } by multiplying the strength of the @xmath89 part by a factor 1.244 . as for @xmath90
, we obtained @xmath91 mev and @xmath92 mev , which explain , respectively , the demachi - yanagi event for the @xmath93 state ( @xmath94 mev ) and the data of ref .
@xcite for the ground state ( @xmath95 mev ; see table 5 of ref.@xcite ) .
the above successful check of the energies of the subsystems encourages us to perform the five - body calculation of @xmath25 with no adjustable parameter , expecting high reliability for the result . in fig . 2 , convergence of the calculated energy of the @xmath96 ground state of @xmath1be
is illustrated with increasing number of five - body basis functions @xmath44 in eq.(1 ) .
the energy is measured from the five - body breakup threshold .
the lowest threshold is @xmath97 , located at @xmath98 mev .
the most important configuration in the ground - state wave function is found to be of the @xmath99 type , namely the configuration described by using the jacobi - coordinate sets @xmath100 in fig . 1 ; here , @xmath101 denotes the @xmath4 three - body degrees of freedom .
the energy point ( a ) is obtained by taking @xmath102 configuration in fig.1 , the point ( b ) is obtained by @xmath102 and @xmath103 and the point ( c ) is obtained by @xmath102 to 4 . the additional @xmath104-mev energy gain from point ( c ) to ( d ) is obtained by including the configurations of the @xmath105 type described with the jacobi coordinates such as @xmath106 and @xmath107 , in which @xmath108 stands for the @xmath6 degrees of freedom .
another @xmath109-mev gain from point ( d ) down to the converged value ( @xmath110 mev with the accuracy of 10 kev totally with @xmath111 basis functions ) is achieved by including all the other types of configurations such as @xmath112 , @xmath113 , @xmath114 , @xmath115 .
thus , the ground state is found to be bound by 5.1 mev below the lowest threshold .
the angular momentum space of @xmath116 was found to be sufficient to obtain convergence for the energy .
0.2 cm cccc ( fm ) & @xmath117 & @xmath118 & @xmath119 + + + @xmath120 @xmath121be @xmath120 & @xmath120 3.68 @xmath120 & @xmath120 3.98@xmath120 & @xmath120 3.54@xmath120 + @xmath38be & @xmath120 3.28@xmath120 & @xmath120 3.53@xmath120 & @xmath120 3.14@xmath120 + @xmath1be & @xmath120 3.10@xmath120 & @xmath120 3.33 @xmath120 & @xmath120 2.94@xmath120 +
+ it is interesting to look at the dynamical change of the nuclear core , @xmath33be , which occurs due to the addition of two @xmath0 particles .
the possibility of nuclear - core shrinkage due to a @xmath0-particle addition was originally pointed out in ref .
@xcite by using the @xmath122 three - cluster model ( @xmath63 , and @xmath34 ) for @xmath123-shell @xmath0 hypernuclei . as for the hypernucleus @xmath124li , the prediction of some 20%-shrinkage , in ref .
@xcite and in an updated calculation @xcite , was actually confirmed by experiment @xcite .
as far as @xmath123-shell double-@xmath0 hypernuclei are concerned , the present authors found @xcite , using an @xmath125 model , that participation of the second @xmath0 particle can induce a further @xmath126% reduction in the distance between the @xmath34 and @xmath127 in the nuclear core .
for the present five - body @xmath2 system , @xmath1be , table i shows the r.m.s .
distances @xmath117 between two @xmath34 particles , @xmath118 between @xmath34 and @xmath55 and @xmath119 between @xmath128 and @xmath55 in @xmath33be , @xmath86 and @xmath25 . when the first @xmath0 particle is added to @xmath33be ,
a reduction of those distances is about 20% , whereas the adding a second @xmath0 reduces the distances by about 6% ; this is similar to the case of double @xmath0 hypernuclei studied with the @xmath125 four - body model @xcite . finally , let us discuss the energy spectra of @xmath1be and its relation to the hida event .
using the same framework and interactions as those in the @xmath96 ground state , we calculated energies and wave functions of the @xmath129 , @xmath83 and @xmath130 states of @xmath1be ; there is no other bound state below the lowest @xmath131 threshold .
the energy level is illustrated in fig .
[ fig : level ] together with that of @xmath33be .
interestingly enough , the order of the @xmath132 and @xmath133 states is reversed from @xmath33be to @xmath25 .
this is because the energy gain due to the addition of the @xmath0-particle(s ) is larger in the compactly coupled state ( @xmath134 than in the loosely coupled state ( @xmath83 ) .
note that the same type of theoretical prediction was reported , in our early work @xcite for @xmath135c based on the @xmath136 model , for the @xmath0 particle addition to the compactly coupled state ( @xmath137 ) and to the loosely coupled state ( @xmath138 ) in @xmath139c . as seen in fig .
[ fig : level ] the calculated value of @xmath140be ) is 18.23 mev for the @xmath96 ground state , while for the excited states the @xmath11 values are calculated to be less than 15.5 mev .
therefore , the observed hida event can be interpreted to be the ground state . when our calculated binding energy is compared with the experimental value of 20.83 mev with a large uncertainty of @xmath141=1.27 mev
, we can say at least that our result does not contradict the data within 2@xmath141 .
so far , the @xmath9 interaction strength has been often estimated rather intuitively by the quantity @xmath142 .
the observed value of @xmath143he ) is 0.67 mev , to which our @xmath9 interaction is adjusted .
the calculated value for the ground state of @xmath29be is 1.32 mev . on the other hand , that for @xmath1be is obtained as only 0.29 mev . here , the @xmath144be )
is given by a weighted sum of the values for ground @xmath145 and excited @xmath146 states of @xmath38be so that there is no contribution of the @xmath16 spin - spin interaction in the double-@xmath0 state .
one should notice here the remarkable difference between @xmath147 values for @xmath29be and @xmath1be .
furthermore , as discussed in the previous paper @xcite , values of @xmath147 include rearrangement effects in nuclear cores due to participation of @xmath0 hyperons .
then , we showed that the @xmath148 defined by eq .
( [ eq : v - bond ] ) was useful for estimating the strength of the @xmath9 interaction @xmath149 where @xmath150 denotes the @xmath11 value calculated by putting @xmath151 .
table ii lists the calculated values of @xmath148 for @xmath18he , @xmath29be and @xmath1be which are similar to each other .
thus , the obtained @xmath9 bond energy in @xmath1be turns out to be reasonable in spite of the small value of @xmath147 .
0.2 cm ccccc + & @xmath18he & @xmath29be & @xmath1be & + + + + @xmath152 & @xmath120 0.54 @xmath120 & @xmath120 0.53@xmath120 & @xmath120 0.56@xmath120 & ( mev)@xmath120 + + + in conclusion , motivated by the recent observation of the hida event for a new double @xmath0 hypernucleus , we have succeeded in performing a five - body calculation of @xmath25 using an @xmath153 cluster model .
the calculated @xmath9 binding energy does not contradict the interpretation that the hida event is an observation of the ground state of @xmath25 . in this model
, we described the core nucleus @xmath33be using the @xmath4 three - cluster model and found a significant structure change ( shrinkage ) of @xmath33be due to the addition of the two @xmath0 particles . for an alternative interpretation of the hida event as the ground ( or any excited ) state of @xmath32be ,
a corresponding six - body @xmath154 model calculation is necessary , but such an undertaking is beyond our present consideration .
more precise data are needed in order to test our present result quantitatively . in the near future , many data for double @xmath0 hypernuclei are expected to be found in the new emulsion experiment e07 at j - parc . then , our systematic predictions including the work of ref .
@xcite will be clearly tested . |
the study of two - component fractional quantum hall ( fqh ) effect has revealed a tremendous amount of new physics .
the earlier studies were performed on gaas systems @xcite , where the land g - factor is small and therefore both components of spin can be active at relatively small magnetic fields .
more recently , two - component fqh effect has been studied in systems where valleys play the role of spin , as in alas quantum wells @xcite and h - terminated si(111 ) surface @xcite ; here the zeeman energy is large enough to freeze the spin degree of freedom for typical experimental parameters . in graphene
, the two components could be either spins or valleys , depending on parameters @xcite .
experiments have shown that in general , fqh states with several spin / valley polarization can occur at a given filling factor , and transitions between them can be caused by tuning the zeeman / valley splitting .
these level crossing transitions are understood in terms a competition between the composite fermion ( cf ) cyclotron energy and the zeeman / valley splitting .
a quantitative understanding of this physics has been achieved through halperin s multi - component wave functions @xcite , and more generally through the theory of spinful composite fermions @xcite . given a rich diversity of strongly correlated states of fermions involving the spin physics ,
it is natural to ask what new physics can be learned from the study of two - component bose gases , such as those made up of two hyperfine spin states of the same atoms , in the fqh regime .
neutral bosons can in principle be driven into the fqh regime by rapid rotation @xcite .
strongly correlation among particles is achieved as the number of vortices @xmath13 in a rotating bose - einstein condensate ( bec ) becomes comparable with the number of atoms @xmath14 as quantified by the filling factor @xmath15 . for simplicity , we will refer to the two components as spins , but the results apply to any two - component bosons for which the interaction is ( approximately ) independent of the component index .
there has been much recent study of bosonic quantum hall states @xcite .
it has been shown that the vortex lattice that forms at large @xmath16 melts and that a series of fqh states appear at various filling factors , which include , for appropriately chosen interactions , laughlin @xcite , jain @xcite , moore - read @xcite and read - rezayi @xcite states .
while fqh effect in cold atom systems has not yet been observed in a convincing manner , substantial progress in that direction has been reported @xcite .
other ingenious methods to simulate the effect of magnetic field have also been explored @xcite .
we consider below two - component bosons in the fqh regime . aside from the experimental interest , a motivation for studying this problem is the possibility of realization of new structures that are not available in electronic fqh effect .
in particular , we will see that some fqh states require a tuning of the interaction , which is more easily accomplished in ultracold atomic systems .
the theoretical study of fqh effect has relied on the notion of formation of emergent quasiparticles , description in terms of which provides a tangible way to understand the physical properties of an inherently hard quantum @xmath14-body problem .
the physics of emergent quasiparticles is captured by appropriate wave functions , which , in turn , represent topological phases . to fully classify all topological phases
is a formidable task , but progress has been made in the context of topological insulators and superconductors @xcite .
however , one can take specific examples and ask if they occur for models with realistic interactions .
we consider in this article several bosonic spin - singlet and spin - partially - polarized states , and ask for what kinds of interaction they would be realized .
some of these support excitations with abelian braid statistics , whereas some with non - abelian braid statistics .
the plan of this paper is as follows . in sec .
[ wave ] we introduce all of the trial wave functions that we study in the subsequent sections . sec .
[ model ] describes our model , and our methods for evaluating the wave functions , exact diagonalization and entanglement spectra .
[ fractional ] presents the results for bosons at fractional fillings and sec .
[ integral ] for integral fillings .
[ conclusion ] summarizes the conclusions of our study .
in this section , we shall use the symmetric gauge on disk geometry where the lowest landau level ( lll ) wave functions are particular simple as given by @xmath17 where @xmath18 is the complex coordinate of particles on the disk .
the ubiqutous exponential factor will be omitted in the rest of this paper .
a bosonic fock state is represented using symmetric monomials and a many - body state is the superposition of all monomials with appropriate quantum numbers .
we use the convention that the coordinates @xmath19 and @xmath20 denote , respectively , spin - up and spin - down particles , whereas @xmath21 denote _ all _ particles . the general wave function of two - component bosons ( with @xmath22 spin - up and @xmath23 spin - down bosons ) at filling factor @xmath16 has the form @xmath24\ ] ] where @xmath25 is the spatial part , @xmath26 and @xmath27 refer to the two components , and @xmath28 denotes symmerization .
it is sufficient to consider @xmath29 provided it satisfies appropriate symmetries . an acceptable wave function with spin @xmath30 must satisfy fock s cyclic condition , which means that the state @xmath31 is annihilated by an attempt to antisymmetrize a spin - down particle with respect to the spin - up particles .
this condition is satisfied for the wave functions considered below .
\(i ) the halperin 221 state at @xmath0 state is given by @xmath32 this form of multi - component wave functions were introduced by halperin for electronic fqh states @xcite .
the halperin 221 wave function vanishes as the third power of distance between particles when two particles are brought together , _
regardless _ of their spin configuration .
it is the exact ground state for the contact interaction @xmath33 .
\(ii ) the jain s cf ( jcf ) states at @xmath34 are given by @xmath35}_{\frac{n}{n+ 1}}(\{z\ } ) & = & { \cal p}_{\rm lll } \biggl [ \phi_{n_\uparrow}(\{z^\uparrow\ } ) \phi_{n_\downarrow}(\{z^\downarrow\ } ) j(\{z\ } ) \biggr ] \label{jaincf1 } \\
n_\downarrow]}_{\frac{n}{n-1}}(\{z\ } ) & = & { \cal p}_{\rm lll } \biggl [ \phi_{-n_\uparrow}(\{z^\uparrow\ } ) \phi_{-n_\downarrow}(\{z^\downarrow\ } ) j(\{z\ } ) \biggr ] \label{jaincf2}\end{aligned}\ ] ] where @xmath36 , @xmath37 and @xmath38 is the jastrow factor for _ all _ particles ; @xmath39 and @xmath40 are two slater determinants for the spin - up and spin - down particles at fillings @xmath41 and @xmath42 , respectively , and their complex conjugates @xmath43 and @xmath44 represent filled ll states in opposite magnetic field ; @xmath45 ; and the symbol @xmath46 represents the lll projection operator . the spin polarization is given by @xmath47 those with @xmath48 are spin - singlet , while those with @xmath49 ( _ i.e. _ odd @xmath50 ) are spin - partially - polarized ( or spin - polarized ) .
these wave functions are closely related to those studied previously for electronic fqh effect@xcite , where they represent the physics of electrons capturing two vortices to turn into composite fermions , which then form integer quantum hall ( iqh ) states . in the present case ,
the bosons capture one vortex each to form composite fermions , which experience a reduced effective magnetic field @xmath51 ( @xmath52 is the external field and @xmath53 is the density ) and condense into iqh states ( with filling factor denoted as @xmath54 ) to produce incompressibility .
an intuitive reason for why bosons convert into composite fermions is because this builds good correlations that keep the particles away from one another and thus reduce the interaction energy . for @xmath55
these wave functions reduce to fully spin polarized bosons which haven been considered previously @xcite . the wave functions in eq .
( [ jaincf1 ] ) and eq .
( [ jaincf2 ] ) are interpreted as the states in which composite fermions fill @xmath41 spin - up and @xmath42 spin - down @xmath56 levels ( @xmath56ls ) , where @xmath56ls are landau - like levels of composite fermions .
a noteworthy aspect of the analogy to iqh effect is that it goes beyond the ground state and also allows construction of wave functions for the excitations of the @xmath34 state in terms of the known excitations of the iqh states .
in fact , the cf theory implies a one - to - one correspondence between the excitations at @xmath57 and those at @xmath16 , because an iqh wave function with a given spin and angular momentum quantum numbers produces , through eqs .
( [ jaincf1 ] ) or ( [ jaincf2 ] ) , a wave function at @xmath16 with the same quantum numbers .
in particular , neutral and charged excitations of the iqh state at @xmath57 produce neutral and charged excitations of the state at @xmath16 . in what follows ,
_ the jcf wave function @xmath58}_{n/(n{\pm}1)}$ ] will collectively represent wave functions for the ground state as well as neutral and charged excitations . _ we study below @xmath59}_{2/3}$ ] , @xmath60}_{4/5}$ ] , @xmath61}_{4/3}$ ] , @xmath62}_{3/4}$ ] , @xmath63}_{3/2}$ ] and @xmath64}_{2}$ ] , including ground state and excitations .
we note that for @xmath65 the ground state wave function is given by @xmath66}_{\frac{2}{3 } , { \rm g.s . } } ( \{z\ } ) = { \cal p}_{\rm lll } \left [ \phi_{1}(\{z^\uparrow\ } ) \phi_1(\{z^\downarrow\ } ) j(\{z\ } ) \right ] \nonumber \\ = \prod_{i < j } ( z^\uparrow_i - z^\uparrow_j ) ( z^\downarrow_i - z^\downarrow_j ) \prod_{i , j}(z_i - z_j)\end{aligned}\ ] ] ( no lll projection is required in this case ) which is identical to the halperin 221 wave function . in other words ,
the halperin-221 state is interpreted as the @xmath12 spin - singlet state of composite fermions .
this interpretation also allows a construction of the excitations of the 2/3 state by correspondence with the excitations of the @xmath12 spin - singlet iqh state @xmath67 .
\(iii ) the simplest non - abelian spin - singlet ( nass ) state @xcite at filling factor @xmath68 can be written as a symmetrized product of @xmath69 copies of the halperin @xmath70 state @xmath71 \label{nasswave}\ ] ] where the particles are divided into @xmath69 groups with coordinates @xmath72 and @xmath73 denotes the _ separate _ symmetrization of the spin - up and spin - down particles .
it may be viewed as a spin - singlet generalization of read - rezayi @xmath74 states @xcite whose excitations obey non - abelian braiding statistics @xcite .
it is the exact zero energy ground state of a model @xmath75-body contact interaction .
it has recently been suggested @xcite that the 4/3 nass state may be realized even for the 2-body contact interaction .
we will also study excitations of this state .
the quasihole excitations , obtained by adding flux quanta , also have zero energy for the @xmath75-body interaction , and can be explicitly constructed @xcite .
the neutral excitations and the quasiparticles of the @xmath75-body hamiltonian are nontrivial and do not have zero energy . to construct trial wave functions for them , we generalize eq .
( [ nasswave ] ) to @xmath76}_{2\over 3 } ( \{z^\alpha\ } ) \psi^{[1,1]}_{2\over 3 } ( \{z^\beta\ } ) \cdots \psi^{[1,1]}_{2\over 3 } ( \{z^k\ } ) \biggr ] \label{nasswave2}\ ] ] this reproduces the wave function of eq .
( [ nasswave ] ) when all factors @xmath59}_{2/ 3}$ ] are chosen as the ground states ( _ i.e. _ the halperin 221 state ) , but also produces excitations by appropriate choice of excited states on the right hand side .
for example , the lowest energy neutral excitations corresponds to a cf exciton in a single factor @xmath59}_{2/3}$ ] .
this approach for constructing excitations follows a multipartite cf " representation investigated recently to study the excitations of the moore - read state @xcite and the read - rezayi @xmath77 state @xcite .
the nass state can also be generalized to produce other candidate incompressible states by replacing the halperin @xmath70 state with @xmath58}_{n/(n\pm 1)}$ ] .
\(iv ) moran _ et al . _
@xcite recently studied the jain spin - singlet ( jss ) wave function for fermions , which they argued contains topological @xmath27-wave pairing structure .
we consider here its bosonic analog at @xmath9 @xmath78 \label{jsswave}\ ] ] where @xmath79 is the wave function of _ two _ filled landau levels .
this does not belong to the @xmath58}_{n/(n\pm 1)}$ ] states considered above , but follows from the parton construction of fqh states @xcite . in this construction ,
each boson is viewed as the bound states of two fictitious species of fermions ( partons ) , one of which carries spin while the other is spinless .
the spinful fermions occupy the spin - singlet state at @xmath11 whereas the spinless ones occupy the fully spin - polarized state at @xmath11 .
the fermionic version of this state ( obtained by multiplication by another full jastrow factor ) describes a spin - singlet incompressible state at @xmath80 ; it was introduced in ref . [ ] and considered as a possible candidate for the spin - singlet @xmath81 fqh state @xcite , but was abandoned when it was realized that the coulomb @xmath81 state is fully spin - polarized .
we will see below in sec .
[ integral ] that this state is realizable for a 2-body interaction .
this result is of interest because @xmath82 is the simplest
parton " state that goes beyond the cf interpretation ( all states of composite fermions admit a parton construction but the converse is not true ) .
the excitations of this state are more complicated .
one may naively expect that the low - lying energy levels can be obtained by creating excitations in either @xmath79 or @xmath83 in eq .
( [ jsswave ] ) . however , it turns out that neither of them gives a very accurate description of the excitations , as we shall see in sec . [ integral ] .
many of the above wave functions involve @xmath84 , the slater determinant wave function of @xmath50 filled lls , on the right hand side . while @xmath84 is uniquely defined for a _ compact _ geometry ,
where the number of single - particle states in each landau level is finite , that is not the case in the disk geometry .
for example , in the disk geometry @xmath79 can be defined with @xmath85 particles in the lowest landau level and @xmath86 particles in the second landau with the constraints that @xmath87 .
different possible choices of @xmath88 complicates the analysis of the edge excitations of the states involving @xmath79 , as has been found to be the case for spin - polarized fermions at 2/5 @xcite .
we consider a bosonic system with two internal states in a rapidly rotating harmonic trap .
these neutral particles experience forces in the rotating reference frame which mathematically has the same description as charged particles moving in a uniform magnetic field .
we specialize to the case where single - particle cyclotron energy is much larger than the many - body gap , so the bosons can be treated as in the lowest landau level only and effects due to landau levels mixing are negelected .
the number of particles , the number of spin - up particles and the number of spin - down particles are denoted using @xmath14 , @xmath89 and @xmath90 , respectively .
we will use the spherical geometry @xcite for most of our calculations . the flux enclosed by the sphere
is denoted as @xmath91 , which is related to the numbers of particles @xmath14 and the filling factor @xmath16 via @xmath92 .
the quantity @xmath93 is called the shift . " sometimes there is an ambiguity when two states at different fillings alias , " _
i.e. _ , occur at the same flux . in such cases , it is important to study several values of @xmath14 to draw unambiguous information .
the compact spherical geometry is very convenient for studying the bulk properties of a fqh state , due to absence of edges . for studying the structure of edge excitations ,
there are two ways of proceeding .
one can study either the states in the disk geometry , or the entanglement spectrum in the spherical geometry @xcite ( see sec . [ model ]
the single - particle eigenstates on a sphere are the so - called monopole harmonics @xcite @xmath94 where @xmath95 ( @xmath50 is the landau level index ) is the angular momentum , @xmath96 is the @xmath97 component of angular momentum , and @xmath98 and @xmath99 are the azimuthal and radial angles .
the spinor coordinates @xmath100 , @xmath101 and the normalization coefficient @xmath102 is @xmath103 when eq .
( [ jaincf1 ] ) and eq .
( [ jaincf2 ] ) are constructed on a sphere , the flux @xmath104 experienced by composite fermions , that is , the flux of the iqh states @xmath105 , is related to the actual flux by @xmath106 .
once the iqh states are constructed using the above single - particle wave functions , we multiply them by the jastrow factor @xmath107 and then project the products to the lll . an efficient jain - kamilla method @xcite has been developed that applies to states of the form @xmath108 , where the projected wave function can be constructed for rather large @xmath14 without the need for expanding it in basis functions .
this method requires _ even _ exponent of @xmath107 for technical reasons . in ref .
[ ] this method was applied to spinless bosons , by writing @xmath109 as @xmath110 .
unfortunately , this method does not work for spin - singlet sates , because @xmath111 is a singular , non - normalizable wave function , as @xmath112 does not vanish when two particles with opposite spins coincide .
therefore , we must evaluate the lll projection by using its expansion in terms of the symmetric monomials for the spin - singlet states @xcite .
the following identity of monopole harmonics discovered by wu and yang @xcite are useful in the lll projection @xmath113 where we have defined the following quantities @xmath114 here @xmath115 , @xmath116 and @xmath117 is the clebsch - gordon coefficient .
the computational time to perform the lll projection grows factorially with the number of particles , since one must consider all possible permutations of the indices . as a result , @xmath118 or
@xmath5 is the maximum number of particles that we can study in a reasonable amount of time .
interaction between particles can be parametrized by the haldane pseudopotential in the 2-body case and their generalizations in the 3-body case @xcite .
we study hamiltonians containing 2-body and 3-body interactions , denoted as @xmath119 and @xmath120 , respectively : @xmath121 \\ h_3 & = & \sum_{ijk } \left [ p_{ijk}(0,3/2 ) + p_{ijk}(0,1/2 ) \right]\end{aligned}\ ] ] where @xmath122 projects out a pair of particles @xmath123 with _ relative _ angular momentum @xmath124 and total spin @xmath125 , and @xmath126 projects out a triple of particles @xmath127 with relative angular momentum @xmath124 and total spin @xmath125 .
the natural interaction for bosons is the contact interaction , which corresponds , in units of @xmath128 , to @xmath129\ ] ] this will be the interaction used unless otherwise stated .
non - zero values for @xmath130 and @xmath131 in @xmath119 , and the 3-body hamiltonian @xmath120 will also be used sometimes , to stabilize certain interesting states .
since the interaction is rotationally invariant and spin - independent , the energy eigenstates are also eigenstates of orbital angular momentum @xmath132 [ with eigenvalue @xmath133 and spin angular momentum @xmath134 [ with eigenvalue @xmath135 . in the figures shown below
, the energy levels are labeled by their angular momentum and spin quantum numbers @xmath124 and @xmath125 and are also shifted horizontally according to their @xmath125 values for clarity .
to study edge excitations , we use the disk geometry .
the hamiltonian can also be represented using 2-body haldane pseudopotentials @xmath136 \nonumber \\ & + & \omega_c ( { \widehat l}_z - l_0)\end{aligned}\ ] ] where @xmath137 is the @xmath97-component angular momentum operator and the term @xmath138 is due to a parabolic confinement potential whose strength is controlled by the parameter @xmath139 .
we choose the coefficients @xmath140 to have the same values as their counterparts in the spherical geometry hamiltonian and tune the coefficient @xmath139 to make sure that the state at angular momentum @xmath141 has the lowest energy , where the counting of edge excitations starts .
in addition to comparing the wave functions with exact eigenstates obtained in finite systems , we also study the entanglement spectrum @xcite in some cases , because it can provide additional insight into the physics of the fqh states .
in particular , it has been found that the entanglement spectrum contains information about the edge excitations ; specifically , entanglement spectrum can reproduce the counting of the edge states ( which provides a method of study edge excitations in the spherical geometry ) . to obtain the entanglement spectrum for an incompressible ground state @xmath142 ,
one divides the hilbert space into two parts labeled as @xmath143 and @xmath52 and then decomposes the ground state as @xmath144 where @xmath145 and @xmath146 are two sets of basis states for @xmath143 and @xmath52 , respectively .
the second step is achieved through a singular value decomposition ( svd ) of the matrix @xmath147 , which also changes the basis states to @xmath148 and @xmath149 . a plot of the eigenvalues " @xmath150 versus the conserved quantum numbers in region @xmath143 comprises the entanglement spectrum .
we shall calculate the real space entanglement spectrum " @xcite ( rses ) , where the cut is made along the equator and the southern hemisphere is chosen as @xmath143 , with @xmath151 ( @xmath152 ) spin - up ( spin - down ) particles .
due to the choice of cut , the levels in the rses can be labeled by the @xmath97 component of the total angular momentum @xmath153 and the total spin quantum number @xmath154 of the particles in @xmath143 . to compare the edge excitations with the rses
, we will calculate energy spectra on disk geometry when the edge counting can not be predicted _
exactly_. for example , the counting of edge excitations of the nass state can be predicted in several ways and does not require exact diagonalization , but the counting of the edge excitations of jcf state and jss state are more complicated .
state for the 2-body contact hamiltonian @xmath155 .
the lines are colored according to their spin quantum numbers and are also shifted in the horizontal direction for clarity .
the same conventions are used in all other figures .
the crosses represent the energies of the wave functions @xmath59}_{2/3}$ ] for the ground and excited states .
the panels correspond to ( a ) @xmath156 , @xmath157 and @xmath158 ; ( b ) @xmath159 , @xmath160 and @xmath161 ; ( c ) @xmath159 , @xmath160 and @xmath162 .
the inset in ( a ) shows the color scheme for all panels .
panels ( a ) and ( b ) correspond to incompressible states where the uniform ground state has @xmath163 and @xmath164 , and the excitations are neutral particle - hole pairs of composite fermions .
panel ( c ) corresponds to a system containing two quasiparticles ; the low energy band contains all possible states of these quasiparticles.,scaledwidth=45.0% ] state for the hamiltonian @xmath155 .
the crosses represent the energies of the wave functions @xmath60}_{4/5}$ ] .
( a ) @xmath156 , @xmath157 and @xmath165 ; ( b ) @xmath166 , @xmath167 and @xmath162 ; ( c ) @xmath159 , @xmath160 and @xmath158 .
the inset in ( a ) shows the color scheme for all panels .
panels ( a ) and ( b ) correspond to incompressible states where the uniform ground state has @xmath163 and @xmath164 , and the excitations are neutral particle - hole pairs of composite fermions . panel ( c ) corresponds to a system containing two quasiholes ; the low energy band contains all possible states of these quasiholes.,scaledwidth=45.0% ] state for the hamiltonian @xmath155 with @xmath168 , @xmath169 and @xmath162 .
the crosses represent the energies of the wave functions @xmath170}_{3/4}$ ] .
( b ) energy spectrum of the @xmath6 state for the hamiltonian @xmath171 with @xmath156 , @xmath172 and @xmath158 .
the crosses represent the energies of the states @xmath173}_{3/2}$ ] .
the insets show the color schemes for the panels.,scaledwidth=45.0% ] state for the hamiltonian @xmath155 .
the crosses represent the energies of the wave functions @xmath61}_{4/3}$ ] .
( a ) @xmath166 , @xmath167 and @xmath158 ; ( b ) @xmath174 , @xmath175 and @xmath176 ; ( c ) @xmath174 , @xmath175 and @xmath177 . the inset of panel ( a ) shows the color scheme for all panels.,scaledwidth=45.0% ] state for the hamiltonian @xmath155 .
the crosses represent the energies of the wave functions @xmath178 obtained from the spinful bipartite cf theory .
( a ) @xmath166 , @xmath167 and @xmath165 ; ( b ) @xmath174 , @xmath175 and @xmath176 ; ( c ) @xmath179 , @xmath169 and @xmath158 . the inset of panel ( a ) shows the color scheme for all panels.,scaledwidth=45.0% ] state for the 3-body hamiltonian @xmath120 .
the crosses represent the energies of the wave functions @xmath178 obtained from the spinful bipartite cf theory .
( a ) @xmath166 , @xmath167 and @xmath165 ; ( b ) @xmath174 , @xmath175 and @xmath176 . the inset of panel ( a ) shows the color scheme for both panels.,scaledwidth=45.0% ] nass state .
( a ) @xmath179 , @xmath169 , @xmath180 , and @xmath181 , using the exact nass state ; ( b ) @xmath179 , @xmath169 , @xmath180 , and @xmath181 , using the ground state of the 2-body hamiltonian @xmath119 ; ( c ) @xmath156 and @xmath157 , energy spectrum on disk geometry of the hamiltonian @xmath182 with confinement potential parameter @xmath183 .
the inset of panel ( a ) shows the color scheme for all panels .
the arrow in panel ( c ) indicates the ground state and the arrows in panel ( a ) and panel ( b ) show the corresponding levels in the rses.,scaledwidth=45.0% ] ground states for the 2-body hamiltonian @xmath155 .
the cross represents the energy of the wave functions @xmath64}_{2}$ ] .
( a ) @xmath166 , @xmath167 and @xmath184 ; ( b ) @xmath174 , @xmath175 and @xmath165 ; ( c ) @xmath179 , @xmath169 and @xmath176 .
the inset of panel ( a ) shows the color scheme for all panels.,scaledwidth=45.0% ] state .
( a ) rses for the exact ground state of the 2-body hamiltonian @xmath155 for @xmath179 , @xmath169 , @xmath180 , and @xmath181 .
( b ) energy spectrum on disk geometry of the hamiltonian @xmath182 for @xmath156 and @xmath157 ; the confinement potential parameter is taken to be @xmath183 .
the inset of panel ( a ) shows the color scheme for both panels .
the arrows in panel ( b ) indicate the ground state and backward - moving edge modes and the arrows in panel ( a ) show the corresponding levels in the rses.,scaledwidth=45.0% ] ground states for the 2-body hamiltonian @xmath119 with @xmath185 , @xmath186 and all other @xmath187 for @xmath188 .
the crosses represent the energies of the wave functions @xmath189 .
( a ) @xmath166 , @xmath167 and @xmath190 ; ( b ) @xmath174 , @xmath175 and @xmath177 . the inset of panel ( a ) shows the color scheme for both panels.,scaledwidth=45.0% ] state . ( a ) rses for @xmath82 for @xmath174 , @xmath175 , @xmath191 , and @xmath192 .
( b ) rses for the ground state of the 2-body hamiltonian @xmath119 for @xmath174 , @xmath175 , @xmath191 , and @xmath192 ; the parameters of the hamiltonian are @xmath185 , @xmath186 and all other @xmath187 for @xmath188 .
( c ) energy spectrum on disk for @xmath168 and @xmath193 for the hamiltonian @xmath182 with @xmath194 and the confinement potential parameter @xmath183 .
the inset of panel ( a ) shows the color scheme for all panels .
the arrows in ( c ) indicate the states obtained with four different choices for @xmath79 in the wave function eq .
( [ jsswave ] ) , which are @xmath195 $ ] , @xmath196 $ ] , @xmath197 $ ] and @xmath198 $ ] ( from left to right ) .
the arrows in ( a ) and ( b ) show the corresponding levels in the rses , which nicely match the starting points of various edge branches.,scaledwidth=45.0% ] [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ]
\(i ) the halperin @xmath70 state is the unique exact zero energy state of the hard - core interaction @xmath155 at flux @xmath199 .
the quasihole states , obtained by adding flux , are also exact zero energy states of @xmath155 , whose counting can be predicted in several ways and the wave functions are also known exactly @xcite .
exact solutions are not known for the neutral excitations and the quasiparticles , which do not have zero energy with respect to @xmath155 . for these
we use the trial wave functions @xmath59}_{2/3}={\cal p}_{\rm lll } [ \phi_1(\{z^\uparrow\ } ) \phi_1(\{z^\downarrow\ } ) j(\{z\ } ) ] $ ] . the lowest energy
neutral excitations correspond to a particle - hole excitation in one of the @xmath200 factors .
when the flux is reduced by one unit , each @xmath200 factor on the right hand side contains one particle in the second ll .
we construct @xmath124 and @xmath125 eigenstates by taking appropriate linear combinations .
[ figure1 ] gives the energies ( shown by crosses ) of the trial wave functions of the neutral excitations in panels ( a ) and ( b ) and of quasiparticle excitations in panel ( c ) .
the overlaps between the trial wave functions and the exact eigenstates are shown in table [ table1 ] .
these comparisons show that the cf theory provides an excellent description of the excitations of the @xmath201 spin - singlet state .
\(ii ) the incompressible @xmath60}_{4/5}$ ] state occurs at @xmath202 .
we find that the system at this flux value is incompressible for up to @xmath203 particles as shown in panels ( a ) and ( b ) of fig .
[ figure2 ] .
we have explicitly constructed the wave function @xmath60}_{4/5}={\cal p}_{\rm lll } [ \phi_2(\{z^\uparrow_i\ } ) \phi_2(\{z^\downarrow_i\ } ) j(\{z\ } ) ] $ ] for the ground states and excitations .
their energies are shown by crosses in fig .
[ figure2 ] , and their overlaps with the corresponding exact states are shown in table [ table1 ] , which have excellent agreement .
we note in passing that another candidate at @xmath204 is a spin - singlet gaffnian state @xcite , but it is likely to describe a gapless or critical state rather than an incompressible state since it is given by the conformal blocks of a non - unitary conformal field theory . for the @xmath201 state ,
the edge energy spectrum is trivial and it has been found that the counting of levels in rses matches predictions @xcite .
in contrast , the edge spectrum of the @xmath1 state is expected to be complicated , containing several branches , because composite fermions occupy two @xmath56 levels .
the studies of fermionic @xmath205 state tell us that such structures can only be seen for a rather large number of particles @xcite .
the systems studied here are too small to bring out the edge physics .
the @xmath3 state @xmath170}_{3/4}$ ] occurs at @xmath207 and the @xmath4 state @xmath173}_{3/2}$ ] occurs at @xmath208 .
these are spin - partially - polarized states .
they are both derived from the spin - partially - polarized iqh state at @xmath209 , one with parallel flux attachment and the other with reverse flux attachment .
[ figure3 ] shows the energy spectra for the contact interaction @xmath155 at these two filling factors and their comparison with the trial wave functions for the ground state as well as neutral excitations .
table [ table2 ] gives the overlaps of the trial states and exact states shown in fig .
[ figure3 ] . in some orbital and spin angular momentum sectors
, there are two trial states and we define the overlap as @xmath210 ^ 2}$ ] where the summation is over the lowest two exact states @xmath211 and trial states @xmath212 .
these results show that the actual @xmath3 state is very well described by the cf theory , whereas this theory is less accurate for @xmath4 .
note that the number of particles in each spin component is fixed ( because the hamiltonian @xmath155 conserves the @xmath97-component of spin ) , so only states with total spin @xmath213 may occur .
( should we allow the spins to flip , these spin - partially - polarized state will not be ground states . )
it is interesting to note that the low energy part of the spectrum contains states with @xmath214 , with the states with higher values of @xmath125 appearing at much higher energies .
this feature is nicely explained by the cf theory as follows .
the @xmath3 and @xmath4 states map into @xmath215 $ ] and @xmath216 $ ] of composite fermions , and the lowest energy excitations ( without changing @xmath217 ) contain a single cf exciton either in the spin - up sector or in spin - down sector .
the resulting states satisfy the fock condition ( all occupied states in the spin - up sector are definitely occupied in the spin - down sector , and therefore the wave function is annihilated upon further antisymmetrization ) , and thus represent states with @xmath218 . to produce a state with @xmath219
one must consider cf configurations containing at least two cf excitons , which are expected to lie at higher energies .
the filling factor @xmath2 has been considered @xcite because it may provide a realization of the simplest nass state @xmath178 . at the same time
, the cf theory provides another candidate @xmath61}_{4/3}$ ] here .
it is therefore of interest to ask what kinds of interaction would favor these states .
the states @xmath61}_{4/3}$ ] and @xmath178 occur at different shifts with @xmath220 and @xmath221 , respectively , on the spherical geometry .
let us first consider the 2-body interaction . for the contact interaction @xmath155 , the spectrum for 12 particles at @xmath220 is shown in fig .
[ figure4](a ) , and for @xmath221 in fig .
[ figure6](a ) .
the overlaps of trial states and exact eigenstates are shown in table [ table3 ] and [ table4 ] , respectively . given that the jcf ground state has a higher overlap ( 0.985 ) than the nass ground state ( 0.918 ) in spite of a larger hilbert space ( 646 independent @xmath222 multiplets as opposed to 79 for nass )
, these comparisons suggest that the states @xmath61}_{4/3}$ ] is favored for the contact interaction .
comparison is also shown for excitations . to test the stability of the jcf and nass states at filling factor @xmath7
, we further test their performances when changing the coefficients @xmath223 for @xmath224 in the hamiltonian @xmath119 .
the results are shown in tables [ table5 ] .
both states remain good approximations for small values of @xmath130 and @xmath131 , but are destroyed at large enough values for these parameters .
we should emphasize that these numbers are not to be compared directly since the two states occur at different shift , and the dimensions of the subspaces with fixed @xmath124 and @xmath125 quantum numbers are different . as mentioned previously
, the nass is the exact ground state for the 3-body contact interaction @xmath120 .
the energy spectra corresponding to the nass shift are shown in fig .
[ figure6 ] for this 3-body interaction . from the energy comparisons shown in this figure , and the overlaps shown in table [ table4 ] ,
the excitations are very well described by the trial wave functions which create cf excitations in individual factors of eq .
( [ nasswave2 ] ) .
we have also compared the rses of the exact nass state and the 2-body ground state in fig .
[ figure7 ] .
the rses are similar , as would be expected from the reasonably high overlaps .
we also show the energy spectrum in the disk geometry , which , however , does not has very similar structure as the rses .
in fact , the energy spectrum in fig .
[ figure7](c ) is better understood as reverse - flux - attached cf state , as described below in sec .
we have not studied the rses for @xmath61}_{4/3}$ ] or the corresponding exact ground state .
since composite fermions occupy two @xmath56 levels in both spin sectors in the @xmath2 state , we do not expect the rses to give very useful information using the system sizes that are accessible to exact diagonalization or for which @xmath61}_{4/3}$ ] can be explicitly generated .
we note that any spectrum in figs .
[ figure4 ] and [ figure5 ] can be interpreted in two different ways .
for example , the @xmath166 , @xmath167 and @xmath165 state in fig .
[ figure4](a ) can be thought of as excitations of @xmath61}_{4/3}$ ] , but here the nass gives a satisfactory account of the exact spectrum . on the other hand , for figs .
[ figure4](b ) and [ figure5](b ) , both interpretations work comparably well ( although they predict different numbers of states ) , as seen from the overlaps in tables [ table3 ] and [ table4 ] . taking all of these results into account , while our studies do not rule out the nass state , they suggest that the 4/3 ground state for the contact interaction is likely to be @xmath61}_{4/3}$ ] with abelian excitations
. it would be useful to compare these two candidate states in the torus geometry where they compete directly .
recently , composite fermion wave functions have been successfully constructed in the torus geometry @xcite for spin - polarized state at filling factors @xmath201 ( for 10 bosons ) and @xmath205 ( for 6 fermions ) . generalizing this method to spinful cases
could be very interesting , although the numerical implementation of such schemes is expected to be very difficult .
we consider the state @xmath64}_{2}={\cal p}_{\rm lll } [ \phi_{-1}(\{z^\uparrow\ } ) \phi_{-1}(\{z^\downarrow\ } ) j(\{z\ } ) ] $ ] , obtained from the @xmath12 spin - singlet state with reverse flux attachment .
this state is analogous to the spin - singlet 2/3 state of fermions @xcite .
it has attracted special interest recently as an example of symmetry protected bosonic integer topological states @xcite , which refer to states with no topological order ( _ i.e. _ , abelian or non - abelian fractional excitations ) but are still topologically non - trivial . as an initial test
, we find that the ground state of the 2-body contact interaction @xmath155 at the @xmath91 values corresponding to @xmath64}_{2}$ ] indeed has @xmath163 and @xmath164 for up to @xmath225 particles .
[ figure8 ] shows the energy spectra .
[ for 18 particles the dimension of the fock space is very large ( with 58,130,756 states in the @xmath226 sector ) , and producing eigenstates by the lanczos method is computationally time consuming ; we have only obtained the lowest few eigenstates to confirm that the ground state has @xmath222 and is separated from the excitations by a reasonable gap . ] the overlaps of trial states and exact eigenstates for @xmath227 and @xmath228 are shown in table [ table6 ] . for @xmath203 ( @xmath229 ) particles , the exact ground state has overlap 0.943 ( 0.888 ) with @xmath64}_{2}$ ] ; for @xmath5 particles we are not able to generate the trial state as explained in sec .
[ model ] b. we also study the stability of the state under addition of longer - range interaction .
table [ table7 ] shows the evolution of overlaps between @xmath64}_{2}$ ] and exact ground state for a range of values of @xmath230 ( with @xmath185 ) , demonstrating that @xmath64}_{2}$ ] remains a good description of the ground state for a wide range of parameters .
we should point out that the trial wave functions for excitations are not as accurate as the ground states as one can see from fig .
[ figure8 ] and table [ table6 ] .
we also study the rses and the edge spectrum .
the cf theory implies a behavior similar to that of the @xmath0 spin - singlet fermionic state , which has been studied in refs . [ ] and [ ] .
in particular , one expects a backward - moving mode that carries spin but no charge , and a forward - moving mode that carries charge but no spin @xcite .
we show in fig .
[ figure9 ] the rses of the ground state of a bosonic system at @xmath11 with @xmath231 particles and the edge excitations of a system with @xmath232 particles on a disk .
( we add a parabolic confinement potential of an appropriate strength in the disk geometry , to ensure that the ground state has the angular momentum given by @xmath64}_{2}$ ] . )
we see a strong similarity between the rses and the spectrum of edge excitations on the disk .
in particular , the rses nicely captures the backward - moving mode marked by the arrows in fig .
[ figure9 ] .
the counting of states for the backward - moving modes is also consistent with that found for the @xmath0 spin - unpolarized fermionic states @xcite .
the forward - moving mode is not clearly idenfiable in both the rses and the disk edge spectrum , as was also the case for spin - singlet @xmath201 state @xcite ; this can be understood by noting that the velocity of this mode is sufficiently large that it rapidly merges into the continuum for the small systems accessible to our study .
the incompressibility at @xmath11 for bosons occurs because of interactions between them , and is therefore closer to the fqh ( rather than the iqh ) of fermions .
one may ask what is the charge of the excitations . identifying an isolated cf particle or cf hole in one of the factors of @xmath200 ,
it is straightforward to see that the charge excess associated with it is equal to a unit charge .
we now explore the validity of @xmath233 .
this state is an excellent description of the ground state at filling factor @xmath9 with @xmath234 if some amount of @xmath131 interaction is turned on , as shown fig .
[ figure10 ] .
the evolution of overlap between trial states and exact ground states with the coefficients @xmath130 and @xmath131 of the hamiltonian @xmath119 is shown in table [ table7 ] .
it is natural to construct wave functions for the excitations of @xmath235 $ ] by analogy to excitations of either the factor @xmath79 or one of the two jastrow factors on the right hand side .
we have constructed such wave functions for the excited states , but neither of them gives very accurate description of the excitations .
we also study the rses at @xmath9 .
[ figure11 ] shows the rses of the jss wave function and the exact ground state wave function for a certain choice of parameters ( @xmath236 and all other @xmath237 s are set to zero ) for 14 particles .
the two have similar low - lying levels . for many trial wave functions that are exact zero energy solutions of certain simple pseudopotential hamiltonians , such as the laughlin or moore - read wave functions , the entanglement spectrum contains only universal levels , _
i.e. _ , all levels represent edge excitations . that , however , is not true in general . for electronic systems , the rses of the exact coulomb eigenstates at @xmath238 or
@xmath81 contain non - universal " levels , as is also true of the either the exact states at @xmath239 or the jcf wave functions for those states .
the trial state @xmath233 also has many non - universal levels as it is not the exact zero energy state of a simple pseudopotential hamiltonian and its construction requires lll projection .
the rses of the exact state contains even more non - universal " levels .
we also show the edge excitation spectrum on disk geometry in fig .
[ figure11 ] , and some similarities between the rses and edge spectrum can be seen even for such a small system .
a noteworthy feature is that there are several branches of edge excitations , and the starting points of these branches ( indicated by arrows in fig .
[ figure11 ] ) match nicely in both the rses and the edge spectrum .
[ note that the minimum value of angular momentum in ( a ) and ( b ) is @xmath240 while the minimum value in ( c ) is @xmath241 , so the positions of the arrows in ( a ) and ( b ) match exactly with those in ( c ) if the angular momentum values in ( a ) and ( b ) are relabeled by adding @xmath242 . ] the starting points of edge excitations can be simply predicted using parton method : they correspond to different choices for the number of particles @xmath243 $ ] in the two @xmath56 levels in the @xmath79 part of eq .
( [ jsswave ] ) , given in the figure caption .
( while the starting points of the edge branches are identifiable , they quickly spread and merge into the non - universal part , making an identification of the edge states difficult . ) the existence of multiple branches in the edge excitation spectrum and the rses have been observed before for spin - polarized fermionic @xmath205 state @xcite , which is also due to the appearance of a @xmath79 factor in the trial wave functions . in short ,
the rses and edge studies provide support to the identification of the exact state with @xmath233 , and , in particular , bring out features that can be understood by analogy to two filled @xmath56ls of composite fermions .
we have carried out an extensive study of quantum hall effect for two - component bosons , studying a number of candidate states at fractional as well as integral fillings . here is a summary of our findings : \(i )
we have shown that for 2/3 and 4/5 , the wave functions @xmath59}_{2/3}$ ] and @xmath60}_{4/5}$ ] provide an accurate representation of the spin singlet states of the contact interaction , for the ground state as well as excitations .
\(ii ) we have also considered partially polarized states at 3/4 and 3/2 . for the former the state @xmath62}_{3/4}$ ] provides an accurate description of the ground state and excitations . for 3/2 , @xmath63}_{3/2}$ ] is not accurate .
\(iii ) for @xmath7 we consider two candidates , @xmath61}_{4/3}$ ] and @xmath178 , which have abelian and non - abelian excitations , respectively .
previous works @xcite suggested that the nass state is realized at this filling factor .
we find , from a direct comparison with the exact solution , that @xmath61}_{4/3}$ ] is more likely for the 2-body contact interaction .
\(iv ) for @xmath7 , the nass state has been known to be the exact ground state for a 3-body interaction .
we find that the exact excited states of this 3-body interaction correspond to cf excitations in the individual factors , confirming a spinful bipartite cf structure for this state .
\(v ) for @xmath11 , @xmath64}_{2}$ ] provides an accurate representation of the exact ground state for in certain parameter range of the 2-body hamiltonian .
the trial wave functions for the excitations are less accurate .
the rses and the disk energy spectrum provide a consistent description of the edge structure , both nicely displaying a backward - moving edge mode ( which is similar to that found previously for the fermionic @xmath201 spin - singlet state @xcite ) .
\(vi ) for @xmath9 , the jss state accurately represents the ground state for a 2-body interaction that contains terms beyond contact interaction .
the rses and edge excitation studies provide further confirmation of the validity of @xmath244 , and , in particular , demonstrate the existene of several edge branches , which are fully consistent with the expectation from an underlying two - filled @xmath56l state .
our trial wave functions are not very accurate for the excitations .
\(vii ) we note a systematic effect as a function of the filling factor : the agreement between the cf and the exact spectra becomes worse with increasing filling factor .
there are two possible reasons for that .
one , the jcf wave functions of the states with large fillings , e.g. @xmath245 and 2 , namely @xmath63}_{3/2}$ ] , @xmath61}_{4/3}$ ] and @xmath64}_{2}$ ] , all require reverse flux attachment .
we have found that for spinful particles , the wave functions involving reverse flux attachment are less accurate than those with parallel flux attachment .
two , from general arguments one expects that bosons at very high fillings are not in the fqh regime , because vortex lattice or other weakly interacting states may be preferred energetically . + + _ note added _ at the time of the completion of this manuscript , we noticed the preprint [ ] , which also studies the @xmath11 state and has some overlap with sec .
[ nu2 ] of our paper .
another preprint [ ] about the @xmath11 state has also appeared since then .
these studies complement one another to an extent , as we briefly describe .
all three works provide evidence for an incompressible state at filling factor @xmath11 for two - component bosons .
[ ] and [ ] report finite - size scaling of gaps and consider the cases where the interaction is not @xmath246 invariant .
[ ] uses chern - simon field theory to interpret the counting on rses ; we compare the rses with the edge spectrum on the disk geometry , and also nd that edge excitations can be understood using cf theory . ref .
[ ] studies the ground state degeneracy on torus to rule out a competing non - abelian state , whereas our observation that incompressible states occur for all cases with @xmath247 in the spherical geometry also rules out the non - abelian state . we have also constructed and studied explicit trial wave functions for the ground states as well as excitations .
we are immensely grateful to the authors , especially n. regnault , of the diagham package for sharing their programs .
we thank kai sun for many useful discussions , the doe for financial support under grant no .
de - sc0005042 , and research computing and cyberinfrastructure , a unit of information technology services at the pennsylvania state university , for providing high - performance computing resources and services .
s. melinte , n. freytag , m. horvati , c. berthier , l. p. lvy , v. bayot , and m. shayegan , phys .
lett . * 84 * , 354 ( 2000 ) ; n. freytag , m. horvati , c. berthier , m. shayegan , and l. p. lvy , _ ibid . _ * 89 * , 246804 ( 2002 ) . | we investigate the feasibility of many candidate quantum hall states for two - component bosons in the lowest landau level .
we identify interactions for which spin - singlet incompressible states occur at filling factors @xmath0 , @xmath1 and @xmath2 , and spin - partially - polarized states at filling factors @xmath3 and @xmath4 , where spin " serves as a generic label for the two components .
we study ground states , excitations , edge states and entanglement spectrum for systems with up to @xmath5 bosons , and construct explicit trial wave functions to clarify the underlying physics .
the composite fermion theory very accurately describes the ground states as well as excitations at @xmath0 , @xmath1 and @xmath3 , although it is less satisfactory for the @xmath6 state . for @xmath7
a non - abelian spin - singlet " state , which is the exact ground state of a 3-body contact interaction , has been proposed to occur even for a 2-body contact interaction ; our trial wave functions are very accurate for the excitations of the 3-body interaction , but they do not describe the excitations of the 2-body interaction very well . instead , we find that the @xmath7 state is more likely to be a spin - singlet state of reverse - flux - attached composite fermions at filling @xmath8 .
we also consider incompressible states at integral filling factors @xmath9 and @xmath10 .
the incompressible state at @xmath9 is shown to be well described by the parton - based jain spin - singlet wave function , and the incompressible state at @xmath11 as the spin - singlet state of reverse - flux - attached composite fermions at @xmath12 , which provides an example of the bosonic integer topological phase . |
we define @xmath2 , where @xmath3 . an additive code @xmath4 over @xmath1 of length @xmath5 is an additive subgroup of @xmath1 .
clearly , @xmath6 contains codewords for some @xmath7 , and can be defined by a @xmath8 generator matrix , with entries from @xmath1 , whose rows span @xmath6 additively .
we call @xmath6 an @xmath9 code .
the _ hamming weight _ of @xmath10 , denoted by @xmath11 , is the number of non zero components of @xmath12 .
the _ hamming distance _ between two vectors @xmath13 , @xmath14 is @xmath15 .
the _ minimum distance _
@xmath16 of a code is defined as the smallest possible distance between pairs of distinct codewords .
the _ conjugation _ of @xmath17 is defined by @xmath18 , and the _ trace map _ is defined by @xmath19 .
hermitian trace inner product _ of @xmath13 and @xmath14 , with @xmath20 , is given by @xmath21 .
we define _ the dual of the code _ @xmath6 with respect to the hermitian trace inner product , @xmath22
. then @xmath23 is _ self - orthogonal _ if @xmath24 , and @xmath6 is _ self - dual _ if @xmath25 .
a graph code is an additive code over @xmath1 that has a generator matrix of the form @xmath26 , where @xmath27 is the identity matrix and @xmath28 is the adjacency matrix of a simple undirected graph .
a code is called _ optimum _ if it meets both lower and upper bounds in the code tables , and a _ proposed optimum _
code if it only meets the lower bound in the code tables .
the distribution of a code is the sequence @xmath29 , where @xmath30 is the number of codewords of weight @xmath31 .
the _ weight enumerator _ of the code is the polynomial @xmath32 let us now introduce some concepts and notions from graph theory .
an _ undirected graph _
@xmath33 is a set @xmath34 of vertices together with a collection @xmath35 of edges , where each edge is an unordered pair of vertices .
the vertices @xmath36 and @xmath37 are _ adjacent _ if @xmath38 is an edge .
then @xmath37 is a _
neighbour _ of @xmath39 .
all the neighbours of vertex @xmath39 in graph @xmath28 form the _ neighbourhood _ of @xmath39 , and it is denoted by @xmath40 .
the _ degree of a vertex _
@xmath41 is the number of vertices adjacent to @xmath41 .
a graph is _ k - regular _ if all vertices have the same degree @xmath42 .
the _ adjacency matrix _
@xmath43 of a graph @xmath33 is a symmetric @xmath44-matrix defined as follows : @xmath45 if the @xmath31-th and @xmath46-th vertices are adjacent , and @xmath47 otherwise .
circulant graphs and their various applications are the objects of intensive study in computer science and discrete mathematics , see @xcite .
recently , monakhova published a survey paper on this subject , see @xcite .
let @xmath48 be a set of integers such that @xmath49 and let the vertices of an @xmath5-vertex graph be labelled @xmath50 .
then the _ ciculant graph _
@xmath51 has @xmath52 adjacent to each vertex @xmath31 .
a _ circulant matrix _ is obtained by taking an arbitrary first row , and shifting it cyclically one position to the right in order to obtain successive rows .
we say that a circulant matrix is generated by its first row .
formally , if the first row of an @xmath5-by-@xmath5 circualant matrix is @xmath53 , then the @xmath54 element is @xmath55 , where subscripts are taken modulo @xmath5 .
the term circulant graph arises from the fact that the adjacency matrix for such a graph is a circulant matrix .
for example , figure [ fig1 ] shows the circulant graph @xmath56 . [ 0.6 ] + figure 1 : the circulant graph @xmath56 [ fig1 ] in @xmath57 , tonchev @xcite first set up a relationship between a linear binary code and the adjacency matrix of an undirected graph . in @xmath0 , danielsen and parker
@xcite proved that every self - dual additive code over @xmath1 is equivalent to a graph code . recently , finding optimum codes from graphs has received a wide attention of many researchers , see @xcite . in @xcite ,
danielsen and parker showed that two codes are equivalent if and only if the corresponding graphs are equivalent with respect to local complementation and graph isomorphism .
they used these facts to classify all codes of length up to @xmath58 . in 2012
, danielsen @xcite focused his attention on additive codes over @xmath59 and transformed the problem of code equivalence into a problem of graph isomorphism . by an extension technique , they classified all optimal codes of lengths @xmath60 and @xmath58 .
in fact , a computer search reveals that circulant graph codes usually contain many strong codes , and some of these codes have highly regular graph representations , see @xcite . in @xcite , danielsen obtained some optimum additive codes from circulant graphs in 2005 .
later , varbanov investigated additive circulant graph codes over @xmath1 , see @xcite . in this paper
, we find out some optimum additive codes from some special circulant graphs .
this paper is organized as follows . in section @xmath61
, we propose a new method to find additive optimum codes from circulant graphs .
inspired by the optimum additive codes obtained by danielsen @xcite , we focus on the dense circulant graphs and get some optimum additive codes in section @xmath62 .
in fact , glynn et al .
@xcite obtained an optimum code from a circulant graph , called 5-valent graph .
recall that @xmath63 is a circulant graph of order @xmath58 ; see figure 1 @xmath64 .
[ 0.6 ] + figure 1 : @xmath64 a @xmath65-valent graph @xmath63 ; @xmath66 the circulant graph @xmath67 . let @xmath68 . for the vertex @xmath69 ,
we let @xmath70 . for the vertex @xmath71
, we just rotate the above vertices and edges , that is , we only permit the existence of the set @xmath72 of edges . for each vertex @xmath73
, we can also obtained the sets @xmath74 of edges .
observe that @xmath75 .
the adjacency matrix of the graph @xmath63 is the following circulant matrix .
@xmath76 the above matrix can also be obtained by the following vector @xmath77 observe that this vector just corresponds to the set @xmath78 of edges , which is an expression of the adjacent relation about the vertex @xmath69 .
we conclude that a @xmath65-valent graph can be determined by the edge set @xmath78 , and the adjacency matrix of this graph is determined by the above vector @xmath79 .
furthermore , the matrix @xmath80 is @xmath81 this matrix is also a circulant matrix , which can be obtained by the following vector @xmath82 from the matrix @xmath80 , we can get a graph code @xmath83 . from the code tables , we know that @xmath84 is an optimum additive code over @xmath1 . then @xmath85 and @xmath86 , where @xmath87 is the minimum distance of the code @xmath83 .
the above statement suggests the following method for finding optimum codes .
* step 1 . * given an even integer @xmath5 .
denote by @xmath88 the lower bound of the additive code ( @xmath5,@xmath89 ) over @xmath1 . from code tables , we find the exact value of @xmath88 for the given @xmath5 .
we now construct a circulant graph @xmath90 by a set @xmath78 of edges as follows .
* step 1.1 . *
arrange the vertices from @xmath91 in a circular order .
* step 1.2 . *
determine the set @xmath78 of edges satisfying @xmath92 or @xmath93 , where @xmath94 . if @xmath93 , then @xmath95 if @xmath92 , then @xmath96 * step 2 .
* by the edge set @xmath78 , we write the vector @xmath97 corresponding to @xmath78 . if @xmath93 , then @xmath98 @xmath99 @xmath100 @xmath101 @xmath102 @xmath103@xmath104 @xmath105 if @xmath92 , then @xmath98 @xmath99 @xmath106 @xmath101 @xmath107 @xmath103@xmath104 @xmath105 * step 3 . * change the first component of the vector @xmath97 to @xmath108 .
denote by @xmath109 the new vector .
we generate a circulant matrix @xmath110 from @xmath109 .
@xmath111 * step 4 .
* by algorithm @xmath112 , we obtain the minimum distance @xmath113 of the code @xmath114 and determine whether @xmath115 .
if so , the code @xmath114 is a proposed optimum code .
below is an algorithm ( running in sage ) .
for more details , we refer to @xcite . ' '' '' + * algorithm * 1 : minimum distance of a circulant graph code + ' '' '' + input : the value of @xmath5 , the generator vector @xmath97 of a circulant graph code @xmath116 + objective : the minimum distance of the circulant graph code @xmath116 + 1 .
input the value of @xmath5 , the generator vector @xmath117 ; + 2 .
obtain the generator matrix @xmath118 of the circulant graph code @xmath116 ; + 3 .
get the minimum distance of the circulant graph code @xmath116 .
+ take an example , let @xmath119 and @xmath120 .
+ the algorithm details are stated as follows : + _ program _ : + @xmath121 + @xmath119 ; + @xmath122 $ ] + @xmath123 $ ] for @xmath31 in @xmath124 $ ] for @xmath42 in @xmath125)$ ] ; + @xmath126,s))$ ] ; + @xmath127 $ ] ; + for @xmath42 in @xmath128 $ ] : + @xmath129 for @xmath31 in subsets @xmath130)$ ] ; + @xmath131 $ ] ; + print @xmath132 ; + _ output _ : [ 8 ] + [ 8 , 8 ] + [ 8 , 8 , 10 ] + [ 8 , 8 , 10 , 8 ] + [ 8 , 8 , 10 , 8 , 8 ] + [ 8 , 8 , 10 , 8 , 8 , 8 ] + [ 8 , 8 , 10 , 8 , 8 , 8 , 8 ] + [ 8 , 8 , 10 , 8 , 8 , 8 , 8 , 10 ] + _ result _ : the minimum element of the last array is the minimum distance @xmath133 of the code @xmath134 , + that is , @xmath135 .
+ ' '' '' inspired by the graph code @xmath136 which corresponds to the @xmath65-valent graph , we hope to find out some other optimum additive codes for @xmath119 .
* recall @xmath137 is the lower bound of the additive code @xmath138 over @xmath1 . from code tables ,
we find the exact value of @xmath137 , @xmath139 .
we now construct a circulant graph @xmath67 by a set @xmath78 of edges as follows .
* step 1.1 . *
arrange the vertices from @xmath140 in a circular order .
* step 1.2 . *
determine the set @xmath78 of edges satisfying @xmath141 or @xmath142 , where @xmath94 .
if @xmath143 , then @xmath144 if @xmath145 , then @xmath146 in this case , the circulant graph @xmath67 can be found out ; see figure 1 @xmath66 . * step 2 .
* by the edge set @xmath78 , we write the vector @xmath147 corresponding to @xmath78 . if @xmath143 , then @xmath148 if @xmath145 , then @xmath149 * step 3 .
* change the first component of the vector @xmath147 to @xmath108 .
denote by @xmath150 the new vector .
we generate a circulant matrix @xmath151(@xmath152 ) from @xmath150(@xmath153 ) .
@xmath154 @xmath155 * step 4 . * by algorithm @xmath112 , we obtain the minimum distance @xmath156 .
therefore , both @xmath157 and @xmath158 are proposed optimum codes over @xmath1 .
the weight enumerators of the codes @xmath157 and @xmath158 are @xmath159 @xmath160 applying the above method , we can obtain optimum graph codes over @xmath1 from the first two generator vector . ' '' '' + @xmath5 @xmath16 @xmath88 first row of generator matrix + ' '' '' + @xmath161 @xmath162 @xmath162 + @xmath163 @xmath164 @xmath164 + @xmath161 @xmath162 @xmath162 + @xmath165 @xmath166 @xmath166 + ' '' ''
from the above section , we see that the circulant graphs under our consideration are all relatively sparse .
so one may think that only sparse circulant graphs produce optimum graph codes .
the following fact gives it a negative answer .
danielsen @xcite obtained an optimum additive code @xmath167 from the vector @xmath168 which corresponds to a circulant graph of order @xmath169 such that its adjacent matrix @xmath170 is generated by the first row @xmath171 it is clear that the circulant graph is @xmath172-regular .
note that the order of this graph is @xmath169 , so the degree of each vertex is relatively large , i.e. , it is a relatively dense graph . | in @xmath0 , danielsen and parker @xcite proved that every self - dual additive code over @xmath1 is equivalent to a graph code .
so , graph is an important tool for searching ( proposed ) optimum codes . in this paper
, we introduce a new method of searching ( proposed ) optimum additive codes from circulant graphs .
+ * ams subject classification 2010 * : 94b05 , 05c50 , 05c25 . |
recently , sahni _ et al .
_ used the luminosity distance @xmath0 , valid in spatially flat friedmann - robertson - walker ( frw ) universes , to propose a diagnostic to tell the cosmological constant from evolving dark energy fields @xcite . for universes dominated by cold matter ( cm ) ,
subscript @xmath1 , and dark energy ( de ) , subscript @xmath2 , with constant equation of state , @xmath3 , friedmann s equation takes the form @xmath4 where @xmath5 , and @xmath6 is the normalized hubble function . in writing
this expression it was implicitly assumed that cold dark matter ( cdm ) and de evolved separately .
as usual , @xmath7 with @xmath8 stand for the energy densities of the components in units of the critical density , @xmath9 , and a zero subscript denotes the present - day value of the corresponding quantity .
sahni _ et al . _
@xcite introduced the redshift dependent function @xmath10 which for spatially flat frw cosmologies ( @xmath11 ) reduces to @xmath12 .
it has the interesting feature that for the @xmath13cdm model ( @xmath14 , @xmath15 ) , it yields @xmath16 , whereas for quintessence fields ( @xmath17 ) and phantom fields ( @xmath18 ) it gives @xmath19 and @xmath20 , respectively .
since this diagnostic does not depend on the present values of the density parameters , @xmath21 and @xmath22 , it circumvents the drawbacks brought about by our comparatively poor knowledge of them .
in fact , it is expected to be useful when accurate data of @xmath23 at different redshifts become available .
this seems more feasible than to substantially upgrade our knowledge of @xmath21 and @xmath22 in the near future . at present
, the @xmath13cdm model fits the observational data , within statistical errors , rather well -see , e.g. @xcite . nonetheless if future data come to reveal that @xmath24 , a number of de alternatives should be considered including scenarios in which the de interacts non - gravitationally with other components . here
, we assume that de and cdm , subscript @xmath25 , interact with each other ( whence they do not evolve separately ) but not with baryonic matter , subscript @xmath26 , since interactions with baryons are strongly constrained by experiments @xcite .
this possibility is being actively considered in the literature -see @xcite and references therein- since it is rather natural @xcite and more general than otherwise , it substantially alleviates the coincidence problem ( i.e. , why are the densities of matter and dark energy of the same order precisely today " ) , and agrees with observation . in this paper
we propose several criteria , based on the evolution of the hubble factor , to help discriminate models in which de and cdm interact with each other non - gravitationally from models in which they do not , and to tell quintessence de energy models from phantom models .
clearly , for these criteria to be useful accurate data of @xmath27 are needed .
while the present data exhibit large error bars @xcite the situation may well improve comparatively soon thanks to upcoming and planned experiments . the baryon oscillation spectroscopy survey ( boss ) 5-years project @xcite aims at measuring the absolute cosmic distance scale and expansion rate with percent - level precision at redshifts @xmath28 and @xmath29 using the standard rule furnished by the baryon acoustic oscillations . the forecast precision for @xmath27 at @xmath30 , @xmath31 and @xmath32 is @xmath33 , @xmath34 and @xmath35 , respectively .
data at intermediate and larger redshifts will partly come by adapting to @xmath36 the method proposed by hogan @xcite which makes use of the gravitational radiation emitted by black hole binary inspiral events to determine @xmath37 with a precision better than @xmath38 .
the method may be extended up to @xmath39 redshifts , and even to @xmath40 by observing massive black hole binaries since there will likely be a direct electromagnetic counterpart of the merger @xcite .
an even more promising method , proposed by corasaniti
_ et al . _
@xcite to constrain cosmological parameters , exploits the sandage - loeb test @xcite .
it consists in measuring the redshift of distant objects at two separate times , say @xmath41 and @xmath42 , at the present epoch .
the redshift variation of the source ( subscript @xmath43 ) between these two instants is , at a very good precision , @xmath44 -see eq . ( 5 ) in @xcite .
this expression , re - written as @xmath45 , provides us with the value of the hubble factor at the time the source emitted the electromagnetic signal that is currently arriving at our detectors ( either telescopes or radio - telescopes ) , modulo we know @xmath37 accurately enough .
the authors of @xcite suggested using the absorption lines of the lyman-@xmath46 forest in the redshift interval @xmath47 and @xmath48 .
altogether , encouraged by the prospects of a greatly improved knowledge of the history of @xmath27 we present several criteria that may help telling apart interacting from noninteracting dark energy models .
section ii presents on phenomenological basis some possible expressions of the interaction term .
sections iii and iv introduce various criteria to distinguish interacting from noninteracting de models .
section v focus on the statefinder parameters and compares their usefulness with the criteria based on the hubble history .
finally , section vi summarizes our findings .
since the dark components are assumed to interact with one another also non - gravitationally but not with baryons the conservation equations take the form @xmath49 where @xmath50 denotes the interaction term and the relationship @xmath51 is understood .
it is worth noticing that when @xmath52 -as we shall consider throughout- the coincidence problem is alleviated since @xmath53 decreases more slowly with expansion and @xmath54 more quickly , and the problem may even be solved in full @xcite . on the other hand ,
if @xmath50 were negative , the second law of thermodynamics could be violated @xcite , and the energy densities of cdm or de could become negative at high or low redshifts . in the absence of a fundamental theory for dark energy the quantity @xmath50 can not be derived from first principles .
however , we may guess likely expressions for it by noting that @xmath50 must be small @xcite ( at least lower than @xmath55 ) and depend on the energy densities multiplied by a quantity with units of inverse of time , a rate .
we shall deal with two different possibilities considered in the literature , namely , @xmath56 that the said rate is proportional to the hubble factor , and @xmath57 that it is just a constant @xcite . in case @xmath56 , we have that @xmath58 . by power law expanding this function and retaining just the first term one follows @xmath59 .
given the lack of information about the coupling , it appears advisable to work with just one parameter rather than two .
thus , three different choices arise , namely , @xmath60 , @xmath61 , and @xmath62 . it is worth mentioning that expressions of this type can be obtained from the scalar - tensor theory of gravity developed by kaloper _
@xcite -see e.g. @xcite . in the next section
we shall consider successively three expressions for @xmath50 with @xmath63 , while keeping @xmath64 constant throughout , and look for diagnostics -based in the history of @xmath27 that may tell interacting from noninteracting models . in case
@xmath57 we have @xmath65 and proceeding as before we write @xmath66 where the @xmath67 ( @xmath68 ) quantities denote constant rates .
this expression was inspired in reheating models @xcite , curvaton decay @xcite , and decay of dm into radiation @xcite . regrettably
, one of the energy densities ( @xmath69 or @xmath54 ) becomes negative either at early or late times @xcite and , generally , it does not solve the coincidence problem .
nevertheless , we will briefly consider it in the section after next .
we begin by assuming that the interaction term takes the form @xmath71 where the factor @xmath72 was introduced for mathematical convenience . one way to explain the present small value of the cosmological constant is to assume a transfer of energy from the quantum vacuum to particles and/or radiation -i.e .
, a decaying cosmological constant ( dcc ) . along the years
a diversity of phenomenological models have been proposed , many of which are summarized in ref .
@xcite . if the quantum vacuum ( @xmath15 ) slowly decays into non - baryonic cdm particles , integration of eqs .
( [ conserv1a])-([conserv1c ] ) yields @xmath73 this kind of models are not purely phenomenological any more for they have received some persuasive backings from the theoretical side @xcite , which makes their consideration all the more interesting . in particular , by means of the renormalization group theory the authors of @xcite arrive to a simple expression for @xmath74 ( @xmath75 in their notation ) in terms of the masses of the super - heavy sub - planckian particles -see their eq .
a rather similar expression was found by removing the flat space - time contribution to the vacuum energy of quantum fields in homogeneous and isotropic universes by employing a technique similar to that used in the computation of the arnowitt - deser - misner mass -see eq .
( 39 ) in ref .
. moreover , consistency both with primordial nucleosynthesis and large scale structure formation leads to the constraint @xmath76 @xcite , @xcite .
recently , dcc models were revisited by basilakos _
_ @xcite and , for some specific choices of @xmath77 , confronted with observational data from supernovae type ia , baryon acoustic oscillations , the cosmic microwave background shift parameter , and the growth rate of galaxy clustering .
as said above , the coincidence problem gets alleviated since @xmath78 < 1 \ , , \ ] ] i.e. , currently the ratio @xmath79 varies more slowly than in the @xmath13cdm model .
this also holds true for the other interaction terms considered below . using friedmann s equation , @xmath80 $ ] , we can write @xmath81\ , , \label{evlambda}\ ] ] where @xmath82
. obviously , in this case @xmath83 with @xmath84 , but rather @xmath85 thus , the @xmath86 criterion would mistake a dcc model by a quintessence field ( phantom field ) if applied at low redshifts ( high redshifts ) .
similar situations occur if the said criterion is used on models considered below in this paper .
the first and second derivatives of @xmath87 with respect to @xmath88 are @xmath89 , and @xmath90 , respectively .
therefore , for large redshifts the first derivative tends to @xmath91 from above .
likewise , @xmath87 is concave ( i.e. , @xmath92 ) .
by contrast , for noninteracting dark energy fields with @xmath64 constant -see eq .
( [ friedmann1])- one has , @xmath93 . for large redshifts
this expression tends to @xmath21 from above if de is of quintessence type and from below if it is phantom . in view that , observationally ,
@xmath91 and @xmath21 are separated by a non small gap ( the latter might be about six or seven times larger than the former ) this may serve discriminate dcc models from noninteracting quintessence and phantom models .
one may think that we would have arrived to the same result just just by dividing @xmath87 by @xmath88 ; i.e. , @xmath94 .
however , this is not true because of the factor @xmath95 in the second term of the derivative which makes the latter tend to @xmath21 faster , and it is key to tell whether the de is quintessence s or phantom s type . as a corollary
, we may say that the sign of the second derivative , may discriminate dcc from noninteracting phantom models , the former being negative and the latter positive , @xmath96 ; only that this criterion can be used also at smaller redshift than the preceding one .
likewise , by measuring @xmath27 at two separated redshifts , @xmath97 and @xmath98 with @xmath99 , they may be discriminated . indeed , for noninteracting phantom models the quantity
@xmath100 is @xmath101\ , + \
, x_{2}^{3(1+w)}\ , - \ , x_{1}^{3(1+w)}\ , , \label{delta1}\ ] ] and it can be negative if the redshifts are sufficiently apart .
by contrast , the corresponding quantity for dcc models , @xmath102 \ , , \label{delta2}\ ] ] will always be positive - definite .
let us assume that , thanks to extra information , we come to know that the right model must be a dcc one .
note that to discriminate which one ( i.e. , its @xmath74 value ) neither direct use of last expression nor of eq .
( [ evlambda ] ) will prove very useful , given the comparatively wide uncertainty in the value of @xmath103 .
however , the following strategy may work .
consider two competing dcc models ( say a and b ) and recall that @xmath91 is reasonably well determined from primordial nucleosynthesis ( @xmath104 , see , e.g. ref .
@xcite ) together with the present value of the hubble factor .
the ratio @xmath105_{a}}{\left[\delta(x_{1 } , x_{2 } ) \ , -\ , \omega_{b0}\ , ( x^{3}_{2}-x^{3}_{1 } ) \right]_{b } } = \left(\frac{1-\epsilon_{b}}{1-\epsilon_{a}}\right)\ , \frac{x^{3(1-\epsilon_{a})}_{2 } -x^{3(1-\epsilon_{a})}_{1}}{x^{3(1-\epsilon_{b})}_{2 } -x^{3(1-\epsilon_{b})}_{1 } } \label{ratio1}\ ] ] depends on @xmath106 and @xmath107 only .
this will enable us to determine which dcc model , a or b , fits better the @xmath27 data .
it should be noted that , in this regard , it looks more advantageous to use this other ratio @xmath108 since it can constrain @xmath74 more easily . at any rate , aside from this specific instance , the @xmath86 function ( eq . ( [ om ] ) ) is not much useful when dealing with interacting models since , in these models , cdm and baryonic matter redshift at different rates with expansion whereby the current fractional densities , @xmath91 and @xmath103 , enter the expression for @xmath109 multiplied by different powers of @xmath110 .
this explains the restricted application range of the said function and we will not resort to it any more in this paper .
for the more general case @xmath111 , one has @xmath112 \ , .
\label{intgrde}\end{aligned}\ ] ] thereby @xmath113\ , , \label{ede1}\ ] ] being @xmath114\ , , \label{fderivative}\end{aligned}\ ] ] and @xmath115 .
\label{sderivative}\end{aligned}\ ] ] in view of the hypothesis @xmath116 and the fact that observation reveals that @xmath117 , the right hand side of ( [ fderivative ] ) is dominated by its two first terms .
likewise , the first term dominates the right hand side of ( [ sderivative ] ) .
therefore , for not large redshifts ( i.e. , @xmath118 ) the normalized hubble function , @xmath87 , is growing and concave . in consequence
, its shape will not tell a decaying energy field ( with @xmath119 ) from dcc models or from noninteracting quintessence or phantom models with constant equation of state .
however , discrimination will likely come at higher redshifts since , as before , the first derivative may tell interacting from noninteracting cosmologies depending on whether @xmath120 tends to @xmath91 or @xmath103 at large redshifts . figure [ fig : sderiv ] shows the evolution of the second derivative of @xmath87 , as given by eq .
( [ sderivative ] ) , for the best fit values of @xmath121 and @xmath74 of model iii of he _
@xcite ( dotted line ) as constrained by data from supernovae , baryon acoustic oscillations , cosmic microwave background , and the value of the hubble constant , @xmath37 . also shown is the noninteracting case ( solid line ) .
the fact that both lines nearly overlap emphasizes the need for accurate data of @xmath27 . in terms of @xmath88 for the best fit values , @xmath122 , @xmath123 , with @xmath124 , and @xmath125 of model iii , @xmath126 , of ref .
solid line : same as before but with @xmath127 ( no interaction ) . ] in the present case @xmath128\ , + \ , \omega_{d0}\ , \left [ x^{3(1+w)}_{2 } - x^{3(1+w)}_{1}\right ] \nonumber \\ & + & \ , \left(\frac{\epsilon}{\epsilon+w}\right ) \omega_{c0}\ , \left[x^{3(1+w)}_{2 } - x^{3(1+w)}_{1 } \ -\ , x^{3(1-\epsilon)}_{2 } \,+\ , x^{3(1-\epsilon)}_{1}\right ] \ , .
\label{delta3}\end{aligned}\ ] ] for @xmath129 and @xmath130 the two last terms are subdominants , and the ratio defined by the first equality in ( [ ratio1 ] ) comes to be @xmath131 thus , thanks to the factor @xmath132 on the right hand side of eq .
( [ ratio1 ] ) we hope discriminate dcc models from interacting dark energy models with @xmath121 close to the cosmological constant value , @xmath133 , and also discriminate within models featuring the interaction ( [ q1 ] ) .
+ we now assume that the interaction term takes the form @xmath135 which may be key to explain the non - vanishing temperature of sterile neutrinos @xcite , @xcite .
then , the conservation equations ( [ conserv1a])-([conserv1b ] ) integrate to @xmath136\,x^{3 } , \nonumber\\ \rho_{d } & = & \rho_{d0}\ , x^{3(1+w+\epsilon ) } \ , .
\label{intgrde2}\end{aligned}\ ] ] hence , @xmath137 \ , .
\label{moresimply}\ ] ] for @xmath15 ( i.e. , dcc with interaction given by eq .
( [ q12 ] ) ) , it reduces to @xmath138 \
, , \label{evlambda1}\ ] ] which is to be compared with ( [ evlambda ] ) .
again , @xmath87 is growing and concave whence its graph will tell neither dcc models with interaction term given by ( [ q12 ] ) from dcc models with interaction term given by ( [ q1 ] ) nor these models from noninteracting quintessence models with @xmath64 constant . expression ( [ evlambda1 ] ) will help constrain @xmath74 without need of knowing @xmath139 since the ratio @xmath140_{x = x_{1}}}{[e^{2}(x)-x^{3}]_{x = x_{2 } } } = \frac{x_{1}^{3}-x_{1}^{3\epsilon}}{x_{2}^{3}-x_{2}^{3\epsilon } } \label{ratio4}\ ] ] does not depend on that quantity . unlike the two previous cases , as inspection of the right hand side of eq .
( [ moresimply ] ) shows , the behavior of the derivative @xmath120 at high redshifts will not discriminate interacting from noninteracting cosmological models .
for the discrimination to be possible @xmath74 should take unrealistic high values . by deriving twice ( [ moresimply ] )
we get @xmath141 when this latter quantity gets accurately determined , we will be able to set useful constraints on @xmath121 and @xmath74 since its ratio for two decaying dark energy ( dde ) models ( say @xmath142 and @xmath143 ) does not depend on the @xmath144 parameters , @xmath145 once again @xmath87 , given by ( [ moresimply ] ) , is a growing function of redshift , concave for quintessence fields and convex for phantom fields . accordingly
, the shape of @xmath87 can tell decaying quintessence fields from decaying phantom fields but not from noninteracting de fields .
figure [ fig : sderiv2 ] depicts the evolution of the second derivative of @xmath87 , as given by eq .
( [ sderivative2 ] ) , for the best fit values of @xmath121 and @xmath74 of model ii of ref .
@xcite ( dotted line ) as constrained by observational data .
the solid line corresponds to the noninteracting case ( solid line ) . as it is apparent the lower the redshift , the easier the discrimination . in terms of @xmath88 for the best fit values , @xmath146 , @xmath147 , with @xmath148 of model ii , @xmath149 , of ref .
solid line : same as before but with @xmath150 ( no interaction ) .
] assuming the interaction term @xmath152 widely considered in the literature -see , e.g. @xcite- , equations ( [ conserv1a])-([conserv1c ] ) integrate to @xmath153\ , x^{\gamma_1 } \
, , \label{intgrde2}\end{aligned}\ ] ] respectively
. here @xmath154\,,\,\;\;\;\;\gamma_2=\frac{3}{2}\,[2+w-\sqrt{w(w+4\epsilon)}]\,,\ ] ] @xmath155\,,\;\,\;\ , c_2=\frac{1}{2\,b}\left[(a+b)\,\rho_{c0}+2\,\epsilon\,\rho_{d0}\right],\ ] ] @xmath156 as in the two previous cases , the three energy densities remain semi - positive definite for all @xmath110 .
therefore , @xmath157 \ , -\omega_{c_2}\,(a - b ) \ , x^{-3b } \right\}\ , x^{\gamma_1}\ , + \omega_{c_2 } \ , x^{\gamma_{2 } } \ , , \label{ede13}\ ] ] where @xmath158\ , , \;\,\;\,\mbox{and}\;\;\;\;\;\;\ ; \omega_{c_2}=\frac{1}{2\,b}\left[\omega_{c0}\,(a+b)+2\,\epsilon\,\omega_{d0}\right]\ , , \ ] ] also stays non - negative as it should .
further , @xmath109 is a growing function of @xmath88 , concave for quintessence and convex for phantom -see fig .
[ figure1]- which will help discriminate phantom from quintessence de in interacting models with @xmath50 given by eq .
( [ q13 ] ) .
however , as is apparent , for @xmath159 the graphs show a nearly straight line behavior whereby one must focus on redshifts below , say , @xmath160 .
inspection of the right hand side of ( [ ede13 ] ) shows that for interacting models ( @xmath161 ) the derivative @xmath120 tends to @xmath91 at high redshifts . accordingly
if observation reveals that behavior , it would be suggestive of interaction .
vs. @xmath88 as given by eq .
( [ ede13 ] ) for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 .
notice that only phantom models present convex curves for reasonable values of the parameters . in drawing the graphs we took @xmath162 , @xmath163 , @xmath164 , and @xmath165.,title="fig : " ] vs. @xmath88 as given by eq .
( [ ede13 ] ) for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 .
notice that only phantom models present convex curves for reasonable values of the parameters . in drawing the graphs we took @xmath162 , @xmath163 , @xmath164 , and @xmath165.,title="fig:",width=307 ] figure [
fig : e2 ] shows the evolution of @xmath109 for the best fit value of model iv ( @xmath166 ) of @xcite ( dotted line ) and the values corresponding to the @xmath167 error , @xmath168 ( dashed ) and @xmath169 ( solid ) ; obviously the latter practically coincides with the noninteracting case .
likewise , fig .
[ fig : sderiv3 ] shows the evolution of @xmath170 , for the best fit values of the same model ( dotted line ) and for the noninteracting case ( solid line ) . as is apparent
the graphs practically overlap . in terms of @xmath88 for the best fit values ,
@xmath146 , @xmath124 , and @xmath125 , @xmath171 , ( dotted line ) and the @xmath74 values corresponding to the @xmath167 errors , @xmath172 ( dashed ) and @xmath173 ( solid ) of model iv , @xmath174 , of ref .
we just depict the @xmath175 interval because the difference between the interacting and noninteracting case augments with redshift.,width=480 ] for the best fit values , @xmath146 , @xmath171 , with @xmath124 , and @xmath125 , of model iv , @xmath176 , of ref .
solid line : same as before but with @xmath177 ( no interaction).,width=480 ]
in case @xmath57 -i.e . , for @xmath50 given by eq .
( [ qroy1])- analytical solution can be found , at most , for one of the conservation equations , either ( [ conserv1b ] ) or ( [ conserv1c ] ) , and this only if @xmath179 , or @xmath180 .
hence , at least one these equations must be solved numerically whereby no analytical expression exists for the hubble factor .
this limits the usefulness of @xmath181 to set diagnostics on the nature of de .
let us consider first the case @xmath180 and @xmath182 . in this instance
, the currently phase of de domination would be a transient one , the smaller @xmath183 , the longer the de domination .
left and right panels of panel of fig .
[ figure2 ] depict the evolution of @xmath109 and its second derivative in terms of @xmath88 for @xmath184 . vs. @xmath88 assuming @xmath185 with @xmath184 for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 . in drawing the graphs we took @xmath162 , @xmath163 , and @xmath164.,title="fig : " ] vs. @xmath88 assuming @xmath185 with @xmath184 for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 . in drawing the graphs we took @xmath162 , @xmath163 , and @xmath164.,title="fig:",width=307 ] as it can be seen , @xmath186 presents a very fast increase ( decrease ) for phantom and dcc ( for quintessence ) at small redshifts ( @xmath187 . this feature is absent in noninteracting models as well as in models whose interaction term is of the general type considered in the previous section . vs. @xmath88 assuming @xmath188 with @xmath189 for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 . in drawing the graphs we took @xmath162 , @xmath163 , @xmath164.,title="fig : " ] vs. @xmath88 assuming @xmath188 with @xmath189 for three values of @xmath121 .
right panel : graphs of the second derivative of @xmath109 with respect to @xmath88 for the same three values of @xmath121 . in drawing the graphs we took @xmath162 , @xmath163 , @xmath164.,title="fig:",width=307 ] as a second example ,
let us consider the interaction term given by eq .
( [ qroy1 ] ) with @xmath190 . as can be seen in fig .
[ figure3 ] , the behavior of the second derivative , @xmath186 , at small redshifts is very similar to the previous case whence we can not tell one interaction from the other .
accordingly , an initial steep slope followed by an abrupt change to a much milder slope appears to be a general characteristic of interacting models in which @xmath50 is proportional to the densities of dm or de or to a combination of them .
we hope this will serve to discriminate this set of interactions from the rest .
years ago sahni _ et al .
_ @xcite and alam _
@xcite - based on the dimensionless , geometrical pair of parameters @xmath191 and @xmath192- gave a somewhat similar diagnostic of dark energy .
these are defined in terms of the third temporal derivative of the scale factor , namely , @xmath193 and @xmath194 , where @xmath195 stands for the deceleration parameter .
the usefulness of the said pair stems from the fact that different models are characterized by different graphs in the @xmath196 plane ; in particular , the @xmath13cdm model is associated to the single point @xmath197 .
so , they may help elucidate the right de candidate .
when these parameters are calculated for cosmological models in which de and dm interact with each other it is seen that the interaction first appears in the third derivative of the scale factor @xcite . in particular , for @xmath198 one has @xmath199 more specifically if one wished to resort to the present value of @xmath191 to discriminate the @xmath13cdm model from the dcc model ( @xmath200 , @xmath201 ) of sub - section iii.a , one would find @xmath202 .
if , instead , the interaction is governed by eq .
( [ qroy1 ] ) one obtains @xmath203 this suggests that once the @xmath144 parameters come to be accurately determined at different redshifts , the statefinder parameters will not only tell whether the interaction really exists but also will help discern the expression for @xmath50 .
however , for the time being , the wide uncertainties about these seriously restrict the utility of the @xmath204 pair in this respect . altogether ,
the diagnostics founded on the @xmath109 and @xmath205 functions appear more advantageous since the interaction parameters ( i.e. , the @xmath74 and @xmath67 quantities ) , enter the hubble factor upon integration ( analytical or numerical ) of the conservation equations ( [ conserv1a])-([conserv1c ] ) .
interacting models of dark energy are well motivated , substantially alleviate the coincidence problem , and show compatibility with observation . thus the question arises , can they be discriminated from noninteracting models ? " for the time being , the answer is not given our still imperfect knowledge of basic cosmological parameters , such as the hubble factor .
the latter is key in constraining models with observational data from baryon acoustic oscillations , the r - shift parameter of cosmic background radiation , the matter growth parameter , etc .
nevertheless , the situation is to improve greatly in the not far future thanks to a variety of upcoming and planed experiments . to extract the invaluable information about the nature of dark energy encrypted in the history of the hubble factor
is does not suffice to have abundant , accurate data of @xmath27 .
one must first set up simple and practical criteria to be used once these data become available .
we proposed several criteria , of notable mathematical simplicity , based on an accurate knowledge of the said history -something we reasonably hope to have at our disposal not very soon but , at least , in the foreseeable future .
specifically if @xmath206 results positive ( negative ) , then the dark energy is of phantom type ( either quintessence or a decaying cosmological constant ) .
only if the de is not - decaying cosmological constant , the graph of @xmath109 vs. @xmath88 will yield a straight line . to discern whether de is interacting ( with @xmath207 ) or not the behavior of @xmath208 at high redshifts must be studied .
if this derivative tends to the present value of the fractional baryon density , @xmath91 , then there will be grounds to believe that is it interacting with the interaction term @xmath50 given either by eq .
( [ q1 ] ) or ( [ q13 ] ) .
if it tends to @xmath21 , then we may strongly suspect that it is not interacting ; and , in particular , if it coincides with @xmath21 independently of redshift , then it will most likely be a conserved cosmological constant -see eq .
( [ friedmann1 ] ) .
finally , if the redshift function @xmath120 does not tend to any of these two quantities , we may conclude that either de is interacting with @xmath50 given by eq .
( [ q12 ] ) or some other law not considered here , or that in reality @xmath121 is not a constant but a function of redshift .
obviously , this would complicate matters very much because it would call for the introduction of further parameters in the analysis -something that we defer to a future work .
if the plot of @xmath170 vs. @xmath88 exhibits a very steep slope at redshifts much smaller than unity , we may conclude that the interaction obeys eq .
( [ qroy1 ] ) ( with @xmath209 and @xmath210 non - negative ) .
its absence excludes such interaction , but it is very hard to specify whether the slope is steep enough . on the other hand , since this criterion deals with the slope of the second derivative extremely accurate data of @xmath181 at these redshifts must be employed .
the statefinder parameters @xcite will be very useful to rule out cosmological models , and tell apart different interacting models , but first the histories of @xmath211 and @xmath212 must be accurately established in a model - independent manner .
this looks more challenging than obtaining an accurate history of the hubble factor .
since the interaction alters the abundance of dark matter in the past two additional ways to ascertain if the de is interacting or not can be implemented , namely , to examine the behavior of the growth factor , and to study the weak lensing power spectrum . if -as we assumed throughout- the de decays into dark matter , then it must have been less dark matter in the past than in the corresponding noninteracting model with the same @xmath121 and , consequently , the growth factor must be smaller . on the other hand , this would imply a decline of the lensing spectrum ( the opposite would hold true if dm decayed into de ) .
both effects have been considered by the authors of ref .
however , they do not take @xmath121 constant but assume the chevalier - polarski - linder ansatz @xcite with very specific values for the parameters entering that expression .
this is why comparison with our work does not seem straightforward . at any rate , both studies may be seen as complementary . while the criteria introduced in this paper should be helpful in deciphering the nature of de they should be used alongside the analysis and interpretation of data coming from a variety of sources : supernovae type ia , cosmic microwave background radiation ( cmbr ) , matter power spectrum , baryon acoustic oscillations , weak lensing , and evolution of galaxy clusters .
the latter lies at the core of recent studies about the dynamical equilibrium of the clusters which seem to favor a decay of de into dm over other possible scenarios @xcite .
in particular , a promising way to test the existence of the interaction is to consider its impact on the integrated sachs - wolfe effect .
since the former alters the evolution rate of gravitational potentials it must affect the frequency shift experienced by cmbr photons crossing collapsing structures .
again , studies of this kind hint at the existence of the interaction with de decaying into dm @xcite .
we restricted ourselves to constant @xmath121 and vanishing spatial curvature . while this automatically limits the scope of our work we feel it does not do it seriously , at least no more than otherwise since it would introduce additional unknown parameters that would compromise the said scope in other ways .
99 v. sahni , a. shafieloo , and a.a .
starobinsky , phys .
d * 78 * , 103502 ( 2008 ) .
p. serra , a. cooray , d.e .
holz , a. melchiorri , s. pandolfi , and d. sarkar , phys .
d * 80 * , 121302(r ) ( 2009 ) .
peebles and b. ratra , rev .
phys . * 75 * , 559 ( 2003 ) ; k. hagiwara _ et al .
d * 66 * , 010001(2002 ) . c. wetterich , nucl.phys .
b * 302 * , 668 ( 1988 ) ; w. zimdahl , d. pavn , and l.p .
chimento , phys .
b * 521 * , 133 ( 2001 ) ; l. amendola and d. tocchini - valentini , phys .
d * 66 * , 043528 ( 2002 ) ; g. farrar and p.j.e .
peebles , astrophys .
j * 604 * , 1 ( 2004 ) ; s. das , p.s .
corasaniti , and j. khoury , phys .
d * 73 * , 083509 ( 2006 ) ; s. del campo , r. herrera , and d. pavn , phys .
d * 74 * , 023501 ( 2006 ) ; r. horvat , phys .
b * 648 * , 347 ( 2007 ) ; r. manini and s. bonometto , jcap 06(2007)020 ; z .- k .
guo , n. ohta , and s. tsujikawa , phys .
d. * 76 * , 023508 ( 2007 ) ; f.e.m .
costa , j.s .
alcaniz , and j.m.f .
maia , phys .
d * 77 * , 083516 ( 2008 ) ; g. izquierdo and d. pavn , phys .
b * 688 * , 115 ( 2010 ) ; f. atrio - barandela and d. pavn , interacting dark energy " in _ dark energy - current advances and ideas _ , edited by j.r .
choi ( research signpost , trivandrum , kerala , india , 2010 ) ; e. abdalla , r. abramo , and j.c.c .
de souza , phys .
d * 28 * , 023508 ( 2010 ) .
a.m. polyakov , nucl.phys .
b * 797 * , 199 ( 2008 ) .
a.m. polyakov , nucl.phys .
b * 834 * , 316 ( 2010 ) .
j. simon , l. verde and r. jimnez , phys .
d * 71 * , 123001 ( 2005 ) .
daly , s.g .
djorgovski , k.a .
freeman , m. mory , c.p .
odea , p. kharb , and s. baum , astrophys .
j. * 677 * , 1 ( 2008 ) .
e. gaztaaga , a. cabr , and l. hui , mon . not .
* 399 * , 1663 ( 2009 ) .
d. schlegel , m. white , and d. eisenstein , arxiv:0902.4680 [ astro-ph.co ] . c. hogan , phys .
d * 77 * , 043512 ( 2008 ) . c. hogan , personal communication .
corasaniti , d. huterer , and a. melchiorri , phys .
d * 75 * , 062001 ( 2007 ) .
a. loeb , astrophys .
j. * 499 * , l111 ( 1998 ) .
s. del campo , r. herrera , and d. pavn , phys .
d * 78 * , 021302(r ) ( 2008 ) ; _ ibid _ jcap 01(2009)020 .
d. pavn and b. wang , gen .
* 41 * , 1 ( 2009 ) .
if it were large and positive , dm would dominate the expansion today . on the other hand ,
if @xmath50 were large and negative , the universe would have been dominated by de practically from the outset and galaxies would not have come into existence .
g. caldera - cabral , r. maartens , and l.a .
urea - lpez , phys .
d * 79 * , 063518 ( 2009 ) . n. kaloper and k.a .
olive , phys .
d * 57 * , 811 ( 1988 ) .
r. curbelo , t. gonzlez , and i. quirs , class .
quantum grav .
* 23 * , 1585 ( 2006 ) .
h. zhang and z .- h .
zhu , phys .
d * 73 * , 043518 ( 2006 ) .
turner , phys .
d * 28 * , 1243 ( 1983 ) .
malik , d. wands , and c. ungarelli , phys .
d * 67 * , 063516 ( 2003 ) .
r. cen , astrophys .
j. * 546 * , l77 ( 2001 ) .
overduin and f.i .
cooperstock , phys .
d * 58 * , 043506 ( 1998 ) .
i. l. shapiro , j. sol , c. espaa- bonet , and ruiz - lapuente , phys .
b * 574 * , 149 ( 2003 ) . c. espaa - bonet , p. ruiz - lapuente , i.l .
shapiro , andj .
sol , jcap 02(2004)006 .
p. wang and x.h .
meng , class .
quantum grav .
* 22 * , 283 ( 2005 ) .
m. maggiore , arxiv:1004.1782v2 [ astro-ph.co ] .
n. bilic , arxiv:1004.4984v2 [ hep - th ] .
fabris , i.l .
shapiro , and j. sol , jcap 02(2007)016 .
s. basilakos , m. plionis , and j. sol , phys .
d * 80 * , 083511 ( 2009 ) .
tasi lectures on dark matter " , arxiv : astro - ph/0301505 .
j .- h . he , b. wang , and e. abdalla , testing the interaction between dark energy and dark matter via latest observations " , arxiv:1012.3904 . s. hansen , j. lesgourgues , s. pastor , and j. silk , mon . not .
333 * , 544 ( 2002 ) .
j. zhou , bin wang , pavn , and e. abdalla , mod .
a * 24 * , 1689 ( 2009 ) .
chimento , a.s .
jakubi , d. pavn , and w. zimdahl , phys .
d * 67 * , 083513 ( 2003 ) ; b. wang _ et al .
b * 637 * , 357 ( 2006 ) ; g. olivares , f. atrio - barandela , and d. pavn , phys .
d * 77 * , 063513 ( 2008 ) .
v. sahni , t.d .
saini , a.a .
starobinsky , and u. alam , jetp lett . *
77 * , 201 ( 2003 ) .
u. alam , v. sahni , t.d .
saini , and a.a .
starobinsky , mon . not .
. soc . * 344 * , 1057 ( 2003 ) .
w. zimdahl and d. pavn , gen .
. grav . * 36 * , 1483 ( 2004 ) .
g. caldera - cabral , r. maartens , and b.m .
schaefer , jcap 07(2009)027 .
m. chevallier and d. polarski , int . j. mod .
d * 10 * , 213 ( 2001 ) ; e.v .
linder , phys .
lett . * 90 * , 091301 ( 2003 ) .
e. abdalla , l.r .
abramo , l. sodr jr . , and b. wang , phys .
b * 673 * , 107 ( 2009 ) .
g. olivares , f. atrio - barandela and d. pavn , phys .
d * 77 * , 103520 ( 2008 ) .
he , b. wang , and p. zhang , phys .
d * 80 * , 063530 ( 2009 ) . | the two dominant components of the cosmic budget today , pressureles matter and dark energy , may or may not be interacting with each other . currently , both possibilities appear compatible with observational data .
we propose several criteria based on the history of the hubble factor that can help discern whether they are interacting and whether dark energy is phantom or quintessence in nature . |
the kepler mission was launched by nasa in 2009 with the driving aim of discovering earth - like planets orbiting other stars ( devore et al . , 2009 ) .
a number of prominent us research institutes are involved in the data recovery and reduction , but the ames research center , in particular , has a central developmental role .
the kepler satellite has continuously surveyed a selected area of about 10@xmath010 degrees in the cygnus - lyra region of the galactic field to determine the proportion of stars , particularly of main sequence type , showing planetary transits photometrically .
it could be reasonably anticipated that this proportion should not be extremely small ( budding et al . , 2005 ) .
in fact , from examination of the @xmath1150000 stars in the target list of the kepler mission it can be asserted that the fraction of stars showing likely planetary transits is a few per cent .
no particular requirement for host star composition appears required for planetary formation ( buchave et al . , 2012 ) .
kepler s first results were announced in january 2010 , with the early planets having relatively short periods . at the time of writing , more than 3000
planetary transits have been recorded and analyzed . mostly these are attributed to planets larger than the earth , although about 10% of candidates hitherto are of a size comparable to that of the earth .
the majority of known examples are smaller than jupiter , although around 10 percent are of about the same size or larger . about 5%
have been located in the ` habitable zones ' of their parent stars .
nasa announced the positive identification of earth - sized planets towards the end of 2011 , and an early survey of the mission s findings was that of borucki et al .
( 2011b ) .
it should also be noted that a fair proportion of initially announced planet finds , perhaps more than @xmath130% have since been marked as false positives ( matijevic et al . , 2012 ; bryson et al . , 2013 ) .
the kepler telescope has a 1.4 m diameter main mirror with fast optics enabling its relatively wide diameter field of view .
there are 21 separate ccd pairs , arranged in a 5@xmath05 square array with the corner elements missing .
photometric accuracy is related to the position of a stellar image on this array , but for a mean flux count of about 186000 per minute for a mag 12.0 star ( cf .
table [ tab : no2 ] , below ) a measuring standard deviation equivalent to about 383 ppm was estimated by the kepler mission engineers ( gilliland et al . , 2011 ) .
this is close to our finding from the best fitting models pursued in this paper , although certain examples appear to show a wider deviation , for reasons we consider later .
software for the primary photometric analysis used to produce the kepler mission s findings so far has been developed from , or related to , modeling procedures for eclipsing binary stars .
a primary aim of the present study is to check the parameters of the selected planetary candidates and host stars found in the nasa exoplanet archive ( nea ) ( which for the candidates we have selected are the values found by batalha et al .
( 2013 ) using alternative , or independent , curve - fitting techniques .
given the open access to the nea , it is useful to compare different methods and check on the closeness of separate modeling results in a way that will add to general confidence on the analysis programs .
regarding exoplanet photometry and its analysis , burrows ( 2012 ) noted the importance of continued awareness of technical developments , while referring also to an established background , citing the well - known wilson - devinney ( 1971 ) program based on the roche approximation for binary component shapes . at the same time , southworth ( 2012 ) discussed the merits of the ebop program ( nelson & davis , 1972 ; etzel , 1980 ; milone , 1993 ) , when developed as a robust and versatile tool for planetary transits .
classical formulae applying to the eclipses of spherical bodies , but directed specifically to the star and planet configuration , were presented in a somewhat generalized form by mandel and agol ( 2002 ) and applied to the hst light curve of hd209458 , while kang et al .
( 2012 ) adopted a very simple ` box - car ' fitting function , but tailored for fast application to large numbers of light curves with the purpose of statistical analysis in mind .
other useful points relevant to transiting planet photometry were noted by seager and malln - ornelas ( 2003 ) .
specifically , in this article , we report on our modifications of the previously produced gui - based curvefit close binary system analysis program ( zeilik et al . , 1988 ) to the form winkepler and its application to 16 selected planetary candidate light curves from the nea . a @xmath2-minimizing algorithm ( budding & najim , 1980 ; budding & zeilik , 1987 ; budding 1993 ) is built into winkepler .
the ` best - fit ' model corresponds to the least value of @xmath2 , defined as @xmath3 ( bevington , 1969 ) , where @xmath4 and @xmath5 are the observed and calculated light levels at a particular phase and @xmath6 is the error estimate for the measured values of @xmath4 .
the nea provides such estimates in the archive , and we discuss later how this may be checked .
the program contains some user - applied options controlling the approach to the minimum , but , broadly , this follows a marquardt - levenberg strategy , with occasional search resettings when an advantageous direction for simultaneous parameter improvements is located .
the formal determinacy of the underlying model can be checked from examination of the error matrix in the vicinity of the optimum fitting .
this error matrix ( formed by inversion of the @xmath2 hessian ) should be positive definite for a formally determinate fitting .
the eigenvalues and eigenvectors of the @xmath2 hessian are also evaluated as checks on the degree of interdependence of the selected parameter set .
adequacy of the model entails that the value of @xmath7 , where @xmath8 is the number of degrees of freedom of the data - set , should be acceptably close to unity at the optimum .
if this ratio lies outside , say , a 90% confidence limit , which can be judged from standard tables of the @xmath9 distribution , then we can reasonably deduce that there is something amiss .
this may be due to some other effect in the data that the model does not take into account , an inappropriately assigned value of @xmath10 , or failure of the optimization algorithm to converge properly on the best set of parameters .
speed of evaluation of an algebraic form of fitting function is an advantage when exploring a wide range of parameter space and evaluating the parameter error matrix .
relevant experience was gained by the present authors with the mauna kea photometry of hd 209458 ( cf .
budding , rhodes & sullivan , 2005 ) .
one interesting question concerned whether there is sufficient information content in the data to permit independent empirical checks on computed models of limb - darkening with sufficient confidence .
this can provide an additional objective for the present analysis .
winkepler is freeware , and can be downloaded from the website : http://home.comcast.net/.17exmichael.rhodes/. a screenshot of winkepler s main page is shown in fig [ pic : winkepler ] .
winkepler continues the prescription for tidal and rotational distortions of the components as given by curvefit .
this formulation adopts the classical approach to finding the shape of a body distorted by forces associated with rotation and tides by referring to equipotential surfaces .
these surfaces can be described in terms of spherical harmonic series .
clauraut s theorem for bodies in equilibrium ( cf .
e.g. pressley , 2001 ) can then be used to evaluate the coefficients of the terms in the series , with dependence on the structure of the distorted body .
such coefficients ( @xmath11 ) have been evaluated for recent models of stars by inlek and budding ( 2012 ) . for the ` first order ' solution of clairaut s equation , in which the interaction of tides on tides is neglected ,
the structural dependence is summarized by the three tidal terms @xmath12 , @xmath13 , @xmath14 , and the rotational term @xmath15 .
the values of @xmath16 and @xmath17 coefficients in this account of the stellar distortion are derived from numerical integration of a first - order differential equation that kopal ( 1959 ) called radau s equation .
we may therefore refer to the underlying model as the ` radau ' model , to distinguish it from others ( in the ` roche ' model , for example , the foregoing @xmath18s would all revert to unity ) .
the dependent variable in radau s equation is @xmath19 : a logarithmic derivative appearing in clairaut s equation that , in turn , reflects the internal stellar density distribution and is directly related to other parameters ( e.g. @xmath18 , as above , or @xmath20 ) also used in this context .
the star s angular rotation @xmath21 is factored in the above terms by @xmath22 ; @xmath23 corresponding to the default condition ( @xmath24 ) of synchronized rotation .
further details on winkepler its formalism , method and utilization are given by rhodes ( 2013 ) .
other aspects of the fitting procedure emerge in the applications to individual cases reported on later .
the light curve data - sets , consisting mainly of short cadence , normalized pdscap_flux values at given times , were downloaded from the nea .
details on the light curve selections are found in the later discussion sections for each candidate .
input parameters used to set up the analyses were also downloaded from the nea , and are shown in tables [ tab : no1 ] and [ tab : no2 ] ) .
light curves were selected without any particular preconceptions other than studying a reasonably typical group . in choosing later examples on the list , however , we tended to look for smaller sized planets , in order to find out how well determined their parameters could be .
we used only a part of each data - set , that was centered on the minimum and twice the width of the transit duration .
normally , we did no preliminary binning or detrending . for koi 42.01 , however , where initial values of @xmath7 were relatively high and a clear downward trend could be seen in the residuals across the eclipse , we carried out a linear detrending .
the issue is discussed in section 3.12 .
individual transits turn out to have high diagnostic power for the main geometric elements when the data has a reasonably good s / n ratio ( the eclipse depth @xmath110 @xmath0 the scatter level , say , and the data - set including at least @xmath1200 points ) . but a number of the light curves show additional effects apart from just the planetary transits .
those effects are not pursued in any detail in this paper .
still , it is necessary to be aware that pulsational microvariability or maculation may be present in the raw data and affect the quality of the transit fittings .
we discuss specific examples in what follows .
& & & & & & & & & + 1.01 & 0.995 & 1.06 & 5814 & 2.4706132 & 0.116 & 4.38 & 0.036 & 1e3 & 0.56 + 2.01 & 1.660 & 2.71 & 6264 & 2.2047355 & 0.000 & 3.79 & 0.039 & 5e3 & 0.51 + 3.01 & 0.786 & 0.74 & 4766 & 4.8878003 & 0.000 & 4.59 & 0.052 & 8e-5 & 0.69 + 4.01 & 1.610 & 2.60 & 6391 & 3.8493724 & 0.232 & 3.81 & 0.056 & 1e3 & 0.50 + 5.01 & 1.145 & 1.42 & 5861 & 4.7803288 & 0.116 & 4.19 & 0.058 & 1e6 & 0.55 + 7.01 & 1.124 & 1.27 & 5858 & 3.2136641 & 0.085 & 4.28 & 0.044 & 5e5 & 0.54 + 10.01 & 1.140 & 1.56 & 6025 & 3.5224991 & 0.128 & 4.11 & 0.047 & 2e2 & 0.52 + 12.01 & 1.260 & 1.40 & 6419 & 17.8551483 & 0.035 & 4.26 & 0.144 & 1e3 & 0.50 + 13.01 & 2.270 & 2.70 & 8848 & 1.7635877 & 0.141 & 3.94 & 0.038 & 3e3 & 0.44 + 17.01 & 1.138 & 1.08 & 5826 & 3.2346996 & 0.000 & 4.42 & 0.045 & 1e3 & 0.55 + 20.01 & 1.172 & 1.38 & 6011 & 4.3796300 & 0.161 & 4.23 & 0.056 & 4e3 & 0.52 + 42.01 & 1.170 & 1.36 & 6170 & 17.834381 & 0.193 & 4.24 & 0.111 & 1e5 & 0.51 + 72.01 & 0.905 & 1.00 & 5627 & 0.8374903 & 0.812 & 4.39 & 0.017 & 1e6 & 0.50 + 117.01 & 1.183 & 1.18 & 5949 & 14.749102 & 0.092 & 4.37 & 0.125 & 1e6 & 0.53 + 377.01 & 1.053 & 1.01 & 5777 & 19.2739380 & 0.170 & 4.45 & 0.143 & 2e6 & 0.57 + 388.01 & 1.092 & 1.58 & 5707 & 6.1493616 & 0.000 & 4.08 & 0.068 & 4e7 & 0.56 + in table [ tab : no1 ] @xmath25 is the host star s mass ( solar masses ) , @xmath26 stellar radius ( solar radii ) , @xmath27 effective temperature ( k ) , @xmath28 - orbital period ( in days ) , @xmath29 stellar metallicity ( taken to be zero if not given in the source material ) , @xmath30
log@xmath31(surface gravity ) of the star , @xmath32 semi - major axis in aus , @xmath33 ratio of planet to star masses , @xmath34 stellar ( linear ) limb - darkening coefficient .
all these parameters , with the exception of mass ratio and limb - darkening coefficient , are taken from the nea . in making comparisons with parameters published in the nea website and those from elsewhere it
should be noticed that the former were occasionally changed during the period we have been working with the data , and such source information may change again .
this may give a temporary quality to any particular set of detailed numerical results , though this is understandable given the very large ongoing programme of the kepler mission .
at the present time , we are using the nea data for the input information relevant to our selected candidates , corresponding to the values listed by batalha et al ( 2013 ) .
even so , comparison of individual curve - fitting results , at any time , reflects the purpose of independent , alternative analysis mentioned before .
the relative scale of the error estimates is also relevant here .
the nea has tended to list parameters always to the same number of decimal places , whereas it is clear that the light curve quality varies considerably from example to example . in many cases
, we found that 3 decimal digits suffice to quantify parameter estimates realistically .
sometimes this may reduce to 2 or increase to 4 . the preliminary planetary mass ratios ( @xmath33 a formally required fitting input parameter , though usually of negligible effect ) in table [ tab : no1 ] generally come from the archive - listed planet radius , to which a simple scaling from the known mass ratios of solar system planets was applied.50% more than that of the formula . ] for planets of greater size , the mean density was assumed equal to that of jupiter and scaled according to the radius cubed .
we derived this quantity and the other parameter at the right of table [ tab : no1 ] , i.e. the coefficient of limb - darkening in the linear approximation @xmath34 , separately , but using the nea information .
for the mass ratio estimate , in view of its essentially approximate nature and very low influence , one significant digit is appropriate , at least initially .
table [ tab : no2 ] gives further sourced information on the selected examples : the kepler input catalog ( kic ) number , an assigned mean planet temperature @xmath35 , followed by photometric information , such as the reference out - of - transit flux count used in the fittings .
that is a function of the detection system that can be presumed linear with the signal .
the depth of the transit is listed in parts per million ( ppm ) .
a ` kepler ' magnitude has been assigned to accord with the normal astronomical magnitude system , although the wide bandpass of the detection system complicates its interpretation .
we list also a ` poisson factor ' ( 1/@xmath36 ) .
if the mean fluxes of column 4 were actually arrival rates of photons , these numbers would reflect expected errors of measurement .
but the counts are to some extent integrated , scaled and averaged by the detection system s internal electronics
. the appropriate error measures for the light curve data would then be these numbers divided by some quantity @xmath37 , say , whose value we derive later .
the distance @xmath38 , in pc ( column 7 ) , can be estimated from the formula ( budding & demircan , 2007 ) @xmath39 where @xmath40 is the johnson @xmath40-magnitude of the star .
@xmath41 is the visual flux parameter defined by @xmath42 where @xmath43 is the stellar effective temperature and @xmath44 is the bolometric correction , which is reasonably well - known for main sequence stars of a given @xmath43 ( cf . budding & demircan , 2007 ; table 3.1 ) . alternatively , there are accurately known empirical relationships between @xmath41 and the @xmath45 colour of the star that have been developed since the work of barnes & evans ( 1976 )
. we can consider a transformation from the ` kepler ' magnitudes listed in table [ tab : no2 ] to the standard @xmath40 magnitude , since details of the filter used were published by rowe et al .
, ( 2009 ) , but the effects of the wide bandwidth are not small . in this case , a representative conversion formula , using a linearized approximation involving standard magnitudes and colours , becomes inaccurate for colour indices appreciably different from zero . on the other hand , the effective wavelength for the relatively small range of near solar temperatures involved does not vary by much from its average value of about 0.61 @xmath46 m . from the given temperatures and corresponding bolometric corrections
, we can then use the barnes - evans colour correlation in reverse , to derive @xmath45 colours for the candidate stars . using the photometric gradient concept ( budding & demircan , 2007 , eqn 3.13 ) and a derived effective wavelength of the kepler mission filter for a given effective temperature ( budding & demircan , 2007 , eqn 3.30 )
, we can then estimate @xmath47 colours ( where @xmath48 is here the kepler magnitude ) and thence suitable @xmath40 magnitudes to substitute into eqn 1 .
temperature estimates in the kepler input catalogue have been discussed by pinsonneault et al ( 2012 ) , and some of the original nea values appear to be in a process of revision . given also the non - linearity of the relationship of the kepler magnitudes to monochromatic magnitudes , our derived distances , which neglect interstellar extinction , should be regarded as coarse estimates only
. & & & & & & & + 11446443 & 1.01 & 1394 & 4.03265e+05 & 14224 & 11.338 & 250 & 0.00157 + 10666592 & 2.01 & 2303 & 1.03525e+06 & 6686 & 10.463 & 500 & 0.00098 + 10748390 & 3.01 & 794 & 2.88809e+06 & 4281 & 9.147 & 41 & 0.00059 + 3861595 & 4.01 & 1623 & 3.71398e+05 & 1313 & 11.432 & 780 & 0.00164 + 8554498 & 5.01 & 1279 & 3.19394e+05 & 944 & 11.665 & 400 & 0.00177 + 11853905 & 7.01 & 1386 & 1.76070e+05 & 740 & 12.211 & 460 & 0.00238 + 6922244 & 10.01 & 1532 & 4.93732e+04 & 9245 & 13.563 & 1100 & 0.00450 + 5812701 & 12.01 & 876 & 4.19316e+05 & 9271 & 11.353 & 400 & 0.00154 + 9941662 & 13.01 & 3320 & 1.72255e+06 & 4646 & 9.958 & 710 & 0.00076 + 10875245 & 17.01 & 1260 & 7.40452e+04&10586 & 13.000 & 560 & 0.00367 + 11804465 & 20.01 & 1314 & 5.74039e+04&16297 & 13.438 & 930 & 0.00417 + 8866102 & 42.01 & 845 & 2.68825e+06 & 334 & 9.364 & 150 & 0.00061 + 11904151 & 72.01 & 1903 & 5.23104e+05 & 191 & 10.961 & 190 & 0.00138 + 10875245 & 117.01 & 807 & 1.37001e+05 & 495 & 12.487 & 500 & 0.00270 + 3323887 & 377.01 & 678 & 3.61538e+04 & 4887 & 13.803 & 740 & 0.00509 + 3831053 & 388.01 & 1214 & 6.00530e+04 & 362 & 13.644 & 1050 & 0.00408 + of the parameters given in table [ tab : no1 ] , the orbital period is the most accurately known . at first , we allowed optimization of the period for given data - sets covering several transits , but as more data accumulates this becomes less purposeful , as the technique for determining periodicity is different from fitting phased data . using a separate procedure to find only orbital ephemerides
has an information advantage over deriving these parameters with others from a single fitting function .
given the period , if we had also the stellar mass , kepler s third law would yield a fairly reliable separation of planet from star ( since we can reasonably neglect the mass of the planet ) .
but prior estimates of the stellar masses came from indirect evidence that depends on additional assumptions .
these masses can probably only be estimated to within @xmath110 percent of their true values .
this point has an important bearing on subsequent quantification , since the absolute sizes of objects determined from light curve fittings can not be more precisely known than the separation of the components .
although the separation has only a weak dependence ( power 1/3 ) on mass , this uncertainty means that , at best , we could expect only three meaningful digits in the semi - major axes , and usually fewer than three .
the values of @xmath32 listed in table [ tab : no1 ] and calculated in this way can differ from other possible derivations that we discuss shortly . we have retained the third significant digit in table [ tab : no1 ] to allow assessment of such differences .
differences in the absolute separations of star and planet occur when we use the listed surface gravity and radius values to calculate masses from the formula @xmath49 ( masses and radii in solar units ) .
this formula can produce quite unlikely masses for main sequence stars , that are reflected in significantly different @xmath32 values , despite the low sensitivity to mass .
this problem was recognized by batalha et al ( 2013 ) , where the observed gravities were ` corrected ' to those appropriate for main sequence stars of the same effective temperature , composition and mean age of the solar neighbourhood .
the masses derived from such corrected gravities are the masses given in table 1 .
inclusion of the gravity constraint , even corrected , will reflect some natural evolutionary spread about the main sequence in the listed masses . here , however , another issue arises , in that the constraint brings in also the stellar radius .
this can , in principle , be derived from applying another condition , this time using the photometrically fitted relative radius ( @xmath50 in conventional terminology ) , which allows an estimate of the mean density ( @xmath51 ) , since @xmath52 .
the absolute radius then follows from @xmath53 .
note here , though , the sensitivity of @xmath51 to the derived @xmath50 value , as well as the imprecisely known gravity .
besides , we can reasonably expect that @xmath51 would vary by @xmath11 order of magnitude for solar type stars on the main sequence .
the uncertainty thus introduced into the absolute radius must affect masses derived from the gravity constraint pejoratively , and an indication of this was reported by muirhead et al ( 2012 ) .
two different paths were followed to set up our light - curve models : ( 1 ) we estimated preliminary parameters from measurements of light curve segments shown in the nea display facility , aided by the basic formulae given in chapter 7 of budding & demircan ( 2007 ) ; ( 2 ) we adopted trial parameters from those given in the nea .
relevant quantities are listed in table [ tab : no3 ] .
we have generally optimized the curve - fitting using the five parameters @xmath54 , @xmath55 , @xmath50 , @xmath56 and @xmath57 listed in table [ tab : no3 ] .
the first two of these locate the reference points for the phase and flux axes , respectively .
these are of interest for time of mid - transit variation ( ttv ) studies , or absolute flux calibrations .
the other 3 parameters correspond to the usual main geometric ` elements ' .
after each parameter value we give the leading digit of our error estimate as it would affect the last tabulated digit of the parameter value .
thus , the first entry of table [ tab : no3 ] means @xmath58 . in
will be quickly seen that the light curves vary a great deal in the relative scale of their scatter , with natural consequences for parameter determinability . in a coarse way
, we could thus assign relatively good quality to the light curves of koi 1.01 , 2.01 , 3.01 , 12.01 , 13.01 ; moderate quality to those of koi 4.01 , 5.01 , 10.01 , 17.01 , 20.01 , 377.01 ; and poor quality to koi 7.01 , 42.01 , 72.01 , 117.01 and 388.01 .
such assessments can guide general expectations on the results . & & & & & & & & + 1.01 & 1.00083 3 & 0.00001 2 & 0.118 & 0.122 2 & 0.124 & 0.1275 7 & 84.22 & 84.3 2 + 2.01 & 1.0044 1 & 0.000003 3 & 0.214 & 0.2155 3 & 0.075 & 0.0761 2 & 88.24 & 87.2 5 + 3.01 & 0.99400 1 & 0.00002 2 & 0.060 & 0.0891 2 & 0.058 & 0.0585 2 & 89.95 & 89.9 1 + 4.01 & 1.00396 2 & 0.0005 2 & 0.223 & 0.2232 4 & 0.042 & 0.037 9 & 77.92 & 78 2 + 5.01 & 1.001 2 & -0.0001 2 & 0.132 & 0.133 1 & 0.037 & 0.036 1 & 82.51 & 82.8 5 + 7.01 & 1.0046 8 & 0.0005 3 & 0.223 & 0.2226 4 & 0.027 & 0.0269 7 & 80.79 & 80.6 6 + 10.01 & 1.00137 2 & -0.0001 1 & 0.133 & 0.146 2 & 0.093 & 0.0972 8 & 85.37 & 84.0 4 + 12.01 & 1.0001 1 & 0.00001 1 & 0.050 & 0.05420 6 & 0.088 & 0.0912 3 & 89.95 & 88.7 1 + 13.01 & 1.00401 8 & -0.00009 3 & 0.227 & 0.2315 2 & 0.078 & 0.0869 3 & 85.37 & 88.1 6 + 17.01 & 1.00112 7 & 0.00001 6 & 0.133 & 0.1366 1 & 0.094 & 0.0953 6 & 89.95 & 87.7 4 + 20.01 & 1.00084 6 & -0.00001 3 & 0.123 & 0.126 1 & 0.117 & 0.1202 6 & 89.95 & 88.6 5 + 42.01(t ) & 1.00053 1 & -0.000673 & 0.052 & 0.053 1 & 0.018 & 0.0180 2 & 87.66 & 87.7 1 + 42.01(d ) & 0.99875 1 & -0.00066 3 & 0.052 & 0.051 1 & 0.018 & 0.0176 2 & 87.66 & 87.7 1 + 72.01 & 1.010 4 & 0.0006 8 & 0.280 & 0.2797 6 & 0.013 & 0.015 2 & 87.09 & 87 3 + 117.01 & 1.0010 2 & 0.00003 3 & 0.131 & 0.131 1 & 0.023 & 0.023 2 & 83.08 & 83 1 + 377.01 & 1.0004 7 & -0.00087 3 & 0.028 & 0.032 2 & 0.075 & 0.079 3 & 89.38 & 88.9 4 + 388.01 & 1.0010 1 & 0.00066 6 & 0.125 & 0.135 3 & 0.017 & 0.018 2 & 87.09 & 87.8 2 + & & & & & & + 1.01 & 8.445 & 8.177 & 0.82 & 0.82 & 0.56 & 0.80 8 + 2.01 & 4.681 & 4.690 & 0.13 & 0.23 & 0.51 & 0.46 1 + 3.01 & 16.680 & 16.892 & 0.03 & 0.02 & 0.69 & 0.63 1 + 4.01 & 4.481 & 4.480 & 0.95 & 0.94 & 0.50 & 0.6 6 + 5.01 & 7.560 & 7.508 & 0.95 & 0.94 & 0.55 & + 7.01 & 4.486 & 4.492 & 0.71 & 0.74 & 0.54 & 0.54 2 + 10.01 & 7.500 & 6.873 & 0.64 & 0.72 & 0.52 & 0.59 9 + 12.01 & 20.040 & 18.450 & 0.03 & 0.43 & 0.50 & 0.35 1 + 13.01 & 4.400 & 4.170 & 0.35 & 0.47 & 0.44 & 0.51 1 + 17.01 & 7.530 & 7.321 & 0.03 & 0.29 & 0.55 & 0.53 3 + 20.01 & 8.133 & 7.937 & 0.02 & 0.19 & 0.52 & 0.48 2 + 42.01(t ) & 19.400 & 19.015 & 0.77 & 0.78 & 0.51 & 0.5 3 + 42.01(d ) & 19.400 & 19.473 & 0.77 & 0.77 & 0.51 & 0.54 9 + 72.01 & 3.575 & 3.575 & 0.17 & 0.20 & 0.50 & 0.6 4 + 117.01 & 7.620 & 7.634 & 0.92 & 0.92 & 0.53 & 0.5 3 + 377.01 & 36.000 & 30.769 & 0.35 & 0.61 & 0.57 & + 388.01 & 8.020 & 7.446 & 0.39 & 0.29 & 0.56 & 0.3 3 + we have worked with ` normalized ' light curves , for which the measured fluxes are divided by a representative mean out - of - eclipse value , so that @xmath55 ( ` unit of light ' ) is nominally unity , at least initially .
@xmath54 is in units of the complete range 0 - 1 .
the relative radius @xmath50 is the radius of star @xmath59 in units of the semi - major axis @xmath32 , @xmath56 is the ratio of planet to stellar radii ( @xmath60 ) , and @xmath57 denotes the inclination ( in deg ) of orbital axis to the line of sight . for koi 42.01
we list two sets of results : ( t ) the initial result , for which a linear trend across the transit phase range was noted ( see fig 24 ) , and ( d ) after this trend had been removed from the input data .
this arrangement holds also for subsequent table entries for koi 42.01 . as well as the main results in conventional notation for binary light curves , table [ tab : no4 ] lists : @xmath61 ; the inverse of the star radius in units of the semi - major axis , as listed by the nea , @xmath62 the same quantity from the present paper s fittings , @xmath63 impact parameter @xmath64 , as in the nea , and @xmath65 as found by us .
we also compare the stellar limb - darkening coefficient calculated by interpolating from the tables of van hamme ( 1993 ) , @xmath66 , with the optimized value from the transit fitting , @xmath67 .
a dash indicates that a fully determinate solution ( positive definite error matrix at the optimum ) was not found . in most cases ,
the optimized value is close to the van hamme value .
the abnormally high value found for koi 1.01 may be a result of it having a ( visual ) binary companion ( daemgen et al , 2009 ) , thereby affecting the assigned ( near solar ) temperature . table [ tab : no5 ] allows assessment of the quality of the light - curve fittings .
this involves @xmath7 which should be close to unity for a reasonably probable model fitting .
@xmath68 is the nea - derived data measure of datum accuracy , i.e. ( mean error)/(mean flux ) , while @xmath69 is the poisson factor given in table [ tab : no2 ] divided by the scaling constant @xmath70 , which is the mean value of individual poisson factor to @xmath68 ratios . the ratios @xmath71 are also listed .
the @xmath10s , whether from the nea or the poisson factors , scale the @xmath7 values to @xmath1 1 reasonably well in most cases .
there are anomalies in some instances , however .
the @xmath68 values take some account of variation in intrinsic scatter in the source counts at a given mean brightness ( gilliland et al .
2011 ) , but if this is higher than normal in a particular case , an additional intrinsic variability , perhaps of a pulsational type , could be the explanation .
this seems feasible for koi 10.01 , where there is an unexpectedly high @xmath7 ratio , as well as a significant excess of @xmath71 .
the other examples of high @xmath7 are koi 2.01 , 3.01 , 7.01 and 42.01 . in the cases of koi 2.01 and 3.01 , there are clear signs in the data of maculation effects : the eclipses of individual surface features being apparent in the light curves .
such features are not taken account of in the modelling , so the excess of @xmath2 will reflect this easily identified extraneous effect . for koi 7.01 ,
the scatter in the residuals curve shown in figure 13 , whilst apparently uniform , appears noticeably wider than the 0.0004 value assigned .
this could then be another example of micro - variability , although , in this case , the nea error measure has not been appreciably affected by the excess scatter from the candidate star .
something similar may apply also for the brighter koi 42.01 ( see figure 24 ) , though there are also longer term sources of flux variation in that case .
this example required us to deal with the issue of ` cleaning ' or detrending the data , which we discuss in more detail in subsection 3.11 below .
& & & & & + 1.01 & 240 & 0.946 & 0.00026 & 0.00026 & 1.00 + 2.01 & 304 & 1.500 & 0.00015 & 0.00016 & 0.94 + 3.01 & 259 & 1.436 & 0.00009 & 0.00010 & 0.90 + 4.01 & 342 & 1.012 & 0.00026 & 0.00027 & 0.96 + 5.01 & 241 & 0.797 & 0.00028 & 0.00029 & 0.97 + 7.01 & 498 & 1.512 & 0.00040 & 0.00040 & 1.00 + 10.01 & 395 & 1.780 & 0.00097 & 0.00075 & 1.29 + 12.01 & 895 & 1.025 & 0.00026 & 0.00026 & 1.00 + 13.01 & 378 & 1.066 & 0.00011 & 0.00013 & 0.85 + 17.01 & 424 & 1.142 & 0.00062 & 0.00061 & 1.02 + 20.01 & 567 & 1.104 & 0.00077 & 0.00069 & 1.12 + 42.01(t ) & 566 & 1.675 & 0.00009 & 0.00010 & 0.90 + 42.01(d ) & 566 & 1.175 & 0.00009 & 0.00010 & 0.90 + 72.01 & 218 & 1.083 & 0.00022 & 0.00023 & 0.96 + 117.01 & 818 & 1.061 & 0.00044 & 0.00045 & 0.98 + 377.01 & 506 & 0.935 & 0.00103 & 0.00085 & 1.21 + 388.01 & 672 & 1.083 & 0.00080 & 0.00068 & 1.18 + & & & & + 1.01 & 1.06 & 0.95 @xmath72 0.02 & 14.40 & 13.17 @xmath72 0.20 + 2.01 & 2.71 & 1.81 @xmath73 0.003 & 22.30 & 15.01 @xmath72 0.03 + 3.01 & 0.74 & 0.66 @xmath72 0.002 & 4.68 & 4.23 @xmath72 0.02 + 4.01 & 2.60 & 2.69 @xmath72 0.005 & 11.80 & 10.70 @xmath72 0.20 + 5.01 & 1.42 & 1.66 @xmath72 0.01 & 5.66 & 6.45 @xmath72 0.10 + 7.01 & 1.27 & 2.11 @xmath72 0.004 & 3.72 & 6.18 @xmath72 0.20 + 10.01 & 1.56 & 1.47 @xmath72 0.02 & 15.90 & 15.60 @xmath72 0.20 + 12.01 & 1.40 & 1.68 @xmath72 0.002 & 13.40 & 16.70 @xmath72 0.20 + 13.01 & 2.70 & 1.84 @xmath72 0.002 & 23.00 & 17.46 @xmath72 0.03 + 17.01 & 1.08 & 1.32 @xmath72 0.001 & 11.04 & 13.74 @xmath72 0.04 + 20.01 & 1.38 & 1.52 @xmath72 0.01 & 17.60 & 19.90 @xmath72 0.20 + 42.01(d ) & 1.36 & 1.59 @xmath72 0.05 & 2.71 & 3.06 @xmath72 0.10 + 72.01 & 1.00 & 1.02
@xmath72 0.002 & 1.38 & 1.64 @xmath72 0.01 + 117.01 & 1.18 & 3.52 @xmath72 0.03 & 2.93 & 10.19 @xmath72 0.40 + 377.01 & 1.01 & 1.00 @xmath72 0.060 & 8.28 & 8.65 @xmath72 0.50 + 388.01 & 1.58 & 1.96 @xmath72 0.040 & 3.01 & 3.75 @xmath72 0.40 + in table [ tab : no6 ] , the absolute radii of host stars ( @xmath26 ) are given in solar units and those of the planets ( @xmath74 ) in units of that of the earth .
in the second column , the host star radii are the nea values using the equation @xmath26 = @xmath75 . in column 3 the radii @xmath76 are derived from multiplying the fitted @xmath50 values by the nea value of the semi - major axis , @xmath32 .
the difference between the two @xmath26 estimates can be seen to be quite appreciable in some cases , with the larger range of values generally corresponding to the @xmath77 numbers , as previously suggested .
the exceptional case of koi 117 is discussed in section 3.13 , below .
the planetary radii are obtained by multiplying the stellar radii by the ratio of radii , i.e. @xmath78 = @xmath79 and @xmath80 = @xmath81 . for koi 42.01
we retain only the final detrended results , that have significantly reduced @xmath7 values .
in what follows we give summary notes on each of the examples studied .
koi 1.01 , also known as tres-2b and kepler-1b , was identified as a transiting exoplanet well before the kepler mission ( sozetti et al , 2007 ) .
the star system has been identified as a binary consisting of a g0 star , around which the planet orbits , and a k4.5-k6 red dwarf ( daemgen et all , 2009 ) .
it appears to be a relatively uncomplicated light curve , in which a ` hot jupiter ' transits the parent star towards its limb .
the light curve data for koi 1.01 consisted of the section from bjd 122.68 to 122.84 of the short cadence data set from quarter 0 .
an initial solution using only the first few transits seemed better fitted by a partial transit ( @xmath82 , where @xmath83 ) , but the rounded bottom of the eclipse becomes clear with sufficient coverage .
the result is a complete eclipse , although the transit is quite near to the limb .
the nea gives the relative radius of the star @xmath50 as 0.118 , compared to our [email protected] .
the physical sizes are : @xmath84 = 0.95 @xmath85 , and @xmath80 = 13.11 @xmath86 .
this is in fair agreement with the nea listing of 1.06 @xmath85 and 14.40 r@xmath87 respectively .
koi 2.01 (= hat - p-7 , kepler-2 ) has become a well - known exoplanet system since its discovery by pal et al .
the relatively large star in this system with its close - in planet makes for a fairly wide transit of duration around 10% of the complete phase cycle .
our light curve for koi 2.01 comes from section jd 121.25 - 121.46 of the short cadence data of quarter 0 .
the nea value for @xmath50 is 0.214 is in essential agreement with our value of [email protected] .
but we find a significantly lower value of the inclination , [email protected] deg versus the nea value of 88.24 deg .
our ratio of radii , 0.0761 , is not far from that of the nea value , 0.075 , so our solution would yield a fairly large planet , 14.40 @xmath86 which is significantly smaller than the nea value of 22.30 @xmath86 .
the transit minimum shows clear signs of a spot eclipse .
this lasts for about 8 deg of phase during the ingress , or about 0.14 of the stellar diameter , i.e. this spot seems likely to have a significantly larger size than the planet ( @xmath14.5 @xmath88 ) .
it must therefore be much larger than known solar spots .
the modelling of spot eclipses considered by budding ( 1988 ) could not be used directly here , since that applied only to the occultations of spots by relatively large bodies .
koi 2.01 s relatively massive planet is close in to the star in a configuration comparable to that of koi 13.01 , for which proximity effects are measurable from the light curve ( see below ) .
we therefore sought to fit the full light curve for this system using the radau model and the ` black body ' approximation for the light redistribution ( budding & demircan , 2007 ; p319 ) to scale trial proximity effects .
these effects appear masked by the maculation effect and perhaps other contributions to the light curve segments we studied , rendering our proximity term evaluation unreliable .
however , welsh et al ( 2010 ) , after detailed processing of 15 short cadence light curves , derived a mass ratio of about 0.013 using the roche model in this way .
koi 3.01 is another object whose planetary status was studied by the earlier hatnet project ( bakos et al , 2002 , 2009 , 2010 ) .
the star is also known as hat - p-11 and , as a confirmed planet , kepler-3 .
the light curve of koi 3.01 , shown in fig 3 , and taken from bjd 144.27 - 144.45 of the short cadence data of quarter 1 , corresponds to the wider separation and smaller star and planet than for the first two examples .
the star is somewhat brighter than in koi 2.01 , but it is a lower mass object .
again there are signs of surface inhomogeneities , this time relatively small in size .
our finding that the planet is about 4.23 times bigger than the earth in radius agrees closely with the nea value of 4.68 r@xmath87 , and in apparent exact agreement with that of bakos et al .
( 2010 ) .
bakos et al ( 2009 , 2010 ) found an eccentricity of 0.198 to the planetary orbit , having a periastron longitude of @xmath1355 with a formal error of about 15 .
in other words , the major axis of the elliptical orbit is not far from the plane of the sky , the planet exiting from the transit more quickly than its entrance .
we have checked this in our fitting program .
the transit s fitting is improved by taking its possible asymmetry into account , and the determined value of the optimal periastron longitude corresponding to the assigned eccentricity @xmath89 is about 8 .
the determination is relatively good , if we suppose all the other parameters are already well - defined , because the degree of asymmetry of the transit is close to the maximum possible for the assigned value of @xmath89 .
however , this result should probably not be taken at face value .
the improvement in the transit fitting can be associated with a better match of the model of light variation through the region of the eclipsed spot . disentangling the photometric effects of maculation from those of orbital eccentricity
is not straightforward for the empirical analysis of a single transit , as we are doing .
bakos et al .
( 2009 ) also noted the complications to parameter estimation arising from the interactions of such varied causes of data variation . for the present , therefore , this apparent confirmation of the orbital eccentricity should be regarded cautiously .
we took the light curve for koi 4.01 from section bjd 172.80 - 173.04 of the short cadence data in quarter 2 .
with an transit depth about 1/4 that of the first few examples , koi 4.01 , a fainter source than the others so far , exhibits a much less well - defined light curve .
the derived planet size turns out to be quite comparable to that in koi 3.01 , while the star itself is appreciably bigger , so the transit effect is smaller .
there is nothing obvious in these data that would immediately indicate a false - positive , but koi 4 was given this status , apparently due to inclusion of a background eclipsing binary in the photometry .
this points up the possibility that other _ prima facie _ planetary light curves turn out to be false positives . here
we note that a false positive occurs when the star identified with a planetary transit is not so eclipsed .
there may still be a planetary transit within the photometric aperture , but small shifts in the light centre during the eclipse indicate the effects relate to another star , not the one identified in the catalogue ( see bryson et al , 2013 , for a recent review ) .
in fact , a plausible alternative solution was found , corresponding to a background algol system about 6.4 mag fainter than the target star .
the small relative radius of the eclipsed star , required by the low phase range of the eclipse , combined with the 3.85 d period , would require the background system to be relatively massive
perhaps @xmath15 solar masses or so , but the resulting @xmath7 value ( 1.07 ) is hardly different from the planetary transit model in terms of probability . there is a distinct maculation effect seen in the out - of - transit light of this star .
periodogram analysis ( available as an on - screen option on the nea data display webpage ) indicated a likely value of @xmath15.8 days for the stellar rotation period as a result . in the case of koi 5.01 , there was a reasonable agreement between our starting trial value of @xmath50 , using the @xmath90 and @xmath32 values of table [ tab : no1 ] , and the value obtained from the curve - fitting .
it is thus not surprising that our final value of @xmath74 , at 6.45 r@xmath87 , is not far from the nea value ( 5.66 r@xmath87 ) .
the transit is complete , although close to the limb .
the data used for this candidate are from section bjd 171.13 - 171.30 of the short cadence measures from quarter 2 .
koi 7.01 , also known as kepler-4 , was reviewed by borucki et al .
( 2011a ) .
with koi 7.01 we again find a relatively small planet in a not - too - distant orbit .
although the loss of light is similar to koi 4.01 , the flux is only 1/3 of that , so the low s / n of the light curve makes for a generally imprecise fitting .
nevertheless , our value for the planetary radius , 6.18 @xmath86 , is significantly larger than the nea value of 3.72 @xmath86 , and is also larger than than the 3.99 @xmath86 found by borucki et al .
( 2011a ) .
as indicated in the notes on table 5 in the previous section , the scatter of the data in the uniform distribution of residuals is appreciably larger than the nea numbers suggest and micro - variability of some kind may be the explanation .
we followed the option of carrying out a periodogram analysis for this data - set , which comes from section bjd 171.65 - 171.99 of the short cadence run in quarter 2 .
we found the power spectrum having spectral components of much lower frequency than those corresponding to small pulsations . the present paper ,
whilst leading to suggestions about candidate micro - variables , does not address the detailed properties of such stars . the flux for koi 10.01 is an order of magnitude less than for koi 1.01 , though the depth of transit is comparable in this data section from bjd 173.82 - 174.09 of the short cadence , quarter 2 run .
we therefore do not expect so good a solution for this large planet , bigger than jupiter by @xmath140% . the scatter in the solution
is @xmath15 times bigger than that of koi 1.01 . as with koi 7.01
, the star may therefore be showing some additional effect that the modelling does not take into account , perhaps small - scale , short - period inherent fluctuations .
the relatively large value of @xmath7 in table [ tab : no5 ] supports the same inference .
the planet in this system has the greatest separation from its host star in our sample , at just over 30 solar radii , but its estimated mean surface temperature at over 800 k is still too high for the ` habitable zone ' . the star is not bright , so the combination of short transit phase range and scatter makes for a relatively poor solution .
the preset value of the limb - darkening coefficient given in table [ tab : no1 ] ( 0.50 ) , produced a systematic trend in the residuals .
our optimized value ( 0.35 ) has effectively removed this irregularity . since optimized limb - darkening coefficients , when allowed to be free parameters , generally agree with those coming from the van hamme ( 1993 ) code , this disagreement is suggestive that one of the underlying parameters for the code , probably the star s effective temperature , has been wrongly assigned in the nea .
the used light curve data are from bjd 128.43 - 129.05 of the short cadence set from quarter 0 .
koi 13.01 has become a well - known candidate of special interest , since it was noticeable , even from the preliminary inspections , that the light curve shows a secondary minimum and proximity effects .
this accords with reasonable expectation for a relatively large planet at fairly low separation .
photometry of this object is , however , compromised by the presence of a close companion of comparable brightness , which entails slight shifts of the light centre on the detection array that may affect the processing statistics .
some data - sets also show indications of additional light variations , perhaps related to maculation effects , or the role of the additional low mass companion in the host star multiple system recently identifed by santerne et al .
welsh et al .
( 2010 ) mentioned also the possibility of a focus drift .
although the transit fitting , that corresponds to data selected from bjd 169.81 - 170.07 of the short cadence second quarter run , may look good , judging from a fairly uniform scatter of residuals , such difference curves have sometimes shown residual short - term irregularities ( blips ) at the inner contact points . in this connection , we anticipated a possible failure of the limb - darkening approximation very close to the limb . a second order limb - darkening term was therefore included in the fitting function
however , it became clear that this enhancement , whilst slightly lowering the optimal @xmath2 value , did nothing to remove the blips .
later experiments have shown that such effects are associated with a poor selection of the orbital inclination .
this controls the rate at which the planet passes over the limb , and that plays a critical role for the blips .
the presence of blips in an otherwise satisfactory set of residuals lends a good diagnostic quality about the action of the optimization , and for the inclination in particular .
removal of the end - point blips in fitting the transit gave a definitive quality to the corresponding main geometrical elements .
figure [ fig : koi13 ] shows the fitting to the primary transit used to derive the parameters given in tables [ tab : no3 ] and [ tab : no4 ]
. there also appears a slight residual trend both through the transit region and a little way on either side .
this may be associated with the flux centroid drifting , mentioned above ( see also mislis & hodgkin , 2012 ) . on the whole , we regard our modelling of this complex system hitherto as preliminary .
the object will merit much closer individual attention .
as mentioned , koi 13.01 is associated with normal proximity effects throughout its orbital cycle . to examine these effects we used normalized long cadence data from quarter 2 extending from bjd 208.507 to 210.612 .
figure [ fig : koi13ecl ] shows the out - of - transit region of the light curve of koi 13 on an expanded scale .
the primary minimum region has been clipped .
the secondary eclipse of the reflected light is visible , as are the proximity effects .
the first maximum is seen to be slightly higher than the second one , in agreement with the effect of ` doppler beaming ' .
doppler beaming is programmed to correct the assigned relative luminosity of the host star @xmath91 ( essentially unity , unless another star is contributing to the measured flux ) , so that instead of @xmath91 in the fitting function ( budding & demircan , 2007 ; eqn 9.17 ) we use @xmath92 , say , through the formula @xmath93 here the line of sight component of the ( stellar ) orbital velocity is given by @xmath94 , @xmath95 being the mass ratio planet to star , @xmath96 the orbital phase and @xmath57 the inclination .
the velocity of light @xmath97 is set at 299792 s@xmath98 .
the orbital velocity is determined from known parameters as @xmath99 . using the value of the planetary mass from table [ tab : no1 ]
the effect turns out to improve the fitting , but only at a marginal level of significance : @xmath2 decreasing by an amount of order unity after the replacement of @xmath91 by @xmath92 in the fitting function . as noted by mislis & hodgkin ( 2012 ) ,
the proximity effects offer a way to determine the mass ratio , although various other parameters are involved in fixing the scale of these effects .
one parameter is @xmath50 , which scales the ` ellipticity effect ' with a third power dependence so clearly having relatively high sensitivity .
we found the out - of - eclipse variation tending to bias @xmath50 to a slightly larger number than that corresponding to the optimal fitting of the transit alone .
however , since the transit phases are relatively clear of proximity effects , we felt it legitimate to use the @xmath50 determined from the transit to throw the weight of the complete light curve fitting onto the mass ratio .
actually , the mass ratio ( @xmath95 ) and gravity - brightening coefficient ( @xmath100 another unknown parameter producing the out - of - eclipse ` ellipticity effect ' light variation ) are both involved in the same linear combination , so if either is decreased the other must be correspondingly increased to produce essentially the same result .
the structural parameters also influence the scale of the ellipticity effect in the same kind of linearly correlated way as @xmath100 , so their inclusion tends simply to scale down the value of @xmath95 in proportion to the increase in the @xmath101 coefficients ( see introduction section ) above unity .
we adopted the value of @xmath100 according to its ` black - body ' formula given in winkepler , allowing the ellipicity effect to determine only @xmath95 .
the value of @xmath95 found in this way ( [email protected] ) compares well with that ( 0.0029@xmath72 0.0005 ) given by mislis & hodgkin ( 2012 ) .
we should not regard this close numerical agreement as definitive , however . as we have seen the light curves of koi 13.01
are complicated by other effects not included in our present model , while even within that model there is scope for the variation of a combination of other parameters that can mimic the effect of a given mass ratio .
hence , the formal error on @xmath95 is really only a lower limit , corresponding to what could be estimated as the error if all the other parameters were regarded as precisely known .
still , the general consistency about the result is encouraging ( see also welsh et al .
, 2010 ; mazeh et al . ,
2012 ) .
the jupiter - sized planet , koi 17.01 , produces a clear and relatively deep primary minimum , even if over a small phase range .
the photometry , corresponding to bjd 173.09 - 173.39 of the short cadence second quarter run , shows a rather higher level of scatter in the fitting than might have been anticipated ( cf .
figs 21 and 22 ) .
there may then be some residual variation in the light curve , though not as pronounced as for some other examples .
although a somewhat fainter and noisier data - set than that of koi 17.01 , the light curve shows a relatively deep eclipse ( in fact the deepest in our selection ) caused by the near - central transit of large ( @xmath11.6 r@xmath102 ) planet in a comparable configuration .
our light curve for koi 20.01 came from from bjd 170.81 - 171.20 of the short cadence data set from quarter 2 .
this planet with radius at about 2.7 r@xmath87 , is one of the smaller ones in our list .
taking into account also the wide separation of close to 30 solar radii , the determinacy of the transit is not high .
the flux level is relatively high , however , so a solution is possible in which the planet passes not far from the limb of the host star .
the fitting is shown in figure [ fig : koi42 ] .
the data were taken from bjd 145.36 - 145.75 of the short cadence first quarter run .
the residuals curve in fig 24 shows a small downward trend , and an additional variability for this star was mentioned in section 2.2 in connection with the anomalously high value of @xmath7 . following the approach of zeilik et al ( 1988 ) in dealing with light curves affected by starspots , a ` cleaning ' stage after the first fitting was considered .
the residuals curve may be fitted by a linear regression , the original transit data then ` corrected ' to a ` clean ' input file , and the fitting run again .
the main geometric parameters all have an even - function effect in varying the light curve about zero phase though , so , while the residuals would be reduced by such a correction and the @xmath7 anomaly reduced , the optimal parameter - set of the model should not be greatly affected .
the downward trend across the range is not more than @xmath120% of the mean scatter , so we would expect the maximum effect on the @xmath170% excess of @xmath7 in the original fitting to be not more than @xmath140% . the initial impression from fig 24
is that random scatter is greater than the assigned 0.0001 , and a small pulsational behaviour can be anticipated .
in fact , the fitting to the detrended data - set did reduce @xmath7 by most of its original excess , while there were also small changes to the main parameters , particularly @xmath50 .
it is known that the optimization convergence is slowed by the presence of extraneous effects , so it can be presumed that the detrending accelerated the fitting process to an inherently better set of parameter estimates .
the @xmath117% remaining excess of @xmath7 may still reflect some inherent short - period variability .
this system has the shortest period of our sample , and the planet s radius is found to be only about 1.64 r@xmath87 , making the planet a ` hot earth ' , orbiting close in to a sunlike star .
the transit is relatively wide in phase , therefore , but not in time .
figure [ fig : koi72 ] shows the fast ingress and egress , with data points evenly spaced .
the data are from bjd 263.82 - 263.97 of the short cadence observations in the second quarter . the transit , which appears shallow against the scatter , is not well resolved in this way .
even with this non - optimal monitoring arrangement , however , figures 25 & 26 show that an acceptable solution is possible .
a shallow and narrow eclipse makes for a relatively poorly defined parameter - set for this rather faint system that resembles koi 42.01 .
our light curve is from bjd 271.24 - 271.80 of the short cadence data set from quarter 2 .
the transit appears free of additional complications and the @xmath7 ratio and distribution of residuals imply the main photometric elements , which are not so different from those of the nea data - base , are reliable .
yet the originally assigned nea radius of the host star ( from the gravity formula ) is appreciably different from the value found by our program .
this latter radius is relatively large : at around 3 solar radii it is the largest star in the sample , and presumably corresponds to a relatively old object now well on towards the end of its main sequence state .
koi 377.01 is also known as kepler 9-b and formed part of a detailed review by torres et al .
koi 377 contains at least one similar , but more distant , planet as well as probably a third inner and smaller one .
torres et al . ( 2011 ) , concerned with the occurrence of false positives , or non - planetary configurations that might simulate similar transit - like effects , were able to show with confidence that koi 377.01 is a saturn - sized planet .
time of transit variations in this system were previously studied by holman et al .
( 2010 ) . the data used for koi 377.01 came from bjd 644.89 - 645.24 of the short cadence run from quarter 7 , from which we can clearly see that this solar - type star is affected by significant ( @xmath10.5% of the mean flux ) out - of - transit variability , probably maculation .
there are also a couple of rather high ` rogue points ' .
such occasional discordant individual points have been noticed in the data - sets examined for this study .
they may perhaps result from cosmic ray effects not removed in preliminary processing of the archived data . or , in keeping with the idea of enhanced magnetic activity associated with strong maculation , one could perhaps also expect occasional flares .
rogue points tend to slow convergence in the modelling and it is advisable to remove them when possible . the regular approach to maculation effects has been to detrend beforehand ( cf .
koi 42.01 above ) , although the phase range of the transit is often so small as make the maculation variation of very little effect during a single transit , as here , particularly as the light loss is relatively deep for this central transit
. figure [ fig : koi377 ] shows an 8 deg phase range about the transit displaying a good fit for the model .
periodogram fitting to the out - of - transit variation indicated a @xmath18.4 d period , which can be associated with the stellar rotation .
the planet , with less than 700 k mean temperature , is the coolest in our sample , but still quite far from being a ` habitable zone ' candidate .
a significant eccentricity was given for this system in the ( unsourced ) nasa ames research center website ( http://kepler.nasa.gov/mission/discoveries/ ) .
as with koi 3.01 , we tested the listed value , @xmath89 = 0.15 , with a transit fitting , having first produced a trial value for the longitude parameter @xmath21 . in practice , it is the mean anomaly at phase zero @xmath103 which requires specification for an eccentric light curve fitting , but @xmath21 is then determined by the formula @xmath104 90@xmath105 ( budding & demircan , 2007 , p314 ) . however , our transit fitting experiments were not decisive on this .
a wide range of values of @xmath103 seemed to give almost the same final @xmath2 value , while if we allow also the zero phase parameter @xmath54 also to be simultaneously adjustable the solution loses determinacy . basically , the scatter is too large in this data - set for a definitive assessment of the system s possible eccentricity from transit fitting alone , and even if numbers are produced for the eccentricity parameters they must be compromised by the aforementioned additional light variation effects .
the planet s longitude of periastron @xmath21 = 269.4 @xmath106 corresponding to the marginally lowest value of @xmath2 , has the major axis of the ellipse close to the line of sight , i.e. an almost symmetrical transit .
this point by itself argues for poor determinacy of the eccentricity parameters from transit fitting alone , in this case .
figure [ fig : koi388 ] shows the transit of a relatively small planet ( ` super - earth ' ) about one of the lower mass stars in our sample , that has a relatively low flux level .
the transit is thus very noisy , like koi 72.01 but worse , yet a 4-parameter determinate solution ( the relatively independent phase correction @xmath54 having been eliminated in a preliminary run ) was found .
high determinacy for such a data - set can not be expected , but it is worth considering in the context of small planet discoveries that the kepler mission addresses .
the light curve data for koi 388.01 was drawn from bjd 741.22 - 741.68 of the short cadence data in quarter 8 .
we have presented an independent approach to the analysis of archival photometric data from the kepler mission , applying alternative curve - fitting techniques with a modified version of the curvefit program for close binary systems ( zeilik et al . , 1988 ) :
the fitting function involves tidal and rotational distortions of the components that depend on structural coefficients derived for modern stellar models .
this permits , in principle , checks on stellar structure or the role of rotation that is non - synchronized to the planet s orbital motion . the important issue of determinacy for the underlying model is also checked in winkepler by examination of the curvature hessian at the optimum .
several results of interest have emerged .
for example , the data on koi 7.01 , 10.01 and 42.01 point to additional microvariability , apart from the main transit phenomena analysed .
similarly , the light curves of koi 2.01 and 3.01 show clear evidence of surface inhomogeneities .
such effects , not included in the present model , are reflected in high optimal values of @xmath7 .
the case of koi 42.01 raised the issue of detrending .
since we dealt mainly with transits that often correspond to very narrow intervals of phase , the fitting function s inherent symmetry across the eclipse would often not be compromised by a slight linear gradient in the observations . for koi 42.01
the value of @xmath7 was significantly affected , however , and linear detrending before the final fitting was advantageous for a more rapid convergence towards a realistic @xmath7 value and parameter - set .
the difference in the resulting geometric elements after the detrending was quite small in comparison to the derived errors , however .
we have shown that in most cases there is sufficient information in the transit sections of the light curves alone to allow a 5 , often 6 , element specification of the model .
more information would be contained in light curves having a complete phase range , but for many of these sunlike stars there are relatively strong maculation effects that are separate from the parameter - sets of direct relevance to the present study and complicate the out - of - eclipse variations on a relatively large scale . in a few cases , we could estimate a likely rotation period using the maculation effect .
for koi 4.01 we found a solution that was close to that previously published in the nea , and essentially comparable to all the other normal star and planet combinations in this paper .
however , matijevic et al . (
2012 ) cast doubt on this candidate as an exoplanet , indicating a more likely background close binary star .
if such an ambiguity can exist among such normal - seeming candidates the implications for discoveries of exoplanets by the transit method could be seriously compromised , unless checks for background eclipses associated with active pixel offsets are simultaneously available ( bryson et al . , 2013 ) . in that case
, it may be possible to confirm the real source of the variation ascribed to a planetary transit of a given star .
bryson et al . ( 2013 ) show that the planetary transit model for such a light curve may give an obviously poor fitting result , that would be signalled by the @xmath7 value at the optimum .
it is also still possible for the eclipse to result from a planet , but not of the target star .
the situation can be checked , in the curve - fitting context , by relaxing the condition that the fractional light of the eclipsed star is effectively unity , as in the normal case . for koi 4.01
, we thus found that the light curve could be modelled by a background classical algol as effectively as by a planetary transit . in the absence of independent evidence coming from apparent shifts of the light centre on the detector array , ambiguity on the cause of the eclipses
would remain .
checks of theoretical linear limb - darkening coefficients by perturbing initially set values and allowing subsequent optimization have shown the values of van hamme ( 1993 ) to be generally supported to within reasonable accuracy limits , though this was not the case for koi 1.01 and 12.01 .
the assigned temperatures may be inappropriate for some reason , or the effects of interstellar extinction considered , in such examples .
the inclusion of a second order term in the limb - darkening approximation had only a marginal effect on @xmath7 : the expected amplitude of this term would normally only be of the order of accuracy to which the linear term can be specified .
koi 13.01 is noticeably different in several respects . here
, we were able to test the program , including its newly added doppler beaming component , to the complete light curve .
the significance of this latter component was demonstrated by a small reduction of the fitting s @xmath2 .
the effects of structural coefficients in the fitting function were also checked for this example . since the proximity effects scale mainly as a direct product of @xmath95 and @xmath107 a few percent decrease in @xmath95
is observed in moving from the roche ( @xmath108 ) to radau ( @xmath109 ) models .
we expect the significance of this point to become enhanced in future cases of well - defined light curves containing large ( jupiter sized or greater ) planets of short period , whose mass ratios may be independently checked spectroscopically . in figure
[ fig : ratioofr_1 ] we show the trend of our results for the relative radius @xmath110 against those published by the nea @xmath111 . the regression coefficient
( at 0.9978 ) looks reassuring , although we may notice that the @xmath110 values of the present paper are generally slightly greater than those of the data - base @xmath111 by a few percent .
the recent addition of error estimates to the nea parameter listings , together with those of the foregoing tables , allow a better perspective on the generally low significance of the differences in @xmath50 values .
the situation is similar with the ratio of radii ( fig [ fig : ratioofradii ] ) , apart from koi 13 , where we find a distinctly smaller planet than that of the nea .
although , for some reason , the nea error estimates for @xmath56 appear about an order of magnitude lower than our ( interdependent ) ones , the discrepancy for koi 13.01 is outside the range of assigned errors , unlike koi 4.01 and 10.01 .
it is with the distribution of inclinations ( fig [ fig : ratioofinclination ] ) that we find a small but significantly different trend , in the sense that the nea inclinations are peaked more towards the 90 limit than ours . running a check for a random distribution of detectable ( totally ) eclipsing binaries , i.e. satisfying @xmath112 for a given value of @xmath50 , demonstrates the expected flatness of the distribution for high values of @xmath57 .
it is well known that , in general , the distribution of observed inclinations of binary orbits on the sky is proportional to @xmath113 .
this means that the distribution of observed inclinations can be reproduced from a uniform distribution of numbers in the range 0 to 1 by considering @xmath114 for those numbers .
alternatively , the tendency to constancy of the parent distribution @xmath113 , when @xmath57 is close to 90 degrees , implies that the numbers in each degree interval close to 90 , for a given representative value of @xmath50 , should be approximately similar . if a significant number of long period planets are included , the observed inclinations would compact towards 90 , due to the decreasing value of @xmath50 , but these cases are less frequently detected , and we have only 3 planets with a period longer than 10 d. planets of solar type stars that cluster around an average 3 d period should have their inclinations distributed reasonably uniformly over the range @xmath182 - 90 degrees . we have found , in fitting procedures that start with a value of @xmath50 that is too small , that most of the transit can be well fitted with a high value of @xmath57 .
residual small blips at the ingress and egress points in the difference curve may remain , however , even though their effect on the net value of @xmath2 is small .
this possible ` @xmath2-valley ' in the optimization process may have occurred for nea solutions whose inclinations appear somewhat clustered towards 90 degrees .
our independent testing of the parametrization of light curves from the nea shows a good measure of agreement in most cases .
where differences were found , we have inquired into their causes .
we believe some differences arise from the more physically detailed fitting function and thorough optimization sequences of our findings .
it is possible that local minima for @xmath2 occur for certain parameter combinations .
the issue shows up in the difference ( residuals ) curve as fairly uniform scatter over most of the range , but sometimes with a few short - term irregularities , typically in the ingress and egress regions .
further optimization experiments with small random shifts in initial parameter sets near known approximate solutions can remove these .
some of the differences in our absolute parameters may come from a different procedure for finding the host star radius .
the nea appears to have retained @xmath115 values using the gravity estimates , as mentioned in section 2.1 .
here we can keep in mind the relative accuracies and sensitivities .
semi - major axes calculated using the very precisely known period may have a reasonable accuracy despite the imprecisely known host star mass , since the sensitivity of @xmath32 to @xmath116 is low . on the other hand ,
the mean density estimate is sensitive to @xmath50 and the gravity can not be determined photometrically with sufficient precision at present .
our values of @xmath7 adopted as optimal are mostly consistent with a high probability to the underlying ( eclipsing binary system ) model .
this is on the basis of error estimates for the data - sets supplied directly by the nea and checked by reasonable statistical arguments on the flux detection .
a few fittings result in @xmath7 values that are unusually high , pointing to the occurrence of inherent microvariability , surface maculation or both .
the possibility of empirical checks of limb - darkening , gravity - brightening , structural and rotational effects ( in selected examples ) , gave additional objectives to the present work .
only koi 13.01 , in our sample , shows a sufficient scale of proximity effects ( above other causes of flux variation ) to allow testing of the relevant parameters ( gravity darkening , reflection and doppler beaming ) .
the mass ratio could be optimized and the empirical result is within reasonable error limits of other determinations of this value , although there are still reasons to be cautious about the value of @xmath95 for the complex system koi 13.01 at present .
a main argument in this article has been that it is useful to compare different analytical approaches .
we have presented the new , purpose - designed , software winkepler to this aim .
this will help to make finally accepted stellar and planetary parameters and their corresponding error estimates , as obtained from transit photometry , more robust and better understood .
our initial sample of 16 planetary candidate light curves is perhaps rather small , but a methodologically useful avenue into the special issues raised by planet eclipses .
we plan to continue with more detailed studies of individual data - sets using this software .
mr and eb attended the carl sagan summer workshop at the california institute of technology , pasadena during the week july 22 - 29 , 2012 , which was a considerable stimulation for much of the work in this paper .
the meeting was organized by the nasa exoplanet science institute ( nexsci ) , who supported eb s registration and accommodation .
eb appreciates also the rasnz s kingdon tomlinson fund for its partial support of his travel costs .
bakos , g. , lzr , j. , papp , i , sri , p , green , e.m .
, 2002 , pasp , 114 , 974 bakos , g. , et al .
( 9 authors ) , 2009 , iau symp . , 253 , 21 bakos , g. , et al .
( 24 authors ) , 2010 , apj , 710 , 1724 barnes , t.g . ,
evans , d.s . , 1976 , mnras , 174 , 489 batalha , n.m . , et al .
( 77 authors ) , 2013 , apjss , 204 , 24 borucki , w.j . , et al .
( 19 authors ) , 2011a , apj , 713 , 2 borucki , w.j . , et al .
( 69 authors ) , 2011b , apj , 736 , 19 brown , t.m . , latham , d.w . ,
everett , m.e .
, esquerdo , g.a . , 2011 ,
aj , 147 , 112 buchave , l.a . , et al . ( 29 authors ) , 2012 , nature , 486 , 375 budding , e. , 1988 , apss , 143 , 1 budding , e. , rhodes , m. , sullivan d. , 2005 , in _ transits of venus : new views of the solar system and galaxy _ , iau coll 196 , ed d.w .
kurtz , cup .
p386 budding , e. , demircan , o. , 2007 , _ introduction to astronomical photometry _ , cup burrows , a. , 2012 , in _ from interacting binaries to exoplanets : essential modelling tools _ , iau symp 282 , eds m.t .
richards & i. hubeny , cup , p562 daemgen , s. , hormuth , f. , brandner , w. , bergfors , c. , janson , m. , hippler , s. , henning , t. , 2009 , a&a , 498 , 567 devore , e. , koch , d. , gould , a. , harman , p. , 2009 ,
baas , 41 , 412 etzel , p. , 1981
, in _ photometric and spectroscopic binary systems _ , eds e.b .
carling & z. kopal , d. reidel pub .
, p111 gilliland , r.l .
( 17 authors ) , 2011 , apjs , 197 , 6 holman , m.j . , et al .
( 41 authors ) , 2010 , science , 330 , 51 inlek , g. , budding , e. , 2012 , apss , 342 , 365 kang , y - w . , 2012 , in _ from interacting binaries to exoplanets : essential modelling tools _ , iau symp 282 , eds m.t .
richards & i. hubeny , cup , p329 kipping , d. , bakos , g. , 2011 , apj , 730 , 50 koch , d.g./ et al .
( 47 authors ) , 2010 , arxiv 1001/1001.0268.pdf kopal , z. , 1959 , _ close binary systems _ , chapman & hall , london & new york mandel , k. , agol , e. , 2002 , apj , 580 , 171 matijevic , g. , zwitter , t. , munari , u. , 2012 , in _ assembling the puzzle of the milky way _
, ed c. reyl , a. robin , m. schultheis , v19 , i d 09006 milone , e. , 1993 , _ light curve modeling of eclipsing binary stars _ , springer mislis , d. , hodgkin , s. , 2012 , mnras , 422 , 1512 muirhead , p.s . ,
hamren , k. , schlawin , e. , rojas - ayala , b , covey , k.r . ,
lloyd , j.p . , 2012 , apj , 750 , 37 nelson , b. , davis , w.d . , 1972 ,
apj , 174 , 617 pl , a. , et al .
( 17 authors ) , 2008 , apj , 680 , 1450 pinsonneault , m.h . , deokkuen , a. , molenda - zakowicz , j. , chaplin , w.j . ,
metcalfe , t.s .
, bruntt , h. 2012 , apjss , 199 , 30 pressley , a. , 2001 , _ elementary differential geometry _
, springer rhodes , m. , 2013 , winkepler manual , http://home.comcast.net/@xmath1michael.rhodes/ santern , a. , et al .
( 9 authors ) , 2012 , a&a , 544 , 12 seager , s. , malln - ornelas , g. , 2003 , ap j. , * 585 * , 1038 .
southworth , j. , 2012 , in _ from interacting binaries to exoplanets : essential modelling tools _ , iau symp 282 , eds m.t .
richards & i. hubeny , cup , p131 sozzetti , a. , et al .
( 8 authors ) , 2007 , apj , 664 , 1190 torres , g. , et al .
( 30 authors ) , 2011 , apj , 727 , 24 van hamme , w. , 1993 , aj , 106 , 2096 welsh , w. , et al .
( 8 authors ) , 2010 , apj , 713 , l145 wilson , r. e. , devinney , e.j . , 1971 , ap j. , 166 , 605 zeilik , m. , de blasi , c. , rhodes , m. , budding , e. , 1988 , apj , 332 , 293
* see figs . 2 - 39 .
* , so involve also the errors of the relative radius @xmath50 .
the results are comparable from both sources in most cases .
the difference in the solution for koi 13.01 is again evident , but the derived k error is sufficiently large as to reduce its significance . ] | we have modified the graphical user interfaced close binary system analysis program curvefit to the form winkepler and applied it to 16 representative planetary candidate light curves found in the nasa exoplanet archive ( nea ) at the caltech website http://exoplanetarchive.ipac.caltech.edu , with an aim to compare different analytical approaches .
winkepler has parameter options for a realistic physical model , including gravity - brightening and structural parameters derived from the relevant radau equation .
we tested our best - fitting parameter - sets for formal determinacy and adequacy .
a primary aim is to compare our parameters with those listed in the nea .
although there are trends of agreement , small differences in the main parameter values are found in some cases , and there may be some relative bias towards a 90 value for the nea inclinations .
these are assessed against realistic error estimates .
photometric variability from causes other than planetary transits affects at least 6 of the data - sets studied ; with small pulsational behaviour found in 3 of those . for the false positive koi 4.01
, we found that the eclipses could be modelled by a faint background classical algol as effectively as by a transiting exoplanet .
our empirical checks of limb - darkening , in the cases of koi 1.01 and 12.01 , revealed that the assigned stellar temperatures are probably incorrect . for koi 13.01 ,
our empirical mass - ratio differs by about 7% from that of mislis and hodgkin ( mon . not .
r. astron .
soc .
422:1512,2012 ) , who neglected structural effects and higher order terms in the tidal distortion .
such detailed parameter evaluation , additional to the usual main geometric ones , provides an additional objective for this work . |
the electronic structure calculations presented in this paper were performed using ab initio density functional theory ( dft ) as implemented in the vienna ab initio simulation package ( vasp)@xcite .
this method uses a plane wave basis set along with the projector augmented waves ( paw)@xcite in the ionic core region .
we used the generalized gradient approximation with the perdue - burke - ernzerhof ( pbe ) parametrization @xcite for the exchange - correlation functional .
the optimum values of the energy cutoff and the size of the k - point mesh were found to be 450 ev and @xmath107 , respectively , and were accordingly employed in our calculation .
the calculations of the rashba coefficients were carried out by taking into account the spin - orbit coupling term perturbatively on top of the fully optimized charge densities . the slab geometry used in the rashba calculations
was created by first optimizing the bulk ktao@xmath0 structure and the cell parameters .
further relaxations of the slab was carried out by keeping the cell dimensions fixed , but allowing all atoms to relax according to the hellman - feynman forces on each atom with the tolerance value of @xmath108 ev@xmath9 .
the atomic displacements are significant only near the first few surface layers as seen from fig .
[ str ] , where we have shown the cation - anion displacements for the ko and tao@xmath19 planes in the first six layers near the surface .
they are close to zero in the middle five layers in the simulation cell and are not shown here .
both surfaces of the slab are tao@xmath19 terminated and for the calculations with the electric fields , a v - shaped potential ( fig .
[ v ] ) was applied .
the linear muffin tin orbitals ( lmto ) method@xcite was used along with the same exchange - correlation functional as above to obtain the partial contributions to the rashba coefficients shown in fig . 2 of the main text .
to obtain these partial contributions , the spin - orbit interaction was retained within a specified radius of the nuclei or on atoms in specific layers .
as emphasized already in the main text , the contribution of the surface electric field to the rashba coefficient is negligible as compared to what is observed in the experiments or computed from the density - functional calculations .
much of the contribution comes from the nuclear regions of the heavy atoms near the surface , where the inversion symmetry is broken . to assess the effect of the surface field
, the surface potential was calculated from the density - functional theory .
we computed the surface potential by averaging the kohn - sham potential over the @xmath69-plane parallel to the surface .
this planar - averaged potential , shown by the red lines in fig .
1 in the main text , has strong oscillations due to the presence of the atomic planes .
averaging it further over a unit cell length along the @xmath109 direction , we get the average surface potential @xmath110 , which may be fitted to a modified jones - jennings - jepsen form@xcite : @xmath111^{-1 } [ -1+e^{-\lambda(z - z_0 ) } ] & z > z_0 \\ u - v e^{\kappa(z - z_0 ) } & z < z_0 .
\end{array } \right . \label{v}\ ] ] the potential has the image form in the vacuum region @xmath112 , with @xmath113 being the image plane location , while inside the material , we have the thomas - fermi exponential form for metallic screening with the inverse screening length @xmath114 . for the relaxed kto in the absence of an external field ,
the parameters are : @xmath115 ev , @xmath116 ev , @xmath117 , @xmath118 @xmath9 , and @xmath119 @xmath9 . the screening is much weaker than the thomas - fermi uniform electron gas result @xmath120 , @xmath121 , since the electron gas is present only in a small region near the surface .
the exponential decay of the electrons into the vacuum region shown in fig .
1 in the main text was obtained by solving the 1d schrdinger equation @xmath122\psi = e\psi$ ] with this potential .
under the @xmath27 crystal field , the five @xmath20 orbitals split into the two - fold @xmath123 ( @xmath124 , @xmath125 ) and the three - fold @xmath126 ( @xmath127 ) states ( also referred to as @xmath28 and @xmath29 , respectively).@xcite when the spin - orbit coupling term @xmath128 is included , the @xmath29 states ( six states including spin degeneracy ) break into a spin - orbit coupled set of states , viz .
, a two - fold @xmath30 and a four - fold @xmath31,@xcite with the partner functions as follows : @xmath129 @xmath130 while the @xmath28 states remain four - fold degenerate with the corresponding partner functions : @xmath131 when the symmetry is reduced from @xmath27 to @xmath33 corresponding to a surface , we have : @xmath132 and @xmath133 , with the corresponding partner functions for the @xmath33 group indicated on the right hand side of eqs .
( [ two ] ) - ( [ four ] ) .
thus , we have two @xmath34 states within the @xmath29 manifold , which are allowed to mix by symmetry .
the amount of mixing depends on their relative energies , which are determined by the crystal - field energy @xmath102 and the spin - orbit coupling strength @xmath17 . for convenience of discussion
, we have used the nomenclatures @xmath34 , @xmath134 , @xmath135 , and @xmath136 to indicate different mixtures between these @xmath29-derived @xmath34 states in table i of the main text , where also the corresponding partner functions are listed without the normalization factors .
( the @xmath28-derived @xmath34 state does not mix significantly owing to the large energy denominator @xmath137 . ) for instance , in table i , the partner functions for @xmath134 are : @xmath138 while the corresponding orthogonal partners spanning the @xmath29 space are @xmath139 as can be easily checked , these two sets are obvious linear combinations of the two @xmath29-derived @xmath34 states given in eqs .
( [ two ] ) - ( [ three ] ) .
the level scheme of the crystal - field split states is shown in fig .
[ levels ] .
the rashba spin - orbit splitting is possible when both spin - orbit interaction and broken inversion symmetry are present as indicated by the rashba split parabolas for a @xmath34 state in fig .
[ levels ] .
the strength of the rashba splitting , characterized by the rashba coefficient @xmath5 , depends on the proximity of the other crystal - field - split levels of the same symmetry , as the splitting comes from admixture of states of the same symmetry .
thus , for example , the @xmath16 states in fig . [ levels ] should have small rashba coefficients because they are separated by the large crystal - field energy @xmath137 from each other , allowing thereby only a small admixture between them .
in fact , the rashba coefficient for these states scales as @xmath140 as seen from the table i in the main text due to this energy denominator effect .
the only states that have a relatively larger @xmath5 are the two @xmath29-derived @xmath34 states owing to their proximity in energy .
in order to understand the rashba effect for the different ta @xmath20 orbitals , we considered a tight - binding model for a cubic lattice terminated by a surface and subject to an electric field . the electron confinement in the direction normal to the surface due to the polar field and the surface potential leads to individual subbands , while the parallel momentum @xmath141 still remains a good quantum number .
it is convenient to consider the rashba effect for each subband , which may be described by taking the square lattice for the surface with the atomic " orbitals being the appropriate wannier functions for the subband . to this
we add the spin - orbit interaction term @xmath142 and the external electric field term @xmath143 , resulting in the minimal model hamiltonian @xmath144 using the basis set : @xmath145 ,
@xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 , the tight - binding hamiltonian in the momentum space is given by where @xmath156 , @xmath157 , @xmath158 , @xmath22 and @xmath23 are the hopping integrals@xcite between the @xmath20 orbitals and @xmath137 , @xmath102 , and @xmath159 are the crystal field parameters .
the @xmath160 matrix in the spin space in the above equation is a unit matrix because the tight - binding hopping does not cause any spin - flip scattering .
we take @xmath161 and the remaining energy shifts between the various @xmath20 orbitals due to surface confinement are incorporated into the crystal field parameters , viz . ,
@xmath162 and @xmath102 , without any loss of generality .
the spin - orbit interaction part @xmath163 in the same basis set , is given by @xmath164 where there is no momentum dependence because the spin - orbit interaction is a local term that couples the orbital and spin angular momenta on the same site only .
finally , the electric field leads to non - zero matrix elements between several orbitals due to the field - induced asymmetry of the orbital lobes , as illustrated in fig .
5 of the paper , and these matrix elements are expected to be roughly proportional to the local electric field .
we define these matrix elements as @xmath165 , @xmath166 , and @xmath167 , where the second orbital is located on the nearest - neighbor site along the direction indicated by the subscript .
other matrix elements such as @xmath168 from symmetry .
it is important to note that these matrix elements change sign if the hopping is from right to left or vice versa , which results in the sine factors in the bloch sum ( as opposed to a cosine factor which leads to the usual @xmath95 band dispersion in the tight - binding theory ) .
these sine factors eventually show up as the linear - k terms in the rashba hamiltonian for small momentum .
we thus have @xmath169 _ lwdin downfolding _ the hamiltonian eq .
[ eqh ] results in the spin - orbit split bands and we obtained the form of the band structure by downfolding the effects of other bands further away in energy via the perturbative lwdin downfolding@xcite .
the lwding downfolding is in essence a perturbative method , which works if the energies of the orbitals to be downfolded are far removed from the main orbitals of interest .
the lwdin downfolding procedure for solving the eigenvalue problem @xmath170 is to partition the hamiltonian into blocks @xmath171 where the two blocks are well separated in energy .
we are not interested in the states in the block @xmath172 , but include its effects on the block @xmath173 by perturbation theory . the exact result for the effective hamiltonian for states in the @xmath173 subspace is given by @xmath174 which however involves the eigenvalue @xmath175 of the full hamiltonian .
it can be shown that an iterative solution of eq .
[ downfolding ] produces the brillouin - wigner perturbation series,@xcite which to the lowest order yields the result : @xmath176 where @xmath177 and @xmath178 belong to the subspace @xmath173 and @xmath18 belongs to the subspace @xmath172 . in most cases
@xmath175 can be replaced by the diagonal elements @xmath179 and eq .
[ heff ] is valid if @xmath180 .
we can now use the lwding downfolding to derive an effective hamiltonian for states that are well - separated in energy from the remaining states in the hamiltonian eq .
( [ eqh ] ) .
we are interested in bands near the @xmath101 point ( @xmath181 ) , where the bands are quadratic in momentum without the rashba effect .
we therefore retain only the linear terms in momentum in the final expression for the downfolded hamiltonian .
for example , by downfolding the effect of the remaining eight orbitals in the full hamiltonian into the ( @xmath182 ) subspace via eq .
[ heff ] with @xmath183 , we get the result , for small @xmath18 around the @xmath101 point : @xmath184 in terms of the pauli matrices , this is written as @xmath185 where in writing the second term , we have ignored the smaller energies @xmath186 in the denominator of the off - diagonal matrix elements in the downfolded hamiltonian eq .
[ df ] . the second term in eq .
[ r ] describes the momentum - dependent spin splitting and can be written in the familiar rashba form , viz . , @xmath187 , with the rashba coefficient @xmath188 as given in the first line of table i in the main text .
the rashba coefficients for the other orbitals were obtained in a similar manner . _ second - order perturbation theory _ it is instructive to illustrate the origin of the rashba hamiltonian using the second - order brillouin - wigner perturbation theory eq .
( [ heff ] ) .
we illustrate this for the rashba spin splitting of the @xmath124 orbital .
referring to fig .
[ pert ] , the tight - binding band structure is indicated by the parabolas , which are spin - degenerate .
the spin degeneracy is removed due to the interaction via the intermediate orbitals @xmath189 or @xmath190 .
the various matrix elements can be read off from the hamiltonians , eqs .
( [ tb ] - [ eqe ] ) , and two of them have been shown in fig .
thus the off - diagonal matrix element @xmath191 due to the interaction with the @xmath192 state is given by the product of the two interactions indicated in fig .
[ pert ] divided by the energy denominator , viz .
, @xmath193 .
note that the first interaction is due to the electric - field induced asymmetry , while the second term is due to the spin - orbit interaction . clearly , both are necessary to produce a non - zero result .
adding up the contributions from the three other intermediate orbitals , viz . ,
@xmath194 , @xmath195 , and @xmath196 , we get the desired result @xmath197 and , in the same manner , we can obtain the remaining matrix elements to yield @xmath198 where the rashba coefficient @xmath199 as before . | we show that the rashba effect at the polar perovskite surfaces and interfaces can be tuned by manipulating the two - dimensional electron gas ( 2deg ) by an applied electric field , using it to draw the 2deg out to the surface or push it deeper into the bulk , thereby controlling the surface - sensitive phenomenon .
these ideas are illustrated by a comprehensive density - functional study of the recently - discovered polar ktao@xmath0 surface .
analytical results obtained with a tight - binding model unravel the interplay between the various factors affecting the rashba effect such as the strength of the spin - orbit interaction and the surface - induced asymmetry .
our work helps interpret the recent experiments on the ktao@xmath0 surface as well as the srtio@xmath0/laalo@xmath0 interface .
= 1 the rashba effect describes the momentum - dependent spin splitting of the electron states at a surface or interface and is the combined result of the spin - orbit interaction ( soi ) and the inversion - symmetry breaking@xcite .
it is commonly described by the hamiltonian @xmath1 where @xmath2 and @xmath3 are the electron momentum and spin , @xmath4 is along the surface normal , and @xmath5 is the rashba coefficient , which leads to the linear spin splitting in the band structure @xmath6 .
the control of the rashba effect by an applied electric field is at the heart of a class of proposed spintronics devices for manipulating the electron spin@xcite . the perovskite interfaces@xcite are expected to have a much larger rashba effect than their semiconductor counterparts@xcite , owing to the presence of high z elements and a strongly localized 2deg formed by the polar catastrophe .
in fact , a strong rashba effect was recently observed in the / interface@xcite , which also showed an ill - understood asymmetric dependence on the direction of the applied electric field . in this letter , we show that the polar perovskite structures constitute an excellent system for the field control of the rashba effect , aided by the relative ease with which the 2deg can be manipulated in these polar structures . detail density - functional results are presented for the ktao@xmath0 ( kto ) surface to illustrate the ideas . _ 2deg at the kto surface _
the ktao@xmath0 ( kto ) surface is an ideal system for the study of the rashba effect because ta is a high z element with strong soi , a polar - catastrophe induced 2deg has been observed there recently@xcite similar to the lao / sto interface , and finally a surface rather than an interface is more easily amenable to external electric fields . fig .
[ densityprofile ] shows the basic features of the 2deg formed at the kto surface obtained from our calculations using density - functional theory ( dft ) , performed with the gga functional and the projector augmented wave pseudopotential method as implemented in the vienna ab initio simulation package @xcite . to simulate the -terminated surface
, we used a slab geometry consisting of 17 and 16 ko alternating layers corresponding to the formula unit ( kto)@xmath7 and 24 of vacuum .
we studied the rashba effect by applying a series of electric fields and by fully relaxing the crystal structure in each case.@xcite for the kto surface , the alternating charged layers , nominally ( ) @xmath8 and ( ko)@xmath9 , lead to the polar catastrophe just like in lao / sto and as a result a 2deg forms in the surface region terminated by .
considerable structural relaxation , as expected for a polar surface , spreads the 2deg several layers into the bulk .
the relaxations , which produce local dipole moments screening out the surface polar field , decay rapidly within about six kto layers and beyond that , the ionic positions return to their bulk values . _ the rashba effect _ the microscopic origin of the rashba effect is the relativistic soi @xmath10 where @xmath11 is the potential gradient . for a spherically symmetric potential , such as the field from a nucleus , it assumes the familiar form : @xmath12 in the presence of a symmetry - breaking surface electric field @xmath13 , the first expression for @xmath14 leads to eq .
( [ rashba ] ) , with the rashba coefficient @xmath15 .
however , this coefficient is severely underestimated , if one naively identifies the electric field with the surface potential gradient .
rather , the correct picture is that the rashba soi originates in the nuclear region due to the large nuclear field gradient there@xcite .
the applied electric field polarizes the orbital wave functions , changing their weights at the nucleus , so that the electron experiences a different nuclear field .
we illustrate this in fig .
[ partialrashba ] by computing the various contributions to the rashba splitting for the @xmath16 bands in fig .
[ bands ] ( a ) .
we have isolated these contributions by keeping the soi @xmath17 either ( i ) on atoms in specific layers or ( ii ) on all atoms but within a specified spherical nuclear region and then by performing a single iteration with the self - consistent dft potential obtained with all interactions present .
as fig .
[ partialrashba ] shows , the dominant contribution comes from the nuclear region of the surface atoms .
this in turn suggests the tuning of the rashba effect by an electric field by moving the 2deg in and out of the surface layers .
_ electric field tuning _ we have calculated the rashba splitting for the kto surface by applying a series of electric fields . as seen from fig .
( [ bands ] ) , the unrelaxed surface with zero field shows a very strong linear-@xmath18 rashba splitting because the 2deg is sharply localized at the surface due to the strong polar field , extending to just three tao@xmath19 layers ( fig .
[ densityprofile ] ) .
relaxation of the surface atoms screens out the polar field and as a result , the 2deg spreads deeper into the bulk region , thereby washing away the rashba effect .
this explains why in the arpes experiments@xcite on kto , the rashba splitting has not been seen despite the presence of a large spin - orbit coupling . on the contrary ,
application of an electric field draws the 2deg towards the surface ( fig .
[ contours ] ) , restoring back the rashba effect .
an electric field in the opposite direction drives the 2deg deeper into the bulk and the rashba splitting quickly becomes very small .
thus , we have demonstrated the field tuning of the rashba effect as well as the very interesting asymmetric dependence on the direction of the applied electric field ( fig .
[ bands ] ( e ) ) .
such an asymmetric dependence was recently observed in the lao / sto interface@xcite .
note that the asymmetry is not expected for a non - polar surface such as ag and a symmetric rashba effect has been predicted there.@xcite _ tight - binding description _ the rashba splitting differs widely within the @xmath20 orbital manifold , which may be understood in terms of the tight - binding ( tb ) model@xcite on the cubic lattice that includes the surface asymmetry and the electric field : @xmath21 the kinetic energy part contains the standard @xmath22 and @xmath23 hopping between the @xmath20 orbitals and the crystal field energies : @xmath24 @xmath25 , where @xmath26 denotes the site - orbital - spin index .
the @xmath27 cubic field splits the d states into @xmath28 plus @xmath29 states . with
the soi included , the six - fold degenerate @xmath29 states ( including spin ) split into a two - fold @xmath30 and a four - fold @xmath31 state , while the e@xmath32 remains unsplit with @xmath31 symmetry .
the surface reduces the cubic symmetry into @xmath33 with @xmath30 going into @xmath34 , while the @xmath31 state splits into @xmath35 , both two - fold degenerate.@xcite note that there are just two double representations @xmath16 and @xmath34 for the @xmath33 group ; we have used primes on @xmath34 to indicate its different orbital character ( see table i ) due to the symmetry - allowed mixing between the two @xmath34 states in the @xmath29 manifold.@xcite [ t ] ll | ll | c & & rashba + & & rashba pseudo - spin & coefficients + & symmetry & partner functions & @xmath36 + @xmath28 & @xmath37 & @xmath38 & @xmath39 & @xmath40 + & & @xmath41 & @xmath42 & @xmath43 + + @xmath30 & @xmath44 & @xmath45 & @xmath46 & @xmath47 + @xmath31 & @xmath48 & @xmath49 & @xmath50 & @xmath51 + & & @xmath52 & @xmath53 & @xmath54 + + @xmath30 & @xmath44 & @xmath55 & @xmath56 , & @xmath57 + & & & @xmath58 & + @xmath31 & @xmath59 & @xmath60 & @xmath61 , & @xmath62 + & & & @xmath63 & + & & @xmath52 & @xmath64 & @xmath54 + the remaining parts , @xmath65 and @xmath66 , in eq .
( [ htb ] ) are the inversion symmetry breaking fields crucial for the rashba effect .
the surface field @xmath65 is modeled by an asymmetric energy for the surface orbitals : @xmath67 and @xmath68 , an asymmetry that may come from strain , the electric field via the atomic relaxation it produces , or the hopping differential between the orbitals , e.g. , @xmath69 and @xmath70 @xcite and as such has a complex dependence on the electric field .
the electric field part @xmath66 induces new hoppings ( fig . [ hopping ] ) between atoms : @xmath71 , @xmath72 , and @xmath73 , whose strengths are roughly proportional to the local electric field . here
the subscript denotes the direction of the nearest - neighbor on which the second orbital is located .
typical parameters for kto are@xcite : @xmath74 ev , @xmath75 ev , @xmath76 ev , @xmath77 ev , while @xmath78 and @xmath79 are @xmath80 mev at the surface layer .
as one goes into the bulk , the surface asymmetry parameters @xmath81 and @xmath82 rapidly go to zero , so that the rashba effect comes just from the first few surface layers .
we obtain the rashba splitting from eq .
[ htb ] by lwdin downfolding@xcite of the effects of the higher - energy bands .
the results can be expressed in the rashba form @xmath83 for most bands and the corresponding rashba coefficients and the partner functions for the pseudo spin @xmath3 are listed in table i. however , for near - degenerate cases , where the soi is strong ( @xmath84 ) but the surface field does not sufficiently lift the four - fold degeneracy of the @xmath31 state ( @xmath85 or @xmath86 ) , the lwdin downfolding fails and the rashba soi can only be written as a @xmath87 matrix spanning the @xmath31 subspace : @xmath88 here , we have included the quadratic - k terms ( the band mass ) , with @xmath89 , @xmath90 , @xmath91 , @xmath92 , and @xmath93 , and the order of the basis is the same as the order of appearance of the four @xmath31 partner functions in case 2 , table i. the rashba part @xmath94 is simply eq .
( [ 44 ] ) minus the @xmath95 terms .
eq .
( [ 44 ] ) is valid for strong @xmath17 and for any @xmath96 and @xmath97 . if @xmath98 , one recovers the results of case 2 , table i using lwdin downfolding .
fig .
[ tb-44 ] shows the tb bands for cases relevant to the rashba splitting seen in the dft bands . note from table
i that even though the linear - k rashba splitting is always present , its magnitude is very small ( @xmath99 ) for the @xmath28 bands as well as for the @xmath29-derived @xmath16 bands . for these bands ,
the higher - order @xmath100 term may in fact be dominant as has been seen in the srtio@xmath0 surface@xcite and also suggested by zhong et al@xcite . also as table i shows , the pseudo - spin partner functions are sometimes not spin entangled at the @xmath101 point , but they always become so away from @xmath101 due to the spin mixing via the rashba hamiltonian . returning to the @xmath29 bands , which make up the 2deg , for small soi @xmath17 relative to the surface field @xmath102 ( case 1 in the table ) , the rashba coefficient can be small if @xmath102 is large ( note that @xmath103 can be varied widely in a material due to lattice relaxation , electric field , or strain , while @xmath17 is more or less fixed ) .
this is the reason for the relatively weak rashba splitting seen in the tb bands ( fig .
[ tb-44 ] ) and also in the dft bands for the relaxed case with @xmath104 ( fig .
[ bands ] ( a ) ) . the rashba effect is enhanced significantly in the opposite limit ( @xmath84 ) , if at the same time the surface field @xmath102 is small or comparable to the electric - field - induced hopping @xmath105 . in this scenario ,
the rashba effect is described by eq .
( [ 44 ] ) .
this is the case for fig .
[ tb-44 ] ( b ) and ( c ) , where a large rashba splitting is seen for the @xmath106 bands , and also for the dft bands ( fig .
[ bands ] ( a ) and ( c ) ) .
thus the electric field changes the rashba effect in two ways : _ one _ , by changing the density of the 2deg in the surface layers and _ two _ , by altering the surface asymmetry field @xmath102 and reorienting the orbital energies . in conclusion , we showed that the rashba effect can be tuned in the polar perovskite oxides by manipulating the 2deg profile by an external electric field .
these results are relevant not just for the kto surface , but also for polar materials in general that contain surface or interface @xmath20 electrons .
we also note that since the energies of the @xmath20 orbitals are sensitive to the applied strain , this suggests another means of tailoring the rashba effect .
we thank zoran popovic for stimulating discussions and computational help .
this work was supported by the office of basic energy sciences of the u. s. department of energy through grant no .
de - fg02 - 00er45818
. 13 e. i. rashba , sov .
phys .
solid state 2 , 1109 ( 1960 ) .
r. winkler , _ spin - orbit effects in two - dimensional electron and hole systems _
( springer , new york , 2003 ) .
a. ohtomo , d. a. muller , j. l. grazul , and h. w. hwang , nature ( london ) * 419 * , 378 ( 2002 ) .
a. ohtomo and h. w. hwang , nature ( london ) * 427 * , 423 ( 2004 ) .
j. nitta , t. akazaki , h. takayanagi , and t. enoki , phys .
rev .
lett .
* 78 * , 1335 ( 1997 ) .
m. b. shalom , m. sachs , d. rakhmilevitch , a. palevski , and y. dagan , phys .
rev .
lett . * 104 * , 126802 ( 2010 ) .
a. d. caviglia , m. gabay , s. gariglio , n. reyren , c. cancellieri , and j .-
m .
triscone , phys .
rev .
lett .
* 104 * , 126803 ( 2010 ) .
p. d. c. king , r. h. he , t. ekpanakul , _ et al .
_ , phys .
rev .
letts . * 108 * , 117602 ( 2012 ) .
a. f. santander - syro , c. bareille , f. fortuna , o. copie , m. gabay , f. bertran , a. taleb - ibrahimi , p. le fvre , g. herranz , n. reyren , m. bibes , a. barthlmy , p. lecoeur , j. guevara , and m. j. rozenberg , phys .
rev .
b * 86 * , 121107 ( 2012 ) .
natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , *
* , ( ) .
, * * , ( ) . see supplementary materials for the tight - binding derivation of the rashba coefficients as well as for details of the symmetry representations , the density - functional methods used in our calculations , and the structural relaxation for the kto surface obtained from the dft calculations .
g. bihlmayer , yu . m. koroteev , p. m. echenique , e. v. chulkov , and s. blgel , surf .
sci .
* 600 * , 3888 ( 2006 ) r. o. jones , p. j. jennings , and o. jepsen , phys .
rev .
b * 29 * , 6474 ( 1984 ) ; b. r. k. nanda , s. satpathy , and m. springborg , phys . rev .
lett .
* 98 * , 216804 ( 2007 ) .
l. petersen and p. hedagard , surf .
sci . * 459 * , 49 ( 2000 ) .
s .- j .
gong , c .- g .
duan , y. zhu , z .- q .
zhu , and j .- h .
chu , phys .
rev .
b * 87 * , 035403 ( 2013 )
. g. f. koster , j. o. dimmock , r. g. wheeler , and h. statz , _ properties of the thirty - two point groups _
( mit press , cambridge , 1963 ) . , , , * * , ( ) .
g. jellison , i. paulauskas , l. boatner , and d. singh , phys .
rev .
b * 74 * , 155130 ( 2006 ) .
t. neumann , g. borstel , c. scharfschwerdt , and m. neumann phys .
rev .
b * 46 * , 10623 ( 1992 ) .
p. lwdin , j. chem . phys .
* 19 * , 1396 ( 1951 ) .
h. nakamura , t. koga , and t. kimura , phys .
rev .
lett . * 108 * , 206601 ( 2012 ) .
z. zhong , a. tth , and k. held , phys . rev .
b * 87 * , 161102 ( 2013 ) . |
we study dimension properties of sets of furstenberg type .
in particular we are interested to be able to construct very small furstenberg sets in a given class .
the notion of size of a set is its hausdorff dimension , denoted by @xmath21 .
we begin with the definition of classical furstenberg sets . for @xmath0 in
@xmath1 $ ] , a subset @xmath2 of @xmath3 is called _ furstenberg set _ of type @xmath0 or @xmath4-set if for each direction @xmath5 in the unit circle there is a line segment @xmath6 in the direction of @xmath5 such that the hausdorff dimension of the set @xmath7 is greater or equal than @xmath0 .
we will also say that such set @xmath2 belongs to the class @xmath4 .
it is known ( @xcite , @xcite , @xcite ) that @xmath22 for any @xmath4-set @xmath23 and there are examples of @xmath4-sets @xmath2 with @xmath24 .
hence , if we denote by @xmath25 then @xmath26.\ ] ] in @xcite the left hand side of this inequality has been extended to the case of more general dimension functions , i.e. , functions that are not necessarily power functions ( @xcite ) .
a function @xmath27 will be called _ dimension function _ if it belongs to the following class : @xmath28 if one looks at the power functions , there is a natural total order given by the exponents .
if we denote by @xmath29 , then @xmath30 is _ dimensionally smaller _ than @xmath31 if and only if @xmath32 . in @xmath33 ,
however , by extending this natural notion of order we only obtain a _ partial _ order . [
def : order ] let @xmath34 be two dimension functions .
we will say that @xmath35 is dimensionally smaller than @xmath27 and write @xmath36 if and only if @xmath37 we will be particularly interested in the special subclass of dimension functions that allows us to classify zero dimensional sets .
a function @xmath38 will be called `` zero dimensional dimension function '' if @xmath39 for any @xmath40 .
we denote by @xmath41 the subclass of those functions . as usual , the @xmath27-dimensional ( outer )
hausdorff measure @xmath42 will be defined as follows . for
a set @xmath43 and @xmath44 , write @xmath45 the @xmath27-dimensional hausdorff measure @xmath42 of @xmath2 is defined by @xmath46 we will consider functions @xmath34 such that there exist constants @xmath47 with @xmath48 for all @xmath49 to be equivalent , even though they would not lead to the exact same measures .
in that case we write @xmath50 .
the hausdorff dimension of a set @xmath51 is the unique real number @xmath52 characterized by the following properties : * @xmath53 for all @xmath54 .
* @xmath55 for all @xmath56 . therefore , to prove that a given set @xmath2 has dimension @xmath52 , it is enough to check the preceding two properties , independently if @xmath57 is zero , finite and positive , or infinite .
however , in general it is not true that , given a set @xmath2 , there is a function @xmath38 , such that if @xmath58 then @xmath10 , and if @xmath59 , then @xmath60 .
the difficulties arise from two results due to besicovitch ( see @xcite and references therein ) .
the first says that if a set @xmath2 has null @xmath42-measure for some @xmath38 , then there exists a function @xmath35 which is dimensionally smaller than @xmath27 and for which still @xmath10 .
symmetrically , the second says that if a _ compact _
set @xmath2 has non-@xmath61-finite @xmath42 measure , then there exists a function @xmath58 such that @xmath2 has also non-@xmath61-finite @xmath62 measure .
these two results imply that if a set @xmath2 satisfies that there exists a function @xmath27 such that @xmath63 for any @xmath59 and @xmath10 for any @xmath58 , then it must be the case that @xmath64 .
this is the particular case in which @xmath2 is a so called @xmath27-set ( generalizing the notion of @xmath52-set for those @xmath52-dimensional sets that satisfy @xmath65 .
but if we look for example at the set @xmath66 of liouville numbers , this set is not an @xmath27-set for any @xmath38 .
these sets are called _ dimensionless _ ( see @xcite ) .
in addition , it is shown in @xcite that there are two proper nonempty subsets @xmath67 of dimension functions such that @xmath68 for all @xmath69 and @xmath70 for all @xmath71 .
this motivates the following notion of dimension partition ( see @xcite ) . by the _ dimension partition _ of a set @xmath2
we mean a partition of @xmath72 into ( three ) sets : @xmath73 with * @xmath74 .
* @xmath75 . *
@xmath76 . note that by the previous example , it is clear that there are sets @xmath2 for which @xmath77 is empty , reflecting the dimensionless nature of @xmath2 . on the other hand , @xmath77 is never empty for an @xmath27-set , but it is not easy to determine this partition in the general case
. we also remark that it is possible to find non - comparable dimension functions @xmath78 and a set @xmath2 with the property of being a @xmath35-set and an @xmath27-set simultaneously .
it follows that even for @xmath27-sets the dimension partition , and in particular @xmath77 , is not completely determined .
note that the results of besicovitch cited above imply that , for compact sets , @xmath79 and @xmath80 can be thought of as open components of the partition , and @xmath77 as the `` border '' of these open components .
an interesting problem is then to determine some criteria to classify the functions in @xmath72 into those classes . to detect where this `` border '' is
, we will introduce the notion of _ chains _ in @xmath72 .
this notion allows to refine the notion of hausdorff dimension by using an ordered family of dimension functions .
more precisely , we have the following definition . a family @xmath81 of dimension functions
will be called a _ chain _ if it is of the form @xmath82 that is , a totally ordered one - parameter family of dimension functions .
suppose that @xmath38 belongs to some chain @xmath83 and satisfies that , for any @xmath84 , @xmath63
if @xmath59 and @xmath10 if @xmath58 . then , even if @xmath85 , in this chain , @xmath27 does measure the size of @xmath2 .
it can be thought of as being `` near the frontier '' of both @xmath79 and @xmath80 .
for example , if a set @xmath2 has hausdorff dimension @xmath0 but @xmath86 or @xmath87 , take @xmath29 and @xmath88 . in this chain
, @xmath30 is the function that best measures the size of @xmath2 .
we look for finer estimates , considering chains of dimension functions that separate sets of `` the same hausdorff dimension '' . the goal will be to find very fine chains to give precise bounds on how far from an expected dimension function the hausdorff measure drops to zero or remains positive . in this setting
, the sharpness of the results are associated to the `` fineness '' of the chains .
consider , for example , a set @xmath89 such that @xmath90 , in which case it is clear that @xmath91 for all @xmath92 .
compare the chain @xmath93 ( which only detects the hausdorff dimension of @xmath89 ) to @xmath94 , in which we can detect if there exists @xmath95 for which @xmath96 and @xmath97 for all @xmath92 . to measure this distance between to dimension functions
, we introduce the following notion : let @xmath98 with @xmath59 .
define the `` gap '' between @xmath35 and @xmath27 as @xmath99 from this definition and the definition of partial order , we always have that @xmath100 , and therefore the speed of convergence to zero of @xmath101 can be seen as a notion of distance between @xmath35 and @xmath27 .
the analogous definition of furstenberg sets in the setting of dimension functions is the following .
[ def : furs ] let @xmath102 be a dimension function .
a set @xmath103 is a furstenberg set of type @xmath102 , or an @xmath104-set , if for each direction @xmath105 there is a line segment @xmath6 in the direction of @xmath5 such that @xmath106 .
in @xcite we proved that the appropriate dimension function for an @xmath104 set @xmath2 must be dimensionally not much smaller than @xmath107 and @xmath108 ( this is the generalized version of the left hand side of ) , the latter with some additional conditions on @xmath102 .
there we also presented precise bounds on the dimensional gaps .
these results are valid for all dimension functions @xmath102 - minding the small restrictions on the decay of @xmath102 for the second bound - regardless of the fact that @xmath102 is ( or is not ) zero - dimensional .
we also exhibited an example of a zero dimensional furstenburg - type set with @xmath109 points in each direction , proving that for this case , @xmath110 contradicting the nave limiting argument in equation , that @xmath111 .
( it seems feasible to find an argument to also construct a zero dimensional @xmath112-set for @xmath113 points as remarked by keleti and mth @xcite ) however , if we consider zero - dimensional furstenberg - type sets belonging to @xmath114 , where @xmath115 is defined by @xmath116 , we showed that any @xmath19 must have @xmath117 . in the present work we look at a refinement of the upper bound for the dimension of furstenberg sets .
since we are looking for upper bounds on a class of furstenberg sets , the aim will be to explicitly construct a very small set belonging to the given class .
we first consider the classical case of power functions , @xmath118 , for @xmath119 .
recall that for this case , the known upper bound implies that , for any positive @xmath0 , there is a set @xmath9 such that @xmath120 for any @xmath92 .
we improve this estimate in terms of logarithmic gaps by showing that there is no need to take a `` power like '' step from the critical dimension .
for any @xmath40 consider @xmath121 .
then , there is an @xmath4-set @xmath2 such that @xmath122 for any @xmath12 ( ) .
note that @xmath123 for all @xmath92 .
we will also focus at the endpoint @xmath124 , and give a complete answer about the size of a class of furstenberg sets by proving that , for any given @xmath16 , there exists a set @xmath125 such that ( ) @xmath126 note that , when @xmath127 , the dimension functions @xmath128 increase , `` approaching '' from below the positive - dimensional case .
however , when @xmath129 , the functions decrease , `` approaching '' the sets having finitely many ( or countably many ) points in each direction .
in this latter case , a zero - dimensional furstenberg set can be constructed @xcite .
we will provide an example of an even smaller furstenberg set @xmath130 ( which naturally depends on the choice of @xmath131 ) that has @xmath109 points in each direction , but such that @xmath132 ( see ) . as usual , we will use the notation @xmath133 to indicate that there is a constant @xmath134 such that @xmath135 , where the constant is independent of @xmath89 and @xmath136 .
by @xmath137 we mean that both @xmath133 and @xmath138 hold .
the paper is organized as follows .
in , for a given @xmath8 , we develop the main construction of small furstenberg @xmath139 sets , and obtain dimension estimates for the class @xmath4 , @xmath40 . in we
first show that we can refine the argument of the previous section to include the zero - dimensional functions @xmath140 defined in .
we then continue to improve on the construction given in @xcite to construct extremely small furstenberg sets that contain 2 points in each direction and we look at the same problem from the point of view of packing dimension and compare the very different behaviour .
is devoted to the proof of an important lemma regarding the size of zero dimensional fibers .
in this section , given @xmath119 we will focus our attention to the upper dimension estimate for @xmath141 sets . as stated in it
was shown that if @xmath142 is the infimum of all the possible values for the hausdorff dimension of @xmath4-sets , then @xmath143.\ ] ] we will concentrate on the right hand side of this inequality , which has been proved by showing that there exists a set @xmath2 in @xmath141 , such that @xmath144 for any @xmath145 . however , by the result of besicovitch cited in the introduction , we know that it is not going to be true that @xmath146 for any @xmath147 . by refining the arguments of wolff in , if @xmath148 we are able to exhibit a set @xmath2 in @xmath141 , such that for @xmath149 we have that @xmath150 .
note that @xmath151 for any @xmath152 .
we begin with a preliminary lemma about a very well distributed ( mod 1 ) sequence .
[ lem : discre ] for @xmath153 and any real number @xmath154 $ ] , there is a pair @xmath155 such that @xmath156 this lemma is a consequence of theorem 3.4 of @xcite , p125 , in which an estimate is given about the discrepancy of the fractional part of the sequence @xmath157 where @xmath0 is a irrational of a certain type .
we also need to introduce the notion of @xmath158-sets , a common ingredient in the construction of kakeya and furstenberg sets .
a @xmath158-set is a compact set @xmath159 which is contained in the strip @xmath160 such that for any @xmath161 $ ] there is a line segment contained in @xmath2 connecting @xmath162 with @xmath163 of slope @xmath164 , i.e. @xmath165\ \exists \ b\in{\mathbb{r } } : mx+b\in e,\ \forall \ x\in [ 0,1].\ ] ] finally we need some notation for a thickened line . given a line segment @xmath166 , we define the @xmath167-tube associated to @xmath168 as @xmath169 now we are ready to prove the main result of this section .
[ thm : logfalpha ] for @xmath170 $ ] and @xmath171 , define @xmath172 .
then , if @xmath12 , there exists a set @xmath9 with @xmath173 . fix @xmath174 and let @xmath175 be a sequence such that @xmath176 .
we consider @xmath177 to be the set defined as follows : @xmath178 : \forall j \
\exists\ p , q\ ; q\le n_j^\alpha ; it can be seen that @xmath179 ( see , ) . if @xmath180 and @xmath181\right)$ ] , we have that @xmath182 $ ] is bi - lipschitz .
therefore the set @xmath183 also has hausdorff dimension @xmath0 .
the main idea of our proof , is to construct a set for which we have , essentially , a copy of @xmath184 in each direction and simultaneously keep some optimal covering property .
define , for each @xmath153 , @xmath185 , q\le n^\alpha\right\}\ ] ] and @xmath186 to count the elements of @xmath187 ( and @xmath188 ) , we take into account that @xmath189 therefore , @xmath190 . for @xmath155 ,
define the line segments @xmath191,\ ] ] and their @xmath192-tubes @xmath193 with @xmath194 .
we will use during the proof the notation @xmath195 instead of @xmath193 . also define @xmath196
note that , by , all the @xmath197 are @xmath158-sets . for each @xmath198 , we look at the points @xmath199 , and define the set @xmath200 . clearly , @xmath201 .
but if we note that , if @xmath198 , then @xmath202 we can bound @xmath203 by the number of non - negative rationals smaller than 2 of denominator @xmath204 .
since @xmath205 , we have @xmath206 . considering _ all _ the elements of @xmath188 , we obtain @xmath207 .
let us define @xmath208 for each @xmath209 , take @xmath194 .
then @xmath210 can be covered by @xmath211 balls of radio @xmath192 with @xmath212 . to see this
, it suffices to set a parallelogram on each point of @xmath213 for each @xmath214 in @xmath188 .
the lengths of the sides of the parallelogram are of order @xmath215 and @xmath216 , so their diameter is bounded by a constant times @xmath216 , which proves the claim
. we can now begin with the recursive construction that leads to the desired set .
let @xmath217 be a @xmath158-set written as @xmath218 ( the union of @xmath219 @xmath220-thickened line segments @xmath221 with appropriate orientation ) .
each @xmath222 to be constructed will be a @xmath158-set of the form @xmath223 having constructed @xmath222 , consider the @xmath224 affine mappings @xmath225\times[-1,1]\rightarrow s_{\ell^j_{i}}^{\delta^j}\qquad 1\le i\le m_j,\ ] ] defined by @xmath226 we choose @xmath227 large enough to satisfy @xmath228 and apply @xmath229 to the sets @xmath230 defined in to obtain @xmath231 since @xmath230 is a union of thickened line segments , we have that @xmath232 for an appropriate choice of @xmath233 , @xmath234 and @xmath233 line segments @xmath235 . from the definition of the mappings @xmath229 and since the set @xmath230 is a @xmath158-set , we conclude that @xmath236 is also a @xmath158-set .
define @xmath237 to cover @xmath238 , we note that if @xmath239 , then @xmath240 , and therefore there exists a rational @xmath241 with @xmath242 therefore @xmath243 , so we conclude that @xmath238 can be covered by @xmath244 balls of diameter at most @xmath245 .
since we chose @xmath175 such that @xmath246 , we obtain that @xmath247 admits a covering by @xmath248 balls of the same diameter .
therefore , if we set @xmath249 and @xmath250 we obtain that @xmath251 for @xmath252 .
therefore , for any @xmath12 , the last expression goes to zero .
in addition , @xmath253 is a @xmath158-set , so it must contain a line segment in each direction @xmath161 $ ] .
if @xmath168 is such a line segment , then @xmath254 the final set of the proposition is obtained by taking eight copies of @xmath2 , rotated to achieve _ all _ the directions in @xmath255 .
in this section we will focus on the class @xmath4 at the endpoint @xmath124 .
note that all preceding results involved only the case for which @xmath8 . introducing the generalized hausdorff measures , we are able to handle an important class of furstenberg type sets in @xmath217
. we will look first at the @xmath256-dimensional case , but in which we request the set to contain uncountably many points . for this
, it will be necessary to put in each direction some set with many points but with certain structure .
the proof relies on , but we must replace the set @xmath177 by a generalized version of it . more precisely , we will need the following lemma . [
lem : fiber ] let @xmath257 and consider the sequence @xmath258 defined by @xmath259 , the function @xmath260 and the set @xmath178\setminus { \mathbb{q } } : \forall j \
\exists\ p , q\ ; q\le { \mathfrak{f}}(n_j ) ; |x-\frac{p}{q}|<\frac{1}{n_j^2}\right\}.\ ] ] then we have that @xmath261 for @xmath262 .
we postpone the proof of this lemma to the next section . with this lemma
, we are able to prove the main result of this section . we have the next theorem .
[ thm : sqrth3/2 ] let @xmath263 .
there exists a set @xmath264 such that @xmath265 .
we will use essentially a copy of @xmath177 in each direction in the construction of the desired set to fulfill the conditions required to be an @xmath104-set .
let @xmath177 be the set defined in .
define @xmath184 as @xmath266 where @xmath267 is the same bi - lipschitz function from the proof of .
then @xmath184 has positive @xmath268-measure .
let us define the corresponding sets of for this generalized case .
@xmath185 , q\le { \mathfrak{f}}(n)\right\},\ ] ] @xmath186 now the estimate is @xmath269 , since @xmath270 for each @xmath198 , define @xmath200 . if @xmath198 , following the previous ideas , we obtain that @xmath271 and therefore @xmath272 now we estimate the size of a covering of @xmath208 for each @xmath209 , take @xmath194 . as before , the set @xmath210 can be covered with @xmath211 balls of radio @xmath192 with @xmath273 .
once again , define @xmath222 , @xmath253 , @xmath238 and @xmath2 as before .
now the sets @xmath222 can be covered by less than @xmath274 balls of diameter at most @xmath245 .
now we can verify that , since each @xmath197 consist of @xmath275 tubes , we have that @xmath276
. we can also verify that the sequence @xmath277 satisfies the relation @xmath278 , and therefore we have the bound @xmath279 where @xmath280 stands for the lower box dimension .
finally , for any @xmath161 $ ] we have a line segment @xmath168 with slope @xmath164 contained in @xmath253 .
it follows that @xmath281 .
we remark that the argument in this particular result is essentially the same needed to obtain the family of furstenberg sets @xmath17 for @xmath116 , @xmath282 , such that @xmath283 announced in the introduction .
we will now look at the case of finite number of points in each direction .
let us begin with the definition of the class @xmath112 .
[ def : fk ] for @xmath284 , @xmath285 , a set will be a @xmath112-set or a furstenberg set of type @xmath286 if for any direction @xmath105 , there are at least @xmath286 points contained in @xmath2 lined up in the direction of @xmath5 .
this case contrasts with the previous case , since we show that in this case @xmath18 is not longer the smallest possible dimension . already in @xcite
we proved that there is a @xmath287-set with zero hausdorff dimension .
we will generalize this example to obtain even smaller @xmath287 sets .
namely , for any @xmath288 , there exists @xmath158 in @xmath287 such that @xmath289 .
it is clear that the set @xmath158 will depend on the choice of @xmath27 .
[ ex : f2haus_h0 ] given a function @xmath38 , we will construct two small sets @xmath290 $ ] with @xmath291 and such that @xmath292{\subseteq}e+f$ ] .
consider now @xmath293 .
clearly , we have that @xmath289 , and contains two points in every direction @xmath294 $ ] . for , if @xmath294 $ ] , let @xmath295 , so @xmath296 $ ] . by the choice of @xmath2 and @xmath253
, we can find @xmath297 and @xmath298 with @xmath299 .
the points @xmath300 and @xmath301 belong to @xmath158 and determine a segment in the direction @xmath302 .
we define @xmath303 : r_j=0 \text { if } m_k+1\le j\le m_{k+1 } ; k \text { even } \right\}\ ] ] and @xmath304 : r_j=0 \text { if } m_k+1\le j\le m_{k+1 } ; k \text { odd } \right\}.\ ] ] here @xmath305 is an increasing sequence such that @xmath306 .
now we estimate the size of the set @xmath2 . given @xmath307 , @xmath308 even , define @xmath309 .
it is clear that @xmath2 can be covered by @xmath310 intervals of length @xmath311 .
therefore , if the sequence @xmath312 increases fast enough , then @xmath313 since the same argument shows that @xmath314 , this estimate proves that the set @xmath158 has hausdorff dimension equal to zero .
now , for the finer estimate on the @xmath42-measure of the set , we must impose a more restrictive condition on the sequence @xmath315 .
recall that the covering property implies that , for a given @xmath38 , we have that @xmath316 therefore we need to choose a sequence @xmath317 , depending on @xmath27 , such that the above quantity goes to zero with @xmath308 . since @xmath318 , we can define recursively the sequence @xmath315 to satisfy the relation @xmath319 this last condition is equivalent to @xmath320 . as an concrete example , take @xmath321 . in this case
we obtain that the sequence @xmath315 can be defined as @xmath322 it is worthy to note here that if we were to measure the size of furstenberg sets with the packing dimension , the situation is absolutely different .
more precisely , for @xmath285 , any @xmath112-set @xmath323 must have @xmath324 . for
, if @xmath2 is an @xmath287 set , then the map @xmath267 defined by @xmath325 is lipschitz when restricted to @xmath326 .
roughly , we are considering the map that recovers the set of directions but restricted `` off the diagonal '' .
it is clear that we can assume without loss of generality that all the pairs are the endpoints of unit line segments .
therefore , since @xmath2 is an @xmath112-set , @xmath327 if @xmath328 is small enough .
we obtain the inequality @xmath329 the key point is the product formulae for hausdorff and packing dimensions .
we obtain that @xmath330 and then @xmath331 .
it also follows that if we achieve small hausdorff dimension then the packing dimension is forced to increase .
in particular , the @xmath287-set constructed in @xcite has hausdorff dimension 0 and therefore it has packing dimension 1 . by means of an example
we illustrate the relation between hausdorff and packing dimension for furstenberg sets . it can be understood as optimal in the sense of obtaining the smallest possible dimensions , both hausdorff and packing .
recall that from we have that for any @xmath332 , any furstenberg set @xmath2 of the class @xmath112 must satisfy @xmath333 in fact , there is an @xmath287 set @xmath2 such that @xmath334 .
[ ex : f2haus = pack=1/2 ] the construction is essentially the same as in , but we use two different sets to obtain all directions . let @xmath89 be the set of all the numbers whose expansion in base @xmath335 uses only the digits @xmath256 and @xmath336
. on the other hand , let @xmath136 the set of those numbers which only uses the digits @xmath256 and @xmath109 .
both sets have packing and hausdorff dimension equal to @xmath18 and @xmath292\subseteq a+b$ ] .
the construction follows then the same pattern as in the previous example .
the purpose of this section is to prove .
our proof relies on a variation of a jarnk type theorem on diophantine approximation .
we begin with some preliminary results on cantor type constructions that will be needed . in this section
we introduce the construction of sets of cantor type in the spirit of @xcite . by studying two quantities , the number of children of a typical interval and some separation property , we obtain sufficient conditions on these quantities that imply the positivity of the @xmath27-dimensional measure for a test function @xmath38
. we will need a preliminary elemental lemma about concave functions .
the proof is straightforward .
[ lem : lemaconcava ] let @xmath337 be a concave dimension function .
then @xmath338 we consider two separate cases : * if @xmath339 then @xmath340 . * if @xmath341 , then @xmath342 by concavity of @xmath27 .
the following lemma is a natural extension of the `` mass distribution principle '' to the dimension function setting .
[ lem : massh ] let @xmath51 be a set , @xmath38 and @xmath343 a probability measure on @xmath2 .
let @xmath92 and @xmath344 be positive constants such that for any @xmath345 with @xmath346 we have @xmath347 then @xmath348 .
for any @xmath167-covering we have @xmath349 then @xmath350 and therefore @xmath348 .
now we present the construction of a cantor - type set ( see example 4.6 in @xcite ) .
[ lem : cantorgeneral ] let @xmath351 be a decreasing sequence of closed subsets of the unit interval
. set @xmath352 $ ] and suppose that the following conditions are satisfied : 1 .
each @xmath353 is a finite union of closed intervals @xmath354 .
[ eq : mk ] each level @xmath355 interval contains at least @xmath312 intervals of level @xmath308 .
we will refer to this as the `` children '' of an interval .
3 . the gaps between the intervals of level @xmath308 are at least of size @xmath356 , with @xmath357 .
let @xmath358 .
define , for a * concave * dimension function @xmath38 , the quantity @xmath359 if @xmath360 , then @xmath348 .
the idea is to use the version of the mass distribution principle from .
clearly we can assume that the property of holds for _ exactly _ @xmath312 intervals .
so we can define a mass distribution on @xmath2 assigning a mass of @xmath361 to each of the @xmath362 intervals of level @xmath308 .
now , for any interval @xmath363 with @xmath364 , take @xmath308 such @xmath365 .
we will estimate the number of intervals of level @xmath308 that could have non - empty intersection with @xmath363 .
for that , we note the following : * @xmath363 intersects at most one @xmath366 , since @xmath367 . therefore it can intersect at most @xmath312 children of @xmath366 .
* suppose now that @xmath363 intersects @xmath368 intervals of level @xmath308
. then it must contain @xmath369 gaps of size at least @xmath356
. therefore , @xmath370 .
consequently @xmath371 intersects at most @xmath372 intervals of level @xmath308 . from these two observations ,
we conclude that @xmath373 now , by the concavity of @xmath27 , we obtain @xmath374 in addition ( also by concavity ) , @xmath27 is doubling , so @xmath375 and then @xmath376 finally , if @xmath377 , there exists @xmath378 such @xmath379 for @xmath380 and we can use the mass distribution principle with @xmath381 and @xmath382 . in the particular case of @xmath383 , @xmath384
we recover the result of @xcite , where the parameter @xmath52 can be expressed in terms of the sequences @xmath312 and @xmath356 .
for the set constructed in , we have @xmath385
the central problem in the theory of diophantine approximation is , at its simplest level , to approximate irrational numbers by rationals .
a classical theorem due to jarnk in this area is the following ( see @xcite ) , which provides a result on the size of the set of real numbers that are well approximable : [ thm : jarnikb ] for @xmath386 , define the following set : @xmath387\setminus { \mathbb{q } } : \|xq\|<\frac{1}{q^{\beta-1}}\text { for infinitely many } q\in{\mathbb{z}}\right\}.\ ] ] then @xmath388 .
in fact , there are several results ( see @xcite , @xcite ) regarding the more general problem of estimating the dimension partition for the set @xmath389\setminus { \mathbb{q } } : \|xq\|\le\frac{q}{{\mathfrak{g}}(q)}\text { for infinitely many } q\in{\mathbb{n}}\right\},\ ] ] where @xmath390 is any positive increasing function .
we can therefore see as a result on the size of a set of well approximable numbers .
we will derive the proof from the following proposition , where we prove a lower bound estimate for the set @xmath391 .
precisely , for @xmath392 we will find conditions on @xmath393 with @xmath394 to ensure that @xmath395 .
[ pro : bglow ] let @xmath390 be a positive , increasing function satisfying @xmath396 and @xmath397 define @xmath391 as in and let @xmath398 such that @xmath399 . consider a sequence @xmath400 that satisfies : * @xmath401 . *
@xmath402 .
if @xmath403 satisfies @xmath404 then @xmath395 .
define @xmath405\setminus { \mathbb{q } } : \|xq\|\le \frac{q}{{\mathfrak{g}}(q)}\}.\ ] ] for each @xmath406 , @xmath407 is the union of @xmath408 intervals of length @xmath409 and two more intervals of length @xmath410 at the endpoints of @xmath292 $ ] : @xmath411 now define for each @xmath412 @xmath413 now , for each @xmath174 consider two prime numbers @xmath414 such that @xmath415 .
we will prove that @xmath416 and @xmath417 are disjoint and well separated .
note that if @xmath418 and @xmath419 are centers of two of the intervals belonging to @xmath416 and @xmath417 , we have @xmath420 since @xmath421 .
therefore , taking into account this separation between the centers and the length of the intervals , we conclude that for @xmath422 and @xmath423 , @xmath424 let @xmath425 be the set of all the prime numbers between @xmath164 and @xmath209 and define @xmath426 then @xmath427 is the union of intervals of length at least @xmath428 that are separated by a distance of at least @xmath429 .
now we observe the following : if @xmath430 is an interval with @xmath431 , then at least @xmath432 of the intervals of @xmath433 are completely contained on @xmath430 . to verify this last statement ,
cut @xmath430 into three consecutive and congruent subintervals .
then , in the middle interval there are at least @xmath432 points of the form @xmath434 .
all the intervals of @xmath433 centered at these points are completely contained in @xmath430 , since the length of each interval of @xmath433 is @xmath435 .
in addition , by the prime number theorem , we know that @xmath436 , so we can find @xmath437 such that @xmath438 hence , if @xmath430 is an interval with @xmath431 , then there are at least @xmath439 intervals of @xmath427 contained on @xmath430
. now we will construct a cantor - type subset @xmath2 of @xmath391 and apply .
consider the sequence @xmath400 of the hypothesis of the proposition and let @xmath352 $ ] .
define @xmath440 as the union of all the intervals of @xmath441 contained in @xmath442 .
then @xmath353 is built up of intervals of length at least @xmath443 and separated by at least @xmath444 . moreover , since @xmath445 , each interval of @xmath442 contains at least @xmath446 intervals of @xmath353 .
now we can apply to @xmath447 . recall that @xmath356 denotes
the separation between @xmath308-level intervals and @xmath312 denotes the number of children of each of them .
consider @xmath448 .
then @xmath449 now we note that @xmath450 and , by hypothesis @xmath451 , we also have that @xmath452 . in addition , @xmath27 is doubling , therefore it follows that we can bound the first factor to obtain that @xmath453 since , by hypothesis @xmath454 , @xmath455 satisfies @xmath456 and @xmath390 satisfies .
now , again by hypothesis @xmath454 , @xmath457 thus , if @xmath458 then @xmath348 and therefore @xmath395 .
we also have an example to illustrate this last result .
[ ex : loglow ] define @xmath459 for @xmath460 and consider the set @xmath461 .
then @xmath462 will be an expected lower bound for the dimension function for the set @xmath461 .
consider the chain @xmath463 ( @xmath464 ) , which satisfy @xmath465 . in this context ,
define the sequence @xmath455 as follows : @xmath467 clearly the sequence is admissible , since * @xmath401 , and * @xmath402 .
inequality now becomes @xmath468 finally , for any @xmath92 and @xmath469 , @xmath455 satisfies , for large @xmath308 , @xmath470 so we conclude that @xmath377 .
therefore the set @xmath471 constructed in the proof above satisfies @xmath472 for all @xmath473 . for the proof of we will need a different but essentially equivalent formulation of jarnk s theorem .
we first recall that in the proof of we have used the following theorem ( see for example @xcite ) [ thm : jarnika ] let @xmath474 be a increasing sequence with @xmath475 for all @xmath476 . for @xmath477 ,
if @xmath478 is defined as @xmath479\setminus { \mathbb{q } } : \forall j\ \exists\ p , q\ ; q\le n_j^\alpha ; |x-\frac{p}{q}|<\frac{1}{n_j^2}\right\},\ ] ] then @xmath480 . for the proofs of and
, we refer the reader to @xcite , @xcite , @xcite , @xcite , and @xcite .
now we want to relate the sets @xmath481 and @xmath482 and their generalized versions .
it is clear that for any @xmath483 $ ] , we have the inclusion @xmath484 . for @xmath483
$ ] , if @xmath485 then for each @xmath476 there exists a rational @xmath486 with @xmath487 such that @xmath488 , which is equivalent to @xmath489 .
therefore @xmath490 .
observe that if there were only finite values of @xmath412 for a given @xmath491 , then @xmath491 has to be rational . for if @xmath492 for all @xmath493 , then @xmath494 and this implies that @xmath495 .
we conclude then that , for any @xmath485 , @xmath496 for infinite many @xmath412 and therefore @xmath497 .
however , since the dimension of @xmath478 coincides with the one of @xmath498 , one can expect that both sets have approximately comparable sizes .
we introduce the following definition , which is the extended version of the definition of the set @xmath478 in .
[ def : jarnikh ] let @xmath474 be any increasing nonnegative sequence of integers .
let @xmath499 be a increasing function defined on @xmath500 .
define the set @xmath501\setminus { \mathbb{q } } : \forall j\ \exists\ p , q\ ; q\le { \mathfrak{f}}(n_j ) ; |x-\frac{p}{q}|<\frac{1}{n_j^2}\right\}.\ ] ] the preceding observation about the inclusion @xmath502 can be extended to this general setting . for a given @xmath390 as in the definition of @xmath391 ,
define @xmath503 .
then the same calculations show that @xmath504 .
we will need a converse relation between those sets , since we want to prove a lower bound for the sets @xmath505 from the estimates provided in .
[ lem : eksubaf ] let @xmath390 and @xmath391 be as in .
define @xmath503 . then ,
if @xmath506 is the sequence defining the set @xmath2 in the proof of , then the set @xmath2 is contained in @xmath507 , where @xmath508 .
recall that in the proof of we define the sets @xmath509 as a union of intervals of the form @xmath510 .
the sets @xmath427 were defined as @xmath511 , where @xmath512 is the set of primes between @xmath209 and @xmath513 .
we can therefore write @xmath514 now , given a sequence @xmath506 , for each @xmath308 , the set @xmath353 is defined as the union of all the intervals of @xmath515 that belong to @xmath442 , where @xmath352 $ ] . if @xmath447 , any @xmath297 is in @xmath353 and therefore in some of the @xmath516 .
it follows that there exists integers @xmath517 and @xmath412 , @xmath518 such that @xmath519 therefore @xmath520 .
we remark that the above inclusion implies that any lower estimate on the size of @xmath2 would also be a lower estimate for @xmath521 .
we now conclude the proof of by proving : let @xmath262 . for @xmath257 , consider the function @xmath522 , the sequence @xmath506 and the set @xmath523 as in .
define @xmath524 , @xmath525 and @xmath526 as in .
it follows that @xmath260 , @xmath259 and @xmath501\setminus { \mathbb{q } } : \forall j\ \exists\ p , q\ ; q\le { \mathfrak{f}}(n_j ) ; |x-\frac{p}{q}|<\frac{1}{n_j^2}\right\}.\ ] ] note that says that the inclusion @xmath527 _ always _ holds , for any defining function @xmath390 , where @xmath2 is the _ substantial portion _ of the set @xmath391 ( see ) .
but we need the positivity of @xmath528 to conclude that the set @xmath507 also has positive @xmath268-measure . for the precise choices of @xmath102 and @xmath390
, we obtain this last property from .
precisely , @xmath529 and therefore the set @xmath505 has positive @xmath268-measure .
this concludes the proof of and therefore the set constructed in the proof of fulfills the condition of being an @xmath104-set .
we would like to thank to pablo shmerkin for many interesting suggestions an remarks about this problem . | for @xmath0 in @xmath1 $ ] , a subset @xmath2 of @xmath3 is called _ furstenberg set _ of type @xmath0 or @xmath4-set if for each direction @xmath5 in the unit circle there is a line segment @xmath6 in the direction of @xmath5 such that the hausdorff dimension of the set @xmath7 is greater or equal than @xmath0 . in this paper
we show that if @xmath8 , there exists a set @xmath9 such that @xmath10 for @xmath11 , @xmath12 , which improves on the the previously known bound , that @xmath13 for @xmath14 .
further , by refining the argument in a subtle way , we are able to obtain a sharp dimension estimate for a whole class of _
zero_-dimensional furstenberg type sets .
namely , for @xmath15 , @xmath16 , we construct a set @xmath17 of hausdorff dimension not greater than @xmath18 .
since in a previous work we showed that @xmath18 is a lower bound for the hausdorff dimension of any @xmath19 , with the present construction , the value @xmath18 is sharp for the whole class of furstenberg sets associated to the zero dimensional functions @xmath20 .
[ section ] [ theorem]lemma [ theorem]corollary [ theorem]proposition [ theorem]claim [ theorem]conjecture [ theorem]definition [ theorem]example [ theorem]remark |
recently a novel paradigm was suggested for the design of quantum algorithms for solving combinatorial search and optimization problems based on quantum adiabatic evolution @xcite . in the quantum adiabatic evolution algorithm ( qaa )
a quantum state is closely following a ground state of a specially designed slowly time - varying control hamiltonian @xmath1 . at the beginning of the algorithm
the control hamiltonian @xmath4 has a simple form with a known ground state that is easy to prepare , and at the final moment of time it coincides with the `` problem '' hamiltonian @xmath5 which ground state encodes the solution of the classical optimization problem in question @xmath6 here @xmath7 is a cost function defined on a set of @xmath8 binary strings @xmath9 @xmath10 , each containing @xmath0 bits .
the summation in ( [ hp ] ) is over the @xmath8 states @xmath11 forming the computational basis of a quantum computer with @xmath0 qubits .
state @xmath12 of the @xmath13-th qubit is an eigenstate of the pauli matrix @xmath14 with eigenvalue @xmath15 .
if at the end of the qaa the quantum state is sufficiently close to the ground state of @xmath5 then the solution to the optimization problem can be retrieved by the measurement .
it has been shown recently @xcite that the query complexity argument that lead to the exponential lower bound for the unstructured search @xcite can not be used to rule out the polynomial time solution of np - complete satisfiability problem by the quantum adiabatic evolution algorithm ( qaa ) .
a set of examples of the 3-satisfiability problem has been recently constructed @xcite to test analytically the power of qaa . in these examples
the cost function @xmath7 depends on a bit - string @xmath16 with @xmath0 bits , @xmath17 , only via a hamming weight of the string , @xmath18 , so that @xmath19 where the function @xmath20 is in general non - monotonic and defines a particular instance of this `` hamming weight problem '' ( hwp ) . in @xcite the original version of qaa @xcite
was applied to the hwp where the control hamiltonian is a linear interpolation in time between the initial and final hamiltonians . in this case , it was shown @xcite that the system can be trapped during the qaa in a local minimum of the cost function for a time that grows exponentially in the problem size @xmath0 . it was also shown @xcite that an exponential delay time in the quantum adiabatic algorithm can be interpreted in terms of the quantum - mechanical tunnelling of an auxiliary large spin between the two intermediate states . the above example has a significance greater than just being a particular simplified case of a binary optimization problem with symmetrized cost .
indeed , one can argue that it shows a generic mechanism for setting `` locality traps '' in the 3-satisfiability problem @xcite .
but most importantly , this example demonstrates that exponential complexity of qaa can result from a _ collective phenomenon _ in which transitions between the configurations with low - lying energies can only occur by simultaneous flipping of large clusters containing order - n bits . in spin glasses
, there is typically an exponential number of such configurations , the so - called local ground states
. a similar picture may be applicable to random satisfiability problems @xcite . in some cases
, these transitions can be understood and described in terms of macroscopic quantum tunnelling .
a tunnelling of magnetization was observed in large - spin molecular nanomagnets @xcite and in disordered ferromagnets @xcite .
the paper @xcite suggests that large tunnelling barriers can be avoided in qaa by using multiple runs of qaa with realizations of the control hamiltonians @xmath1 sampled from a random ensemble .
this ensemble is chosen in a sufficiently simple and general form that does not depend on the specific instance of the optimization problem .
different hamiltonians @xmath1 correspond to different paths of the unitary evolution that begin and end in the same initial and final states ( modulus phase factors ) .
the complexity of qaa with different paths for the hwp was tested numerically in @xcite using an ensemble of random 8@xmath218 matrices .
the results indicate that the hwp may be solved in polynomial time with finite probability . in case when the random paths @xmath1 preserve the bit - permutation symmetry of the problem it is natural to describe the random ensemble of @xmath22 in terms of the dynamics of a spin-@xmath3 system .
this approach allows for a general theoretical analysis of the algorithm . in the present paper ,
we perform this analysis for the random version of hwp ( over - constrained 3-satisfiability problem ) by mapping the dynamics of qaa onto the motion of a quantum particle in a 1d effective potential .
this allows us to compute the statistical weight of the successful evolution paths in the ensemble and provide a complete characterization of such paths .
in a qaa with different paths @xcite , one specifies the time - dependent control hamiltonian @xmath23 @xmath24 where the control parameter @xmath25 plays the role of dimensionless time .
this hamiltonian guides the quantum evolution of the state vector @xmath26 according to the schrdinger equation @xmath27 from @xmath28 to @xmath29 , the _ run time _ of the algorithm .
@xmath5 is the `` problem '' hamiltonian given in ( [ hp ] ) . @xmath30 and @xmath31 are ` driver' hamiltonians designed to cause the transitions between the eigenstates of @xmath5 . an initial state of the system @xmath32 is prepared as a ground state of the initial hamiltonian @xmath4 .
it is typically constructed assuming @xmath33 knowledge of the solution of the classical optimization problem and related ground state of @xmath5 . in the simplest case @xmath34 where @xmath35 is a pauli matrix for @xmath36-th qubit and @xmath37
is some scaling constant .
the ground state of @xmath38 has equal projections on any of the @xmath8 basis states @xmath39 ( [ basis ] ) .
consider instantaneous eigenstates @xmath40 of @xmath1 with corresponding eigenvalues @xmath41 arranged in non - decreasing order at any value of @xmath42@xmath43 provided the value of @xmath44 is large enough and there is a finite gap for all @xmath45 between the ground and exited state energies , @xmath46 , quantum evolution is adiabatic and the state of the system @xmath26 stays close to an instantaneous ground state , @xmath47 ( up to a phase factor ) . because @xmath48 the final state @xmath49 is close to the ground state @xmath50 of the problem hamiltonian .
therefore a measurement performed on the quantum computer at @xmath51 will find one of the solutions of combinatorial optimization problem with large probability .
quantum transition away from the adiabatic ground state occurs most likely in the vicinity of the point @xmath52 where the energy gap @xmath53 reaches its minimum ( avoided - crossing region ) .
the probability of the transition is small provided that @xcite @xmath54 where @xmath55 , \label{mingap1}\end{aligned}\ ] ] the r.h.s . in eq .
( [ mingap ] ) gives an upper bound estimate for the required runtime of the algorithm and the task is to find its asymptotic behavior in the limit of large @xmath56 .
the numerator in ( [ mingap ] ) is of the order of the largest eigenvalue of @xmath57 , which typically scales polynomially with @xmath0 .
however , @xmath58 can scale down exponentially with @xmath0 and in such cases the required runtime of the quantum adiabatic algorithm to find a solution grows exponentially fast with the size of the input .
one should note that the second term in the r.h.s . of ( [ h_tot ] ) is zero at @xmath59 and @xmath60 .
therefore , by using different driver hamiltonians @xmath31 one can design a family of ( possibly random ) adiabatic evolution paths that start at @xmath59 in the same generically chosen initial state and arrive at the ground state of @xmath5 at @xmath60 . in general , different paths will correspond to different minimum gaps @xmath61 and one can introduce the distribution of minimum gaps .
this distribution can be used to compute the fraction of the adiabatic evolution paths @xmath62 that arrive at the ground state of @xmath5 within polynomial time , @xmath63 for a successfully designed family of paths the fraction @xmath62 is bounded from below by a polynomial in @xmath64 which leads to the average polynomial complexity of qaa .
consider a binary optimization problem defined on a set of @xmath0-bit strings @xmath65 with the cost function @xmath7 in the following form : @xmath66 this cost is symmetric with respect to the permutation of bits , it depends on a string @xmath16 only through the number of unit bits in the string @xmath67 ( the hamming weight ) . in this paper
we consider the cost function ( [ symcost ] ) in the following form which is generalization of the cost introduced in @xcite @xmath68 here the sum is over all possible 3-bit subsets of the @xmath0-bit string @xmath65 .
a subset @xmath69 contributes to the total cost a weight factor @xmath70 where @xmath13 is a number of units bits in the subset .
a set of weights @xmath71 defines an instance of this generalized hamming weight problem ( hwp ) .
one can formulate a random version of hwp , e.g. , by drawing numbers @xmath71 independently from a uniform distribution defined over a certain range . in the limit of large @xmath56
the cost function ( [ 3sat_gen ] ) takes the following form : @xmath72 here @xmath73 and we only keep the terms of the leading order in @xmath0 . the coefficients @xmath74 in ( [ gp ] )
are linear combinations of @xmath75 @xmath76+\frac{1}{6 } \binom{3}{% k}\left[p_0+(-1)^k p_3\right ] .
\label{beta_1}\ ] ] here @xmath77 for @xmath78 and @xmath79 for @xmath80 .
the function @xmath81 in ( [ gp ] ) is a third degree polynomial in @xmath82 , and the form of the function depends on the coefficients @xmath83 ( @xmath84 ) . it is easy to show that there is a finite size region in the parameter space @xmath71 where @xmath81 is a non - monotonic function of @xmath82 that has global and local minima on the interval @xmath85 .
those minima are separated by a finite barrier with width @xmath86 .
the barrier separates strings that have close values of the cost @xmath7 but are at large hamming distance from each other : they have @xmath87 distinct bits .
this property can lead to exponentially small minimum gaps in qaa due to the onset of low - amplitude quantum tunnelling @xcite . )
@xmath88 _ vs _
@xmath82 for different choices of the weights @xmath71 .
curve @xmath89 corresponds to @xmath90 @xmath91 @xmath92 @xmath93 , and the cost function @xmath81 has a global minimum at @xmath94 , corresponding to the string @xmath16 with the hamming weight zero , @xmath95 .
it also has a local minimum at @xmath96 corresponding to the bit string with hamming weight @xmath0 , @xmath97 .
the curve @xmath89 yields the particular form of the cost function @xmath81 considered in @xcite , @xcite .
curve @xmath98 corresponds to @xmath99 @xmath100 @xmath101 @xmath102 , it has a global minimum at @xmath103 inside of the interval @xmath104 .
this minimum corresponds to approximately @xmath105 bit strings @xmath16 that all have the same hamming weight @xmath106.,width=316 ]
it is natural to consider the control hamiltonians ( [ h_tot ] ) for solving the hwps that are symmetric with respect to permutation of individual bits ( [ basis ] ) . in
what follows , we use the normalized components of the total spin operator @xmath107 for the system of @xmath0 individual spins-@xmath108 @xmath109 here @xmath110 are the projections of the total spin operator on the @xmath36-th axis ( @xmath111 ) and @xmath112 are pauli matrices for the @xmath113-th spin . for the sake of bookkeeping , in ( [ s ] ) and also throughout the paper we use `` hats '' for the spin operators , such as @xmath114 , @xmath115 , and some others , in order to distinguish them from their corresponding eigenvalues ( @xmath116 and @xmath117 , respectively , in the above example ) . to obtain the problem hamiltonian ( [ hp ] )
we make use of the obvious connection between the values of the hamming weight function @xmath118 of an @xmath0-bit string @xmath16 and corresponding eigenvalues @xmath119 of the spin projection operator @xmath120 @xmath121 then from eqs .
( [ basis]),([gp ] ) and ( [ sz ] ) we obtain @xmath122 we chose the driver @xmath38 in a bit - symmetric form that coincides with ( [ hd ] ) ( up to a constant term)@xmath123 it was proposed in @xcite that @xmath31 can be constructed using some generic ensemble of random matrices .
the bit - symmetric random drivers for the cost functions of the type ( [ 3sat_gen ] ) can be constructed as follows @xcite .
one generates an @xmath125 random hermitian matrix @xmath126 with zero diagonal elements and non - diagonal elements that are independent random numbers identically distributed in a certain interval .
matrix elements of @xmath127 can be enumerated by all possible configurations of a 3-bit string @xmath128 .
then @xmath31 takes the form @xmath129 here @xmath130 are computational basis states eq .
( [ hp ] ) corresponding to bit - strings @xmath9 , and string @xmath131 has three of its bits flipped at the positions @xmath113,@xmath36 and @xmath13 as compared to the string @xmath132 ( i.e. , @xmath133 , etc ) .
each randomly selected @xmath126 generates @xmath31 and therefore a random path modification of the qaa . from the above discussion
, it follows that the matrix of the operator @xmath31 ( [ hea ] ) is symmetric with respect to the bit permutations and therefore it commutes with the operator of a total spin @xmath134 of a system of @xmath135 spins @xmath136 .
this means that @xmath31 acts independently in each of the sub - spaces corresponding to certain values of the total spin @xmath137 @xcite .
it follows from ( [ h_p_z ] ) and ( [ hdfarhi ] ) that the same is true for the total control hamiltonian ( [ h_tot ] ) @xmath138 = 0,\quad \tau \in ( 0,1 ) .
\label{commut}\ ] ] since in our case the initial state ( [ hd ] ) is a totally symmetric combination of all states and therefore corresponds to the maximal spin @xmath139 , our system always stays in this sub - space during the algorithm .
therefore in the analysis of the complexity of qaa one can reduce the @xmath140 matrix of @xmath1 to the @xmath141 matrix that only involves the states with different spin projections of the _ maximum _ total spin @xmath139 .
binary strings corresponding to the quantum states from this subspace are distinguished from each other by their hamming weight only . in appendix
[ matrix ] , we show that in the case of real - valued symmetric matrices @xmath126 and in the large - spin limit , the bit - symmetric driver @xmath31 ( [ hea ] ) can be presented as a linear combination of 6 operators expressed in terms of the large spin operator components @xmath142 acting in the subspace with @xmath139 . using this fact , and also eqs .
( [ h_p_z ] ) and ( [ hdfarhi ] ) one can write a bit - symmetric control hamiltonian ( [ h_tot ] ) in the following form @xmath143 @xmath144 where @xmath145 ( @xmath146 ) are independent real coefficients given in eq .
( [ gammas_1 ] ) . as we show in appendix [ matrix ] , any random realization of the real matrix @xmath126 can be mapped onto combinations of drivers ( [ h_e_3 ] ) by the appropriate choice of the real coefficients @xmath147 .
we note that @xmath124 in eq .
( [ h_e_3 ] ) does not have any terms involving @xmath148 operator .
the reason for that is that we chose matrix elements of @xmath149 ( [ hea ] ) to be real numbers .
then matrix elements of @xmath124 in @xmath132 basis are real as well . in this case @xmath124 can only involve terms with * even * powers of @xmath150 . in ( [ h_e_3 ] ) we have used a conservation of the total spin ( see discussion above ) and substituted @xmath151 . the form of the total hamiltonian in ( [ h_tot_4 ] ) allows us to analyze the minimum gap in qaa with different paths ( [ h_tot ] ) using the wkb analysis of the dynamics of a spin-@xmath152 in the large spin limit ( @xmath56 ) .
our analysis in this section is a particular application of the wkb - type approach commonly used for the description of quantum spin tunnelling in magnetics @xcite , @xcite , @xcite , @xcite .
this approach is applicable for the large spins @xmath153 , which is the case of interest for us .
we choose @xmath154 as a quantization axis and following the standard procedure to obtain the effective quasi - classical hamiltonian in polar coordinates @xmath155 with @xmath156 $ ] and @xmath157 $ ] .
we make use of the villain transformation @xcite @xmath158 where azimuthal angle operator @xmath159 satisfies the commutation relation @xmath160 = i\epsilon .
\label{comm1}\ ] ] in a change of notation we introduce a coordinate @xmath82 and canonically - conjugate momentum @xmath161 ( cf .
@xcite ) @xmath162 ( @xmath163 ) . expanding ( [ villain_1 ] ) in the large spin limit @xmath164
, we obtain @xmath165 finally , we write the scaled hamiltonian of the system ( [ h_tot_4 ] ) in terms of the new variables @xmath166 where @xmath167 @xmath168 = i\,\epsilon , \label{[qp]}\ ] ] and @xmath169 is a small correction @xmath170 ( here @xmath171 has the same arguments as @xmath172 in ( [ hqp ] ) ) . the stationary schrdinger equation ( [ adiab ] ) in the new basis @xmath173 can be solved in the wkb approximation with the small parameter @xmath174 playing the role of a plank constant
. then the wave function @xmath175 takes the form @xmath176 , \label{wkb}\ ] ] where in the leading order in @xmath177 the action function @xmath178 satisfies the following hamilton - jacobi equation @xmath179 this equation describes a 1d auxiliary mechanical system with coordinate @xmath82 , momentum @xmath180 , energy @xmath181 , and hamiltonian function @xmath182 .
classical orbits satisfy the hamiltonian equations @xmath183 where @xmath184 and @xmath185 stand for the partial derivatives of @xmath186 with respect to @xmath82 and @xmath180 , respectively .
stationary points of the dynamics @xmath187 correspond to the elliptic and saddle points of the hamiltonian function @xmath188 elliptic points are minima ( or maxima ) of @xmath189 on the ( @xmath190 ) plane .
they satisfy the condition @xmath191 where @xmath192 is understood as a second derivative of @xmath186 with respect to @xmath180 , etc .
saddle points correspond to @xmath193 in ( [ d ] ) . in the limit @xmath164 the adiabatic ground state @xmath194 ( [ schh ] )
is localized in the small vicinity of the fixed points @xmath195 corresponding to the global minimum of @xmath196 at a given value of @xmath25 . to logarithmic accuracy
the wkb - asymptotic ( [ wkb ] ) of the ground - state wave function is determined by the mechanical action for the imaginary - time instanton trajectory @xmath197 emanating from the fixed point @xmath187 @xmath198 , \notag \\ & & q(-i\infty ) = q_{\ast } , \quad p(-i\infty ) = p_{\ast } , \quad q(0)=q . \label{inst}\end{aligned}\ ] ] integration in ( [ inst ] ) is along the imaginary axis @xmath199 .
the instanton trajectory obeys eq .
( [ ham_eq ] ) with the boundary conditions given above and @xmath200 corresponding to the line of integration in ( [ inst ] ) .
the choice of the final instant , @xmath28 , is arbitrary since the instanton trajectory is degenerate with respect to a shift of the time axis .
we note that the wkb asymptotic ( [ inst ] ) decays exponentially fast as the coordinate @xmath82 in ( [ inst ] ) moves away from its value at the global minimum @xmath201 into the classically inaccessible region .
this corresponds to the growth of the imaginary part of the action in ( [ inst ] ) , similar to the conventional quantum tunnelling in the potential . in the vicinity of @xmath187
the ground - state wave function @xmath202 takes the form similar to that of harmonic oscillator : @xmath203 , \notag \\ & & m^{\ast } = \frac{1}{|h_{pp}(q_{\ast } , p_{\ast } , \tau ) | } , \label{osc}\end{aligned}\ ] ] here @xmath204 is defined in ( [ d ] ) .
similarly , the energy spectrum in that region corresponds to the classical elliptic orbits with oscillation frequency @xmath205 @xmath206 we note that the frequency @xmath205 depends on @xmath25 and determines the time - varying instantaneous gap between the ground and first exited states , @xmath207 .
it can be seen from eq .
( [ hqp ] ) that the global minimum of @xmath196 will correspond to @xmath208 ( @xmath209 ) provided that the following condition holds for all @xmath210 : @xmath211 where the positive and negative signs of @xmath212 correspond to even and odd values of @xmath13 , respectively .
the value of @xmath201 in ( [ dgdnx ] ) corresponds to the global minimum of the effective potential @xmath213 @xmath214 under the above condition the hamiltonian function of the system near the global minimum @xmath215 exactly corresponds to that of the harmonic oscillator with effective frequency @xmath205 ( [ d ] ) and mass @xmath216 ( [ osc ] ) @xmath217 in the wkb picture the ground state of the system correspond to the particle performing zero - level oscillations near the bottom of the slowly varying potential @xmath213 .
there are two types of the bifurcations that can destroy the above adiabatic picture : assume that at some instant of time @xmath218 the effective mass @xmath219 goes to infinity . in the vicinity of this point the hamiltonian function ( [ hqp ] )
can be approximated as follows : @xmath220 where @xmath221 in the above equations all functions are evaluated at the point @xmath222 .
equation ( [ locb ] ) corresponds to @xmath223 bifurcation point @xcite .
it can be seen from ( [ locb ] ) that for @xmath224 the single global minimum of @xmath189 splits into the two minima with nonzero momenta @xmath225 due to the symmetry @xmath226 the two global minima with nonzero @xmath227 will stay symmetric with respect to the @xmath82-axis at later times .
it follows from ( [ d ] ) , ( [ osc1 ] ) ) that the linear oscillation frequency vanishes at the bifurcation point , @xmath228 , however the energy gap @xmath229 . by solving the schrdinger equation ( [ adiab ] ) at this point in the representation of the momentum
@xmath180 one can find the eigenfunctions @xmath230 and eigenvalues @xmath231 corresponding to a 1d quantum system moving in a quartic potential ( cf .
( [ locb ] ) ) .
this analysis yields an estimate for the value of the gap , and the characteristic localization range @xmath232 for @xmath233 @xmath234 the size of the energy barrier in momentum @xmath180 separating the two global minima in ( [ locb ] ) grows with time for @xmath235 and this leads to a rapid decrease of the energy gap . sufficiently far from the bifurcation point ,
@xmath236 , each of the global minima @xmath237 gives rise to its own wkb asymptotic ( [ inst ] ) localized at the minimum . the ground state and the first exited state correspond to their symmetric and anti - symmetric combinations , respectively @xmath238 for @xmath236
the tunnelling splitting of energy levels for the symmetric and antisymmetric states determines the value of the gap @xmath53 and decreases exponentially fast with @xmath235 .
away from the bifurcation region , @xmath239 , the gap scales down exponentially with @xmath0 ( note that @xmath240 ) . as a result of the local bifurcation , the purely adiabatic evolution in qaa collapses .
the amplitude of staying in the adiabatic ground state for @xmath241 is nearly equally split between the states ( [ split ] ) with the two lowest eigenvalues . in general
, this may reduce the probability of finding a system in a ground state at @xmath60 by a factor of 2 .
we note that the control hamiltonian ( [ h_tot_4 ] ) is at most a cubic polynomial in @xmath210 , @xmath119 , and therefore the number of local bifurcation events during qaa is of the order of one . in the worst case
they will cause the reduction of the success probability in qaa by a constant factor . for a given instance of the cost function ( [ gp ] ) defined by the coefficients @xmath74 ( or @xmath75 ) the onset of local bifurcations ( [ locb ] ) depends on the choice of the driver hamiltonian @xmath124 ( [ h_e_3 ] ) .
there are a number of ways to select coefficients @xmath147 s in the driver hamiltonian ( [ h_e_3 ] ) to avoid local bifurcations during qaa in a broad range of values of the coefficients @xmath70 .
for example , to completely suppress local bifurcations ( [ locb ] ) one can keep in ( [ h_e_3 ] ) only terms linear in @xmath242 and set @xmath243 the hamiltonian function @xmath244 defines a 3d surface over a 2d plane @xmath245 and the shape of this surface varies with time @xmath25 .
we consider global bifurcations of this surface where the energies of its two minima cross each other at some instant of time @xmath218 while the distance between the minima on the @xmath245 plane remains finite at the crossing point .
for @xmath246 the minima exchange their roles : global minimum becomes local and vise versa .
before and after the intersection in the energy space the two minima are uniquely identified with the ground and first exited states of the system s hamiltonian ( [ hqp ] ) .
the corresponding wave functions @xmath247 are well approximated by their asymptotic expressions([inst]),([osc ] ) .
the small vicinity of the global bifurcation point can be described within the standard 2-level avoiding - crossing picture .
there @xmath248 are given by symmetric and antisymmetric superpositions of the wkb - asymptotic corresponding to intersecting minima .
the value of the gap changes with time as @xmath249 where @xmath250 is some constant and the minimum gap is determined by the overlap of the wkb asymptotic . to logarithmic accuracy
it is given by the imaginary part of the mechanical action ( [ inst ] ) along the instanton trajectory connecting the two minima @xmath251 @xmath252 here @xmath253 are coordinates of the two minima ; @xmath254 , and the instanton trajectory obeys the eqs .
( [ ham_eq ] ) . the analytical expression for the minimum gap
was studied in @xcite , @xcite for the case @xmath255 , using a simplified version of the hamming weight problem ( [ 3sat_gen ] ) .
below we identify certain geometrical properties of the global bifurcations in the case @xmath255 that will be used later in the selection of the drivers @xmath124 for the successful qaa .
@xmath82 for @xmath256 , @xmath257 and @xmath258 are represented by the curves 1,2 , and 3 , respectively.,width=316 ] in the case @xmath259 @xmath260 , the hamiltonian has a minimum at @xmath261 and the value of @xmath262 corresponds to the global minimum of the effective potential @xmath213 ( [ u ] ) .
we use eq .
( [ h_tot_4 ] ) and also the condition @xmath263 to obtain the following equation for @xmath262 @xmath264 this equation holds until the global bifurcation point at @xmath265 where @xmath262 changes discontinuously in time ( see fig . 2 ) . at the minimum of the potential @xmath266 and therefore the direction of the motion of @xmath262 entirely depends on the direction of the `` force '' , @xmath267 . at @xmath59
the potential @xmath268 has a unique minimum at the point @xmath269 .
it is clear that with this initial condition equation ( [ dq*dt ] ) can lead to a `` wrong '' minimum of @xmath81 that lies above the global minimum , and such cases will give rise to a global bifurcation .
this effect is illustrated in fig.[fig : cost ] where the two different cost functions correspond to the same direction of motion for @xmath270 .
the value of @xmath262 may either smoothly approach the global minimum of the cost ( curve 2 ) , or move toward a wrong local minimum ( curve 1 ) , leading to the global bifurcation and exponentially small gap in qaa .
adding @xmath31 to the control hamiltonian can invert the direction of motion of @xmath262 toward the global minimum of @xmath271 .
this can be seen from the fact the eq.([dq*dt ] ) in presence of @xmath31 possesses the additional term @xmath272 ( here we drop for sake of brevity the argument @xmath25 in @xmath270 ) .
clearly , the successful @xmath273 should _ not _ possess reflection symmetry with respect to @xmath119 .
therefore we should only select the terms in ( [ h_e_3 ] ) that contain odd powers of @xmath119 .
taking into account ( [ avoid ] ) we arrive at the following form of the driver hamiltonian @xmath274 this driver can remove the potential barrier between the two competing global minima of @xmath213 by shifting the original minimum at @xmath275 towards the true global minimum of the cost function @xmath81 ( cf .
[ fig : cost ] ) . in the classical picture ( [ hqp ] )
the driver ( [ gamma4 ] ) corresponds to an external field parallel to @xmath154-axis which can destroy the tunnelling barrier along this direction .
the mechanism of such tunnelling avoidance is similar to the one considered in @xcite , where the external field generated by the driver ( [ gamma4 ] ) compensates the effective field due to the linear term proportional to the coefficient @xmath276 in the problem hamiltonian ( [ gp ] ) . in general
, one can expect that a complete suppression of the tunnelling barrier at all values of @xmath25 requires a certain magnitude ( and sign ) of the coefficient @xmath277 depending on the choice of the coefficients @xmath278 in the cost function @xmath81 .
the transition to the tunnelling regime can be described as an @xmath223 bifurcation point , illustrated in fig.2 .
the effective potential @xmath279 changes parametrically with @xmath280 and @xmath281 . near the bifurcation point @xmath282
, the potential has the form @xmath283 where @xmath284 are deviations from the bifurcation point in @xmath285 and @xmath277 , respectively .
the corresponding conditions for the @xmath223 bifurcation point are : @xmath286 taking into account ( [ u ] ) and ( [ h_tot_4]),([h_e_3]),([gamma4 ] ) , the above equation yields @xmath287 ^{2 } } , \label{bifurcation_1 } \\
\gamma _ { 4c}\left ( 1-\tau _ { c}\right ) & = & \frac{\left ( 1-\tau _ { c}\right ) \left ( 3\beta _ { 1}+\beta _ { 3}\right ) + \tau _ { c}\beta _ { 2}\beta _ { 3}}{\tau _ { c}\beta _ { 2}-2\left ( 1-\tau _ { c}\right ) } , \notag\end{aligned}\ ] ] these equations should be solved for @xmath288 and @xmath289 for the given set of the coefficients @xmath83 .
the bifurcation is avoided when @xmath290 for example , in the particular case of the hwp ( [ 3sat_gen ] ) considered in @xcite , @xcite , we have @xmath291 in this case the example of the driver hamiltonian @xmath31 that allows to avoidance of tunnelling in qaa was given in @xcite where the value of @xmath292 was used . according to ( [ avoid_farhi ] )
this value is way below the critical value @xmath288 .
we performed numerical simulations with the effective potential ( [ u ] ) checking for the onset of tunnelling for all @xmath293 $ ] , @xmath294 .
the numeric simulations confirm that the situation discussed above is typical for the general hwp , implying that ( [ gamma4 ] ) is the only driver term that can be fundamentally responsible for the tunnelling avoidance in a general case , if the coefficient @xmath277 is defined appropriately .
in particular , one of the two drivers ( [ gamma4 ] ) with @xmath295\quad k=0,1,2,3 , \label{avoid3}\end{aligned}\ ] ] always suppresses tunnelling in the qaa . _ vs _ domain size @xmath296.,width=316 ] we also solve the eqs .
( [ bifurcation_1 ] ) numerically for coefficients @xmath297 taking values on a dense grid of points in the cube @xmath298 $ ] ( @xmath13=0,1,2,3 ) . for each size of the cube @xmath296 we select the point with largest value of @xmath288 denoted below as @xmath299 .
the results are presented in fig.3 .
the critical value @xmath300 is monotonically increasing in @xmath296 , and the dependence is close to linear for sufficiently large @xmath296 , but it is non - linear in the range @xmath301 . it can be inferred from eq .
( [ bifurcation_1 ] ) that a nonlinear dependence of @xmath300 on the scale @xmath296 is due to the fact that the critical time @xmath289 also depends on @xmath296 .
the linear dependence of @xmath302 for large @xmath296 has a simple intuition . according to ( [ gp ] ) and ( [ beta_1 ] ) ,
the magnitude of @xmath81 is proportional to @xmath296 . according to eq.([dq*dt ] )
, the maximal magnitude of the coefficient @xmath276 presents a force that can possibly move a system into the local minimum at small @xmath25 . from ( [ beta_1 ] ) , we conclude that @xmath303}\beta _ { 1}=2l$ ] . in the limit of large @xmath296 ,
the role of the driver @xmath304 in ( [ bifurcation_1 ] ) becomes unimportant .
therefore , the only competing terms are the driver @xmath273 and the problem hamiltonian @xmath305 . the term ( [ avoid_ad ] ) generated by @xmath31 compensates ( [ avoid_ad ] ) the force
@xmath276 when @xmath306 , and therefore in this limit we have @xmath307}\gamma _ { 4c}\approx \left| \beta _ { 1}\right| _ { \max } = 2l
. \label{largel}\ ] ] one should note that among the effective potentials generated by choosing different @xmath308 , there are two subsets that can be mapped onto each other by means of the mirror reflection about the @xmath82-axis , @xmath309 .
we note that the same driver @xmath31 can not simultaneously suppress tunnelling barriers in each of the two mutually symmetric potentials : if the tunnelling barriers are not suppressed with @xmath310 , they will be suppressed with @xmath311 , and vice versa .
this gives a simple intuition for the tunnelling barrier suppression boundary ( [ avoid_gamma4 ] ) .
finally we conclude , that it is possible to indicate the range of value of @xmath312 will play the role of a * universal * driver that guarantees polynomial performance of the qaa for all instances of the generalized hamming weight problem ( [ 3sat_gen ] ) provisory to the mirror - reflection symmetry in the possible choice of the cost functions and the common normalization factor @xmath296 .
using the analysis from the previous section one can estimate the probability of success for the qaa with random paths proposed in @xcite . in that algorithm ,
the ensemble of random drivers @xmath31 was generated using random 3@xmath3133 matrices @xmath314 ( [ hea ] ) .
it is shown in appendix [ matrix ] that for the bit - symmetric optimization problem ( [ 3sat_gen ] ) the above ensemble is identical to the ensemble of independent uniformly - distributed random coefficients @xmath315 ( k=@xmath316 ) that appears in the large - spin representation of the driver @xmath317 ( [ h_e_3 ] ) .
then for any instance of optimization problem in ( [ 3sat_gen ] ) defined by the set of the coefficients @xmath71 one should compute the fraction @xmath318 of the domain of the coefficients @xmath319 where the following conditions are satisfied : 1 .
condition for the nonzero effective mass ( [ dgdnx ] ) .
2 . the condition ( [ avoid_gamma4 ] ) for the complete avoidance of the tunnelling barriers in combination with eq.([bifurcation_1 ] ) for the bifurcation boundary . here
we compute the fraction @xmath62 for the particular instance of the optimization problem ( [ 3sat_gen ] ) considered in @xcite . in this case
( [ dgdnx ] ) takes the form @xmath320 \neq 0 , \label{effmass_2}\end{aligned}\ ] ] where @xmath321 and @xmath201 provides global minimum of @xmath213 ( [ u ] ) .
the effective mass is non - zero if @xmath322 , and ( [ effmass_2 ] ) yields an estimate on the range of @xmath323 as @xmath324 following @xcite we assume that the non - diagonal matrix elements @xmath314 are distributed in the interval @xmath325 $ ] .
making use of ( [ gammas_1 ] ) , we obtain@xmath326 therefore , the probability that inequality ( [ range ] ) is satisfied is estimated as 1 - 15@xmath327 0.71875 .
on the other hand , the values of @xmath277 in ( [ gammas_1 ] ) belong to the range , @xmath328 . using the value of
@xmath329 -0.95 given in ( [ avoid_farhi ] ) we estimate the probability of @xmath330 to be approximately equal to @xmath331 0.46 . making an approximation that the cases when the effective mass is non - zero are statistically independent from the cases when @xmath330 , we obtain the total probability of success as @xmath3320.46 @xmath313 0.71875@xmath333 1/3 , which is in qualitative agreement with the numerical results of @xcite .
this estimate can be generalized to the case when the matrix elements @xmath314 are distributed in the interval @xmath334 $ ] for sufficiently large @xmath335 . in this case
, the probability that ( [ range ] ) is satisfied remains the same , @xmath331 0.71875 , while the probability that @xmath330 is estimated as @xmath336 . with the assumption of statistical independence , the total probability of success
is @xmath337 , and in the limit of large @xmath338 we have @xmath339 which exceeds slightly the value for @xmath340 .
in absence of tunnelling , the dynamics of the large spin can be characterized by classical equations of motion for the spin projections treated as c - numbers in the form @xcite @xmath341 , \label{heis_dyn_1}\ ] ] with @xmath342 in coordinate form and in terms of the dimensionless spin projections , this yields @xmath343 where we took into account that since @xmath186 does not contain the @xmath344 component , @xmath345 . in the case
when the effective magnetic field
@xmath346 does not explicitly depend on time , the system ( [ heis_dyn_1 ] ) , ( [ heis_dyn_2 ] ) has two independent integrals of motion @xmath347 the first ( [ int2 ] ) reflects the conservation of total spin and also holds for an arbitrary time - dependent field @xmath348 , whereas the second integral corresponds to the adiabatic invariant of the system ( [ heis_dyn_1 ] ) , ( [ heis_dyn_2 ] ) . since in our case @xmath348 is parametrically time - dependent , the adiabatic invariant is conserved approximately for sufficiently slow parametric evolution .
note that the adiabatic solutions always play the role of envelope solutions .
this means that on average , the spin closely follows the adiabatic solution , but there are fast oscillatory - type motions superimposed on the slow adiabatic evolution .
basically , the adiabatic approximation in the classical case is applicable when the slow motion is much slower than the fast oscillatory motion .
this exactly corresponds to the adiabatic evolution of the spin system in the quantum case @xcite .
making use of ( [ int2 ] ) and taking into account that at the instant @xmath349 , the total spin was parallel to the @xmath350-axis , we obtain @xmath351 , or @xmath352 implying that the total spin is always parallel the effective magnetic field @xmath346 .
therefore , the adiabatic evolution of the large spin can be simply described as the situation when the spin follows the effective field ( on average ) .
we note that at this level , there is a direct correspondence between the adiabatic classical solution and the quasiclassical wave functions of the large spin parallel to @xmath353 . from ( [ adiabatic_1 ] )
, it follows that this direction can be identified with the effective magnetic field @xmath354 .
this justifies the variational approach introduced in @xcite , identifying the variational wave functions with the adiabatic ground states along the evolution paths when the total spin is parallel to @xmath355 .
therefore , one can observe that in the absence of tunnelling , the general hwp is solved essentially by the classical paths of the qaa .
we apply the quantum adiabatic evolution algorithms with different paths @xcite to the generalized hamming weight problem that corresponds to the specific case of the random satisfiability problem defined in ( [ 3sat_gen ] ) .
we show that any random evolution path produced by this algorithm for the hwp can be obtained by using 6 specific deterministic basis operators with random weights and therefore is parameterized by 6 independent random numbers .
therefore , the approach to qaa with different paths can still be reduced to the large spin dynamics for the hwp .
we show that only one of these generators can be a universal
driver fundamentally responsible for tunnelling suppression for arbitrary hwp and therefore the problem of constructing such a universal driver reduces to the definition of its weight @xmath277 . due to the possible reflection symmetry of the cost function , any particular case of the general hwp
can be solved with one of the two values of the weight with @xmath356 that lead to the collapse of the adiabatic evolution .
the global bifurcations correspond to the onset of tunnelling in qaa and lead to the failure of the algorithm .
in contrast , the local bifurcations while still corresponding to exponentially small minimum gap only lead to the decrease of the probability of success by a factor of 2 . since in a given problem function @xmath357 is a low degree polynomial in its arguments there are only a few local bifurcations possible .
however , the phenomenon of local bifurcations may become important for more difficult random optimization problems . assuming the number of such bifurcations @xmath358 is large the probability of success is reduced by a factor of @xmath359 .
for @xmath360 that scales up with @xmath0 that would lead to the failure of the algorithm .
we want to thank edward farhi ( mit ) for useful discussion .
this work was supported in part by the national security agency ( nsa ) and advanced research and development activity ( arda ) under army research office ( aro ) contract number xxxxxx - xx - x - xxxx , we also want to acknowledge the support of nasa is revolutionary computing algorithms program ( project no : 749 - 40 ) .
the random real symmetric @xmath125 random matrix @xmath126 introduced in ( [ hea ] ) , describes the transitions between each of the @xmath361 states for each clause involving @xmath362 bits @xcite .
this matrix has @xmath363 independent matrix elements and can be presented in the form @xmath364 where @xmath365 correspond to the transitions involving one , two and three bits , respectively . for each realization
, we have @xmath366 where @xmath367 are the real coefficients . the indices
@xmath368 label the bits , @xmath369 @xmath370 and @xmath371 are the pauli sigma - matrices of raising / lowering , @xmath350-projection and @xmath154-projection , respectively and @xmath372 is a spin projection variable .
note that the operator @xmath373 is a projector onto the spin state @xmath374 for the bit @xmath13 .
clearly , the number of independent parameters in ( [ a_c_2 ] ) is @xmath375 , where the three terms of the sum correspond to @xmath376 and @xmath377 , respectively .
note that this number of parameters equals the number of independent matrix elements of @xmath126 estimated above . in the matrix form
, the representation ( [ a_c_2 ] ) yields @xmath378 , \label{a_c_3}\ ] ] where the vector @xmath379 of @xmath380 basis states is @xmath381^t , \label{basis_1}\ ] ] and the lower left portion of the symmetric matrix @xmath382 is obtained by reflection with respect to the diagonal .
the driver @xmath31 is obtained by summation over all clauses . in doing this summation , we take into account that now the bit indices @xmath383 run through all @xmath0 bits , whereas the indices @xmath384 characterizing the realization of @xmath385 still run through the @xmath362 bits ( since the same realization of @xmath385 is applied to all triples of bits ) . the driver @xmath31 is given by @xmath386 where @xmath387 correspond to the transitions involving one , two and three bits , analogous to ( [ a_c_1 ] ) .
@xmath388 one should note that the second term on the r.h.s . for @xmath389 gives a contribution , which is diagonal in @xmath390 representation and therefore leads to the effective re - definition of the cost function .
following the logic of @xcite , we disregard such terms .
also , the commutation relations between the total spin components give contributions @xmath391 to the effective potential and can be neglected in the large - spin limit .
taking this into account , we obtain from ( [ h_e_2 ] ) in the large - spin limit @xmath392 where @xmath393 is a dimensionless spin projection on @xmath394-axis and the coefficients @xmath145 are given by @xmath395 in particular , the deterministic driver considered in @xcite corresponds to @xmath396 , @xmath397 and @xmath398 it follows from ( [ gammas_1 ] ) that in this case , the only non - zero coefficient in ( [ gammas_1 ] ) is @xmath292 .
this corresponds to @xmath399 , which is equivalent to @xmath400 in the large - spin limit according to the above discussion .
taking into account only the @xmath277 term in @xmath31 and expanding up to the 4th order , we obtain the conditions @xmath401 for the @xmath223 bifurcation point @xmath402 in the form @xcite ( cf .
( [ conda3 ] ) ) @xmath403 , \notag \\ \tau _ { c}\left ( 2\beta _ { 2}+6\beta _ { 3}x\right ) & = & -\left ( 1-\tau _ { c}\right ) \left [ 2 - 3x^{2}+\gamma _ { c}\tau _ { c}\left ( -3x\right ) \right ] , \label{bif_2 } \\ 6\tau _ { c}\beta _ { 3 } & = & \left ( 1-\tau _ { c}\right ) \left ( 6x+3\gamma _ { c}\tau _ { c}\right ) , \notag\end{aligned}\ ] ] which have to be solved for @xmath300 , @xmath289 and @xmath350 for the given @xmath281 . solving for @xmath350
, we obtain condition ( [ bifurcation_1 ] ) in the text . r. monasson and r. zecchina , `` entropy of the k - satisfiability problem '' , phys .
lett . , * 76 * , p.3881 ( 1996 ) ; _ ibid _ , `` statistical mechanics of the random k - satisfiability problem '' , phys .
e * 56 * , p.1357 ( 1997 ) . | we provide a theoretical study of the quantum adiabatic evolution algorithm with different evolution paths proposed in @xcite .
the algorithm is applied to a random binary optimization problem ( a version of the 3-satisfiability problem ) where the @xmath0-bit cost function is symmetric with respect to the permutation of individual bits .
the evolution paths are produced , using the generic control hamiltonians @xmath1 that preserve the bit symmetry of the underlying optimization problem . in the case where the ground state of @xmath2 coincides with the totally - symmetric state of an @xmath0-qubit system the algorithm dynamics
is completely described in terms of the motion of a spin-@xmath3 .
we show that different control hamiltonians can be parameterized by a set of independent parameters that are expansion coefficients of @xmath1 in a certain universal set of operators .
only one of these operators can be responsible for avoiding the tunnelling in the spin-@xmath3 system during the quantum adiabatic algorithm .
we show that it is possible to select a coefficient for this operator that guarantees a polynomial complexity of the algorithm for all problem instances .
we show that a successful evolution path of the algorithm always corresponds to the trajectory of a _ classical _ spin-@xmath3 and provide a complete characterization of such paths . |
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.